Peptides and proteins in dendritic assemblies · 1.2 Multivalency in peptide and protein chemistry...

Peptides and proteins in dendritic assemblies

Citation for published version (APA):Baal, van, I. (2007). Peptides and proteins in dendritic assemblies. Eindhoven: Technische UniversiteitEindhoven. https://doi.org/10.6100/IR629916

DOI:10.6100/IR629916

Document status and date:Published: 01/01/2007

Document Version:Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can beimportant differences between the submitted version and the official published version of record. Peopleinterested in the research are advised to contact the author for the final version of the publication, or visit theDOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and pagenumbers.Link to publication

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, pleasefollow below link for the End User Agreement:www.tue.nl/taverne

Take down policyIf you believe that this document breaches copyright please contact us at:[email protected] details and we will investigate your claim.

Download date: 24. May. 2020

https://doi.org/10.6100/IR629916

https://doi.org/10.6100/IR629916

https://research.tue.nl/en/publications/peptides-and-proteins-in-dendritic-assemblies(9450e1ea-0a96-4afa-80bf-fa953514ce26).html

PEPTIDES AND PROTEINS IN

DENDRITIC ASSEMBLIES

PEPTIDES AND PROTEINS IN

DENDRITIC ASSEMBLIES

PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de

Technische Universiteit Eindhoven, op gezag van de

Rector Magnificus, prof.dr.ir. C.J. van Duijn, voor een

commissie aangewezen door het College voor

Promoties in het openbaar te verdedigen

op dinsdag 20 november 2007 om 16.00 uur

door

Ingrid van Baal

geboren te Tilburg

Dit proefschrift is goedgekeurd door de promotor:

prof.dr. E.W. Meijer

Copromotor:

dr. M. Merkx

The research described in this thesis was financially supported by the National Research

School Combination Catalysis

Omslagontwerp: Ingrid van Baal en Koen Pieterse

Druk: GildePrint, Enschede

A catalogue record is available from the Eindhoven University of Technology Library

ISBN: 978-90-386-1110-5

TABLE OF CONTENTS

Chapter 1 1

Chemical Biology Approaches to Multivalent Peptides and Proteins

1.1 Introduction 2

1.2 Multivalency in peptide and protein chemistry 2

1.3 Synthetic tools in peptide and protein chemistry 8

1.4 Recent advances in semi-synthetic multivalent peptides and proteins 13

1.5 Aim and outline of this thesis 15

1.6 References 16

Chapter 2 19

Multivalent Peptides and Proteins Using Native Chemical Ligation

2.1 Introduction 20

2.2 Synthesis of cysteine dendrimers 21

2.3 Synthesis of multivalent peptide dendrimers 22

2.4 Recombinant expression and purification of fluorescent proteins with 24

a C-terminal thioester

2.5 Synthesis of a peptide-protein hybrid dendrimer 26

2.6 Synthesis of protein tetramers 28

2.7 Discussion and conclusions 34

2.8 Experimental section 35

2.9 References 40

Chapter 3 43

Native Chemical Ligation on a Biosensor Chip

3.1 Introduction 44

3.2 Native chemical ligation on a biosensor surface: concept and synthesis 46

3.3 Biosensor surface modification 48

3.4 Immobilization of a peptide-thioester via native chemical ligation 49

3.5 Immobilization of a protein-thioester via native chemical ligation 52

3.6 Immobilization of S-peptide enables measurement of thermodynamic parameters 54

3.7 Discussion and conclusions 56


3.9 References 60

Chapter 4 61

From Phage Display to Dendrimer Display

4.1 Introduction 62

4.2 Selection of streptavidin-binding peptide via phage display 63

4.3 Synthesis of multivalent peptides 65

4.4 Binding studies with streptavidin-binding peptides using SPR 68

4.5 Discussion 72

4.6 Conclusion 76


4.8 References 82

Chapter 5: 85

Non-Covalent Synthesis of Multivalent RNase S Assemblies

5.1 Introduction 86

5.2 Studying the S-peptide–S-protein interaction 88

5.3 Synthesis of a multivalent RNase S dendrimer 93

5.4 Characterization of the RNase S tetramer using mass spectrometry 96

5.5 Development of building blocks for S-protein assemblies 98

5.6 Conclusion and outlook 101


5.8 References 106

Summary 108

Curriculum Vitae 111

Dankwoord 112

1

1

CHEMICAL BIOLOGY APPROACHES TO

MULTIVALENT PEPTIDES AND PROTEINS

Chapter 1

2

1.1 Introduction

Polypeptides play a pivotal role in many biological processes and structural

assemblies, including signal transduction, catalytic activity, immune response and the

structure of the extracellular matrix. Biological and chemical tools to synthesize and modify

proteins and peptides are essential to study and manipulate their structure, function and

interactions with other biomolecules. The strength of biological synthesis is the ability to build

large macromolecules with high fidelity and the application of evolutionary schemes to

optimize binding properties. Despite recent progress in the incorporation of non-natural

amino acids, protein engineering approaches are still limited to a relatively small set of

building blocks and topologies.[1] Synthetic chemistry allows the construction of molecular

assemblies with a much wider variety of functional groups and 3-dimensional topologies, but

is less efficient in the error-free synthesis of large macromolecules. Semi-synthetic or hybrid

molecules consisting of biomacromolecules and synthetic components offer the opportunity

to merge the strengths of both approaches.[2-4] In recent years, the combination of chemical

and biological tools has developed into a new field of research termed chemical biology.[5,6]

Chemical biology tools can provide access to complex protein and peptide assemblies to

facilitate the research of protein and peptide function on a molecular level.

1.2 Multivalency in peptide and protein chemistry

In biological systems, multiple interactions are often employed to increase affinity and

specificity. These multiple interactions are often much stronger than the corresponding

monovalent interaction, an effect referred to as multivalency.[7,8] To understand and employ

this multivalent effect, synthetic organic chemistry approaches have been used to closely

mimic natural multivalent systems. The use of multivalency has promising prospects for

biomedical applications such as vaccines, molecular imaging and target-specific drug

delivery systems. In addition, (semi-)synthetic multivalent structures can be used as model

systems to gain insight into biological interactions.

1.2.1 Theory of multivalency

The valency of a compound can be defined as the number of separate connections it

can form with its complementary counterpart (Figure 1). Multivalent binding can be described

using the additivity of free energies.[9,10] The standard free energy for multivalent binding

!G°multi can be defined as:

!

"Gmulti

o= N"G

mono

o+ "G

interaction

o [1]

where !G°mono is the standard binding free energy of the corresponding monovalent

interaction, N is the valency of the complex and !G°interaction is a balance between the

favorable and unfavorable effects of tethering.


3

Figure 1: Schematic representation of mono- and multivalent binding of a receptor and ligand.

Multivalent binding is often mistaken for cooperative binding, which takes place when

the binding of one ligand influences the binding strength of a receptor toward a subsequent

ligand (or ligands). Cooperativity in biological systems has been defined in the case of

multiple intermolecular binding of monovalent ligands to a multivalent receptor molecule (e.g.

the binding of four O2 molecules to tetrameric hemoglobin shows positive cooperativity).

However, positive cooperativity in multivalent systems has not been characterized and

quantified to date. Whitesides and co-workers introduced an emprical parameter to quantify

the enhanced affinity of multivalent interactions. The affinity enhancement factor ! is defined

as Kmulti /Kmono, where Kmulti and Kmono are the association constants for the multi- and

monovalent complexes, respectively. This enhancement factor ! can be used to compare

efficiency of multivalent ligands, where a molecule with a higher ! value is a more efficient

ligand.[7]

The multivalent binding mode described above assumes a multivalent receptor, as

depicted in Figure 2a. In vivo, other multivalent binding modes may contribute to the potency

of a multivalent ligand. Some proteins possess binding subsites next to the primary binding

site, which can be occupied by a multivalent ligand (Figure 2b). In the case of receptors that

are not oligomeric, multivalent binding can occur via two-dimensional diffusion of receptors in

the fluid bilayer of a cell surface, which may induce signal transduction (Figure 2c).[11,12]

Figure 2: Multivalent binding modes on a cell surface: a) Binding of a divalent ligand to a divalent

receptor; b) Subsite binding of a divalent ligand; c) Receptor clustering via two-dimensional diffusion

on a cell surface. (Adapted from [11])

Chapter 1

4

In literature, most examples of multivalent ligand design, synthesis and binding

properties concern carbohydrate ligands. Fan et al. developed a pentavalent ligand of heat-

labile enterotoxin from E. coli, based on galactose. The pentavalent ligand was designed to

match the spatial arrangement of the receptor (having a 5-fold symmetry) by a rigid synthetic

scaffold, to which the galactose ligands were attached via flexible linkers of varying length.

They demonstrated that the ligand with spacers that matched the binding site geometry in

the target best, showed the highest affinity, with a 104-fold gain compared to the monovalent

ligand.[13] Another example of large multivalent effects is provided by the work of Kitov et al.,

who designed a decavalent carbohydrate ligand of Shiga-like toxin I, capable of binding two

pentavalent toxin assemblies. This multivalent ligand showed an in vitro inhibitory activity that

was 1–10-million-fold higher than that of the monovalent ligands.[14] Kiessling and co-workers

have explored the use of multivalency to probe and control cell surface recognition

events.[11,15,16] For example, they used low-affinity multivalent interactions to selectively

recognize tumor cells over normal cells.[17]

1.2.2 Multivalent proteins via recombinant strategies

The synthesis and application of multivalent peptides and proteins has been much

less explored, due to synthetic challenges and, for proteins, difficulties of maintaining a

correctly folded structure during chemical conversion. Recombinant antibodies and their

fragments are frequently used due to their high affinity and selectivity. Native antibodies

display multiple antigen-binding sites: IgG is divalent, IgM has 10 binding sites available.

Antibodies have been reduced in size by separating the minimal binding fragments, yielding

single chain fragments (scFv) or single-domain fragments (sdAb, originating from heavy-

chain only antibodies). These smaller fragments display improved penetration into tissues or

tumors, and can be obtained via recombinant expression in bacteria. However, these

antibody fragments are monovalent and they often show decreased affinity compared to

native antibodies. To restore the multivalent binding character and thereby increasing the

affinity, various approaches towards multivalent antibody fragments have been investigated.

Single chain antibody fragments can be linked via short oligopeptide spacers which enable

the self-assembly of multiple fragments into dia-, tria- and tetrabodies, depending on the

linker length (Figure 3a).[18] Other approaches include linking the different antibody-fragments

via disulfides or thioether linkages, but reproducible production of these assemblies can be

difficult. Another more general multimerization strategy is based on the self-assembly of

antibody fragments by oligomerization domains. Examples of self-associating domains that

have been applied to multimerize antibody fragments are leucine zippers, the tetramerization

domain of p53 and the shiga-like toxin B-subunit that self-assembles into a pentameric

structure.[19-21] Multimerization increased both affinity and stability for all constructs. To enable

a multimerization step on demand, complementary binding partners can be used. For

example, the tight interaction between barnase and barstar has been used to make well-

defined antibody fragment dimers and trimers, showing improved affinity and target tissue


5

accumulation in vivo compared to the monovalent antibody fragment (Figure 3b).[22] The

strong interaction between biotin and streptavidin (Kd ~ 10–14 M)[23] has also been used to

multimerize antibody fragments. Streptavidin is a homo-tetrameric protein that can bind four

biotin molecules. Upon biotinylation of the C-terminus of single chain antibody fragments,

these fragments were complexed to streptavidin and the resulting tetrameric “streptabodies”

showed approximately 20-fold higher affinities than monovalent antibody fragments.[24] Using

a similar approach, major histocompatibility complex (MHC) proteins have been multimerized

using the biotin–streptavidin interaction and these resulting assemblies display a higher

affinity than their monovalent counterparts.[25,26]

a) b)

Figure 3: a) Domain-swapped antibody fragments, assembled in dia-, tria- and tetrabodies; [18] b)

Trivalent antibody fragments, assembled via the barnase–barstar interaction. (Adapted from [22])

All of the preceding examples are based on non-covalent interactions, however.

Furthermore, often a large protein is used as scaffolding core, which can limit penetration into

tissue and decrease clearance rates. Providing a well-defined, synthetic scaffold to which

multiple bioactive peptides and proteins can be attached would provide much more control

over valency and stability of the resulting assemblies. To obtain multivalent peptides and

proteins, either covalent or non-covalent synthesis can be considered. Covalently linked

multimers are well-defined and stable structures, but cannot adapt to the conformation and

valency of their binding partner. This in contrast to non-covalent peptide and protein

assemblies, which can display dynamic behavior, but at the cost of less control over structure

and stability.

1.2.3 Synthetic scaffolds for multivalent peptides and proteins

A (semi-)synthetic multivalent ligand consists of a synthetic scaffold bearing multiple

ligands via linkers or spacers. Different scaffolds can be considered for the synthesis of

multivalent peptides and proteins, ranging from well-defined low molecular weight scaffolds

Chapter 1

6

(e.g. oligosaccharides, peptides, benzene derivatives, calixarenes) to polymers and

supramolecular constructs such as liposomes and micelles, with higher but less defined

valency numbers (Figure 4).[27,28] For quantitative multivalency studies, scaffolds displaying a

defined number of ligands are preferred. Dendrimers are highly branched monodisperse

macromolecules with a regular three-dimensional architecture. They are synthesized via an

iterative sequence of reaction steps, in which each sequence leads to a higher generation

material, with a higher number of functional end groups.[29] Dendrimers (and dendritic

wedges) are therefore perfectly suited to display multiple ligands in a spatially well-defined

manner.[30]

Figure 4: Synthetic scaffolds for multivalent display of peptides and proteins: a) linear polymer; b)

dendrimer; c) dendritic wedge; d) micelle; e) liposome.

1.2.4 Synthetic multivalent peptides

There are already a number of examples in literature of multivalent peptides based on

synthetic scaffolds. Probably, the most widely known biological application of multivalent


7

peptides is the use of multiple antigen peptides (MAPs) as immunogens to induce an

immune response in vivo. The synthetic scaffolds used for MAPs are lysine-based dendritic

wedges, which are prepared using solid phase peptide synthesis. Subsequently, the

antigenic peptide sequence is synthesized on the multivalent scaffold and the entire

construct is cleaved from the solid support. MAPs allow the tandem assembly of B and T

helper epitopes on the small lysine core, thereby inducing humoral response and subsequent

efficient production of antibodies. Since the first report of the synthesis and use of MAPs by

Tam[31] they have been employed routinely as immunogens for antibody production and

cytotoxic immune responses.[32]

The tripeptide RGD is a common recognition element for "v#3-integrin receptor, which

plays an important role in human metastasis and tumor-induced angiogenesis.[33-35] Cyclo-(-

RGDfK-) is a cyclic derivative that displays improved affinity for integrin[36] and is used for

tumor targeting and imaging applications. Multimerization of the cyclic peptide on a lysine-

based scaffold was used to enhance integrin targeting, resulting in improved tumor uptake

and excretion behavior.[37,38]

Multivalent peptides can also be obtained by employing non-covalent, dynamic

scaffolds, via the self-assembly of amphiphilic peptides into liposomes or micellar structures.

These peptides can be prepared with lipid alkyl tails using an solid phase strategy. Alkyl

chains can be attached either at the N- or C-terminus of the peptide, depending on the

orientation required for appropriate display of the recognition element. A nice example of self-

assembly of lipidated peptides into cylindrical micelles is the work of Hartgerink et al.[39] They

synthesized a peptide with an alkyl tail to promote self-assembly into cylindrical micelles.

Four consecutive cysteine residues were included to cross-link the resulting cylindrical

architecture. A phosphoserine residue was incorporated to allow mineralization of

hydroxyapatite and the integrin-binding RGD-sequence was included to promote cell

adhesion. The nanofibers were able to direct mineralization to form a composite material,

aligned in a similar fashion as observed for collagen fibrils and hydroxyapatite crystals in

bone. Another example of non-covalent synthesis of multivalent peptides shows the potential

for affinity increase by peptide multimerization. Basha et al. described the incorporation of an

anthrax toxin inhibitory peptide into liposomes, thereby increasing the potency of the peptide

by >50000-fold in vitro and enabling neutralization of pentavalent anthrax toxin in vivo. In this

work, peptides that bind to cellular anthrax receptors were identified using phage display

selection and were synthesized with an additional cysteine residue to allow attachment to

thiol-reactive liposomes.[40]

The aforementioned methods are applicable only to the chemical synthesis of

multivalent peptides, however. For many applications, the use of folded proteins is preferred,

due to their generally higher affinity and better proteolytic stability. Recombinant proteins may

be conjugated to synthetic scaffolds via endogenous lysine or cysteine residues, but this

often results in heterogeneity of the resulting assemblies. Furthermore, modification of

Chapter 1

8

endogenous residues may result in reduced activity or decreased stability of the protein. A

more general synthetic strategy based on chemoselective coupling would provide access to

well-defined multivalent proteins and peptides, indifferent of the source of polypeptide.

1.3 Synthetic tools in peptide and protein chemistry

A variety of chemical reactions can be used to further modify or process peptides and

proteins. The nascent side chain functional groups can be used for conjugation. Most often

the primary amine of lysine or the sulfhydryl of cysteine are utilized. This approach does not

require the adjustment of the native structure, but conjugation is aspecific. In recent years,

much effort has been put into the development of bioorthogonal synthetic strategies.

Bioorthogonal labeling strategies for in vivo visualization and detection of target biomolecules

have proven their relevance and will play a key role in the field of chemical biology.[41] In

addition to in vivo applications, bioorthogonal chemistries allow the synthesis of complex

biohybrid structures based on peptides and proteins, and enable the site-specific

immobilization of biomolecules on surfaces.

1.3.1 Oxime ligation

In the early 1990s different methods were developed to connect synthetic peptides to

form larger polypeptides and proteins. Most of these ligation strategies comprised the use of

mutually reactive functional groups based on thiol and imine chemistry to form a non-peptide

bond, thereby circumventing the need for protective groups.[42] One ligation strategy that has

recently gained more attention is the reaction of an aldehyde or ketone with an aminooxy

group to form an oxime bond.[43,44] Dirksen et al. described the use of an aniline catalyst to

accelerate oxime ligation.[45,46] Furthermore, Gilmore et al. reported an efficient way to modify

the N-terminus of proteins via a transamination reaction with pyridoxal-5-phosphate to afford

an aldehyde or ketone (Scheme 1).[47] Recently, Maynard and co-workers described the use

of oxime ligation to immobilize ketone-modified proteins onto polymer films that were micro-

patterned with aminooxy groups.[48] The oxime bond formed in the ligation is stable under

physiological conditions, but is not as stable as an amide bond at acidic or basic pH.

H2NR

HOOC

O

NH3C

HO

CHO

OPO32-

H3CR

O

O R'ONH2

H3CR

O

N

O

R'

Scheme 1: Transamination of a protein/peptide N-terminus, followed by oxime ligation.[47]

1.3.2 Native chemical ligation

The pH lability of non-amide bond linkages directed research towards the

development of chemoselective ligation strategies creating stable amide linkages. In 1994,

Dawson et al. described the reaction of an N-terminal cysteine residue with a C-terminal


9

peptide thioester to form, via an intramolecular rearrangement, a native peptide bond

(Scheme 2).[49] The native chemical ligation reaction proceeds fast and selective at neutral

pH in buffered aqueous media.

O

NH

H2N

HS

R'R S

O R S

O

O

NH

H2NR'

R

O

R'HN

NH

SH

O

+

Scheme 2: Native chemical ligation of an N-terminal cysteine with a C-terminal thioester, yielding a

native peptide bond.[49]

C-terminal thioester peptides can be obtained using Boc-mediated peptide synthesis

using a mercaptopropionic acid derivatized resin. The peptide chain is built up from the free

sulfhydryl group, generating a thioester at the C-terminus of the growing peptide chain. HF-

cleavage releases the thioester-modified peptide, which can react in native chemical

ligation.[50] Peptide thioesters can also be obtained via Fmoc-mediated strategies, but the

poor stability of thioesters under basic conditions necessitate that the thioester is introduced

subsequent to assembly of the peptide chain.[51] Native chemical ligation has proven its

usability in the total synthesis of full-length proteins from such synthetic peptide thioester

fragments.[49,50,52-56]

1.3.3 Expressed protein ligation

The development of a recombinant approach to generate thioesters extended the

applicability of the native chemical ligation reaction towards full-length proteins. In expressed

protein ligation, an expression and purification strategy based on protein splicing is used to

acquire proteins with a C-terminal thioester. Protein splicing is a post-translational process in

which an internal protein fragment (termed intein) is spliced from a precursor protein in a

series of intramolecular rearrangements.[57,58] These inteins can be engineered to promote

only the first step of protein splicing. Proteins that are expressed as N-terminal fusions to

such an engineered intein domain can be cleaved by addition of thiols, releasing the

recombinant protein containing a C-terminal thioester.[59] Subsequent reaction with an N-

terminal cysteine results in a native peptide bond (Scheme 3).

Chapter 1

10

Protein NH

O

HS

Intein CBD Protein S

O

Protein

O

HN

HS

Intein CBD

Protein

O

HSSO3

-

SSO3

-

HSSO3

-

NH

R

HS H2N R

HS

H2N Intein CBD

Scheme 3: Intein-based fusion protein expression system and expressed protein ligation.[59] The

fusion protein contains a chitin-binding domain (CBD) for affinity purification.

Expressed protein ligation (EPL) has been used to incorporate various functionalities

into recombinant proteins, including fluorescent probes, oligonucleotides, unnatural amino

acids and isotope labels. EPL has also been used to immobilized proteins on surfaces in a

chemoselective fashion.[60] In addition, EPL can be used in the semi-synthesis of post-

translationally-modified proteins. Tolbert and Wong incorporated a glycosylated amino acid

into a recombinant model protein (maltose binding protein, MBP) by expressed protein

ligation of a glycosylated peptide with the MBP-thioester. Subsequent reaction of the

glycosylated protein with UDP-galactose in the presence of galactosyltransferase resulted in

extension to a more complex oligosaccharide-modified protein.[61] Waldmann and co-workers

developed an efficient semi-synthetic strategy towards prenylated Rab-proteins using

expressed protein ligation. The family of Rab-proteins has been identified as key regulators

of intracellular vesicular transport in both exocytic and endocytic pathways in eukaryotic

cells.[62] Membrane association is achieved via two geranylgeranyl groups covalently

attached to the C-terminus of the protein. The Rab-protein was expressed and purified using

intein-mediated thioester-formation, and subsequently a prenylated peptide was ligated to the

C-terminus of the protein, affording sufficient amounts of protein for structure elucidation of

complexes with interacting proteins.[63] Bertozzi and co-workers described a similar strategy

towards phospholipidated proteins, where Green Fluorescent Protein (GFP) was modified

with a cysteine-functionalized phospholipid tail. The lipidated protein could be incorporated

into supported lipid bilayers and its behavior in the bilayer was characterized using

fluorescence microscopy.[64]


11

1.3.4 Staudinger ligation

A widely used functional group in bioorthogonal chemistry is the azide. Azides are

abiotic and effectively unreactive towards all biomolecules. In the Staudinger ligation, an

azide reacts with a phosphine to form an aza-ylide, followed by an intramolecular reaction to

form a stable amide bond between the two reactants (Scheme 4a).[65,66] Staudinger ligations

can also be performed using a phosphinothioester instead of a methyl ester-containing

phosphine, resulting in the formation of an amide with no residual atoms (Scheme 4b).[67,68]

This traceless variation of the Staudinger ligation can be used in protein synthesis orthogonal

to native chemical ligation, having the advantage that no cysteine is required at the site of

ligation.[69]

Azides can be incorporated in synthetic peptides as a side-chain functionality or by

post-assembly diazo transfer on the solid phase, prior to cleavage from the resin.[70] The

azido-group can also be incorporated in recombinant proteins by replacing methionine by

azidohomoalanine in the cell culture medium.[71] Alternatively, an azido-functionality can be

introduced at the C-terminus of a protein via a variation of the expressed protein ligation

described above. The intein fusion protein obtained after bacterial expression can be reacted

with a hydrazine-modified azide derivative.[72] These azido-modified proteins can be

immobilized onto solid surfaces for microarray applications.[73,74] The traceless Staudinger

ligation has been applied in combination with native chemical ligation to synthesize full-length

proteins.[69] The Staudinger ligation has also been used for in vivo labeling of biomolecules,

as the azide is a small molecular probe that does not interfere with biological processing and

can be visualized using a phosphino-modified imaging label.[75-77]

R N3 +

R' P

OCH3

O

PhPh

N2

R' P+

OCH3

O

PhPh

N- R

CH3OH

R' P+

N

O

PhPh

RH2O

R' P

NH

O

PhPh

R

O

a)

b)

R N3 + S

PPh Ph N2

H2O

O

R' S

+PPh2O

R'

-N

R

O

R'

+PPh2

NR

-S

O

R' NH

R

PPh2

HS

+

O

Scheme 4: a) The Staudinger ligation as proposed by Saxon and Bertozzi [65]; b) Alternative approach

using a phosphinothioester, yielding an amide with no residual atoms of the phosphinothiol.[67,68]

Chapter 1

12

1.3.5 [3+2] cycloadditions

Azides can also participate in Huisgen-type [3+2] cycloadditions, such as the copper-

catalyzed click-reaction described by Sharpless and co-workers.[78] In this click-reaction, an

azide reacts fast and regioselective with an alkyne derivative to form exclusively the 1,4-

disubstituted 1,2,3-triazole (Scheme 5a). The Cu(I) catalyst necessary for the click reaction is

toxic to bacterial and mammalian cells, however, and can give rise to protein unfolding and

precipitation. The use of an activated alkyne substrate, for example by employing ring strain

as in cyclooctyne-derivatives, precludes the necessity of a catalyst (Scheme 5b).[79] Alkynes

can be incorporated in recombinant proteins by replacing methionine by 2-amino-5-hexynoic

acid in the cell culture medium.[80] Click chemistry is of particular use in proteomic

applications such as activity-based protein profiling, due to its high sensitivity.[81,82] In addition,

click chemistry can be used to immobilize proteins on surfaces onto either azido- or alkyne-

modified solid supports.[83,84]

R N3 + R'

Cu(I),ligand

R'

NN

N R

R N3 +NN

N

R'

R

R'

a)

b)

Scheme 5: Huisgen-type [3+2] cycloadditions: a) copper-catalyzed click reaction [78]; b) strain-

promoted cycloaddition.[79]

1.3.6 Chemoselective synthesis of protein-polymer biohybrid structures

A beautiful example of combining different bioorthogonal coupling strategies to

synthesize a biohybrid protein-polymer is the chemical synthesis of a homogeneous polymer-

modified erythropoiesis protein. Human erythropoietin (Epo) is a natural glycoprotein

hormone that regulates erythroid cell proliferation, differentiation and maturation. In the

synthetic protein, the four natural glycans were replaced by two monodisperse, branched

poly(ethylene glycol) chains bearing in total eight negative charges (Figure 5) The synthetic

scheme towards the full-length protein with appending polymer chains comprised both native

chemical ligation (to construct the peptide backbone) and oxime ligation (to introduce the

polymer side chains). The protein was constructed from four peptide fragments, requiring

three cysteine residues at the ligation sites. The cysteines at positions 89 and 117 were

alkylated after the first two ligation steps, rendering them inactive in disulfide formation during

protein folding. The fully synthesized and folded protein was tested for biological activity and

showed increased potency and circulation time in vivo compared to natural Epo.[85]


13

Figure 5: a) Molecular structure of synthetic polymer-modified erythropoiesis protein; b) Structure of

the branched, negatively charged monodisperse polymer sidechain; c) Synthetic strategy towards the

synthetic protein. (Adapted from [85])

1.4 Recent advances in semi-synthetic multivalent peptides and proteins

In recent years, different chemoselective strategies have been employed to

synthesize multivalent peptides and proteins in a covalent fashion. The integrin recognition

peptide sequence RGD is a favorite model ligand for multivalency studies and has therefore

been used to illustrate a range of chemoselective multimerization strategies. Liskamp and co-

workers used microwave-assisted click-chemistry to synthesize multivalent c[RGDfK] peptide

dendrimers from alkyne-modified dendrimers.[86] In a further study, they synthesized

radiolabeled di- and tetravalent c[RGDfK]-modified dendritic wedges, and compared the

affinities for "v#3 integrin to the affinity of the monovalent c[RGDfK] peptide in a competitive

binding assay, showing some multivalent effect (4-fold increase in affinity for tetravalent

peptide). In vivo studies showed enhanced tumor uptake of the tetravalent peptide wedge.[87]

Garanger et al. used oxime ligation to synthesize multivalent c[RGDfK] peptides, grafted on a

cyclic peptide scaffold.[88] Dirksen et al. showed that native chemical ligation can be used to

prepare multivalent RGDS peptides using poly(lysine) dendritic wedges.[89] After synthesis of

the peptide wedge, the thioproline residue at the focal point of the wedge could be converted

into an N-terminal cysteine, allowing a second native chemical ligation reaction with another

Chapter 1

14

dendritic wedge, equipped with multiple DTPA ligands at its periphery and a thioester

functionality at the focal point.

An example of covalent clustering of proteins onto a synthetic scaffold is the

synthesis of hemoglobin dendrimers described by Kluger and Zhang.[90] They used a fourth

generation PAMAM dendrimer to cluster 3–5 cross-linked hemoglobins together. Cross-links

were made by reacting the $-amino groups of Lys-82 of the two #-subunits of hemoglobin,

leaving one ester of the cross-linking reagent available for reaction with primary amines of

the dendrimer periphery. The clustering of hemoglobin to the dendrimer caused an altered

oxygen binding behavior beyond that caused by the modification itself. This strategy provides

a structurally defined but polydisperse protein assembly. Moreover, the synthetic strategy

used is very specific for hemoglobin conjugation and not applicable to other peptides and

proteins in a straightforward manner.

Equipping proteins with phospholipid tails enables the formation of multivalent protein

assemblies on the surface of liposomes or micellar structures, similar to the peptide example

described above. Reulen et al. recently described the synthesis of a phospholipid-

functionalized collagen-binding protein (CNA-35) and its subsequent assembly into

liposomes, resulting in a 150-fold increase in affinity for collagen (Figure 6).[91] This type of

protein-modified liposomes can find application as drug-delivery containers or carriers of

contrast agents (e.g. MRI, ultrasound) in molecular imaging.

Figure 6: a) Incorporation of phospholipid tail in collagen binding protein CNA-35 via native chemical

ligation; b) Liposome functionalized with CNA-35 proteins.[91]


15

1.5 Aim and outline of this thesis

The use of synthetic scaffolds to synthesize multivalent proteins provides a freedom

in the organization of protein domains that is not possible using protein engineering

approaches alone. Ideally, such a scaffold should allow the synthesis of well-defined

multivalent ligands of both recombinant proteins and synthetic peptides using a conjugation

chemistry that is unlikely to interfere with the biological activity. Covalent conjugation

strategies can facilitate the initial characterization of these new biomolecular constructs and

allow mechanistic studies on multivalent protein/peptide interactions. The aim of this

research was to develop general methods to obtain well-defined protein and peptide

assemblies, and to study multivalent interactions of these assemblies in a controlled fashion.

Ideally, the synthetic strategy to be used is chemo- and regioselective and reaction rates are

fast. Furthermore, catalysts or other reagents that can compromise protein stability must be

avoided, as well as harsh conditions (e.g. high temperature, organic solvents, basic of strong

acidic pH). Native chemical ligation was selected as the synthetic strategy to synthesize

multivalent peptides and proteins, since it is a fast and efficient reaction that is suitable for

both peptide and protein conjugation.

Chapter 2 describes the development of a general method to obtain multivalent

peptides and proteins using native chemical ligation to a dendritic scaffold. Different

generations of poly(propylene imine) dendrimers were functionalized with N-terminal cysteine

residues enabling the native chemical ligation of multiple peptide and protein thioesters.

Using this approach, dendrimers with 4–16 peptides were prepared efficiently and in a well-

defined manner. Native chemical ligation of a protein thioester with a cysteine-modified

dendrimer followed by ligation with peptide thioesters gave access to novel hybrid peptide-

protein dendrimers. Ligation of 4 equivalents of protein thioester with a cysteine-modified

dendrimer yielded a branched, multivalent protein tetramer.

In Chapter 3, peptide and protein thioester building blocks are used to functionalize

biosensor chip surfaces via native chemical ligation. A streptavidin-binding peptide was

immobilized via its C-terminus onto a cysteine-modified biosensor chip, and subsequent

binding experiments with streptavidin showed specific and reproducible binding to the

peptide surface. Short ligation steps of peptide thioester were alternated with streptavidin

binding experiments on a single chip. This provided an increased peptide loading after each

ligation step, yielding enhanced protein-binding capacity. Also, recombinant proteins were

immobilized via their C-terminus onto the biosensor chip. Immobilization of S-peptide onto

the biosensor chip allowed for characterization of the binding kinetics with S-protein to form

ribonuclease S.

In Chapter 4, native chemical ligation is used to prepare multivalent peptides based

on a streptavidin-binding peptide sequence derived from phage display. The synthetic

multivalent scaffolds were used to mimic the multivalent character of the peptides on the

head of a phage, without the presence of the phagemid coat proteins or genetic information.

