The Protective Antigen Component of Anthrax Toxin Forms...

doi:10.1016/j.jmb.2009.07.037 J. Mol. Biol. (2009) 392, 614–629

Available online at www.sciencedirect.com

The Protective Antigen Component of Anthrax ToxinForms Functional Octameric Complexes

Alexander F. Kintzer1,2, Katie L. Thoren1, Harry J. Sterling1,Ken C. Dong1, Geoffrey K. Feld1, Iok I. Tang1, Teri T. Zhang3,Evan R. Williams1,2, James M. Berger2,3 and Bryan A. Krantz1,2,3⁎
1Department of Chemistry,University of California,Berkeley, CA 94720, USA2California Institute forQuantitative BiomedicalResearch (QB3),University of California,Berkeley, CA 94720, USA3Department of Molecular &Cell Biology, University ofCalifornia, Berkeley,CA 94720, USA
Received 16 June 2009;received in revised form11 July 2009;accepted 13 July 2009Available online20 July 2009

*Corresponding author. DepartmenUniversity of California, Berkeley, CE-mail address: [email protected] used: Atx, anthrax

antigen; LF, lethal factor; EF, edemaEM, electron microscopy; nanoESI,ionization; MS, mass spectrometry;soluble Atx receptor; msATR, monoD4, domain 4; MIL, membrane inseuniversal bilayer buffer.

0022-2836/$ - see front matter © 2009 E

The assembly of bacterial toxins and virulence factors is critical to theirfunction, but the regulation of assembly during infection has not beenstudied. We begin to address this question using anthrax toxin as a model.The protective antigen (PA) component of the toxin assembles into ring-shaped homooligomers that bind the two other enzyme components of thetoxin, lethal factor (LF) and edema factor (EF), to form toxic complexes. Todisrupt the host, these toxic complexes are endocytosed, such that the PAoligomer forms a membrane-spanning channel that LF and EF translocatethrough to enter the cytosol. Using single-channel electrophysiology, weshow that PA channels contain two populations of conductance states,which correspond to two different PA pre-channel oligomers observed byelectron microscopy—the well-described heptamer and a novel octamer.Mass spectrometry demonstrates that the PA octamer binds four LFs, andassembly routes leading to the octamer are populated with even-numbered,dimeric and tetrameric, PA intermediates. Both heptameric and octamericPA complexes can translocate LF and EF with similar rates and efficiencies.Here, we report a 3.2-Å crystal structure of the PA octamer. The octamercomprises ∼20–30% of the oligomers on cells, but outside of the cell, theoctamer is more stable than the heptamer under physiological pH. Thus, thePA octamer is a physiological, stable, and active assembly state capable offorming lethal toxins that may withstand the hostile conditions encounteredin the bloodstream. This assembly mechanism may provide a novel meansto control cytotoxicity.

© 2009 Elsevier Ltd. All rights reserved.

Keywords: anthrax toxin; cell-surface assembly; translocation; lethal toxin;oligomerization
Edited by J. Bowie
Introduction

Anthrax toxin1 (Atx) is a binary toxin (A2B)comprised of three nontoxic proteins that are secretedby Bacillus anthracis and combine on eukaryotic host

t of Chemistry,A 94720, USA.edu.toxin; PA, protectivefactor; WT, wild type;nanoelectrospraydsATR, dimericmeric soluble ATR;rtion loop; UBB,

lsevier Ltd. All rights reserve

cell surfaces to make noncovalent, toxic complexes(SupplementaryData Fig. S1). Protective antigen (PA)is the 83 kDa, cell-binding B component thatultimately forms a translocase channel for the two∼90 kDa, enzymatically active, A components—lethal factor (LF) and edema factor (EF). Followingsecretion, PA binds to the host cell via one of twoknown Atx receptors, ATR12 and ATR23, and is thencleaved by a furin-type protease to make theproteolytically activated form, called nPA. The63 kDa, receptor-bound portion of nPA then self-assembles into a ring-shaped homooligomer, or pre-channel, which has been shown to be heptameric.4-8

The pre-channel can bind LF or EF to make lethal oredema toxins, respectively. These toxin complexes areendocytosed and brought to an acidic compartment.9

Under acidic conditions, the PA pre-channel inserts

d.

