Stabilizing biocatalysts.pdf

32
6534  Chem. Soc. Rev.,  2013,  42, 6534--6565  This journ al is  c  The Royal Society of Chemistry 2013 Cite this:  Chem. Soc. Rev., 2013, 42, 6534 Stabilizing biocatalysts Andreas S. Bommarius* a and Marie ´tou F. Paye b The area of biocatalysis itself is in rapid development, fueled by both an enhanced repertoire of protein engineering tools and an increasing list of solved problems. Biocatalysts, however, are delicate materials that hover close to the thermodynamic limit of stability. In many cases, they need to be stabilized to survive a range of challenges regarding temperature, pH value, salt type and concentration, co-solvents, as well as shear and surface forces. Biocatalysts may be delicate proteins, however, once stabilized, they are eciently active enzymes. Kinetic stability must be achieved to a level satisfactory for large-scale proc ess appli catio n. Kine tic stability evokes resis tance to degr adati on and maintained or incr eased catalytic eciency of the enzyme in which the desired reaction is accomplished at an increased rate. Howev er, beyon d these limitatio ns, stabl e bioca talys ts can be oper ated at high er temperat ures or co-solvent concentrations, with ensuing reduction in microbial contamination, better solubility, as well as in many cases more favorabl e equi lib ri um, and can serve as more eec ti ve templates for comb inatorial and data -driv en prote in engin eerin g. To incr ease thermodyn amic and kinet ic stabi lity , immo biliz ation , prot ein engin eerin g, and medium engin eerin g of bioca talys ts are available, the main focus of this work. In the case of protein engineering, there are three main approaches to enhancing the stability of protein biocatalysts: (i)  rational design, based on knowledge of the 3D-structure and the catalytic mechanism, (ii)  combinatorial design, requiring a protocol to generate diversity at the genetic level, a large, often high throughput, screening capacity to distinguish ‘hits’ from ‘misses’, and (iii)  data- driven design, fueled by the increased availability of nucleotide and amino acid sequences of equivalent functionality. a School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Parker H. Petit Institute of Bioengineering and Bioscience, 315 Ferst Drive,  Atlanta, GA 30332-0363, USA. E-mail: [email protected]; Fax:  +1 404-894-2291 b School of Chemistry and Biochemistry, Georgia Institute of Technology, 901 Atlantic Drive, Atlanta, GA 30332-0400, USA  Andreas S. Bommarius  Andreas (Andy) S. Bommarius is a pro fes sor of Che mic al and  Biomolecular Engineering as w el l of C he mi st r y a nd B io - chemistry at the Georgia  Institute of Technology in  Atlanta, GA, USA. He received hi s di pl oma in Ch emis tr y in 1984 at the Technical University of Munich , Ge rmany and hi s Chemical Engi ne er ing BS and  PhD degrees in 1982 and 1989 at MIT, Cambridge, MA, USA.  From 1990–2000, he led the  Laboratory of Enzyme Catalysis at Degussa (now Evonik) in Wol fgan g, Ger many. At Geo rgi a Tec h since 2000 , his res ear ch inter ests cover green chemi stry and biomo lecul ar engin eerin g, specifically biocatalyst development and protein stability studies. Marie ´tou F. Paye  Marie ´tou F. Paye obtained her BA de gr ee in Bi oche mi stry an d a minor in French from Middlebury College in Middlebu ry, Vermont, USA. She was awarded a Bill and  Melinda Gates Schola rship for her un de rg ra du ate and gr aduate stud ies . At the Geor gia Inst itut e of Technology in Atlanta, GA, she works toward s her Ph D in the Sc ho ol of Chemistry and Bi o- ch emis tr y in the Bommar ius  group, focusi ng on novel synt heses towards beta -lac tam antibiotics. Recei ved 18th April 2013 DOI: 10.1039/c3cs60137d www.rsc.org/csr Chem Soc Rev REVIEW ARTICLE    P   u    b    l    i   s    h   e    d   o   n    2    7    J   u   n   e    2    0    1    3  .    D   o   w   n    l   o   a    d   e    d    b   y    O   p   e   n    U   n    i   v   e   r   s    i    t   y   o   n    0    8    /    0    7    /    2    0    1    3    1    4   :    1    7   :    3    5  . View Article Online View Journal | View Issue

Transcript of Stabilizing biocatalysts.pdf

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 1/32

6534   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

Cite this:  Chem. Soc. Rev., 2013,

42, 6534

Stabilizing biocatalystsAndreas S. Bommarius*a and Marietou F. Payeb

The area of biocatalysis itself is in rapid development, fueled by both an enhanced repertoire of protein

engineering tools and an increasing list of solved problems. Biocatalysts, however, are delicate materials

that hover close to the thermodynamic limit of stability. In many cases, they need to be stabilized to

survive a range of challenges regarding temperature, pH value, salt type and concentration, co-solvents,

as well as shear and surface forces. Biocatalysts may be delicate proteins, however, once stabilized, they

are efficiently active enzymes. Kinetic stability must be achieved to a level satisfactory for large-scale

process application. Kinetic stability evokes resistance to degradation and maintained or increased

catalytic efficiency of the enzyme in which the desired reaction is accomplished at an increased rate.

However, beyond these limitations, stable biocatalysts can be operated at higher temperatures or

co-solvent concentrations, with ensuing reduction in microbial contamination, better solubility, as well

as in many cases more favorable equilibrium, and can serve as more effective templates for

combinatorial and data-driven protein engineering. To increase thermodynamic and kinetic stability,

immobilization, protein engineering, and medium engineering of biocatalysts are available, the main

focus of this work. In the case of protein engineering, there are three main approaches to enhancing

the stability of protein biocatalysts: (i)  rational design, based on knowledge of the 3D-structure and the

catalytic mechanism, (ii)  combinatorial design, requiring a protocol to generate diversity at the genetic

level, a large, often high throughput, screening capacity to distinguish ‘hits’ from ‘misses’, and (iii)  data-

driven design, fueled by the increased availability of nucleotide and amino acid sequences of equivalent

functionality.

a School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Parker H. Petit Institute of Bioengineering and Bioscience, 315 Ferst Drive,

 Atlanta, GA 30332-0363, USA. E-mail: [email protected]; Fax:  +1 404-894-2291b School of Chemistry and Biochemistry, Georgia Institute of Technology, 901 Atlantic Drive, Atlanta, GA 30332-0400, USA

 Andreas S. Bommarius

 Andreas (Andy) S. Bommarius

is a professor of Chemical and 

 Biomolecular Engineering as

well of Chemistry and Bio-

chemistry at the Georgia

 Institute of Technology in

 Atlanta, GA, USA. He received 

his diploma in Chemistry in1984 at the Technical University

of Munich, Germany and his

Chemical Engineering BS and 

 PhD degrees in 1982 and 1989

at MIT, Cambridge, MA, USA.

 From 1990–2000, he led the

 Laboratory of Enzyme Catalysis at Degussa (now Evonik) in

Wolfgang, Germany. At Georgia Tech since 2000, his research

interests cover green chemistry and biomolecular engineering,

specifically biocatalyst development and protein stability studies.

Marietou F. Paye

 Marietou F. Paye obtained her BA

degree in Biochemistry and a

minor in French from Middlebury

College in Middlebury, Vermont,

USA. She was awarded a Bill and 

 Melinda Gates Scholarship for her 

undergraduate and graduate

studies. At the Georgia Instituteof Technology in Atlanta, GA, she

works towards her PhD in the

School of Chemistry and Bio-

chemistry in the Bommarius

 group, focusing on novel 

syntheses towards beta-lactam

antibiotics.

Received 18th April 2013

DOI: 10.1039/c3cs60137d

www.rsc.org/csr 

Chem Soc Rev 

REVIEW ARTICLE View Article OnlineView Journal | View Issue

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 2/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6535

1. Need for biocatalyst stabilization

 Justification for current review. The area of biocatalysis itself is in

rapid development, fueled by both an enhanced repertoire of 

protein engineering tools (for more reviews, see citations1,2)

and an increasing list of solved problems, such as the synthesis

of the side chains of statins such as Lipitors and Crestors,3 as

 well as the active pharmaceutical ingredients (APIs) of Lyricas

(pregabalin)4 and Januvias/Janumet s (sitagliptin)5 (Fig. 1) thepathway from glucose to 1,3-propanediol,6 and the process to

acrylamide and nicotinamide from their respective nitriles. In

parallel, the subject of stabilizing biocatalysts has developed

alongside and certainly deserves a renewed look. In addition to

this review, readers are referred to the overview of enzyme

stability by Polizzi et al.,7 the focus on stabilization through the

surrounding medium by Bommarius and Broering,8 and lastly 

the works of Eijsink on both rational design9 and directed

evolution for comprehensive reviews on the topic of stabilizing 

biocatalysts.10

 Advantages of biocatalysts.  Biocatalysis, defined as reactions

catalyzed by macromolecules such as isolated enzymes and

 whole cells, has shown to be favorable in several ways.14–16

Biocatalysts are able to catalyze diverse sets of reactions includ-

ing: (a) formation or hydrolysis of peptides, esters, and amides

by hydrolases, (b) enantio- and regioselective oxidation of 

alcohols, and reductions of ketones & double bonds by oxido-

reductases, (c) enantio, chemo-, and regioselective reactions by 

transferases and isomerases, (d) and formation of C–C bonds

by lyases.16–20 Essentially, biocatalysis offers industrial-scale

synthesis of compounds within a controllable environment.This biochemical method of synthesis relieves many of the

disadvantages presented in the production of complex macro-

molecules such as lack of reactive specificity and in the use of 

chemocatalysts such as metals during organic synthesis.16,19,21

 Additionally, the shift towards green chemistry, the need for the

use of biodegradable materials, mild reaction conditions,

controlled reactions conditions, and reduced cost and

manpower have rendered biocatalysis increasingly timely in

industrial settings.15,16,19,22,23 Biocatalysis encompasses a wide

range of chemical reactions, including redox reactions and

carbon–carbon formations, as illustrated by the following two

examples.

Fig. 1   (a) Structure of enzymes studied and evolved to catalyze the production of the side chains of pharmaceutically available medicines. 12-Oxophyto-

dienoate reductase—synthesis of Lyricas (pregabalin);11 PDB: 3HGR.12 (b)   D-Amino acid aminotransferase—synthesis of Januvias/Janumets (sitagliptin);5

PDB: 3LQS.13

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 3/32

6536   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

Oxidation of double bonds by cytochrome P45024

Cytochrome P450 from Rattus norvegicus: naturally a dimer; however, only the monomer is shown here. PDB: 1AMO; 25 in black are

loops and coils, in green are helices, and in blue are  b-sheets.24

Carbon–Carbon bond formation by transketolase26

Transketolase from E. coli K-12: naturally a dimer; however, only monomer is shown here. PDB: 2R8O;27 in black are loops and coils,

in green are helices, and in blue are   b-sheets.26 This enzyme catalyzes carbon–carbon bond formation between 3-hydroxy-

2-oxopropanoate and  b/g-hydroxylated aldehyde yielding two enantiomers.

 A prime example for the need of biocatalysis.   Penicillin, a

b-lactam antibiotic discovered by Alexander Fleming in 1928,

 was found to be effective against a range of infections.28  As

propitious as this event presented itself, the organic chemists

of the World War II era considered the total synthesis of 

penicillin to be an endearingly impossible task.28  With the

conglomerate work of 1000 chemists from 39 distinct labora-

tories, the synthesis of Penicillin V was published by John C.

Sheehan in 1957, thirty years after its discovery.22,28  With the

extensive manpower, time, knowledge, and money invested in

the production of penicillin, it became an objective to create

cost and time effective total synthesis of penicillin.22 However,

13 years preceding, the employment of biocatalysis for the

synthesis of penicillin yielded an astonishing 16 million doses

of the drug in the singular year of 1944 (Fig. 2).28 Today,

penicillin and various other b-lactam antibiotics can be synthe-

sized and purified within days by an individual when using 

biocatalysis in lieu of the traditional chemocatalysis.

 Advantages of stabilizing biocatalysts.  Besides the previously 

mentioned benefits with biocatalyst productivity, stable bio-

catalysts can be operated at higher temperatures or co-solvent 

concentrations, with ensuing reduction in microbial contam-

ination, better solubility, and often a more favorable equili-

brium position, and can serve as more effective templates

for combinatorial and data-driven protein engineering, a

point firstly succinctly made by Bloom and Arnold32 and

Fig. 2   Biosynthesis of b-lactam ring from ACV catalyzed by Isopenicillin N Synthase. The bonds highlighted in red make-up the b-lactam nucleus, which is responsible

for antibiotic activity against targeted penicillin binding proteins.29–31

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 4/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6537

recently discussed as a key step in the development of indus-

trial biocatalysts.33

1.1 Enzymes are delicate materials, an issue of 

thermodynamics

 Although enzymes designed for industrial settings are opti-

mized for efficient synthesis of compounds, the route to

obtaining these stable biocatalysts is often challenging since

naturally enzymes have not evolved for industrial environ-ments.17,34,35 Biocatalysts have to be adapted to synthesis

requirements, i.e. they must be engineered to maintain proper

folding for catalytic efficiency under ambient conditions of 

temperature and pressure in the presence of high substrate

and product concentrations.36,37 Thermodynamic stability 

(T m, see Section 2.1) is a required component for industrial

applications; this important parameter concerns the folding 

and unfolding process enzyme.38 For example, enzymes

involved in the production of organic compounds such as

ethanol for fuel must be thermodynamically stabilized in

solution with high concentration of organic solvent to prevent 

misfolding. Enzymes such as DhA of the haloalkane dehalo-genase family from Rhodococcus rhodochrous are often used for

the synthesis of enantiopure organic compounds for organic

synthesis.39,40 DhA, however, with 1,2-dibromomethane as its

natural substrate, is unstable in solvent with high concen-

tration of organic materials such as DMSO.39 This destabilizing 

phenomenon was eliminated by introducing bulkier and

mostly hydrophobic amino acids in DhA’s access tunnel, which

prevented the hydrophilic organic solvent from entering the

active site. Therefore, there were more of the natural substrate

and less of the deactivating DMSO in the active site, which

 yielded kinetic stability.39 Similarly,  a-amino ester hydrolases

have been shown to have great potential for industrial synthesis

of   b-lactam antibiotics when compared to the industrially available penicillin G acylase.41 However, these hydrolases are

thermally instable as they are easily degraded into smaller peptides.

Their rapid degradation at just slightly above ambient temperatures

requires further stabilization of the enzyme (Fig. 3).42

Thermodynamic instability of enzymes can be partly attrib-

uted to the lack of rigidity within the tertiary structure. Such

lack of rigidity often is caused by a large fractions of less stable

and very flexible random coils or  a-helices less stabilized due

to lower ionic interactions, in contrast to the more stable

b-sheets.38 The added rigidity through protein engineering,

must not, however, reduce the flexibility of the enzyme neces-

sary for catalysis. Subsequently, even in reactions occurring inorganic solvents, water plays a significant role in maintaining 

the entropic stability of the protein structure, requiring a care-

ful balance of water in the organic medium.43 Additionally, the

role of salts, pH, ionic strengths, temperature, and chaperones

may need to be explored when stabilizing a biocatalyst.44,45

Multimeric enzymes often dissociate at unfavorable pH values

or ionic strengths, which in case of obligate multimers inacti-

 vates the enzyme. As more parameters are to be optimized, the

more challenging it becomes to achieve thermodynamic stability.

However, these steps of stabilization generally must be determined

before attempting to achieve increased kinetics stability of 

the biocatalyst.

1.2 Stable biocatalysts are a prerequisite for active

biocatalysts, an issue of kinetics

Biocatalysts may be delicate proteins, however, once stabilized,

they are enzymes efficient in catalyzing their respective reactions.36

Kinetic stability (k d, t 1/2, see Section 2.2) must be achieved to a

level satisfactory for large-scale process application. Kinetic

stability evokes resistance to degradation and maintained or

increased reaction efficiency of the enzyme. Nature has been

shown to be a master of evolving enzymes to kinetically and

thermodynamically adapt to their environments as shown by 

psychrophiles, mesophiles, thermophiles, and hyperthermo-philes whose orthologous enzymes evolved to have mutated

amino acids that allow for distinct molecular interactions.48,49

The theory of neutral drift states that there are several types of 

mutations that occur within an enzyme over time: (a) positive

mutations that drifted to maintain functionality (advantageous

mutations), (b) mutations with no effect on functionality 

(neutral mutations), and (c) ‘‘negative’’ mutations that drifted

to alter the function of the enzyme (i.e. deleterious mutations).32,50,51

Neutral drift in the literature is not linked to thermal stabili-

zation. Nonetheless, neutrally drifted and thermostabilized

templates have higher tolerance for mutations; they are able to

tolerate considerable number of changes towards new properties. When attempting to stabilize an enzyme, all three types of 

mutations are encountered, increasing the challenge of protein

engineering. In the laboratories, this process of neutral drift is

achieved by introducing random mutations through error-

prone PCR or DNA shuffling, and selecting those variants with

 wild-type activity.51–53 However, the application of neutral drift 

and finding of purifying mutations is very dependent on an

effective high-throughput screening or selection assay to obtain

the library of advantageous mutations. However, moving 

toward a kinetically stabilized template based on neutral drift 

Fig. 3   Both of the above enzymes belong to the   a/b-hydrolases family

of enzymes. (a) A psychrozyme named   a-amino ester hydrolase (AEH) from

 Xanthomonus citri   (PDB 1MPX) and (b) hyper-thermozyme from   Archaeon

archaeoglobus fulgidus   named carboxylesterase (PDB 1JJI).46 AEH from

 Xanthomonus campestris, an orthologs, has a half-life of 30 minutes at 26.8   1C.42

It is a very fragile enzyme whose tertiary structure is composed of large amounts

of random coils. Carboxylesterase has significantly less random coils in its

structure and compared to its less thermostable counterparts, this enzyme has

shorter loops or random coils.47 Both enzymes are tetrameric.

