Anthrax Toxin: Receptor Binding, Internalization, Pore...

ANRV313-BI76-11 ARI 28 April 2007 16:21

Anthrax Toxin: ReceptorBinding, Internalization,Pore Formation, andTranslocationJohn A.T. Young1 and R. John Collier2

1Infectious Disease Laboratory, The Salk Institute for Biological Studies, La Jolla,California 92037; email: [email protected] of Microbiology and Molecular Genetics, Harvard Medical School,Boston, Massachusetts 02115; email: [email protected]

Annu. Rev. Biochem. 2007. 76:243–65

First published online as a Review in Advance onMarch 2, 2007

The Annual Review of Biochemistry is online atbiochem.annualreviews.org

This article’s doi:10.1146/annurev.biochem.75.103004.142728

Copyright c© 2007 by Annual Reviews.All rights reserved

0066-4154/07/0707-0243$20.00

Key Words

Brownian ratchet, proton gradient, self-assembly, receptor

AbstractAnthrax toxin consists of three nontoxic proteins that self-assembleat the surface of receptor-bearing mammalian cells or in solution,yielding a series of toxic complexes. Two of the proteins, calledLethal Factor (LF) and Edema Factor (EF), are enzymes that acton cytosolic substrates. The third, termed Protective Antigen (PA),is a multifunctional protein that binds to receptors, orchestrates theassembly and internalization of the complexes, and delivers themto the endosome. There, the PA moiety forms a pore in the en-dosomal membrane and promotes translocation of LF and EF tothe cytosol. Recent advances in understanding the entry processinclude insights into how PA recognizes its two known receptorsand its ligands, LF and EF; how the PA:receptor interaction influ-ences the pH-dependence of pore formation; and how the pore func-tions in promoting translocation of LF and EF across the endosomalmembrane.

243

Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

Contents

INTRODUCTION. . . . . . . . . . . . . . . . . 244STRUCTURES OF NATIVE PA

AND THE HEPTAMERICPREPORE. . . . . . . . . . . . . . . . . . . . . . . 245Native PA: PA83 . . . . . . . . . . . . . . . . . . 246The Heptameric PA63 Prepore . . . . 246

ANTHRAX TOXIN RECEPTORS . 247Two Receptors: ANTXR1 and

ANTXR2 . . . . . . . . . . . . . . . . . . . . . 247Binding of PA to its Receptors . . . . 247

PROTEOLYTIC ACTIVATION OFPA AND OLIGOMERIZATIONOF PA63 . . . . . . . . . . . . . . . . . . . . . . . . . 249Furin Cleavage and Release

of PA20 . . . . . . . . . . . . . . . . . . . . . . . . 249Self-Association of PA63 . . . . . . . . . . . 250Association of Toxin-Receptor

Complexes with Lipid Rafts . . . . 250BINDING OF LF AND EF TO

THE PA63 PREPORE. . . . . . . . . . . . 251The Stoichiometry of Binding . . . . 251The LF/EF Binding Site on PA63 . . 251

INTERNALIZATION ANDTRAFFICKING OFTOXIN-RECEPTORCOMPLEXES . . . . . . . . . . . . . . . . . . . 252

CONVERSION OF THE PA63

PREPORE TO THE PORE . . . . . 252pH-Dependence and Mechanism

of Pore Formation. . . . . . . . . . . . . 253Receptor Dependence of Pore

Formation . . . . . . . . . . . . . . . . . . . . 254Mutations in PA that Affect Pore

Formation and Translocation . . 255TRANSLOCATION OF LF/EF

THROUGH THE PORE. . . . . . . . 255pH-Dependent Unfolding of

LF/EF . . . . . . . . . . . . . . . . . . . . . . . . 255Translocation Is N to C Terminal . 256A Transmembrane Proton

Gradient Drives Translocation . 257The Phenylalanine Clamp and Its

Role in Translocation . . . . . . . . . . 258

Binary toxin: atoxin in which twocompounds (proteinsin this context)combine to elicit atoxic response

Lethal factor (LF):a Zn2+ protease thatcleaves andinactivates mostMAP kinase kinases

Edema factor (EF):a Ca2+- andcalmodulin-dependent adenylatecyclase

Protective antigen:the transport proteinof anthrax toxin,which combines withLF and/or EF toform toxic complexes

INTRODUCTION

Pathogenic bacteria have evolved a varietyof ways to disable cells of the mammalianimmune system in order to create a nichein which they can multiply and disseminate.Among the most interesting ways, from abiochemical perspective, is for the bacteriato release toxic proteins that are capable ofpenetrating to the cytosol of host cells andenzymically modifying key substrates withinthat compartment. This mode of action neces-sitates that the toxin, or at least an enzymicallyactive piece of it, penetrate a host-cell mem-brane at some level. How this occurs is notfully understood for any toxin, but for some,including anthrax toxin, there has been signif-icant progress recently [for recent reviews ofthis and related subjects, see (1–6)].

Anthrax toxin belongs to a family of tox-ins, called binary toxins, that are produced by

certain Bacillus and Clostridium species (7).Each binary toxin consists minimally of twonontoxic proteins that combine at the sur-face of a receptor-bearing eukaryotic cell orin solution to form toxic complexes. One ofthese proteins is an enzyme (the A moiety),and the other is a receptor-binding and pore-forming protein (the B moiety). Bacillus an-thracis, the causative agent of anthrax, secretesthree monomeric, plasmid-encoded proteinsthat are collectively called anthrax toxin. Twoare enzymes: Lethal Factor (LF; 90 kDa), aZn2+ protease that specifically cleaves and in-actives MAP kinase kinases (8, 9), and EdemaFactor (EF; 89 kDa), a Ca2+- and calmodulin-dependent adenylyl cyclase (10). The third,Protective Antigen (PA83; 83 kDa), named forits effectiveness in inducing protective immu-nity against anthrax, is the vehicle for deliveryof LF and EF to the cytosol.

244 Young · Collier

Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

Figure 1Entry of anthrax toxin into cells.

The following general model of anthraxtoxin action at the cellular level has emergedin recent years (Figure 1). Toxin action be-gins when PA83 binds to either of two knownreceptors and is proteolytically activated bya member of the furin family of cellularproteases. This cleavage removes a 20-kDapiece (PA20) from the N terminus, leaving thecomplementary 63-kDa piece (PA63) boundto the receptor. Receptor-bound PA63 thenself-associates to form a ring-shaped hep-tameric complex—the prepore—that is capa-ble of binding up to three molecules of LFand/or EF competitively. Because each boundligand molecule effectively occludes two sub-units, only three ligand molecules can bind tothe heptamer simultaneously. The resultingcomplexes are endocytosed and delivered to

Prepore: aheptameric form ofPA63 formed byspontaneousself-association;binds sevenmolecules ofreceptor and up tothree molecules ofLF and/or EF

an endosomal compartment. There, under theinfluence of low pH, the prepore forms a porein the bounding membrane, initiating translo-cation of LF and EF to the cytosolic compart-ment. The cargo molecules are believed to un-fold, partially or completely, in order to passthrough the narrow (∼15 A) pore and there-fore must refold in the cytosolic compartmentto become enzymically competent.

STRUCTURES OF NATIVE PAAND THE HEPTAMERICPREPORE

The crystallographic structure of native PAhas defined its folding domains and pro-vided the basis for analyzing its multiple func-tions. The structure of the heptameric PA63

www.annualreviews.org • Anthrax Toxin Entry into Cells 245

Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

prepore serves as a model of an important in-termediate, which is capable of binding EFand LF and forming the transmembrane porethat translocates them to the cytosol.

Native PA: PA83

A 2.1 A crystallographic model of monomericPA (PA83; 83 kDa; 735 residues) shows a long,flat molecule of dimensions ∼100 × 50 ×30 A, containing four folding domains (11)(Figure 2).

Domain 1 (residues 1–258) consists of a β-sandwich with jelly-roll topology, four small

Domain 3

ANTXR2I domain

β4–α4 loop

2β2–2β3 loop

Domain 1b

Tyr119Divalent cation

Domain 4

Domain 2

Domain 1a

Phe427

Figure 2Monomeric PA complexed with the ANTXR2 I domain. This figure isadapted from PTBIT6B (13). The divalent cation bound to the receptorMIDAS is shown as a gray circle.

helices, and two calcium ions. Proteolytic acti-vation by furin occurs at the sequence RKKR(residues 164–167) within a surface loop, gen-erating a nicked form of PA83 (nPA83) as theprecursor of the heptameric PA63 prepore.

Domain 2 (residues 259–487) consists ofa β-barrel core with elaborate excursions anda modified Greek-key topology. The majorfunction of this domain is to form a trans-membrane pore to serve as the portal of en-try of EF and LF into the cytosol. Recently,domain 2 has also been shown to partici-pate in binding PA83 to cellular receptors(12, 13). In order to insert stably into mem-branes, PA63 must oligomerize to the hep-tameric state, forming the prepore, and do-main 2 must undergo a major pH-dependentconformational rearrangement. This allowsthe seven domains 2 to combine to form a14-strand transmembrane β-barrel. The large2β2–2β3 loop of domain 2, which is disor-dered in the monomer, forms the transmem-brane portion of this β-barrel (14, 15).

Domain 3 (residues 488–595) has aferridoxin-like fold and is believed to mediateself-association of PA63. Mutations in a loop(residues 510–518) of this domain strongly in-hibit oligomerization by PA63, and the struc-ture of the heptameric PA63 prepore revealsthat this loop inserts into a cleft in domain1′ of the adjacent subunit (residues 192–205)(12, 16).

Domain 4 (residues 596–735) con-sists primarily of a β-sandwich with animmunoglobulin-like fold and is primarilyinvolved in binding to cellular receptors.This domain has relatively little contact withthe rest of the protein (17).