http://dx.doi.org/10.1016/j.jmb.2009.07.037

615Protective Antigen Forms Functional Octamers

into themembrane to forma translocase channel.10 LFand EF translocate through the channel to enter thecytosol, where they catalyze reactions that disrupt thehost cell (Supplementary Data Fig. S1).11

Analogous to the staphylococcal α-hemolysinpore,12 PA assumes a similar mushroom-shapedarchitecture8 and β-barrel transmembrane motif.13,14The β-barrel of the PA channel is similarly narrow(∼15-Å in diameter8,15) but considerably longer(∼100 Å) than the α-hemolysin. The narrow channelrequires LF and EF to unfold before translocation.16-18

Acidic endosomal conditions serve two purposes:first, they destabilize LF and EF by aciddenaturation;18 and second, they drive translocationvia a proton gradient (ΔpH).11 PA also contains arequired ring of phenylalanine side chains, or ϕclamp, which catalyzes translocation,19 exemplifyinghow the structure of the channel is crucial to itstranslocase function.Assembly is paramount to Atx function20,21 and its

cellular internalization.22 ATR receptors are interna-lized slowly by the host cell, but PA-bound ATR caninternalize rapidly once it dimerizes,22 making properoligomerization a critical step in the internalizationpathway. ATR is dimeric, further complicating the

brane for the nPA+LFN sample. The δ histogram was fit to oneusingA(δ)=A1/σ1√(π/2) exp(–2δ2/σ1

2)+A2/σ2√(π/2) [exp(-2((δ0.96, respectively, yielding best-fit parameters: peak area A1=deviation, σ1=5.5 (±0.5) pS, σ2=9 (±2) pS. (See also Supplemesmaller (black) and larger (red) PA channels in 100 mMKCl, pHwere acquired at 400 Hz and filtered further with a 100 point/σ

assembly mechanism, since PA oligomers arebelieved to be odd-numbered and heptameric.23,24

ATR2 andAtxwere recently implicated as factors thathelp the B. anthracis bacterium escape the acidicphagolysosome following spore germination, sug-gesting Atx components may assemble in hostileenvironments as well.25 Another potentially interest-ing extracellular assembly mechanism has becomeapparent in the investigation of animals infected withB. anthracis, where it has been shown that anthraxtoxin accumulates in the blood of animals.26 Specifi-cally, LF is found alongside proteolytically activatednPA,27,28 implying that the PA is potentiated forassembly. Relative to toxin produced in vitro, the toxinproduced in vivo (i.e., isolated from the blood ofanimals suffering from anthrax) forms unique assem-blies, as evidenced by their unique resistance toantibody binding, and the in vivo-derived toxin ismore lethal.29 We probe assembly in various cellularand extracellular contexts using electrophysiology,electron microscopy, mass spectrometry, and crystal-lography, and we conclude that the activity of thetoxin may be regulated through assembly, potentiallyaffecting the degree of cytotoxicity throughout thestages of anthrax.

Fig. 1. Heterogeneous PA chan-nel conductance distributions. (a)Two PA samples were analyzed: (i)PA is nicked by trypsin tomake nPA;a 20 kDa piece (PA20) dissociates,allowing PA to oligomerize into thepre-channel on a Q-Sepharose col-umn, making QPA; and (ii) nPA ismixed with LFN to drive oligomer-ization, making nPA+LFN. Eitherpre-channel oligomer formsa channelupon inserting into the membrane.(b) Example of 200 Hz-filtered, sin-gle-channel data collected at aΔψ of20 mV, 100 mM KCl, pH 6.60; γvalues computed by γ= i/Δψ arelisted next to each channel insertion.(c) Normalized histograms of theestimated single-channel γ valuesfor the QPA and nPA+LFN samples.Data bins are 1 pS wide, and thenumber of channels, n, in eachsample are normalized for compar-ison. The samples, QPA (n=360;black bars) and nPA+ LFN (n=107;red bars), are statistically distinct bya non-parametric, lower-tailed,Whitney-Mann test (pN0.95). (d)Histogram of all the pairwise differ-ences, δ, betweenmeasured γ valuesidentified within the same mem-

(dotted line) and two Gaussian (continuous line) functions,–μ2)/σ1)

2)+exp(–2((δ+μ2)/σ1)2)], withR values of 0.89 and

470±80, A2=140±40; mean, μ2=8 (±2) pS; and standardntary Data Table S2.) (e) Single-channel current records for6.6. Arrows indicate the two respective channel sizes. DataGaussian filter to better reveal conductance sub-states.