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 5/32

6538   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

may results in libraries of enzymes with altered substrate

specificity. Depending on the precise screening method,

neutral drift might result in active and stable variants, or just 

optimize the finding of active but not necessarily more stable

 variants. ‘Neutral’ mutations in and around the active site may 

result in increased promiscuity.32,51 In addition to improved

enzymes’ half-lives, improved thermostability has been linked to

enzymes’ increased tolerability for destabilizing mutations.32,51–53

 Focus of current article.   To increase thermodynamic and

kinetic stability, various methods including immobilization,

protein engineering, and medium engineering of biocatalysts

are to be considered.54 In the case of protein engineering, there

are three main approaches to enhancing the stability of protein

biocatalysts (Fig. 4): (i)   rational design, based on knowledge of 

the 3D-structure and the catalytic mechanism, (ii)  combinatorial 

design, requiring a protocol to generate diversity at the genetic

level, a large, often high throughput, screening capacity todistinguish ‘hits’ from ‘misses’, and (iii)   data-driven design,

fueled by the increased availability of nucleotide and amino

acid sequences of equivalent functionality.

2. Different concepts of protein biocatalyst

stability

2.1 Three types of criteria are used to judge stability 

The   thermodynamic stability   of a protein is indicated by the

melting temperature, T m (K or   1C), of the protein or the Gibbs

free energy of unfolding, DG (kJ mol

1

) (Fig. 5). For a two-stateunfolding process (eqn (1)) at the melting temperature,   T m,

the equilibrium constant  K  is equal to unity. Thus,  DGu = 0 at 

T m, eqn (3).

Biocatalysts also exhibit a degree of  kinetic stability at a given

temperature and salt concentration, reported as half-life   t1/2(h or d) or, preferably, as an intrinsic ( i.e.  mechanism-based)

deactivation rate constant   k d   (h1 or d1). As elsewhere in

kinetics, such intrinsic rate constants gave to be carefully 

distinguished from observed rate constants. For enzymes with

a first-order rate of deactivation, the observed deactivation rateconstant, k d,obs is linked to the observable half-life:

t1=2  ¼lnð0:5Þ

kd;obs

(4)

 As shown in eqn (11), the observed deactivation rate constant,

k d,obs depends on the unfolding equilibrium and thus not just 

on kinetics.

 Process stability   or   operating stability   indicates the active

lifetime of a biocatalyst active during the process of the reac-

tion. Two criteria have been employed to characterize process

stability, the total turnover number TTN (eqn (5)) and the

productivity number PN or consumption number (eqn (6)).The more defined and more generalizable dimension of merit 

is the dimensionless total turnover number TTN, which scales

the amount (mass) of product formed per unit mass of (bio)-

catalyst and which thus indicates the average number of 

catalytic events per active site during the entire lifetime of the

biocatalysts. For enzymes under prevalent process conditions

(for details, see Sections 2.4 and 3.3), the TTN can be deter-

mined via  eqn (5):

TTN ¼  ½Product

½ðBioÞcatalyst 0s active site¼kcat;obs

kd;obs

(5)

PN ¼  mass of product

;  kg

mass of prepared catalyst   (6)

The observed specific catalytic activity  k cat,obs  is based on the

active,   i.e.  native, enzyme concentration [N] at given tempera-

ture and time after starting the experiment or production run

(see Rogers and Bommarius).55 Too often, experimenters

assume that [N] equals the initial enzyme concentration [E]0but that is valid only at or close to initial conditions and at 

temperatures far below the melting temperature.

Good productivity numbers (PNs) and total turnover num-

bers (TTNs) of more than 10 000 to 100 000 are not uncommon.

2.2 Thermodynamic stability 

It is well established that the deactivation of a biocatalyst by 

thermal or chemical means may proceed   via   multiple inter-

mediate states.55,100 Generally, any significant conformational

change to the native fold (hence catalytically active) of an

enzyme can be expected to eliminate activity. If one or more

of these alternate conformations are thermodynamically stable,

then there may exist in the system multiple enzyme species,

each with a varied degree of tertiary and/or secondary structure;

only one of the species is assumed to remain catalytically active.

Upon exposure to the system conditions, equilibrium in the

Fig. 4   Three modes of protein stabilization: (i) protein engineering, (ii) immo-

bilization, and (iii) formulation (medium engineering).

Fig. 5   N = native state; U = Unfolded state; K   is the equilibrium constant;  DG is

the Gibbs free energy.

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 6/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6539

 vast majority of cases, especially when not involving dissocia-

tion of multimeric enzymes, is rapidly established between all

of these stable states, according to Fig. 6.In Fig. 6, the native, folded state N is shown in equilibrium

 with n  stable intermediate states. It is implied that the ‘‘early’’

intermediate states (n  = 1, 2. . .) represent smaller departures

from the native state; thus, they have lower energy barriers than

intermediates further to the right. Each of the equilibrium

constants K n, where  K n = [I]n/[I]n1, may be modeled as a van’t 

Hoff equilibrium according to eqn (8), in which   D H n   is the

enthalpy change at the transition state conformation, T n is the

characteristic temperature at which K n equals unity (analogous

to a melting temperature   T m) and   R   is the universal gas

constant.

lnK n  ¼DH 

nR

1

T n

 1

T    (7)

 At least one of the transitions, usually the last one, In1  to In,

leads to complete unfolding of the protein. The temperature of 

this transition commonly is referred to as the melting tempera-

ture, T m, observable via  several techniques (see Chapter 3.1). If 

the temperature difference between  T n  and  T n+1  is sufficiently 

large, several transitions can be observed within one run.

Due to the rapid establishment of equilibrium, the apparent 

initial catalytic activity in the system will be that of [N]0, the

initial concentration of   active   enzyme. This value must be

distinguished from [E]0, the concentration of total enzyme

initially charged to the system, in terms of all equilibriumconstants K n. The model in Fig. 6 generates the initial condition

shown in eqn (8):

½N0  ¼  ½E0

1 þPn j ¼1

P j i ¼1

K i 

(8)

This relationship has two important ramifications for the reporting 

of enzyme specific activity under a given set of conditions:55

– The starting enzyme activity is not [E]0 but [N]0, owing to

the instantaneous equilibration of the enzyme population, and

– An incremental increase in temperature causes a doubling 

of the apparent  specific activity, it must be noted that a smaller

fraction of the original enzyme is contributing to the observed

activity (a fraction of the enzymes is now inactive due to the

shift in equilibrium). Hence, the magnitude of increase of the

intrinsic   specific activity, according to the above correction

factor, would be expected to be:

2 1 þXn j ¼1

X j i ¼1

K i 

!  (9)

In many cases, techniques such as differential scanning calori-

metry (DSC), differential scanning fluorimetry (DSF), circular

dichroism (CD) spectroscopy, or ANS fluorescence (see Section 3)

may be used to determine the number of intermediate states

involved in the deactivation of the biocatalyst, thereby removing 

unnecessary degrees of freedom from the above model.

2.3 Kinetic stability 

In addition to the loss of activity through a rapid shift in

thermodynamic equilibrium, one must consider the entropically-

driven and time-dependent denaturation of a biocatalyst in

parallel to catalytic events. From the general case set forth in

Fig. 7, it can be assumed that any of the stable intermediates

(or the native state) may deteriorate to the permanently unfolded state, D, and that each may do so according to a

different kinetic rate constant (states which are already ‘‘loosened’’

in terms of tertiary structure may deteriorate more readily). The

rate law for denaturation is assumed to be first order, which

leads to the following time-dependent model:

Each of the first-order deactivation rate constants may be

modeled by transition state theory according to eqn (10), where

DGn  is the Gibbs free energy change upon denaturation,  k B  is

the Boltzmann constant,   h   is Planck’s constant,   R   is the

universal gas constant, and   T  is temperature, the underlying 

independent variable that dictates the value of the rate constant 

and, therefore, the speed of time-dependent denaturation at 

any given temperature.

kd;n  ¼K BT 

h  e

DGn

RT    (10)

It was shown previously that the population of native enzyme at 

time   t , therefore, depends on both the equilibria between

available intermediate states and the first-order deactivation

rate constants for each of those states.55 Eqn (11) shows the

expression for [N](t ) in terms of the model in Fig. 7 and the

initial value of [N]0 given in eqn (3).

½NðtÞ ¼ ½N0e

kD;0þ

Pn

 j ¼1

kD; j 

Q j 

i ¼1

K i 

1þPn j ¼1

Q j i ¼1

K i 

0BB@

1CCAt

(11)

 With regard to process evaluation, the expression, in the

exponential multiplied to  t , is essentially a weighed average of 

the first-order rate constants among all of the species present in

the system. This weighted average is the rate constant, which

 would be observed in an isothermal deactivation and will

hereby be abbreviated k d,obs.

½NðtÞ ¼ ½N0e   kd;obsð Þt (12)

Fig. 7   Time-dependent denaturation from multiple equilibrium intermediates.

Fig. 6   General scheme for unfolding via  multiple intermediate states.

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 7/32

6540   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

2.4 Process stability 

 Process or operational stability  relates the kinetic stability of a

(bio)catalyst to its catalytic output throughout the (bio)-

catalyst’s lifetime. A typical process is run isothermally and at 

saturating concentrations of substrate. At such conditions,

V max,t   =   k cat [N]t   =   k cat,obs[N]0. The product yield (y), in moles,

during the lifetime of the biocatalyst is then equaled:55

Z 10

V max;t dt   (13)

 With eqn (12):

 y ¼ kcat;obs½N0

Z 10

ekd;obst dt   (14)

 which simplifies to:

 y ¼ kcat;obs½N0

1

kd;obs

(15)

The TTN scales product yield to the initial biocatalyst concen-tration [N]0:

TTN ¼  y

½N0

¼kcat;obs

kd;obs

(16)

Use of eqn (12) assumes first-order deactivation kinetics of 

the biocatalyst. Under thermal stress or in the presence of 

chaotropes, this assumption is justified in the overwhelming 

number of cases. First-order deactivation cannot be assumed if 

aggregation (a higher-order process) is the method of deactiva-

tion.56  At a constant TTN, a very active but rather unstable

catalyst is equivalent to a slower catalyst with improved stability 

over time.

3. Measuring stability of biocatalysts

3.1 Thermodynamic stability: DSC, DSF, CD, and fluorescence

Thermodynamic instability typically registers conformational

change of a protein from native state N to an unfolded state U

resembling a random coil. This change is modeled as reversi-

ble, N2 U, though by far not every protein unfolding event is

reversible. Upon unfolding, hydrophobic residues are exposed

to the solvent, allowing the application of fluorescence to

record a strong increase in emission upon unfolding. Similarly,

Circular dichroism (CD) uses circularly polarized light to

measure changes in ellipticity upon the disappearance of secondary structure elements,  i.e.  a-helices and  b-sheets.

 Additional methods of tracking thermodynamic instability 

include differential scanning calorimetry (DSC) and differential

scanning fluorimetry (DSF). DSC measures the change in heat 

during the unfolding of the protein. However, it has been a

popular method for non-high throughput studies. DSF, in

contrast, allows the analysis of samples in well plates. DSF

can be conducted with any instrument that allows accurate

measurement and ramping of temperature, such as a real-time

PCR machine. Major conformational changes are picked up  via

fluorescence interactions with a dye probe molecule, such as

SYBER orange. The fitting procedure to calculate the midpoint 

of unfolding, i.e. the melting temperature (T m) is similar to the

procedure used in fluorescence experiment. The procedure has

been described in detail by Niesen  et al .57 Differential scanning 

fluorimetry (DSF) has been compared with differential static

light scattering (DSLS) and isothermal denaturation (ITD) for

applications in protein stability and aggregation measure-

ments.58 DSF has also been applied to screen for ionic liquidsolvents suitable for stabilizing proteins.59

3.2 Kinetic stability: half-life,   s1/2

Measuring residual activity over time at given conditions of 

temperature, pH value, & salt and solvent composition allows

one to follow biocatalyst deactivation and will result in an

activity-time plot that can be fitted to an empirical rate law.

In most cases,  i.e.  if unfolding or chemical degradation is the

determining event, a first-order rate law will be observed with

respect to biocatalyst concentration, allowing one to extract an

observed first-order deactivation rate constant , k d,obs (h1 or d1).

Dominance of aggregation or autohydrolysis in case of proteases leads to a concentration-dependent behavior that 

often can be fit   via   a second-order rate law.56 If a first- or

second-order plot does not fit the data, the apparent reaction

order can be obtained via a Levenspiel plot of ln(dimensionless

rate) vs. ln(1-degree of conversion) or ln(d X /dt ) vs. ln(1  X ), the

slope of which equals the apparent reaction order.60 However,

in the authors’ own experience, the accessible range of bio-

catalyst concentration often does not lend itself to accurate

Levenspiel plots.

 An empirical fit of residual activity over time does not reveal

any details about the  mechanism of deactivation. Consequently,

the rate constant is an observed one; to obtain an   intrinsic

deactivation rate constant ,   k d, a mechanistic representation of the events leading to deactivation is required. For the Lumry–

Eyring mechanism N2 U-  D,100 the observed and intrinsic

rate constants,  k d,obs and  k d, respectively, are linked  via:

kd;obs  ¼  kd

1 þ K   (17)

K  ¼½N

½U  (18)

 with eqn (18) as the folding equilibrium constant.61 The crucial

difference between k d,obs and k d is to be noted when simplifying 

eqn (17) in different temperature regimes.(i) High   T , or   T   >   T m: then [U]  c   [N], thus   K  {   1, and

k d,obsE k d(ii) Moderate T , or T o T m: then [U] o [N], thus  K  > 1, and

k d,obsE k d/(1 + K )

(iii) Low   T , or  T  {   T m: then [U]  {   [N], thus  K  c  1, and

k d,obsE k d/ K 

Extreme care is required when attempting to determine  k dby calculating  k d,obs.

Experimentally, it is often more convenient to express bio-

catalyst stability in terms of the half-life   t1/2   (h or d),   i.e.   the

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 8/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6541

time at which 50% of the initial activity is left. For enzymes that 

deactivate according to a first-order rate law, the observed

deactivation rate constant   k d,obs   is linked to the observable

half-life by a  t1/2 = (ln 0.5)/k d,obs, eqn (4). In case of measuring 

T 3050 or  T 6050 (see Section 3.5), the desired temperatures are those

at which the half-lives are 30 or 60 minutes, respectively.

3.3 Process stability: total turnover number TTN

The dimensionless TTN scales the observed catalytic constant k cat,obs   to the observed deactivation rate constant   k d,obs   (for

first-order deactivation processes), as shown in eqn (10).55

Eqn (10) is valid (i) at saturation of all substrates,  i.e. if  k cat and not just an apparent   k cat,app can be measured, (ii) if the

native enzyme concentration at time t , [N]t , is known, (iii) under

isothermal conditions,   i.e.   both   k cat,obs   and   k d,obs   have to be

measured at the same temperature, most often a temperature

meaningful for a process, and (iv) at the same pH value as well

as salt and co-solvent composition.

There are two other methods to determine TTN:

I. If substrate can be supplied over the lifetime of a bio-

catalyst, which usually but not always requires the use of acontinuous reactor, conversion can be followed over time and

the amount of product compared to the amount of initial

(bio)catalyst employed;

II. If temperature deactivation dominates, the biocatalyst 

can be exposed to increasing thermal stress by ramping up the

temperature, measuring the instantaneous conversion, again

often as output of a continuous reactor, and fitting the time–

temperature–conversion data to a kinetic model of both activa-

tion and deactivation. This method has been demonstrated on

the example of pen G acylase by Rogers  et al.62

In applied biocatalysis, such as in many industrial situa-

tions, where the purity of biocatalyst often is not known,

biocatalyst stability is expressed as an   enzyme consumptionnumber  (ECN):

ECN ¼  1

PN¼

amount of enzymes spent ðgÞ

mass of product ðkg or lbÞ  (19)

PN is defined in eqn (6).

The value of ECN depends on process parameters such as

temperature, pH value, & concentrations of substrate(s) and

product(s). With the molar masses of enzyme and product, ECN

and TTN can be interconverted:

TTN /  1

ECN

MWenzyme

MWproduct

(20)

It should be emphasized that TTN is not a completely suitable

quantity for the evaluation of operating stability because the

number of moles of biocatalyst is not a suitable reference for

the complexity and cost of its manufacture. However, the values

for both numerator and denominator in eqn (20) are usually 

known and can be expressed in monetary terms. To assess the

application of a biocatalyst, the contribution of the biocatalyst 

to the overall cost can be assessed readily.

 An empirical value for the amount of enzyme required to

generate a certain amount of product can be gained via observing 

conversion over time in a continuous stirred tank reactor (CSTR).

The relevant parameter for studies of operating stability of 

enzymes is the product of active enzyme concentration, [E]active,

and residence time,   t; [E]activet. In a continuous stirred tank 

reactor (CSTR), the quantities [E]active  and   t  are linked by the

following equation:

½Eactivet

½S0

  ¼  x

rðxÞ

  (21)

 where [S0] denotes the substrate concentration in the feed,

assumed to be constant,   x   is the degree of conversion, and

r (x) the conversion-dependent reaction rate.63–65  A stable

enzyme leads to a constant degree of conversion   x. How-

ever, an unstable enzyme will lead to a decreasing degree of 

conversion over time. To keep up the initial degree of conver-

sion and thus the output of product per unit time, either the

residence time   t   can be increased or, fresh enzyme can be

added, increasing [E]active. Thus, over time, the amount 

of [E]active required per unit mass of product can be determined,

providing a hands-on empirical measure of total turnover

number.

3.4 Quick stability criteria: T 50 and  C 50

Often, the measurement of the half-life or the deactivation rate

constant is not desirable because it would take too long to

deactivate a biocatalyst, or because a library would create too

many data points at once when following deactivation behavior

of many variants simultaneously. In such cases, T 50 (in the case

of temperature-dependent deactivation) or  C 50  (in the case of 

chaotrope-dependent deactivation) are rapid and useful indi-

cators of deactivation, often over 60 min, resulting in T 6050 or C 6050data.   T 6050   is the temperature at which the measured residual

activity after 60 min is 50% of its initial value. The quantity isneither a thermodynamic nor a kinetic one but encompasses a

mixture of both.

By definition,   T 6050   is the temperature at which [N](t )/[N]0  =

0.5, where [N] is the concentration of native, folded (hence

catalytically active) biocatalyst present in the system. Combin-

ing this fact with the result in eqn (12) for  t  = 3600 s gives the

following:

ln(0.5) =  (k d,obs)(3600) (22)

This leads to the idea that, at  T 6050, the observed first-order

rate constant of the biocatalyst takes on a particular value:

k d,obs = 1.925    10

4 s

1 (23)

 As shown in eqn (11), k d,obs is a quantity which depends on

the thermodynamic equilibrium between intermediate states as

 well as the propensity of each state to undergo time-dependent 

denaturation. If information about mechanism or number of 

intermediates is known, a priori , then a restricted model can be

imposed and the T 6050 can be calculated directly. To illustrate, it 

 will be assumed that a biocatalyst follows a deactivation

mechanism akin to the Lumry–Eyring mechanism: N2 U2 D.