Chapter 1

16

Peptides with different valency (from 1 to 4 copies per scaffold) and spacing were prepared

and their affinity for streptavidin was measured using surface plasmon resonance. All

multivalent peptides showed a significant increase in affinity compared to their monovalent

counterpart. A binding model was used to describe the multivalent effect in a quantitative

manner.

The specific and strong interaction between S-peptide and S-protein to form

ribonuclease S is used in Chapter 5 to assemble more complex multivalent protein systems.

Synthetic S-peptide with a C-terminal thioester was reacted with a tetravalent cysteine-

functionalized dendritic wedge, yielding an S-peptide wedge. This S-peptide wedge was

shown to complex four S-proteins, yielding an enzymatically active RNase S tetramer.

Thioester-modified RNase A was obtained via recombinant expression as a precursor in the

synthesis of multivalent S-protein assemblies, which will enable the construction of more

complex protein–peptide assemblies in the near future.

1.6 References

[1] Wang, L.; Schultz, P. G. Angew. Chem., Int. Ed. Engl. 2005, 44, (1), 34-66.

[2] Tsai, C.-J.; Zheng, J.; Aleman, C.; Nussinov, R. Trends Biotechnol. 2006, 24, (10), 449-454.

[3] Merzlyak, A.; Lee, S.-W. Curr. Opin. Chem. Biol. 2006, 10, (3), 246-252.

[4] Zhou, M.; Bentley, D.; Ghosh, I. J. Am. Chem. Soc. 2004, 126, (3), 734-735.

[5] Waldmann, H. Bioorg. Med. Chem. 2003, 11, (14), 3045-3051.

[6] Waldmann, H.; Famulok, M. ChemBioChem 2001, 2, 3-6.

[7] Mammen, M.; Choi, S.-K.; Whitesides, G. M. Angew. Chem., Int. Ed. Engl. 1998, 37, (20),

2755-2794.

[8] Kiessling, L. L.; Strong, L. E.; Gestwicki, J. E., Annual reports in medicinal chemistry, 2000,

35, 321-330.

[9] Kitov, P. I.; Bundle, D. R. J. Am. Chem. Soc. 2003, 125, (52), 16271-16284.

[10] Jencks, W. P. Proc. Natl. Acad. Sci. U.S.A. 1981, 78, (7), 4046-4050.

[11] Kiessling, L. L.; Gestwicki, J. E.; Strong, L. E. Curr. Opin. Chem. Biol. 2000, 4, (6), 696-703.

[12] Bertozzi, C. R.; Kiessling, L. L. Science 2001, 291, (5512), 2357-2364.

[13] Fan, E.; Zhang, Z.; Minke, W. E.; Hou, Z.; Verlinde, C. L. M. J.; Hol, W. G. J. J. Am. Chem. Soc. 2000, 122, (11), 2663-2664.

[14] Kitov, P. I.; Sadowska, J. M.; Mulvey, G.; Armstrong, G. D.; Ling, H.; Pannu, N. S.; Read, R.

J.; Bundle, D. R. Nature 2000, 403, (6770), 669-72.

[15] Gestwicki, J. E.; Strong, L. E.; Cairo, C. W.; Boehm, F. J.; Kiessling, L. L. Chem. Biol. 2002, 9,

(2), 163-169.

[16] Kiessling, L., L.; Gestwicki, J., E.; Strong, L., E. Angew. Chem., Int. Ed. Engl. 2006, 45, (15),

2348-68.

[17] Carlson, C. B.; Mowery, P.; Owen, R. M.; Dykhuizen, E. C.; Kiessling, L. L. ACS Chem. Biol. 2007, 2, (2), 119-127.

[18] Todorovska, A.; Roovers, R. C.; Dolezal, O.; Kortt, A. A.; Hoogenboom, H. R.; Hudson, P. J. J. Immunol. Methods 2001, 248, (1-2), 47-66.

[19] Rheinnecker, M.; Hardt, C.; Ilag, L. L.; Kufer, P.; Gruber, R.; Hoess, A.; Lupas, A.;

Rottenberger, C.; Pluckthun, A.; Pack, P. J Immunol 1996, 157, (7), 2989-2997.

[20] Pack, P.; Plueckthun, A. Biochemistry 1992, 31, (6), 1579-1584.

[21] Zhang, J.; Tanha, J.; Hirama, T.; Khieu, N. H.; To, R.; Tong-Sevinc, H.; Stone, E.; Brisson, J.-

R.; Roger MacKenzie, C. J. Mol. Biol. 2004, 335, (1), 49-56.

[22] Deyev, S. M.; Waibel, R.; Lebedenko, E. N.; Schubiger, A. P.; Pluckthun, A. Nat. Biotech. 2003, 21, (12), 1486-1492.

[23] Green, N. M. Adv. Protein Chem. 1975, 29, 85.


17

[24] Cloutier, S. M.; Couty, S.; Terskikh, A.; Marguerat, L.; Crivelli, V.; Pugnieres, M.; Mani, J. C.;

Leisinger, H. J.; Mach, J. P.; Deperthes, D. Mol. Immunol. 2000, 37, (17), 1067-1077.

[25] Hugues, S.; Malherbe, L.; Filippi, C.; Glaichenhaus, N. J. Immunol. Methods 2002, 268, (1),

83-92.

[26] Dunbar, P. R.; Ogg, G. S. J. Immunol. Methods 2002, 268, (1), 3-7.

[27] Stiriba, S.-E.; Frey, H.; Haag, R. Angew. Chem., Int. Ed. Engl. 2002, 41, (8), 1329-1334.

[28] Lee, Y.; Sampson, N. S. Curr. Opin. Struct. Biol. 2006, 16, (4), 544-550.

[29] Bosman, A. W.; Janssen, H. M.; Meijer, E. W. Chem. Rev. 1999, 99, (7), 1665-1688.

[30] Boas, U.; Heegaard, P. M. H. Chem. Soc. Rev. 2004, 33, (1), 43-63.

[31] Tam, J. P. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, (15), 5409-5413.

[32] Sadler, K.; Tam, J. P. Rev. Mol. Biotechnol. 2002, 90, (3-4), 195-229.

[33] Pierschbacher, M. D.; Ruoslahti, E. Nature 1984, 309, (5963), 30-33.

[34] Pierschbacher, M. D.; Ruoslahti, E. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, (19), 5985-5988.

[35] Ruoslahti, E.; Pierschbacher, M. D. Science 1987, 238, (4826), 491-7.

[36] Aumailley, M.; Gurrath, M.; Muller, G.; Calvete, J.; Timpl, R.; Kessler, H. FEBS Lett. 1991,

291, (1), 50-54.

[37] Thumshirn, G.; Hersel, U.; Goodman, S. L.; Kessler, H. Chem. -Eur. J. 2003, 9, (12), 2717-

2725.

[38] Poethko, T.; Schottelius, M.; Thumshirn, G.; Herz, M.; Haubner, R.; Henriksen, G.; Kessler, H.;

Schwaiger, M.; Wester, H.-J. Radiochimica Acta 2004, 92, (4-6), 317-327.

[39] Hartgerink, J. D.; Beniash, E.; Stupp, S. I. Science 2001, 294, (5547), 1684-1688.

[40] Basha, S.; Rai, P.; Poon, V.; Saraph, A.; Gujraty, K.; Go, M. Y.; Sadacharan, S.; Frost, M.;

Mogridge, J.; Kane, R. S. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, (36), 13509-13513.

[41] Prescher, J. A.; Bertozzi, C. R. Nat Chem Biol 2005, 1, (1), 13-21.

[42] James P. Tam; Yu, Q.; Miao, Z. Peptide Science 1999, 51, (5), 311-332.

[43] Rose, K. J. Am. Chem. Soc. 1994, 116, (1), 30-3.

[44] Gaertner, H. F.; Rose, K.; Cotton, R.; Timms, D.; Camble, R.; Offord, R. E. Bioconjugate Chem. 1992, 3, (3), 262-268.

[45] Dirksen, A.; Hackeng, T. M.; Dawson, P. E. Angew. Chem., Int. Ed. Engl. 2006, 45, (45),

7581-7584.

[46] Dirksen, A.; Dirksen, S.; Hackeng, T. M.; Dawson, P. E. J. Am. Chem. Soc. 2006, 128, (49),

15602-15603.

[47] Gilmore, J. M.; Scheck, R. A.; Esser-Kahn, A. P.; Joshi, N. S.; Francis, M. B. Angew. Chem., Int. Ed. Engl. 2006, 45, (32), 5307-5311.

[48] Christman, K. L.; Broyer, R. M.; Tolstyka, Z. P.; Maynard, H. D. J. Mater. Chem. 2007, 17,

(19), 2021-2027.

[49] Dawson, P. E.; Muir, T. W.; Clark-Lewis, I.; Kent, S. B. Science 1994, 266, (5186), 776-9.

[50] Hackeng, T. M.; Griffin, J. H.; Dawson, P. E. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, (18),

10068-73.

[51] Shin, Y.; Winans, K. A.; Backes, B. J.; Kent, S. B. H.; Ellman, J. A.; Bertozzi, C. R. J. Am. Chem. Soc. 1999, 121, (50), 11684-11689.

[52] Muir, T. W.; Dawson, P. E.; Kent, S. B. H. Methods Enzymol. 1997, 289, (Solid-Phase Peptide

Synthesis), 266-298.

[53] Lu, W.; Starovasnik, M. A.; Kent, S. B. FEBS Lett. 1998, 429, (1), 31-5.

[54] Bang, D.; Chopra, N.; Kent Stephen, B. H. J. Am. Chem. Soc. 2004, 126, (5), 1377-83.

[55] Johnson, E. C. B.; Durek, T.; Kent, S. B. H. Angew. Chem., Int. Ed. Engl. 2006, 45, (20),

3283-3287.

[56] Dawson, P. E. Methods Enzymol. 1997, 287, 34-45.

[57] Kane, P. M.; Yamashiro, C. T.; Wolczyk, D. F.; Neff, N.; Goebl, M.; Stevens, T. H. Science 1990, 250, (4981), 651-7.

[58] Hirata, R.; Ohsumi, Y.; Nakano, A.; Kawasaki, H.; Suzuki, K.; Anraku, Y. J. Biol. Chem. 1990,

265, (12), 6726-33.

[59] Muir, T. W.; Sondhi, D.; Cole, P. A. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, (12), 6705-6710.

[60] Camarero, J. A.; Kwon, Y.; Coleman, M. A. J. Am. Chem. Soc. 2004, 126, (45), 14730-14731.

[61] Tolbert, T. J.; Wong, C.-H. J. Am. Chem. Soc. 2000, 122, (23), 5421-5428.

[62] Bock, J. B.; Matern, H. T.; Peden, A. A.; Scheller, R. H. Nature 2001, 409, (6822), 839-841.

Chapter 1

18

[63] Watzke, A.; Brunsveld, L.; Durek, T.; Alexandrov, K.; Rak, A.; Goody, R. S.; Waldmann, H.

Org. Biomol. Chem. 2005, 3, (7), 1157-1164.

[64] Grogan, M. J.; Kaizuka, Y.; Conrad, R. M.; Groves, J. T.; Bertozzi, C. R. J. Am. Chem. Soc. 2005, 127, (41), 14383-14387.

[65] Saxon, E.; Bertozzi, C. R. Science 2000, 287, (5460), 2007-2010.

[66] Lin, F. L.; Hoyt, H. M.; vanHalbeek, H.; Bergman, R. G.; Bertozzi, C. R. J. Am. Chem. Soc. 2005, 127, (8), 2686-2695.

[67] Nilsson, B. L.; Kiessling, L. L.; Raines, R. T. Org. Lett. 2000, 2, (13), 1939-41.

[68] Saxon, E.; Armstrong, J. I.; Bertozzi, C. R. Org. Lett. 2000, 2, (14), 2141-2143.

[69] Nilsson, B. L.; Hondal, R. J.; Soellner, M. B.; Raines, R. T. J. Am. Chem. Soc. 2003, 125, (18),

5268-5269.

[70] Rijkers, D. T. S.; van Vugt, H. H. R.; Jacobs, H. J. F.; Liskamp, R. M. J. Tetrahedron Lett. 2002, 43, (20), 3657-3660.

[71] Kiick, K. L.; Saxon, E.; Tirrell, D. A.; Bertozzi, C. R. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, (1),

19-24.

[72] Kalia, J.; Raines, R. T. ChemBioChem 2006, 7, (9), 1375-1383.

[73] Watzke, A.; Koehn, M.; Gutierrez-Rodriguez, M.; Wacker, R.; Schroeder, H.; Breinbauer, R.;

Kuhlmann, J.; Alexandrov, K.; Niemeyer, C. M.; Goody, R. S.; Waldmann, H. Angew. Chem., Int. Ed. Engl. 2006, 45, (9), 1408-1412.

[74] Kalia, J.; Abbott, N. L.; Raines, R. T. Bioconjugate Chem. 2007, 18, (4), 1064-1069.

[75] Saxon, E.; Luchansky, S. J.; Hang, H. C.; Yu, C.; Lee, S. C.; Bertozzi, C. R. J. Am. Chem. Soc. 2002, 124, (50), 14893-14902.

[76] Prescher, J. A.; Dube, D. H.; Bertozzi, C. R. Nature 2004, 430, (7002), 873-877.

[77] Dube, D. H.; Prescher, J. A.; Quang, C. N.; Bertozzi, C. R. Proc. Natl. Acad. Sci. U.S.A. 2006,

103, (13), 4819-4824.

[78] Rostovtsev, V. V.; Green, L. G.; Fokin, V. V.; Sharpless, K. B. Angew. Chem., Int. Ed. Engl. 2002, 41, (14), 2596-2599.

[79] Agard, N. J.; Prescher, J. A.; Bertozzi, C. R. J. Am. Chem. Soc. 2004, 126, (46), 15046-

15047.

[80] van Hest, J. C.; Kiick, K. L.; Tirrell, D. A. J. Am. Chem. Soc. 2000, 122, (7), 1282-1288.

[81] Agard, N. J.; Baskin, J. M.; Prescher, J. A.; Lo, A.; Bertozzi, C. R. ACS Chem. Biol. 2006, 1,

(10), 644-648.

[82] Speers, A. E.; Cravatt, B. F. Chem. Biol. 2004, 11, (4), 535-546.

[83] Sun, X.-L.; Stabler, C. L.; Cazalis, C. S.; Chaikof, E. L. Bioconjugate Chem. 2006, 17, (1), 52-

57.

[84] Gauchet, C.; Labadie, G. R.; Poulter, C. D. J. Am. Chem. Soc. 2006, 128, (29), 9274-9275.

[85] Kochendoerfer, G. G.; Chen, S.-Y.; Mao, F.; Cressman, S.; Traviglia, S.; Shao, H.; Hunter, C.

L.; Low, D. W.; Cagle, E. N.; Carnevali, M.; Gueriguian, V.; Keogh, P. J.; Porter, H.; Stratton,

S. M.; Wiedeke, M. C.; Wilken, J.; Tang, J.; Levy, J. J.; Miranda, L. P.; Crnogorac, M. M.;

Kalbag, S.; Botti, P.; Schindler-Horvat, J.; Savatski, L.; Adamson, J. W.; Kung, A.; Kent, S. B.

H.; Bradburne, J. A. Science 2003, 299, (5608), 884-887.

[86] Rijkers, D. T. S.; Esse, G. W. v.; Merkx, R.; Brouwer, A. J.; Jacobs, H. J. F.; Pieters, R. J.;

Liskamp, R. M. J. Chem. Commun. 2005, (36), 4581-4583.

[87] Dijkgraaf, I.; Rijnders, A. Y.; Soede, A.; Dechesne, A. C.; Esse, G. W. v.; Brouwer, A. J.;

Corstens, F. H. M.; Boerman, O. C.; Rijkers, D. T. S.; Liskamp, R. M. J. Org. Biomol. Chem. 2007, 5, (6), 935-944.

[88] Garanger, E.; Boturyn, D.; Coll, J.-L.; Favrot, M.-C.; Dumy, P. Org. Biomol. Chem. 2006, 4,

(10), 1958-1965.

[89] Dirksen, A.; Meijer, E. W.; Adriaens, W.; Hackeng, T. M. Chem. Commun. 2006, (15), 1667-

1669.

[90] Kluger, R.; Zhang, J. J. Am. Chem. Soc. 2003, 125, (20), 6070-6071.

[91] Reulen, S. W. A.; Brusselaars, W. W. T.; Langereis, S.; Mulder, W. J. M.; Breurken, M.;

Merkx, M. Bioconjugate Chem. 2007, 18, (2), 590-596.

19

2

MULTIVALENT PEPTIDES AND PROTEINS USING

NATIVE CHEMICAL LIGATION*

Abstract:

A general synthetic strategy is described to obtain multivalent peptides and proteins

using native chemical ligation. Different generations of poly(propylene imine) dendrimers

were functionalized with N-terminal cysteine residues to facilitate the native chemical ligation

reaction with C-terminal thioesters. Solid phase peptide synthesis was used to obtain

different C-terminal thioester peptides. Subsequent ligation of these peptides with cysteine-

functionalized dendrimers yielded multivalent peptide dendrimers of different generations

with 4 to 16 peptides per dendrimer. This native chemical ligation strategy could be

expanded to recombinant proteins by employing intein-mediated protein expression and

purification to obtain fluorescent proteins modified with a C-terminal thioester. Native

chemical ligation of GFP-MESNA with a cysteine-modified dendrimer followed by ligation with

peptide thioesters gives access to novel hybrid peptide-protein dendrimers. Ligation of 4

equivalents of GFP-MESNA with the cysteine-modified dendrimer yielded a branched,

multivalent protein tetramer. Size exclusion chromatography combined with mass

spectrometry proved to be an invaluable tool to study these complex bio-macromolecules.

* Part of this work has been published: van Baal, I.; Malda, H.; Synowsky, S.A.; van Dongen,

J.L.J.; Merkx, M.; Meijer, E.W., Angew. Chem., Int. Ed. Engl. 2005, 44 (32), 5052-5057.

Chapter 2

20

2.1 Introduction

Multiple, simultaneous interactions are often used in biology to enhance the affinity

and specificity of binding, an effect that is known as multivalency.[1] The principle of

multivalency has been recognized as an important strategy for the development of

(semi)synthetic ligands with high affinity and specificity for biological targets.[1-6] Dendrimers

are well-defined, hyperbranched polymers with a high density of functional groups and are

therefore attractive scaffolds for multivalent display of natural ligands.[7-9] Dendrimers have

been successfully applied in the design of multivalent sugar ligands,[10-12] but much less work

has been reported on multivalent peptide and protein dendrimers. Solid phase peptide

synthesis has been used to construct dendritic peptide wedges by condensation of

successive generations of lysine residues followed by functionalization of the ! and " amine

groups with peptides.[13-15] These so-called Multiple Antigen Peptides (MAPs) typically contain

2–8 peptide chains and have been found to be more immunogenic than single peptides.

Dendrimers functionalized with cysteine-reactive chloroacetyl[16] and maleimide[17] groups

have been used to conjugate cysteine-containing peptides, but this strategy cannot be

applied to peptides and proteins containing functionally important native cysteines. The few

reports of protein ligation to dendrimers typically involve the coupling of a single protein (e.g.

antibody) using non-specific conjugation chemistry resulting in conjugation at multiple sites

on the protein. Recently, the preparation of multivalent hemoglobin dendrimers was reported

that involved specific conjugation of a 4th generation PAMAM dendrimer to hemoglobin using

a hemoglobin-specific cross-linking agent.[18] However, no general synthetic strategy is

currently available that allows conjugation of dendrimers with both oligopeptides and

recombinant proteins in a chemoselective manner.

Native chemical ligation was first reported by Dawson et al. as a unique method to

connect two unprotected peptide fragments to form a native peptide bond.[19] This

chemoselective reaction occurs spontaneously between a peptide with a C-terminal thioester

and a peptide with an N-terminal cysteine residue under aqueous conditions at neutral pH

(Scheme 1). Native chemical ligation has allowed the chemical synthesis of large proteins by

multistep ligation of several peptide fragments, the synthesis of proteins with synthetic

moieties such as fluorescent dyes, biotin tags etc., and the immobilization of peptides and

proteins on material surfaces and on liposomes.[20-25] The application of native chemical

ligation was recently extended to recombinantly expressed proteins by the development of

expression systems based on self-cleavable intein domains that generate proteins containing

N-terminal cysteine or C-terminal thioester groups.[26,27] In this chapter, native chemical

ligation is presented as a general synthetic strategy to conjugate both oligopeptides and

recombinantly expressed proteins to dendrimers resulting in multivalent peptide and protein

dendrimers.

Multivalent Peptides and Proteins using Native Chemical Ligation

21

O

NH

H2N

HS

S

O

HN

O

NH2

O

Peptide 1

Peptide 2

S

O

Peptide 1

S

O

Peptide 1

O

NH

H2N Peptide 2

O

Peptide 1 Peptide 2HN

NH

SH

O

Thiol exchange

Spontaneous Rearrangement

HS

Scheme 1: Native chemical ligation reaction between a C-terminal thioester and an N-terminal

cysteine residue.

2.2 Synthesis of cysteine dendrimers

Dendrimers can either be functionalized with thioester groups or with cysteine

residues to allow native chemical ligation. We chose the latter strategy because the

synthesis of these cysteine dendrimers is more straightforward. Cysteine dendrimers of

generations 1–3 were synthesized by reaction of the amine end groups of poly(propylene

imine) dendrimers with succinimide-activated and trityl-protected cysteines (Scheme 2).

Deprotection of the trityl-protected amine and thiol groups using trifluoroacetic acid (TFA)

afforded the cysteine-functionalized dendrimers 1–3 in good yields (Figure 1).

Scheme 2: Synthetic strategy towards cysteine-functionalized poly(propylene imine) dendrimers.

Reaction conditions: i) Et3N, CH2Cl2; ii) TFA, triethylsilane (2.5% v/v), water (2.5% v/v).

Chapter 2

22

Figure 1: First, second and third generation cysteine-functionalized poly(propylene imine) dendrimers

C1, C2 and C3.

2.3 Synthesis of multivalent peptide dendrimers

To assess the reactivity of the cysteine dendrimers in native chemical ligation

reactions we performed ligation reactions with different peptides functionalized with a

thioester at their C-terminus. Boc-mediated solid phase peptide synthesis was used as


23

described by Hackeng et al. to obtain peptides with a mercaptopropionic acid leucine (MPAL)

thioester functionality.[20] The synthesis starts with the introduction of Boc-Leu on the HF-

labile 4-methylbenzhydryl amine (MBHA) resin. Deprotection of the Leu, followed by coupling

of trityl-protected mercaptopropionic acid yields the trityl-associated mercaptopropionic acid

leucine (TAMPAL) resin, which after deprotection of the trityl functionality can be used for the

synthesis of any desired peptide sequence using standard coupling conditions. Using this

protocol, a model peptide sequence, Ac-LYRAG-MPAL was synthesized and purified using

RP-HPLC (Figure 2a). Ligation of C1 with 4.4 equivalents of Ac-LYRAG-MPAL resulted in the

formation of the peptide tetramer (LYRAG)4C1 (Figure 2b). Liquid Chromatography-Mass

Spectrometry (LC-MS) analysis showed that the reaction went to completion as no

(LYRAG)1–3C1 were detected (not shown). Purification using HPLC yielded pure peptide

dendrimer (LYRAG)4C1 as confirmed by mass spectrometry (Figure 3a).

Figure 2: a) Peptide thioester Ac-LYRAG-MPAL.; b) Native chemical ligation of Ac-LYRAG-MPAL with

cysteine-dendrimer C1: i) 6 M Gu·HCl, 70 mM Tris·HCl (pH 7), thiophenol (2% v/v), benzyl mercaptan

(2% v/v), 37 °C, 2 h.

Similar results were obtained for other oligopeptides such as Ac-GRGDSGG-MPAL

(Figure 3b). The RGD sequence present in this peptide is known to bind to various

extracellular integrin receptors and can thus be used to attach dendrimers to cell surfaces.[28]

Ligation reaction for higher generations of cysteine dendrimers were investigated in

detail by H. Malda, as were ligations with non-water-soluble peptide sequences.[29] Ligations

of peptide thioesters with the second generation cysteine-modified dendrimer were also

successful, resulting in peptide octamers. Ligations with the third generation dendrimer

proved to be much more difficult, possibly due to steric crowding. In addition, the

characterization of these constructs was troublesome.

Chapter 2

24

a) b)

Re

lative

Ab

un

da

nce

200018001600140012001000800600400

m/z

[M+2H]2+

[M+3H]3+

[M+4H]4+

785.5

1047.3

1570.5

Re

lative

Ab

un

da

nce

200015001000500

m/z

[M+2H]2+

[M+4H]4+

[M+3H]3+

[M+5H]5+

[M+6H]6+

541.6

649.6

811.7

1081.8

1622.0

Figure 3: Mass spectra of first generation peptide dendrimers (a): (LYRAG)4C1 (calcd. mass 3139.8

Da, deconvoluted mass 3139.0 Da) and (b): (GRGDSGG)4C1 (calcd. mass 3243.5 Da, deconvoluted

mass 3242.4 Da).

2.4 Recombinant expression and purification of fluorescent proteins with a

C-terminal thioester

In addition to solid phase synthesis of thioester peptides, recombinant proteins

bearing C-terminal thioesters can be obtained via recombinant expression in bacteria. An

expression and purification strategy based on protein splicing can be used to acquire

proteins with this non-natural functionality. Protein splicing is a post-translational process in

which an internal protein fragment (termed intein) is spliced from a precursor protein in a

series of intramolecular rearrangements.[30,31] Inteins can be engineered to promote only the

first step of protein splicing. Proteins that are expressed as N-terminal fusions to such an

engineered intein domain can be cleaved by addition of thiols, releasing the recombinant

protein containing a C-terminal thioester.[26] At New England Biolabs, the IMPACTTM

expression and purification system has been developed, which combines intein-catalyzed

cleavages with an affinity-based purification tag to obtain thioester-terminated recombinant

proteins.

Green Fluorescent Protein (GFP) was chosen as a model protein to study the ligation

of folded proteins to cysteine dendrimers. GFP is a stable protein of 27 kDa that is highly

fluorescent upon illumination.[32,33] Various mutants of GFP have been developed, with

different fluorescent properties, which often have improved quantum yields.[34] Two examples

are enhanced Cyan Fluorescent Protein (eCFP) and enhanced Yellow Fluorescent Protein

(eYFP), which are used frequently as FRET partners in fluorescence microscopy applications

(FRET stands for fluorescence resonance energy transfer).[35] The expression and

purification procedure is described for GFP here, but is identical for eCFP and eYFP.[36]

The gene coding for GFP was cloned into the IMPACT vector pTXB1 to yield an E.

coli expression vector for a fusion protein of GFP with an intein domain and a CBD. Due to


25

the cloning strategy used here the GFP contained 8 additional residues at the C-terminus of

GFP (i.e. NEFLEGSS). After ligation of the target gene in the multiple cloning site of the

IMPACT vector, the plasmid was transfected into E. coli and the GFP fusion protein was

overexpressed. The fusion protein contains a chitin-binding domain (CBD) that can be

employed in protein purification via affinity chromatography (Figure 4). The CBD of the fusion

protein binds to chitin-beads in a column, while other proteins are washed away. Upon

addition of a thiol (e.g. mercaptoethane sulphonic acid, MESNA) the intein domain catalyzes

cleavage of the target protein, resulting in the release of the thioester-terminated target

protein. Overnight incubation of the GFP-intein-CBD protein bound to the chitin column with

MESNA resulted in cleavage of the peptide bond between GFP and the formation of GFP

thioester.

Figure 4: Schematic representation of the purification and cleavage of thioester-functionalized

proteins using the IMPACT system.

The protein was isolated in a yield of approximately 40 mg L–1. SDS-PAGE and ESI-

MS analysis showed the presence of a single protein with a mass of 27705 Da that is

consistent with GFP-MESNA (Figure 5).

Chapter 2

26

a) b)

Figure 5: Characterization GFP-thioester: (a) SDS-PAGE analysis (10%, reducing conditions); (b)

deconvoluted mass spectrum of GFP-MESNA (Calcd. mass 27704 Da), inset: m/z spectrum.

2.5 Synthesis of a peptide-protein hybrid dendrimer

To test whether GFP-MESNA was active in native chemical ligation reactions with

cysteine-modified dendrimers, the GFP thioester was first ligated to dendrimer C1 using a

large excess of the dendrimer. This large excess of dendrimer yields preferential formation of

dendrimers with only one protein attached. ESI-MS of the reaction mixture after incubation

for 20 h at room temperature indeed showed the clean conversion of GFP thioester to

(GFP)1C1, with no indication for the presence of unreacted GFP thioester or dendrimer with

more than one GFP (Figure 6a). An interesting feature of the resulting protein-dendrimer is

that three cysteine-residues are still available for reaction with thioester peptides or proteins,

giving access to hybrid structures containing different sequences (Scheme 3).

The large difference in molecular weight between (GFP)1C1 and C1 allowed the easy

removal of excess of C1 from the ligation mixture by repeated concentration/dilution steps

using a centrifugal filter with a 10 kDa molecular weight cut-off. Subsequently, (GFP)1C1 was

reacted with an excess of Ac-GRGDSGG-MPAL.


27

Scheme 3: Synthesis of a hybrid peptide-protein dendrimer: (GFP)1(GRGDSGG)3C1.

After 20 h incubation, characterization using ESI-MS showed the presence of a

single, high molecular weight peak at 30173 Da, which is exactly the mass expected for

(GFP)1(GRGDSGG)3C1 assuming formation of two intramolecular disulfide bonds (Figure

6b).

a) b)

Figure 6: Deconvoluted mass spectra of (a) protein dendrimer (GFP)1C1 (Calcd. mass 28291 Da),

inset: m/z spectrum and (b) peptide-protein dendrimer (GFP)1(GRGDSGG)3C1 (Calcd. mass

30177 Da), inset: m/z spectrum.

This synthetic methodology should be generally applicable to prepare dendrimers

with (exactly) one copy of any recombinant protein and multiple copies of any peptide or

other synthetic ligand with a thioester functionality.

Chapter 2

28

2.6 Synthesis of protein tetramers

In order to test whether native chemical ligation is efficient enough to obtain

multivalent protein dendrimers, 25 #M of dendrimer C1 was incubated with 4 or 8 equivalents

of GFP-MESNA and the ligation reaction was monitored using SDS-PAGE (Scheme 4).

Scheme 4: Native chemical ligation of GFP-MESNA with C1: i) 0.1 M phosphate buffer (pH 7),

thiophenol (2% v/v), benzyl mercaptan (2% v/v), 20 °C.

SDS-PAGE analysis of the reaction after 20 h showed the presence of 4 distinct

bands at 27 kDa, 55 kDa, ~ 120 kDa and ~ 170 kDa that we attribute to respectively GFP

thioester and (GFP)1C1, (GFP)2C1, (GFP)3C1, and (GFP)4C1 (Figure 7).

Figure 7: SDS-PAGE analysis of reaction mixtures after 20 h incubation: Lane 1, 3: molecular weight

markers (kDa); Lane 2, 4: different ratios dendrimer to protein (1:4 and 1:8).

The bands of (GFP)3C1 and (GFP)4C1 run at a higher apparent molecular weight than

expected, which is probably due to the branched structure of these multivalent proteins. A

similar effect of branching on the apparent molecular weight has been reported for

ubiquitinated proteins, which are a natural form of branched proteins.[37] From the intensities

of the different bands in SDS-PAGE it can be concluded that the reaction is far from

completion. Several attempts to increase tetramer yield by changing variables such as

temperature, pH, reaction time and addition of denaturants did not improve conversion.

The limited solubility of GFP under the conditions suitable for native chemical ligation

(100–200 #M GFP thioester) probably makes the ligation of GFP slower than the peptide


29

ligation reactions that are typically performed at 2–10 mM concentrations. A second factor

that may prevent the full conversion to (GFP)4C1 is increased steric crowding that occurs

when attaching four 27 kDa proteins around a 0.7 kDa dendritic core, as can been seen from

molecular modeling experiments with the GFP tetramer (Figure 8). Modeling the system with

a rigid peptide linker provides enough space for multiple proteins around the dendritic core.

However, when modeling the system with a random-coil linker (which is a more reasonable

assumption) steric crowding is evident.

a) b)

Figure 8: Molecular models of GFP tetramer. (a) Model assuming a rigid peptide linker between

dendritic core and protein; (b) Model assuming a random coil peptide linker between dendritic core

and protein.

To provide definitive evidence for the formation of the GFP tetramer, the ligation

product was further purified using size exclusion chromatography and analyzed using Electro

Spray Ionization Time-of-Flight Mass Spectrometry under native conditions (Figure 9).

Chapter 2

30

a)

b)

Rela

tive A

bundance

500045004000350030002500m/z

monomer (31–34.5 min)

dimer (29–30.5 min)

trimer (27–28 min)

tetramer (26–27 min) [M + 23H]23+

[M + 20H]20+

[M + 16H]16+

[M + 10H]10+

Figure 9: a) Size exclusion chromatography of the ligation reaction mixture (1:4) on an analytical

Superdex 200 column showing the total ion current (solid line) and the selected ion current for

(GFP)4C1 (dashed line); b) Mass spectra of the different ligation intermediates and final product

(GFP)4C1.