The Heptameric PA63 Prepore

A 4.5 A structure of the water-soluble PA63

heptamer (11), and a later 3.6 A structure (12),show a hollow ring, 160 A in diameter and85 A high, with the subunits packed like piewedges. There are no major conformationalchanges from the structure of monomericPA83. Domains 1′ and 2 are on the inside of the


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

ring, and domains 3 and 4 are on the outside.The subunit:subunit interface is comprisedmainly of charged or polar residues from do-mains 1′ and 2. The lumen averages ∼35 Ain diameter, narrowing to ∼20 A in places,and is lined mainly with polar and negativelycharged amino acids. A conserved phenylala-nine (Phe427) in a mobile, solvent-exposedloop in the lumen is a prominent exception,and as described below; this residue plays amajor role in protein translocation throughthe pore (18).

Two other functionally important featuresof the prepore are a large, flat hydrophobicsurface exposed on domain 1′ at the top ofthe structure, which serves as the binding sitefor LF and EF, and the 2β2–2β3 loop of do-main 2, which is involved in pore formation.Whereas the 2β2–2β3 loop is disordered inPA83, the 3.6 A structure of the prepore showsthis loop to project outward, away from themonomer, and to be packed between domains2 and 4 of the neighboring monomer (12).This implies that domains 2 and 4 must moveapart during pore formation, a transition thatinvolves a major structural rearrangement andrelocation of the 2β2–2β3 loop.

ANTHRAX TOXIN RECEPTORS

Cell biological and molecular genetic studieshave revealed the existence of two relatedcellular receptors for PA, both of which arecapable of mediating anthrax toxin action.Many aspects of their interaction with PA83

and the prepore are now understood, to thelevel of crystallographic detail for one of thereceptors.

Two Receptors: ANTXR1 andANTXR2

Two related cell surface receptors that bindPA83 have been identified: ANTXR1 (tumorendothelial marker-8) and ANTXR2 (cap-illary morphogenesis protein 2) (Figures 2and 3) (19, 20). Alternative mRNA splicinggives rise to three different protein isoforms

ANTXR1 andANTXR2: anthraxtoxin receptors 1 and2, respectively, theonly known types ofreceptor for thistoxin

VWA or I domain[von WillebrandFactor A, orintegrin inserted(I) domain]: adoptsa Rossman foldstructure

FRET: fluorescenceresonance energytransfer

of each receptor. The long (564 amino acids)and medium-length (368 amino acids) formsof ANTXR1, and the long (489 and 488amino acids) forms of ANTXR2, are type 1membrane proteins that function as anthraxtoxin receptors (19–22). By contrast, theshort forms of these proteins (333 aminoacids for ANTXR1, and 368 amino acids forANTXR2), lack this function (21, 22). PA83

binds to an approximately 190 amino acid–long extracellular domain of ANTXR1 andANTXR2, which is related to von Willebrandfactor type A (VWA) domains and integrininserted (I) domains. The cytoplasmic tails ofthese receptors do not play a critical role intoxin entry since this region of ANTXR1 canbe removed without abrogating intoxication(21).

Both ANTXR1 and ANTXR2 appear tobe ubiquitously expressed and bind collagenα3(VI) and collagen IV/laminin, respectively(23, 24). Both receptors are also expressedin vein endothelial cells where they appearto have important roles in regulating angio-genic processes (23–30). Mutations in theANTXR2 gene are linked to two humanautosomal recessive diseases, juvenile hyalinefibromatosis and infantile systemic hyalinosis(31, 32).

Binding of PA to its Receptors

PA83 binds the soluble ANTXR2 I do-main (residues 35–225) at a 1:1 ratio, asshown by isothermal titration calorimetryand by FRET (fluorescence resonance en-ergy transfer)-based assays performed with anAF488 (donor fluorophore)-labeled version ofPA83 and an AF546 (acceptor fluorophore)-labeled version of the soluble ANTXR2 I do-main (33) and by crystallography (Figure 2)(13). The heptameric PA63 prepore can bindup to 7 molecules of the receptor I domain,as shown by the same FRET-based assayand by X-ray crystallography (Figure 3) (12,33). Receptor binding does not influence therate of PA63 heptamer formation in solution(33).


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

PA domain 4

PA domain 2PA 2ββ2–2β3 loop

ANTXR2I domain

Figure 3Top-down view ofProtective Antigen(PA) domains 2 and4 complexed withthe ANTXR2 Idomain. Thisfigure is adaptedfrom PDB ITZN(12).

MIDAS: metalion-dependentadhesion site

Like other VWA/I domains, those ofANTXR1 and ANTXR2 contain a metalion-dependent adhesion site (MIDAS),which, in the anthrax toxin receptors, com-prises residues Asp50, Ser52, Ser54, Thr118,and Asp148 (ANTXR2)/Asp150 (ANTXR1)(19, 20). The receptor MIDAS coordinatesa divalent cation, which plays a role inPA-receptor binding (12, 13, 33, 35, 37).

PA-receptor interactions bear some sim-ilarity to those of I domain-containing α-integrins and their natural ligands. Inte-grin I domains can exist in either of twoconformations, closed (low-affinity ligand-binding state) or open (high-affinity ligand-binding state) (36). Conversion from closedto open states requires the following: re-arrangement of the MIDAS residue-metalion contacts; metal ion coordination by acarboxylate-containing amino acid side chainof the bound ligand; movement of two con-served phenylalanines that bind within I-domain hydrophobic pockets; and a 10 Adownward movement of the C-terminal

α-helix of the I domain away from theMIDAS-containing face (36). The metal ion-coordinating residues in the PA-ANTXR2 Idomain complex are arranged in an open con-figuration, with PA residue D683 and the re-ceptor’s MIDAS threonine and both MIDASserines located within the primary metal ioncoordination sphere (13, 34). Consistently,both receptor MIDAS aspartate residues arelocated in the secondary coordination sphere,where they bind the metal ion indirectly viawater molecules (12, 13). The placement ofthe conserved ANTXR2 residues, Phe181 andPhe203, and of the C-terminal α-helix of theI domain in the PA-ANTXR2 complex is alsoconsistent with an open conformation (12,13, 34). In the case of ANTXR1, an openI-domain conformation appears to be essen-tial for PA binding, since this interaction wasabolished by substituting either the receptorMIDAS threonine or PA residue Asp683 (17,35). However, although probably optimal forPA binding, such a conformation is not abso-lutely essential for ANTXR2; that receptor,


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

unlike ANTXR1, can bind and support intoxi-cation mediated by D683N or D683K mutantforms of PA (37). Indeed, the D683K mutantform of PA can also mediate lethal toxin killingof Fischer 344 rats, albeit severalfold less ef-ficiently than wild-type PA, indicating thatANTXR2 plays an important role in lethaltoxin killing in vivo (37).

The PA-binding affinities of the ANTXR1and ANTXR2 I domains are strikingly differ-ent from each other. The affinity of ANTXR1[KD = 1.1 μM or 130 nM for the Mg2+- andCa2+-bound complexes, respectively; (38)] ison a par with those of integrin-ligand inter-actions (36), but that of ANTXR2 is approx-imately 1000-fold higher (KD = 170 pM or780 pM for the Mg2+- and Ca2+-bound com-plexes, respectively) (33). Consistent with itshigher binding affinity, the soluble ANTXR2I domain is the more effective receptor decoy–based inhibitor of anthrax intoxication (38).

X-ray structural studies revealed why PAbinds ANTXR2 with such high affinity. Com-pared to the 1300 A of buried protein surfacethat is typically associated with α-integrin-ligand interactions, there is 1900–2000 A ofburied protein surface at the PA-ANTXR2interface, involving I domain contact withPA domain 4 and with the 340–380 loop ofPA domain 2 (Figure 2) (12, 13). Homolog-scanning mutagenesis studies have subse-quently revealed that receptor residues lo-cated in the β4-α4 loop region of the Idomain, which is involved in binding PAdomain 2, contribute to the difference inPA-binding affinity between ANTXR1 andANTXR2 (37).

The high binding affinities seen with thePA-ANTXR2 interaction in the presence ofthe metal ions are primarily due to very slowdissociation rates: 9.2 × 10−6 s−1 (Mg2+) or8.4 × 10−5 s−1 (Ca2+) (33). Although the slowrate of PA dissociation from ANTXR2 is notcommonly seen with other VWA domain-ligand interactions, it is also a feature ofhookworm neutrophil inhibitory factor (NIF)binding to CD11b (39). Given the high bind-ing affinity that exists between PA and its re-

ceptors, especially ANTXR2, it seems likelythat PA binding could interfere with the as-sociation between these receptors and theirphysiological ligands. If so, disruption of thesereceptor-ligand interactions might also playa role in anthrax pathogenesis, an idea thatremains to be formally tested.

PROTEOLYTIC ACTIVATION OFPA AND OLIGOMERIZATION OFPA63

The proteolytic activation of native PA83,which is effected in vivo by furin-family pro-teases, may be replicated in vitro with trypsin,allowing studies of the activation and the self-association of PA63 to be performed in solu-tion. These studies have been complementedby studies of the behavior of receptor-boundwild-type and mutated forms of PA in culturedcells.

Furin Cleavage and Release of PA20

The N-terminal region of PA83, correspond-ing to PA20, sterically inhibits self-associationof the protein and thus also blocks later stepsin toxin action. To remove the blockage,receptor-bound PA83 is cleaved by a furin-family protease (40, 41), generating a nickedform of the protein (nPA83) that subsequentlydissociates, releasing PA20 into the medium.Dissociation of PA20 (residues 1–167) fromPA63 requires a β-sheet to be ruptured and alarge hydrophobic surface to be exposed. PA20

and PA63 tend to remain associated via non-covalent forces (Kd ∼ 190 nM) (42) and con-fractionate on size-exclusion columns.

Domain 1′, the segment of domain 1that forms the N terminus of PA63, gener-ates the binding sites for EF and LF uponPA63 oligomerization. Two calcium atoms arefirmly bound within domain 1′ (t1/2 of ex-change with free Ca2+ ∼ 4 h) and are thoughtto stabilize domain 1′ in a conformation thatfacilitates oligomerization of PA63 and bind-ing of EF and LF (43). The dissociation rateconstant of PA20 and PA63, koff, has been


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

DRM:detergent-resistantmembranemicrodomain

estimated by FRET measurements of dye-labeled fragments in solution to be ∼3 × 10−2

s−1 at 20◦C, corresponding to a t1/2 of 21 s(42). Thus, nPA83 is largely dissociated withina minute after proteolytic activation, and dis-sociation is unlikely to be a rate-limiting stepin toxin action, which occurs over tens of min-utes at the cellular level. Dissociation of PA20

from PA63 in solution was also examined bystopped-flow FRET, and two steps were iden-tified: a fast bimolecular step and a slower uni-molecular step. The first step may correspondto dissociation of the PA20:PA63 complex andthe second to an isomerization in the PA63

moiety that could potentiate it for oligomer-ization and subsequent steps in toxin action.