For such a system,   k d,obs   at the   T 6050   can be expressed as a

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 9/32

6542   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

function of individual deactivation rate constants and the

equilibrium constant between N and U:

kd;obs  ¼  kd;1K 1

1 þ K 1

¼ 1:925 104 (24)

The above equation may be rewritten by expressing the

deactivation rate constants and the equilibrium constant in

terms of their transition state and van’t Hoff definitions,

respectively:

1:925 104 ¼

kBT 

h  e

DG1

RT 

  e

DH 1R

1T 1

1T 

!

1 þ e

DH 1R

1T 1

1T 

0BBBBB@

1CCCCCA (25)

 Assuming typical parameters for unfolding and

deactivation,62,66,67

D H 1 = 108 kJ mol1

DG1 = 98 kJ mol1 T 1 = 325  K 

the value of   T , which is equivalent to  T 6050, is then found by an

iteration with the desired level of convergence (here, to match

the value of  k d,obs within 1 107) to be T 6050 = 319.1 K. So even

at 6 K below the melting temperature  T m of 325 K, the enzyme is

stable only for an hour.

3.5 Additional stability indicators:  T opt  or  T max 

T opt  for enzyme reactions: Peterson  et al.67 define a parameter

T eq, an equivalent of the melting temperature,   T m. Peterson

demonstrates that the optimum temperature,   T opt , for an

enzyme reaction,   i.e.   the temperature at which a maximum

rate is observed that is determined from dV max 

/dT   with an

analytical expression, is very close to the observable maximum

in a plot of rate as a function of temperature:

T opt E T m(1    aT ) (26)

a ¼  R

DH mln

  DH m

DGacat

1

  (27)

Since typically, 5 105o ao 104, a T m{ 1; and thus

T opt  is approximately equal to T m with a difference between T opt and  T m of less than 0.018   1C. This fact had been surmised by 

many practitioners but had never been supported with experi-

mental results. While the significance of   T opt    may seem

obvious to anyone trying to optimize the rate of an enzymaticreaction, Peterson   et al.   point out the significance of the

melting temperature   T m   next to   T opt : the magnitude of   T mreflects thermostability, as does the deactivation rate constant 

k d, and also reflects evolutionary adaptation as the melting 

temperatures of proteins often correlates with the preferred

habitat of the organism harboring the proteins.

Sometimes the   T opt    is referred to as   T max ; while   T opt commonly is utilized in connection with impending unfolding 

upon further heating and   T max   can occur at temperatures

significantly below the melting temperature, the difference

between T max  and  T opt  often is rather semantic.  T max  for enzyme

reactions:   Burkholderia cepacia   lipase immobilized on porous

ceramic particles, lipase PS-C II (Amano Enzyme Inc.), gave an

enantiopure product at 40–120   1C in the kinetic resolution of 

1,1-diphenyl-2-propanol, with the highest conversion (39%) at 

80–90   1C.68

4. Modes of stabilizing biocatalysts:immobilization, medium engineering,

protein engineering

4.1 Mode 1: immobilization

Overview.  Stabilization of biocatalysts through immobilization

have been employed by industries since the 1960s for several

reasons.69 Immobilization of biocatalyst is a method by which

an enzyme is bound to a solid support through covalent or ionic

interactions, or   via   entrapment or crosslinking, to eliminate

diffusion of the enzyme in solution. Site-directed mutagenesis

can be employed to introduce amino acids that would facilitate

these desired types of interactions in the desired regions of the

protein. Immobilization allows for the recyclability of bio-catalysts as well for creation of robustly bound enzymes.45

 Additionally, immobilization can be used in reactions requiring 

biphasic systems where the product of the reaction is immedi-

ately and automatically isolated from the reaction phase. In the

case of using enzymes as biosensors, immobilization provides

the optimal control of the enzyme’s orientation as needed by 

the system.34 For biodiesel production, immobilization of 

enzymes yields an environmentally friendly process since

operating temperature tends to be lower while purification of 

products is eased.70 Most importantly, however, the method

of immobilization can be used to stabilize enzymes both

thermodynamically and kinetically as well as change theirfunctionality.71 This phenomenon was shown by both Grazu

et al. and Cecchini et al. where multipoint covalent immobiliza-

tion of penicillin G acylase resulted in several stable variants of 

the enzyme (Fig. 8).72,73 Both groups showed that by combining 

site-directed mutagenesis and immobilization of enzyme in

Fig. 8   Grazu et al. showed that multipoint covalent immobilization can stabilize

the protein depending on the locations  vis-a-vis  the active site.72

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 10/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6543

specific regions, generally far from the active site, a stable

 variant with improved reaction specificity can be created. Using 

this method, an astonishing kinetic stability of penicillin G

acylase was shown by the significant reduction of the undesired

secondary hydrolysis.73

 Integrative chemistry. Integrative chemistry is a newly emer-

ging concept where the designs of complex hierarchal struc-

tures inspired by nature are constructed with interdisciplinary 

tools including those of chemistry.74,75 Brun  et al.  used inte-grative chemistry-based rational design to create the first lipase

hybrid macrocellular biocatalyst, triacylglycerol acyl hydro-

lase.74 Immobilization was used as a method of stabilization

and also to create recyclable reaction materials. Lipases from

Candida rugosa  (CRl) and   Thermomyces lanuginosus  (TLl) were

used in their study; the former enzyme is known for its high

substrate specificity and reactivity and the latter for its high

reactivity in transesterification reactions involving the produc-

tion of biodiesel. Both enzymes were stabilized with porous and

(3-glycidyloxypropyl) trimethoxysilane (Glymo)-functionalized

silica-based hybrid macrocellular monolithic foams that are

capable of stabilizing the interface between water and oil; thissystem was dubbed Si(HIPE).

Surface display. Another example of immobilization concerns

recombinant whole-cell biocatalysis with the enzyme expressed

at the surface of the cells. The final result is an active enzyme

anchored at the cell surface.76 Kim   et al.   used aga2 protein

fused to epoxide hydrolase to anchor the protein to the surface

of the Rosetta strain of  E. coli . Using this method, the authors

eliminated isolation and purification of epoxide hydrolase,

 which allows for the maximum use of the enzyme during the

initial period of high activity right after protein synthesis and

folding. In addition, the enzyme, while in the cell, is stabilized

by multiple factors that may be unknown to the experimenter.

Kim   et al.   also found that the enzyme had an improved   K M value of 2.6 mM with this process.

 Fusion protein.   Immobilization of a fusion protein of a

cellulose-binding domain (CBD) and the catalytic domain of 

horseradish peroxidase (HRP) onto cellulose beads preserved

both binding and catalytic function and resulted in six-fold

enhanced stability of the construct in buffer. Cellulose-bound

CBD-enzyme was used for the first time in aqueous-organic

solvents.77 With increasing THF, acetone, acetonitrile, or ethanol

concentration, activity decreased but stayed above the level of 

soluble enzyme forms. The soluble CBD-HRP fusion protein

featured lower activity than native HRP but higher thermal and

solvent stability. The immobilized fusion protein was found tobe more stable in aqueous acetone of any composition.

Single-walled carbon nanotubes (SWNTs). Single-walled carbon

nanotubes (SWNTs) were found to stabilize enzymes against 

high temperatures and organic solvents than flat surfaces. The

curvature of SWNTs, which suppresses unfavorable protein–

protein lateral interactions, is the cause of the stabilization, as

 was found through both experiments and theoretical analysis.

Curvature-based stabilization should not be limited to SWNTs

but should be a general property of nanomaterials.78 Similarly,

highly curved surface of C60 fullerenes enhances the stability of 

soybean peroxidase at 95   1C. The half-life at that temperature of 

soybean peroxidase adsorbed onto fullerenes was measured to be

2 h, compared to 25 min when adsorbed on graphite flakes, a flat 

support, and 10 min for the native, soluble enzyme.79 This

phenomenon is not limited to fullerenes, but can also be extended

to other nanoscale supports including silica and gold nano-

particles, or highly active and stable polymer-nanocomposite films.

4.1.1. More examples of stabilization through immobilization.

Case I. Penicillin G acylase (PGA) was stabilized through variousmeasures: (i) insertion of a cysteine, each in different regions of 

the lysine-rich enzyme surface and oxidation of the sulfhydryl

sidechain groups to form disulfide bonds, which stabilize the

enzyme via  freezing of configurational entropy, (ii) subsequent 

immobilization on disulfide/epoxide supports, and (iii) a series

of point mutations to increase resistance against heat and

organic solvents. The best immobilized PGA variant was

300 000 times more stable than the soluble wild-type, as judged

by a combination of increased temperature and half life, and

thus very suitable for various reactions of interest in fine

chemistry and pharma, including penicillin G hydrolysis,

enantioselective hydrolysis and synthesis reactions. Immobili-zation on disulfide-epoxide (Eupergit) and glyoxyl-agarose

supports rendered biocatalysts of comparable stability. The

stereochemistry of the PGA variants did not vary significantly 

from that of the wild-type.80

Case II.  PGA has also been encapsulated into a rigid poly-

 vinylalcohol matrix, termed Lentikats, in the form of cross-

linked enzyme aggregates (CLEAs) to compensate for the

inadequate mechanical properties of CLEAs. While encapsula-

tion in Lentikats decreased CLEA activity by 40%, its improved

stability more than compensated for the decreased activity.

Partitioning studies of the dioxane solvent into the Lentikats

found that the solvent concentration inside the Lentikats was

lower than in the bulk and lent credence to the hypothesis that the hydrophilic environment in the matrix is primarily respon-

sible for increased solvent stability; indeed, thermal stability 

 was comparable to the corresponding CLEA. The PGA-CLEA 

Lentikats with their favorable substrate and product partitioning 

 yielded a significantly higher reaction rate as well as enhanced

 yield in the thermodynamically controlled synthesis of cephalo-

sporin C (ceph C) from phenylacetic acid and 7-amino-

deacetoxycephalosporanic acid (7-ADCA).81

Case III.   In the next example, immobilization of formate

dehydrogenase (FDH) on highly activated glyoxyl agarose, in

contrast to other matrices, increased its stability against several

distorting influences, such as extreme pH value, high tempera-ture, presence of organic solvent, or contact with hydrophobic

interfaces as a consequence of stirring. Biocatalyst optimized

 with respect to activation of the support, time and temperature

of immobilization retaining half the initial activity was stabi-

lized 50-fold at high temperatures and neutral pH value and at 

10   1C with pH 4.5. At acidic pH value, where inactivation is

preceded by subunit dissociation, immobilization under such

optimized conditions stabilizes the quaternary structure of 

dimeric FDH and stabilizes the glyoxyl-agarose-FDH by hun-

dreds of times over the wild-type.82

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 11/32

6544   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

Case IV.   Brun   et al.   had created a significantly improved

efficiency of CRl in the synthesis of butyloleate ester, which is

used as a lubricant in biodiesel.73 Stabilized CRl showed an

overall 26 fold improvement in reactivity. In addition, the enzyme

 was recycled 19 times during which it maintained 100% catalytic

efficiency! TLl was shown to also have increased catalytic effi-

ciency of 80% hydrolysis of triester of olive oil in non-aqueous

medium after four rounds of recycling and up to 65% in 8 rounds.

4.2 Mode 2: medium engineering 

 A second method of stabilization of proteins relies in formulating 

favorable buffer conditions. Here, both salt concentration and

salt type are important.

Salt concentration. The concepts for the quantitative descrip-

tion of the influence of salt concentration have been developed

since the 1880s but mainly in the 1930s. In 1889, Setschenow 

published his law on the dependence of the logarithm of 

protein solubility on salt concentration.83

log  P0

P ¼ ks½Csalt

  (28)

 with [P]0   and [P] the protein concentrations in presence and

absence of salt, respectively, and the k s   the Setschenow coeffi-

cient. Soon after, it became clear that not the ion concentration

[C] but the ionic strength,  I , was the appropriate descriptor of 

salt concentration. In the 1920s, Cohn and Edsall developed

their law for fractional crystallization of proteins, which was

published in 1943:84

I  ¼1

2

XZ 2mi ; in mol L1

log  E 0ð Þ ¼ log  S 0ð Þ K sI 

(29)

Debye and H +uckel treated a point charge in a medium of 

uniform dielectric constant  e  and based on this concept devel-

oped the Debye–H +uckel law (eqn (30)) and, at high ionic

strength, the extended Debye–H +uckel law (eqn (31))85  written

here for proteins and salts:

log( y) = A|Z proteinZ salt | I 1/2 (30)

log( y) = [ A|Z proteinZ salt | I 1/2] + [ AA*|Z proteinZ salt | I ] (31)

 with [E] = enzyme solubility [g/L], b  = protein solubility at zero

ionic strength   I   (log[E]0), and   K s   denoting the salting-out 

constant. Except for utilizing   I   and [C]salt , eqn (28) and (29)

are mathematically equivalent.

Debye and H +uckel treated a point charge in a medium of 

uniform dielectric constant   e   and based on this concept 

developed the Debye–H +uckel law (eqn (30)) and, at high ionic

strength, the extended Debye–H+uckel law (eqn (31)).85 To

make use of the Debye–H +uckel law for enzyme deactivation,

 we apply transition state theory, which predicts that   k d   =

k d,0(gproteingsalt /ga), so eqn (32) results:

log(k d) = log(k d,0)   A|Z proteinZ salt | I 

1/2

(32)log(k d) = log(k d,0)    [ A|Z proteinZ salt | I 1/2] + [ AA*|Z proteinZ salt | I ]

(33)

Surprisingly few enzymes have been analyzed with respect to

the influence of salt concentration on stability or activity,

 whether with the (extended) Debye–H+uckel model, or other-

 wise. Examples include glucose dehydrogenase86 and formate

dehydrogenase.87

Salt type: influence of ions on biological molecules: the Hofmeister 

series.   The type of salt in most instances is even much more

important than its concentration. The significance of ion type

for protein solubility and precipitation has been known sincethe 1880s, when a pharmacist, Franz Hofmeister, observed

the dependence of the precipitation of hen-egg white lysozyme

on the type of salt 83 and qualitatively ranked the influence

of ions.

The non-specific influence of ions on many biophysical

phenomena, such as salting-in and salting-out of proteins,83,88

melting of gels,89 DNA helix-coil-transitions,90 and ion

channel formation,91 has been described qualitatively by the

Hofmeister series (1888),88 which classifies and ranks ions as chao-

tropes (water-structure breaker) or kosmotropes (water-structure

former).

Fig. 9 ranks some common anions and cations in terms of 

chaotropicity and kosmotropicity, starting with the strong chaotropes Cs+ and thiocyanate (SCN) via   the indifferent Cl

and Na+ ions to the strongly kosmotropic phosphate and Mg 2+

ions. In the case of well-folded enzymes, addition of destabilizing 

(chaotropic) ions tends to lead to increased unfolding or

deactivation, whereas addition of stabilizing (kosmotropic)

ions tends to support preservation of activity.

Two of the most relevant of the few attempts towards a

quantification of the Hofmeister series are the   dipole-cavity

model  and the   ion hydration model .

The dipole-cavity model , treating the protein as a dipole in a

high ionic strength aqueous solution (Kirkwood model),92

attributes the influence of ionic strength on a protein to

Fig. 9   The Hofmeister series.

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 12/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6545

electrostatic (DGelectrostatic) and surface energy (DGcavity ) forces93

and arrives at the following eqn (34):

DG0  ¼ DGcavity þ DGelectrostatic

¼ A   BI 0:5

1 þ CI 0:5 ðL OgÞI    (34)

( A,   B,   C   and   D   are constants,   L   and   Og   represent the

electrostatic and hydrophobic terms, respectively, and  g  is the

surface tension of the solvent.) As the surface tension  g  can be

 written as  g  =  g0 +  Ds, with Ds as the surface tension increment 

attributable to the salt, the Debye term becomes unimportant 

at high salt concentrations and eqn (34) simplifies to eqn (35),

 written here for rate constants:

log(k ) = log(k 0)   (L   ODs) I    (35)

Eqn (35) resembles the Setschenow (1889)83 (eqn (28))

and the Cohn–Edsall eqn (29) (1943)84 for the solubility 

of proteins.

The ion hydration model 84,94–97 describes the influence of salt 

as an interaction of chaotropes and kosmotropes with layers of 

hydration on the surface of the respective ion, captured by the Jones–Dole equation:98

Z

Z0

¼ 1 þ Ac0:5 þ Bc   (36)

Z   and   Z0   are the viscosities in salt-containing and salt-free

medium, respectively; again, the Debye–H+uckel term,   Ac0.5,

becomes insignificant at high salt concentration (c  E   1 M),

so that relative viscosity   Z / Z0  depends linearly on ion–solvent 

interactions exerted by chaotropes (with a Jones–Dole  B  coeffi-

cient o 0) or kosmotropes ( B  > 0).  B  values are approximately 

additive for individual ion species and are available from the

literature. Usually, only anions have to be considered, as they are more strongly hydrated than cations.99

The observed kinetic deactivation constants,   k d,obs, for the

deactivation reaction of native protein (N) to some deactivated

state (or ensemble of states) (D), eqn (37), strongly correlate

 with the Jones–Dole   B-viscosity coefficient 98 of the anion of 

chaotropic salts ( B o 0) and show little variation in solutions of 

kosmotropic salts ( B   > 0).100 As   B-viscosity coefficients are

indicative of ion hydration,95,101,102 this observation suggests

that ion hydration strongly influences a salt’s effect on a

protein and that   B-viscosity coefficients may be useful for

predicting salt effects on protein deactivation.

N  !kd;obsD

Specifically, the model should predict invariance of   k d,obs with kosmotropic B-viscosity coefficients as well as the observed

logarithmic dependence of   k d,obs   with chaotropic   B-viscosity 

coefficients. This bimodal behavior suggests that the observeddeactivation might be the result of competing, parallel pro-

cesses, rather than unfolding and irreversible processes in

series, as implied by the Lumry–Eyring model.103,104 In addi-

tion, it would be convenient to incorporate dependencies of 

k d,obs   on water activity (i.e.   salt concentration) as well as

temperature into the model so that salt-induced deactivation

can be predicted under a variety of solution conditions.