616 Protective Antigen Forms Functional Octamers

Results

PA can form two different channel sizes

PA channels were inserted into planar lipid bilayersfollowing one of two different assembly methods(outlined in Fig. 1a). A pre-oligomerized sample,called QPA (trypsin-nicked PA assembled on Q-Sepharose anion-exchange resin10) was applied tothe bilayer, and discrete single-channel steps wereobserved (Fig. 1b). QPA channels had a meanconductance of 95.5 pS (n=360). Single-channelconductance values (γ) for wild type (WT) PAchannels were reported in the range 85–110 pS.30 Weobserved that the QPA sample contained two discretesizes of channel: a prevalent, smaller one and a rarer,larger one. The overall γ-value distribution recordedfrommany individualmembraneswasbroad (Fig. 1c).To determine if the method of PA assembly affected

the γ-value distribution, we studied trypsin-nickedPA (nPA) samples assembled in the presence of theamino-terminal domain of LF (LFN), called nPA+LFN.Again, larger-conductance channels appeared amongsmaller-conductance channels in two discrete sizes,but the frequency of observing larger channelsincreased. The mean γ value for nPA+LFN (98 pS,n=107) increased relative to that observed for QPA.Moreover, a lower-tailed Whitney-Mann test (WMT)showed that QPA and nPA+LFN distributions are

Fig. 2. EM studies of heptameric and octameric PA. EM imassembled either in vitro (upper panels a–f) or in vivo on celoctamers (left) and heptamers (right) are shown. The total numheptamers and octamers are given. The proportions of heptamred, respectively. In vitro samples include: (a) nPA assemblheptamer; 2% octamer); (b) nPA assembled in the presence of so74% heptamer; 26% octamer); (c) nPA assembled in the preseheptamer; 1% octamer); (d) nPA assembled in the presence of Lassembled in the presence of EFN; (+EFN; n=5363; 81% hepassembled on an anion-exchange column (QPA S170C; n=2933cells: (g) His6-PA assembled on cells expressing ATR2 (C-CHO;of octamers and heptamers are shown, resulting from referencea) is consistent for all images. The percentages of oligomers aralignments unless noted otherwise. (Specific percentages are l

significantly different (pN0.95). Finally, three separateQPA samples contained a consistently lower meanconductance of∼95pS relative to twoother nPA+LFNsamples, which had a mean conductance of ∼98 pS.Thus, the method of assembly shifts the single-channel conductance distribution (Fig. 1c).Inherent variability in bilayer thickness, combined

with the presence of another larger-conductanceconformation, likely contributes to the broad aggre-gate distributions (Fig. 1c). To circumvent thisproblem, we examined the discrete differences inchannel conductance within each membrane bytabulating the set of all the pairwise differences (δ)in γ values/membrane, {γi–γj}, where i≠ j. A histo-gram of δ values recorded for the nPA+LFN sampleshows that two sizes of channel conductance arepresent, as the distribution fits best to a two-Gaussiandistribution (Fig. 1d). Based upon our fit, these twopopulations of γ values differ from one another by 8(±2) pS, or about ±10%,where the larger conductancestate represents 25 (±6)%of the population. This resultled us to conclude that PA forms two discreteconductance states that may be populated differen-tially, depending upon the method of assembly.We testedwhether large-conductance channelswere

a substate of small-conductance channels due to aconformational rearrangement. To test this possibility,we measured ∼6 min recordings of single channels(Fig. 1e). While fluctuations to a more conductingsubstate are observed, these fluctuations are relatively

ages of negatively stained samples of WT PA oligomersl surfaces (lower panel). Representative class averages ofber of particles assessed, n, and the relative percentages ofers and octamers are indicated by bars colored black anded on an anion-exchange column (QPA; n=12589; 98%luble dimeric ATR2 at 4:1 stoichiometry (+dsATR; n=837;nce of soluble monomeric ATR2 (+msATR; n=9401; 99%FN (+LFN; n=8409; 72% heptamer; 28% octamer); (e) nPA

tamer; 19% octamer); and (f) disulfide-bonded PA S170C; 78% heptamer; 22% octamer). Oligomers extracted fromn=4729; 74% heptamer; 26% octamer), where three classes-based analysis. The scale bar representing 5 nm (shown ine means of reference-free and crystal structure-referencedisted in Supplementary Data Table S2.)