The m/z spectrum of the first species that elutes from the column shows a limited

number of peaks that can be deconvoluted to a single mass of 110971 Da (Figure 10). This

mass corresponds almost exactly to the average mass calculated for (GFP)4C1 with two

disulfide bonds (110973 Da). The other detected ions originated from GFP (hydrolyzed

thioester), (GFP)2C1, and (GFP)3C1 and clearly reveal that size-exclusion liquid

chromatography coupled to mass spectrometry under native conditions is an ideal analytical

tool to separate and analyze these protein mixtures. Moreover, the high m/z values and the

narrow charge distribution indicate that the proteins are still properly folded.[38,39] The absence


31

of protein thioester in the reaction mixture shows that thioester hydrolysis competes with

native chemical ligation in this synthesis.

Figure 10: Deconvoluted mass spectrum of first generation GFP dendrimer (GFP)4C1 (Calcd. mass

110977 Da), inset: m/z spectrum.

Although the mass spectra of the protein trimer and tetramer appear to be from single

species and only show some tailing due to residual salt adducts, the mass spectra of the

monomer and dimer peaks in the chromatogram show heterogeneity. Zooming in on the m/z

spectrum of the monomer peak, two distinct species can be observed, with deconvoluted

masses of 27561 and 27684 Da (Figure 11a).

a) b)

Rela

tive A

bundance

28002700260025002400

m/z

2757.1

2769.3

2517.7

2506.6

Rela

tive A

bundance

38003700360035003400

m/z

3691.8

3721.7

3491.8

3461.1

Figure 11: Enlargements of the monomer and dimer GFP mass spectra.

The mass difference compared to the calculated mass for the protein without

thioester (27579 Da) is –18 Da and +104 Da, respectively. The 18 Da difference corresponds

to the loss of water from the protein, however, it is unclear via which mechanism this would

occur. The 104 Da difference is believed to originate from GFP with an extra cysteine residue

at the C-terminus, which has a calculated mass of 27683 Da. This hypothesis is supported

by the mass spectrum of the dimer peak, in which, next to the expected (GFP)2C1 (calcd.

Chapter 2

32

mass 55853 Da, deconvoluted mass 55850 Da), a species with mass 55362 is present

(Figure 11b). This corresponds to the disulfide-bridged dimer of GFP-Cys, which has a

calculated mass of 55364 Da and would elute at the same retention time as (GFP)2C1.

To confirm that this GFP-Cys was not originating from a contamination of free

cysteine in the dendrimer starting material, the protein tetramer (GFP)4C1 was purified on a

small scale and injected again on the size exclusion column, yielding a single peak. After

incubation of the tetramer for 5 days at 20 °C, the sample was again injected on the column,

now showing three peaks, corresponding to tetramer, dimer and monomer GFP assemblies.

Upon reduction of this sample using 2% v/v 2-mercaptoethanol, the dimer peak disappeared

and only tetramer and monomer were present (Figure 12).

Inte

nsity

2624222018time / min

tetramer, t = 0 tetramer, t = 5 days tetramer, t = 5 days, reducing conditions

Figure 10: Size exclusion chromatograms of purified tetramer (solid line), tetramer after 5 days

incubation (dotted line) and tetramer after 5 days incubation, injected under reducing conditions

(dashed line).

These results show that the GFP-Cys is originating from degradation of the GFP

dendrimer after ligation of the protein to the cysteine on the dendrimer. This degradation has

never been observed for the peptide dendrimers, however. A possible explanation may be

that the tertiary amines of the dendrimer interior provide a local basic environment, resulting

in increased nucleophilicity of the cysteine sulfhydryl groups towards the amide bond

between the cysteine and the dendritic core.

Although the aforementioned degradation issue has not been resolved yet, it was

decided to further explore the possibilities of the synthetic strategy to obtain protein

multimers. The methodology used to obtain peptide-protein hybrid structures should also

allow the synthesis of mixed protein assemblies. Attractive targets for understanding multi-

protein systems are FRET partner proteins, capable of energy transfer (e.g. eCFP and

eYFP). Both eCFP and eYFP were obtained with a MESNA thioester in a similar fashion as

GFP. The strategy to obtain mixed protein assemblies is outlined in Scheme 5.


33

Scheme 5: a) Synthesis of a mixed protein dendrimer: (eCFP)1(eYFP)3C1: first, eCFP-MESNA is

ligated to an excess of C1 dendrimer, after which unreacted dendrimer is removed via dialysis. Next, 3

equivalents of eYFP-MESNA are ligated to the (eCFP)1C1 dendrimer.

Upon native chemical ligation of eCFP-MESNA with an excess of C1, (eCFP)1C1 was

formed. Removal of unreacted dendrimer and subsequent reaction with 3 equivalents of

eYFP-MPAL yields a hybrid protein dendrimer (eCFP)1(eYFP)3C1. Similar to the preparation

of (GFP)4C1, the reaction went not to completion. Analytical separation using size exclusion

chromatography and characterization using SEC-MS showed formation of (eCFP)1(eYFP)3C1

(Figure 13a). In a similar way, starting from reaction of eYFP-MESNA with an excess of C1,

(eYFP)1(eCFP)3C1 could be formed (Figure 13b). The synthesis of these mixed protein

assemblies shows the versatility of the strategy. However, preparative purification of the

protein tetramers was not pursued, and therefore fluorescence energy transfer

measurements have not been attempted yet.

Chapter 2

34

a) b)

Figure 13: a) Deconvoluted mass spectra of a) (eCFP)1(eYFP)3C1 (Calcd. mass 111456 Da; inset:

m/z spectrum); b) (eYFP)1(eCFP)3C1 (Calcd. mass 111284 Da; inset: m/z spectrum).

2.7 Discussion and conclusions

A general method to functionalize dendrimers with both oligopeptides and

recombinantly expressed proteins was developed, employing the native chemical ligation

reaction. The approach presented allows chemoselective dendrimer conjugation to the C-

terminus of a protein, which is much less likely to interfere with protein function than other

less specific protein ligation strategies. The modular approach presented here provides

access to a wide variety of well-defined multivalent peptides and proteins that are attractive

both for understanding the fundamental mechanisms of multivalency in biological interactions

and for biomedical applications in targeted drug delivery, molecular imaging and

immunology. Peptide dendrimers were synthesized efficiently and purification using

preparative HPLC was straightforward. Dendrimers were derivatized with exactly one copy of

a recombinant protein, and the remaining cysteines could then be further functionalized with

oligopeptides. The synthesis of branched protein multimers was illustrated on analytical

scale. Both tetramers containing four identical proteins and dendrimers displaying a mix of

two different sequences were obtained. A difficulty in the synthetic strategy was shown to be

the incomplete reaction of large proteins with the cysteine-modified dendrimers. Complicating

factors are steric hindrance, resulting in competing reactions such as hydrolysis of the

protein thioester, and degradation of protein dendrimers following ligation. This degradation

issue was not investigated further. Although the presented work shows the feasibility of

synthesizing these branched protein dendrimers using native chemical ligation, optimization

of the scaffold and reaction conditions are desired. Other types of dendritic scaffolds may

prevent degradation to occur, and incorporation of extra spacing between branching points

and reactive cysteine moieties may reduce steric crowding, thereby increasing ligation

reaction rates. The use of water-soluble thiol catalysts can also improve ligation yields.

Recently, 4-mercaptophenylacetic acid was reported as an alternative to thiophenol


35

and benzyl mercaptan in native chemical ligation.[40] In addition, a possibility to circumvent

the slow reaction kinetics of large proteins to the dendrimer is to use a non-covalent

approach, e.g. first covalently attach a small peptide or protein domain to the scaffold using

NCL, followed by association with a tight binding partner. An exploration of this method is

discussed in Chapter 5.

2.8 Experimental section

General

Unless stated otherwise, all solvents (p.a. quality) and other chemicals were obtained from

commercial sources and used as received. Water was demineralized prior to use. Dichloromethane

was obtained by distillation from P2O5. Amine-terminated poly(propylene imine) dendrimers were

obtained from SyMO-Chem BV (Eindhoven, The Netherlands) and dried prior to use. Trityl-protected

cysteine (Trt-Cys(Trt)-OH) was obtained from Bachem (Bubendorf, Switserland). S-trityl 3-

mercaptopropionic acid (Sigma), triisopropylsilane (Aldrich), p-cresol (Aldrich) were all used as

supplied. All t-Boc-protected amino acids and MBHA resins (0.92 mmol g–1) were obtained from

NovaBiochem. Standard 1H NMR and 13C NMR experiments were performed on a Varian Gemini-2000

300 MHz spectrometer, a Varian Mercury Vx 400 MHz spectrometer, and a Varian Unity Inova 500

MHz spectrometer at 298 K. Chemical shifts are reported in parts per million relative to

tetramethylsilane (TMS). Reversed phase high pressure liquid chromatography (RP-HPLC) was

performed on a Varian Pro Star HPLC system coupled to a UV-Vis detector probing at 214 nm using a

Vydac protein & peptide C18 column. A gradient of acetonitrile in water (both containing 0.1% TFA)

was used to elute products. MALDI-TOF spectra were obtained on a Perspective Biosystems Voyager

DE-Pro mass spectrometer using !-cyano-4-hydroxycinnamic acid as matrix. ESI-MS spectra were

measured on either an Applied Biosystems ESI Mass Spectrometer API-150EX, a Micromass Q-TOF

Ultima Global Mass Spectrometer, or a Thermo Finnigan LCQ Deca XP MAX, all in positive mode.

Native mass spectrometry was performed in positive mode using a Micromass LCT. The protein was

injected onto a Superdex 200 H/R size-exclusion column (3.2 mm $ 300 mm; Amersham Biosciences)

using a mobile phase of 50 mM ammonium acetate, pH 6.7 at a flow rate of 50 #L min–1. The post-

column eluent was guided into the electrospray source using a fused silica emitter. The electrospray

source was optimized for transmission of (GFP)xC1 complexes. Mass determinations were performed

under conditions of increased pressure in the source and intermediate pressure regions in the mass

spectrometer.[41] Micromass MaxEnt 1 software was used for deconvolution of m/z spectra. All

reported calculated masses are of fully reduced species.

N-Trt-S-Trt-L-cysteine N-hydroxysuccinimidyl ester

Trt-Cys(Trt)-OH (2.5 g, 4.1 mmol) was dissolved in 20 mL acetonitrile while pyridine (0.37 mL,

4.5 mmol) and disuccinimidyl carbonate (2.1 g, 8.2 mmol) were added. The mixture was stirred

overnight and acetonitrile was evaporated using reduced pressure. The crude product was dissolved

in ethylacetate and filtered to remove the precipitate (pyridine salt). Subsequently, the clear solution

was washed with a sodium bicarbonate solution. The organic layer was dried over magnesium sulfate

and, after filtration; the solvent was evaporated using reduced pressure. Column chromatography

(silica, gradient EtOAc:heptane 1:3 – 1:1) afforded a white fluffy solid in 77% yield as the product. 1H NMR (400MHz, CDCl3): % = 7.12–7.52 (m, 30H, Trt), 3.72 (dt, 1H, CH), 2.67 (s, 4H, CH2, NHS), 2.6

(d, 1H, NH), 2.5 (dd, 1H, CH2), 2.4 (dd, 1H, CH2).

Chapter 2

36

13C NMR (CDCl3, 100 MHz): % =168.1, (2 $ Cq, NHS), 167.7 (C=O), 144.8 (3 $ Cq, NHTrt), 144 (3xCq,

NHTrt), 129.4 (3 $ Cq, STrt), 129.3 (3 $ Cq, STrt), 128 (3 $ CH, STrt), 127.7 (3 $ CH, NHTrt), 127.4 (3

$ CH, NHTrt), 126.9 (3 $ CH, STrt), 126.2 (3 $ CH, STrt), 126.1 (3 $ CH, NHTrt), 70.7 (Cq, NHTrt),

66.4 (Cq, STrt), 53.4 (CH), 36.6 (CH2), 25 (CH2, NHS).

MALDI-TOF: m/z [C45H38N2O4S + Na]+ Calcd. 725.2 Da, Obsd. 725.1 Da; [2M + 2Na]2+ Calcd. 1427.5

Da, Obsd. 1426.9 Da.

General synthesis of cysteine dendrimers C1, C2 and C3

PPI dendrimer (G1, G2, or G3) was dissolved in 5 mL dichloromethane and 5 (G1), 10 (G2) or 15 (G3)

equivalents of triethylamine were added. 4, 8, or 16 equivalents, respectively, of N-Trt-S-Trt-L-cysteine

N-hydroxysuccinimidyl ester were added and the reaction mixture was stirred for 2 h at 20 °C.

Subsequently the reaction mixture was washed twice with a potassium carbonate solution (pH 10) and

once with a saturated solution of potassium bisulfate solution (pH 0.2). The organic layer was dried

over magnesium sulfate and after filtration the solvent was evaporated using reduced pressure. The

obtained compound was kept at 0 °C using ice while a mixture of TFA with 2.5% of Et3SiH and 2.5% of

demiwater was added to remove the trityl protective groups. After stirring for 1 h, demiwater was

added and the solution was washed three times with diethyl ether. The water layer was lyophilized to

obtain the product. The crude product was purified with RP-HPLC using a gradient of acetonitrile in

water containing 0.1% TFA.

Dendrimer C1 1H NMR (300MHz, CD3OD): % = 4.01 (t, 4H, CH), 3.48 (q, 8H, CH2NH), 3.26–3.17 (m, 12H, CH2N),

3.09–2.94 (m, 8H, CH2S), 1.97 (m, 8H, NCH2CH2CH2N), 1.81 (m, 4H, NCH2CH2CH2CH2N)

LC-MS: m/z [C28H60N10O4S4 + H]+ Calcd. 729.4 Da, Obsd. 729.4 Da.

Dendrimer C2

LC-MS: m/z [C64H136N22O8S8 + 3H]3+ Calcd. 533.3 Da, Obsd. 533.4 Da; [M + 2H]2+ Calcd. 799.4 Da,

Obsd. 799.4 Da; [M + H]1+ Calcd. 1597.9 Da, Obsd. 1598.7 Da;

Dendrimer C3

LC-MS: m/z [C136H288N46O16S16 + 6H]6+ Calcd. 556.7 Da, Obsd. 557 Da; [M + 5H]5+ Calcd. 667.8 Da,

Obsd. 668 Da; [M + 4H]4+ Calcd. 834.5 Da, Obsd. 835 Da; [M + 3H]3+ Calcd. 1112.3 Da, Obsd. 1112

Da; [M + 2H]2+ Calcd. 1667.9 Da, Obsd. 1668 Da.

General synthesis of MPAL-activated peptides

All peptides were prepared by manual solid phase peptide synthesis on a 0.25 mmol scale using the in

situ neutralization/HBTU activation procedure for t-Boc chemistry[42]. Each synthetic cycle consisted of

3–4 minutes activation of the t-Boc-amino acid (1.1 mmol) by HBTU (1.0 mmol; 0.5 M in DMF) in the

presence of N,N-diisopropylethylamine (DIPEA) (3 mmol). All activated amino acids were coupled for

10 min, except for Arg, Ser and Asn, which required a coupling time of 20 min. Unbound amino acids

were removed by a DMF flow wash (2 $ 20 s of rinsing). The Boc group was removed by TFA (2 $ 1

min) followed by a second DMF flow wash. Additional flow washes with dichloromethane were

performed directly before and after the TFA deprotection step of Asn residues. Peptide were

synthesized on a TAMPAL resin to yield C-terminal MPAL thioesters.[20] The TAMPAL resin was

prepared as follows. Boc-Leu was coupled to the 4-methylbenzhydrylamine (MBHA) resin, following

the same synthetic cycle as mentioned above. Next, 1.1 mmol of S-tritylmercaptopropionic acid was

activated with 1.0 mmol of HBTU in the presence of 3 mmol DIPEA and coupled for 30 min to Leu-

MBHA resin. The protecting trityl group was removed by the addition of TFA containing 2.5%

triisopropylsilane and 2.5% H2O. The thioester bond was formed after coupling of the next coupling of

a C-teminal amino acid to the resin. After formation of the MPAL thioester on the resin, the remaining


37

amino acids were coupled in the typical manner. After synthesis, the peptide was flow-washed with

dichloromethane (DCM) and 50% MeOH in DCM to remove all DMF and dried in vacuo. HF treatment

over 1 h at 0 °C with 4% p-cresol was used to cleave the peptide from the resin and to remove the

protecting groups from the side chains (except for the Dnp groups on the histidine side-chains). After

cleavage, the peptide was precipitated with ice-cold diethyl ether, dissolved in acetonitrile and

lyophilized.

Ac-LYRAG-MPAL

After the synthesis of MPAL-thioester on the resin, respectively, Boc-Gly, Boc-Ala, Boc-Arg(Tos), Boc-

Tyr(2-Br-Z) and Boc-Leu(·H2O) were coupled using standard coupling procedure. The N-terminus of

the peptide was acetylated using 0.5 M acetic anhydride and 0.5 M pyridine in DMF (2 $ 2 min). The

peptide was cleaved from the resin using liquid HF and was purified using RP-HPLC with a gradient of

20–40% acetonitrile in water over 90 min. The pure peptide was obtained in 82% yield (95 mg).

LC-MS: m/z [C37H60N10O9S + H]+ Calcd. 821.4 Da, Obsd. 821.4 Da.

Ac-GRGDSGG-MPAL

After the synthesis of MPAL-thioester on the resin, respectively, Boc-Gly and Boc-Gly, Boc-Ser(Bzl),

Boc-Asp(OcHx), Boc-Gly, Boc-Arg(Tos) and Boc-Gly were coupled using standard coupling

procedure. The N-terminus of the peptide was acetylated using 0.5 M acetic anhydride and 0.5 M

pyridine in DMF (2 $ 2 min). The peptide was cleaved from the resin using liquid HF and was purified

using RP-HPLC with a gradient of 5–30% acetonitrile in water over 90 min. The pure peptide was

obtained in 80% yield (89 mg).

LC-MS: m/z [C32H54N12O13S + H]+ Calcd. 847.4 Da, Obsd. 847.5 Da.

(LYRAG)4C1

Ac-LYRAG-MPAL (5.7 mg, 6.9 #mol) was dissolved in 0.5 mL Tris buffer (70 mM, pH 8.0) containing

6 M guanidine and cysteine-functionalized dendrimer C1 (1 mg, 1.3 #mol) was added. Thiophenol (2%

v/v) and benzyl mercaptan (2% v/v) were added and the pH was checked (7). After 2 h at 37 °C the

reaction was complete. The peptide dendrimer was analyzed by RP-HPLC combined with ESI-MS.


Obsd. 1047.3 Da; [M + 2H]2+ Calcd. 1569.8 Da, Obsd. 1570.5 Da.

(GRGDSGG)4C1

Ac-GRGDSGG-MPAL (53 mg, 62 #mol) was dissolved in 2.5 mL Tris buffer (70 mM, pH 8.0)

containing 6 M guanidine and cysteine-functionalized dendrimer C1 (11 mg, 15 #mol) was added.

Thiophenol (2% v/v) and benzyl mercaptan (2% v/v) were added and the pH was checked (6) and

adjusted to 7.5 using 10 #L NaOH (5 M). After 2 h at 37 °C the reaction was complete. The peptide

dendrimer was purified by RP-HPLC with a gradient of 5–25% acetonitrile in water over 90 min. The

pure product was obtained in 40% yield (19.5 mg).


Obsd. 649.6 Da; [M + 4H]4+ Calcd. 811.4 Da, Obsd. 811.7 Da; [M + 3H]3+ Calcd. 1081.5 Da, Obsd.

1081.8 Da; [M + 2H]2+ Calcd. 1621.7 Da, Obsd. 1622.0 Da.

Recombinant expression and purification of GFP thioester

Cloning of expression plasmid for GFP-intein-CBD fusion protein

The expression vector pBAD-GFPuv (Clontech) was used as the source of the GFP gene in this

study. GFPuv is a GFP variant that is optimized for high level expression in E. coli and high

Chapter 2

38

fluorescent intensity when illuminated by UV light. It contains 3 amino acid mutations relative to wt

GFP: F99S, M153T, and V163A. Site-directed mutagenesis (QuickChange site-directed mutagenesis

kit, Stratagene) was used to delete the NdeI restriction site within the GFP gene using the primers 5&-

CCCGTTATCCGGATCATGAAACGGCATGAC-3& and 5&-GTCATGCCGTTTCATGTGATCCGGATA

ACGGG-3&. The mutant plasmid was digested with Nde I and EcoR I, and the GFP fragment was

ligated into the Nde I and EcoR I sites of pET28a yielding pET28GFP. Site-directed mutagenesis using

primers 5&CGGAGCTCGAATTCATTTTTGTAGAGCTCATCC-'3 and 5&-CGATGAGCTCTACAAAAATG

AATTCGAGCTCCG-3& was used to delete the first nucleotide of the TAA stop codon. The mutated

plasmid was again digested with Nde I and EcoR 1 and the GFP fragment was inserted into the MCS

of pTXB1 (New England Biolabs) yielding pGFPX1. This cloning strategy inserts a stretch of 8 amino

acids (NEFLEGSS) between the original C-terminus of GFP and the intein cleavage site. DNA

sequencing using T7 promoter and intein specific reversed primers (New England Biolabs) confirmed

the correct in-frame fusion of GFP with the intein sequence.

Protein expression and purification

pGFPX1 was transformed into chemically competent E. coli BL21(DE3) (Novagen) and plated on LB

agar plates containing 100 mg L–1 ampicillin. Single colonies were used to inoculate 2 mL LB medium

containing 100 mg L–1 ampicillin. Cultures were incubated overnight at 37 °C and subsequently used to

start 200 mL culture containing 100 mg L–1 ampicillin. At OD600 = 0.5 the temperature was lowered to

15 °C and 0.3 mM IPTG was added to induce expression of the target protein. Cells were collected

after overnight expression at 15 °C and 250 rpm by centrifugation, resuspended in BugBuster

(Novagen) lysis buffer and incubated for 20 min at 20 °C. A clarified cell extract was obtained by

centrifugation at 40000 $ g for 45 min. The supernatant was loaded onto a 10 mL chitin column (New

England Biolabs) that was equilibrated with 20 mM sodium phosphate, 0.5 M NaCl, 0.1 mM EDTA, pH

8.0 (column buffer). The column was washed with 10 volumes (100 mL) of column buffer to remove

non- and weak binding proteins. Subsequently, 3 volumes (30 mL) of cleavage buffer (200 mM sodium

phosphate, 0.5 M NaCl, 0.1 mM EDTA, 50 mM MESNA, pH 6.0) were flushed quickly through the

column. After overnight incubation of the column at 20 °C, the MESNA thioester of GFP was eluted

from the column using 1 volume of cleavage buffer. SDS-PAGE analysis of the eluted protein showed

a single band at ~ 27 kDa. This procedure typically yields 40 mg of pure GFP-MESNA from 1 L of E.

coli culture.

ESI-MS: deconvoluted mass [C1234H1891N329O382S8] Calcd. 27704 Da, Obsd. 27705 Da.

Recombinant expression and purification of eCFP-MESNA and eYFP-MESNA

The expression and purification of eCFP-MESNA and eYFP-MESNA are described in P.Y.W. Dankers!

thesis.[36]

(GFP)1C1

GFP-MESNA was diluted to a final concentration of 75 #M in a buffer containing 0.1 M sodium

phosphate, 2% (v/v) thiophenol and 2% (v/v) benzyl mercaptan (pH 7). C1 dendrimer was added (2

mM) and the reaction mixture was incubated at 20 °C for 20 h. Remaining unreacted dendrimer was

removed by repeated concentration and dilution of the reaction mixture using an Amicon Ultra-4

centrifugal concentrator with a 10-kDa cut-off.



39

(GFP)1(GRGDSGG)3C1

Ac-GRGDSGG-MPAL was added to the (GFP)1C1 solution (approximately 75 #M) to a final

concentration of 375 #M. 2% (v/v) thiophenol and 2% (v/v) benzyl mercaptan were added to catalyze

native chemical ligation and the reaction mixture was incubated at 20 °C for 20 h. Remaining

unreacted peptide thioester was removed by repeated concentration and dilution of the reaction

mixture using an Amicon Ultra-4 centrifugal concentrator with a 10-kDa cut-off.


(GFP)4C1

GFP-MESNA was diluted to a final concentration of either 200 #M or 100 #M in a buffer containing

0.1 M sodium phosphate, 2% (v/v) thiophenol and 2% (v/v) benzyl mercaptan (pH 7). C1 dendrimer

was added (25 #M) and the reaction mixture was incubated at 20 °C for 20 h. Precipitates were

removed after the reaction by centrifugation. SDS-PAGE analysis showed bands corresponding to one

through four proteins ligated to the cysteine-functionalized dendrimer.

SEC-ESI-MS (tetramer peak): deconvoluted mass [C4956H7600N1326O1520S28] Calcd. 110977 Da, Obsd.

110971 Da.

(eCFP)1C1

eCFP-MESNA was diluted to a final concentration of 40 #M in a buffer containing 0.1 M sodium






(eCFP)1(eYFP)3C1

eYFP-MESNA was added to the (eCFP)1C1 solution (approximately 40 #M) to a final concentration of

120 #M. 2% (v/v) thiophenol and 2% (v/v) benzyl mercaptan were added to catalyze native chemical

ligation and the reaction mixture was incubated at 20 °C for 20 h.


111454 Da.

(eYFP)1C1

eYFP-MESNA was diluted to a final concentration of 40 #M in a buffer containing 0.1 M sodium






(eYFP)1(eCFP)3C1

eCFP-MESNA was added to the (eYFP)1C1 solution (approximately 40 #M) to a final concentration of

120 #M. 2% (v/v) thiophenol and 2% (v/v) benzyl mercaptan were added to catalyze native chemical

ligation and the reaction mixture was incubated at 20 °C for 20 h.


111280 Da.

Chapter 2

40

2.9 References

[1] Mammen, M.; Choi, S.-K.; Whitesides, G. M. Angew. Chem., Int. Ed. Engl. 1998, 37, (20),

2755-2794.

[2] Kiessling, L. L.; Strong, L. E.; Gestwicki, J. E., Annual reports in medicinal chemistry, 2000,

35, 321-330.

[3] Kiessling, L. L.; Gestwicki, J. E.; Strong, L. E. Curr. Opin. Chem. Biol. 2000, 4, (6), 696-703.

[4] Fan, E.; Zhang, Z.; Minke, W. E.; Hou, Z.; Verlinde, C. L. M. J.; Hol, W. G. J. J. Am. Chem. Soc. 2000, 122, (11), 2663-2664.

[5] Kitov, P. I.; Sadowska, J. M.; Mulvey, G.; Armstrong, G. D.; Ling, H.; Pannu, N. S.; Read, R.

J.; Bundle, D. R. Nature 2000, 403, (6770), 669-72.

[6] Rao, J.; Lahiri, J.; Isaacs, L.; Weis, R. M.; Whitesides, G. M. Science 1998, 280, (5364), 708-

11.



[9] Mitchell, J. P.; Roberts, K. D.; Langley, J.; Koentgen, F.; Lambert, J. N. Bioorg. Med. Chem. Lett. 1999, 9, (19), 2785-8.

[10] Reuter, J. D.; Myc, A.; Hayes, M. M.; Gan, Z.; Roy, R.; Qin, D.; Yin, R.; Piehler, L. T.; Esfand,

R.; Tomalia, D. A.; Baker, J. R., Jr. Bioconjugate Chem. 1999, 10, (2), 271-8.

[11] Vrasidas, I.; De Mol, N. J.; Liskamp, R. M. J.; Pieters, R. J. Eur. J. Org. Chem. 2001, (24),

4685-4692.

[12] Turnbull, W. B.; Stoddart, J. F. Rev. Mol. Biotechnol. 2002, 90, (3-4), 231-255.

[13] Rao, C.; Tam, J. P. J. Am. Chem. Soc. 1994, 116, (15), 6975-6.

[14] Sadler, K.; Tam, J. P. Rev. Mol. Biotechnol. 2002, 90, (3-4), 195-229.


1669.

[16] Sakamoto, M.; Ueno, A.; Mihara, H. Chem. Commun. 2000, (18), 1741-1742.


[18] Kluger, R.; Zhang, J. J. Am. Chem. Soc. 2003, 125, (20), 6070-6071.

[19] Dawson, P. E.; Muir, T. W.; Clark-Lewis, I.; Kent, S. B. Science 1994, 266, (5186), 776-9.

[20] Hackeng, T. M.; Griffin, J. H.; Dawson, P. E. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, (18),

10068-73.

[21] Tolbert, T. J.; Wong, C.-H. J. Am. Chem. Soc. 2000, 122, (23), 5421-5428.

[22] Lue, R. Y. P.; Chen, G. Y. J.; Hu, Y.; Zhu, Q.; Yao, S. Q. J. Am. Chem. Soc. 2004, 126, (4),

1055-1062.

[23] Yeo, D. S. Y.; Srinivasan, R.; Chen, G. Y. J.; Yao, S. Q. Chem. -Eur. J. 2004, 10, (19), 4664-

4672.

[24] Camarero, J. A.; Kwon, Y.; Coleman, M. A. J. Am. Chem. Soc. 2004, 126, (45), 14730-14731.

[25] Reulen, S. W. A.; Brusselaars, W. W. T.; Langereis, S.; Mulder, W. J. M.; Breurken, M.;

Merkx, M. Bioconjugate Chem. 2007, 18, (2), 590-596.


[27] David, R.; Richter Michael, P. O.; Beck-Sickinger Annette, G. Eur. J. Biochem. 2004, 271, (4),

663-77.

[28] Ruoslahti, E.; Pierschbacher, M. D. Science 1987, 238, (4826), 491-7.

[29] Malda, H., Thesis, Eindhoven University of Technology, 2006.

[30] Hirata, R.; Ohsumi, Y.; Nakano, A.; Kawasaki, H.; Suzuki, K.; Anraku, Y. J. Biol. Chem. 1990,

265, (12), 6726-33.

[31] Kane, P. M.; Yamashiro, C. T.; Wolczyk, D. F.; Neff, N.; Goebl, M.; Stevens, T. H. Science 1990, 250, (4981), 651-7.

[32] Zimmer, M. Chem. Rev. 2002, 102, (3), 759-782.

[33] Chalfie, M.; Tu, Y.; Euskirchen, G.; Ward, W. W.; Prasher, D. C. Science 1994, 263, (5148),

802-805.

[34] Heim, R.; Tsien, R. Y. Current Biol. 1996, 6, (2), 178-182.

[35] Truong, K.; Ikura, M. Curr. Opin. Struct. Biol. 2001, 11, (5), 573-578.

[36] Dankers, P. Y. W., Thesis, Eindhoven University of Technology, 2006.


41

[37] Borodovsky, A.; Ovaa, H.; Kolli, N.; Gan-Erdene, T.; Wilkinson, K. D.; Ploegh, H. L.; Kessler,

B. M. Chem. Biol. 2002, 9, (10), 1149-1159.

[38] Heck, A. J. R.; van den Heuvel, R. H. H. Mass Spectrom. Rev. 2004, 23, (5), 368-389.

[39] van den Heuvel, R. H. H.; Heck, A. J. R. Curr. Opin. Chem. Biol. 2004, 8, (5), 519-526.

[40] Johnson, E. C. B.; Kent, S. B. H. J. Am. Chem. Soc. 2006, 128, (20), 6640-6646.

[41] Tahallah, N.; Pinkse, M.; Maier, C. S.; Heck, A. J. R. Rapid Commun. Mass Spectrom. 2001,

15, (8), 596-601.

[42] Schnolzer, M.; Alewood, P.; Jones, A.; Alewood, D.; Kent, S. B. Int. J. Pept. Protein Res. 1992, 40, (3-4), 180-93.

Chapter 2

42

43

3

NATIVE CHEMICAL LIGATION ON A BIOSENSOR CHIP*

Abstract:

The use of surface plasmon resonance (SPR) biosensors enables real-time detection

and monitoring of biomolecular binding events. The efficient and selective immobilization of

proteins and peptides on an SPR sensor chip can be difficult using conventional

immobilization strategies. In this chapter, a chemoselective immobilization strategy for

Biacore SPR sensor chips is presented, based on native chemical ligation. First, a thioproline

was introduced on the surface, which could be deprotected using mild conditions to an N-

terminal cysteine residue. A streptavidin-binding peptide was immobilized via its C-terminus

onto the biosensor chip, and subsequent binding experiments with streptavidin showed

specific and reproducible binding to the peptide surface. Short ligation steps of peptide

thioester were alternated with streptavidin binding experiments on a single chip. This

provided an increased peptide loading after each ligation step, yielding enhanced protein-

binding capacity. As an example of a recombinant protein, green fluorescent protein (GFP)

was immobilized on a biosensor surface. Again, binding experiments with an antibody

directed against GFP showed the specificity and robustness of the coupling strategy. The

immobilization of S-peptide via native chemical ligation was used to illustrate the possibility of

obtaining kinetic information from the specific interaction between S-peptide and S-protein.

This approach allows for efficient immobilization of both recombinant proteins and synthetic

peptides with high control over the degree of functionalization of the surface.

* Part of this work has been accepted for publication: Helms, B.A.; van Baal, I.; Merkx, M.; Meijer, E.W. ChemBioChem, 2007.