 As seen in Fig. 10, a three-parameter exponential relation-

ship with the form of eqn (37) is able to model these

characteristics:

k d,obs = [k p +  k c exp{o B}] (37)

The theoretical curves of eqn (38) generated using assumed

 values for  k p,  k c  and  o  shown in Fig. 10 bear striking resem-

blance to the experimental log (k d,obs) vs. B-viscosity coefficient 

data and suggest that the model might be used to describe salt-

induced protein deactivation.

The overall model from salt-induced protein deactivation

gives:104

kd;obs  ¼ kp þ kc exp   Bo0 þ o0 ln aw

RT 

  (38)

Provided k p,  k c,  o0, and  o 0 are known for a protein–solvent 

combination, the model presented can be used to predict thedeactivation constants of proteins in varying salt solutions at 

 varying water activities and temperatures. Table 1 summarizes

the model parameters obtained for HL-ADH,  a-chymotrypsin,

and mRFP.100

More recently, the dichotic behavior of the deactivation or

misfolding rate constant as a function of the Jones–Dole

 B-viscosity coefficient has been extended to the misfolding of 

 yeast prions105 and monoclonal antibodies.106 In the following 

section, several other examples of the model will be presented.

Fig. 10   Correlation of deactivation rate constants of HL-ADH (60 1C), a-chymotrypsin (50 1C), and mRFP (80  1C) with B-viscosity coefficients. All deactivations occurred

at pH 7.0. Deactivation rate constants were measured in salt solutions with water activities of  aw = 0.95 (K), 0.97 (O), and 0.99 (.). Deactivation constants appear to

vary linearly with chaotropic  B-viscosity coefficient (B o 0) but are relatively unaffected by kosmotropic salts (B > 0). Taken from Broering.100

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 13/32

6546   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

4.2.1. Examples of stabilization of biocatalysts through

medium engineering.   Case I.   To understand in details the

mechanism of action of trehalose, a known exceptional stabi-

lizer of proteins retaining enzyme activity in solution as well as

in freeze-dried state, a thorough investigation of its effect on

the thermal stability of solutions of RNase A was conducted

(Fig. 11).107 A 2 M trehalose solution raised T m of RNase A by as

much as 18  1C and the Gibbs free enthalpy (DG) by 20.1 kJ mol1

at pH 2.5, while decreasing the heat capacity of protein dena-

turation (DC p) in trehalose solutions. Such findings point 

toward a general stabilization mechanism due to the elevationand broadening of the RNase A stability curve (DG vs. T ). Four to

seven molecules of trehalose were found to be excluded from

the protein surface upon denaturation at 1.5 M sugar. Investi-

gation of stability as a function of pH value suggested that 

 while the charge of RNase A can contribute significantly to its

stability, trehalose acts as a universal stabilizer of protein

conformation due to its exceptional effect on the properties

of solvent water around the protein.107

Case II.  The irreversible thermo-inactivation of savinase at 

70   1C due to autodigestion limits its industrial utilization in

detergents. The effect of sorbitol and trehalose on the stability 

of savinase revealed that presence of either osmolyte prevented

autolysis of savinase at 70   1C and increased the protein’sstructural and kinetic stability.109

Case III.   The apparent melting temperature,   T m, of papain

shifted from a control value of 831   to a value of >90   1C in

presence of 40% sorbitol, as measured by fluorescence spectro-

scopy and differential scanning calorimetry (DSC); maximum

stabilization was seen in case of 30% sorbitol. Transitions of 

both domains of papain shifted to higher temperatures. Activity 

measurements with papain indicate several-fold increase in the

protein’s thermal stability in sorbitol, sucrose, xylose, or glycerol.

Partial specific volume measurements of papain in presence of these co-solvents revealed that the preferential interaction para-

meter (x3) was negative in all co-solvents and that maximal

hydration was observed in the case of glycerol where the

preferential interaction parameter was 0.165 g g 1. These results

suggest that papain is thermally stabilized in presence of polyol

co-solvents as a result of preferential hydration.110

Case IV. To investigate the effect of polyols on thermostability 

and the anomalous surface tension behavior of glycerol, a

highly thermolabile two-domain protein, yeast hexokinase A,

has been monitored via differential scanning calorimetry (DSC)

and loss of activity over time. Upon addition of polyols, the  DT m

for domain 1 was larger than  DT m  for domain 2, increasingly proportionally with the number of hydroxyl groups in polyols,

 with sorbitol as the best stabilizer against both thermal stress

and chemical denaturation by urea. The increase in  T m (DT m),

the retention of activity, and the increase in the surface tension

of polyol solutions correlate very well, except glycerol. However,

the   DT m   values show a linear correlation with apparent 

molal heat capacity and volume of aqueous polyol solutions,

including glycerol. These results indicate that interfacial proper-

ties do not always correlate well with stabilizing effects, while

bulk solution properties contribute significantly to protein

stabilization. Various weak binding and exclusion effects of the

osmolytes scaled by the effects of water form a subtle balance

and influence stabilization. These effects must be understoodfor rational design of stable protein formulations.111

Table 1   Values for the three-parameter model given by eqn (28) using protein-

specific values for k p and k c. Dimensions of k p and k c forchymotrypsin are (mM min)1.

Taken from Broering  et al.104

HLADH Chymotrypsin mRFP

k p (min1) 6.85   104 52.1 2.13   103

k c (min1) 3.13   104 40.8 8.03   104

o , a w  = 0.95 (M) 115   4.0 27.8   13.2 61.2   2.9o , a w  = 0.97 (M) 96.7   7.0 33.1   3.6 44.8   1.6o , a w  = 0.99 (M) 70.0   3.1 23.2   3.2 31.6   2.2

Fig. 11   View of RNase A from Bos taurus  at three different angles. PDB: 2E3W.108 The disaccharide is trehalose, which was shown to stabilize RNase A.

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 14/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6547

Case V.  The hydrolytic activity of penicillin V acylase (PVA,

EC 3.5.1.11) can be enhanced by adding monophasic organic

solvents but this addition also in many cases decreases the

stability of the enzyme (Fig. 12).  Streptomyces lavendulae  PVA’s

threshold monophasic organic solvent concentration, at 

 which half of the initial activity in purely aqueous systems is

recovered, in water–organic monophasic systems correlateslinearly with the free enthalpy of denaturation at 40  1C. Solvents

 with log( P ) values o1.8 exert protective effects, those with log 

 P   values >1.8 have a denaturing effect; both effects are

exacerbated at higher solvent concentrations. Deactivation rate

constants of PVA at 40  1C can be predicted in any water–organic

monophasic system.112,113

4.3 Mode 3: protein engineering 

Protein engineering has been employed since the 1980s,

starting in the laundry detergent industry 115,116 This method

of stabilization differs from immobilization in that it requires

changes to the primary structure of the enzyme to confer thedesired features. Quite often, however, protein engineering is

used in conjunction with immobilization to stabilize proteins

as shown by Cecchini et al. Over the years, as described by the

theory of neutral drift, nature has evolved to select for certain

existing mutations (neutral, deleterious, or advantageous)

 within orthologous proteins, which tend to suit their respective

host organisms.117,118 The laboratory settings attempt to mimic

nature, a process known as laboratory evolution.119 However,

rather than selecting for already present mutations within a sea

of bacteria medium, researchers or protein engineers study 

orthologous enzymes and use other methods to introduce new 

mutations for creating the desired kinetically and thermo-

dynamically enhanced protein variants. Three methods of 

protein engineering are often used: rational design, combinatorial

design, and data-driven protein design.

4.3.1. Rational protein design.   Rational design of bio-

catalysts is one of the most commonly used methods of proteinengineering primarily because of the efflux of available data

on protein sequence, structures, and more. The ideology of 

rational design is based on studying available structures and

sequences of already known stable proteins; based on the

interactions of their respective amino acids, which are assumed

to participate in stabilization, a new variant of the protein of 

interest is created through site-directed mutagenesis.120

Respectively, this method requires an extensive knowledge of 

the protein of interest.1,121 However, public databases are now 

rich of protein structures, although not all proteins are found

 within that database and not all proteins have stable orthologs,

i.e.  a-amino ester hydrolase.122

In addition, it is often difficult to determine which of these mutations found in the ortho-

logous enzymes were selected as neutral, deleterious or advan-

tages through evolution.118 Even if those mutations were

determined, the complex interactions between the primary,

secondary, and tertiary structures are often difficult to predict 

or fully account for with increasing size of the protein.123 Two

theories are currently given as to how nature evolves these

enzymes: (a) the neo-Darwinian hypothesis that states that 

positive mutations have already been installed in the primary 

sequence of the protein and that neutral and deleterious

Fig. 12   Penicillin V Acylase from  Lysinibacillus sphaericus. PDB: 3PVA.114

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 15/32

6548   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

mutations are then added over time; (b) the competing hypo-

thesis is that which states that neutral mutations confer flexibility 

for evolution at the addition of deleterious or advantageous

mutations, which are selected under certain conditions.117

However, both theories agree that the sole examination of 

protein structures and sequence will not always yield the

desired answers. Thus, other methods are used in addition to

rational design. Combinatorial design, discussed in the follow-

ing section, is used to account for some of the structuralinteractions within an enzyme.124

4.3.2. Examples of rational biocatalyst design. Case I.  The

stability of prolipase from   Rhizopus oryzae   (proROL) toward

lipid oxidation products such as aldehydes was insufficient for

use in the oleochemical industry. To increase stability, 6 Lys

and all 6 His residues except for the catalytic histidine out of 22

amino acid residues (15 Lys and 7 His), prone to react with

aldehydes, were subjected to saturation mutagenesis. Active

 variants were prescreened by an activity staining method on

agar plates to quickly and reliably identify mutants within the

resulting libraries that lead to more stable variants. Active

mutants were expressed in  E. coli  in a 96-well microtiter plateformat and challenged with octanal as a model-deactivating 

agent. The most stable histidine variant (H201S) increased

stability by 60%, or even 100% in combination with a lysine

 variant (H201S/K168I). Interestingly, the variants still displayed

essentially wild-type catalytic activity.125

Case II. LipC from Pseudomonas aeruginosa is a cold-adapted

and cold-acting lipase, highly salt- and heavy metal-tolerant,

 with different substrate specificity at higher temperatures; all

these properties are biotechnologically and environmentally 

relevant, where often activity at low temperature is sought.

However, low thermostability of LipC prevents its application

in long-term processes. Specific sites on the sequence were

picked for modification according to the highest flexibility onthe 3D model structure to increase LipC thermostability but 

 without decreasing cold-adapted properties. Eight mutant 

libraries plus two point mutations were constructed and more

than 3000 mutant clones screened for successful LipC variants:

a LipC double variant with conserved cold-adapted properties

and 7-fold increased thermostability over the wild-type LipC

 was found and has the desired properties for use in processes at 

low temperatures (4–20   1C).126

Case III.  Large-scale synthetic applications of cyclohexane

monooxygenase (CHMO) are hampered, however, by the

instability of the enzyme mostly owing to oxidation of cysteine

and methionine residues. Exhaustive mutation of all methio-nine and cysteine residues in the wild-type CHMO revealed that 

oxidation-labile residues are mostly either surface-exposed or

located within the active site. Combinatorial mutation of two

methionine residues identified for thermostabilization but 

buried within the folded protein led to two variants with either

oxidative or thermal stability, while keeping specific activity 

and enantioselectivity. The most oxidation-stable variant 

retained nearly 40% of its activity after incubation with H2O2

(0.2 M), whereas wild-type activity was completely lost at 5 mM

H2O2. Therefore, oxidation-stable CHMO variants might be

required for subsequent thermostabilization, so that lab-

evolved thermostability in CHMO might be masked by a high

degree of oxidation instability.127

Case IV.  In the case of pharmaceutical industry, the use of 

rational design to stabilize biocatalysts for the production of 

organic precursors is becoming a widely valued approach.

However, during the synthesis of many active pharmaceutical

ingredients, the use of cofactors such as NADH and NADPH,

(simplified notation: NAD(P)H) is widely common. Sincethe cost of these co-factors are expensive, Park   et al.   have

engineered an NAD(P)H oxidase biocatalyst named NoxV from

 Lactobacillus plantarum, which is able to regenerate the

NAD(P)H cofactor using oxygen with water as the byproduct.128

They showed that the enzyme has inproved Michaelis–Menten

kinetics with improved stability when compared to other

NAD(P)H oxidases.128 Based on structural comparison of NoxV 

 L. plantarum   to NAD(P)H oxidase of   L. sanfranciscensis, Park 

et al.   found that site-directed mutagenesis of NoxV in the

binding pocket with single mutation L179R and double muta-

tions L179R/G178R would yield two enzyme variants capable of 

accepting NADH and NAD(P)H while maintaining the enzyme’sspecific activity.128 There are few other cases where rational

design is used to create an oxidase that is capable of accepting 

 various types of cofactors for the use of the synthesis of 

pharmaceutically viable compounds.129

Case V.  Often, high throughput screening can require lots

of resources and, in addition, may yield less than 0.001%

of desired variants and more than 33% chance of creating 

deleterious variants.130 Ito  et al., using rational design, trans-

formed a mesophilic cellulase, cellobiohydrolase II (CBH2),

into a thermostable enzyme.130 Ito   et al.  revealed 45 distinct 

amino acids not conserved using primary structure alignment 

of five thermophilic fungal CBH2 and Phanaerochaete chrysosporium

CBH2. They were able to introduce 15 additive mutations ontothe wild type, which conferred thermostability and improved

kinetics stability. Ito   et al.   also showed the efficacy of using 

sequence homology to avoid the creation of random library that 

may yield no desirable outcome while requiring extensive usage

of high throughput materials.130

Case VI.   Tackling the issue of organic solvent and bio-

catalysts, Park   et al.   studied and devised a method to create

organic solvent-stable Candida antarctica lipase (CalB) (Fig. 13).

Enzymes involved in organic synthesis must be stabilized in the

presence of their substrates, products, and solvents. Use of 

Fig. 13   Lipase CalB from  Candida antarctica.  PDB: 3ICV.132

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 16/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6549

organic solvents in reactions presents many benefits to indus-

try because they allow for sterile reaction conditions since

microbes are not able to grow in organic mediums, hydro-

phobic substrates can be accommodated, and the equilibrium

can be shifted towards product production.131 Park  et al.  were

able to create sets of rules for stabilizing enzyme in organic

medium based on previous literatures on directed evolution of 

enzymes in hydrophilic solvents, and based on their study of 

and results from CalB stabilization. One guidance recommendsmutating amino acids from the surface of the proteins; the

chosen surface residues must be capable of either shortening 

the lengths of or increasing the number H-bonds between the

proteins hydrophilic organic solvent; increasing or strengthening 

H-bonds helps prevent unfolding when water molecules are

removed from the protein. Three mutants were created: N97Q,

N264Q, and D265E, all of which had stability in organic solvent 

 when compared to the wild type.131 Their half-lives were

increased up to 1.5 fold while activity was increased up to

17%. Park  et al.  demonstrated that by understanding solvent–

enzyme interactions and using rational design, one could

construct variants that have improved functionality and stability in organic solvents.

Case VII.   Similarly, Badoei-Dalfard   et al.   performed two

mutations on   Salinivibrio   zinc-metalloprotease (SVP) to

increase active site polarity and therefore maintain its hydra-

tion while the enzyme is in organic solvents such as DMF and

methanol.133 The two mutations in the active sites of SVP are

 A195E and G203D. Additionally, to stabilize a surface loop, they 

performed A268P. For A195E, Badoei-Dalfard   et al.   created a

 variant with improved kinetics of 11%, 26%, 32%, and 41% in

methanol, DMF, isopropanol, and   n-propanol, respectively.

 Additionally, with this mutation, rate of irreversible deactivation

 was reduced by 31–33% in DMF and  n-propanol. Although with

G203P the improvements in kinetics stability were similar tothose of A195E, its rate of inactivation was higher by 17% and

23.1% in DMF and  n-propanol, respectively.133

Both Park  et al.   and Badoei-Dalfard  et al.   introduce muta-

tions using acidic amino acids Q and E to confer stability of 

enzyme in organic solvents.133 One used these amino acids to

stabilize hydrophilic interactions with organic solvent to avoid

unfolding due to removal of water while the other’s goal was to

maintain water within the active sites while in organic solvents.

These methods of rational design for enzyme in organic solvents

maybe found effective in various other enzymes. However, it is

important to note that both groups have proposed seemingly 

opposite hypotheses while yielding the same results: stable

enzymes. However, the goal of Park  et al.  was to remove water

molecules   from the surface   and increase hydrophilic inter-

actions with the organic solvent, while Badoei-Dalfard   et al.strive to keep water molecules   within the active site   when in

organic solvent. Therefore, both groups were able to confer

stability using one method: rational design revolving around

the strengthened H-bonds.

4.3.3. Combinatorial protein design. Although it is nearly 

impossible to account for all the possible interactions within an

enzyme and to satisfactorily determine the role of every evolu-

tionary mutation, combinatorial protein engineering is able to

assess the suitability of a mutation based on energetic

features.124 This computational method coupled with labora-

tory assays, often high-throughput assays, can help find

mutants with the desired enzymatic functions.

124

 When setsof possible mutants are determined, often, high-throughput 

systems are required to eliminate undesired variants since this

approach can yield 104–1012  variants based on the number of 

amino acids being considered for mutations.123  Although it 

seems to be a laborious process, many valuable industrial

catalysts have been discovered through this method. One of 

the most advantageous features of using combinatorial protein

design is that extensive knowledge of the protein structure is

not always required. Primary sequence is the main tool for

determining the various combinations of mutations that can be

implemented to yield a satisfactory biocatalyst using computa-

tional algorithm and consideration of the energy landscape.

4.3.4. Examples of combinatorial protein design.  Case I. A drug produced by Merck to combat asthma is Singulair or

montelukast sodium.134 This molecule is sensitive to many 

chemical processes including hydrogenations. One of the

compounds used during its synthesis is ()- B-chlorodiisopino-

campheylborane, () DIP-Cl, to create the one stereocenter in

the molecule (Fig. 14). () DIP-Cl is, however, corrosive and

Fig. 14   From Liang et al.; the synthesis of Singulair  via  () DIP-Cl as shown by Liang  et al.  Produced through the reduction of molecule  1, (E )-methyl 2-(3-(3-(2-(7-

chloroquinolin-2-yl)vinyl)phenyl)-3-oxopropyl) benzoate, Singulair has one stereocenter as shown by molecule two labeled ( S), stating the chirality.