Self-Association of PA63

After PA20 dissociates from nPA83 it is free todiffuse in three dimensions, whereas receptor-bound PA63 is constrained to diffuse in thetwo dimensions of the membrane. EffectivelyPA63 therefore becomes concentrated relativeto PA20. Depending on a number of other fac-tors, this could accelerate its self-association.

When PA63 is purified from trypsin-activated PA83 by anion-exchange chromatog-raphy, it is found to have spontaneously self-associated into a ring-shaped heptamer (44).As long as the pH is maintained above 8, theheptamer remains soluble, in the so-calledprepore state. The association-dissociationequilibrium strongly favors the heptamer,and neither monomeric PA63 nor subhep-tameric oligomers have been detected. Re-cent FRET measurements, performed withprepore prepared from PA preparations la-beled with donor and acceptor fluorescentdyes, have yielded a dissociation rate constant,kd, of 1 × 10−6 s−1, corresponding to a t1/2

of ∼7 days (45). In contrast, a ternary com-plex, containing the PA-binding domain of LF(LFN) bound to a PA63 dimer formed fromtwo nonoligomerizing mutants, dissociatedrapidly (t1/2 ∼ 1 min). The remarkable stabil-ity of the prepore relative to the ternary com-plex presumably derives from each subunit

of the ring being bound to two neighboringsubunits.

Association of Toxin-ReceptorComplexes with Lipid Rafts

PA63 oligomerization causes the receptors tocluster in detergent-resistant membrane mi-crodomains (DRMs, or lipid rafts) (46). Theprecise nature of these DRMs is not known.However, PA63 and antibody-clustered formsof PA83 colocalize with GPI-anchored pro-teins, but not with caveolin-1, indicating thatthey are not caveolae (46). Consistently, PA63

can form pores and can internalize LF in cellswithout caveolae (46).

Lipid-raft association is regulated by S-palmitoylation of the receptors. ResiduesCys347 and Cys481 of the long isoform ofANTXR1 have been identified as major sitesof S-palmitoylation, and residue Cys521 ofthis protein is a minor palmitoylated residue(47). A mutant receptor lacking all possibleS-palmitoylation sites localizes constitutivelyto DRMs, indicating that this posttransla-tional modification negatively regulates re-ceptor association with DRMs in the absenceof PA binding (47). DRM association of thismutant receptor remained sensitive to in-hibition by the palmitoylation inhibitor, 2-bromopalmitate, implicating the involvementof another palmitoylated protein in targetingthe toxin-receptor complex to DRMs (47).

Oligomerization of PA and its cluster-ing in lipid rafts are important for the sub-sequent internalization of PA:receptor com-plexes into cells. A furin-resistant PA mutant(PASSSR) was found to remain at the cell sur-face (48), and another such mutant (PASNKE)was internalized into CHO cells only afterantibody-induced clustering (46). Moreover,antibody-induced clustering and internaliza-tion of wild-type PA63, and LF transloca-tion, were blocked by β-methylcyclodextrin(MCD) treatment, which extracts cholesterolfrom membranes, disrupting DRMs (46). Ithas been proposed that the slow intrinsicreceptor internalization rate, and the rapid


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

uptake after heptamerization and LF/EFbinding, help to facilitate intoxication (46).

BINDING OF LF AND EF TOTHE PA63 PREPORE

The availability of PA63 prepore and its en-zymic ligands in pure form, and the ability toprepare mutated forms of these proteins eas-ily, have permitted the formation of toxic, het-erooligomeric complexes to be examined insolution. While no crystallographic structureof such a complex has been reported, otherapproaches have yielded a detailed model ofhow the prepore binds these ligands.

The Stoichiometry of Binding

LF, EF, and LFN bind to PA63 with high affin-ity (Kd ∼ 1–2 nM; kon ∼ 3 × 105 M−1 s−1; koff

∼ (3–5) × 10−4 s−1), as shown by surface plas-mon resonance measurements with the pu-rified proteins and by binding studies on L6cells with radiolabeled forms of these proteins(49). LF and EF compete for the same siteson the PA63 heptamer, and a single heptamercan bind both of these proteins simultaneously(50). Even though the heptameric PA63 pre-pore has radial sevenfold symmetry, measure-ments by two independent methods showedthat the heptamer binds only three moleculesof ligand under saturating conditions (51).Later studies revealed that neither of two dif-ferent oligomerization-deficient PA mutants,with lesions on different subunit:subunit con-tact surfaces, bound ligand; but a mixture ofthe two mutant proteins did (52). Mixing thetwo mutant forms of proteolytically activatedPA with LFN in solution yielded a ternarycomplex containing a single copy of each ofthese proteins. Thus, two molecules of PA63

must come together to form a high-affinityLF/EF binding site. A mutagenesis study (53)confirmed that the ligand binding site spansthe intersection between two PA63 subunits,and further, that a single ligand moleculewould be expected to occlude the binding sitesof two subunits. This explained the finding

that the PA63 heptamer binds a maximum ofthree ligand molecules. The saturated hep-tamer is therefore asymmetric, with six of theseven subunits blocked and the remaining onefree. At variance with this picture, a study bycryoelectron microscopy of binding of LF tothe PA63 heptamer yielded results suggestingthat a single molecule of LF interacts with foursuccessive PA63 monomers within the preporeand partially unravels the heptamer (54).

The LF/EF Binding Site on PA63

Although no crystallographic model of thebinding of EF or LF to PA63 has been re-ported, directed mutagenesis and other meth-ods have revealed how the binding domainsof these ligands dock onto the PA63 heptamer.Substitution of Ala for residues of LFN thatwere selected from the crystallographic struc-ture of LF, based on a sequence compari-son with EF, allowed the face of this do-main that recognized PA63 to be identified(55). When Cys residues were introduced intothis face and the binding surface of PA63,three contact points were identified by disul-fide cross-linking; and two additional contactpoints were found by directed charge-reversalmutations (56). These contact points wereconsistent with the lowest energy LFN-PAcomplex by computational protein-proteindocking with the Rosetta program. Enhancedpeptide amide hydrogen/deuterium exchangemass spectrometry, and the demonstrationthat residues within this surface interact withLys197 of two adjacent PA subunits simulta-neously, allowed the PA-interactive surface onLFN to be defined more precisely (57). Theresulting model showed that EF and LF bindthrough a highly electrostatic interface, withtheir disordered, charged N termini poisedat the entrance of the heptameric prepore, andthus positioned to initiate translocation in anN- to C-terminal direction.

The way in which EF and LF bind to PA63

has ramifications for how these proteins affectthe kinetics of assembly of heterooligomerictoxin complexes. Oligomerization of PA63


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

in solution is accelerated by concentrationsof LFN up to 50 nM, but is inhibited athigher concentrations (45). A prepore lig-anded with LF/EF contains both liganded andunliganded PA63 subunits, and of those sub-units that are liganded, there exist two sub-populations, each with different contacts withthe ligand. High concentrations of LF/EFcan be conceived to inhibit oligomerizationby fostering conversion of PA63 to a lig-anded PA63 dimer, thereby depleting freePA63, which is necessary to form a completeprepore. Alternatively, the inhibition couldresult from weak interactions of LF/EF withPA63 monomers, yielding a population of het-erodimers that is biased toward one of thetwo subpopulations of ligand:PA63 interac-tions within the liganded heptamer.

INTERNALIZATION ANDTRAFFICKING OFTOXIN-RECEPTORCOMPLEXES

Following lipid-raft association, anthraxtoxin-receptor complexes are believed tobe internalized primarily through clathrin-coated pits and then trafficked to early en-dosomes. Supporting this model, antibody-cross-linked forms of PA83 colocalize withclathrin-coated pits and vesicles early after in-ternalization and expression of the dominant-negative E�95/295 form of Eps15, whichspecifically interferes with clathrin-mediatedendocytosis and blocks internalization (46).However, other cellular endocytic pathwayshave also been implicated in toxin uptake(58).

Anthrax toxin uptake is dependent uponubiquitination of a conserved receptor lysineresidue (Lys352 in ANTXR1 and Lys350 inANTXR2), mediated by the cellular E3 lig-ase Cbl (47). The cell surface protein LRP6 isalso involved, acting as a toxin coreceptor (59).The extracellular domain of LRP6 interacts,either directly or indirectly, with ANTXR1and ANTXR2 in a PA-independent manner.This coreceptor stimulates PA-binding as well

as toxin internalization, and the latter prop-erty requires the LRP6 cytoplasmic tail (59).

LRP6 is a component of cellular Wnt-signaling pathways, associating with Wntligands and the frizzled receptors, and itcan downregulate this pathway by regulat-ing internalization of LRP6/Kremen(Krm)/Dickoppf (DKK) complexes (60). It remainsto be seen if additional Wnt-signaling com-ponents are associated with LRP6-ANTXR1/ANTXR2 complexes. LRP6 appears to playa very specific role in toxin entry, since thehighly related LRP5 protein, which also func-tions in Wnt-signaling, does not appear ableto functionally substitute for LRP6 in anthraxtoxin internalization (59). LRP6-specific anti-bodies interfere with anthrax intoxication, in-dicating that this newly identified coreceptormight be a viable target for the developmentof novel antitoxins (59).

ARAP3, a phosphoinositide-binding pro-tein with two GTPase activating (GAP) do-mains, one for ARF6 GTPase and anotherfor Rho GTPase (61), also plays a role intoxin uptake after PA-receptor binding (62).Because ARF6 recruits clathrin adaptor pro-tein 2 (AP2) and clathrin to coated pits(63, 64), it seems likely that ARAP3 impactsclathrin-dependent endocytosis of the toxin-receptor complexes. However, the precise roleof ARAP3 in anthrax toxin internalization re-mains to be determined. Following internal-ization, the toxin-receptor complexes are traf-ficked to an early endosomal compartmentcontaining Rab5, a small GTPase regulatoryprotein involved in sorting endosome forma-tion and function (65, 66).