Chapter 3

44

3.1 Introduction

Interactions between proteins and other biomolecules (e.g. peptides, carbohydrates,

nucleic acids, proteins) are of importance for a better understanding of many biological

functions. Therefore, studying these interactions has been a key subject of biochemical

research in the past decades. A wide variety of techniques have been developed, ranging

from co-immunoprecipitation and the yeast-two-hybrid screen to isothermal titration

calorimetry and solid phase binding assays (e.g. Enzyme-Linked ImmunoSorbent Assay,

ELISA).[1-4] Surface plasmon resonance (SPR) is an advanced and relatively new technique

to study biomolecular interactions quantitatively in real-time.

The SPR biosensor is based on surface plasmon resonance, a physical phenomenon

that occurs when light is reflected off thin metal films. A fraction of the light energy incident at

a sharply defined angle can interact with the delocalized electrons in the metal film

(plasmon), thereby reducing the reflected light intensity. The angle of incidence at which this

occurs depends on the refractive index close to the backside of the metal film, to which target

molecules can be immobilized. The binding of a ligand results in a change of the refractive

index on the sensor surface, which is measured as a change in resonance angle or

resonance wavelength, yielding a sensorgram.[5]

Figure 1: Schematic drawing of a surface plasmon resonance experimental setup, as devised by

Biacore AB. On top, a ligand is flowed over a gold surface covered with analyte molecules. Below, the

incoming polarized light beam is reflected and the change in angle of incidence of the reflected light is

a measure of the amount of binding on the chip surface.

SPR has become a routine technique in biomolecular research as a result of

successful commercialization by Biacore AB (now a part of General Electric Healthcare). In a

Biacore instrument, one reactant is immobilized onto a biosensor chip surface and the other

reactant is flowed across the surface, while the binding of the two reactants is followed in real

time by SPR technology (Figure 1). SPR enables not only the measurement of

thermodynamic parameters of an interaction, such as Kd and !H values, but can also provide

kinetic information (i.e. association and dissociation rates).


45

The biosensor surfaces used in Biacore apparatus are constructed on a microfluidic

chip and consist of a gold substrate (for SPR) coated with a non-fouling dextran layer. The

most widely used chip type bears carboxymethyl groups on the dextran matrix. Activation of

the carboxymethyl groups as N-hydroxysuccinimide-esters enables immobilization of

proteins via the primary amines of endogenous lysine residues[6] (Figure 2). However, this

immobilization strategy has some disadvantages.[7] First, because most proteins contain

multiple lysine residues, reaction with activated esters results in a heterogeneous surface.

Also, lysines are often present in or near the active site of a protein, which can result in

reduced binding upon conjugation. Finally, since pre-concentration is required on the surface

for efficient coupling reaction, the immobilization of small peptides can be difficult due to poor

pre-concentration prior to coupling.

Figure 2: General immobilization strategy for proteins on a biosensor surface via EDC/NHS activation.

Next to the standard amine-coupling strategy, ligands can be immobilized via intrinsic

sulfhydryl groups or via disulfides introduced on the protein prior to immobilization. For this,

the carboxymethylated surface is first modified with either an activated disulfide (for reaction

with sulfhydryl groups) or an aliphatic sulfhydryl group (for reaction with an activated

disulfide). For a more specific immobilization strategy, two dedicated chip types are

available, having either a streptavidin-modified surface, which immobilizes biotinylated

compounds, or an NTA-functionalized surface, capable of capturing His-tagged proteins via

metal chelation. All these strategies have certain disadvantages, however. The sulfhydryl

immobilization is not compatible with proteins bearing reduced cysteines in the native

structure, whereas both the streptavidin and NTA strategies are non-covalent coupling

methods, which can compromise the robustness of the modified surface. It would therefore

be useful to have access to a site-specific, covalent immobilization strategy for Biacore

biosensor chips valid for both synthetic peptides and recombinant proteins. Various chemo-

selective strategies have become available for immobilization of biomolecules on surfaces,

for example the Staudinger ligation, native chemical ligation or the Huisgen azide-alkyne

cycloaddition.[8-18] Of these, native chemical ligation displays the fastest kinetics, which is of

Chapter 3

46

importance in a continuous-flow system with low quantities of biomolecules available. In

addition, the synthetic and recombinant accessibility of thioester-modified peptides and

proteins required for this reaction, combined with the mild reaction conditions makes native

chemical ligation the ideal candidate for site-specific immobilization of peptides and proteins

onto biosensor surfaces. In this Chapter, the development of an immobilization strategy

based on native chemical ligation is described.

3.2 Native chemical ligation on a biosensor surface: concept and synthesis

In collaboration with Brett Helms of our laboratory, a site-specific immobilization

strategy for Biacore sensor chips was developed, based on native chemical ligation. In this

chemoselective reaction, an N-terminal cysteine and a C-terminal thioester form a native

peptide bond.[19,20] This approach allows for immobilization of both recombinant proteins and

synthetic peptides with high control over the degree of functionalization of the surface. The

experimental design requires a biosensor surface with N-terminal cysteines for direct

conjugation to proteins and peptides bearing C-terminal thioesters via NCL (Figure 3).

Figure 3: Preparation of cysteine-modified biosensor surface for protein and peptide immobilization

via native chemical ligation.


47

The carboxymethylated dextran brush of the sensor chip is readily modified with

amines using standard carbodiimide coupling in an aqueous format. Beyond that, any

chemistry required for the preparation of the cysteine-modified chips or subsequent ligation

reactions needs to be performed under mild conditions with aqueous soluble reagents over a

relatively narrow temperature range (10–40 °C). Harsh conditions (e.g. TFA) can be avoided

in the preparation of cysteine-modified biosensor surfaces by immobilizing onto the dextran

layer a thiazolidine derivative, which carries a nascent N-terminal cysteine. This thiazolidine

ring can be deprotected using methoxyamine at pH 4, yielding the cysteine-modified surface

ready for native chemical ligation reaction with both synthetic peptide thioesters and

recombinant protein thioesters.[21]

The thiazolidine derivative can be reacted on the surface via reaction of an amine

with an NHS-activated ester. A short synthesis of thiazolidine-derivative 4 from the readily

available NHS-ester of N-Boc-thiazolidine-4-carboxylic acid and mono-Boc-protected

ethylenediamine was devised (Scheme 1). The amide adduct 3 could be deprotected after

treatment with TFA, liberating the thiazolidine derivative 4 with a primary amine functionality.

The primary amine will react with the activated ester preferentially over the secondary amine

present in the thiazolidine ring.

Scheme 1: Synthesis of a nascent-cysteine for biosensor surface modification. Reaction conditions: i)

NHS, HBTU, DIPEA, DMF (67%); ii) tert-butyl phenylcarbonate, EtOH (65%); iii) CH2Cl2, DIPEA (95%);

iv) 1:1 CH2Cl2:TFA (97%).

The deprotection of the thiazolidine ring was first examined in solution. This was done

because the deprotection results in a loss of only a methyl group, which gives a mass

difference probably not visible on the sensorgram. Thiazolidine derivative 4 was prepared as

a 25 mM solution in 50 mM acetate buffer (pH 4.0) containing 250 mM methoxyamine

hydrochloride. The deprotection of the thiazolidine ring was monitored by LC-MS over 10 h.

Figure 4 shows the conversion of 4 to the free N-terminal cysteine with time at the two

different temperatures. At 37 °C, the reaction was determined to be greater than 90%

complete after 2 h. Therefore, these conditions were used in the surface-deprotection step,

assuming nearly complete deprotection there as well.

Chapter 3

48

100

80

60

40

20

0

% F

ree C

yste

ine

6005004003002001000

time / min

20 ºC 37 ºC

Figure 4: Deprotection kinetics of the thiazolidine ring in 4 at different temperatures.

3.3 Biosensor surface modification

The biosensor surface (CM5 chip, GE Healthcare) was activated with EDC and NHS

before treatment with a 250 mM solution of 4 in 50 mM borate buffer at pH 8.5. The

remaining activated esters were quenched with 2-ethanolamine (1.0 M, pH 8.5) before

proceeding. The thiazolidine rings were then deprotected with a 250 mM solution of

methoxyamine hydrochloride in 50 mM sodium acetate buffer at pH 4.0 for 2 h at 37 °C, in

accordance with the measurements of the deprotection kinetics. Regeneration of the surface

with successive treatments of 50 mM NaOH in 500 mM NaCl was required to remove non-

covalently bound organics from the surface. At the end of this procedure, approximately 150

resonance units (RU) of the reactive cysteine was reproducibly immobilized onto the dextran

layer coating the sensor surface. Resonance units can be approximately converted to mass

of immobilized compound using the relation 1 RU = 1 pg mm–2.[22] One flow cell has an active

area of 1.1 mm2, and the dextran layer is approximately 100 nm high. Combining this data in

Equation [1] gives an estimated molar concentration in the flow cell of 9 mM.[23]

!

conc (M) "conc (RU)

100 #Mw [1]

A reference surface, functionalized solely with 2-ethanolamine for negating buffer

effects and non-specific protein/peptide adsorption in the SPR set-up, was also prepared.

Typical sensorgrams for chip preparation are given in Figure 5.


49

a) b) 54000

52000

50000

48000

46000

44000

42000

40000

Response / R

U

18001600140012001000800600400200time / s

i ii

iii

54000

52000

50000

48000

46000

44000

42000

40000

Response / R

U

18001600140012001000800600400200time / s

i

iii iii

c) d)

40000

39000

38000

37000

Response / R

U

80006000400020000time / s

Sample Channel Reference Channel

43000

42000

41000

40000

39000

38000

Response / R

U

25002000150010005000time / s

Sample Channel Reference Channel

Figure 5: Typical sensor responses during chip preparation. (a) Modification of the sample channel

with thiazoline derivative 4: i) activation with EDC and NHS; ii) injection of 4; iii) injection of

ethanolamine. (b) Modification of the reference channel with ethanolamine. (c) Deprotection of the

thiazolidine ring with 250 mM methoxyamine at 37 °C. (d) Regeneration of the surface with repeated

injections of 50 mM NaOH and 0.5 M NaCl.

3.4 Immobilization of a peptide-thioester via native chemical ligation

Although the previous results indicated successful modification of the surface with

cysteines, a native chemical ligation reaction of a thioester-modified peptide would provide

the unambiguous proof of principle. As a first example, a streptavidin-binding peptide

sequence identified from phage display containing the well-described HPQ motif was used

for immobilization.[24] In this experiment, the ligation of the SLLAH(Dnp)PQGGG-MPAL

thioester to the biosensor surface was performed over both the reference channel and the

sample channel containing the surface-bound N-terminal cysteines, operating in tandem.

This allowed for monitoring the specific immobilization of peptide onto the sensor surface that

was functionalized with the reactive cysteine partner. Ligation was performed using an

injection of SLLAH(Dnp)PQGGG-MPAL (5 mM) in HBS-EP buffer containing 50 mM 4-

mercapto-phenylacetic acid (MPAA) and 10 mM tris(carboxyethyl)phosphine (TCEP) at a pH

Chapter 3

50

of 7.4.[20] The sensorgram clearly shows the specific immobilization of the decapeptide, only

in the channel that has been functionalized with the reactive cysteine moiety (Figure 6). After

regeneration of the surface to remove non-covalently bound material, the immobilization level

of the streptavidin-binding peptide in the sample channel was measured at 750 RU whereas

the reference channel does not show any appreciable change. This immobilization level

corresponds to approximately 8 mM peptide concentration in the dextran layer (Eq. [1]),

implying a very efficient reaction with over 85% of cysteine residues reacted.

43000

42000

41000

40000

39000

38000

Response / R

U

70006000500040003000200010000time / s

i ii

Ref.

Sample

750 RU

Figure 6: Ligation of a streptavidin-binding peptide to a cysteine-modified biosensor surface: i) 350 !L

injection of 5 mM SLLAH(Dnp)PQGGG-MPAL, 50 mM MPAA, 10 mM TCEP (HBS-EP; pH 7.4; 5 !L

min–1); ii) surface regeneration using repeated injections of 50 mM NaOH and 0.5 M NaCl.

To confirm that the streptavidin-binding peptide was successfully immobilized and

that it could engage in specific binding to its target, a binding experiment was performed

whereby a solution of the protein was introduced over both the reference and sample

channels. The response for the binding of streptavidin (SA) to the SLLAHQPGGG-modified

surface in the sample channel was much higher (! = 3720 RU) than that for the reference

channel, indicating the specificity of the peptide-conjugated surface for SA (Figure 7).

43000

42000

41000

40000

39000

38000

Response / R

U

6004002000

time / s

i ii

Sample

Ref.

Figure 7: Binding response of peptide-modified surface towards streptavidin: i) 50 !L injection of 2 !M

streptavidin (HBS-EP; 2 mM TCEP; pH 7.4; 5 !L min–1); ii) surface regeneration using 50 mM NaOH

and 0.5 M NaCl.


51

Regeneration of the surfaces brought the sensor responses in both channels back to

their initial values. Over several cycles of SA binding and subsequent regeneration, the

sensor responses were reproducible within 1–2% (Figure 8a). The binding of SA was also

concentration dependent, although the particulars of this multivalent interaction is outside the

scope of this chapter (Figure 8b).

a) b)

4000

3000

2000

1000

Re

sp

on

se

/ R

U

7000600050004000300020001000

time / s

6000

5000

4000

3000

2000

1000

Response / R

U

1000800600400200

time / s

32 !M

200 nM

Figure 8: (a) Repeated injections of 2 !M streptavidin over peptide-modified surface (signal: sample

channel – reference channel); (b) Binding kinetics of SA to peptide-modified biosensor chip.

We were also interested in whether the NCL chips would perform well during

repeated cycles of ligation followed by analyte binding. The cysteine-modified surfaces are

designed such that they remain active towards ligations so long as the sulfhydryls are in the

reduced state. To demonstrate this aspect the SLLAHQPGGG peptide was used as well.

Four sequential ligation steps were performed, each followed by a SA binding experiment.

During these experiments, the surface remained active towards further NCL of the HPQ-

peptide and the SA binding capacity was shown to increase monotonically with increasing

HPQ-ligand density (Figure 9). The sensor response in the reference channel remained both

invariant and small, pointing to an unresponsive surface. Taken together, these results

indicate that these cysteine-modified chips are extremely versatile in that analyte binding

experiments may be conducted on the same surface whose immobilized ligand density may

be increased stepwise at any point in the chip"s lifetime.

Chapter 3

52

Figure 9: a); Schematic representation of sequential ligations of the streptavidin-binding peptide

SLLAH(Dnp)PQGGG-MPAL on the same sensor surface. Between ligation steps, reversible binding

with streptavidin can be measured; b) Immobilization levels upon repeated injection of

SLLAH(Dnp)PQGGG-MPAL (5 !L injection of 0.5 mM SLLAH(Dnp)PQGGG-MPAL, 50 mM MPAA, 10

mM TCEP (HBS-EP; pH 7.4; 5 !L min–1)); c) Streptavidin-binding responses to the HPQ-modified

surface (light grey) and a reference surface without peptide (dark grey) after each ligation cycle (50 !L

injection of 2 !M streptavidin in HBS-EP; pH 7.4; 5 !L min–1).

3.5 Immobilization of a protein-thioester via native chemical ligation

Having established the NCL technique as viable to immobilize peptide thioesters onto

cysteine-functionalized biosensor surfaces, we were interested whether we could apply the

same methodology for the immobilization of recombinant proteins. To do so, we employed


53

the well-known green fluorescent protein (GFP) with a C-terminal MESNA thioester obtained

after expression and intein-mediated cleavage with MESNA on a chitin column, as described

in Chapter 2 of this thesis. Ligation of the GFP-MESNA thioester proceeded smoothly over

60 min, again with complete specificity for the sample channel with surface-bound cysteines.

The overall immobilization level in the sample channel was measured at 480 RU while in the

reference channel the sensor response indicated no significant change from the initial level

prior to ligation (Figure 10a). Compared to the immobilization levels reached for the small

peptide sequence, this reaction results in a lower surface covering (approximately 170 #M,

calculated from Eq. [1]). The slower kinetics of this reaction can be explained by the lower

concentration used (100 #M protein thioester, compared to 5 mM peptide thioester) and the

bulkiness of the protein, giving rise to sterical hindrance. A higher level of immobilization can

be easily reached by prolonged reaction, as no plateau level was reached during reaction.

To test whether GFP had been immobilized in the sample channel as intended, a

binding experiment was performed whereby a solution of an antibody directed against GFP

(JL-8, Clontech) was introduced to both the reference and the sample channels. The binding

of anti-GFP to the GFP-modified surface was pronounced and specific (Figure 10b); only

slight non-specific binding was observed in the reference channel. This was also an

indication that the immobilized GFP was still in a conformation that presented a suitable

binding surface for the antibody recognition and that the NCL protocol had not adversely

affected its folding.

a) b)

41000

40000

39000

38000

Re

sp

on

se

/ R

U

50004000300020001000

time / s

i

Sample

Ref.480 RU

250

200

150

100

50

0

Rel. R

esponse / R

U

120010008006004002000

time / s

ii

Sample

Ref.

Figure 10: (a) Ligation of green fluorescent protein to a cysteine-modified biosensor surface: i) 350 !L

injection of 0.1 mM GFP-MESNA, 50 mM MPAA, 10 mM TCEP (HBS-EP; pH 7.4; 5 !L min–1); (b)

Binding of monoclonal antibody directed against GFP to protein-modified surface: ii) 10 !L injection of

14 nM JL-8 anti-GFP in HBS-EP; pH 7.4; 5 !L min–1.

Chapter 3

54

3.6 Immobilization of S-peptide enables measurement of thermodynamic

parameters

Biosensor surfaces modified with proteins or peptides using the NCL strategy can

also be used to extract kinetic and thermodynamic parameters from the interaction between

the immobilized ligand and an analyte of interest. This aspect was investigated using the

bimolecular interaction between S-protein and S-peptide, whose assembly into RNase S has

been first described by Richards and coworkers.[25] The pair derives from RNase A, which

can be cleaved in the presence of the protease subtilisin to generate the 104 aa S-protein

and the 20 aa S-peptide, and is one of the best-studied interactions in biochemistry. In

addition, the choice to study this interaction in the Biacore set-up was made because of our

interest in using the S-peptide and S-protein to build non-covalent assemblies. More about

this research can be found in Chapter 5.

The S-peptide was successfully immobilized onto the cysteine-modified biosensor

surface from its MPAL thioester with a short diglycine spacer (i.e. KETAAAKFENQH(Dnp)

MDSSTSAAGG-MPAL). To mitigate mass transport limitations during the SPR kinetics

experiments, low immobilization levels of the peptide ligand on the surface are required. This

was achieved with a single, one-minute injection of the peptide at a relatively low

concentration (0.5 mM) over the reference and sample channels. After this treatment, the

aimed immobilization level of 110 RU was reached on the cysteine-modified surface in the

sample channel (Figure 11a).

In all previous binding experiments, supplementing 2 mM TCEP to running buffers

ensured the reduced state of all cysteines. However, because the S-protein contains four

disulfides, which are essential for proper structure and binding to S-peptide, running the

binding experiments under reducing conditions was not possible. Therefore, the free

sulfhydryl groups in the sample channel were reacted with Ellman"s reagent prior to binding

experiments to ensure a stable baseline during the kinetics experiments.[26] In the

sensorgram, a clear increase of approximately 35 RU in the sample channel is visible,

whereas no change in response in the reference channel was observed (Figure 11b).


55

a) b)

41500

41000

40500

40000

39500

39000

38500

38000

Re

sp

on

se

/ R

U

140120100806040200

time / s

Sample

Ref.

i110 RU

Inset:37860

37840

37820

37800

37780

37760

37740

Response / R

U

500400300200100

time / s

Sample

Ref.

ii

35 RU

Figure 11: (a) Ligation of S-peptide thioester to cysteine-modified chip surface: i) 5 !L injection of 0.5

mM S-peptide-MPAL, 50 mM MPAA, 10 mM TCEP (HBS-EP; pH 7.4; 5 !L min–1). The inset shows a

magnification of the sensorgram. Here, the kinetics of ligation in the sample channel is clearly visible;

(b) Blocking of free sulfhydryl groups by Ellman"s reagent: ii) 25 !L injection of 2 mM 5-5"-dithio-bis(2-

nitrobenzoic acid) (HBS-EP; pH 7.4; 5 !L min–1).

The binding kinetics of the S-protein and S-peptide interaction at pH 7.4 were

investigated by following the reference subtracted sensor response in the sample channel

during 5 minute injections of the S-protein over a concentration range of 10 nM to 1 #M. The

dissociation phase was also monitored for 5 minutes and was followed by a regeneration

step of 10 mM glycine at pH 1.5 before the next injection of S-protein. An elevated flow rate

of 70 #L min–1 was used to diminish mass transport limitation effects. The sensorgrams

(Figure 12) were fit to a 1:1 binding model as dictated by the stoichiometry of the interaction.

300

250

200

150

100

50

0

Response / R

U

700600500400300200100

time / s

10 nM

1 !M

Figure 12: Kinetics of S-protein binding to a surface modified with the S-peptide. Fits to experimentally

obtained sensorgrams (subtracted signal Ch2 – Ch1; black) are shown in grey. S-protein

concentration ranged from 10 nM to 1 !M. (HBS-EP; pH 7.4; 70 !L min–1).

The fitted value for kon was 6.1 $ 105 M–1 s–1 with the koff at 2.4 $ 10–3 s–1 and an overall Kd of

3.9 nM, concomitant with previous accounts of the high affinity interaction between the S-

Chapter 3

56

protein and the S-peptide.[27] In Chapter 5, the S-peptide–S-protein interaction is discussed in

more detail.

3.7 Discussion and conclusions

A facile and direct strategy has been presented for the immobilization of peptides and

proteins bearing C-terminal thioesters on cysteine-modified biosensor chip surfaces via

native chemical ligation. The ligand density at the surface can be controlled at any point in

the chip"s construction, even between analyte binding experiments. The ligand-modified

surfaces showed analyte-specific responses in all of the cases presented, and the kinetic

and thermodynamic data extracted from binding experiments are consistent with accounts in

the literature for the model systems used. A reversible inactivation of sulfhydryls on the

surface was achieved via reaction with Ellman"s reagent, thereby making this immobilization

strategy suitable for use in binding experiments with disulfide-containing proteins. However,

the use of disulfide-containing proteins for immobilization on the surface has not been

attempted yet, and might be complicating due to the reducing conditions used during the

immobilization procedure. Removing TCEP completely from the ligation mixture and lowering

the concentration of MPAA used, may prevent reduction of disulfides prevented for the most

part. Being able to follow the immobilization reaction on the surface in time provides the

opportunity for detailed studies concerning the kinetics of native chemical ligation on

surfaces, which can be of value for various applications requiring peptide- or protein-modified

surfaces.


General

Unless stated otherwise, all solvents (p.a. quality) and other chemicals were obtained from commercial sources and used as received. Water was demineralized prior to use. Dichloromethane was obtained by distillation from P2O5. DMF was stored over 3 Å molecular sieves. Water was deionized prior to use. HBTU (Iris Biosciences), DIPEA (Aldrich), N-hydroxysuccinimide (Iris Biosciences or GE Healthcare), tert-butyl phenylcarbonate (Aldrich), ethylenediamine (Aldrich),

trifluoroacetic acid (Aldrich), EDC (GE Healthcare), 4-mercaptophenylacetic acid (Aldrich), tris(carboxyethyl)phosphine hydrochloride (TCEP, Aldrich), methoxyamine hydrochloride (Aldrich), ribonuclease S (grade XII-S, Sigma), Ellman"s reagent (Merck), S-trityl 3-mercaptopropionic acid (Sigma), triisopropylsilane (Aldrich), p-cresol (Aldrich) were all used as supplied. All t-Boc-protected amino acids and MBHA resins (0.92 mmol g–1) were obtained from NovaBiochem. Standard 1H NMR and 13C NMR experiments were performed on a Varian Gemini-2000 300 MHz spectrometer, a Varian Mercury Vx 400 MHz spectrometer, and a Varian Unity Inova 500 MHz spectrometer at 298 K. Chemical shifts are reported in parts per million relative to tetramethylsilane (TMS). Reversed phase high pressure liquid chromatography (RP-HPLC) was performed on a Varian Pro Star HPLC system coupled to a UV-Vis detector probing at 214 nm using a Vydac protein & peptide C18 column. A gradient of acetonitrile in water (both containing 0.1% TFA) was used to elute products. ESI-MS spectra were recorded on an Applied Biosystems Single Quadrupole Electrospray Ionization Mass


57

Spectrometer API-150EX in positive mode. For peptides, all reported values are exact masses, for proteins, average masses are calculated. All SPR experiments were performed on a Biacore T100 instrument (GE Healthcare) using CM5 type chips. HBS-EP was prepared freshly from commercially available 10 $ HBS-EP (GE Healthcare) by diluting 1:9 with demiwater and subsequent filtering. N-(tert-butyloxycarbonyl)thiazolidine-4-carboxylic acid N-hydroxysuccinimidyl ester 1 The title compound was prepared from N-(tert-butyloxycarbonyl)thiazolidine carboxylic acid (1.00 g, 4.3 mmol), N-hydroxysuccinimide (542 mg, 4.7 mmol), HBTU (1.63 g, 4.3 mmol) and DIPEA (1.66 g, 12.9 mmol) in 10 mL DMF. After 12 h, the reaction mixture was concentrated in vacuo and the residue dissolved in 100 mL ether. The ethereal layer was washed with sat. KCl (5 $ 100 mL), sat. NaHCO3 (5 $ 100 mL), and deionized water (10 $ 100 mL) before drying over Na2SO4. After filtration and concentration, 1 was obtained as a colorless solid (950 mg, 67%). 1H NMR (CDCl3, 400 MHz): (3:1 mixture of diastereomers) % = 5.16 (br, 1H, N-CH-C(=O)NHS, minor), 4.91 (t, 1H, J = 5.7 Hz, N-CH-C(=O)NHS, major), 4.61 (dd, 2H, J = 54 Hz, J = 9.2 Hz, N-CH2-S, major), 4.54 (dd, 2H, J = 65 Hz, J = 9.0 Hz, N-CH2-S, minor), 3.50 (m, 2H, CH-CH2-S, minor), 3.41 (m, 2H, CH-CH2-S, major), 2.83 (s, 4H, -CH2-CH2-), 1.47 (s, 9H, -CH3).

13C NMR (CDCl3, 100 MHz): % = 168.27, 166.34, 152.27, 82.10, 81.54, 59.33, 49.02, 48.18, 34.70, 33.08, 27.86, 27.63, 25.25. IR (cm–1): " = 3000, 2943, 1816, 1784, 1739, 1696, 1377, 1364, 1349, 1311, 1271, 1258, 1207, 1164,

1145, 1107, 1085, 1059, 1045, 991, 948, 932, 895, 866, 855, 811, 785, 772, 761, 732. EI-MS: Calc. for C13H18N2O6S: m/z 330.08; Obsd: m/z 229, 188, 132, 88, 57, 41. N-(tert-butyloxycarbonyl)thiazolidine-4-carboxylic acid (2-[N-tert-butyloxycarbonyl] aminoethyl)

amide 3 A solution of mono-N-(tert-butoxycarbonyl)ethylenediamine 2 (361 mg, 2.25 mmol) in 2 mL CH2Cl2

was added to a flask containing 1 (676 mg, 2.05 mmol) and DIPEA (793 mg, 6.14 mmol) in 3 mL of CH2Cl2. After 6 h, the solvent was removed in vacuo and the residue taken up in 50 mL ethyl acetate before extraction with sat. NaHCO3 (5 $ 50 mL), deionized water (5 $ 50 mL), and brine. The organic layer was dried over Na2SO4, filtered and concentrated to give 3 as a colorless oil (735 mg, 95%). 1H NMR (CDCl3, 400 MHz): % = 6.92 (br, 1H, -C(=O)-NH-), 4.92 (br, 1H, Boc-NH-), 4.61 (m, 2H, N-CH2-S), 4.38 (br, 1H, N-CH-C(=O)), 3.36 (m, 2H, CH-CH2-S), 3.23 (m, 4H, Boc-NH-CH2-CH2-NH-), 1.47 (s, 9H, -CH3), 1.43 (s, 9H, -CH3).

13C NMR (CDCl3, 100 MHz): % = 170.64, 156.37, 153.72, 81.56, 79.39, 76.89, 62.95, 49.21, 40.46, 39.90, 33.28, 28.06, 27.97. IR (cm–1): " = 3322, 2977, 2933, 1682, 1524, 1365, 1272, 1250, 1158, 1112, 999, 860, 758.

EI-MS: Calc. for [C16H29N3O5S]: m/z 375.18; Obsd: m/z 319, 263, 246, 215, 159, 115, 88, 57, 41. Thiazolidine-4-carboxylic acid (2-aminoethyl)amide bisTFA adduct 4 Compound 3 (730 mg, 1.94 mmol) was dissolved in 10 mL of a 1:1 mixture of CH2Cl2 and TFA. After 3 h, the reaction mixture was concentrated in vacuo and the residue taken up in 25 mL of deionized water. The aqueous layer was washed with CH2Cl2 (3 $ 25 mL) before freeze drying. The product was isolated as a pale brown oil (330 mg, 97%). 1H NMR (d6-DMSO, 400 MHz): % = 8.64 (s, 1H, -C(=O)-NH-), 7.83 (br, 3H, -CH2-NH3), 6.0-5.0 (br, 2H, -RCH-NH2-CHS), 4.28 (dd, 2H, J = 15.4 Hz, J = 6.6 Hz, N-CH2-S), 4.22 (dd, 1H, J = 15.8 Hz, J = 6.2 Hz, N-CH-C(=O)), 3.34 (m, 2H, H3N-CH2-CH2), 3.25-3.12 (m, 2H, CH-CH2-S), 2.88 (m, 2H, H3N-CH2-CH2).

13C NMR (d6-DMSO, 100 MHz): % = 168.74, 63.71, 50.54, 38.76, 37.21, 34.06. IR (cm–1): " = 3061, 1778, 1665, 1567, 1434, 1173, 1131, 838, 797, 722, 706.

LC-MS: m/z [C6H13N3OS + H]+ Calcd. 176.08 Da, Obsd. 176.20 Da.

Chapter 3

58

Deprotection kinetics of thiazolidine 4 in aqueous buffer Thiazolidine deravitive 4 was prepared as a 25 mM solution in 50 mM acetate buffer (pH = 4.0) containing 250 mM methoxyamine hydrochloride. The deprotection of the thiazolidine ring at either 25 °C or 37 °C was monitored by LC-MS over 10 h. At 37 °C, the reaction was determined to be greater that 90% complete after 2 h. NCL chip preparation All experiments were performed on a Biacore T100 (GE Healthcare) on CM5 chips. The running buffer was HBS-EP at pH 7.4 at a flow rate of 10 #L min–1 unless otherwise stated. Thiazolidine derivative 4 was immobilized in flow channels 2 and 4 at 25 °C using the following three step procedure: (1) 7 min injection of an equimolar solution of EDC and NHS (1.0 M); (2) 7 min injection of 250 mM 4 in 50 mM borate buffer at pH 8.5; (3) 7 min injection of 1.0 M ethanolamine HCl at pH 8.5. For the reference surfaces, ethanolamine HCl was immobilized in flow channels 1 and 3 using a similar approach. The temperature of the chip was raised to 37 °C before 2 $ 70 min injections of a 250 mM methoxyamine·HCl in 50 mM acetate buffer at pH 4.0 over flow channels 1–4 at a flow rate of 5 #L min–1. Two separate injections were needed to reach a 2 h incubation time (Biacore software only permits injections lasting 60 min or less). After cooling the chip to 25 °C, regeneration of the surface was initiated using 10 # 30 s pulses of 50 mM NaOH in 0.5 M NaCl over all channels at a flow rate of

10 #L min–1. Chips prepared according to the above procedure were used directly in ligation experiments, or were stored at 4 °C in HBS-EP buffer at pH 7.4 containing 2 mM TCEP until needed. General synthesis of MPAL-activated peptides The general synthesis of MPAL-activated peptides is described in Chapter 2.