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 17/32

6550   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

moisture sensitive as well as failing to be environmentally 

friendly. Thus, Liang  et al.  set out to create a biocatalyst that 

can replace () DIP-Cl. KetoREDuctase (KRED) is the enzymethey mutated using error prone PCR. The mutated enzyme

is able to catalyze the reduction of an asymmetric ( E )-methyl

2-(3-(3-(2-(7-chloroquinolin-2-yl)vinyl)phenyl)-3-oxopropyl) benzoate

to its (S)-alcohol (Fig. 15). The enantioselectivity of the enzyme

 was increased to more than 99.9%, a ‘self-driven’ process due to

the precipitation of the product out of the reaction solution.134

In addition, the enzyme is water soluble and stable in 70%

organic solvents at 45   1C with a loading of 100 g L1.134 This

property yields a still active enzyme because layers of hydration

are present at the enzyme’s active site; therefore, the required

environment for catalysis is maintained while in organic–water

solution; the hydrophobic organic solvent is water miscible and

organic substrate, which would otherwise be insoluble in water, will become accessible to the enzyme under these conditions.

 Additionally, the enzyme uses NAD or NADPH as a cofactor,

 which is also regenerated during the reaction by KRED. This

feature is highly advantageous for industrial applications since

these cofactors can be highly costly.

Case II.   Similar to enzymes used in therapeutic practices,

phytases, also known as myo-inositol hexakisphosphate phos-

phohydrolases, are enzymes supplemented in animal feed to

help increase phosphorous absorption while decreasing its

secretion.135 The positive attributes of this enzymes included

optimum in acidic pH, pepsin digestion resistance, and efficiency 

in metabolizing phylate. However, to create a thermostable variant of the enzyme, directed evolution through error-prone

PCR was used and yielded a variant with an increased 6–7   1C

in T m. Additionally, thermostability was improved to 10 minutes

at 80   1C, keeping in mind that the enzyme’s environment 

in-body with temperature of  B37   1C! At pH 3.5, the catalytic

efficiency was improved by 56% and 152% for the two superior

 variants, K46E and K65E/K97M/S209G. The mutations that 

rendered these stabilizing features were found to strengthened

local interactions within the protein.135 Error-prone PCR can

 yield a myriad of variants, some of which are the desired stable

proteins as shown by the example from Kim and Lie.135 While a

process that is easily introduced into any lab, selecting the

stable variants can be laborious.

Case III.  To improve the half-life of wild-type  b-glucosidase

from   P. polymyxa   of 15 minutes at 35   1C and no detectable

activity at 55   1C, some random variants that enhanced thermal

resistance of the enzyme were combined to achieve a half-life of 

12 min at 65   1C, with retention of kinetic parameters. Analysis

of the putative causative agents for the increased thermalresistance were (i) the formation of an extra salt bridge, (ii)

the replacement of an Asn residue exposed to the solvent, (iii)

stabilization of the hydrophobic core due to tighter packing,

and stabilization of the quaternary structure of the protein.136

Case IV. A xylanase for animal feed as well as pulp and paper

applications was sought by Diversa (San Diego, CA) that could

 withstand very high temperatures, above 90   1C. In the first 

round, all possible 19 amino acid substitutions at each residue

position were generated by Diversa’s own gene site saturation

mutagenesis (GSSM) and screened for activity after a tempera-

ture challenge. After identifying nine mutations that were

responsible for enhanced thermostability, all 512 possiblecombinatorial variants of the 9 mutations (= 29) were then

generated and screened for improved thermal tolerance. A total

of 11 variants were found with melting temperatures   T mincreased from 61  1C to as high as 96  1C! Thus, the consecutive

application of two evolution strategies enabled the rapid

discovery of the enzyme variant with the highest degree of 

fitness,   i.e.   greater thermal tolerance at constant activity at 

comparable temperatures.137

4.3.5. Data-driven protein design.   Data-driven protein

design is similar to rational design in that it is based on

acquired information on the protein of interest or on ortho-

logous or paralogous proteins. The difference between the two

methods is that data-driven protein design’s predictions aremore accurate. Additionally, data-driven protein design is

advantageous over combinatorial protein design because it 

 yields a smaller size of protein library that is to be screened

 while still allowing for ‘error-prone PCR’ types of muta-

tions.138,139 Unlike rational design, data-driven’s predictions

are obtained from experiments that monitor the fold or unfolding 

sequence of the protein to determine ‘‘weak’’ spot, in the case

of rendering a protein more thermostable.140 Once that infor-

mation is determined, site-directed mutagenesis is performed.

By using this method, high-throughput assays are rarely 

needed, which reduces the cost and labor.

4.3.6. Examples of data-driven protein design. Case I. Data-driven design has been used by Hirose   et al.   to improve

solubility of proteins. Biocatalysts are often tagged to ease

purification or to improve stability.141 However, these tags

can also have detrimental effects on the protein’s structure

and functions.141 Thus, Hirose designed data-driven tags, or

DDTs, by considered the effect of hydrophobic and hydrophilic

tags. Unsurprisingly, their experiments showed that hydro-

philic tags increased solubility of proteins. They also showed

tags conferring solubility were negatively charged while those

that made protein less soluble were positively charged with

Fig. 15   Liang  et al.   depicted the reaction catalyzed by KRED. In the place of

()DIP-Cl is KRED. The enzyme regenerates thecofactors NAD(P)Hin thepresence

of isopropanol, which is oxidized into acetone.

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 18/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6551

arginine rich sequence.141 In addition, while the authors did

not focus on any specific biocatalyst, in fact they do not 

explicitly mention any of them, they showed that data-driven

protein design could be used to create stable biocatalyst and to

also help determine satisfactory tags, proteins or peptides that 

can help with the purification of a specific protein.

Case II.   The kinetic stability of chitinase B from   Serratiamarcescens (ChiB) (Fig. 16) was stabilized using semi-automated

design methods through rigidifying mutations such as Gly - Ala

and Xxx - Pro.142 Of the 15 stabilizing single variants, those

 with significant stabilization clustered in one surface-exposed

region of the enzyme, with the double variant G188A/A234P

showing tenfold increase in half-life at 57   1C and a 4.2   1C

increase in the apparent   T m. Likely, the mutations led to

entropic stabilization of ChiB, with partial unfolding in the

critical region in question, which triggered thermal deactiva-

tion coupled with aggregation.

Case III. In an attempt to further elucidate protein stability,

Rathi et al. used data-driven process to understand what allowsthermal stability or instability (Fig. 18).139 Citrate synthases

(CS) from five distinct host organisms each with optimal

growth temperature (T p) of 37   1C, 59   1C, 75   1C, 87   1C, and

100   1C were chosen for the study. Their sequences were

compared and their structures analyzed to find areas within

the proteins that would be more susceptible to mutations.

The Ensemble-based Computer Network Analysis (CNS) was

modified to take into account the idea that increased hydro-

phobic interactions, hydrogen bonds, and salt bridges result in

a more thermostable protein. Using two of their data sets

shown in Fig. 17, Rathi   et al.139 showed that the thermal

unfolding of CS happens in stepwise manners where the

structurally stable enzyme (Fig. 17II(a)) which has dimer inter-

actions between F, G, M, and L, decomposes into a transition

state in which there is strong core interactions between G, I, L,

and M regions of the enzyme, corresponding to 330–360 K or an

enthalpy (H) of about 1 (Fig. 17II(b) and (c)). Next, with highertemperatures, the enzyme breaks down at the core region into

final smaller rigid structures connected by flexible links

(Fig. 17II(d) and (e)).

 Additionally, Rathi showed that the thermal stability and

unfolding mechanisms of the CS orthologs had no correlation

 while the thermostability of these orthologs showed to have a

significant correlation with the  T m or  T p of the host organisms

( R2 = 0.88). They also found that each orthologs had different 

sets of weak spots, which are defined as the region of tertiary 

structures whose interactions are disturbed at the T p. Sequence

analysis showed that these regions of weak spots were mutated

in the higher stable orthologs. In all, using data-driven strate-gies, they showed that by understanding the thermostability 

rather than thermal unfolding, one could predict amino acids

involved in stabilization of thermozyme orthologs.

Case IV, V. One of the most popular drug problems is cocaine

addiction. Much research has been devoted to developing drugs

that combat cocaine addictions and one of the avenues is the

engineering of cocaine esterase (CocE) and butyrylcholinesterase

(BChE),   a/b   hydrolases.145,146 CocE is an enzyme found in

bacteria,   i.e. Rhodococcus sp   strain MB1, that use cocaine as

their sole carbon source.147 The enzyme hydrolyzes cocaine into

Fig. 16   Chitinase B from  Serratia marcescens.  PDB: 2WK2.143 Mechanism of Chitin hydrolysis by Chitinase B was adapted from Tews  et al.144

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 19/32

6552   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

benzoic acid and Ecgonine methyl ester. On the other hand,

BChE is a human protein that catalyzes the hydrolysis of 

acetylcholine into acetate and choline (Fig. 18).148 Several

stable variants of both enzymes have been generated using 

site-directed mutagenesis based on rational design.73,74 How-

ever, Huang  et al.  and Zhen et al.  were able to design thermo-

stable variants of CocE and BChE, respectively, based on data

obtained from computer simulations.149,150

Fig. 17   (I) Rathi et al.139 showed that a cartoon of CS (top figures I-A and I-B, 901 rotation of I-A) in which each letter points to the section of the protein it represents.

(II) Bottom figure shows the melting curve of CS, which happens in a step wise manner. Structure (a) represents a stable form of the protein with all interactions intact;

(b) and (c) are intermediate while (d) and (e) are inactive and lost significant local interactions.

Fig. 18   (a) Cocaine (molecule 1) hydrolyzed by butyrylcholinesterase (BChE), human liver carboxylesterase (hCE-1/2), or Cocaine Esterase (CocE).147 (b) Deacetylation

of choline esterase, a reaction catalyzed by BChE.148

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 20/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6553

In the case of CocE, Huang   et al.   performed molecular

dynamics simulations on wild type CocE, which has a half-life

of 11 minutes at 37   1C. Based on literatures and simulations,

Huang   et al.   determined that the lack of rigidity within the

active site is responsible for instability of the enzyme. The data

suggested that mutation L169K would yield a thermostable

enzyme, which was confirmed by wet-lab experiments with a

determined half-life of 570 minutes at 37  1C, equaling to more

than 50-fold or 5000% increase in stability! A finding that canbe used towards the stabilizations of other a/b hydrolases such

as a-amino ester hydrolase and platelet-activating factor acetyl-

hydrolase, a lipase from  Streptomyces exfoliates.147,151

 While Huang   et al.   focused on thermal stability of CocE,

Zheng  et al. tackled the kinetic stability of BChE metabolism of 

cocaine since the therapeutic enzyme must be of superior

thermostability and catalytic efficiency. Zheng   et al.   focused

on understanding the mechanism of catalysis with the goal to

stabilize the transition state and improve the kinetics of the

rate-limiting step.149 The group produced a quadruple variant 

of BChE, A199S/S287G/A328W/Y332G, which has a 456-fold

improvement in catalytic efficiency, a promising outcome foranti-cocaine addiction therapeutics developments!149

Both Huang  et al. and Zheng  et al. showed that by increasing 

rigidity of the active site without disrupting the structure of the

enzyme and by understanding the catalytic mechanism of the

biocatalysts, one can stabilize the key parameters through

the mutation of one or more amino acids. The key is to stabilize

the active site and the binding of the intermediate substrate

through charge–charge interactions, hydrophobic interactions,

etc.  These methods of computational studies with data-driven

approach can be applied in the design of myriad of drugs, and

other industrial biocatalysts.149,150

Stabilization based on major algorithms. Case VI.  Moleculardynamics (MD) simulations were employed to identify flexible

regions in haloalkane dehalogenase (DhlA), which can serve as

a target for enhancing stability  via   introduction of a disulfide

bond. MD simulations of the  a/b-hydrolase fold protein DhlA 

showed high mobility in a helix-loop-helix region in the five

a-helix cap domain, involving residues 184–211. A disulfide

cross-link was engineered between residue 201 of this flexible

region and residue 16 of the main domain. In ‘wet-lab’ experi-

ments, the mutant enzyme showed substantial changes in

both thermal and urea denaturation. The apparent transition

temperature T m,app of the oxidized form of the variant increased

from 47.5 to 52.5   1C, whereas the T m,app of the reduced variant decreased by more than 8  1C compared to the wild-type enzyme;

denaturation with urea followed a similar trend. The Gibbs free

energy of unfolding was decreased by 0.43 kcal mol1 in the

 variant compared to the wild-type enzyme, also indicating that 

the helix-loop-helix region was involved early in the unfolding 

process. Thus, MD simulations can identify mobile protein

domains that can be targeted successfully for stabilization  via

disulfide cross-linkage.152

Case VII.   To increase the efficiency of data-driven protein

engineering based on iterative saturation mutagenesis (ISM),

adaptive substituent reordering algorithm (ASRA) was intro-

duced. ISM presented as an alternative to traditional quantita-

tive structure–activity relationship (QSAR) methods for

identifying potential protein mutants with desired properties

from minimal sampling of focused libraries. ASRA delivers

predictions without explicit knowledge of the structure–

property relationships by identifying the underlying regularity 

of the protein property landscape. In a proof-of-principle study,

 ASRA identified all or most of the best enantioselective variantsof  Aspergillus niger  epoxide hydrolase.153

 B-FIT concept. Case VIII.   Iterative saturation mutagenesis

(ISM) was applied to thermally stabilized lipase A (LipA) from

 Bacillus subtilis   (Fig. 19),   i.e.   sites in LipA were subjected to

saturation mutagenesis and the best hit of any given library is

then used as a template for randomization at other sites, and

the procedure is repeated until the desired improvement level

of the chosen property has been achieved. The randomization

sites in LipA were chosen on the basis of the highest B-factors

available from X-ray crystallography data of the wild-type (WT)

LipA, termed the B-FIT approach. After five rounds, this proce-dure resulted in pentuple variants with melting temperatures

T m  increased by 45   1C that also showed improved reversibility 

after heating to temperatures above 65   1C and subsequent 

cooling.

The fitness-pathway landscape constructed from every 

possible sequence of the five point mutations,   i.e.   5! = 120

pathways, from the WT LipA to the thermostabilized variant XI,

 was analyzed for epistatic interactions. The analysis revealed

significant cooperative effects between distal residues, which

are especially pronounced in the final steps towards variant XI.

MD simulations uncovered the stepwise formation of an exten-

sive H-bond/salt-bridge network on the surface of the enzyme.

Most importantly, while all five point mutations are necessary for high thermostability, only one participates directly in the

extended network. The authors suggest that evolution and then

identification of structural and functional amino acid networks

deserve attention.154

Combined thermal deactivation profiles, CD and NMR 

experiments, as well as X-ray structure analyses on the three

best LipA variants and the wild-type yielded variations of 

surface amino acid residues that counteracted propensity to

precipitate at elevated temperatures. Thus, not increased

conformational stability (higher   T m) of the evolved variants

but rather reduction of precipitation of unfolding inter-

mediates seem to be responsible for improved thermal stressbehavior.155

Case IX.   Three positions on the surface of   Pseudomonas

 fluorescens   aryl esterase (Fig. 20) were selected according to

high flexibility as discerned from the B-factor iterative test 

(B-FIT) principle and targeted first   via   site-saturation muta-

genesis and then via restricted libraries utilizing only consensus-

based mutations to increase its thermostability. Restricted

libraries reduce the library size significantly while ensuring a

high hit rate. The best variants demonstrate significantly 

improved thermostability, 8   1C higher  T m  compared with the

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 21/32

6554   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

 wild type without compromising specific activity, whereas

subsequent iterative saturation mutagenesis (ISM) even gave a

 variant with a 9   1C increased   T m, again with unchanged

catalytic properties.156

Case X.   The B-FIT concept was applied to increase thethermal robustness of the epoxide hydrolase from  Aspergillus

niger . Several rounds of ISM resulted in the best variant show-

ing a 21  1C increase in the T 6050 value, an 80-fold improvement in

half-life at 60 1C, and a 184 kJ mol1 improvement in the energy 

of deactivation; seven other variants yielded lesser but still

respectable improvements. The best variants during the ISM

process surprisingly were obtained from neutral or even infer-

ior parent variants, from a library with few or no improved

 variants. Tapping into such libraries apparently avoids dead

ends in evolution, i.e. local minima in the fitness landscape.158

4.3.7. Stabilization via the consensus concept. The concept 

of stabilization via  the consensus method determines the most 

prevalent amino acid at a given position of an alignment of 

orthologous proteins. The resulting sequence of most prevalent 

amino acids is termed consensus sequence. Lehmann et al.159–161

and Amin   et al.162 later applied the consensus approach to

generate a stable fungal phytase and  b-lactamase, respectively.

 Watanabe   et al.163 used a phylogenetic approach to identify 

ancestral amino acids that ultimately conferred thermal stabi-

lity to a 3-isopropylmalate dehydrogenase. Binz et al.164 used a

consensus approach to develop a conserved scaffold (i.e.,residues important for maintaining repeat structure) where

potential interacting amino acids where randomized en route

to improving interactions between repeating units of the

ankyrin protein. The result is an increase in expression, solu-

bility, and stability of the ankyrin repeat protein. DiTursi

et al.165 used Bayesian sequence-based algorithms on serine

protease sequences to identify a stabilizing motif in subtilisin

E, improving its melting temperature by 13   1C. In cases of low 

level of sequence identity and/or few available amino acid

sequences, however, the consensus residue of a specific

position cannot be found easily or not at all. Thus, the

structure-guided consensus approach was developed, whichuses structural information (in addition to other criteria, see

below) to reduce the number of residues in a given protein to be

mutated and checked for stabilization, and demonstrated

thermostabilization on penicillin G acylase (PGA) in 47%

(10 of 21) of investigated single variants (Fig. 21).166 Towards

this goal, there were only 8 sequences ranging from 32.5 to

87.7% and 28.6–87.6% amino acid identity for the   a- and

b-chains, respectively. Moreover, PGA is a large, heterodimeric

entity with a complicated maturation process to the active

protein (23 kDa and 63 kDa in  a- and  b-subunit, respectively)

Fig. 20   Aryl Esterase from  Pseudomonas fluorescens. PDB: 3T4U.157

Fig. 19   LipA from Bacillus subtilis.  PDB: 1ISP.