CONVERSION OF THE PA63PREPORE TO THE PORE

The conformational transition of the PA63

prepore to the pore has been shown to bedependent both on acidic pH and the spe-cific receptor to which it is bound. Mutationalanalysis of PA and its receptors has yielded in-sights into the mechanism by which receptors


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

influence the pH threshold of prepore-to-pore conversion.

pH-Dependence and Mechanism ofPore Formation

Addition of small amounts of trypsin-activated PA83 or purified PA63, but not na-tive PA83, to planar phospholipid bilayersor liposomes causes cation-selective channels(pores) to form (67–69). Channel formationwas observed with membranes composed ofasolectin or diphytanoyl phosphatidylcholine,and there have been no indications that chan-nel formation or function depends on specificlipids. The rate at which PA63 forms chan-nels is accelerated by mild acidification of thecis compartment of planar bilayers (the com-partment to which the protein was added).Consistent with this, one can demonstratepermeabilization of the plasma membrane ofmammalian cells to 86Rb+ and 22Na+ by nPA83

or PA63 under acidic conditions (69). PA hasalso been shown to mediate the translocationacross the plasma membrane of EF, LF, and fu-sion proteins derived from their N-terminal,PA63-binding domains (70). This transloca-tion is pH dependent, with a pH optimumof ∼5.5. Consistent with these findings, PA63

pore formation and translocation occur withinacidic endosomal compartments (1, 22, 46).Recent findings in a liposomal system indi-cate that an important role of receptors isto favor partitioning of PA63 into membranesby maintaining the prepore in close proxim-ity to the target membrane, and presumablyin an optimal orientation, when it undergoespH-dependent conformation to the pore state(71). When the prepore is exposed to acidicpH in solution, it rapidly loses the ability toinsert into membranes.

The crystallographic structures of PA83

and the prepore revealed the 2β2–2β3 loopas a candidate for formation of a transmem-brane 14-strand β-barrel similar to that seenin the pore formed by the α-toxin of Staphy-lococcus aureus (11). Evidence in favor of thismodel came from experiments in which pores

formed in planar lipid bilayers from PA mu-tants containing Cys substitutions in the 2β2–2β3 loop were exposed to the thiol-reactiveprobe, methanethiosulfonate ethyltrimethy-lammonium (MTS-ET) (14). Because re-action with this reagent adds a positivelycharged group, modification of those Cysresidues with side chains projecting into thelumen would be predicted to inhibit conduc-tance of the cation-selective pore. Conversely,Cys residues with side chains facing the lipidbilayer should be inaccessible to MTS-ETand thus should have no effect on conduc-tance. The pattern of effects of MTS-ET ob-served coincided with that predicted by theβ-barrel model. Thus, strong inhibitory ef-fects were seen at alternating positions overtwo stretches corresponding to the descend-ing and ascending strands of the putative β-barrel, and also at four consecutive residuescorresponding to the turn region.

Additional support for this model of mem-brane insertion has come from studies invivo with PA labeled at Cys residues inthe 2β2–2β3 loop with the environment-sensitive fluorophore, 7-nitrobenz-2-oxa-1,3-diazole (NBD) (15). Membrane insertion wasmonitored in CHO cells by following the in-crease in fluorescence intensity of NBD asit entered the nonpolar environment of thebilayer.

Major conformational rearrangements ofdomain 2 are required for the 2β2–2β3 loopsof the seven PA63 subunits of the prepore to in-teract to form a 14-strand β-barrel. Because,in the prepore, the 2β2–2β3 loop is sand-wiched between domains 2 and 4 of a neigh-boring subunit, these domains must moveapart for the loop to be relocated to the baseof the heptamer (12). In addition, because the2β2–2β3 loop is part of a Greek key motif andextends laterally from approximately the ver-tical midpoint of domain 2, the flanking 2β2and 2β3 strands must be stripped out of thatdomain for the relocation to occur (11).

In an extension of the original studiesin planar bilayers, the approach of couplingMTS-ET inhibition of conductance with


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

Cys-scanning mutagenesis was applied to se-quences flanking the 2β2–2β3 loop (72). Theresults suggest that the β-barrel structure ex-tends well beyond the 2β2–2β3 loop and in-corporates the 2β2 and 2β3 strands and the2β1–2β2 and 2β3–2β4 loops. If residues 303–324 comprise the transmembrane region, aspredicted, then the entire β-barrel, predictedto encompass residues 275–352, could be overtwice as long as the 14-strand β-barrel of theα-hemolysin. A model of the PA63 pore basedon this information and energy minimizationcalculations has been published (73).

Receptor Dependence of PoreFormation

The crystal structures of ANTXR2-PA com-plexes revealed that this receptor acts as amolecular clamp, binding PA domains 2 and4, to restrict toxin pore formation at neutralpH (12, 13). Indeed, the pH profile of PA63

pore formation is consistent with titration ofhistidine residues in the toxin-receptor com-plex: There are 4 histidines in the 285–340stretch of PA domain 2, and receptor residueHis121, which is conserved in both ANTXR1and ANTXR2, is located at the PA-binding in-terface (12, 13). Until recently, it was thoughtthat prepore-to-pore conversion would occurunder the same low-pH conditions if the toxinwas bound to either ANTXR1 or ANTXR2.However, a striking 1.0 pH unit difference wasfound in the pH thresholds for pore formationwhen PA is bound to ANTXR1 (∼pH 6.2)versus ANTXR2 (∼pH 5.2) (22, 74). Mul-tiple lines of evidence support this conclu-sion, including immunoblot and electrophys-iological patch-clamp analyses, used to studyPA63 pore formation on cell surfaces; excimerfluorescence measurements of pyrene-labeledAsn306Cys PA-receptor complexes; and thefact that ANTXR2-dependent toxin entry isinhibited by the lysosomotropic agent ammo-nium chloride, whereas ANTXR1-dependententry is not (22, 74). These findings raise thepossibility that pore formation occurs withindistinct endosomal compartments, depend-ing upon whether the toxin is associated with

ANTXR1 or ANTXR2: the intraluminal pHof the sorting endosome is in the range of pH5.9–6.0, whereas that of late endosomes is typ-ically pH 5.0–6.0 (66). Given these findings,it will be of interest to determine whetherPA63 heptamers can form mixed complexeswith both ANTXR1 and ANTXR2 and if so,to establish how differences in the ratios ofthese receptors in the complexes influence thepH threshold for toxin pore formation. Fur-thermore, although it was proposed that thereceptor might remain bound to form a struc-tural support for the newly formed pore (13),coimmunoprecipitation studies have failed todetect receptors associated with PA63 pores,indicating that the receptors may be released(22). Indeed, pore formation in the absence ofreceptor binding is consistent with the factthat (PA63)7 pores can be formed in mem-branes that do not contain anthrax toxin re-ceptors (67).

Mutational studies have identified thekey determinants that distinguish the activi-ties of the ANTXR1 and ANTXR2 clamps.The key regulator of acid pH-dependentpore formation was residue Tyr119. Replace-ment of this residue of ANTXR2 with ei-ther a Phe or an Ala led to a dramatic0.6–0.8 shift in the pH threshold requiredfor toxin pore formation (H. Scobie et al.,submitted). It is not yet known why thehydroxyl moiety of this side chain, which is lo-cated at the binding interface with PA domains2 and 4 (Figure 2), is so important for this pro-cess, although it is predicted to make H-bondswith the polypeptide backbone at PA residueAla341 and/or with the Arg side chain of PAresidue 659. Additionally, residues located inthe β4-α4 loop region of the ANTXR2 I do-main, which bind PA domain 2 (Figure 2),have been identified as major determinants ofthe 1.0 pH unit difference in the receptor-specific pH threshold of pore formation (H.Scobie et al., submitted). Together, these stud-ies support a model in which toxin pore for-mation minimally involves receptor release ofPA domain 2 and the neighboring edge of PAdomain 4.


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

Mutations in PA that Affect PoreFormation and Translocation

Point mutations that strongly inhibit thetranslocation function of PA were identifiedby site-directed mutagenesis in two solvent-accessible loops of domain 2 (2β7–2β8 and2β10–2β11), which lie on the luminal surfaceof this domain (75). Subsequently, a scan ofthe entire PA63 fragment by Cys-replacementmutagenesis revealed a total of 33 sitesat which the mutation inhibited toxin actionat least 100-fold in cell culture (76). The factthat 22 of the 33 sites lie in domain 2 showsthe remarkable sensitivity of this domain tomutational disruption. This sensitivity mayreflect the major conformational rearrange-ments in domain 2 needed for pore formationand translocation.

Two classes of mutations in domain 2 havebeen characterized: (a) those that block con-version of the prepore to the pore, and (b)those that allow pore formation, but blockthe pore’s ability to translocate its substrates(75, 77, 78). The first class is illustrated byreplacements for Asp425, a residue in the2β10–2β11 loop. The Asp425Ala mutation,for example, blocks the ability of PA to per-meabilize the plasma membrane of CHO cellsto 86Rb+ and the ability to convert from theSDS-dissociable prepore to the SDS-resistantpore (75).

Some of the mutations in domain 2 thatinhibited pore formation and/or transloca-tion are dominantly negative (DN) (76–78). Thus, when heptameric prepores wereformed from mixtures of mutated PA andwild-type PA, the mutated form stronglyinhibited pore formation and/or transloca-tion. Asp425Lys and Phe427Ala, both ofwhich are in the 2β10–2β11 loop, exhib-ited strong dominant-negative properties, andother DN mutations have been identified inthe nearby 2β6 and 2β7 strands, suggest-ing that this region of domain 2 undergoessignificant reorganization during prepore-to-pore conversion. Some of the mutations thatblock pore formation and translocation are

not dominantly negative, however, such asLys397Asp.