small; and large-conductance channels (N102 pS) didnot interconvert to smaller ones (b95 pS) during theseand all other recordings. Therefore, either two uniquechannel sizes exist in PA samples or the timescale ofthe conformational rearrangements between the large-and small-conductance states is slow.

Electron microscopy reveals two PA oligomers

Under the assumption that the ∼10% difference inconductance states may correspond to slow-timescalestructural changes in the channel diameter,19,31 weexamined the QPA oligomers for structural hetero-geneity by electron microscopy (EM). Reference-freeanalysis of ∼104 negatively stained particles identi-fied two classes of ring-shaped oligomers—theheptamer and a novel octamer (Fig. 2a).

Mass spectrometry reveals Atx heterogeneity

To further establish the observed heterogeneity, weused nanoelectrospray ionization mass spectrometry(nanoESI-MS). We assembled WT nPAwith WT LFNand then analyzed the mixtures by nanoESI-MS(Fig. 3a). We identified two large molecular mass

Fig. 3. Mass spectrometry stu-dies of Atx assembly. (a) NanoESImass spectrum of nPA co-assembledwith LFN. Multiple charge-state dis-tributions are observed that corre-spond to five different PA-LFNcomplexes. Charge state and mole-cular mass were calculated asdescribed.69 The PA7(LFN)3 distri-bution clearly has thehighest relativeabundance and is shown abovelabeledwith charge states. The insetsshow distributions of less intenseoligomers PA7(LFN)2 and PA8(LFN)4,and low abundances of PA2LFN andPA4(LFN)2 that may be stable inter-mediates in the formation of thehigher-order complexes. The mole-cular mass was assigned from thecharge-state distribution thatresulted in the smallest standarddeviation in calculated molecularmass. (Supplementary Data TableS1 summarizes the observed massesanddescribes the solvent correction.)(b) Assembly kinetics for the 63 kDaPAmonomer, PA7(LFN)3, PA8(LFN)4,and PA4(LFN)2 from a solution ofnPAmixedwith excess LFN. Data forall but PA4(LFN)2 were fit withexponential functions to guide theeye. The appearance of the oligo-mers, PA7(LFN)3 and PA8(LFN)4,coincides with the disappearance ofPA monomer. Data at early timesindicate a rapid increase in theabundance of PA4(LFN)2 followed

by slow decay for t≥5 min., suggesting it is an intermediate in tline is given for PA4(LFN)2 also to guide the eye. All four analyt

species, 537,082 (±186) Da and 631,167 (±217) Da,corresponding to the oligomers, PA7(LFN)3 and PA8(LFN)4, respectively, as well as two minor (and likelyintermediate) complexes of 158,193 (±38) Da and315,395 (±27) Da, corresponding to PA2LFN and PA4(LFN)2, respectively (Supplementary Data Table S1).The PA7(LFN)3 species has been reported.6 Moreover,similar results were obtainedwhen full-length LF andEF were used as ligands (data not shown). The PA8(LFN)4 complex is novel and corroborates the presenceof the octameric PA species, and it demonstrates thatan octamer can carry a payload of four LFs or EFs—onemore than its heptameric counterpart. Finally, thepreviously undetected forms, PA2LFN and PA4(LFN)2,suggest a pathway of assembly for the octamer viaeven-numbered intermediates.We then probed the kinetics of PA assembly with

nanoESI-MS (Fig. 3b). Here, oligomerization wasinitiated by mixing nPA and LFN in a 1:1 molarratio, using the assembly method described in thesingle-channel studies. The ion abundances in the ESImass spectra were recorded at several time points for∼1 h. The abundances of both PA7(LFN)3 and PA8(LFN)4 increase concomitantly. Also, the appearanceof either oligomer is correlated to a decrease in the

he formation of the higher-order complexes; an interpolatedes reach steady-state levels in ∼30 min.


relative ion abundances for free PAmonomer and PA4(LFN)2 species. We observed an initial burst in theabundance of the PA4(LFN)2 species followed by asubsequent decrease with time; the decrease wasconcomitant with the increase in the formation of PA7(LFN)3 and PA8(LFN)4. The trend in PA4(LFN)2abundance suggests that it is a stable intermediatein the formation of the higher-order complexes. PA8(LFN)4, however, is not an intermediate species in theheptamerization pathway, since its relative abun-dance did not decrease during the experiment.