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 22/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6555

and thus difficult to access through rational or combinatorial

design. In the present work, the structure-guided consensus

approach was extended towards a simple, broadly applicable,

and highly successful approach on an important, homo-

tetrameric biocatalyst with cofactor-dependent catalysis, glucose

dehydrogenase.

In summary, it was demonstrated that the stabilizing effect 

of consensus mutations in multiple natural and diverse

 variants of the same functionality. Additionally, it was demon-

strated that the   structure-guided consensus concept   is a very successful and widely applicable technique for generating 

thermostable proteins160,166,168—further demonstrated in the

improvement in the stability of GDH. The stability of the

his-tagged   B. subtilis   GDH was improved from  B20 minutes

at 25 1C toB3.5 days at 65  1C, a 106-fold improvement.169 Using 

this approach it was possible to identify residues critical to

stability with considerably less effort than traditional directed

molecular evolution: of order 10  vs.  order 104. The number of 

investigated variants can be tuned by the experimentalist; the

two most influential factors are the amino acid identity, range

of the selected sequences, and the chosen consensus level (here

50%). In contrast to directed evolution, where key amino acidsare first identified  via   high-throughput screening (HTS), each

 variation from the parent sequence has to be incorporated into

the parent gene, and the variant purified and characterized. In

the case of difficult and/or expensive substrates, not amenable

to HTS, a small number of variants can be characterized

directly on the substrate of interest, a strong advantage

(‘‘you get what you screen for’’).170 The success of the struc-

ture-guided consensus concept depends on the diversity of 

available sequences. The number of consensus candidates

can then be reduced with available information, such as

(i) data on stability from thermostable or labile homologs,171,172

(ii) crystallographic structures, including B-factor analysis173

and Ramachandran plots,174,175 and (iii) information about the

deactivation mechanism, such as crucial subunit interactions,

 which might prefer surface residues. Often, use of all available

data and tools optimizes output while minimizing effort. As

discussed, if the search was limited to the subunit–subunit 

interface, the success rate would have increased from 46% to

86%. The development of thermostable enzymes is of interest 

to the specialty chemical and pharma industries but also to theprotein engineer, since recent work has shown that evolvability 

is linked to stability.176,177 This finding supports the idea that 

protein engineering projects will benefit from starting with a

stable protein scaffold. Thus, straightforward, time-saving tech-

niques—such as the structure-guided consensus concept—is

projected to prove invaluable to protein engineering efforts.

4.3.8. Stabilization   via   SCHEMA.   The discovery of exons

and introns gave birth to one evolutionary theory, which states

that proteins have evolved through nonhomologous gene

recombination.178 In 1978, W. Gilbert proposed that RNA 

splicing presents a faster opportunity for novel proteins to be

created from older ones when compared to rare single point mutations.179,180 This concept essentially lead to the idea that 

heteromeric proteins may have evolved from the recombination

of the exons of separates genetic material whether through

horizontal or vertical gene transfer.178,181 The structure of the

individual domains from the exons of these proteins would

therefore be independently stables. Enzyme stabilization

through rational design, combinatorial design, or data-driven

design can require an extensive amount of work in an expansive

time frame. Laboratory directed evolution uses combinatorial

shuffling of DNA to create stable enzymes in a timely fashion.178,182

Fig. 21   Penicillin G Acylase from E. Coli.  PDB: 1GM9.167

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 23/32

6556   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

This method of stabilization is dubbed SCHEMA.183,184

‘Schema’ is the original computer algorithm, which searched

for genetic materials whose crossovers would yield favorable

interactions. While ‘schema’ focused on genetic materials, Voigt 

et al.   created SCHEMA, which applied the same algorithmic

concept onto protein sequence; the program attempts to identify 

structural domains from distinct proteins that must have origi-

nated from one ancestor, relating back to Gilbert’s hypothesis.179,184

 Additionally, by scoring the interactions between amino acidsof two parent proteins, SCHEMA can help determine which

interactions, when removed, would not cause the instability 

of the parent proteins’ structures and would yield a stable

chimeric protein from the recombination of those parent 

proteins.183,185  Voigt calculated the disruption,  E ab, of hybrid

protein, from the recombination of fragment   a   of parent 

protein 1 and fragment   b   of parent protein 2 and where

 when residue i and j are within a specified distance,   C ij   = 1

and if not, C ij  = 0. P ij  simply accounts for the possibilities of no

disruption.184 This method of stabilization is more efficient 

that error-prone PCR used in combinatorial design since large

amounts of amino acids are changes simultaneously to create anew hybrid protein with desired features.185

E ab  ¼Xi 2a

Xi 2b

cij Pij    (39)

4.3.9. Examples of stabilization   via   SCHEMA.   Case I.

SCHEMA has already been employed in research to create

biocatalysts more stable than their already available industrial

counterparts.186–188 In an attempt to understand the stability of 

arginase, which is an enzyme that catalyzes the hydrolysis of 

L-arginine into ornithine, Romero et al. used SCHEMA. Arginases

(Fig. 22) are potential therapeutic agents for   L-arginine auxo-

troph tumors and hyperargininemea. Thus, making them stable

is valuable.189 Recombination of Arginase I and II showed that pI

of the proteins is very significant for long-term thermostability.

The higher the pI of the arginase, the lower the kinetic stability 

of the chimera.189 Essentially, SCHEMA can be used to not only 

create novel and stable proteins but to understand how evolution

have chosen certain amino acids, as stated in the neutral drift 

theory, to yield suitable enzymes for the host organism.

Case II.   Similarly, Production of biofuels and biomass

requires a source of sugar, which is commonly obtained fromthe breakdown of cellulose by cellulase.186–188 Heinzelman

et al. have shown that by using structure-guided recombination

of three fungal class II cellobiohydrolases (CBH II), several

cellulases with thermostability of 7   1C to 15   1C higher along 

 with increased pH profile all with increased activity were

created.187,188 that a single mutation, C313S, was responsible

for the stabilization; stable variants were created when this

mutation is implemented in the parent proteins,   Hypocrea

  jecorina  and   H. insolens  CBH II, and in a distant protein with

only 55–56% identical features,   Phanerochaete chrysosporium

CBH II.187 These findings demonstrate that results from

structure-guided recombination SCHEMA can inform us of mutations that were selected through evolution to improve

thermodynamics and kinetics stability of orthologous enzymes.

Case III.  Often, the disadvantage of biodegradation of bio-

mass is the thermostability of the cellulase. A thermostable

cellulase can be efficient at higher temperature and have longer

half-life, which both equals to greater amount of biodegrada-

tion.190 Heinzelman et al. showed that SCHEMA can sometimes

produce combinations that require high throughput screen-

ings.190 They were able to develop a method for finding stable

blocks of amino acids; a thermostable recombinant cellobio-

hydrolase class I (CBH I) was isolated from 390625 (58) possible

chimeras. Heinzelman   et al.   begun by analyzing the parent 

Fig. 22   Arginase from  Themus thermophilus.  PDB: 2EF4.

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 24/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6557

enzymes and the singular units, which were considered for

recombination. They identified 36–40 stable blocks that con-

densed into 16 thermostable CBH Is, which were more stable

than the most thermostable CBH I (Talaromyces emersonii ) at 

70   1C.190 SCHEMA can be used to create novel stable recombi-

nant with improved or diversified functions. By studying and

choosing thermostable blocks of amino acids that make up the

chimera and/or by incorporating various catalytic domains, one

can make a desirable recombinant protein with diverse functionscan be created and implemented in industrial applications.

Case IV.   Industrial processing of biomass relies on the

mixture of various cellulases and glycoside hydrolases with

diverse functions.191 Thus, Smith   et al.   demonstrated that 

understanding structural differences within a family of protein

could help interpret their evolution as it relates to their

specificity as well as create myriads of protein with divergent 

functions. Thus, using structure-guided recombination, Smith

et al.   were able to increase the library of examined Cel48

(Fig. 23) from 13 to 73 by using the catalytic domains of three

glycoside hydrolase family 48 bacterial cellulases (Cel48). The

three parent enzymes include   Clostridium cellulolyticum  CelF,Clostridium stercorarium   CelY, and   Clostridium thermocoellum

CelS and yield 60 recombinants with various functionalities

governed by pH, temperature, stability, and specific activity 

 with crystalline cellulose. Similar to Heinzelman et al., Smith

et al. studied the individual parent protein fragments and found

that the stability of each fragments is additive in a recombinant,

based on their linear models of sequence property.191

Case V. One of the most important tools when working with

proteins is protein tags. Tags can be used for stabilization or

purification of a protein of interest. Avidin and steptavidin

(Fig. 24) are two of the most common tags used for protein

purification. Either enzymes can be linked to a protein of 

interest and during purification, they are able to bind to

anchored biotins from which they will only be released at the

addition of a certain salt concentration and ionic strength. The

reverse of that scenario is also applicable: a protein modified

 with biotin can bind to an anchored avidin. Maatta   et al.

produced a thermostable avidin called chimeric avidin by 

recombining avidin and avidin-related protein 4.194  When the

recombinant protein was studied in organic solvents and

elevated temperatures, it showed to have superior stability in

extreme pHs starting from pH 1 to 13 as well as improvedstability in organic solvents such as DMF and methanol; avidin

 was stable at pH 13 while streptavidin was stable at pH 1.

 Additionally, the binding energy of chimeric avidin is energe-

tically favorable with an enthalpy of  112.6 kJ mol1, which is

5% higher than that of avidin. Maatta et al.  were able to show 

that the conferred stabilizations are due to increased stability of 

the tetrameric structures because of increased interactions

between the bridging amino acids.194 Maatta  et al.  developed

a tag that could be used for extended levels of purification due

to tag’s stability in very extreme conditions.

Case VI.  While the examples given thus far use SCHEMA to

create one or two stable variants of their target protein, Otey et al. used this method to create a library of 6561 of which are

3000 distinct properly folded variants of cytochrome P450

(Fig. 25).198 Each variant have different level of stability and

capability to incorporate a heme cofactor and peroxygenases

activity; some of the properly folded enzymes are inactive. The

recombination of three cytochromes P450s (CYP102A,

CYP102A2, and CYP102A3) at seven locations yielded variants

that are more than 73% dissimilar to any of the known

sequences of cytochromes P450, which is an astonishing 4500

sequences. Comparing the sequence of correctly folded variants

to the mis/unfolded variants and inactive yet well folded

recombinants can give insights as to what regions are necessary 

to maintain native fold and stability along with understanding 

Fig. 23   Cel48F from   Clostridium cellulolyticum. PDB: 1F9D.192 Reaction was adapted from the project of the month for the School of Theoretical Chemistry and

Biology, Royal Institute of Technology.193

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 25/32

6558   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

 what confers catalytic specificity and efficiency. This paperpresent a fine example of studying how enzymes have been

evolved and determining what amino acids are deleterious,

neutral, or advantageous using SCHEMA.198

Case VII.   The low thermostability of sucrose phosphorylase

from Bifidobacterium adolescentis  is a significant drawback for

applications in glycosylations. After substituting the most 

flexible residues with the consensus amino acids, & based on

careful inspection of the crystal structure, introducing residues

that promotes electrostatic, the half-life of a hexuple variant at 

the industrially relevant temperature of 60   1C more than

doubled over that of the wild type. Although mostly at low 

temperatures, the hexuple variant also was more stable against organic solvents.199

4.5 Improvement of process stability 

The operational stability of an immobilized lipase from

Thermomyces lanuginosa, commercially available as lipozyme

TL IM was investigated in the interesterification of two lipid

blends, blend A, a 55: 25: 20 wt% mixture of palm stearin

(POS), palm kernel oil (PK), and sunflower oil, and blend B,

formed by a 55: 35: 10 wt% mixture of POS, PK, and highly 

concentrated (n     3)-polyunsaturated fatty acids (PUFAs).

The continuous packed-bed bioreactor operated in neat mediumat 70   1C at a residence time of 15 min for 580 h (blend A) and

390 h (blend B), respectively. The inactivation of the biocatalyst 

followed a first-order deactivation model, with estimated half-

lives of 135 h and 77 h for blends A and B, respectively. The

higher levels of oxidation-prone PUFAs in blend B may explain

the lower lipase stability compared to blend A.200

5. Case studies

5.1 Record-setting stabilizations

In an especially successful example of thermal stabilization

through the application of two different methods of combina-torial protein engineering, gene site saturation mutagenesis

(GSSM) and exhaustive mutation among nine sites with a

xylanase gene, Palackal et al.137 improved the melting tempera-

ture T m of the xylanase from 61   1C to as high as 96   1C!

 Among data-driven protein engineering works, two examples

stood out:

 With the help of the B-FIT approach, Reetz  et al.154 found

that the melting temperature   T m   of a pentavariant of lipase

 A (LipA) from Bacillus subtilis, was increased by 45   1C, accom-

panied by improved reversibility. Simulations lend credence to

Fig. 24   (a) Avidin from  Gallus gallus  PDB: 1VYO;195 (b) Steptavidin from  Streptomyces avidinii; PDB: 3T6F.196 (c) Avidin–biotin complex from PDB: 2AVI.197

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 26/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6559

the notion that an extended network of charges and H-bonds is

responsible for the enhanced thermal stability.154

Lastly, the structure-guided consensus approach yielded a

 vastly stabilized glucose dehydrogenase (GDH). After analyzing 24 single variants identified by this approach, Vazquez-

Figueroa et al.169 succeeded to stabilize the his-tagged B. subtilis

GDH from a half-life of B20 minutes at 25   1C to B3.5 days at 

65   1C, a 106-fold improvement.

5.2 Simultaneous stabilization against heat and other stresses

such as organic solvents

 Whether stabilization against one stress,   i.e.   increasing 

temperature, also helps to stabilize against other stresses such

as those stemming from organic solvents, interfaces, or high

concentrations of chaotropes or salts in general, remains an

important hypothesis of protein stability. As results disproving 

this hypothesis often tend not to be disseminated, any report 

confirming simultaneous stabilization must be read with caution,as generalizability might not be assured.

5.2.1. Examples of simultaneous enzyme stabilization.

Case I. Cowan et al.201,202 showed a positive correlation between

thermal stability and tolerance to organic solvents in naturally 

occurring homologs of several different thermophilic proteins.

Hao and Berry demonstrated how a thermostable fructose

bisphosphate aldolase found via directed evolution also featured

increased stability in organic solvents.203

Case II.   Mutants of the lipase from Bacillus subtilis, pre-

 viously engineered for enhanced thermostability using directed

Fig. 25   (a) Otey  et al.198 show a cartoon depiction of the recombinant protein composed of 8 distinct protein fragments. (b) Depiction of how the numerous

recombinants were derived from the shuffling of the 7 distinct interaction regions.

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 27/32

6560   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

evolution based on the B-FIT method, show significantly 

increased tolerance to hostile organic solvents.204

Case III. A thermophilic Old Yellow Enzyme (‘TOYE’) (Fig. 26)

isolated from  Thermoanaerobacter pseudethanolicus E39 is stable

at high temperatures (T m   > 701) and simultaneously features

increased resistance to denaturation in water-miscible organic

solvents compared to a closely related mesophilic OldYellow Enzyme family member, pentaerythritol tetranitrate

reductase.205  According to sedimentation velocity and multi-

angle laser light scattering (MALS) measurements, TOYE adapts

higher-order oligomeric states, such as octamers and dodeca-

mers in solution, under stress compared to a tetrameric state at 

ambient conditions, likely a major reason for increased stability.

Case IV. Penicillin V acylase (EC 3.5.1.11) from  Streptomyces

lavendulae showed both enhanced activity and thermal stability 

in mixed water–glycerol and water–glycol solvents. Both activity 

and thermal stability increased proportionally to the amount of 

added glycerol and glycol co-solvents up to a critical concen-

tration of these co-solvents, with further addition leading to

gradual protein deactivation. Thus, reaction conditions that allow simultaneously enhanced activity and stability in the

S. lavendulae  Pen V acylase-catalyzed hydrolysis of penicillin V 

could be defined.113

Case V. Thermolysin-like protease from Bacillus stearothermo-

 philus   was engineered for extreme thermostability   via   intro-

duction of a disulfide bond G8C/N60C (double variant, DV) plus

six other amino acid substitutions in the exposed loop region,

amino acids 56–69, to generate a protein dubbed ‘Boilysin,

BLN’. Both DV and BLN were probed for stability against water-

miscible organic solvents and detergents. The solvent concen-

trations at which 50% of initial enzyme activity were irreversibly 

lost,   C 50, decreased in the order methanol > 2-propanol >dimethylsulfoxide > dioxane > acetonitrile > DMF > acetone.

However, the   C 50   values were remarkably higher for the BLN

and DV thermostable variants than for the wild-type enzyme,

pointing to a similar mechanism for irreversible thermal and

solvent deactivation. However, the differences of the  C 50 values

of DV and BLN were not significant, while the corresponding 

T 50   values of DM and BLN differed by 10 K. 207 Detergents

affected DV and BLN equally, inactivating (SDS, sulfobetaine)

or strongly activating (CTAB, CHAPS), and depending in degree

on detergent concentration but always destabilizing; Triton

X-100 and Tween 20 started inactivating beyond a threshold

concentration. Thus, the mechanism of detergent inactivation

seems to differ from that of thermal inactivation.208

Case VI.   Encapsulation in a carboxymethylcellulose (CMC)-

Ca alginate liquid membrane optimized at 4% CMC, 1% CaCl2,

and 3% alginate led to >60% activity retention of the cysteine

protease cathepsin B purified from goat brain during fivebatches. The optimum pH value, maximum temperature  T max ,

and apparent   K m   value of the free and immobilized enzyme

 were measured to be 6.0, 50  1C, and 1.52 mM, & 6.0–6.5, 55  1C,

and 2.3 mM, respectively. Microencapsulation significantly 

stabilized the enzyme against pH, thermal, and also solvent 

instability, opening up the potential for use in catalyzing 

transesterifications or transamidations.209

Case VII.   Formate dehydrogenase (FDH) from   Candida

boidinii , as cofactor-regeneration enzyme, had been found to

be much less stable when regenerating NAD+ to NADH than

the production enzyme, an amino acid dehydrogenase, when

reacting NADH to NAD+. Already in 2000, variants with free Cys

residues mutated to Ser or Ala were found to stabilize FDH.210 Asthe variants C23S and C23S/C262A were found to be advanta-

geous even in the absence of oxygen, other stresses were tested.