With the most strongly inhibitory of theDN mutants, the presence of a single mu-tant subunit within the prepore may be suf-ficient to ablate its activity, suggesting thatDN forms of PA could be used as novel an-titoxins (77, 78). Thus in an anthrax infec-tion, injected DN-PA would coassemble withwild-type PA to yield inactive prepores orpores, preventing EF and LF from enteringthe cytosol. Furthermore, since the DN mu-tations tested did not impair immunogenicityof PA, the mutated proteins could also poten-tially induce a long-term protective immuneresponse. Hence, DN-PA might represent acombination therapeutic and vaccine for an-thrax. Initial experiments in the Fischer 344rat model of lethal toxin action showed thatsubstoichiometric amounts of either of twoDN forms of PA, one with the Phe427Alapoint mutation and another with two muta-tions, Lys397Asp and Asp425Lys, completelyblocked the effects of a lethal combination ofLF and PA.

TRANSLOCATION OF LF/EFTHROUGH THE PORE

Among the many toxins that act by enzymi-cally modifying cytosolic substrates, anthraxtoxin has proven perhaps the most tractableto the study of the translocation process. Var-ious approaches, but in particular electrophys-iological measurements in planar phospho-lipids bilayers, have elucidated key features oftranslocation, including the role of the N ter-minus of EF and LF, the function of the PA63

pore, and the energetics of translocation.

pH-Dependent Unfolding of LF/EF

By analogy with the staphylococcal α-hemolysin structure, the lumen of the puta-tive 14-strand β-barrel formed by PA63 is pre-dicted to be ∼15 A in diameter. The β-barrelis lined principally by residues with small


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

Molten globule: acompact, partiallyfolded state observedunder mildlydenaturingconditions, such aslow pH; much of theoriginal secondarystructure is retained

side-chain volumes, such as Ser and Thr, andhas been predicted by modeling to accom-modate LFN in an α-helical configuration(79). Thus, the β-barrel could accommodateα-helix or extended polypeptide, and largerstructures would need to unfold in order tobe translocated. Indeed, there is evidence thatimpediments to unfolding block translocationin this system (70). Fusions of LFN with thecatalytic domain of diphtheria toxin (DTA) orwith dihydrofolate reductase (DHFR) wereshown to be translocated by PA across theplasma membrane of CHO cells when the pHof the medium was lowered, but translocationwas blocked either by introduction of an arti-ficial disulfide bridge into the DTA moiety ofLFN-DTA or by adding a small-molecular-weight ligand of DTA (adenine) or DHFR(methotrexate) that stabilized the native con-formation.

The low-pH conditions of the endosomecan be conceived to facilitate unfolding of LFand EF as a prelude to their entry into the PA63

pore, and the unfolding properties of the iso-lated N-terminal LFN and EFN domains insolution suggested that this may be the case(79). The pH-dependence of the unfolding ofeach of these domains was examined by fluo-rescence and circular dichroism in the pres-ence and absence of chemical denaturants.Both proteins were shown to unfold via a four-state mechanism: N↔I↔J↔U, N being thenative state, U the unfolded state, and I andJ the intermediate states. The acid-inducedN→I transition occurs over a pH range ofpH 5–6, approximating the pH range of theendosome. The I state, which predominates atlower pH values, is compact and has the char-acteristics of a molten globule. The J and Ustates are populated significantly only in thepresence of denaturant. Thus, the moltenglobular I state is the primary candidate as anintermediate in the unfolding process and inentry of EF and LF into the PA pore. Cellu-lar studies performed with ammonium chlo-ride revealed, however, that LF translocationcan also occur under near-neutral pH levelsin ANTXR1-expressing cells (22). The rate-

limiting step of entry has not yet been de-termined, and before the importance of theacid-induced unfolding of ligands can be de-termined, it will be important to measure therates of various steps in this process as a func-tion of pH.

LF and EF are believed to be delivered ini-tially into the lumen of intraluminal vesicleswithin the endosomal pathway (65). Thesevesicles are then trafficked to late endosomeswhere they fuse with the limiting endosomalmembrane, releasing the enzymatic toxin sub-units into the cytosol (65). Consistent withthis model, toxin delivery is impaired by noco-dazole treatment, which disrupts trafficking ofvesicles from early to late endosomes; by dis-ruption of late endosome function through ex-pression of the Asn125Ile dominant-negativeform of Rab 5; by a lysobisphosphatidic acid(LBPA)-specific antibody that inhibits forma-tion of these vesicles; and by knocking-downexpression of ALIX, a homolog of the yeastclass E vps31 protein, involved in multivesic-ular body (MVB) sorting and biogenesis (65).Also, electron microscopic analysis has con-firmed that PA63 pores are found on the mem-branes of these vesicles (65). Consistently,SDS-resistant PA63 heptamer formationwas blocked at the restrictive temperature inldlF cells that harbor a temperature-sensitivedefect in the epsilon COPI coatamer subunitthat interferes with MVB formation (65). Ithas been proposed that this pathway of toxindelivery into cells may be required for theefficient interaction of LF with its MEK1substrate, located within a scaffolding com-plex which is associated with late endosomes(65).

Translocation Is N to C Terminal

The crystallographic structure of LF revealedthat the N-terminal 30 amino acids of LFN

comprise a disordered region containing ahigh density of acidic and basic residues (80).This flexible region is connected to a helix thatextends outward from the main body of thedomain and, according to the current model


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

of LFN docked to the prepore, positions theflexible N terminus close to the mouth ofthe pore. N-terminal truncations of 27 or36 residues strongly impaired acid-inducedtranslocation of LFN across the plasma mem-brane without significantly affecting its bind-ing to PA63 (81). The same truncations alsoablated the protein’s ability to block ion con-ductance of PA63 pores formed in planar bilay-ers at a small positive voltage (+20 mV). Fus-ing a hexahistidine tag to the N terminus ofthe truncated proteins restored both translo-cation activity and channel-blocking activity.Also, at +20 mV, hexahistidine and biotin tagsat the N terminus were accessible to Ni2+ andstreptavidin, respectively, added to the transcompartment of a planar bilayer.

Taken together, these findings suggest thatthe N terminus of bound LF or EF enters thePA63 pore under the influence of acidic pHand a positive membrane potential, and initi-ates translocation of the unfolded polypeptidein an N- to C-terminal direction. By proto-nating His and acidic residues, low pH is pre-dicted to destabilize LFN, give the N-terminalflexible region of LF and EF a net positivecharge, and perhaps also attenuate the electro-static component of the nM affinity of theseproteins for oligomeric PA63.

The role of a positively charged N termi-nus in initiating the threading of ligands intothe PA63 pore may explain the finding that fus-ing short tracts of cationic residues—Lys, Arg,or His—to the N terminus of DTA enabledthis protein to be translocated to the cytosolby PA, causing the inhibition of protein syn-thesis (82). Lys8DTA showed ∼10% the ac-tivity of LFNDTA and was the most effectiveof the tagged proteins tested. A high concen-tration (∼1 mM) of a short synthetic peptidecontaining a Lys6 tract gave measurable pro-tection against the action of Lys6DTA, butLFN did not, even at saturating concentra-tions (1 μM). These findings suggest that theelectrostatic attraction of a polycationic tagmay be sufficient to guide a protein with nosignificant affinity for the LF/EF site on PA63

into the prepore.

A Transmembrane Proton GradientDrives Translocation

It is possible to measure translocation of PA63

ligands, such as LFN, across planar lipid bilay-ers by monitoring ion conductance, and thisprovides a means to address the question ofwhat drives translocation in this system. LFN

enters the PA63 pore at small positive voltagesand blocks ion conductance (81, 83, 84). Atlarge positive voltages it is translocated to theopposite side of the membrane, thereby un-blocking the pore (84). Thus all of the translo-cation machinery is apparently contained inthe PA63 pore, and, at least in this system,translocation does not require cellular pro-teins or ATP. Using this in vitro system asa model of translocation across the endoso-mal membrane, one can investigate the ap-plied transmembrane potential (��) and aproton gradient (�pH) as potential sourcesof energy to drive translocation. When thesetwo parameters were varied independently,translocation of LFN and of full-length LFand EF was found to be markedly stimulatedby �pH—that is, when the pH of the transside of the membrane (corresponding to thecytosol) was greater than that of the cis side(corresponding to the endosomal lumen) (85).For example, at a low �� of +20 mV notranslocation of LFN occurred at pH 5.5 if�pH = 0, whereas there was rapid transloca-tion (t1/2 ∼10 s) when the trans pH was raisedto 6.2. Little translocation of whole LF orEF occurred at symmetric pH 5.5, even at amoderate �� of +50 mV, but introduction ofa �pH resulted in significant translocation.In comparing the �� dependence of LFN

translocation in the presence and absence ofa one-unit �pH (pHtrans = 6.5, pHcis = 5.5),the �pH was better able to promote translo-cation at lower voltages than at higher volt-ages. Extrapolating the translocation t1/2 backto 0 mV revealed that the �pH would accel-erate translocation ∼100-fold relative to thatexpected at symmetric pH. It is generally ac-cepted that the sign of �� across the endoso-mal membrane is positive (promoting passage


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

Brownian ratchet:in the currentcontext, amechanism oftranslocation across amembrane involvingbiased randomthermal motion anddriven by a chemicalasymmetry betweenthe two sides of themembrane

Phenylalanineclamp: a structure,formed by the sevenPhe427 side chainswithin the PA63 porethat is essential fortranslocation and isbelieved to interactdirectly with thetranslocatingpolypeptide

of cations from the cis to the trans compart-ment and anions in the opposite direction),but estimates of the magnitude of �� varywidely (86–88). The translocation process inplanar bilayers closely matches results seen invivo when the applied �� is small (0–20 mV),suggesting that the �� across the endosomalmembrane is within this range.