Stabilizing dimeric PA intermediates promotesoctamer formation

To further investigate how even-numbered inter-mediates influence assembly, we conducted EMstudies using PA oligomers prepared in the presenceof a dimeric soluble Atx receptor domain (dsATR),LFN, or EFN (EF's PA-binding, amino-terminaldomain). A dimeric ATR construct was also chosenbased upon evidence that the receptor may exist in adimeric state on cell surfaces.24 PA pre-complexed todsATR (Fig. 2b), but not to monomeric ATR (msATR)(Fig. 2c), showed increased proportions of octamersupon assembly. Further studies demonstrated thatwhen dsATR was loaded under less saturatingconditions, the octamer levels decreased (Supplemen-tary Data Fig. S2B). Thus the more saturatingconditions allow the dsATR sites to fully populatewith PA before assembly, which increases the prob-ability of forming the even-numbered, octameric form.LF or EF can form a ternary complex with PA

dimers,21 and we have observed this species in ourmass spectrometry experiments (Fig. 3a; Supplemen-tary Data Table S1). When LFN or EFN is used toassemble nPA into oligomers (Fig. 2d and e), ∼25% ofthe populationbecameoctameric. This increase is 5- to10-fold more than that observed for PA oligomerizedin the absence of LFN or EFN. Also the S170C PAmutant, which can form a disulfide-bonded homo-dimer (as modeled in Supplementary Data Fig. S2C),increases the proportion of octamers relative tounlinked WT PA (Fig. 2f). Finally, our reference-freeEM analysis was supported, when possible, withreference-based analysis, mass spectrometry, andelectrophysiology (Supplementary Data Table S2).Therefore, we conclude that the observation of anincrease in the proportion of octamers was the resultof increasing the population of even-numbered PA2-precursor complexes.

PA forms octamers on cells

These in vitro results led us to probe the oligomer-ization pathway on cell surfaces. We used a Chinesehamster ovary (CHO) cell line, expressing ATR2,called C-CHO.32 PAwith a carboxy-terminal His6 tag,called His6-PA, was added to the extracellularmedium of cultured C-CHO cells and incubated toassemble. Endosomal acidification was blocked withammonium chloride to prevent the conversion of thepre-channel oligomers to the channel state. The cells

were harvested, lysed in detergent and purified onHis6-affinity resin. SDS-PAGE of His6-pure extractswere Western blotted, confirming that the purifiedfraction contained His6-PA in the 63 kDa form(Supplementary Data Fig. S3).EM images of these extracts identified oligomeric

rings consistent with the size and shape of PAoligomers; however, these complexes were less welloriented than the other in vitro samples, perhaps dueto the presence of cellular components, like full-lengthATR, which may form the observed extensions fromthe oligomeric structure. Several tilted class averageswere obtained to capture this heterogeneity. Fromthese, we determined that ∼20–30% of the oligomerswere octameric (Fig. 2g). Control experiments showthat His6-PA on its own forms ∼1% octamer(Supplementary Data Fig. S2A), confirming that theHis6 tag is not responsible for the high levels ofoctamer observed on cells.We examined extracts fromT-CHO cells (expressing ATR1) and fromC-CHO cellsthatwere treatedwith bothPAandLFN; andwe foundthat octamer levelswere 20∼30%under all cell-surfaceconditions tested (data not shown). Thus, PA forms amixture of octamers and heptamers on cell surfaces.