The two changes were found also to stabilize FDH against 

interfaces with nitrogen bubbles, shear stress in Couette

 viscometers, and shear stress and cavitation in gear pumps.87

Case VIII.   Thermostable glucose dehydrogenase (GDH) var-

iants developed   via   an amino acid sequence-based consensus

method also showed improved stability in solutions with high

concentrations of either kosmotropic or chaotropic salts as well

as water-miscible organic solvents. Only the most stable variants

showed little deactivation dependence on salt-types along the

Hofmeister series and on salt concentration. Kinetic stability,expressed by the deactivation rate constant  k d,obs, did not always

correlate with thermodynamic stability of variants, as measured

by melting temperature,   T m. However, a strong correlation

( R2 > 0.95) between temperature stability and organic solvent 

stability was found when plotting  T 6050 versus C 6050 values.211

6. Conclusions

Nowadays, a vast set of tools exists to stabilize biocatalysts,

consisting mostly of immobilization, medium engineering, and

Fig. 26   Old yellow enzyme from  Thermoanaerobacter pseudoethanolicus. PDB: 3KRU.205 Reaction was adapted from Brenna  et al.206

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 28/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6561

protein engineering. When measuring enzyme stability, thermo-

dynamic stability, measured by the melting temperature   T m,

has to be differentiated from both kinetic stability, measured by 

the deactivation rate constant  k d  or half-life   t, or operational

stability, also termed process stability, measured by the total

turnover number TTN, which scales enzyme activity over kinetic

stability. Protein engineering has been found to increase T m by 

35–45   1C. Through data-driven protein engineering, kinetic

stabilization of several 100 000-fold has been achieved. TTNhas been found to improve by the same order of magnitude. In

contrast to stabilization against detergents or other specific

chemical entities, which often feature interactions with specific

moieties on the enzyme surface, stabilization against one stress

factor, such as high temperature, was in many cases found to

stabilize against others, such as high concentrations of chao-

tropes as well as high shear through pumping, stirring, or

cavitation. Given the vast number of successful cases of enzyme

stabilization, almost any well-folded protein should be amen-

able to such procedures.

References

1 U. T. Bornscheuer and M. Pohl,   Curr. Opin. Chem. Biol.,

2001,  5, 137–143.

2 S. Lutz, Curr. Opin. Biotechnol., 2010,  21, 734–743.

3 S. K. Ma, J. Gruber, C. Davis, L. Newman, D. Gray, A. Wang,

 J. Grate, G. W. Huisman and R. A. Sheldon, Green Chem.,

2010,  12, 81–86.

4 C. A. Martinez, S. Hu, Y. Dumond, J. Tao, P. Kelleher and

L. Tully,  Org. Process Res. Dev., 2008,  12, 392–398.

5 C. K. Savile, J. M. Janey, E. C. Mundorff, J. C. Moore,

S. Tam, W. R. Jarvis, J. C. Colbeck, A. Krebber, F. J.

Fleitz, J. Brands, P. N. Devine, G. W. Huisman andG. J. Hughes,  Science, 2010, 329, 305–309.

6 C. E. Nakamura and G. M. Whited,  Curr. Opin. Biotechnol.,

2003,  14, 454–459.

7 K. M. Polizzi, A. S. Bommarius, J. M. Broering and

 J. F. Chaparro-Riggers,   Curr. Opin. Chem. Biol., 2007,   11,

220–225.

8 S. Bommarius and J. M. Broering,   Biocatal. Biotransform.,

2005,  23, 125–139.

9 V. G. H. Eijsink, A. Bjork, S. Gaseidnes, R. Sirevag,

B. Synstad, B. van den Burg and G. Vriend,   J. Biotechnol.,

2004,  113, 105–120.

10 V. G. H. Eijsink, S. Gaseidnes, T. V. Borchert and B. vanden Burg,  Biomol. Eng., 2005, 22, 21–30.

11 C. K. Winkler, D. Clay, S. Davies, P. O’Neill, P. McDaid,

S. Debarge, J. Steflik, M. Karmilowicz, J. W. Wong and

K. Faber, J. Org. Chem., 2013, 78, 1525–1533.

12 C. Breithaupt, R. Kurzbauer, F. Schaller, A. Stintzi,

 A. Schaller, R. Huber, P. Macheroux and T. Clausen,

 J. Mol. Biol., 2009,  392, 1266–1277.

13 B. W. Lepore, D. Liu, Y. Peng, M. Fu, C. Yasuda,

 J. M. Manning, R. B. Silverman and D. Ringe, Biochemistry,

2010, 49, 3138–3147.

14 M. Winn, J. M. Foulkes, S. Perni, M. J. H. Simmons,

T. W. Overton and R. J. M. Goss,   Catal. Sci. Technol.,

2012, 2, 1544–1547.

15 B. Rosche, X. Z. Li, B. Hauer, A. Schmid and K. Buehler,

Trends Biotechnol., 2009,  27, 636–643.

16 T. Hudlicky and J. W. Reed,   Chem. Soc. Rev., 2009,   38,

3117–3132.

17 M. Schrewe, M. K. Julsing, B. Buhler and A. Schmid, Chem.

Soc. Rev., 2013, DOI: 10.1039/C3CS60011D.18 J. E. Leresche and H.-P. Meyer, Org. Process Res. Dev., 2006,

10, 572–580.

19 J. A. Littlechild, J. Guy, S. Connelly, L. Mallett, S. Waddell,

C. A. Rye, K. Line and M. Isupov,   Biochem. Soc. Trans.,

2007,  35, 1558–1563.

20 F. Secundo, Chem. Soc. Rev., 2013, DOI: 10.1039/C3CS35495D.

21 M. T. Reetz and S. Wu,   J. Am. Chem. Soc., 2009,   131,

15424–15432.

22 T. Newhouse, P. S. Baran and R. W. Hoffmann, Chem. Soc.

 Rev., 2009,  38, 3010–3021.

23 M. Wang, T. Si and H. Zhao, Bioresour. Technol., 2012, 115,

117–125.24 D. Wistuba and V. Schurig,   Angew. Chem., Int. Ed. Engl.,

1986, 25, 1032–1034.

25 M. Wang, D. L. Roberts, R. Paschke, T. M. Shea,

B. S. S. Masters and J.-J. P. Kim,   Proc. Natl. Acad. Sci.

U. S. A., 1997, 94, 8411–8416.

26 D. Yi, T. Devamani, J. Abdoul-Zabar, F. Charmantray,

 V. Helaine, L. Hecquet and W.-D. Fessner, ChemBioChem,

2012, 13, 2290–2300.

27 P. Asztalos, C. Parthier, R. Golbik, M. Kleinschmidt,

G. Hubner, M. S. Weiss, R. Friedemann, G. Wille and

K. Tittmann, Biochemistry, 2007, 46, 12037–12052.

28 T. S. Institute,  Penicillin, http://www.scripps.edu/nicolaou/

pdfs/Sample_Chapter13.pdf.29 K. S. Goo and T. S. Sim,   Curr. Comput.-Aided Drug Des.,

2011, 7, 53–80.

30 P. L. Roach, I. J. Clifton, C. M. H. Hensgens, N. Shibata,

C. J. Schofield, J. Hajdu and J. E. Baldwin,  Nature, 1997,

387, 827–830.

31 S. S. Weber, R. A. L. Bovenberg and A. J. M. Driessen,

 Biotechnol. J., 2012,  7, 225–236.

32 J. D. Bloom and F. H. Arnold,  Proc. Natl. Acad. Sci. U. S. A.,

2009, 106, 9995–10000.

33 A. S. Bommarius, J. K. Blum and M. J. Abrahamson,  Curr.

Opin. Chem. Biol., 2011, 15, 194–200.

34 K. Hernandez and R. Fernandez-Lafuente, Enzyme Microb.Technol., 2011, 48, 107–122.

35 J. L. Foo, C. B. Ching, M. W. Chang and S. S. J. Leong,

 Biotechnol. Adv., 2012,  30, 541–549.

36 S. Robic,  CBE Life Sci. Educ., 2010,  9, 189–195.

37 R. Burioni, D. Cassi, F. Cecconi and A. Vulpiani,  Proteins:

Struct., Funct., Bioinf., 2004, 55, 529–535.

38 W. F. Li, X. X. Zhou and P. Lu,  Biotechnol. Adv., 2005,  23,

271–281.

39 T. Koudelakova, R. Chaloupkova, J. Brezovsky, Z. Prokop,

E. Sebestova, M. Hesseler, M. Khabiri, M. Plevaka,

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 29/32

6562   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

D. Kulik, I. K. Smatanova, P. Rezacova, R. Ettrich,

U. T. Bornscheuer and J. Damborsky,   Angew. Chem., Int.

 Ed., 2013,  52, 1959–1963.

40 T. Koudelakova, S. Bidmanova, P. Dvorak, A. Pavelka,

R. Chaloupkova, Z. Prokop and J. Damborsky,  Biotechnol.

 J., 2013,  8, 32–45.

41 J. K. Blum and A. S. Bommarius,   J. Mol. Catal. B: Enzym.,

2010, 67, 21–28.

42 J. K. Blum, M. D. Ricketts and A. S. Bommarius, J. Biotechnol., 2012,  160, 214–221.

43 Y. Harano, Entropy, 2012,  14, 1443–1468.

44 J. L. England and G. Haran, in  Annual Review of Physical 

Chemistry, ed. S. R. Leone, P. S. Cremer, J. T. Groves and

M. A. Johnson, 2011, vol. 62, pp. 257–277.

45 R. Singh, M. Tiwari, R. Singh and J.-K. Lee, Int. J. Mol. Sci.,

2013, 14, 1232–1277.

46 T. R. M. Barends, J. J. Polderman-Tijmes, P. A. Jekel,

C. M. H. Hensgens, E. J. de Vries, D. B. Janssen and

B. W. Dijkstra,  J. Biol. Chem., 2003,  278, 23076–23084.

47 G. De Simone, V. Menchise, G. Manco, L. Mandrich,

N. Sorrentino, D. Lang, M. Rossi and C. Pedone,   J. Mol. Biol., 2001, 314, 507–518.

48 A. Cipolla, F. Delbrassine, J.-L. Da Lage and G. Feller,

 Biochimie, 2012,  94, 1943–1950.

49 P. L. Wintrode and P. H. Arnold, in   Advances in Protein

Chemistry, Vol 55: Evolutionary Protein Design, ed. F. H.

 Arnold, 2001, vol. 55, pp. 161–225.

50 M. Camps, A. Herman, E. Loh and L. A. Loeb,  Crit. Rev.

 Biochem. Mol. Biol., 2007,  42, 313–326.

51 S. Bershtein, K. Goldin and D. S. Tawfik, J. Mol. Biol., 2008,

379, 1029–1044.

52 S. Bershtein and D. S. Tawfik, Curr. Opin. Chem. Biol., 2008,

12, 151–158.

53 S. G. Peisajovich and D. S. Tawfik,   Nat. Med., 2007,   4,991–994.

54 D. P. C. de Barros, P. Fernandes, J. M. S. Cabral and

L. P. Fonseca,   J. Chem. Technol. Biotechnol., 2010,   85,

1553–1560.

55 T. A. Rogers and A. S. Bommarius,  Chem. Eng. Sci., 2010,

65, 2118–2124.

56 P. Johnson and T. L. Whateley,   Biochem. J., 1981,   193,

285–294.

57 F. H. Niesen, H. Berglund and M. Vedadi,   Nat. Protocols,

2007, 2, 2212–2221.

58 G. A. Senisterra and P. J. Finerty,  Mol. BioSyst., 2009,   5,

217–223.59 J. V. Rodrigues, V. Prosinecki, I. Marrucho, L. P. N. Rebelo

and C. M. Gomes,   Phys. Chem. Chem. Phys., 2011,   13,

13614–13616.

60 O. Levenspiel,  Ind. Eng. Chem. Res., 1999,  38, 4140–4143.

61 R. O. Jenkins,  Appl. Organomet. Chem., 2004,  18, 373.

62 T. A. Rogers, R. M. Daniel and A. S. Bommarius,

ChemCatChem, 2009,  1, 131–137.

63 K. D. A. S. Bommarius, U. Groeger and C. Wandrey,

 Membrane Bioreactors for the Production of Enantiomerically

 Pure  a-Amino Acids, Wiley, Chichester, England, 1992.

64 A. S. Bommarius, K. Drauz, H. Klenk and C. Wandrey, in

 Enzyme Engineering Xi , ed. D. S. Clark and D. A. Estell,

1992, vol. 672, pp. 126–136.

65 C. Wandrey, habilitation thesis, TH Hannover, Germany,

1977.

66 R. M. Daniel, M. J. Danson and R. Eisenthal,   Trends

 Biochem. Sci., 2001,  26, 223–225.

67 M. E. Peterson, R. Eisenthal, M. J. Danson, A. Spence and

R. M. Daniel,  J. Biol. Chem., 2004, 279, 20717–20722.68 T. Ema, M. Kageyama, T. Korenaga and T. Sakai,   Tetra-

hedron: Asymmetry, 2003,  14, 3943–3947.

69 N. End and K. U. Schoning,   Immobilized Catalysts, 2004,

242, 273–317.

70 B. H. Zhang, Y. Q. Weng, H. Xu and Z. P. Mao,   Appl.

 Microbiol. Biotechnol., 2012,  93, 61–70.

71 D. A. Cowan and R. Fernandez-Lafuente,  Enzyme Microb.

Technol., 2011, 49, 326–346.

72 V. Grazu, F. Lopez-Gallego, T. Montes, O. Abian,

R. Gonzalez, J. A. Hermoso, J. L. Garcıa, C. Mateo and

 J. M. Guisan, Process Biochem., 2010,  45, 390–398.

73 D. A. Cecchini, R. Pavesi, S. Sanna, S. Daly, R. Xaiz,M. Pregnolato and M. Terreni,  Appl. Microbiol. Biotechnol.,

2012,  95, 1491–1500.

74 N. Brun, A. B. Garcia, H. Deleuze, M. F. Achard, C. Sanchez,

F. Durand, V. Oestreicher and R. Backov,   Chem. Mater.,

2010,  22, 4555–4562.

75 R. Backov, Soft Matter , 2006,  2, 452–464.

76 H. S. Kim, S. J. Lee and E. Y. Lee, J. Mol. Catal. B: Enzym.,

2006,  43, 2–8.

77 A. Fishman, I. Levy, U. Cogan and O. Shoseyov,   J. Mol.

Catal. B: Enzym., 2002,  18, 121–131.

78 P. Asuri, S. S. Karajanagi, H. C. Yang, T. J. Yim, R. S. Kane

and J. S. Dordick,  Langmuir , 2006,  22, 5833–5836.

79 P. Asuri, S. S. Karajanagi, A. A. Vertegel, J. S. Dordick andR. S. Kane,   J. Nanosci. Nanotechnol., 2007,  7, 1675–1678.

80 V. G. S. G. Bonavia, J. Manuel, R. F. Lafuente, O. A. Franco,

T. M. Fernandez, C. M. Gonzalez, R. G. Garcia, J. A. Hermoso

Domingues and J. L. Garcia Lopez, WO 2007138148 A1

20071206, 2007.

81 L. Wilson, A. Illanes, B. C. C. Pessela, O. Abian,

R. Fernandez-Lafuente and J. M. Guisan,   Biotechnol.

 Bioeng., 2004,  86, 558–562.

82 J. M. Bolivar, L. Wilson, S. A. Ferrarotti, R. Fernandez-

Lafuente, J. M. Guisan and C. Mateo,  Biomacromolecules,

2006, 7, 669–673.

83 J. Z. Setschenow,  Z. Phys. Chem., 1889,  4, 117–125.84 E. J. Cohn,  Proteins, amino acids and peptides as ions and 

dipolar ions, Reinhold Publishing Corporation, New York,

1943.

85 J. Hine,   Physical Organic Chemistry: Second Edition,

McGraw-Hill, New York, 1962.

86 C. H. Wong, D. G. Drueckhammer and H. M. Sweers, J. Am.

Chem. Soc., 1985,  107, 4028–4031.

87 A. S. Bommarius and A. Karau,  Biotechnol. Prog., 2005, 21,

1663–1672.

88 F. Hofmeister, Arch. Exp. Pathol. Pharmakol., 1888, 24, 247–260.

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 30/32

This journal is   c   The Royal Society of Chemistry 2013   Chem. Soc. Rev., 2013,   42, 6534--6565   6563

89 H. V. H. Peter and K.-Y. Wong, Science, 1964, 145, 577–580.

90 O. Sinanogl and S. Abdulnur, Fed. Proc., 1965, 24, S12–S23.

91 P. A. Grigorjev and S. M. Bezrukov,  Biophys. J., 1994,  67,

2265–2271.

92 J. G. Kirkwood,  Proteins, amino acids and peptides as ions

and dipolar ions, Reinhold Publishing Corporation,

New York, 1943.

93 W. Melander and C. Horvath,   Arch. Biochem. Biophys.,

1977, 183, 200–215.94 K. D. Collins,   Proc. Natl. Acad. Sci. U. S. A., 1995,   92,

5553–5557.

95 K. D. Collins,  Biophys. J., 1997, 72, 65–76.

96 K. D. Collins,  Methods, 2004,  34, 300–311.

97 K. D. Collins and M. W. Washabaugh,   Q. Rev. Biophys.,

1985, 18, 323–422.

98 G. Jones and M. Dole, J. Am. Chem. Soc., 1929, 51, 2950–2964.

99 J. M. Broering and A. S. Bommarius,  Biochem. Soc. Trans.,

2007, 35, 1602–1605.

100 J. M. Broering and A. S. Bommarius, J. Phys. Chem. B, 2005,

109, 20612–20619.

101 K. D. Collins, Proc. Natl. Acad. Sci. U. S. A., 1995, 92, 5553–5557.102 M. W. Washabaugh and K. D. Collins,   Q. Rev. Biophys.,

1985,  18, 323–422.

103 R. Lumry and H. Eyring, J. Phys. Chem., 1954, 58, 110–120.

104 J. M. Broering and A. S. Bommarius, J. Phys. Chem. B, 2008,

112, 12768–12775.

105 V. Yeh, J. M. Broering, A. Romanyuk, B. X. Chen,

Y. O. Chernoff and A. S. Bommarius,   Protein Sci., 2010,

19, 47–56.