Streptavidin-binding HPQ peptide MPAL thioester – SLLAH(Dnp)PQGGG-MPAL After the synthesis of MPAL-thioester on the resin, respectively, Boc-Gly, Boc-Gly, Boc-Gly, Boc-Gln, Boc-Pro, Boc-His(Dnp), Boc-Ala, Boc-Leu, Boc-Leu and Boc-Ser(Bzl) were coupled. The peptide was cleaved from the resin using liquid HF and was purified using RP-HPLC with a gradient of 20–40% acetonitrile in water over 90 min. The pure peptide was obtained in 82% yield (95 mg). LC-MS: m/z [C55H83N17O18S + 2H]2+ Calcd. 651.8 Da, Obsd. 651.9 Da; [M + H]+ Calcd. 1302.6 Da, Obsd. 1302.6 Da. S-Peptide MPAL thioester - KETAAAKFERQH(Dnp)MDSSTSAAGG-MPAL After the synthesis of MPAL-thioester on the resin, respectively, Boc-Gly, Boc-Gly, Boc-AlA, Boc-Ala, Boc-Ser(Bzl), Boc-Thr(Bzl), Boc-Ser(Bzl), Boc-Ser(Bzl), Boc-Asp(OcHxl), Boc-Met, Boc-His(Dnp), Boc-Gln, Boc-Arg(Tosyl), Boc-Glu(OcHxl), Boc-Phe, Boc-Lys(ClZ), Boc-Ala, Boc-Ala, Boc-Ala, Boc-Thr(Bzl), Boc-Glu(OcHxl) and Boc-Lys(ClZ) were coupled. The peptide was cleaved from the resin using liquid HF, removing also the protecting groups from the side chains (except for the Dnp groups). The peptide was purified by RP-HPLC using a gradient of 15–35% acetonitrile in water over 90 min. The pure peptide was obtained in 54% overall yield. LC-MS: m/z [C108H168N34O40S2 + 3H]3+ Calcd. 882.7 Da, Obsd. 883.0 Da; [M + 2H]2+ Calcd. 1323.6 Da, Obsd. 1324.3 Da. Purification of S-protein

Preparative RP-HPLC was used to isolate S-protein from RNase S. RNase S (20 mg, 1.46 #mol) was dissolved in deionized water and eluted using a linear gradient of 20–40% acetonitrile in water (containing 0.1% TFA) over 20 min. The S-protein was obtained in 71% yield (12.0 mg, 1.04 #mol).


59

Mass spectrometry revealed that the protease subtilisin does not hydrolyze the amide bond between residues 20 and 21 with complete selectivity; the bond between residues 21 and 22 is also cleaved, although to a much lesser extent.[28] ESI-MS: deconvoluted mass [C486H759N143O161S11, S21–124, predominant species, oxidized] Calcd. 11534.0 Da, Obsd. 11534.6 Da. [C483H754N142O159S11, S22–124, minor species, oxidized] Calcd. 11446.9 Da, Obsd. 11447.8 Da. Native Chemical Ligation of SLLAH(Dnp)PQGGG-MPAL

SLLAH(Dnp)PQGGG-MPAL (5 mM) was preincubated with 4-mercaptophenylacetic acid (50 mM) and TCEP (10 mM) in HBS-EP at pH 7.4 (30 min, 20 °C). At a flow rate of 5 #L min–1, the ligation mixture was injected over flow channels 1 and 2 for 60 min. Subsequent regeneration using 50 mM NaOH in 0.5 M NaCl yielded the surface with 750 RU protein immobilized. Binding experiments with streptavidin (Sigma-Aldrich) were performed at different concentrations in HBS-EP at a flow rate of 5 #L min–1. Sequential HPQ immobilization on the same biosensor surface SLLAH(Dnp)PQGGG-MPAL (0.5 mM) was preincubated with 4-mercaptophenylacetic acid (50 mM) and TCEP (10 mM) in HBS-EP at pH 7.4 (30 min, 20 °C). At a flow rate of 5 #L min–1, the ligation mixture was injected over flow channel 1 and 2 for 60 sec. Subsequent regeneration using 50 mM NaOH in 0.5 M NaCl yielded a surface with 105 RU peptide immobilized. Streptavidin (2 #M) was injected for 600 s, followed by regeneration of the surface with 50 mM NaOH in 0.5 M NaCl. This cycle

was repeated 3 times. Native Chemical Ligation of GFP-MESNA

GFP-MESNA was obtained as described in Chapter 2. The protein thioester (0.1 mM) was preincubated with 4-mercaptophenylacetic acid (50 mM) and TCEP (10 mM) in HBS-EP at pH 7.4 (30 min, 20 °C). At a flow rate of 5 #L min–1, the ligation mixture was injected over flow channels 1 and 2 for 60 min. Subsequent regeneration using 50 mM NaOH in 0.5 M NaCl yielded the surface with 450 RU protein immobilized. Binding experiments with anti-GFP (JL-8, Clontech) were performed at a 500-fold dilution (approx. 14 nM) in HBS-EP at a flow rate of 5 #L min–1. S-Peptide immobilization S-Peptide (0.5 mM) was preincubated with 4-mercaptophenylacetic acid (50 mM) and tris(carboxyethyl)phosphine (10 mM) in HBS-EP at pH 7.4 (30 min, 20 °C). At a flow rate of 5 #L min–1, the ligation mixture was injected over flow channels 1 and 2 for 60 s. Subsequent regeneration using 50 mM NaOH in 0.5 M NaCl yielded the surface with 110 RU of immobilized S-peptide. A 0.2 mM solution of 5,5'-dithio-bis(2-nitrobenzoic acid) (DTNB, Ellman"s reagent) was injected to block all free sulfhydryl groups on the surface, giving an additional increase in response in the sample channel of 35 RU. Again, regeneration with 50 mM NaOH in 0.5 M NaCl removed all non-covalently bound organics from the chip. S-Protein binding S-Protein was dissolved in HBS-EP and diluted into the proper concentrations for kinetic binding experiments (1 #M to 10 nM). A kinetic experiment was set up using the Biacore T100 method builder. A flow rate of 70 #l min–1 with an association phase of 300 s, a dissociation phase of 300 s, and a single regeneration step of 10 mM glycine at pH 1.5 were employed. Kinetic data were fitted to a standard 1:1 binding model using BIAevaluation software.

Chapter 3

60

3.9 References

[1] Fields, S.; Song, O.-K. Nature 1989, 340, (6230), 245-246. [2] Phizicky, E. M.; Fields, S. Microbiol Rev 1995, 59, (1), 94-123. [3] Engvall, E.; Perlman, P. Immunochemistry 1971, 8, (9), 871-4. [4] Wiseman, T.; Williston, S.; Brandts, J. F.; Lin, L.-N. Anal. Biochem. 1989, 179, (1), 131-137. [5] Lofas, S.; Malmqvist, M.; Ronnberg, I.; Stenberg, E.; Liedberg, B.; Lundstrom, I. Sensors and

Actuators B: Chemical 1991, 5, (1-4), 79-84. [6] Johnsson, B.; Lofas, S.; Lindquist, G. Anal. Biochem. 1991, 198, (2), 268-77. [7] Kortt, A. A.; Oddie, G. W.; Iliades, P.; Gruen, L. C.; Hudson, P. J. Anal. Biochem. 1997, 253,

(1), 103-11. [8] Yousaf, M. N.; Mrksich, M. J. Am. Chem. Soc. 1999, 121, (17), 4286-4287. [9] Lesaicherre, M.-L.; Uttamchandani, M.; Chen, G. Y. J.; Yao, S. Q. Bioorg. Med. Chem. Lett.

2002, 12, (16), 2079-2083. [10] Kindermann, M.; George, N.; Johnsson, N.; Johnsson, K. J. Am. Chem. Soc. 2003, 125, (26),

7810-7811. [11] Cheung, C. L.; Camarero, J. A.; Woods, B. W.; Lin, T.; Johnson, J. E.; De Yoreo, J. J. J. Am.

Chem. Soc. 2003, 125, (23), 6848-9. [12] Soellner, M. B.; Dickson, K. A.; Nilsson, B. L.; Raines, R. T. J. Am. Chem. Soc. 2003, 125,

(39), 11790-1. [13] Camarero, J. A.; Kwon, Y.; Coleman, M. A. J. Am. Chem. Soc. 2004, 126, (45), 14730-14731. [14] Watzke, A.; Koehn, M.; Gutierrez-Rodriguez, M.; Wacker, R.; Schroeder, H.; Breinbauer, R.;

Kuhlmann, J.; Alexandrov, K.; Niemeyer, C. M.; Goody, R. S.; Waldmann, H. Angew. Chem., Int. Ed. Engl. 2006, 45, (9), 1408-1412.

[15] Dantas de Araujo, A.; Palomo, J. M.; Cramer, J.; Koehn, M.; Schroeder, H.; Wacker, R.; Niemeyer, C.; Alexandrov, K.; Waldmann, H. Angew. Chem., Int. Ed. Engl. 2006, 45, (2), 296-301.

[16] Sun, X.-L.; Stabler, C. L.; Cazalis, C. S.; Chaikof, E. L. Bioconjugate Chem. 2006, 17, (1), 52-57.

[17] Kwon, Y.; Coleman, M. A.; Camarero, J. A. Angew. Chem., Int. Ed. Engl. 2006, 45, (11), 1726-9.

[18] Gauchet, C.; Labadie, G. R.; Poulter, C. D. J. Am. Chem. Soc. 2006, 128, (29), 9274-9275. [19] Dawson, P. E.; Muir, T. W.; Clark-Lewis, I.; Kent, S. B. Science 1994, 266, (5186), 776-9. [20] Johnson, E. C. B.; Kent, S. B. H. J. Am. Chem. Soc. 2006, 128, (20), 6640-6646. [21] Bang, D.; Kent, S. B. H. Angew. Chem., Int. Ed. Engl. 2004, 43, (19), 2534-2538. [22] Stenberg, E.; Persson, B.; Roos, H.; Urbaniczky, C. J. Colloid Interface Sci. 1991, 143, (2),

513-526. [23] Muller, K. M.; Arndt, K. M.; Pluckthun, A. Anal. Biochem. 1998, 261, (2), 149-158. [24] Katz, B. A. Biochemistry 1995, 34, (47), 15421-9. [25] Richards, F. M.; Vithayathil, P. J. J. Biol. Chem. 1959, 234, 1459-65. [26] Ellman, G. L. Arch. Biochem. Biophys. 1958, 74, (2), 443-50. [27] Schreier, A. A.; Baldwin, R. L. J. Mol. Biol. 1976, 105, (3), 409-426. [28] Doscher, M. S.; Hirs, C. H. Biochemistry 1967, 6, (1), 304-12.

61

4

FROM PHAGE DISPLAY TO DENDRIMER DISPLAY Abstract:

Phage display is widely used for the selection of target-specific peptide sequences.

Although the phage presents multiple copies of the peptide, the selected peptides are mostly

used in a monovalent fashion, which often results in an unanticipated decrease in affinity and

specificity. This chapter describes the use of a synthetic multivalent scaffold to mimic the

phage, maintaining the peptide valency. Using phage display, a peptide sequence was

selected for specific binding to streptavidin. Native chemical ligation was used to prepare

multivalent peptides based on this sequence similar to the multivalent peptide dendrimers

described in Chapter 2. Peptides with different valency (from 1–4 copies per scaffold) and

spacing were prepared and their affinity for streptavidin was measured using surface

plasmon resonance. All multivalent peptides showed a significant increase in affinity

compared to their monovalent counterpart and a binding model was used to describe the

multivalent effect in a quantitative manner. However, the peptide dendrimers still showed

considerably lower affinity than the streptavidin-binding phage. Possible reasons for this

difference are discussed as well as suggestions for further improvement of the dendrimer

display by optimization of both scaffold rigidity and spacing of ligands.

Chapter 4

62

4.1 Introduction

Selective targeting of compounds (e.g. fluorescent imaging probes, therapeutics,

contrast agents) to structures or receptors in biological systems is important for the

understanding and treatment of disease. For example, the delivery of contrast agents and

chemotherapeutics to tumor cells in vivo is an important research area. Historically,

antibodies have been the targeting element of choice due to their high affinity and

selectivity.[1,2] Monoclonal antibodies are complicated and large protein assemblies containing

multiple fragments that linked by disulfide bonds, making their production a complicated

process. In recent years, target-specific oligopeptides have been identified for a variety of

biomarkers (e.g. antibodies, enzymes, cell receptors).[3] The major advantage of peptides

over antibodies is their ease of synthesis on a solid support. However, the identification of a

peptide sequence with a high selectivity and affinity for a certain target can be difficult. One

method of enhancing peptide affinity is by attaching multiple peptides to a multivalent

scaffold.[4]

In phage display bacteriophages are employed to rapidly screen a large library of

random peptides or proteins for binding to a specific target. Different types of phage display

libraries are available, varying in both the type of phage used and the coat protein to which

the peptide or protein is fused. For oligopeptide selection the filamentous M13 phage is often

used, with the peptide sequence fused to either the pVIII or the pIII coat protein.[5-8] Because

five copies of the pIII protein are present on the head of this phage, also five copies of the

peptide will be presented on each phage. Libraries containing five copies of random

heptapeptide or dodecapeptides sequences fused to the N-terminus of the pIII protein on

M13 phages are commercially available (Figure 1).

Figure 1: Schematic representation of an M13 bacteriophage, showing the different coat proteins. The

ssDNA is encapsulated by five different coat proteins, of which pVIII is the most abundant (2700

copies). Five copies of the pIII protein are located on the phage head (left).

When using peptide sequences selected from phage display it is often neglected that

the initial selection is based on a multivalent peptide assembly rather than on a single copy.

Most applications rely on monovalent display of the peptide sequence for targeting

applications, which often results in an unanticipated decrease in affinity and specificity

compared to the phage. In this chapter, a synthetic scaffold based on dendrimers will be

used to reconstitute the multivalent character of the phage display-derived peptides, without


63

the presence of the phage coat proteins or genetic information. The native chemical ligation

strategy presented in Chapter 2 will be used to connect peptides to the synthetic scaffolds.

Streptavidin is used as a well-defined multivalent model target system to provide a proof of

principle for the concept of dendrimer display and gain a quantitative understanding of

multivalent interactions.

4.2 Selection of streptavidin-binding peptide via phage display

To test whether multivalent phage-derived peptides display improved binding

characteristics, streptavidin was used as a model target system. Streptavidin consists of four

identical subunits, with a total mass of about 60 kDa. Streptavidin is used predominantly

because of its specific interaction with biotin, yielding a tight complex of streptavidin and four

biotin molecules (Ka ~ 1014 M–1, Figure 2).[9-13] Libraries of small peptide sequences have been

screened for streptavidin binding using phage display. This has led to the identification of the

streptavidin-binding motif His-Pro-Gln (HPQ), which binds in the biotin-binding site of

streptavidin with a Kd of 50–150 !M.[14-16]

Figure 2: a) Structure and schematic representation of streptavidin (PDB: 1SWB) [17]; b) Structure and

schematic representation of biotin; c) Schematic representation of the binding of four biotin molecules

to one streptavidin protein.

To generate phages displaying this HPQ-peptide, we performed a phage display

selection experiment using immobilized streptavidin as target surface. In the selection

experiment, a plate was coated with streptavidin. After washing and blocking of remaining

surface with BSA, a phage library presenting linear 7-mer peptides was incubated on the

surface. A biotin solution was used to remove bound phages by competitive elution. Two

more rounds of phage selection were performed, each time starting with the enriched and

amplified phage pool from the previous panning round. After round three, six plaques were

used for amplification of single phage clones. DNA from each phage clone was extracted and

sequenced, resulting in the peptide sequences reported in Table 1.

Chapter 4

64

Table 1: Sequences of the six phage clones selected for streptavidin binding after three rounds of

panning. Peptides are connected to the pIII protein via their C-terminus.

Phage clone Amino acid sequence

SA1 S L L A H P Q

SA2 N L L N H P Q

SA3 S T H T S A Q

SA4 N L L N H P Q

SA5 S L I A H P Q

SA6 T L L A H P Q

Table 1 shows that five out of six peptides display the HPQ motif, and this motif is

predominantly preceded by a stretch of hydrophobic residues (i.e. S, L, I, A), corresponding

to the sequences reported in literature.[14-16] Phage clone SA1 was selected for further

experiments, as it best represents the consensus sequence. The binding constant of phage

SA1 was determined using an enzyme-linked immunosorbent assay (ELISA) solid phase

binding experiment. A 96 well plate was coated with streptavidin and a dilution range of

phage SA1 was incubated in the streptavidin-coated wells. After washing of the plate, bound

phages were detected using an anti-M13 antibody labeled with horseradish peroxidase

(HRP). This antibody was incubated on the plate, followed by washing and addition of

substrate hydrogen peroxide and a hydrogen donor (ABTS) that changes color during the

reaction. Monitoring the absorption at 405 nm in time gives a measure of phage bound to the

surface. The result of the solid phase binding assay is depicted in Figure 3. The data points

in Figure 3 were fitted using a one-site binding model, yielding a Kd of 1 " 1010 phages mL–1,

which corresponds to a molar Kd of 16 pM.[18] This sub-nanomolar Kd indicates very strong

binding of the phage to streptavidin on the surface.

0.20

0.15

0.10

0.05

0.00

!O

D (

405 n

m)

105

106

107

108

109

1010

1011

1012

pfu / mL Figure 3: Solid phase binding assay of phage SA1 on a streptavidin-coated surface. Error bars

represent ± S.D. for triplo measurements. The solid line represents the fit to a one-site binding model,

giving a Kd of 1010 plaque-forming units (pfu) mL–1 and a (!OD405)max of 0.153.


65

4.3 Synthesis of multivalent peptides

To make multivalent streptavidin-binding peptides that mimic in a simple way the

phage head, the consensus sequence obtained from phage display was coupled to a

multivalent synthetic scaffold using native chemical ligation. To get insight into the influence

of peptide valency, a peptide tetramer based on the cysteine dendrimer C1 (Chapter 2) was

synthesized. In addition, a peptide dimer was obtained starting from the divalent cysteine-

linker L1, which was synthesized from hexanediamine (Scheme 1).

H2NNH

SH

OHN

O

NH2

HS

H2NNH2

HN

NH

S

OHN

O

NH

S

Ph

Ph Ph

Ph

PhPh

PhPh

Ph

Ph

Ph Ph

i

ii

L1 Scheme 1: Synthesis of the divalent cysteine-linker L1: i) N-Trt- S-Trt-Cys (NHS ester), Et3N, CH2Cl2;

ii) TFA, triethylsilane (2.5% v/v), water (2.5% v/v).

The consensus sequence from phage display experiments, Ser-Leu-Leu-Ala-His-Pro-

Gln, was synthesized using Boc-mediated solid phase chemistry. At the C-terminus a

thioester functionality was introduced to facilitate native chemical ligation with an N-terminal

cysteine. Since on the phage, the peptides are connected to the pIII protein via a three-

glycine spacer, three glycines were placed between the MPAL thioester and the streptavidin-

binding motif. In addition to this sequence, a second peptide was synthesized having an

extra stretch of 3 residues (i.e. Ser-Gly-Gly) between the phage display-derived sequence

and the thioester. After synthesis on the resin, both peptides SLLAH(Dnp)PQGGG-MPAL

and SLLAH(Dnp)PQGGGSGG-MPAL were cleaved using liquid HF, followed by purification

using RP-HPLC. The dinitrophenyl (Dnp) protecting group of histidine is typically removed

using thiophenol prior to cleavage of the peptide from the resin, but this would result in

thiotransesterification of the MPAL thioester and release the peptide from the resin.

Therefore, the Dnp group was maintained on the peptide, as it is removed readily under

native chemical ligation conditions.

Recently, Johnson and Kent described the use of a water-soluble thiol catalyst, 4-

mercaptophenylacetic acid (MPAA).[19] It was shown that ligations of small peptides using

MPAA proceed to completion within a few hours. Therefore, it was decided to use MPAA for

ligation reactions with the HPQ peptide thioesters instead of thiophenol and benzyl

mercaptan. The kinetics of the ligation of SLLAH(Dnp)PQGGG-MPAL with L1 were followed

Chapter 4

66

using LC-MS (Figure 4). Tris(carboxyethyl)phosphine hydrochloride (TCEP) was added to

ensure the reduced state of all cysteines.[20]

876543time / min

SLLAH(Dnp)PQGGG-MPALSLLAHPQGGG-MPAL

t = 3 h

t = 2 h

t = 0 h

t = 1 h

(SLLAHPQGGG)2L1

(SLLAHPQGGG)1L1

Figure 4: Total ion chromatograms monitoring the ligation of SLLAH(Dnp)PQGGG-MPAL with L1 in

time.

After 3 h of incubation, the chromatogram of the reaction mixture showed that all

dicysteine linker had reacted to form (SLLAHPQGGG)2L1, and that only a small fraction of

unreacted peptide was present (due to the small excess of peptide over L1). It is also evident

that the Dnp group is efficiently removed under these ligation conditions. Therefore, the same

conditions were used to synthesize all peptide multimers containing the HPQ sequence.

(SLLAHPQGGG)2L1, (SLLAHPQGGG)4C1, (SLLAHPQGGGSGG)2L1 and

(SLLAHPQGGGSGG)4C1 were synthesized on preparative scale and purified using RP-

HPLC. All products were obtained in reasonable yields (23–55%) and characterized by mass

spectrometry (Figure 5).

a)

NH

HN

NH

HNH2N

HO

O

O

O

O

HN

NH

HN

NH

HN

OH2N

O

O

O

O

N

O

O

HN N

NH

HS

O HN

NH

HN

NH

NH2

OH

O

O

O

O

NH

HN

NH

HN

NH

O NH2

O

O

O

O

N

O

O

NHN

HN

SH

O

NH

HN

NH

HNH2N

HO

O

O

O

O

HN

NH

HN

NH

HN

OH2N

O

O

O

O

N

O

O

HN NHS

O

HN

NH

HN

NH

NH2

OH

O

O

O

O

NH

HN

NH

HN

NH

O NH2

O

O

O

O

N

O

O

NHNSH

O

NH

HN

NH

HNH2N

HO

O

O

O

O

HN

NH

HN

NH

HN

OH2N

O

O

O

O

N

O

O

HN NHS

O

HN

NH

HN

NH

NH2

OH

O

O

O

O

NH

HN

NH

HN

NH

O NH2

O

O

O

O

N

O

O

NHNSH

O

NN

HN

NH

HN

NH

(SLLAHPQGGG)2L1

(SLLAHPQGGG)4C1 Figure 5: a) Structures of (SLLAHPQGGG)2L1 and (SLLAHPQGGG)4C1.


67

b)

HN

SH

O

NH

HS

O

HN

SH

O

NH

HS

O

NN

NH

NH

HN

HN

(SLLAHPQGGGSGG)2L1

(SLLAHPQGGGSGG)4C1

HN

NH

HN

NH

NH2

OH

O

O

O

O

NH

HN

NH

HN

NH

O NH2

O

O

O

OHN N

O

O

NHN

NH

O

OHO

O

NH

HN

HS

O

HN

NH

SH

O

NH

HN

NH

HNH2N

HO

O

O

O

O

HN

NH

HN

NH

HN

OH2N

O

O

O

O

NH

N

O

O

HN N

HN

O

OOH

O

HN

NH

HN

NH

NH2

OH

O

O

O

O

NH

HN

NH

HN

NH

O NH2

O

O

O

OHN N

O

O

NHN

NH

O

OHO

O

HN

NH

HN

NH

NH2

OH

O

O

O

O

NH

HN

NH

HN

NH

O NH2

O

O

O

OHN N

O

O

NHN

NH

O

OHO

ONH

HN

NH

HNH2N

HO

O

O

O

O

HN

NH

HN

NH

HN

OH2N

O

O

O

O

NH

N

O

O

HN N

HN

O

OOH

O

NH

HN

NH

HNH2N

HO

O

O

O

O

HN

NH

HN

NH

HN

OH2N

O

O

O

O

NH

N

O

O

HN N

HN

O

OOH

O

c) d)

Re

lative

Ab

un

da

nce

200015001000500m/z

[M+2H]2+[M+3H]

3+

[M+4H]4+

1080.1720.5

540.5

Re

lative

Ab

un

da

nce

200015001000500

m/z

[M+5H]5+

[M+3H]3+

[M+4H]4+

1101.4

1467.8

881.3[M+7H]

7+

[M+6H]6+

[M+8H]8+

551.3

629.8

734.7

e) f)

Re

lative

Ab

un

da

nce

200015001000500

m/z

[M+2H]2+

[M+3H]3+

[M+4H]4+

1280.8

854.3

641.0

Re

lative

Ab

un

da

nce

200015001000500

m/z

[M+5H]5+

[M+8H]8+

1041.9

579.5

[M+7H]7+

[M+6H]6+

[M+9H]9+

744.5

868.3

651.7

Figure 5, continued: b) Structures of (SLLAHPQGGGSGG)2L1 and (SLLAHPQGGGSGG)4C1;

c) Mass spectrum of (SLLAHPQGGG)2L1 (calcd. mass 2157.2 Da, deconvoluted mass 2158.2 Da);

d) Mass spectrum of (SLLAHPQGGG)4C1 (calcd. mass 4398.3 Da, deconvoluted mass 4400.4 Da);

e) Mass spectrum of (SLLAHPQGGGSGG)2L1 (calcd. mass 2559.2 Da, deconvoluted mass

2559.6 Da); f) Mass spectrum of (SLLAHPQGGGSGG)4C1 (calcd. mass 5202.5 Da, deconvoluted

mass 5204.5 Da).

Chapter 4

68

For binding experiments with streptavidin, the monovalent peptide is required without

the Dnp and thioester functionality present. Both were removed by incubating

SLLAH(Dnp)PQGGG-MPAL with 2-mercaptoethanol (4% v/v) at pH 7.5 for 2 h, yielding the

pure SLLAHPQGGG after RP-HPLC purification. In a similar fashion, SLLAHPQGGGSGG

was prepared.

4.4 Binding studies with streptavidin-binding peptides using SPR

Surface plasmon resonance (SPR) was used to investigate binding of the mono- and

multivalent peptides to streptavidin surfaces. In an SPR experiment, one binding partner is

immobilized on the biosensor chip surface, whereas the other binding partner is flowed over

the surface through a micro-fluidic channel.[21] Streptavidin-coated chips are commercially

available for use in Biacore SPR machines. On this chip, the streptavidin is covalently

attached to a dextran-matrix covering the chip metal surface via primary amines of lysine

residues. Therefore, the precise amount of surface loading and the number of sites available

per streptavidin molecule are not known. A reference channel is required to correct for buffer

effects and aspecific binding responses. The reference surface was prepared by blocking the

binding site of streptavidin with biotin.

Monovalent peptide SLLAHPQGGG was injected over both the reference and sample

channels at a concentration of 500 !M. As can be seen from the sensorgram in Figure 6a,

the response in the sample channel is significantly higher than the response in the reference

channel. The subtracted signal shows the fast kinetics of this interaction (Figure 6b). The fast

kinetics did not allow us to obtain the Kd from the association and dissociation rate constants.

Instead, steady state binding responses were used to measure dissociation constants of the

HPQ peptide.

a) b)

43350

43300

43250

43200

43150

43100

Re

sp

on

se

/ R

U

200150100500

time / s

Sample

Ref.

250

200

150

100

50

0

Re

sp

on

se

/ R

U

200150100500

time / s

Net Response

Figure 6: Binding of 500 !M SLLAHPQGGG to a streptavidin surface in HBS-N, pH 7.4, at 5 !L min–1:

a) Response in sample channel (black line, streptavidin surface) and reference channel (grey line,

biotin-blocked streptavidin surface); b) Net response (sample signal – reference signal).


69

The peptide was injected over the streptavidin surface at different concentrations,

ranging from 5–1000 !M, for 2 minutes with a 20-minute delay between two injections. Also,

a scrambled peptide sequence was synthesized (i.e. SALQLPHGGGC) and tested for

binding to the streptavidin-coated chip. For SLLAHPQGGG, the binding response increase

with increasing peptide concentration can be fitted to a one-site binding model, with a Kd of

142 ± 15 !M and an Rmax 187 ± 7 RU (Figure 7). This Kd value is comparable with values

reported in literature for monovalent peptides[16], but it is 107-fold lower than the phage Kd

estimated from the solid phase binding assay. For the scrambled sequence no significant

binding was observed, showing that binding is specific for the HPQ motif.

200

150

100

50

0

Re

sp

on

se

/ R

U

1000800600400200[peptide] / !M

SLLAHPQGGG scrambled peptide

Figure 7: Steady state binding response (RU) at different peptide concentrations for the streptavidin

binding peptide SLLAHPQGGG and scrambled sequence SALQLPHGGGC in HBS-N, pH 7.4, at 5 !L

min–1. The line represents the fit to a one-site binding model.

The dissociation constants of the divalent and tetravalent peptides were measured in

a similar fashion as the monomeric peptides, both for the short spacer (Figure 8) and for the

longer spacer including the additional three residues (Figure 9). All concentrations are

calculated on a mole basis per construct and not per peptide. All binding responses were

fitted to a one-site binding model. Both peptide dimer and tetramer bind stronger to the

streptavidin surface than the monomer, with respectively a Kd of 11.2 ± 1.4 !M and a Kd of

3.8 ± 0.5 !M. Even if these dissociation constants are corrected for the number of peptides

per ligand, yielding a Kd of 22.4 !M for the dimer and a Kd of 15.2 !M for the tetramer, these

values are significantly lower than the monovalent dissociation constant. Although the fit for

the peptide tetramer at higher concentrations was not perfect, inclusion of an additional

binding site in the model did not significantly improve this fit.

Chapter 4

70

450

400

350

300

250

200

150

100

50

0

Response / R

U

5004003002001000[Ligand] / !M

SLLAHPQGGG

(SLLAHPQGGG)2

L1

(SLLAHPQGGG)4

C1

Figure 8: Steady state binding response (RU) at different molar concentrations for binding of peptide

assemblies SLLAHPQGGG, (SLLAHPQGGG)2L1 and (SLLAHPQGGG)4C1 to streptavidin in HBS-N,

pH 7.4, at 5 !L min–1. The lines represent fits to a one-site binding model (SLLAHPQGGG: Kd = 142 ±

15 !M; (SLLAHPQGGG)2L1: Kd = 11.2 ± 1.4 !M; (SLLAHPQGGG)4C1: Kd = 3.8 ± 0.5 !M).

To check whether the spacer length affects the affinity for streptavidin, dissociation

constants for the set of peptides with an extra Ser-Gly-Gly strech (i.e. SLLAHPQGGGSGG,

(SLLAHPQGGGSGG)2L1 and (SLLAHPQGGGSGG)4C1) were measured using the SPR

experimental setup (Figure 9). All binding responses were fitted to a one-site binding model,

showing increased affinity for both peptide dimer and tetramer (Kd = 8.4 ± 1.7 !M and Kd =

3.7 ± 0.9 !M, respectively) compared to peptide monomer (Kd = 164 ± 18 !M). Again,

inclusion of an additional binding site in the model did not significantly improve the fit for the

tetramer peptide binding data. These results show that no considerable effect of linker length

on the dissociation constants is observed.

500

450

400

350

300

250

200

150

100

50

0

Response / R

U

5004003002001000[Ligand] / !M

SLLAHPQGGGSGG

(SLLAHPQGGGSGG)2

L1

(SLLAHPQGGGSGG)4

C1

Figure 9: Steady state binding response (RU) at different molar concentrations for peptide assemblies

SLLAHPQGGGSGG, (SLLAHPQGGGSGG)2L1 and (SLLAHPQGGGSGG)4C1 to streptavidin in HBS-

N, pH 7.4, at 5 !L min–1. The lines represent fits to a one-site binding model (SLLAHPQGGGSGG: Kd

= 164 ± 18 !M; (SLLAHPQGGGSGG)2L1: Kd = 8.4 ± 1.7 !M; (SLLAHPQGGGSGG)4C1: Kd = 3.7 ± 0.9

!M).


71

Because the affinities of all multivalent peptides are much lower than the affinity of the

phage for streptavidin, a competition experiment was performed to see whether the peptides

can compete for binding with the phage. Similar to the experiment with only phage present, a

hydrophobic plate was coated with streptavidin, yielding a densely coated streptavidin

surface. Different amounts of peptides SLLAHPQGGG, (SLLAHPQGGG)2L1 and

(SLLAHPQGGG)4C1 were mixed with phage SA1 (final concentration 16 pM) and the mixture

was incubated on the streptavidin surface. After thorough washing, the amount of phage

bound to the plate was quantified using an anti-M13 antibody labeled with HRP (Figure 10).

0.035

0.030

0.025

0.020

0.015

0.010

0.005

0.000

!O

D (

40

5 n

m)

1 10 100 1000[Ligand] / "M

SLLAHPQGGG

(SLLAHPQGGG)2 L1

(SLLAHPQGGG)4 C1

Figure 10: Competition experiment showing binding of phage SA1 to a streptavidin-coated surface in

the presence of various concentrations of peptide assemblies SLLAHPQGGG (rounds),

(SLLAHPQGGG)2L1 (triangles) and (SLLAHPQGGG)4C1 (squares). Individual points represent duplo

measurements. The lines represent fits to a Hill-equation (SLLAHPQGGG: IC50 = 360 !M;

(SLLAHPQGGG)2L1: IC50 = 215 !M; (SLLAHPQGGG)4C1: IC50 = 100 !M). All concentrations are

calculated on a per mole basis.

All three peptides compete with the phage for streptavidin-binding, but only at

relatively high concentration (IC50 values 100–360 !M). The peptide tetramer is most efficient

in competition with the phage, followed by the peptide dimer and monomer. The results can

be corrected for the number of peptides per ligand, yielding an IC50 value of 430 !M for the

peptide dimer and an IC50 value of 400 !M for the peptide tetramer. These results contradict

the direct binding studies using SPR, where a clear multivalent effect is present. A possible

explanation for this inconsistency can be found in the different streptavidin surface used in

both assays; the streptavidin in the solid phase assay might have only one binding site left at

one side of the protein, due to the hydrophobic adsorption on the plate. This would mean the

peptide dimer and tetramer have no second binding site within reach, whereas the phage

can span multiple streptavidin molecules due to its 7 nm diameter and therefore can still bind

in a multivalent fashion. Further research, including competition experiments on a better-

defined streptavidin surface, is needed to test this hypothesis.