Taken together, these findings suggest thattranslocation across the endosomal mem-brane is driven primarily by the proton gra-dient, although �� may play a role, partic-ularly in the initial threading of the cationicN-terminal region through the pore to thecytosol. How might the proton gradientdrive translocation? A charge-state Brownianratchet mechanism has been proposed (85).The pore is cation selective (67); hence it dis-favors the passage of anions and thus also ofnegatively charged segments of a translocat-ing polypeptide. The relative rates of proto-nation and deprotonation of acidic side chainson the cis and trans sides of the membraneshould thus bias Brownian fluctuations of thetranslocating polypeptide, driving the chainthrough the pore in a unidirectional manner.A stretch of anionic polypeptide can only en-ter the cation-selective portion of the pore af-ter it is protonated and enough of the negativecharge is neutralized to make the stretch neu-tral or cationic. As the stretch exits the poreand enters the higher pH of the trans com-partment (cytosol), the acidic residues becomedeprotonated, giving the segment a negativecharge and blocking back diffusion into thepore. The result is a unidirectional diffusionof the polypeptide across the membrane, ac-companied by unfolding of the polypeptideas it is threaded into the pore and ultimatelyby refolding after it has emerged into the cy-tosol. Refolding would not be expected to bea significant driving force until relatively longsegments of polypeptide had emerged fromthe pore, and experimental evidence supportsthis assumption (85). The extant data do notexclude the possibility that chaperonins, ei-ther endosomal or cytosolic, may play a rolein translocation in vivo.

The Phenylalanine Clamp and ItsRole in Translocation

What role does the PA63 pore play in thetranslocation of EF and LF across the en-dosomal membrane? Recent studies indicatethat the pore does not serve simply as apassive, water-filled conduit, but rather, ac-tively catalyzes the passage of substrate pro-teins across the membrane. This conclusionderives primarily from studies of Phe427 ofPA (18). Mutation of Phe427 to Ala wasfound to have a drastic inhibitory effect onthe ability of PA to mediate translocationof LFNDTA into cells. Indeed, this muta-tion was dominantly negative, although itdid not block formation of pores in pla-nar bilayers. In the crystallographic structureof the prepore, the seven Phe427 residuesare seen to be luminal, near the base ofthe structure, and ∼20 A apart in neighbor-ing subunits. Three lines of evidence indi-cate that these residues remain luminal andsolvent-exposed when the prepore converts tothe pore. (a) Single-channel ion conductancewas diminished by replacing Phe427 with abulkier residue, such as Trp, and increased by aless bulky residue, such as Ala. (b) When porescontaining the Phe427Cys mutation were re-acted with MTS-ET, a reagent that cova-lently links a positively charged group to thethiol, ion conductance was strongly inhibited.(c) When PA Phe427ys was reacted with a spinprobe, the electron paramagnetic signatureshowed only weak spin-spin interaction in theprepore, but strong interaction in the pore.The data from the spin-spin interaction stud-ies indicate that the Phe427 side chains comecloser together, perhaps within 10 A of oneanother, as the prepore converts to the pore.

How the Phe427 side chains are reposi-tioned during prepore-to-pore conversion re-mains to be determined, but recent evidencesuggests the existence of a scaffold in whichan acidic residue, Asp426, immediately adja-cent to Phe427, forms a salt bridge with a ba-sic residue, Lys397, in an adjacent subunit ofthe pore (Figure 4) (89). While the members


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

of this salt bridge are conserved in other bi-nary toxins produced by Bacillus species, bothresidues are Gln in Clostridial homologues ofPA, indicating that hydrogen bonds replaceionic interactions in the Clostridial toxins.

When various amino acid replacements forPhe427 were tested for effects on the kineticsof translocation in the planar bilayer system,the results correlated well with the effects seenon toxin action in cultured cells (18). Phe atposition 427 was most effective in promot-ing translocation, but Leu, Trp, and Tyr alsoshowed strong activity. Ile was much less ac-tive than Leu, and Ala, Val, Val, Asp, Ser,and Gly were essentially completely inactive.Thus, the presence of a hydrophobic residueat position 427 is important in promotingtranslocation, but not just any hydrophobicresidue will function, as both Ile and Val werevirtually inactive. This suggests that a discretestructure involving Phe427 forms within thepore and that the structure is sensitive to smalldifferences in side-chain structure at thatposition.

The results described indicate that theseven Phe427 residues form a structure in thelumen of the pore that facilitates transloca-tion of LF and EF. Where is this structure lo-cated, and how does it facilitate translocation?The pattern of MTS-ET effects in the vicinityof position 427 appears to be inconsistent withPhe427 being in the 14-strand β-barrel of thepore, and it is thus in the cap region, perhapsnear the proximal end of the β-barrel (18).Precisely how the Phe427 structure functionsremains uncertain, but clues have come fromplanar bilayer studies. In single-channel mea-surements, LFN is seen to produce a com-pletely and continuously closed state of ionconductance. In contrast, with the Phe427Alachannel, LFN produces a dynamic “flickering”between an open state, a fully closed state,and multiple partly closed substates. The sim-plest interpretation of these findings is thatthe wild-type Phe427 structure “clamps” theflexible N terminus of LFN in such a mannerthat passage of hydrated K+ ions is prevented;hence, the term phenylalanine clamp (Phe

Figure 4(a) Ribbon model of a PA63 monomer, showing the locations of Lys397,Asp426, and Phe427. (b) Conceptual model of the formation of a scaffoldfor Phe427 in the PA63 pore by the interaction of the Lys397 and Asp426side chains. This figure is adapted from (89).

clamp or φ clamp) has been coined for thestructure. With Ala at position 427, the sub-strate polypeptide is not stably bound withinthe lumen and is unable to block conductanceeffectively.

Precisely how the Phe clamp interactswith substrate polypeptides to foster translo-cation remains uncertain. However, the clampwas found to be the major conductance-blocking site for certain hydrophobic drugsand hydrophobic model cations, such as tetra-butylammonium (TBA) ions (90–92). For


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

example, the affinity of TBA for Phe427Alachannels relative to wild-type channels wasreduced by a factor of 4000, and the affinityof the polyaromatic, 4-aminoquinolone drug,quinacrine (93), was reduced by 1000 (18). Ex-amination of a library of 35 quaternary am-monium and phosphonium ion compoundsled to the conclusion that the Phe clamp doesnot recognize specific geometric or steric fea-tures of the substrates. Instead, it appears torecognize compounds primarily by nonspe-cific hydrophobic interactions, although itsπ-electron clouds may contribute electrostat-ically through aromatic-aromatic, π-π, andcation-π interactions.

The characteristics of the Phe clamp ininteracting with small molecules, combinedwith the fact that substrate polypeptides mustunfold in order to be translocated, are consis-tent with a chaperone-like function in whichthe clamp’s phenyl rings interact primarilywith exposed hydrophobically dense segmentsof the substrate polypeptides (18). In LF andEF, as is true for globular proteins in general,hydrophobically dense segments occur peri-odically along the chain. As the molten globu-lar substrate protein undergoes transient un-folding and is drawn incrementally into thechannel, such hydrophobic segments wouldbe expected to bind favorably to the Phe clampand thus to pause. However, given that thePhe clamp site actually catalyzes transloca-tion, it must reduce some other energy bar-

rier, such as unfolding of the substrate pro-tein. In essence, the Phe clamp site may createan environment that mimics the hydropho-bic core of the unfolding molten globularprotein. Transient interaction of hydropho-bic segments of the translocating polypeptidewith the Phe clamp would reduce the energypenalty of exposing hydrophobic side chainsto the solvent or the hydrophilic lumen of thechannel.

Alternatively, the Phe clamp may functionprimarily to form a seal around the translocat-ing polypeptide, blocking the passage of ions,as shown in single-channel conductance ex-periments (18, 85). This seal would preservethe proton gradient across the membrane, atleast in the microenvironment of an individ-ual pore, and thereby maintain this gradient asa potential energy source for driving translo-cation. It is conceivable that the Phe clampserves both as a seal and a chaperonin. The�pH-driven charge-state ratchet may work intandem with the Phe clamp, which has beenproposed to be a hydrophobic ratchet.

In the final analysis, the PA63 pore may beconceived to function as an enzyme, with thesubstrates being LF and EF bound at the poreentrance in the endosome, the products be-ing these proteins delivered to the cytosol,the active site being the Phe clamp, and thedriving force being the transmembrane pHgradient. In a sense, the pore functions as aproton/protein symporter.

SUMMARY POINTS

1. Pathogenic bacteria have evolved a number of systems to deliver harmful enzymesinto host cells. The tripartite anthrax toxin is one of the most tractable of such systemsfor study of the entry process.

2. Entry begins when the pore-forming component, protective antigen (PA), binds to acellular receptor (ANTXR1 or ANTXR2), is proteolytically activated, and oligomer-izes to form a ring-shaped heptamer, called the prepore (Figure 1). The preporebinds one or both of the two enzymatic components, lethal factor (LF) or edema fac-tor (EF), and the resulting complexes are trafficked to the endosome. There, underthe influence of acidic pH, the prepore forms a transmembrane pore, and LF and EFunfold and cross the endosomal membrane via the pore to the cytosol.


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

3. The pore is believed to have a mushroom shape, with a globular cap, where LF and EFbind via their N-terminal domains, and a 14-strand transmembrane β-barrel formedby the interaction of 7 β-hairpins from domains 2 of the heptamer.

4. Both receptors, ANTXR1 and ANTXR2, have a VWA/I domain which binds PA.Contacts involve the domain 2 of PA as well as its canonical receptor-binding domain(domain 4). ANTXR2 binds PA more tightly than ANTXR1, and a lower pH isrequired for pore formation when the prepore is bound to the former.

5. PA forms cation-selective pores in planar phospholipid bilayers in the absence ofreceptors. The translocation of LF/EF through these pores can be replicated in thissystem under conditions close to those in the endosome.

6. When LF (or EF) binds the pore in the planar bilayer system, its flexible positivelycharged N terminus enters, blocking ion conductance and initiating an N- to C-terminal translocation process. Translocation is driven largely by a transmembraneproton gradient, from an acidic compartment (the endosome) to a neutral or lessacidic one (the cytosol).

7. Rather than serving as a passive, water-filled conduit, the PA pore plays an active rolein translocating LF and EF. The seven Phe427 side chains within the lumen of thepore form a structure called the Phe clamp, which is required for protein translocationthrough the pore. The Phe clamp may perform either or both of two functions: i)by forming a seal around the translocating polypeptide, it may prevent ion passageand thereby preserve the force-generating proton gradient; ii) by transiently bindinghydrophobic segments of the translocating polypeptide, it may reduce the energybarrier to unfolding of the translocating polypeptide.