106 M. Hall, J. Rubin, S. H. Behrens and A. S. Bommarius,

 J. Biotechnol., 2011,  155, 370–376.

107 J. K. Kaushik and R. Bhat,   J. Biol. Chem., 2003,   278,

26458–26465.

108 D. J. Boerema, V. A. Tereshko and S. B. H. Kent, Pept. Sci.,2008,  90, 278–286.

109 A. Nasiripourdori, H. Naderi-Manesh, B. Ranjbar and

K. Khajeh,  Int. J. Biol. Macromol., 2009,  44, 311–315.

110 H. A. Sathish, P. R. Kumar and V. Prakash,   Int. J. Biol.

 Macromol., 2007, 41, 383–390.

111 A. Tiwari and R. Bhat, Biophys. Chem., 2006,  124, 90–99.

112 M. Arroyo, R. Torres-Guzman, I. de la Mata, M. P. Castillon

and C. Acebal, Enzyme Microb. Technol., 2000, 27, 122–126.

113 M. Arroyo, R. Torres-Guzman, I. de la Mata, M. P. Castillon

and C. Acebal,  Biotechnol. Prog., 2000,  16, 368–371.

114 C. G. Suresh, A. V. Pundle, H. SivaRaman, K. N. Rao,

 J. A. Brannigan, C. E. McVey, C. S. Verma, Z. Dauter,E. J. Dodson and G. G. Dodson,  Nat. Struct. Biol., 1999,  6,

414–416.

115 D. A. Estell, T. P. Graycar, J. V. Miller, D. B. Powers,

 J. P. Burnier, P. G. Ng and J. A. Wells, Science, 1986,  233,

659–663.

116 J. A. Wells, D. B. Powers, R. R. Bott, T. P. Graycar and D. A.

Estell, Proc. Natl. Acad. Sci. U. S. A., 1987,  84, 1219–1223.

117 J. P. Colletier, A. Aleksandrov, N. Coquelle, S. Mraihi,

E. Mendoza-Barbera, M. Field and D. Madern,  Mol. Biol.

 Evol., 2012,  29, 1683–1694.

118 M. Soskine and D. S. Tawfik,   Nat. Rev. Genet., 2010,   11,

572–582.

119 F. H. Arnold, P. L. Wintrode, K. Miyazaki and

 A. Gershenson, Trends Biochem. Sci., 2001, 26, 100–106.

120 H. E. Schoemaker, D. Mink and M. G. Wubbolts,  Science,

2003, 299, 1694–1697.

121 A. A.-A.-F. Amara, Pak. J. Pharm. Sci., 2013, 26, 217–232.

122 R. Kourist, H. Jochens, S. Bartsch, R. Kuipers, S. K. Padhi,

M. Gall, D. Bottcher, H. J. Joosten and U. T. Bornscheuer,ChemBioChem, 2010,  11, 1635–1643.

123 J. G. Saven,  Curr. Opin. Struct. Biol., 2002,  12, 453–458.

124 S. G. Kang and J. G. Saven,  Curr. Opin. Chem. Biol., 2007,

11, 329–334.

125 M. Di Lorenzo, A. Hidalgo, R. Molina, J. A. Hermoso,

D. Pirozzi and U. T. Bornscheuer,  Appl. Environ. Microbiol.,

2007, 73, 7291–7299.

126 S. Cesarini, C. Bofill, F. I. J. Pastor, M. T. Reetz and P. Diaz,

 Process Biochem., 2012,  47, 2064–2071.

127 D. J. Opperman and M. T. Reetz, ChemBioChem, 2010, 11,

2589–2596.

128 J. T. Park, J.-I. Hirano, V. Thangavel, B. R. Riebel and A. S.Bommarius, J. Mol. Catal. B: Enzym., 2011,  71, 159–165.

129 M. Katzberg, N. Skorupa-Parachin, M.-F. Gorwa-Grauslund

and M. Bertau, Int. J. Mol. Sci., 2010,  11, 1735–1758.

130 Y. Ito, A. Ikeuchi and C. Imamura,  Protein Eng., Des. Sel.,

2013, 26, 73–79.

131 H. Park, J. Joo, K. Park and Y. Yoo,  Biotechnol. Bioprocess

 Eng., 2012,  17, 722–728.

132 Z. Qian, J. R. Horton, X. Cheng and S. Lutz,   J. Mol. Biol.,

2009, 393, 191–201.

133 A. Badoei-Dalfard, K. Khajeh, S. M. Asghari, B. Ranjbar and

H. R. Karbalaei-Heidari,  J. Biochem., 2010,  148, 231–238.

134 J. Liang, J. Lalonde, B. Borup, V. Mitchell, E. Mundorff,

N. Trinh, D. A. Kochrekar, R. Nair Cherat and G. G. Pai,Org. Process Res. Dev., 2009,  14, 193–198.

135 M.-S. Kim and X. Lei, Appl. Microbiol. Biotechnol., 2008, 79,

69–75.

136 G. Gonzalez-Blasco, J. Sanz-Aparicio, B. Gonzalez,

 J. A. Hermoso and J. Polaina,   J. Biol. Chem., 2000,   275,

13708–13712.

137 N. Palackal, Y. Brennan, W. N. Callen, P. Dupree, G. Frey,

F. Goubet, G. P. Hazlewood, S. Healey, Y. E. Kang,

K. A. Kretz, E. Lee, X. Q. Tan, G. L. Tomlinson,

 J. Verruto, V. W. K. Wong, E. J. Mathur, J. M. Short,

D. E. Robertson and B. A. Steer,   Protein Sci., 2004,   13,

494–503.138 J. F. Chaparro-Riggers, K. M. Polizzi and A. S. Bommarius,

 Biotechnol. J., 2007, 2, 180–191.

139 P. C. Rathi, S. Radestock and H. Gohlke,   J. Biotechnol.,

2012, 159, 135–144.

140 S. Radestock and H. Gohlke, Eng. Life Sci., 2008, 8, 507–522.

141 S. Hirose, Y. Kawamura, M. Mori, K. Yokota, T. Noguchi

and N. Goshima, New Biotechnol., 2011,  28, 225–231.

142 B. Synstad, S. Gaseidnes, D. M. F. van Aalten, G. Vriend,

 J. E. Nielsen and V. G. H. Eijsink, Eur. J. Biochem., 2004,

271, 253–262.

Review Article Chem Soc Rev

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 31/32

6564   Chem. Soc. Rev., 2013,   42, 6534--6565   This journal is   c   The Royal Society of Chemistry 2013

143 J. M. Macdonald, C. A. Tarling, E. J. Taylor, R. J. Dennis,

D. S. Myers, S. Knapp, G. J. Davies and S. G. Withers,

 Angew. Chem., Int. Ed., 2010,  49, 2599–2602.

144 I. Tews, A. C. T. vanScheltinga, A. Perrakis, K. S. Wilson

and B. W. Dijkstra, J. Am. Chem. Soc., 1997, 119, 7954–7959.

145 D. Gao, D. L. Narasimhan, J. Macdonald, R. Brim,

M.-C. Ko, D. W. Landry, J. H. Woods, R. K. Sunahara and

C.-G. Zhan, Mol. Pharmacol., 2009,  75, 318–323.

146 J. M. Turner, N. A. Larsen, A. Basran, C. F. Barbas,N. C. Bruce, I. A. Wilson and R. A. Lerner,  Biochemistry,

2002,  41, 12297–12307.

147 N. A. Larsen, J. M. Turner, J. Stevens, S. J. Rosser,

 A. Basran, R. A. Lerner, N. C. Bruce and I. A. Wilson, Nat.

Struct. Biol., 2002,  9, 17–21.

148 C. J. Rogers, L. M. Eubanks, T. J. Dickerson and

K. D. Janda,  J. Am. Chem. Soc., 2006,  128, 15364–15365.

149 F. Zheng and C.-G. Zhan, J. Comput. Aided Mol. Des., 2008,

22, 661–671.

150 X. Huang, D. Gao and C.-G. Zhan,   Org. Biomol. Chem.,

2011,  9, 4138–4143.

151 Y. Wei, L. Swenson, C. Castro, U. Derewenda, W. Minor,H. Arai, J. Aoki, K. Inoue, L. Servin-Gonzalez and

Z. S. Derewenda, Structure, 1998,  6, 511–519.

152 M. G. Pikkemaat, A. B. M. Linssen, H. J. C. Berendsen and

D. B. Janssen, Protein Eng., 2002,  15, 185–192.

153 X. J. Feng, J. Sanchis, M. T. Reetz and H. Rabitz,

Chem.–Eur. J., 2012,  18, 5646–5654.

154 M. T. Reetz, P. Soni, J. P. Acevedo and J. Sanchis,  Angew.

Chem., Int. Ed., 2009,  48, 8268–8272.

155 W. Augustyniak, A. A. Brzezinska, T. Pijning, H. Wienk,

R. Boelens, B. W. Dijkstra and M. T. Reetz,  Protein Sci.,

2012,  21, 487–497.

156 H. Jochens, D. Aerts and U. T. Bornscheuer,   Protein Eng.,

 Des. Sel., 2010, 23, 903–909.157 K. Kawasaki, H. Kondo, M. Suzuki, S. Ohgiya and S. Tsuda,

 Acta Crystallogr., Sect. D: Biol. Crystallogr., 2002,   58,

1168–1174.

158 Y. Gumulya and M. T. Reetz,   ChemBioChem, 2011,   12,

2502–2510.

159 M. Lehmann, C. Loch, A. Middendorf, D. Studer,

S. F. Lassen, L. Pasamontes, A. van Loon and M. Wyss,

 Protein Eng., 2002,  15, 403–411.

160 M. Lehmann, L. Pasamontes, S. F. Lassen and M. Wyss,

 Biochim. Biophys. Acta, Protein Struct. Mol. Enzymol., 2000,

1543, 408–415.

161 M. Lehmann and M. Wyss,  Curr. Opin. Biotechnol., 2001,12, 371–375.

162 N. Amin, A. D. Liu, S. Ramer, W. Aehle, D. Meijer,

M. Metin, S. Wong, P. Gualfetti and V. Schellenberger,

 Protein Eng., Des. Sel., 2004,  17, 787–793.

163 K. Watanabe, T. Ohkuri, S. Yokobori and A. Yamagishi,

 J. Mol. Biol., 2006,  355, 664–674.

164 H. K. Binz, M. T. Stumpp, P. Forrer, P. Amstutz and

 A. Pluckthun, J. Mol. Biol., 2003, 332, 489–503.

165 M. K. DiTursi, S. J. Kwon, P. J. Reeder and J. S. Dordick,

 Protein Eng., Des. Sel., 2006,  19, 517–524.

166 K. M. Polizzi, J. F. Chaparro-Riggers, E. Vazquez-Figueroa

and A. S. Bommarius,  Biotechnol. J., 2006, 1, 531–536.

167 C. E. McVey, M. A. Walsh, G. G. Dodson, K. S. Wilson and

 J. A. Brannigan, J. Mol. Biol., 2001,  313, 139–150.

168 B. Steipe, B. Schiller, A. Pluckthun and S. Steinbacher,

 J. Mol. Biol., 1994,  240, 188–192.

169 E. Vazquez-Figueroa, J. Chaparro-Riggers and A. S. Bommarius,

ChemBioChem, 2007, 8, 2295–2301.

170 F. H. Arnold and G. Georgiou,   Directed Enzyme Evolution:Screening and Selection Methods, 1st edn, Humana Press,

Totowa, New Jersey, 2003.

171 J. Gomes and W. Steiner,   Food Technol. Biotechnol., 2004,

42, 223–235.

172 M. Podar and A. L. Reysenbach,  Curr. Opin. Biotechnol.,

2006,  17, 250–255.

173 M. T. Reetz, J. D Carballeira and A. Vogel,   Angew. Chem.,

 Int. Ed., 2006,  45, 7745–7751.

174 K. Gunasekaran, C. Ramakrishnan and P. Balaram, J. Mol.

 Biol., 1996,  264, 191–198.

175 A. E. Serov, E. R. Odintzeva, I. V. Uporov and V. I. Tishkov,

 Biochemistry, 2005,  70, 804–808.176 W. Besenmatter, P. Kast and D. Hilvert,   Proteins: Struct.,

 Funct., Bioinf., 2007,  66, 500–506.

177 J. D. Bloom, S. T. Labthavikul, C. R. Otey and F. H. Arnold,

 Proc. Natl. Acad. Sci. U. S. A., 2006,  103, 5869–5874.

178 L. Riechmann and G. Winter, Proc. Natl. Acad. Sci. U. S. A.,

2000, 97, 10068–10073.

179 W. Gilbert,  Nature, 1978,  271, 501.

180 W. F. Doolittle,  Nature, 1978,  272, 581–582.

181 C. C. F. Blake,  Nature, 1978,  273, 267.

182 L. D. Bogarad and M. W. Deem, Proc. Natl. Acad. Sci. U. S. A.,

1999, 96, 2591–2595.

183 J. J. Silberg, J. B. Endelman and F. H. Arnold, Protein Eng.,

2004, 388, 35–42.184 C. A. Voigt, C. Martinez, Z. G. Wang, S. L. Mayo and F. H.

 Arnold, Nat. Struct. Biol., 2002,  9, 553–558.

185 L. Yuan, I. Kurek, J. English and R. Keenan, Microbiol. Mol.

 Biol. Rev., 2005, 69, 373–392.

186 N. D. Clarke,  Curr. Opin. Struct. Biol., 2010,  20, 527–532.

187 P. Heinzelman, C. D. Snow, M. A. Smith, X. L. Yu,

 A. Kannan, K. Boulware, A. Villalobos, S. Govindarajan,

 J. Minshull and F. H. Arnold,   J. Biol. Chem., 2009,   284,

26229–26233.

188 P. Heinzelman, C. D. Snow, I. Wu, C. Nguyen, A. Villalobos,

S. Govindarajan, J. Minshull and F. H. Arnold,  Proc. Natl.

 Acad. Sci. U. S. A., 2009,  106, 5610–5615.189 P. A. Romero, E. Stone, C. Lamb, L. Chantranupong,

 A. Krause, A. E. Miklos, R. A. Hughes, B. Fechtel,

 A. D. Ellington, F. H. Arnold and G. Georgiou, ACS Synth.

 Biol., 2012, 1, 221–228.

190 P. Heinzelman, R. Komor, A. Kanaan, P. Romero, X. Yu,

S. Mohler, C. Snow and F. Arnold,  Protein Eng., Des. Sel.,

2010, 23, 871–880.

191 M. A. Smith, A. Rentmeister, C. D. Snow, T. Wu,

M. F. Farrow, F. Mingardon and F. H. Arnold,   FEBS J.,

2012, 279, 4453–4465.

Chem Soc Rev Review Article

View Article Online

7/26/2019 Stabilizing biocatalysts.pdf

http://slidepdf.com/reader/full/stabilizing-biocatalystspdf 32/32

192 G. Parsiegla, C. Reverbel-Leroy, C. Tardif, J. P. Belaich,

H. Driguez and R. Haser, Biochemistry, 2000, 39, 11238–11246.

193 T. C. A. Biology, Royal Institute of Technology School of 

Biotechnology, 2008.

194 J. A. E. Maatta, Y. Eisenberg-Domovich, H. R. Nordlund,

R. Hayouka, M. S. Kulomaa, O. Livnah and V. P. Hytonen,

 Biotechnol. Bioeng., 2011,  108, 481–490.

195 S. Repo, T. A. Paldanius, V. P. Hytonen, T. K. M. Nyholm,

K. K. Halling, J. Huuskonen, O. T. Pentikainen,K. Rissanen, J. P. Slotte, T. T. Airenne, T. A. Salminen,

M. S. Kulomaa and M. S. Johnson,   Chem. Biol., 2006,  13,

1029–1039.

196 L. Baugh, I. Le Trong, D. S. Cerutti, N. Mehta, S. Gulich, P. S.

Stayton, R. E. Stenkamp and T. P. Lybrand,   Biochemistry,

2011, 51, 597–607.

197 O. Livnah, E. A. Bayer, M. Wilchek and J. L. Sussman, Proc.

 Natl. Acad. Sci. U. S. A., 1993,  90, 5076–5080.

198 C. R. Otey, M. Landwehr, J. B. Endelman, K. Hiraga,

 J. D. Bloom and F. H. Arnold, PLoS Biol., 2006, 4, e112.

199 A. Cerdobbel, K. De Winter, D. Aerts, R. Kuipers,

H.-J. Joosten, W. Soetaert and T. Desmet,   Protein Eng., Des. Sel., 2011, 24, 829–834.

200 N. M. Osorio, M. M. R. da Fonseca and S. Ferreira-Dias,

 Eur. J. Lipid Sci. Technol., 2006, 108, 545–553.

201 D. A. Cowan,   Comp. Biochem. Physiol., A: Mol. Integr.

 Physiol., 1997, 118, 429–438.

202 R. K. Owusu and D. A. Cowan,  Enzyme Microb. Technol.,

1989, 11, 568–574.

203 J. J. Hao and A. Berry,   Protein Eng., Des. Sel., 2004,   17,

689–697.

204 M. T. Reetz, P. Soni, L. Fernandez, Y. Gumulya and

 J. D. Carballeira, Chem. Commun., 2010,  46, 8657–8658.

205 B. V. Adalbjornsson, H. S. Toogood, A. Fryszkowska,C. R. Pudney, T. A. Jowitt, D. Leys and N. S. Scrutton,

ChemBioChem, 2010,  11, 197–207.

206 E. Brenna, F. G. Gatti, D. Monti, F. Parmeggiani and

S. Serra, Adv. Synth. Catal., 2012, 354, 105–112.

207 B. Van den Burg, A. de Kreij, P. Van der Veek, J. Mansfeld

and G. Venema, Biotechnol. Appl. Biochem., 1999, 30, 35–40.

208 J. Mansfeld and R. Ulbrich-Hofmann,   Biotechnol. Bioeng.,

2007, 97, 672–679.

209 S. Sharma, A. Mittal, V. K. Gupta and H. Singh,   Enzyme

 Microb. Technol., 2007, 40, 337–342.

210 H. Slusarczyk, S. Felber, M. R. Kula and M. Pohl,   Eur. J.

 Biochem., 2000, 267, 1280–1289.211 E. Vazquez-Figueroa, V. Yeh, J. M. Broering, J. F. Chaparro-

Riggers and A. S. Bommarius,  Protein Eng., Des. Sel., 2008,

21, 673–680.

Review Article Chem Soc Rev

View Article Online