Chapter 4

72

4.5 Discussion

In this research synthetic multivalent scaffolds were used to reconstitute the

multivalent character of phage-derived peptides. A streptavidin-binding peptide was selected

and different multivalent peptides containing the streptavidin-binding sequence were

synthesized. The phage-derived peptides displayed on synthetic multivalent scaffolds

showed improved binding characteristics compared to the monovalent peptide, but their

affinity to streptavidin was still significantly weaker than that of the original phage displaying

five copies of the same sequence. To understand the behavior of the peptide assemblies, a

multivalent binding model, combined with an analysis of the orientation of the different

binding sites on streptavidin is used. The multivalent binding events are described in a

model, from which the correlation between the experimental equilibrium constants Ka of the

peptide assemblies and the effective molarity can be determined.

Although streptavidin in principle has four binding sites, these sites are oriented in

pairs, at opposite ends of the protein. X-ray structures of several streptavidin–peptide

complexes have been determined, giving information about the distances between adjacent

binding sites. Figure 11 shows the representation of the X-ray structure of streptavidin with

the NWS-HPQ-FEK sequence bound. Two binding sites are in close proximity and the

distance between the two glutamine residues is approximately 27 Å. Because the peptide is

somewhat buried into the protein, the distance to be spanned by the linker is approximately

30 Å. Considering the linker lengths of our peptide assemblies, only two binding sites are

within reach at the same time. Therefore, the binding scheme that needs to be considered is

a divalent target, to which mono- or multivalent peptides bind.

Figure 11: Structure of streptavidin with four HPQ-peptides bound, showing the distance between two

binding sites is approximately 27 Å (PDB: 1KL3).[22]


73

The choice for a divalent target model is also consistent with the Rmax values

observed for mono-, di- and tetravalent peptides, as summarized in Table 2. The increase in

Rmax is consistent with the increasing molecular weight of the dimer and tetramer: for the

dimer (SLLAHPQGGG)2L1, the linker L1 accounts for 15% of the molecular weight, whereas

for the tetramer (SLLAHPQGGG)4C1, the dendrimer core C1 and two out of four peptides

Table 2: Dissociation and association constants and maximum reponses for peptide mono- and

multimers.

peptide Kd (!M) Ka (M–1) Rmax (RU)

SLLAHPQGGG 142 ± 15 7.0 (± 0.7) ! 103 187 ± 7

SLLAHPQGGGSGG 164 ± 18 6.1 (± 0.7) ! 103 256 ± 9

(SLLAHPQGGG)2L1 11 ± 1.4 9.1 (± 1.1) ! 104 234 ± 8

(SLLAHPQGGGSGG)2L1 8.4 ± 1.7 1.2 (± 0.2) ! 105 331 ± 16

(SLLAHPQGGG)4C1 3.8 ± 0.5 2.6 (± 0.3) ! 105 357 ± 10

(SLLAHPQGGGSGG)4C1 3.7 ± 0.9 2.7 (± 0.7) ! 105 447 ± 24

account for 57% of the molecular weight. Because the binding experiments for the peptides

with longer linkers were done using another streptavidin-coated chip (having a slightly

different immobilization level), the absolute Rmax values for this set of molecules are

somewhat higher, but show the same overall correlation. Figure 12 shows the proposed

binding scheme for mono-, di- and tetravalent peptides.

Figure 12: Schematic representation of the different binding modes for peptide monomer, dimer and

tetramer.

Chapter 4

74

For the monovalent interaction, Kmono is defined as Ka (= Kd–1). The divalent peptide

binds in a two-step mechanism: first, one peptide binds with an association constant Kdi,1 that

is four-fold higher than the monovalent association constant (Eq. [1]). This factor derives

from the four possibilities of binding (two peptides can bind in two binding sites).

!

Kdi,1

= 4 "Kmono

[1]

In the subsequent (intramolecular) binding event, the second peptide fragment binds

to the free binding site with an association constant Kdi,2 (Eq. [2]). The effective molarity EM in

Equation [2] is a measure of affinity enhancement in multivalent interactions, that can be

interpreted as the concentration of free peptide that would give the same amount of complex

formation.[23,24] A statistical factor of 1/2 represents the single association possibility

combined with the two dissociation events possible.

!

Kdi,2

=1

2EM "K

mono [2]

The overall association constant can be expressed as the product of the two

individual association constants (Eq. [3]).

!

Kdi,overall

= Kdi,1"K

di,2= 2 " EM "K

mono

2 [3]

In a similar fashion, the tetravalent peptide binding can be described using a two-step

mechanism. In this case the first binding event displays an 8-fold increase in association

constant relative to the monovalent association constant (Eq. [4], four peptides can bind in

two binding sites). In the second binding step, three peptides can bind to the free site,

whereas dissociation from two sites can occur, resulting in a statistical factor of 3/2 (Eq. [5]).

Again, the overall association constant can be expressed as the product of the two individual

association constants (Eq. [6]).

!

Ktetra,1

= 8 "Kmono

[4]

!

Ktetra,2

=3

2" EM "K

mono [5]

!

Ktetra,overall

= Ktetra,1

"Ktetra,2

=12 " EM "Kmono

2 [6]

From Equations [3] and [6], combined with the experimental Kmono and Koverall values,

the effective molarity EM was calculated for the multivalent peptides (SLLAHPQGGG)2L1,

(SLLAHPQGGG)4C1, (SLLAHPQGGGSGG)2L1 and (SLLAHPQGGGSGG)4C1 (Table 3). The

effective molarities for the tetrameric peptides are slightly lower than their divalent

counterparts. This could indicate that the assumption that all three peptide arms in the

tetramer are equally likely to bind in step 2 of the overall process is possibly incorrect

(meaning that the factor 3/2 in Eq. [5] is incorrect). This can be attributed to the asymmetry in

the dendritic scaffold of the tetramer.


75

Table 3: The calculated effective molarities for peptide multimers (SLLAHPQGGG)2L1,

(SLLAHPQGGG)4C1, (SLLAHPQGGGSGG)2L1 and (SLLAHPQGGGSGG)4C1.

peptide Kd (!M) EM (mM)

(SLLAHPQGGG)2L1 11 0.9

(SLLAHPQGGG)4C1 3.8 0.4

(SLLAHPQGGGSGG)2L1 8.4 1.6

(SLLAHPQGGGSGG)4C1 3.7 0.6

In contrast to the effective molarity, which is used as an empirical measure for affinity

enhancement, the effective concentration Ceff provides a theoretical relation between the

length and flexibility of a linker and the energy of binding. For flexible linkers that display

random-coil behavior, the effective concentration can be calculated using the linker length of

and the distance that needs to be spanned by this linker. It was shown by Zhou that random-

coil loops in proteins can be described by a worm-like chain (WLC) model.[25] Evers et al.

showed recently that the energy transfer between two fluorescent protein domains separated

by a flexible peptide linker could be quantitatively be described by the WLC model, using a

persistence length of 4.5 Å.[26] The effective concentration can be calculated as a function of

the peptide linker length and end-to-end distance re with the WLC model.[27,28] The curves for

different re values are plotted as a function of linker length in Figure 13. From the structures

of our peptide assemblies with a short spacer (i.e. (SLLAHPQGGG)2L1 and

(SLLAHPQGGG)4C1) a linker length of 11 amino acids is estimated. Combining this linker

length with the end-to-end distance between the two binding sites in streptavidin of 30 Å

yields an estimate Ceff of 2 mM.

10-5

10-4

10-3

10-2

Ceff

/ M

50403020100

linker length / aa

30 Å

35 Å

25 Å

20 Å

15 Å

Figure 13: The dependence of the effective concentration Ceff on the linker length and the distance r

that has to be spanned in the protein-peptide complex. Curves were calculated according to the worm-

like chain model with a persistence length of 4.5 Å.[26,28]

Although somewhat lower (at 0.9 and 0.6 mM), the observed effective molarities are

consistent with the prediction of Ceff using this simple model. The peptide assemblies with the

Chapter 4

76

longer spacer, and therefore a longer linker of 17 amino acids, have a calculated effective

concentration that is only slightly higher, at 3.5 mM. This trend is reflected in the actual

effective molarities calculated from the binding experiments, which are higher than the short-

spacer peptides, but still in the same order of magnitude. Overall, the effective

concentrations calculated from the model are somewhat higher than the experimental EM

values, but taking into account the error of measurement, values come close. The curve

corresponding to an re of 30 Å in Figure 13 also shows that increasing the linker length can

only improve the affinity by another factor two, above which the affinity will decrease again.

The effective molarities found for our system are relatively low compared to multivalent

examples from literature, where effective molarities for multivalent biochemical systems of

10–200 mM have been reported, but in these examples either shorter end-to-end distances

on the target or more rigid linkers between the ligands are used.[23,24]

One possibility to increase the affinity considerably is by increasing the valency of the

peptide assembly. For an octavalent peptide dendrimer for instance, our model predicts an

overall Ka of 56 · EM · (Kmono)2. Another important factor is the initial monovalent equilibrium

constant Kmono. Our concept of dendrimer display from phage-display-derived peptides may

perform much better using a system with a higher monovalent affinity for the target. If for

example a divalent peptide ligand can be prepared starting from a peptide with a Kmono of 5

!M instead of 140 !M, a Kdi, overall of 28 nM is predicted using an effective molarity of 0.9 mM,

yielding a 200-fold increase in affinity.

The relatively small effect of increased linker length suggests that other factors, such

as scaffold design and pre-organization of ligands must play a considerable role, as the

phage head displays an extremely high affinity with a pentavalent system. However, the

peptides on the phage may be able to span multiple streptavidin proteins, thereby reaching

more than two binding sites. Therefore a combination of a large, rigid scaffold with the

peptide ligands attached via a flexible linker might be a better phage mimic than our flexible

scaffold and can result in an increased affinity for streptavidin. But this consideration is

specific for a streptavidin-coated surface as a target model; different target receptors will

display a different binding site geometry and valency and therefore other binding modes

need to be considered.

4.6 Conclusion

In this research the first step to the use of synthetic multivalent scaffolds as phage

mimics has been explored, using a streptavidin-binding peptide sequence that was selected

by phage display. Based on this sequence, mono-, di- and tetravalent peptides were

synthesized and binding constants of the peptide assemblies were measured using surface

plasmon resonance. The multivalent peptides displayed a higher affinity for streptavidin

compared to the monovalent peptide but did not come close to the streptavidin-binding

phage. A comparison of the experimental data to a model based on the calculation of


77

effective molarities as a function of linker length and the distance between two adjacent

binding sites explained the multivalent effect observed appropriately and provided an

increased understanding of the factors that determine affinity gain in multivalent peptide

systems, such as Kmono, linker length an receptor distance. Our concept of dendrimer display

from phage display-derived peptides may provide higher affinity gains by an increase of

scaffold valency beyond the tetravalent system discussed here. In addition, starting from a

phage-derived peptide that exhibits a higher monovalent affinity would enhance the

multivalency effect even more.


General

Unless stated otherwise, all solvents (p.a. quality) and other chemicals were obtained from commercial sources and used as received. Water was demineralized prior to use. Hexanediamine (Aldrich), N-Trt-S-Trt-L-cysteine N-hydroxysuccinimidyl ester (NovaBioChem), 4-mercaptophenylacetic acid (Aldrich), tris(carboxyethyl)phosphine hydrochloride (TCEP, Aldrich), p-cresol (Aldrich), bovine serum albumine (BSA, Sigma), biotin (Iris biosciences), streptavidin (Sigma), Ph.D.-7 phage display peptide library kit (New England Biolabs) were all used as supplied. All t-Boc-protected amino acids and MBHA resins (0.92 mmol g–1) were obtained from NovaBiochem. Standard 1H NMR and 13C NMR

experiments were performed on a Varian Gemini-2000 300 MHz spectrometer, a Varian Mercury Vx 400 MHz spectrometer, and a Varian Unity Inova 500 MHz spectrometer at 298 K. Chemical shifts are reported in parts per million relative to tetramethylsilane (TMS). Reversed phase high pressure liquid chromatography (RP-HPLC) was performed on a Varian Pro Star HPLC system coupled to a UV-Vis detector probing at 214 nm using a Vydac protein & peptide C18 column. A gradient of acetonitrile in water (both containing 0.1% TFA) was used to elute products. ESI-MS spectra were recorded on an Applied Biosystems Single Quadrupole Electrospray Ionization Mass Spectrometer API-150EX in positive mode. All SPR experiments were performed on a Biacore T100 instrument (GE Healthcare) using SA type chips. HBS-N buffer (10 mm Hepes, 150 mM NaCl, pH 7.4) was used as running buffer for all experiments. Phage display

Phage selection by solid phase panning Streptavidin (150 !g) was dissolved in 1.5 mL of NaHCO3 (0.1 M, pH 8.6). The solution was transferred to a 60 " 15 mm Petri dish and incubated overnight at 4 °C on a rocking platform. After incubation, the plate was emptied and blocking buffer was added (0.1 M NaHCO3, 0.02% NaN3, BSA (5 mg mL–1), streptavidin (0.1 !g mL–1)). The plate was incubated at 20 °C for 60 min, followed by 6 !

washing with TBST buffer (50 mM Tris, 150 mM NaCl, 0.1% Tween20, pH 7.5). The phage library was diluted in TBST to a final concentration of 2 ! 1011 pfu mL–1 and applied to the streptavidin-coated

surface. The phages were allowed to bind to the surface for 60 min at 20 °C on a rocking platform, after which unbound phages were removed by 10 " washing with TBST. Bound phages were eluted using a solution of 100 !M biotin in TBS (50 mM Tris, 150 mM NaCl, pH 7.5). Phage amplification The eluate containing the phage pool enriched for streptavidin binding was amplified by infection of E. coli. A single colony of E. coli ER2738 (included in the NEB phage display kit) was inoculated in 10 mL

Chapter 4

78

of LB medium (10 g L–1 peptone, 5 g L–1 yeast extract, 5 g L–1 NaCl) supplemented with 20 !g mL–1 tetracycline and incubated overnight at 37 °C, 250 rpm. The overnight culture was diluted 100-fold in 20 mL LB medium and the unamplified eluate was added. After 4.5 h incubation at 37 °C and 250 rpm, the phages were separated from the cells by centrifugation (10 min, 4 °C, 3500 rpm, Sorvall legend RT swing out rotor). The supernatant containing the phages was transferred to a new tube and centrifuged again. 80% of the supernatant was transferred to a new tube and 1/6 volume of 20% (w/v) PEG-8000 (2.5 M NaCl) was added. The phages were allowed to precipitate overnight at 4 °C. The precipitation mixture was centrifuged for 15 min at 4 °C and 3500 rpm, the supernatant was discarded and the tube was briefly centrifuged again. The remaining supernatant was removed by using a pipet. The pellet was suspended in 1 mL of TBS and the suspension was transferred to a microcentrifuge tube and centrifuged (microlitre rotor, Sorvall legend RT) for 10 min at 4 °C and 10,000 rpm to remove remaining cells. The supernatant was transferred to a new tube and 1/6 volume of 20% (w/v) PEG-8000 (2.5 M NaCl) was added. The phages were allowed to precipitate for another hour at 4 °C. The precipitation mixture was centrifuged for 10 min at 4 °C and 3500 rpm, the supernatant was discarded and the tube was briefly centrifuged again. The remaining supernatant was removed by using a pipet. The pellet was suspended in 200 µL of TBS/0.02% NaN3 and centrifuged for 1 min at 4 °C and 10.000

rpm. The supernatant was transferred to a new tube. 200 !L of sterile 60% (v/v) glycerol was added to the amplified phage pool and it was stored at –20 °C. Phage titering A single colony of E. coli ER2738 was inoculated in 5–10 mL of LB and incubated at 37 °C, 250 rpm until mid-log phase (OD600 ~ 0.5, approximately 4 hours). While cells were growing, top agar (10 g L–1 peptone, 5 g L–1 yeast extract, 5 g L–1 NaCl, 7 g L–1 agar and 1 g L–1 MgCl2·6H2O) was melted in a microwave and dispensed into sterile culture tubes, 3 mL per tube, four tubes per phage pool. The

tubes were kept at 45 °C until usage. LB agar plates containing IPTG (40 mg mL–1) and X-gal (40 mg mL–1) were prewarmed at 37 °C. A 10-fold serial dilution of each phage yielding a phage concentration of approximately 5 " 1014 pfu mL–1 pool was prepared in LB. For unamplified pools, dilutions of 101 – 104 were used. For amplified stocks, 109 – 1012 dilutions were used. For every phage pool that was titered, four microcentrifuge tubes were filled with 200 !l of the mid-log phase E. coli culture. From each dilution, 10 !l was added to a tube containing E. coli culture, the tube was vortexed and incubated at room temperature for 1–5 minutes. The infected cells were transferred to a culture tube containing melted top agar. The tube was vortexed quickly and immediately poured onto a prewarmed LB/IPTG/X-gal plate. The top agar was evenly spread by tilting the plate. The plates were allowed to cool for 5 min, inverted, and incubated overnight at 37 °C. Blue plaques were counted on plates that contained ~100 plaques. From this number the titer of each phage pool in plaque forming units per mL (pfu mL–1) was calculated. Isolation of phage clones and DNA isolation The LB agar plate that had been used to titer the phage pool was also used to isolate individual clones within 16 hours after being put in the incubator. A sterile toothpick was used to touch a plaque on the plate to transfer a phage clone to a tube containing 1 mL of a 1:100 diluted overnight E. coli ER2738 culture in LB supplemented with 20 !g mL–1 tetracycline. This was done for 6 clones. The cultures were incubated for 5 h at 37 °C and 200 rpm. The tubes were centrifuged (14000 rpm, 30 s). The supernatant (500 !L), containing the phage, was transferred to a new tube and centrifuged again for 30 s at 14000 rpm. The supernatant was transferred to a new tube and 200 !L of 20% (w/v) PEG-8000/2.5 M NaCl was added. The phages were allowed to precipitate for 20 min at 20 °C. The precipitation mixture was centrifuged for 10 min at 4 °C and 14000 rpm, the supernatant was


79

discarded and the pellet was centrifuged again for 10 min at 4 °C and 14000 rpm. The pellets were suspended in 100 !L iodide buffer (10 mM Tris-HCl (pH 8.0), 1 mM EDTA, 4 M NaI) and 250 !L absolute ethanol. This mixture was incubated at 20 °C for 20 min and then centrifuged for 10 min at 14000 rpm at 4 °C. The supernatant was discarded and 0.5 mL 70% ethanol was added. After centrifuging again for 10 min at 14000 rpm and 4 °C, the supernatant was discarded. After the pellet was dried overnight at 20 °C, 30 !L of demi water was added. 5 !L of the sample was tested on a 1% agarose gel with a 1 kb marker to check the efficiency of DNA extraction. The remaining 25 !L was sent for DNA sequencing by BaseClear, The Netherlands. The –96 glll sequencing primer (5#-CCC TCA TAG TTA GCG TAA CG- 3#) from New England Biolabs that was included in the phage display kit was used. Enzyme-Linked ImmnoSorbent Assay (ELISA) Streptavidin (270 !g) was dissolved in 27 mL TBS. In a high binding 96 well plate (Corning), wells were supplemented with 35 !L streptavidin solution and the plate was incubated overnight at 4 °C on a rocking platform. After incubation, the plate was emptied and 200 !L blocking buffer per well was added (TBS with 2% BSA). The plate was incubated at 37 °C for 60 min, followed by 3 ! washing with

TBST. The phage stock (clone SA 1) was diluted ranging from 105–1011 pfu mL–1 in TBST containing 2% BSA. The phage dilutions were brought in the wells (35 !L per well) and phages were allowed to bind for 2 h at 20 °C. The wells were washed 6 ! with TBST, followed by thorough emptying of the

wells by slapping the plate on a clean towel. Bound phages were detected by incubation with 35 !L of a 1:1000 anti-M13-HRP (Amersham Pharmacia Biotech) in TBS. The plate was rinsed six times with 200 !L of TBST. Finally 70 !L of ABTS (2,2$-Azino- bis(3-ethylbenzthiazoline)-6-sulfonic acid) solution (77.2 mM Na2HPO4, 61.4 mM citric acid, 0.22 mg mL–1 ABTS, 12.4 mM H2O2) was added to each well and the increase in absorbance at 405 nm was determined in time using a platereader (Thermo Multiskan). Competition experiment (ELISA) Streptavidin was coated on a 96 well plate as described above. The phage stock (clone SA 1) was diluted to 2 ! 1010 pfu mL–1 in TBST containing 2% BSA. The phage dilution was mixed with a dilution

range of peptides SLLAHPQGGG, (SLLAHPQGGG)2L1 and (SLLAHPQGGG)4C1 (ranging from 1–1000 !M). The phage/peptide mixtures were brought in the wells (35 !L per well) and were allowed to bind for 2 h at 20 °C. The wells were washed 6 ! with TBST, followed by thorough emptying of the

wells by slapping the plate on a clean towel. Bound phages were detected by incubation with 35 !L of a 1:1000 anti-M13-HRP (Amersham Pharmacia Biotech) in TBS. The plate was rinsed 6 ! with 200 !L

of TBST. Finally 70 !L of ABTS (2,2$-Azino- bis(3-ethylbenzthiazoline)-6-sulfonic acid) solution (77.2 mM Na2HPO4, 61.4 mM citric acid, 0.22 mg mL–1 ABTS, 12.4 mM H2O2) was added to each well and the increase in absorbance at 405 nm was determined in time using a platereader (Thermo Multiskan). Data were fitted to a Hill-equation.

General synthesis of MPAL-activated peptides The general synthesis of MPAL-activated peptides is described in Chapter 2. Streptavidin-Binding HPQ Peptide MPAL Thioester – SLLAH(Dnp)PQGGG-MPAL The synthesis of SLLAH(Dnp)PQGGG-MPAL is described in Chapter 3.

Chapter 4

80

HPQ Peptide MPAL Thioester with spacer – SLLAH(Dnp)PQGGGSGG-MPAL

After the synthesis of MPAL-thioester on the resin, respectively, Boc-Gly, Boc-Gly, Boc-Ser(Bzl), Boc-Gly, Boc-Gly, Boc-Gly, Boc-Gln, Boc-Pro, Boc-His(Dnp), Boc-Ala, Boc-Leu, Boc-Leu and Boc-Ser(Bzl) were coupled. The peptide was cleaved from the resin using liquid HF and was purified using RP-HPLC with a gradient of 30–50% acetonitrile in water over 90 min. The pure peptide thioester was obtained in 54% yield (92 mg). LC-MS: m/z [C62H94N20O22S + 2H]2+ Calcd. 752.3 Da Obsd. 752.3 Da; [M + H]+ Calcd. 1503.7 Da Obsd. 1503.6 Da. Scrambled peptide – SALQLPHGGGC

Respectively, Boc-Cys(4MeBzl), Boc-Gly, Boc-Gly, Boc-His(Dnp), Boc-Pro, Boc-Leu, Boc-Gln, Boc-Leu, Boc-Ala and Boc-Ser(Bzl) were coupled. The peptide was cleaved from the resin using liquid HF and was purified using RP-HPLC with a gradient of 5–25% acetonitrile in water over 60 min. The pure peptide thioester was obtained in 24% yield (9.5 mg). LC-MS: m/z [C43H71N15O13S + 2H]2+ Calcd. 519.8 Da Obsd. 519.7 Da; [M + H]+ Calcd. 1038.5 Da Obsd. 1038.4 Da. SLLAHPQGGG

SLLAH(Dnp)PQGGG-MPAL (23 mg, 17.7 !mol) was dissolved in 0.5 mL Tris buffer (70 mM, pH 8.0) containing 6 M guanidine. %-Mercaptoethanol (4% v/v) was added and the pH was checked (8.0). After

two hours of reaction time at 37 ˚C, the hydrolyzed peptide was purified by preparative RP-HPLC using a gradient of 10–30% of acetonitrile in water over 90 min. The pure peptide with the Dnp group removed and the hydrolyzed thioester was obtained in 60% yield (10 mg, 10.6 !mol). LC-MS: m/z [C40H65N13O13 + 2H]2+ Calcd. 468.7 Da Obsd. 468.7 Da; [M + H]+ Calcd. 936.5 Da Obsd. 936.5 Da.

SLLAHPQGGGSGG

SLLAH(Dnp)PQGGGSGG-MPAL (26 mg, 17.3 !mol) was dissolved in 0.55 mL Tris buffer (70 mM, pH 8.0) containing 6 M guanidine. %-Mercaptoethanol (4% v/v) was added and the pH was checked (8.0). After two hours of reaction time at 37 ˚C, the hydrolyzed peptide was purified by preparative RP-HPLC using a gradient of 13–33% of acetonitrile in water over 90 min. The pure peptide with the Dnp group removed and the hydrolyzed thioester was obtained in 61% yield (12 mg, 10.5 !mol). LC-MS: m/z [C47H76N16O17 + 2H]2+ Calcd. 569.3 Da Obsd. 569.3 Da; [M + H]+ Calcd. 1137.6 Da Obsd. 1137.5 Da. L-cysteine-functionalized hexanediamine L1

Hexanediamine (190 mg, 3.3 mmol primary amine end groups) and N-Trt-S-Trt-L-cysteine N-hydroxysuccinimidyl ester (2.64 g, 3.8 mmol) were dissolved in dichloromethane (33 mL) and triethylamine (0.9 mL) was added. The mixture was stirred at room temperature overnight. After extraction with a saturated Na2CO3 solution (3 " 33 mL) the organic layer was dried over Na2SO4 and filtered. The solvent was evaporated and the crude intermediate was purified by column chromatography (silica; MeOH/DCM, 98/2, v/v). The obtained compound was kept at 0 °C using ice while a mixture of triethylsilane (3 mL), water (0.9 mL) and TFA (44 mL) was added. The mixture was stirred at room temperature for 50 min. After TFA was removed by nitrogen flow the residue was dissolved in water and extracted with diethylether (2 " 50 mL). The water layer was lyophilized and L1 was obtained as a white powder in 69% yield (366 mg, 1.1 mmol).


81

1H NMR (CD3OD, 400 MHz): & = 3.97 (t, 2H, COCHNH2); 3.26 (m, 4H, CH2CH2NH); 2.96 (m, 4H, CHCH2SH); 1.57 (m, 4H, NHCH2CH2(CH2)2CH2CH2NH); 1.40 (m, 4H,NH(CH2)2(CH2)2(CH2)2NH). LC-MS: m/z [C12H26N4O2S2 + 2H]2+ Calcd. 162.1 Da Obsd. 162.1 Da; [M + H]+ Calcd. 323.2 Da Obsd. 323.2 Da; [2M + H]+ Calcd. 645.3 Da Obsd. 644.9 Da. Cysteine-functionalized dendrimer C1

The synthesis of cysteine-dendrimer C1 is described in Chapter 2. (SLLAHPQGGG)2L1

SLLAH(Dnp)PQGGG-MPAL (14.3 mg, 10 !mol) was dissolved in 1.1 mL Tris buffer (70 mM, pH 8) containing 6 M guanidine and cysteine-functionalized hexanediamine L1 (1.6 mg, 4.9 !mol) was added. 4-Mercaptophenylacetic acid (50 mM) and tris(carboxyethyl)phosphine hydrochloride (TCEP, 10 mM) were added and the pH was checked (7). After 3 h incubation at 20 °C the peptide dimer was isolated by preparative RP-HPLC using a gradient of 20–35% acetonitrile in water over 90 min. The pure product was obtained in 55% yield (5.8 mg, 2.7 !mol). LC-MS: m/z [C92H152N30O26S2 + 4H]4+ Calcd. 540.3 Da Obsd. 540.5 Da; [M + 3H]3+ Calcd. 720.0 Da Obsd. 720.5 Da; [M + 2H]2+ Calcd. 1079.6 Da Obsd. 1080.1 Da. (SLLAHPQGGG)4C1

SLLAH(Dnp)PQGGG-MPAL (15.9 mg, 12.2 !mol) was dissolved in 1.2 mL Tris buffer (70 mM, pH 8)

containing 6 M guanidine and cysteine-functionalized dendrimer C1 (2 mg, 2.8 !mol) was added. 4-Mercaptophenylacetic acid (50 mM) and TCEP (10 mM) were added and the pH was checked (6) and adjusted to 7.5 using 20 !L NaOH (1 M). After 3 h incubation at 20 °C the peptide tetramer was isolated by preparative RP-HPLC using a gradient of 22–37% acetonitrile in water over 90 min. The pure product was obtained in 39% yield (4.7 mg, 1.1 !mol). LC-MS: m/z [C188H312N62O52S4 + 8H]8+ Calcd. 550.8 Da Obsd. 551.3 Da; [M + 7H]7+ Calcd. 629.3 Da Obsd. 629.8 Da; [M + 6H]6+ Calcd. 734.1 Da Obsd. 734.7 Da; [M + 5H]5+ Calcd. 880.7 Da Obsd. 881.3 Da; [M + 4H]4+ Calcd. 1100.6 Da Obsd. 1101.4 Da; [M + 3H]3+ Calcd. 1467.1 Da Obsd. 1467.8 Da.

(SLLAHPQGGGSGG)2L1

SLLAH(Dnp)PQGGGSGG-MPAL (17.5 mg, 11.6 !mol) was dissolved in 1.1 mL Tris buffer (70 mM, pH 8) containing 6 M guanidine and cysteine-functionalized hexanediamine L1 (1.6 mg, 4.9 !mol) was added. 4-Mercaptophenylacetic acid (50 mM) and TCEP (10 mM) were added and the pH was checked (7). After 3 h incubation at 20 °C the peptide dimer was isolated by preparative RP-HPLC using a gradient of 20–35% acetonitrile in water over 90 min. The pure product was obtained in 37% yield (4.5 mg, 1.8 !mol). LC-MS: m/z [C106H174N36O34S2 + 4H]4+ Calcd. 640.8 Da Obsd. 641.0 Da; [M + 3H]3+ Calcd. 854.1 Da Obsd. 854.3 Da; [M + 2H]2+ Calcd. 1280.6 Da Obsd. 1280.8 Da.

(SLLAHPQGGGSGG)4C1

SLLAH(Dnp)PQGGGSGG-MPAL (19.7 mg, 13.1 !mol) was dissolved in 1.2 mL Tris buffer (70 mM, pH 8) containing 6 M guanidine and cysteine-functionalized dendrimer C1 (2.2 mg, 3 !mol) was added. 4-Mercaptophenylacetic acid (50 mM) and TCEP (10 mM) were added and the pH was checked (7.2). After 3 h incubation at 20 °C the peptide tetramer was isolated by preparative RP-HPLC using a gradient of 22–37% acetonitrile in water over 90 min. The pure product was obtained in 23% yield (3.6 mg, 0.7 !mol).

Chapter 4

82

LC-MS: m/z [C216H356N74O68S4 + 10H]10+ Calcd. 521.3 Da Obsd. 521.5 Da; [M + 9H]9+ Calcd. 579.1 Da Obsd. 579.5 Da; [M + 8H]8+ Calcd. 651.3 Da Obsd. 651.7 Da; [M + 7H]7+ Calcd. 744.2 Da Obsd. 744.5 Da; [M + 6H]6+ Calcd. 868.1 Da Obsd. 868.3 Da; [M + 5H]5+ Calcd. 1041.5 Da Obsd. 1041.9 Da.

General procedure for Biacore binding experiments

All experiments were performed on a Biacore T100 (GE Healthcare) on SA chips. The running buffer was HBS-N at pH 7.4 at a flow rate of 5 !L min–1 unless otherwise stated. All chips were cleaned and hydrated prior to use by 10 repetitive injections of 3 !L regeneration solution (10 mM NaOH, 0.5 M NaCl). A reference surface was prepared by injection of 350 !L biotin solution (10 !M in HBS-N) at 20 !L min–1, followed by regeneration with 5 injections of regeneration solution (10 mM NaOH, 0.5 M NaCl). Binding experiments were performed in tandem over two channels, one reference channel (blocked with biotin) and one sample channel. Analyte was injected over two channels for 120 s, followed by a 20 min waiting period. Samples were measured in increasing concentration order. Steady state values were fitted to a one site binding model:

!

RU =Rmax

" [ligand]

(KD + [ligand])

4.8 References

[1] Damle, N. K.; Frost, P. Curr. Opin. Pharmacol. 2003, 3, (4), 386-390. [2] Maison, W.; Frangioni, J. V. Angew. Chem., Int. Ed. Engl. 2003, 42, (39), 4726-4728. [3] Arap, W.; Kolonin, M. G.; Trepel, M.; Lahdenranta, J.; Cardo-Vila, M.; Giordano, R. J.; Mintz,

P. J.; Ardelt, P. U.; Yao, V. J.; Vidal, C. I.; Chen, L.; Flamm, A.; Valtanen, H.; Weavind, L. M.; Hicks, M. E.; Pollock, R. E.; Botz, G. H.; Bucana, C. D.; Koivunen, E.; Cahill, D.; Troncoso, P.; Baggerly, K. A.; Pentz, R. D.; Do, K.-A.; Logothetis, C. J.; Pasqualini, R. Nature Med. 2002, 8, (2), 121-127.

[4] Mammen, M.; Choi, S.-K.; Whitesides, G. M. Angew. Chem., Int. Ed. Engl. 1998, 37, (20), 2755-2794.