8. Translocation of LF and EF through the PA pore is proposed to occur by a charge-stateBrownian ratchet mechanism. Because the pore excludes negatively charged segmentsof polypeptide, acidic residues can enter only after being protonated in the endosome(low pH). After they reach the cytosol (neutral pH), these segments are then rapidlydeprotonated and cannot back diffuse into the pore. This results in a directionallybiased diffusion through the pore by the translocating protein.

FUTURE ISSUES

1. Structure of the PA pore. Although crystallographic structures of monomeric PA,LF and EF, the prepore, and the ANTXR2-bound forms of monomeric PA and theprepore are known, direct structural information about the pore is lacking. Becausethe pore is a integral membrane protein, crystallization is a major challenge.

2. Does the pore remain associated with receptor? The PA prepore can form poresin planar phospholipid bilayers, indicating that pore formation is not dependent onreceptors or other cellular proteins, and there is evidence from studies in cells that thereceptor dissociates once the pore is formed. Nonetheless questions remain about thepossible retention of interaction of the pore with receptors or other cellular proteins.


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

3. Are cellular chaperones involved in the unfolding, translocation and/or refolding ofLF and EF? Although the translocation process can be replicated in planar phospolipidbilayers, it is conceivable that cellular chaperones may aid in this process in vivo.Results from studies in isolated endosomes have suggested that Hsp90 is involved inthe translocation and/or refolding of diphtheria toxin, which also enters the cytosolfrom endosomes (94).

4. Can a single PA pore translocate more than one bound molecule of LF and/or EFafter it enters the endosome? The efficiency of translocation in vivo is in need offurther study.

DISCLOSURE STATEMENT

Both authors hold stock in PharmAthene, Inc. R.J. Collier is a consultant for a company thatmay develop drug treatments for anthrax.

ACKNOWLEDGMENTS

We thank John Naughton for help in preparing the figures. Our experimental research onanthrax toxin and its receptors has been supported by grants from the National Institute ofAllergy and Infectious Diseases (AI22021, AI84849, and AI56013). This review was writtenin part while one of us (R.J.C.) was a guest in the laboratory of Prof. Klaus Aktories at theUniversity of Freiburg, working under a Research Award from the Alexander von HumboldtFoundation.

LITERATURE CITED

1. Abrami L, Reig N, van der Goot FG. 2005. Trends Microbiol. 13:72–782. Agrawal A, Pulendran B. 2004. Cell Mol. Life Sci. 61:2859–653. Collier RJ, Young JA. 2003. Annu. Rev. Cell Dev. Biol. 19:45–704. Moayeri M, Leppla SH. 2004. Curr. Opin. Microbiol. 7:19–245. Mock M, Fouet A. 2001. Annu. Rev. Microbiol. 55:647–716. Scobie HM, Young JA. 2005. Curr. Opin. Microbiol. 8:106–127. Barth H, Aktories K, Popoff MR, Stiles BG. 2004. Microbiol. Mol. Biol. Rev. 68:373–4028. Duesbery NS, Webb CP, Leppla SH, Gordon VM, Klimpel KR, et al. 1998. Science

280:734–379. Vitale G, Pellizzari R, Recchi C, Napolitani G, Mock M, Montecucco C. 1998. Biochem.

Biophys. Res. Commun. 248:706–1110. Leppla SH. 1982. Proc. Natl. Acad. Sci. USA 79:3162–6611. Petosa C, Collier RJ, Klimpel KR, Leppla SH, Liddington RC. 1997. Nature 385:833–3812. Lacy DB, Wigelsworth DJ, Melnyk RA, Harrison SC, Collier RJ. 2004. Proc. Natl. Acad.

Sci. USA 101:13147–5113. Santelli E, Bankston LA, Leppla SH, Liddington RC. 2004. Nature 430:905–814. Benson EL, Huynh PD, Finkelstein A, Collier RJ. 1998. Biochemistry 37:3941–4815. Qa’dan M, Christensen KA, Zhang L, Roberts TM, Collier RJ. 2005. Mol. Cell Biol.

25:5492–9816. Mogridge J, Mourez M, Collier RJ. 2001. J. Bacteriol. 183:2111–16


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

17. Rosovitz MJ, Schuck P, Varughese M, Chopra AP, Mehra V, et al. 2003. J. Biol. Chem.278:30936–44

18. Krantz BA, Melnyk RA, Zhang S, Juris SJ, Lacy DB, et al. 2005. Science 309:777–8119. Bradley KA, Mogridge J, Mourez M, Collier RJ, Young JA. 2001. Nature 414:225–2920. Scobie HM, Rainey GJ, Bradley KA, Young JA. 2003. Proc. Natl. Acad. Sci. USA 100:5170–

7421. Liu S, Leppla SH. 2003. J. Biol. Chem. 278:5227–3422. Rainey GJ, Wigelsworth DJ, Ryan PL, Scobie HM, Collier RJ, Young JA. 2005. Proc. Natl.

Acad. Sci. USA 102:13278–8323. Nanda A, Carson-Walter EB, Seaman S, Barber TD, Stampfl J, et al. 2004. Cancer Res.

64:817–2024. Bell SE, Mavila A, Salazar R, Bayless KJ, Kanagala S, et al. 2001. J. Cell Sci. 114:2755–7325. St Croix B, Rago C, Velculescu V, Traverso G, Romans KE, et al. 2000. Science 289:1197–

20226. Carson-Walter EB, Watkins DN, Nanda A, Vogelstein B, Kinzler KW, St Croix B. 2001.

Cancer Res. 61:6649–5527. Hotchkiss KA, Basile CM, Spring SC, Bonuccelli G, Lisanti MP, Terman BI. 2005. Exp.

Cell Res. 305:133–4428. Rmali KA, Al-Rawi MA, Parr C, Puntis MC, Jiang WG. 2004. Int. J. Mol. Med. 14:75–8029. Bonuccelli G, Sotgia F, Frank PG, Williams TM, de Almeida CJ, et al. 2005. Am. J. Physiol.

Cell Physiol. 288:C1402–1030. Rmali KA, Puntis MC, Jiang WG. 2005. Biochem. Biophys. Res. Commun. 334:231–3831. Hanks S, Adams S, Douglas J, Arbour L, Atherton DJ, et al. 2003. Am. J. Hum. Genet.

73:791–80032. Dowling O, Difeo A, Ramirez MC, Tukel T, Narla G, et al. 2003. Am. J. Hum. Genet.

73:957–6633. Wigelsworth DJ, Krantz BA, Christensen KA, Lacy DB, Juris SJ, Collier RJ. 2004. J. Biol.

Chem. 279:23349–5634. Lacy DB, Wigelsworth DJ, Scobie HM, Young JA, Collier RJ. 2004. Proc. Natl. Acad. Sci.

USA 101:6367–7235. Bradley KA, Mogridge J, Rainey GJA, Batty S, Young JAT. 2003. J. Biol. Chem. 278:49342–

4736. Shimaoka M, Takagi J, Springer TA. 2002. Annu. Rev. Biophys. Biomol. Struct. 31:485–51637. Scobie HM, Wigelsworth DJ, Marlett JM, Thomas D, Rainey GJA, et al. 2006. PLoS

Pathog. 2:949–5538. Scobie HM, Thomas D, Marlett JM, Destito G, Wigelsworth DJ, et al. 2005. J. Infect. Dis.

192:1047–5139. Ustinov VA, Plow EF. 2002. J. Biol. Chem. 277:18769–7640. Klimpel KR, Molloy SS, Thomas G, Leppla SH. 1992. Proc. Natl. Acad. Sci. USA 89:10277–

8141. Molloy SS, Bresnahan PA, Leppla SH, Klimpel KR, Thomas G. 1992. J. Biol. Chem.

267:16396–40242. Christensen KA, Krantz BA, Melnyk RA, Collier RJ. 2005. Biochemistry 44:1047–5343. Gao-Sheridan S, Zhang S, Collier RJ. 2003. Biochem. Biophys. Res. Commun. 300:61–6444. Milne JC, Furlong D, Hanna PC, Wall JS, Collier RJ. 1994. J. Biol. Chem. 269:20607–1245. Christensen KA, Krantz BA, Collier RJ. 2006. Biochemistry 45:2380–8646. Abrami L, Liu S, Cosson P, Leppla SH, van der Goot FG. 2003. J. Cell Biol. 160:321–2847. Abrami L, Leppla SH, van der Goot FG. 2006. J. Cell Biol. 172:309–20


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

48. Beauregard KE, Collier RJ, Swanson JA. 2000. Cell Microbiol. 2:251–5849. Elliott JL, Mogridge J, Collier RJ. 2000. Biochemistry 39:6706–1350. Pimental RA, Christensen KA, Krantz BA, Collier RJ. 2004. Biochem. Biophys. Res. Commun.

322:258–6251. Mogridge J, Cunningham K, Collier RJ. 2002. Biochemistry 41:1079–8252. Mogridge J, Cunningham K, Lacy DB, Mourez M, Collier RJ. 2002. Proc. Natl. Acad. Sci.

USA 99:7045–4853. Cunningham K, Lacy DB, Mogridge J, Collier RJ. 2002. Proc. Natl. Acad. Sci. USA 99:7049–

5354. Ren G, Quispe J, Leppla SH, Mitra AK. 2004. Structure 12:2059–6655. Lacy DB, Mourez M, Fouassier A, Collier RJ. 2002. J. Biol. Chem. 277:3006–1056. Lacy DB, Lin HC, Melnyk RA, Schueler-Furman O, Reither L, et al. 2005. Proc. Natl.

Acad. Sci. USA 102:16409–1457. Melnyk RA, Hewitt KM, Lacy DB, Lin HC, Gessner CR, et al. 2006. J. Biol. Chem.

281:1630–3558. Boll W, Ehrlich M, Collier RJ, Kirchhausen T. 2004. Eur. J. Cell Biol. 83:281–8859. Wei W, Lu Q, Chaudry GJ, Leppla SH, Cohen SN. 2006. Cell 124:1141–5460. He X, Semenov M, Tamai K, Zeng X. 2004. Development 131:1663–7761. Santy LC, Casanova JE. 2002. Curr. Biol. 12:R360–6262. Lu Q, Wei W, Kowalski PE, Chang AC, Cohen SN. 2004. Proc. Natl. Acad. Sci. USA

101:17246–5163. Krauss M, Kinuta M, Wenk MR, De Camilli P, Takei K, Haucke V. 2003. J. Cell Biol.

162:113–2464. Paleotti O, Macia E, Luton F, Klein S, Partisani M, et al. 2005. J. Biol. Chem. 280:21661–6665. Abrami L, Lindsay M, Parton RG, Leppla SH, van der Goot FG. 2004. J. Cell Biol.

166:645–5166. Maxfield FR, McGraw TE. 2004. Nat. Rev. Mol. Cell Biol. 5:121–3267. Blaustein RO, Koehler TM, Collier RJ, Finkelstein A. 1989. Proc. Natl. Acad. Sci. USA

86:2209–1368. Koehler TM, Collier RJ. 1991. Mol. Microbiol. 5:1501–669. Milne JC, Collier RJ. 1993. Mol. Microbiol. 10:647–5370. Wesche J, Elliott JL, Falnes PO, Olsnes S, Collier RJ. 1998. Biochemistry 37:15737–4671. Sun J, Vernier G, Wigelsworth DJ, Collier RJ. 2007. J. Biol. Chem. 282:1059–6572. Nassi S, Collier RJ, Finkelstein A. 2002. Biochemistry 41:1445–5073. Nguyen TL. 2004. J. Biomol. Struct. Dyn. 22:253–6574. Wolfe JT, Krantz BA, Rainey GJ, Young JA, Collier RJ. 2005. J. Biol. Chem. 280:39417–2275. Sellman BR, Nassi S, Collier RJ. 2001. J. Biol. Chem. 276:8371–7676. Mourez M, Yan M, Lacy DB, Dillon L, Bentsen L, et al. 2003. Proc. Natl. Acad. Sci. USA

100:13803–877. Sellman BR, Mourez M, Collier RJ. 2001. Science 292:695–9778. Yan M, Collier RJ. 2003. Mol. Med. 9:46–5179. Krantz BA, Trivedi AD, Cunningham K, Christensen KA, Collier RJ. 2004. J. Mol. Biol.

344:739–5680. Pannifer AD, Wong TY, Schwarzenbacher R, Renatus M, Petosa C, et al. 2001. Nature

414:229–3381. Zhang S, Finkelstein A, Collier RJ. 2004. Proc. Natl. Acad. Sci. USA 101:16756–6182. Blanke SR, Milne JC, Benson EL, Collier RJ. 1996. Proc. Natl. Acad. Sci. USA 93:8437–4283. Neumeyer T, Tonello F, Dal Molin F, Schiffler B, Orlik F, Benz R. 2006. Biochemistry

45:3060–68


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

ANRV313-BI76-11 ARI 28 April 2007 16:21

84. Zhang S, Udho E, Wu Z, Collier RJ, Finkelstein A. 2004. Biophys. J. 87:3842–4985. Krantz BA, Finkelstein A, Collier RJ. 2006. J. Mol. Biol. 355:968–7986. Rybak SL, Lanni F, Murphy RF. 1997. Biophys. J. 73:674–8787. Sonawane ND, Thiagarajah JR, Verkman AS. 2002. J. Biol. Chem. 277:5506–1388. Van Dyke RW, Hornick CA, Belcher J, Scharschmidt BF, Havel RJ. 1985. J. Biol. Chem.

260:11021–2689. Melnyk RA, Collier RJ. 2006. Proc. Natl. Acad. Sci. USA 103:9802–790. Blaustein RO, Finkelstein A. 1990. J. Gen. Physiol. 96:943–5791. Blaustein RO, Finkelstein A. 1990. J. Gen. Physiol. 96:905–1992. Blaustein RO, Lea EJ, Finkelstein A. 1990. J. Gen. Physiol. 96:921–4293. Orlik F, Schiffler B, Benz R. 2005. Biophys. J. 88:1715–2494. Ratts R, Zeng H, Berg EA, Blue C, McComb ME, et al. 2003. J. Cell Biol. 160:1139–5095. Liu S, Leung HJ, Leppla SH. 2007. Cell Microbiol. 9:977–8796. Chen KH, Liu S, Bankston LA, Liddington RC, Leppla SH. 2007. J. Biol. Chem. PMID:

17251181

NOTE ADDED IN PROOF

During the final preparation of this review, the key role of receptor residue Tyr119 in regulatingthe PH threshold of toxin pore formation was also confirmed by another group (95). In addition,other receptor-specific PA variants were described (96).


Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

AR313-FM ARI 8 May 2007 21:56

Annual Review ofBiochemistry

Volume 76, 2007Contents

Mitochondrial Theme

The Magic GardenGottfried Schatz � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �673

DNA Replication and Transcription in Mammalian MitochondriaMaria Falkenberg, Nils-Göran Larsson, and Claes M. Gustafsson � � � � � � � � � � � � � � � � � � �679

Mitochondrial-Nuclear CommunicationsMichael T. Ryan and Nicholas J. Hoogenraad � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �701

Translocation of Proteins into MitochondriaWalter Neupert and Johannes M. Herrmann � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �723

The Machines that Divide and Fuse MitochondriaSuzanne Hoppins, Laura Lackner, and Jodi Nunnari � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �751

Why Do We Still Have a Maternally Inherited Mitochondrial DNA?Insights from Evolutionary MedicineDouglas C. Wallace � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �781

Molecular Mechanisms of Antibody Somatic HypermutationJavier M. Di Noia and Michael S. Neuberger � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �1

Structure and Mechanism of Helicases and Nucleic Acid TranslocasesMartin R. Singleton, Mark S. Dillingham, and Dale B. Wigley � � � � � � � � � � � � � � � � � � � � � � 23

The Nonsense-Mediated Decay RNA Surveillance PathwayYao-Fu Chang, J. Saadi Imam, Miles F. Wilkinson � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 51

Functions of Site-Specific Histone Acetylation and DeacetylationMona D. Shahbazian and Michael Grunstein � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 75

The tmRNA System for Translational Surveillance and Ribosome RescueSean D. Moore and Robert T. Sauer � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �101

Membrane Protein Structure: Prediction versus RealityArne Elofsson and Gunnar von Heijne � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �125

v

Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

AR313-FM ARI 8 May 2007 21:56

Structure and Function of Toll Receptors and Their LigandsNicholas J. Gay and Monique Gangloff � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �141

The Role of Mass Spectrometry in Structure Elucidation of DynamicProtein ComplexesMichal Sharon and Carol V. Robinson � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �167

Structure and Mechanism of the 6-Deoxyerythronolide B SynthaseChaitan Khosla, Yinyan Tang, Alice Y. Chen, Nathan A. Schnarr,

and David E. Cane � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �195

The Biochemistry of Methane OxidationAmanda S. Hakemian and Amy C. Rosenzweig � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �223

Anthrax Toxin: Receptor Binding, Internalization, Pore Formation,and TranslocationJohn A.T. Young and R. John Collier � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �243

Synapses: Sites of Cell Recognition, Adhesion, and FunctionalSpecificationSoichiro Yamada and W. James Nelson � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �267

Lipid A Modification Systems in Gram-negative BacteriaChristian R.H. Raetz, C. Michael Reynolds, M. Stephen Trent,

and Russell E. Bishop � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �295

Chemical Evolution as a Tool for Molecular DiscoveryS. Jarrett Wrenn and Pehr B. Harbury � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �331

Molecular Mechanisms of Magnetosome FormationArash Komeili � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �351

Modulation of the Ryanodine Receptor and Intracellular CalciumRan Zalk, Stephan E. Lehnart, and Andrew R. Marks � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �367

TRP ChannelsKartik Venkatachalam and Craig Montell � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �387

Studying Individual Events in BiologyStefan Wennmalm and Sanford M. Simon � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �419

Signaling Pathways Downstream of Pattern-Recognition Receptorsand Their Cross TalkMyeong Sup Lee and Young-Joon Kim � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �447

Biochemistry and Physiology of Cyclic Nucleotide Phosphodiesterases:Essential Components in Cyclic Nucleotide SignalingMarco Conti and Joseph Beavo � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �481

The Eyes Absent Family of Phosphotyrosine Phosphatases: Propertiesand Roles in Developmental Regulation of TranscriptionJennifer Jemc and Ilaria Rebay � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �513

vi Contents

Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

AR313-FM ARI 8 May 2007 21:56

Assembly Dynamics of the Bacterial MinCDE System and SpatialRegulation of the Z RingJoe Lutkenhaus � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �539

Structures and Functions of Yeast Kinetochore ComplexesStefan Westermann, David G. Drubin, and Georjana Barnes � � � � � � � � � � � � � � � � � � � � � � � �563

Mechanism and Function of Formins in the Control of Actin AssemblyBruce L. Goode and Michael J. Eck � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �593

Unsolved Mysteries in Membrane TrafficSuzanne R. Pfeffer � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �629

Structural Biology of Nucleocytoplasmic TransportAtlanta Cook, Fulvia Bono, Martin Jinek, and Elena Conti � � � � � � � � � � � � � � � � � � � � � � � � � �647

The Postsynaptic Architecture of Excitatory Synapses: A MoreQuantitative ViewMorgan Sheng and Casper C. Hoogenraad � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �823

Indexes

Cumulative Index of Contributing Authors, Volumes 72–76 � � � � � � � � � � � � � � � � � � � � � � � �849

Cumulative Index of Chapter Titles, Volumes 72–76 � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �853

Errata

An online log of corrections to Annual Review of Biochemistry chapters (if any, 1997to the present) may be found at http://biochem.annualreviews.org/errata.shtml

Contents vii

Ann

u. R

ev. B

ioch

em. 2

007.

76:2

43-2

65. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by U

nive

rsity

of

Mas

sach

uset

ts -

Am

hers

t on

02/1

1/08

. For

per

sona

l use

onl

y.

Anthrax Toxin: Receptor Binding, Internalization, Pore...

Documents

Transcript of Anthrax Toxin: Receptor Binding, Internalization, Pore...