[5] Scott, J. K.; Smith, G. P. Science 1990, 249, (4967), 386-390. [6] Smith, G. P.; Petrenko, V. A. Chem. Rev. 1997, 97, (2), 391-410. [7] Devlin, J. J.; Panganiban, L. C.; Devlin, P. E. Science 1990, 249, (4967), 404-406. [8] Sidhu, S. S. Biomol. Eng. 2001, 18, (2), 57-63. [9] Chaiet, L.; Wolf, F. J. Arch. Biochem. Biophys. 1964, 106, (1), 1-5. [10] Wilchek, M.; Bayer, E. A., Avidin-Biotin Technology. 1990; 3-45. [11] Sano, T.; Cantor, C. R. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, (1), 142-146. [12] Weber, P. C.; Ohlendorf, D. H.; Wendoloski, J. J.; Salemme, F. R. Science 1989, 243, (4887),

85-88. [13] Green, N. M. Adv. Protein Chem. 1975, 29, 85. [14] Katz, B. A. Biomol. Eng. 1999, 16, (1-4), 57-65. [15] Katz, B. A. Biochemistry 1995, 34, (47), 15421-9. [16] Chang, Y.-P.; Chu, Y.-H. Anal. Biochem. 2005, 340, (1), 74-79. [17] Freitag, S.; Trong, I. L.; Klumb, L.; Stayton, P. S.; Stenkamp, R. E. Protein Sci 1997, 6, (6),

1157-1166. [18] Barbas, C. F.; Burton, D. R.; Scott, J. K.; Silverman, G. J., Phage Display: A Laboratory

Manual. Cold Spring Harbor, 2001. [19] Johnson, E. C. B.; Kent, S. B. H. J. Am. Chem. Soc. 2006, 128, (20), 6640-6646. [20] Getz, E. B.; Xiao, M.; Chakrabarty, T.; Cooke, R.; Selvin, P. R. Anal. Biochem. 1999, 273, (1),

73-80. [21] Lofas, S.; Malmqvist, M.; Ronnberg, I.; Stenberg, E.; Liedberg, B.; Lundstrom, I. Sensors and

Actuators B: Chemical 1991, 5, (1-4), 79-84. [22] Korndorfer, I. P.; Skerra, A. Protein Sci 2002, 11, (4), 883-893.


83

[23] Rao, J.; Lahiri, J.; Weis, R. M.; Whitesides, G. M. J. Am. Chem. Soc. 2000, 122, (12), 2698-2710.

[24] Mulder, A.; Huskens, J.; Reinhoudt, D. N. Org. Biomol. Chem. 2004, 2, (23), 3409-3424. [25] Zhou, H. X. J. Phys. Chem. B 2001, 105, (29), 6763-6766. [26] Evers, T. H.; van Dongen, E. M. W. M.; Faesen, A. C.; Meijer, E. W.; Merkx, M. Biochemistry

2006, 45, (44), 13183-13192. [27] Zhou, H.-X. J. Mol. Biol. 2003, 329, (1), 1-8. [28] van Dongen, E. M. W. M.; Evers, T. H.; Dekkers, L. M.; Meijer, E. W.; Klomp, L. W. J.; Merkx,

M. J. Am. Chem. Soc. 2007, 129, (12), 3494-3495.

Chapter 4

84

85

5

NON-COVALENT SYNTHESIS OF MULTIVALENT

RNASE S ASSEMBLIES

Abstract:

The synthesis of complex biomolecular systems such as multivalent protein

dendrimers using covalent chemistry has proven to be difficult. Therefore, alternative

strategies using non-covalent chemistry are of high interest. This chapter explores the

suitability of using the S-peptide–S-protein interaction to obtain well-defined, stable protein

dendrimers. Association of S-peptide and S-protein results in the formation of an active

enzyme, ribonuclease S, whereas neither fragment alone displays any enzyme activity.

Native chemical ligation was used to couple four S-peptides via their C-terminal thioester to a

cysteine-functionalized dendritic scaffold to yield a tetravalent S-peptide dendrimer. A fully

functional ribonuclease S tetramer was prepared by addition of four equivalents of S-protein.

Different biophysical techniques (ITC, SPR and mass spectrometry), and a fluorescent

enzyme activity assay were used to quantify complex formation. For the non-covalent

synthesis of more complex dendritic architectures, S-protein building blocks are required.

Thioester-modified RNase A was obtained via recombinant expression as a precursor in the

synthesis of multivalent S-protein assemblies. The non-covalent synthetic strategy presented

in this Chapter may be used to synthesize semi-synthetic protein assemblies, which can find

applications in nanomedicine or functional biomaterials.

Chapter 5

86

5.1 Introduction

Semi-synthetic or hybrid molecules consisting of biological macromolecules and

synthetic components offer the opportunity to merge the strengths of biological and chemical

synthetic approaches. Biological synthesis allows the construction of large macromolecules

with high fidelity and the application of evolutionary schemes to optimize functional

properties, but is limited to a relatively small set of building blocks and structures. Synthetic

chemistry is less efficient in the error-free synthesis of large macromolecules, but does

provide access to a much wider variety of functional groups and 3-dimensional topologies.

For example, complex protein assemblies such as those formed by viral coat proteins and

ferritin have been used to construct nano-sized objects that would be difficult to construct via

purely synthetic approaches.[1] Vice versa, synthetic systems could be used as scaffolds to

organize protein domains.[2,3] This approach would allow the construction of complex protein

structures without having to rely solely on protein–protein interactions[4], and would give

access to protein assemblies with non-natural topologies that may be very hard to obtain via

protein engineering approaches.

Dendrimers and dendritic wedges are well-defined, hyperbranched synthetic

molecules that provide attractive scaffolds for multivalent peptides and proteins.[5-7]

Multivalency plays a major role in cell-surface interactions (cell–virus interaction, receptor

clustering), in the immune system (e.g. antibodies are multivalent) and in protein–protein

interactions. In Chapter 2 we developed a general synthetic strategy that allows conjugation

of cysteine-functionalized dendrimers with both oligopeptides and recombinant proteins using

native chemical ligation.[8] While this approach is efficient and high-yielding for the synthesis

of multivalent peptides, the synthesis of protein multimers proved to be more difficult. Here,

we explored the non-covalent synthesis of protein assemblies using the strong interaction

between the 20-residue S-peptide and the S-protein, a small, but very stable folded protein.

Binding has been reported to be strong and tunable by pH, temperature and ionic strength,

and results in the formation of RNase S. RNase S, which was originally obtained after

cleavage of RNase A by subtilisin, is catalytically active, but neither S-peptide or S-protein

alone display enzymatic activity towards RNA (Figure 1). [9,10] An attractive property of the

RNase S system is the availability of a very sensitive fluorescent assay of enzyme activity

that allows the formation of this complex to be studied under very dilute conditions

(nanomolar range).[11,12] In addition, these complexes have been studied with mass

spectrometry and isothermal titration calorimetry. [13,14] The S-peptide and S-protein can be

modified with a C-terminal thioester to provide the building blocks for a variety of

supramolecular multivalent assemblies. This thioester functionality can be reacted with N-

terminal cysteines on multivalent scaffolds (e.g. dendrimers, dendritic wedges, liposomes).


87

Figure 1: Formation of ribonuclease S from cleavage of ribonuclease A by subtilisin. RNase S can be

separated into S-peptide and S-protein.

In this research, our first aim is to synthesize a multivalent S-peptide, and investigate

whether multiple S-proteins can be complexed to form multivalent RNase S enzymes

(Figure 2). A multivalent S-protein would enable the non-covalent display of S-peptides that

are fused to other functionalities (e.g. proteins, peptides, imaging labels). However, the

synthesis of a multivalent S-protein is not straightforward, due to the four disulfide bonds that

need to be formed after expression in bacteria. Finally, the availability of both S-peptide and

S-protein multimers would provide an attractive model system to study multivalent

interactions in detail (Figure 3).

Figure 2: An RNase S tetramer can be prepared from an S-peptide tetramer via multiple monovalent

interactions.

Chapter 5

88

Figure 3: S-protein tetramers can be used as scaffolds to display any S-peptide-fusion protein via

non-covalent interactions. In addition, S-protein and S-peptide tetramers can form RNase S multimers

via multivalent interactions.

5.2 Studying the S-peptide–S-protein interaction

Although the S-peptide–S-protein interaction is generally considered strong, different

binding constants have been reported in literature, based on different analytical techniques.

Schreier and Baldwin described a large pH-dependence of the S-peptide–S-protein affinity

(Kd ranging from 1.1 ! 10–9 M at pH 5.8 to 3.1 ! 10–11 M at pH 8.3; data obtained from

concentration-dependent hydrogen exchange kinetics).[15] However, the affinities reported in

this study deviate from values reported later by Connelly et al (Kd of 1.1 ! 10–7 M at pH 6.0;

measured with ITC).[14] Therefore, before exploring multivalent RNase S assemblies, first the

monovalent interaction between S-peptide and S-protein was investigated using ITC, SPR

and an enzymatic activity assay.

The S-peptide was synthesized using standard Boc-mediated solid phase peptide

synthesis. S-protein was purified from commercial RNase S using preparative separation on

a C4 HPLC column. Characterization of the purified S-protein using mass spectrometry

revealed two different proteins (Figure 4). The predominant species corresponds to residues

21–124 of the initial RNase A, whereas the minor cleavage product corresponds to residues

22–124. This heterolytic activity of subtilisin has been described in literature.[16,17] No further

purification of S20–124 was attempted, as no difference in affinity for S-peptide is expected.


89

a) b) R

ela

tive

ab

un

da

nce

24002200200018001600140012001000m / z

[M+6H]6+

[M+7H]7+

[M+8H]8+

[M+9H]9+ [M+5H]

5+

2307.9

1923.1

1648.8

1442.8

1282.7

Re

lative

ab

un

da

nce

12500120001150011000

mass / amu

S21-124

11447

11534

S22-124

Figure 4: Mass spectrum of S-protein: a) m/z spectrum; b) deconvoluted mass spectrum (calcd. mass

S21–124 11534 Da; S22–124 11447 Da).

Isothermal titration calorimetry (ITC) is a widely used technique to obtain

thermodynamic parameters of biomolecular interactions.[18,19] ITC directly measures the

energy associated with mixing of two components. Figure 5 shows a typical titration

experiment and corresponding binding curve for the interaction between S-peptide and S-

protein. To quantify the pH-dependence of the interaction, ITC binding curves were

measured at three different pH values (pH 6.0, 7.4 and 9.2; Table 1). The stoichiometric ratio

n was determined to be 1.1 S-peptide per S-protein in all experiments (meaning either the S-

peptide concentration was underestimated slightly, or the S-protein concentration was

overestimated somewhat).[20]

a) b)

-0.8

-0.6

-0.4

-0.2

0.0

heat flow

/ !

cal s

-1

20000150001000050000

time / s

-25

-20

-15

-10

-5

0

kca

l /

mo

le o

f in

jecta

nt

2.52.01.51.00.5S-peptide / S-protein molar ratio

Figure 5: Isothermal titration calorimetry experiment: a) Injection of S-peptide (166.7 !M) in a cell

containing S-protein (8.8 !M) in sodium acetate (50 mM, pH 6.0, 25 °C); b) Heats of injection

(corrected for dilution effects) as a function of the S-peptide–S-protein molar ratio. The solid line

represents the fit to a one-site binding model (Ka = 6.9 (± 1.1) ! 106 M–1).

Chapter 5

90

Table 1: Thermodynamic parameters from ITC experiments with S-peptide and S-protein at different

pH values. S-peptide (166.7 !M) was injected into a solution of S-protein (8.8 !M) at 25 °C.

Buffer system Ka (M–1) Kd (M) !H (cal mol–1) !S (cal mol–1 K–1)

NaOAc, pH 6.0 6.9 (± 1.1) ! 106 1.4 (± 0.2) ! 10–7 –2.8 (± 0.4) ! 104 –62.8 ± 0.4

HBS-EP, pH 7.4 1.6 (± 0.2) ! 107 6.3 (± 0.8) ! 10–8 –3.0 (± 0.3) ! 104 –66.2 ± 1.1

TBS, pH 9.2 1.4 (± 0.3) ! 107 7.1 (± 1.5) ! 10–8 –2.9 (± 0.5) ! 104 –65.9 ± 0.9

Table 2: Kinetic parameters from SPR experiments.

Buffer system kon (M–1 s–1) koff (s

–1) Kd (M) Ka (M–1)

MBS-EP, pH 5.8 1.4 ! 106 2.6 ! 10–3 1.9 ! 10–9 5.2 ! 108

HBS-EP, pH 6.8 1.1 ! 106 2.7 ! 10–3 2.6 ! 10–9 3.8 ! 108

HBS-EP, pH 7.4 6.1 ! 105 2.4 ! 10–3 3.9 ! 10–9 2.6 ! 108

TBST, pH 9.2 3.3 ! 105 3.2 ! 10–3 9.7 ! 10–9 1.0 ! 108

The binding constant Ka of 6.9 ! 106 M–1 determined at pH 6.0 is comparable to values

reported before for the interaction of S-protein with S1–15-peptide of 9.0 ! 106 M–1 (using

identical ITC experimental conditions).[14] The values for !H and !S differ somewhat

(Connelly reported a !H value of –3.9 ! 104 cal mol–1 and a !S of –100 cal mol–1 K–1), which

may be due to the different peptide-fragment used in the literature experiment. While we

used the full-length S-peptide with two additional glycine residues at the C-terminus,

Connelly used only the first 15 residues of the S-peptide. Although we did observe a slight

increase in affinity upon increasing pH, the effect is not as dramatic as reported before: only

a two-fold increase of Ka is observed going from pH 6 to pH 9.2, instead of the 100-fold

increase reported previously by Schreier and Baldwin (using concentration-dependent

hydrogen exchange kinetic measurements).[15]

To investigate the S-peptide–S-protein interaction in more detail, surface plasmon

resonance (SPR) was used. SPR is widely used for biomolecular interaction characterization

and enables the determination of both thermodynamic and kinetic parameters.[21] In Chapter

3, an efficient and chemoselective method for peptide and protein immobilization was

described, based on native chemical ligation. Using this strategy, S-peptide was immobilized

on a biosensor chip. A set of four buffers (ranging in pH from 5.8 to 9.2) was used to

investigate the pH dependence of the S-peptide–S-protein interaction. All buffers were

supplemented with 0.05% detergent (either Tween or P20) to ensure a stable resonance

signal. For each data set, a concentration range of S-protein in the corresponding buffer was

prepared and injected for 10 min over the S-peptide modified chip surface. After injection, the

protein was allowed to dissociate for 10 min, followed by a single regeneration step using 10

mM glycine (pH 1.5). All data were globally fitted to a one-to-one binding model, yielding

kinetic parameters kon and koff, and dissociation constants Kd (Figure 6, Table 2). When

comparing the data obtained with identical buffer composition (HBS-EP, pH 7.4), SPR gives


91

a 16-fold lower Kd of 3.9 ! 10–9 M than that determined with ITC (6.3 ! 10–8 M). An

explanation for these differences may be that the peptide is immobilized on a dextran matrix

in the SPR experiment, which might stabilize the formation of the S-peptide helix prior to

binding with S-protein. This would result in an increased kon, giving rise to a lower Kd. SPR

also shows a weak pH-dependence for the affinity, albeit opposite to the trend found with

ITC.

300

250

200

150

100

50

0

Response / R

U

700600500400300200100time / s

7.8 nM

1 !M

Figure 6: Kinetics of S-protein binding to a surface modified with the S-peptide. Fits to experimentally

obtained sensorgrams (subtracted signal Ch2 – Ch1; black) are shown in grey. S-protein

concentration ranged from 7.8 nM to 1 !M in TBS-Tween, pH 9.2, at 70 !L min–1.

Since ITC and SPR yielded different binding constants for the S-peptide–S-protein

interaction, another method to study RNase S formation in solution was devised, based on

the enzymatic activity of RNase S. Kelemen et al. developed a fluorogenic ribonuclease

substrate, optimized for hydrolysis catalyzed by RNase A. The substrate is labeled with two

fluorescent dyes (6-carboxyfluorescein and 6-carboxytetramethylrhodamine). In the intact

substrate, fluorescein fluorescence is quenched by the nearby rhodamine. Upon cleavage of

the substrate by ribonuclease activity, fluorescence is manifested.[11] The initial velocity of

substrate cleavage was used to quantify enzyme activity. A peptide concentration of 40 nM

was chosen for the activity assay. A high salt buffer of pH 9.2 was used to slow down

enzymatic activity sufficient for accurate determination of the initial rate of reaction.[12] First,

S-protein and S-peptide were preincubated for at least 5 minutes to ensure full complex

formation. Addition of 800 nM substrate followed by rapid mixing started the enzymatic

reaction and fluorescence intensity was monitored in time. The initial reaction rate V0

increased linearly with increasing S-protein concentration, until one equivalent of S-protein

per S-peptide was added (Figure 7). No further increase in V0 was observed at higher S-

protein:S-peptide ratio, indicating that full complex-formation was achieved at a one-to-one

ratio of S-peptide and S-protein. The sharp inflection point of the data supports the SPR

experiments in that the Kd must be well below 40 nM; otherwise, a more gradual transition

would be expected.

Chapter 5

92

0.20

0.15

0.10

0.05

0.00

V0 / c

ounts

s-1

160140120100806040200[S-protein] / nM

Figure 7: Enzyme activity of S-peptide (40 nM) in the presence of different amounts of S-protein and

800 nM substrate in 100 mM Tris, 200 mM NaCl, pH 9.2, at 20 °C.

Complex formation was also studied with size exclusion chromatography (SEC) and native

mass spectrometry to obtain additional biophysical evidence for the association of S-peptide

and S-protein to form RNase S. Size exclusion chromatograms were measured using a

Superdex-75 column equilibrated with 200 mM ammonium acetate (pH 6.8). Upon mixing S-

peptide and S-protein in an equimolar ratio, a clear shift in retention time was observed,

indicating quantitative formation of RNase S (Figure 8a). To confirm the presence of RNase

S, the outlet of the SEC column was connected to an electro spray ionization mass

spectrometer (ESI-MS), showing the presence of both RNase S and S-protein (Figure 8b).

a) b)

Inte

nsity

20191817161514time / min

RNase S S-protein S-peptide

Re

lative

Ab

un

da

nce

210020001900180017001600 m / z

S-protein

[M+7H]7+

S-protein

[M+6H]6+

RNase S

[M+7H]7+

RNase S

[M+8H]8+

Figure 8: a) Size exclusion chromatograms (200 mM ammonium acetate, pH 6.8, ELS detection) of S-

peptide (dotted line), S-protein (dashed line) and RNase S (solid line); b) Native mass spectrum of

RNase S, with S-protein and RNase S present.

The free S-protein can in part be attributed to fragmentation of RNase S upon entering the

gas phase, but the signals originating from free S-protein are slightly more abundant at

higher retention times in the peak at 12.2–13.8 min. This shows that although the S-peptide–


93

S-protein interaction is tight and full complexation is observed in the enzymatic assay, the

complex is dynamic and does not survive the column and the transition into the gas phase

fully intact. The heterogeneity of the S-protein is represented in both S-protein and RNase S

peaks.

5.3 Synthesis of a multivalent RNase S dendrimer

Since all experimental data indicate a tight complex, we used the S-peptide–S-protein

interaction to synthesize a multivalent RNase S dendrimer, starting from a tetravalent S-

peptide wedge. The S-peptide was synthesized using standard Boc-mediated peptide

synthesis, as described in Chapter 2. The peptide was modified with a C-terminal thioester to

allow native chemical ligation with a multivalent cysteine-scaffold, and two glycine residues

were inserted to provide additional spacing. Starting from a tetravalent lysine-based dendritic

scaffold modified with cysteine residues, four S-peptides were coupled (Figure 9; this wedge

was a kind gift of Dr T. M. Hackeng, Dr A. Dirksen and W. Adriaens, University of

Maastricht).[22] Initial attempts to ligate four S-peptides to a first generation poly(propylene

imine) dendrimer functionalized with four cysteines were unsuccessful, as only dendrimers

with two or three peptides coupled could be detected. The reason for this inefficient reaction

is unclear, but it could be due to sterical hindrance or an unfavorable low local pH due to the

tertiary amines present in the dendrimer.

a) b)

NH2

O

NH

HN

O

O HN

HN

HN

NH

O

O

O

NH

NH

HN

NH

HN

O

O

O

O

NH2

SH

HN

NH

HN

O

O

NH

O

O

H2N SH

HN

NH

O

NH

HN

NH

HN

O

O

O

O

NH2

SH

HN

H2N

O

O

O

O

HS

Figure 9: a) Synthesis of an S-peptide tetramer; b) structure of the tetravalent lysine wedge containing

four cysteine endgroups, connected via a spacer of glycine residues.

Ligation of four equivalents of S-peptide thioester with one equivalent of lysine-wedge

resulted in full conversion to the S-peptide tetramer in 1 h. Also, complete removal of the

Dnp protecting group on the histidine residues was achieved. The S-wedge was purified

Chapter 5

94

using preparative RP-HPLC and characterized by mass spectrometry (Figure 10). Attempts

to obtain the second generation S-peptide wedge from an octavalent lysine wedge via the

same synthetic strategy were not successful.

a) b)

Re

lative

ab

un

da

nce

200018001600140012001000800m / z

[M+6H]6+

[M+7H]7+

[M+8H]8+

[M+9H]9+

[M+10H]10+

[M+11H]11+

[M+12H]12+

1796.9

1540.0

898.8

1347.7

1198.2

1078.3

980.4

Re

lative

ab

un

da

nce

11200108001040010000

mass / amu

10775

Figure 10: Mass spectrum of the tetravalent S-peptide wedge: a) m/z spectrum; b) deconvoluted

mass spectrum (calcd. mass 10776.7 Da).

ITC was used to study the interaction between the tetravalent S-peptide wedge and

S-protein at different pH values, similar to the study of the single S-peptide–S-protein

interaction. The S-peptide wedge (41.7 "M, corresponding to 166.7 "M S-peptide) was

titrated to a solution of S-protein (8.8 "M) under constant stirring at 25 °C (Figure 11a). The

reverse approach, titrating S-protein to a solution of S-peptide wedge was attempted too, but

the S-protein solution proved to be very viscous at high concentrations. The values of Ka, !H

and n were determined using a one-site binding model with S-peptide concentration instead

of S-peptide wedge concentration (Figure 11b). A stoichiometric ratio n of 0.3 S-peptide

wedge per S-protein (1.2 S-peptide per S-protein) was obtained, which is slightly higher than

the n measured for the single S-peptide–S-protein interaction (n = 1.1).[23] The data fits well to

the one-site binding model; hence, there is no indication for (anti-)cooperative effects. In

Table 3, all thermodynamic parameters are listed for the different buffer compositions,

showing the same pH dependence of Ka as in the monovalent interaction. Overall, affinities

are approximately 2-fold lower than the monovalent interaction, possibly due to a small effect

of sterical hindrance. SPR experiments were not performed with the S-peptide wedge, as no

S-protein-modified chip is currently available.


95

a) b)

-0.8

-0.6

-0.4

-0.2

0.0

heat flow

/ !

cal s

-1

20000150001000050000

time / s

-25

-20

-15

-10

-5

0

kca

l /

mo

le o

f in

jecta

nt

(S-p

ep

tid

e)

0.80.60.40.20.0S-wedge / S-protein molar ratio

Figure 11: Isothermal titration calorimetry experiment: a) Injection of S-peptide wedge (41.7 !M) in a

cell containing S-protein (8.8 !M) in 200 mM sodium acetate (50 mM, pH 6.0, at 25 °C); b) Heats of

injection (corrected for dilution effects) as a function of the S-peptide-wedge:S-protein molar ratio. The

solid line represents the fit to a one-site binding model (Ka = 2.9 (± 0.3) ! 106 M–1).

Table 3: Thermodynamic parameters from ITC experiments with S-peptide wedge and S-protein.

Buffer system Ka (M–1) Kd (M) !H (cal mol–1) !S (cal mol–1 K–1)

NaOAc, pH 6.0 2.9 (± 0.3) ! 106 3.4 (± 0.4) ! 10–7 –2.8 (± 0.4) ! 104 –66.4

HBS-EP, pH 7.4 8.9 (± 0.9) ! 106 1.1 (± 0.1) ! 10–7 –2.7 (± 0.2) ! 104 –59.3

Tris, pH 9.2 8.7 (± 0.8) ! 106 1.2 (± 0.1) ! 10–7 –2.7 (± 0.2) ! 104 –58.0

Ribonuclease S activity was also used to determine the binding stoichiometry of S-

protein with the S-peptide wedge (Figure 12). The initial reaction rate V0 increased with

increasing S-protein concentration, until four equivalents of S-protein per S-wedge were

added. No further increase in V0 is observed at higher S-protein:S-peptide-wedge ratio,

indicating that full complex-formation is achieved at a one-to-four ratio of S-peptide wedge

and S-protein. The sharp inflection point of the data shows that also for the tetravalent

RNase S wedge the Kd must be well below 40 nM. The initial reaction rates for the RNase S

wedge correspond closely to the values reported for the one-to-one interaction (indicated by

the dashed line in Figure 12). This indicates that bringing multiple enzymes close together on

the dendritic scaffold does not hamper enzyme activity. Two small effects must be noted

however: The first data points (1–3 equivalents of S-protein per S-peptide wedge) seem to

deviate negatively from the line that represents the results of the single S-peptide–S-protein

interaction, suggesting negative cooperativity. In contrast, the initial reaction rate at a one-to-

four ratio of S-peptide wedge to S-protein is slightly higher than for the single S-peptide–S-

protein interaction, which suggests a small positive cooperative effect upon binding of four S-

proteins. Although both effects are minor, the exact reason for this behavior is not known,

yet.

Chapter 5

96

0.20

0.15

0.10

0.05

0.00

V0

/ c

ounts

s-1

160140120100806040200

[S-protein] / nM

Figure 12: Enzyme activity of S-wedge (10 nM) in the presence of different amounts of S-protein and

800 nM substrate in 100 mM Tris, 200 mM NaCl, pH 9.2, at 20°C. The dashed line corresponds to the

data of the one-to-one interaction (Figure 6).

5.4 Characterization of the RNase S tetramer using mass spectrometry

To obtain additional biophysical evidence for the formation of RNase S dendrimers,

complex formation was also studied using size exclusion chromatography (SEC) and native

mass spectrometry. Figure 13 shows SEC traces of S-protein and S-peptide wedge in 200

mM ammonium acetate (pH 9.0). Although the S-protein and the S-peptide wedge have

approximately the same molecular weight, their retention times differ significantly due to the

branched nature of the S-peptide wedge. Upon mixing four equivalents of S-protein with one

equivalent of S-peptide wedge, a clear shift in retention time is observed, indicating formation

of higher molecular weight assemblies. The small peak present at higher retention time is not

due to free S-protein, but can be attributed to residual RNase A, which is present in small

amounts in the S-protein sample.


97

Intensity

S-proteinS-wedge

In

tensity

201816141210time / min

Complex

Figure 13: Size exclusion chromatograms (200 mM ammonium acetate, pH 9.0, ELS detection): Top:

S-wedge (dashed line) and S-protein (solid line); Bottom: Mixture of S-wedge and S-protein in a 1:4

ratio.

To characterize the species present in the complex peak at 13 min, the outlet of the

SEC column was connected to an electro spray ionization mass spectrometer (ESI-MS).

Detection in negative mode did not show any complex, however, and only very weak signals

were detected in positive mode at pH 9.0. To improve the signal to noise in positive mode the

buffer pH was lowered to 6.8. Mass spectra were taken throughout the complex peak,

confirming the presence of all possible non-covalent complexes (1 to 4 S-proteins bound to

the S-peptide wedge, ranging in mass from 22 kDa to 57 kDa) as well as free S-protein

(Figure 14). This demonstrates that these large, non-covalent architectures can be detected

and characterized using mass spectrometry. The RNase S tetramer elutes first from the

column at 11.5 min. In the tetramer spectrum, peaks originating from the RNase S trimer can

be observed as well. This is partly due to the fragmentation of tetrameric protein upon

entering the gas phase, but also the different protein complexes are not separated

completely on the size exclusion column, and are therefore eluting in one peak. This

suggests also that although the complex is tight, the complex is dynamic on the exclusion

timescale, since a homogeneous RNase S tetramer peak is expected if a stable complex is

formed. The free S-protein can be attributed to partial fragmentation of the complexes upon

entering the gas phase and is present throughout the entire complex peak. The

heterogeneity of the S-protein is represented in both S-protein and RNase S wedge peaks,

as was also observed for the one-to-one complex.

Chapter 5

98

Figure 14: a) Total ion chromatogram of RNase S wedge (mixture of S-peptide wedge and S-protein

in a 1:4 ratio), indicating the timeframes used for the mass spectra in frames b)–e); b) Mass spectrum

of (S-protein)4S-wedge (solid squares); c) Mass spectrum of (S-protein)3S-wedge (solid triangles); d)

Mass spectrum of (S-protein)2S-wedge (solid double triangles); e) Mass spectrum of (S-protein)1S-

wedge (solid circles).

5.5 Development of building blocks for S-protein assemblies

The availability of both S-peptide and S-protein multimers would provide an attractive

model system to study multivalent interactions in detail. To obtain S-protein tetramers via

native chemical ligation, the thioester-modified S-protein is necessary. Although in theory the


99

S-protein thioester can be obtained via solid phase peptide synthesis, it is a rather large

fragment to synthesize and refolding the S-protein without the S-peptide present is

complicated. Instead, we started with the recombinant expression of thioester-modified

RNase A using the IMPACT expression system.[24] The plasmid containing the gene coding

for the fusion protein of RNase A with an intein and chitin binding domain was a kind gift of

prof. R.T. Raines (University of Wisconsin–Madison).[25]

The plasmid was transfected into E. coli and the RNase A fusion protein was

overexpressed. After purification and cleavage of RNase A-MESNA, the protein was isolated

in a reasonable yield of 2 mg L–1. However, RNase A from bacterial expression needs to be

folded and oxidized. For refolding of proteins, buffers based on the glutathione redox couple

are commonly used. However, the glutathione in the refolding buffer can transthioesterify

with the MESNA thioester, yielding eventually to hydrolyzed thioester. This would render the

RNase A useless for native chemical ligation reactions. Although the refolding of RNase A is

well-studied, the correctly folded RNase A with an active thioester at its C-terminus has not

been reported, since in literature examples the thioester is reacted prior to refolding and

therefore there is no need to preserve the thioester functionality.[25-27]

A reference experiment (using GFP-MESNA as model protein) was used to quantify

the hydrolysis rate of thioester-modified proteins in conventional refolding buffers containing

30 mM reduced glutathione and 10 mM oxidized glutathione. Overnight incubation in this

buffer resulted in > 30% transthioesterification and subsequent hydrolysis. A different

refolding buffer, based on the MESNA redox couple was devised, to prevent this enhanced

hydrolysis rate during refolding. Oxidized MESNA was prepared according to literature

procedure.[28] Incubating GFP-MESNA in a MESNA refolding buffer containing 30 mM

reduced MESNA and 10 mM oxidized MESNA showed no detectable hydrolysis of the

thioester.

RNase A was refolded using both the MESNA refolding buffer and the glutathione-

based buffer. An enzymatic activity assay similar to the assay described in section 5.2 was

used to assess folding efficiency. Refolding experiments were initially performed with the

commercial RNase A, after unfolding the RNase under reducing and denaturing conditions.

The unfolded RNase A was subsequently diluted into refolding buffer containing either the

glutathione or MESNA redox couple and stirred overnight. Enzyme activity measurement

yielded a refolding efficiency of 56% for the MESNA redox pair and 68% for the glutathione

redox pair, making the MESNA refolding protocol an acceptable substitute (Figure 15a). We

reasoned that the MESNA refolding buffer could also be applied to the unfolded protein

during the MESNA-catalyzed on-column cleavage. Typically, 50–75 mM of reduced MESNA

is used to induce cleavage of the target protein on the column. However, for protein refolding,

redox buffer concentrations of 1–5 mM give the best results. Therefore different

concentrations of refolding buffer were tested for on-column cleavage and refolding of RNase

A fusion protein. A cleavage and refolding buffer containing 30 mM reduced MESNA and 10

Chapter 5

100

mM oxidized MESNA gave the best yield of folded RNase A-MESNA (~ 0.6 mg active protein

per liter culture).

a) b)

30

25

20

15

10

5

Re

lative

In

ten

sity

302520151050time / s

3 mM GSH1 mM GSSG

3 mM red. MESNA1 mM ox. MESNA

no refolding

Refolding in solution

80

60

40

20

0R

ela

tive Inte

nsity

302520151050time / s

30 mM red. MESNA, 10 mM ox. MESNA

75 mM red. MESNA

Refolding on column

Figure 15: a) Enzyme activity of unfolded RNase A and of RNase A after refolding in either

glutathione or MESNA refolding buffer (100 mM Tris, 100 mM NaCl; pH 8.0; 20 °C); b) Initial enzyme

activity of cleaved RNase A-MESNA with and without oxidized MESNA present in the cleavage buffer

(50 mM MOPS, 500 mM NaCl, 0.1 mM EDTA, pH 6.0, at 20 °C).

The RNase A that was refolded on the chitin-column was reacted with a divalent

cysteine linker L1 to check whether the thioester was present and reactive in native chemical

ligation (Figure 16a). RNase A thioester (5.3 "M) was incubated with linker L1 (2.4 "M) and

the ligation reaction was monitored using SDS-PAGE. Analysis of the reaction after 20 h

showed the presence of 3 distinct bands in SDS-PAGE, at 14 kDa (RNase A), 28 kDa

((RNase A)2L1) and a high molecular weight band that may correspond to larger aggregates

of RNase A (Figure 16b). This shows that the protein thioester is present and reactive, and

multivalent RNase A assemblies are within reach. The low conversion to protein dimer may

be due to the relatively low concentration of protein thioester used. We are currently working

on improving the ligation efficiency by using higher concentrations of both protein and

cysteine-linker, and are investigating the conversion of the RNase A dimer to an S-protein

dimer via enzymatic cleavage using subtilisin.


101

a) b)

H2NNH

SH

OHN

O

NH2

HS

L1

Figure 16: a) Structure of divalent cysteine linker L1; b) SDS-PAGE analysis of ligation (15%,

reducing conditions): RNase A-MESNA (lane 1), ligation mixture after 20 h incubation in 100 mM Tris,

100 mM NaCl, 50 mM MPAA, pH 7.5, at 20 °C (lane 2).

5.6 Conclusion and outlook

In conclusion, we have demonstrated an efficient non-covalent method to synthesize

hybrid molecules using the S-peptide S-protein interaction. Isothermal titration calorimetry

and surface plasmon resonance were used to study the S-peptide–S-protein interaction at

different pH values and an enzyme activity assay was used to further study complex

formation. The enzymatic activity assay supported the dissociation constant obtained from

SPR (4 nM at pH 7.4), though ITC experiments gave a 16-fold higher Kd value. This

inconsistency was not investigated further. A tetravalent RNase S dendrimer was prepared

by first synthesizing an S-peptide tetramer via native chemical ligation, followed by binding of

four S-proteins. The enzymatic activity assay revealed a clear one to four binding

stoichiometry, while SEC-MS analysis confirmed formation of the tetravalent complex in a

more direct analysis. Reversing our approach by immobilizing the S-protein onto the scaffold

and binding the S-peptide fragment in a non-covalent manner would enable the binding of S-

peptide fusion proteins, in that way providing a much more general method to synthesize

multivalent protein biohybrids. We started with the recombinant expression of thioester-

modified RNase A, which was refolded and reacted with a divalent scaffold using native

chemical ligation. Future research will focus on the conversion of RNase A multimers to S-

protein multimers via an enzymatic hydrolysis with subtilisin and the synthesis of more

complex non-covalent assemblies based on the S-peptide–S-protein interaction.

Chapter 5

102


General

Water was deionized prior to use. Ribonuclease S (grade XII-S, Sigma), Ribonuclease A (Sigma)

thiophenol (Fluka), benzyl mercaptan (Aldrich), 2-mercaptoethanol (Aldrich), 4-mercaptophenylacetic

acid (Aldrich), 6-FAM-dArUdAdA-6-TAMRA (Integrated DNA Technologies), diethylpyrocarbonate

(Aldrich) were used as received. All t-Boc-protected amino acids were obtained from NovaBiochem.

Reversed phase high pressure liquid chromatography (RP-HPLC) was performed on a Varian Pro Star

HPLC system coupled to a UV-Vis detector probing at 214 nm using a Vydac protein & peptide C18

column. A gradient of acetonitrile in water (both containing 0.1% TFA) was used to elute products.

ESI-MS spectra were recorded on an Applied Biosystems Single Quadrupole Electrospray Ionization

Mass Spectrometer API-150EX in positive mode. Native mass spectrometry was performed in positive

mode using Micromass Q-TOF Ultima Global Mass Spectrometer. Micromass MaxEnt 1 software was

used for deconvolution of m/z spectra. All reported calculated masses are of fully reduced species.

Fluorescence spectroscopy was performed on a Varian Cary Eclipse spectrometer. Isothermal titration

calorimetry was performed using a VP-ITC (MicroCal). SPR experiments were performed on a Biacore

T100 instrument (GE Healthcare) using CM5 type chips.

KETAAAKFERQH(Dnp)MDSSTSAAGG-MPAL

The synthesis of KETAAAKFERQH(Dnp)MDSSTSAAGG-MPAL is described in Chapter 3.

KETAAAKFERQHMDSSTSAAGG

KETAAAKFERQH(Dnp)MDSSTSAAGG-MPAL (100 mg, 38 "mol) was dissolved in 10 mL Tris buffer

(70 mM, pH 8.0) containing 6 M guanidine. #-Mercaptoethanol (1% v/v) was added and the pH was

checked (8.0). After two hours of reaction time at 37 °C, the hydrolyzed peptide was purified by

preparative RP-HPLC using a gradient of 10–30% of acetonitrile in water over 90 min. The pure

peptide with the Dnp group removed and the hydrolyzed thioester was obtained in 49% yield (43.3 mg,

19 "mol). LC-MS: m/z [C93H150N30O35S1 + H]+ Calcd. 2280.5 Da Obsd. 2279.9 Da.

Purification of S-protein

The purification of S-protein from Ribonuclease S is described in Chapter 3.

Cysteine-functionalized lysine-wedge

The cysteine-functionalized lysine wedge was a kind gift from Dr T. M. Hackeng, Dr A. Dirksen and W.

Adriaens.[22]

(KETAAAKFERQHMDSSTSAAGG)4–L-lysine dendritic wedge

KETAAAKFERQH(Dnp)MDSSTSAAGG-MPAL (25.4 mg, 9.6 "mol) was dissolved in 3 mL Tris buffer

(70 mM, pH 8.0) containing 6 M guanidine and the L-lysine dendritic wedge (4.1 mg, 2.4 "mol) was

added. Thiophenol (1% v/v) and benzyl mercaptan (1% v/v) were added and the pH was checked

(7.5). After 4 hours at 37 °C the reaction was complete. Thiols were extracted by hexane and the S-

peptide tetramer was purified by RP-HPLC using a gradient of 10–30% acetonitrile in water over 90

min. The pure tetramer was obtained in 20% yield (5.3 mg, 0.49 "mol).

LC-MS: deconvoluted mass [C434H699N147O159S8] Calcd. 10776.7 Da; Obsd. 10775 Da.


103

Isothermal titration calorimetry

The calorimetric experiments were performed using a VP-ITC (MicroCal) at 25 °C. The reference cell

was filled with degassed buffer, and the S-protein (8.8 "M) was transferred to the sample cell. Either

S-peptide (166.7 "M) or S-wedge (41.7 "M) was used to fill the syringe. After an initial equilibration

period of 10 min, stirring at 300 rpm was begun, followed by another equilibration period of 20 min. S-

peptide was titrated to the S-protein solution in 40 ! 5 "L injections at 10 min intervals. Control

injections of peptide and wedge into buffer and buffer into protein were performed and were

neglectable. HBS-EP was prepared freshly from commercially available 10 ! HBS-EP (GE Healthcare)

by diluting 1:9 with demiwater and subsequent filtering. The resulting buffer of pH 7.4 was used

directly. NaOAc (50 mM NaOAc, 100 mM NaCl, pH 6.0) and Tris (100 mM Tris-HCl, 200 mM NaCl, pH

9.2) were prepared freshly and filtered.

Surface plasmon resonance

The immobilization of S-peptide on a cysteine-functionalized SPR chip is described in Chapter 3.

S-Protein binding

S-Protein was dissolved in buffer and diluted into the proper concentrations for kinetic binding

experiments (1 "M to 7.8 nM). A kinetic experiment was set up using the Biacore T100 method

builder. A flow rate of 70 "L min–1 with an association phase of 300 s, a dissociation phase of 300 s,

and a single regeneration step of 10 mM glycine at pH 1.5 were employed. Kinetic data were fitted to a

standard 1:1 binding model using BIAevaluation software. HBS-EP was prepared freshly from

commercially available 10 ! HBS-EP (GE Healthcare) by diluting 1:9 with demiwater and subsequent

filtering. The resulting buffer of pH 7.4 was used directly, or adjusted to pH 6.8 using 1 M HCl. MBS-

EP (10 mM MES, 150 mM NaCl, 3 mM EDTA, 0.05% v/v Tween, pH 5.8) and Tris-Tween (100 mM

Tris, 200 mM NaCl, 0.05% Tween, pH 9.2) were prepared freshly and filtered.

Enzymatic activity assay RNase S

Diethylpyrocarbonate (0.1%, 100 "L) was dissolved in 100 mL demi-water and incubated overnight on

a rocking platform. Subsequent autoclaving of the solution resulted in ribonuclease-free water. A stock

solution of 6-FAM-dArUdAdA-6-TAMRA in ribonuclease-free water (20 "M) was dispended into small

aliquots and stored at –20 °C. All assays were performed in assay buffer (100 mM Tris-HCl, 200 mM

NaCl, pH 9.2) at 20 °C. Prior to use, S-protein was preincubated with either S-peptide of S-peptide

wedge in assay buffer (minimum 5 min). Upon addition of substrate, fluorescence emission was

monitored in time at 515 nm with excitation at 490 nm. Initial velocity V0 was calculated from a linear fit

of the emission intensity in time during the first 200 s. Throughout all experiments, substrate

concentration was kept constant (800 nM). Also, S-peptide concentration (40 nM) and S-peptide

wedge concentration (10 nM) were the same in all experiments.

Size exclusion chromatography

Size exclusion chromatography was performed using a Superdex 75 column (GE Biosciences)

equilibrated with 200 mM ammonium acetate (pH 6.8 or 9.0). A PL-ELS 2100 evaporative light

scattering detector (Polymer Laboratories) was used with a gas flow rate of 0.8 L min–1 at 50 °C. S-

peptide wedge (100 "M), S-protein (400 "M) and a one to four mixture of S-peptide wedge and S-

protein (100 "M and 400 "M respectively) were injected onto the column equilibrated at either pH 9 or

pH 6.8. A mixture of S-peptide wedge and S-protein (100 "M and 400 "M respectively) was prepared

in ammonium acetate buffer, and injected onto the equilibrated column.

Chapter 5

104

Mass spectrometry

Native electrospray mass spectrometry was performed using a Q-TOF Ultima global mass

spectrometer (Micromass) equipped with a Z-spray source. The flow outlet of the Superdex 75 size

exclusion column was connected to the inlet of the mass spectrometer and the ammonium acetate

buffer (200 mM, pH 6.8) was pumped with a Shimadzu LC-10ADvp at a flow rate of 50 "L min–1.

Electrospray ionization was achieved in the positive ion mode by application of 3.5 kV on the needle.

The source block temperature was maintained at 60 °C and the desolvation gas was heated to 120

°C. Collisional cooling was used to maintain the non-covalent complex in the gas phase.[29] This was

achieved by increasing the pressure in the first stage to 4.5 mbar.

A mixture of S-peptide wedge and S-protein (100 "M and 400 "M respectively) was prepared in

ammonium acetate buffer at pH 6.8, and injected onto the column equilibrated at pH 6.8.

In Table S1 the calculated and measured m/z values of the different RNase S dendrimers are

summarized. For all RNase S dendrimers (one to four S-proteins complexed to the wedge), m/z is

calculated using the most abundant S-protein sequence (S21–S124; 11534.6 Da).

Table S1: Calculated and measured masses for the different RNase S dendrimers.

compound charge m/z (calc.) m/z (meas.)

(S-protein)1S-wedge 8+ 2789.7 2789.4

9+ 2479.8 2481.0

10+ 2231.9 2229.2

(S-protein)2S-wedge 10+ 3385.3 3386.1

11+ 3077.7 3077.9

12+ 2821.3 2822.0

(S-protein)3S-wedge 11+ 4126.2 4126.1

12+ 3782.4 3783.3

13+ 3491.6 3491.5

14+ 3242.2 3242.0

(S-protein)4S-wedge 13+ 4378.8 4372.1

14+ 4066.1 4060.2

15+ 3795.1 3795.3

16+ 3558.0 3558.0

17+ 3348.7 3385.2

Ribonuclease A-MESNA

Protein expression and purification

The plasmid coding for RNase A–intein–CBD fusion protein, kindly provided by prof. R. Raines

(University of Wisconsin, Madison), was transformed into chemically competent E. coli BL21(DE3)

and plated on LB agar plates containing 100 mg L–1 ampicillin. Single colonies were used to inoculate

2 mL LB medium containing 100 mg L–1 ampicillin. Cultures were incubated overnight at 37 °C and

subsequently used to start 200 mL culture containing 100 mg L–1 ampicillin. At OD600 = 0.5 the

temperature was lowered to 15 °C and 0.3 mM IPTG was added to induce expression of the target

protein. Cells were collected after overnight expression at 15 °C and 250 rpm by centrifugation,

resuspended in BugBuster (Novagen) lysis buffer and incubated for 20 minutes at 20 °C. A clarified

cell extract was obtained by centrifugation at 40000 $ g for 45 min. The supernatant was loaded onto

a 10 mL chitin column (New England Biolabs) that was equilibrated with 20 mM MOPS NaOH, 0.5 M

NaCl, 0.1 mM EDTA, pH 6.8 (column buffer). The column was washed with 10 volumes (100 mL) of

column buffer to remove non- and weak binding proteins. Subsequently, 3 volumes (30 mL) of

cleavage buffer (50 mM MOPS NaOH, 0.5 M NaCl, 0.1 mM EDTA, 50 mM MESNA, pH 6.0) were

flushed quickly through the column. After overnight incubation of the column at 20 °C, the MESNA

thioester of RNase A was eluted from the column using 1 volume of cleavage buffer. SDS-PAGE

analyis of the eluted protein showed a single band at ~14 kDa.


105


On-column refolding and protein purification

A 10 mL chitin column was equilibrated with 10 volumes (100 mL) of column buffer. The column was

loaded with cell extract and washed with 10 volumes (100 mL) of column buffer. Subsequently, 2.5

volumes of refolding/cleavage buffer (50 mM MOPS NaOH, 0.5 M NaCl, 0.1 mM EDTA, 30 mM

reduced MESNA, 10 mM oxidized MESNA, pH 6.0) were flushed quickly through the column. The

column was incubated for three days at room temperature and the product was eluted from the column

with 10 volumes of cleavage buffer (50 mM MOPS NaOH, 0.5 M NaCl, 0.1 mM EDTA, pH 6).

Fractions were analyzed by SDS-PAGE and enzyme activity assay, giving an estimate yield of 0.6 mg

L–1 active RNase A. LC-MS analysis was not possible due to low ionization.

GFP-MESNA

The expression and purification of GFP-MESNA is described in Chapter 2.

GFP thioester exchange

GFP-MESNA (1 "M) was incubated in either glutathione refolding buffer (50 "L; 0.1 M Tris·HCl, 0.1 M

NaCl, pH 8 containing 3.3 mM reduced glutathione and 1.1 mM oxidized glutathione) or MESNA

refolding buffer (50 "L; 0.1 M Tris·HCl, 0.1 M NaCl, pH 8, containing 3.3 mM MESNA and 1.1 mM

diMESNA) After 8 h incubation at 4 °C, samples were characterized using LC-MS.

2,2-dithiobis(ethanesulfonate) (oxidized MESNA)

Oxidized MESNA was synthesized according to literature procedure.[28]

Enzymatic activity assay RNase A

RNase A was dissolved in assay buffer (0.1 M Tris, 0.1 M NaCl, pH 8.0) to a 40 "M stock. Upon

addition of substrate (400 nM), fluorescence emission was monitored in time at 515 nm with excitation

at 490 nm. The initial reaction rate V0 was calculated from a linear fit of the emission intensity in time

over the first 30 seconds. A concentration range of RNase A was prepared and measured to obtain a

calibration curve (1–10 nM; dilutions were freshly made before each measurement).

Unfolding of RNase A

RNase A (0.465 mg, 33.9 nmol) was dissolved in 1 mL 4 M Gu·HCl, 100 mM TCEP. The mixture was

incubated overnight at 20 °C. Unfolding resulted in an increase of 8 Da from the 8 reduced cysteine

residues and was characterized using LC-MS.

ESI-MS: deconvoluted mass (reduced) [C575H909N171O193S12] Calcd. 13690 Da; Obsd. 13689 Da.

Refolding by dilution, glutathione vs MESNA redox couple

Unfolded RNase A (500 "L) was added in 20 "L aliquots to 40 mL of refolding buffer (100 mM

Tris·HCl, 100 mM NaCl; pH 8.0) containing either 3 mM reduced glutathione and 1 mM oxidized

glutathione, or 3 mM MESNA and 1 mM diMESNA, and stirred overnight at room temperature.

Refolding efficiencies were measured using the enzymatic activity assay at 5 nM initial RNase A

concentration.

L-cysteine-functionalized hexanediamine L1

The synthesis of cysteine-linker L1 is described in Chapter 4.

Chapter 5

106

(RNase A)2L1

Refolded RNase A-MESNA was diluted to an approximate concentration of 5.3 "M in a buffer

containing 100 mM Tris·HCl, 100 mM NaCl and MPAA (50 mM). L1 (2.4 "M) was added and the

reaction mixture was incubated at 20 °C for 20 h. The reaction mixture was analyzed using SDS-

PAGE.

5.8 References

[1] Merzlyak, A.; Lee, S.-W. Curr. Opin. Chem. Biol. 2006, 10, (3), 246-252.

[2] Fournel, S.; Wieckowski, S.; Sun, W.; Trouche, N.; Dumortier, H.; Bianco, A.; Chaloin, O.;

Habib, M.; Peter, J.-C.; Schneider, P.; Vray, B.; Toes, R. E.; Offringa, R.; Melief, C. J. M.;

Hoebeke, J.; Guichard, G. Nat. Chem. Biol. 2005, 1, (7), 377-382.

[3] Matsuura, K.; Murasato, K.; Kimizuka, N. J. Am. Chem. Soc. 2005, 127, (29), 10148-10149.

[4] Tsai, C.-J.; Zheng, J.; Aleman, C.; Nussinov, R. Trends Biotechnol. 2006, 24, (10), 449-454.




[8] van Baal, I.; Malda, H.; Synowsky, S. A.; van Dongen, J. L. J.; Hackeng, T. M.; Merkx, M.;

Meijer, E. W. Angew. Chem., Int. Ed. Engl. 2005, 44, (32), 5052-5057.

[9] Richards, F. M.; Vithayathil, P. J. J. Biol. Chem. 1959, 234, 1459-65.

[10] Raines, R. T. Chem. Rev. 1998, 98, (3), 1045-1065.

[11] Kelemen, B. R.; Klink, T. A.; Behlke, M. A.; Eubanks, S. R.; Leland, P. A.; Raines, R. T.

Nucleic Acids Res. 1999, 27, (18), 3696-3701.

[12] Park, C.; Kelemen, B. R.; Klink, T. A.; Sweeney, R. Y.; Behlke, M. A.; Eubanks, S. R.; Raines,

R. T. Methods Enzymol. 2001, 341, (Ribonucleases, Part A), 81-94.

[13] Loo, R. R. O.; Goodlett, D. R.; Smith, R. D.; Loo, J. A. J. Am. Chem. Soc. 1993, 115, (10),

4391-2.

[14] Connelly, P. R.; Varadarajan, R.; Sturtevant, J. M.; Richards, F. M. Biochemistry 1990, 29,

(25), 6108-14.

[15] Schreier, A. A.; Baldwin, R. L. J. Mol. Biol. 1976, 105, (3), 409-426.

[16] Doscher, M. S.; Hirs, C. H. Biochemistry 1967, 6, (1), 304-12.

[17] Mendez, T. J.; Johnson, J. V.; Richardson, D. E. Anal. Biochem. 2000, 279, (1), 114-118.

[18] Wiseman, T.; Williston, S.; Brandts, J. F.; Lin, L.-N. Anal. Biochem. 1989, 179, (1), 131-137.

[19] Blandamer, M. J.; Cullis, P. M.; Engberts, J. B. F. N. J. Chem. Soc., Faraday Trans. 1998, 94,

(16), 2261-2267.

[20] Concentrations used in further experiments were not corrected for this deviation of

stoichiometric ratio.

[21] Lofas, S.; Malmqvist, M.; Ronnberg, I.; Stenberg, E.; Liedberg, B.; Lundstrom, I. Sensors and Actuators B: Chemical 1991, 5, (1-4), 79-84.


1669.

[23] Concentrations used in further experiments were not corrected for this deviation of

stoichiometric ratio.


[25] Kalia, J.; Raines, R. T. ChemBioChem 2006, 7, (9), 1375-1383.

[26] Arnold, U.; Hinderaker, M. P.; Raines, R. T. TheScientificWorld 2002, 2, 1823-1827.

[27] delCardayre, S. B.; Ribo, M.; Yokel, E. M.; Quirk, D. J.; Rutter, W. J.; Raines, R. T. Protein Eng. 1995, 8, (3), 261-73.

[28] Brzezinska, E.; Ternay, A. L. J. Org. Chem. 1994, 59, (26), 8239-8244.

[29] Chernushevich, I. V.; Thomson, B. A. Anal. Chem. 2004, 76, (6), 1754-60.

108

SUMMARY

Multiple, simultaneous interactions are often used in biology to enhance the affinity

and specificity of binding, an effect referred to as multivalency. This multivalency can be

mimicked by anchoring multiple peptides and proteins onto synthetic dendritic scaffolds. The

aim of this research was to develop general methods to obtain well-defined protein and

peptide assemblies, and to study multivalent interactions of these assemblies in a controlled

fashion.

In Chapter 2, a general synthetic strategy is described to obtain multivalent peptides

and proteins using native chemical ligation. Different generations of poly(propylene imine)

dendrimers were functionalized with N-terminal cysteine residues to allow the native

chemical ligation reaction with C-terminal thioesters. Ligation of a peptide thioester with

cysteine-functionalized dendrimers yielded multivalent peptide dendrimers of different

generations with 4 to 16 peptides per dendrimer. This native chemical ligation strategy was

expanded to recombinant proteins by employing intein-mediated protein expression and

purification to obtain fluorescent proteins modified with a C-terminal thioester. Native

chemical ligation of GFP-MESNA with a cysteine-modified dendrimer followed by ligation with

peptide thioesters gives access to novel hybrid peptide-protein dendrimers. Ligation of 4

equivalents of GFP-MESNA with the cysteine-modified dendrimer yielded a branched,

multivalent protein tetramer. Size exclusion chromatography combined with mass

spectrometry proved to be an invaluable tool to study these complex bio-macromolecules.

The use of surface plasmon resonance (SPR) biosensors enables real-time detection

and monitoring of biomolecular binding events. Chapter 3 describes a chemoselective

immobilization strategy for Biacore SPR sensor chips, based on native chemical ligation.

First, a thioproline was introduced on the surface, which could be deprotected using mild

conditions to an N-terminal cysteine residue. A streptavidin-binding peptide was immobilized

via its C-terminus onto the biosensor chip, and subsequent binding experiments with

streptavidin showed specific and reproducible binding to the peptide surface. Short ligation

steps of peptide thioester were alternated with streptavidin binding experiments on a single

chip. This provided an increased peptide loading after each ligation step, yielding enhanced

protein-binding capacity. As an example of a recombinant protein, green fluorescent protein

(GFP) was immobilized on the biosensor surface. Again, binding experiments with an

antibody directed against GFP showed the specificity and robustness of the coupling

strategy. The immobilization of S-peptide via native chemical ligation was used to illustrate

the possibility of obtaining kinetic information from the specific interaction between S-peptide

and S-protein. The presented approach allows for efficient immobilization of both

recombinant proteins and synthetic peptides with high control over the degree of

functionalization of the surface.

Summary

109

In Chapter 4, native chemical ligation was used to synthesize multivalent peptides

based on a streptavidin-binding peptide sequence derived from phage display. The synthetic

multivalent scaffolds were used to mimic the multivalent character of the peptides on the

head of a phage, without the presence of the phagemid coat proteins or genetic information.

Peptides with different valency (from 1–4 copies per scaffold) and spacing were prepared

and their affinity for streptavidin was measured using SPR. All multivalent peptides showed a

significant increase in affinity compared to their monovalent counterpart and a binding model

was used to describe the multivalent effect in a quantitative manner. However, the peptide

dendrimers still showed considerably lower affinity than the streptavidin-binding phage.

Possible reasons for this difference are discussed in this Chapter, as well as suggestions for

further improvement of this dendrimer display by optimization of both scaffold rigidity and

spacing of ligands.

Although the covalent conjugation strategy described in Chapter 2 allowed the

synthesis of tetravalent protein dendrimers of 110 kDa, non-covalent synthetic strategies are

required for the development of even more complex protein assemblies. Chapter 5 explores

the suitability of using the S-peptide–S-protein interaction to obtain well-defined, stable

protein dendrimers. Association of S-peptide and S-protein results in the formation of an

active enzyme, ribonuclease S, whereas neither fragment alone displays any enzyme

activity. Native chemical ligation was used to couple four S-peptides via their C-terminal

thioester to a cysteine-functionalized dendritic scaffold to yield a tetravalent S-peptide

dendrimer. A fully functional ribonuclease S tetramer was prepared by addition of four

equivalents of S-protein. Different biophysical techniques (ITC, SPR and mass

spectrometry), and a fluorescent enzyme activity assay were used to quantify complex

formation. For the non-covalent synthesis of more complex dendritic architectures, S-protein

building blocks are required. Thioester-modified RNase A was obtained via recombinant

expression as a precursor in the synthesis of multivalent S-protein assemblies. This non-

covalent synthetic strategy based on ribonuclease S can be used to synthesize semi-

synthetic protein assemblies such as supramolecular polymers, gels and polymer networks

with high control of structural organization, and may find applications in nanomedicine or

functional biomaterials.

111

CURRICULUM VITAE

Ingrid van Baal werd geboren op 21 januari 1979 te Tilburg. Na het

behalen van haar VWO-diploma aan het Pauluslyceum te Tilburg begon

ze aan de studie Scheikundige Technologie aan de Technische

Universiteit Eindhoven. Deze studie werd in 2003 afgerond met een

afstudeerproject in de vakgroep Macromoleculaire en Organische

Chemie onder leiding van prof. dr. E.W. Meijer. Dit project vormde de

aanzet tot het promotieonderzoek wat in dezelfde vakgroep werd

uitgevoerd, onder leiding van dr. M. Merkx en prof. dr. E.W. Meijer. Het

doel van dit onderzoek was het ontwikkelen van een algemene methode

om multivalente peptiden en eiwitten te synthetiseren, om deze vervolgens te gebruiken voor

het bestuderen van multivalente interacties. De belangrijkste resultaten van dit onderzoek

zijn beschreven in dit proefschrift.

Ingrid van Baal was born on January 21, 1979 in Tilburg. After secondary education at the

Pauluslyceum in Tilburg, she started studying Chemical Engineering at the Eindhoven

University of Technology. This study was completed in 2003 with an undergraduate project in

the Laboratory for Macromolecular and Organic Chemistry under guidance of prof. dr. E.W.

Meijer. This research was the starting point for a PhD project in the same laboratory, under

guidance of dr. M. Merkx and prof. dr. E.W. Meijer. In this research project, a general method

for synthesizing multivalent peptides and proteins was developed, and these structures were

used to investigate multivalent interactions. The most important results of this research are

described in this thesis.

112

DANKWOORD

Het zit er op. Na vier mooie jaren als promovendus is het tijd om Eindhoven te

verlaten. Tijdens mijn promotietijd heb ik met veel mensen samengewerkt, en mede dankzij

deze mensen is dit proefschrift geworden tot wat het is.

Veel heb ik te danken aan Bert Meijer. Een betere coach kan ik me niet voorstellen.

Bert, ik ben blij dat je me de kans hebt gegeven om binnen jouw groep aan multivalente

eiwitten en peptiden te kunnen werken, en mij alle kansen hebt gegeven om binnen en

buiten SMO zoveel mogelijk kennis van de chemische biologie te kunnen opdoen. Ik hoop

dan ook dat de nieuwe Chemical Biology groep een groot succes wordt!

Maarten Merkx heeft minstens evenveel bijgedragen aan mijn onderzoek. Vanaf het

moment dat ik tijdens mijn afstudeerproject een keer kwam praten over de mogelijkheid om

eiwitten en synthetische structuren te combineren kreeg mijn onderzoek een geheel nieuwe

dimensie, die we tijdens mijn promotietijd verder verkend hebben. Maarten, ik heb ontzettend

veel van je geleerd, bedankt daarvoor!

I would like to thank prof.dr. H. Waldmann and prof.dr.ir. J.C.M. van Hest for reading

the manuscript and for participating in the committee. Prof.dr. Emmo Meijer, ik heb altijd met

veel plezier met je van gedachten gewisseld over onderzoek in het algemeen en enzymen in

het bijzonder. Ik ben blij dat je bij mijn promotie aanwezig kan zijn. Marcel van Genderen wil

ik bedanken voor de kritische blik op mijn manuscript, en de opbouwende kritiek tijdens vele

werkbesprekingen. Mijn promotiecommissie zou niet compleet zijn zonder de Nederlandse

pionier van de native chemical ligation: Tilman Hackeng. Tilman, ik ben de afgelopen jaren

regelmatig in jouw lab te gast geweest, en heb me daar altijd welkom gevoeld. Bedankt voor

al je hulp de afgelopen jaren.

De metingen aan het GFP-tetrameer zijn uitgevoerd in het lab van prof. A.J.R. Heck

aan de Universiteit Utrecht. Ik wil naast prof. Heck ook Robert van den Heuvel en Silvia

Synowsky bedanken voor de samenwerking. I would like to thank prof. R.T. Raines and Jeet

Kalia from the University of Wisconsin–Madison for fruitful discussions during the Spetses

Chemical Biology Summer School and for providing the RNase A plasmid.

Ik heb het voorrecht gehad om tijdens mijn promotie een groot aantal studenten te

mogen begeleiden: Rachel, Cheng, Rob, Anke, Sanne, Linda, Edith, Sander en Maartje. Ik

wil jullie allemaal bedanken voor jullie inzet en de prettige samenwerking.

Sanne en Edith, ik kan me geen kundigere paranimfen indenken en ben daarom erg

blij dat jullie me bij willen staan. Ik wens jullie allebei heel veel succes met jullie verdere

promotie, de multivalente peptiden en eiwitten zijn bij jullie in goede handen!

Massa spectrometrie heeft een belangrijke rol gespeeld in mijn onderzoek, en daarbij

hebben Joost en Lou me altijd met raad en daad bijgestaan. Voor het phage display

onderzoek, alsmede de goede organisatie van het Bio-lab ben ik Peggy erg dankbaar.

Wencke, eerst in Maastricht en later in Eindhoven stond je altijd klaar om me te helpen met

Dankwoord

113

de peptide synthese en daarvoor ben ik je veel dank verschuldigd. I also would like to thank

Brett for an enjoyable collaboration, which lead to the work described in Chapter 3. Tom wil ik

bedanken voor de goede discussies over multivalentie, en uiteraard ook voor het delen van

zijn kennis van de literatuur op dat gebied. Koen Pieterse wil ik bedanken voor zijn hulp en

creatieve inbreng bij het fabriceren van mooie plaatjes. Verder wil ik alle mensen van de

protein engineering groep en het dendrimeren-team bedanken voor een plezierige

werkomgeving, en daarbij ook de mensen van Lab 2 en mijn kamergenoten betrekken. Joke,

Hanneke en Hans wil ik bedanken voor alle kleine dingen die het leven van een AIO

makkelijker maken.

Naast mooie wetenschap is er de afgelopen jaren altijd tijd geweest voor ontspanning

en gezelligheid, in de vorm van SMO-uitjes, gezamenlijke congresbezoeken, vakanties en

koffie-, thee-, en lunchpauzes. Hiervoor wil ik bedanken: Sagitta, Kelly, Jos, Tom, Suzan,

Sanne, Monica, Jan, Peggy, Wencke, Ronald, Eva, Patricia, Linda, Nicole, Pim, Hitomi,

Daniela, Dirk, Jeroen, Freek, Oren en Bas.

Buiten SMO hebben mijn vrienden en familie altijd veel belangstelling getoond voor

mijn onderzoek en gezorgd voor de nodige ontspanning en afleiding op z!n tijd. Aukje en

Sabine, ik ben blij met zulke goede vriendinnen en kijk alweer uit naar onze komende

skivakantie, natuurlijk ook samen met Jan en Geert. Petra, Harrie, Ruud en Marietje, bedankt

voor jullie interesse in mijn onderzoek en alle gezelligheid.

Tot slot is er niets zo belangrijk als het thuisfront. Mama en Papa, jullie steun en

interesse betekent heel veel voor me, bedankt voor alles. Yvonne, ik vind het keileuk dat je

ook de academische wereld gaat verkennen, en dit combineert met het starten van een

eigen bedrijf. Ik ben trots op je en wens jou en Jeroen veel succes met de toekomst!

De laatste regels zijn uiteraard voor jou, Jeroen. Bedankt voor alles de afgelopen

jaren. Wij samen kunnen de hele wereld aan.

Ingrid

Peptides and proteins in dendritic assemblies · 1.2 Multivalency in peptide and protein chemistry...

Documents

Transcript of Peptides and proteins in dendritic assemblies · 1.2 Multivalency in peptide and protein chemistry...