Crystal structure of the octamer

Mutations were then introduced into PA to probethe molecular mechanism of assembly. The mostinteresting mutations identified disrupted the inter-face between two PA subunits at the interface ofdomain 4 (D4) and the neighboring membraneinsertion loop (MIL) in the adjacent PA subunit. Onemutant replaced the MIL (i.e., residues 305–324) witha type II turn, deleting all possible hydrophobicinteractions between L668 of D4 and F313 and F314 inthe MIL. This type of PA oligomer (PAΔMIL) wasreported to form heptameric rings;33 but our versionmade an enriched source of octameric rings, asobserved by EM.WeusedPAΔMIL oligomers as a concentrated source

of octameric PA, and solved the crystal structure of theoctamer to 3.2-Å resolution (Fig. 4a; SupplementaryData Table S3).Molecular replacement identified eightPA monomers arranged as a ring in the asymmetricunit. We find that the octamer is best described ashaving fourfold noncrystallographic symmetry,because there are two types of PA monomerconformations, called A and B (Fig. 4c), which occupyalternating positions around the ring (Fig. 4d). Fromstructural alignments, these two conformers mainlydiffer in the orientation of D4 (Fig. 4c). This confor-mational heterogeneity is notable, since D4 interactswith the MIL and may be a structural feature in theassembly mechanism. The flexibility of D4 is alsoconsistent with the higher than average B-factorsobserved there and in the MIL of the heptamerstructure.7 Thus, plasticity in these two regions mayprovide a mechanism for octamer formation, wheremodulation of the structure in these regions mayoccur either (i) via ATR binding, which reorients D4relative to D27,34 or (ii) via exposure to a more acidicpH, which may alter the conformation of the MIL.

Fig. 4. X-ray crystal structure of PA in the octameric oligomerization state. Axial views of: (a) the PAΔMIL octamer(PDB 3HVD) side-by-side with (b) the WT PA heptamer (PDB 1TZO).7 Monomer subunit chains are colored uniquely.The MIL is depicted with spheres in the latter structure of WT heptamer. (c) A backbone alignment of two adjacent PAmonomers, chains A and B, called conformation A (red) and B (blue), showing the displacement of D4. (d) Superimposedon a surface rendering of half an octamer is the square planar arrangement of symmetrically related A and B conformerscalculated from the center of mass of each chain. Chains A, C, E, G are conformation A; and chains B, D, F and H areconformation B. Adjacent A-B pair center of masses are 64.3 (±0.1) Å apart at angles of 90 (±0.1)°. Domains are colored as:D1' (magenta), D2 (green), D3 (gold), and D4 (blue). (e) An A-B oligomerization interface split apart to compare relativedifferences in surface area burial at the oligomerization interface of the heptamer and octamer structure among the fourdomains. The domains (upper panel) are colored as in d. The relative degrees of surface area buried (lower panel) arecolored as follows: green, surface buried equally (i.e., to within 10%) in either structure; red, surface buried 10% moreburied in the heptamer; blue, surface buried 10%more in the octamer; and white, surface buried b75% in both structures.All molecular graphics were rendered using CHIMERA.64


Consistent with our single-channel data, the meanpore diameter of the octamer pre-channel is ∼10%larger than that of the heptamer (46 (±4) Å and 40 (±4)Å, respectively) (Fig. 4a and b). The residues lining thepre-channel of the octamer are similar to those in theheptamer; all types of chemistry are represented,though the charge composition is more anionicoverall. The octamer buries ∼3300 Å2 of solvent-accessible surface area per monomer, which is∼800 Å2 less than the heptamer; the MIL and itsinteractions with its neighboring docking groove inD4 largely account for this difference. The octamerforms additional contacts (not found in the heptamer)at sites more proximal to the central pore, accountingfor ∼350 Å2 of additional buried surface per dimerinterface (Fig. 4e). For WT octameric PA, the MILshould form analogous interactions with this D4-docking groove; therefore, the octamer may bury at

least 350 Å2 more surface area per monomer than theheptamerwhen theMIL is present. Thus,WToctamerwill bury ∼6100 Å2 of additional surface relative tothe heptamer, when including the eighth subunit andadditional increases in burial per monomer.The angles between n adjacent monomers arranged

symmetrically about a ring (θ) are ideally equal to180–360/n, and these angles are widened by ∼6° forthe octamer (Fig. 4a) with respect to the heptamer(Fig. 4b). We found this is accomplished by a subtleshift in the inter-monomer packing interfaces (Fig. 4e),where the octamer buriesmore surface area in regionsproximal to the central channel. During assembly, thestericmass of theMILmay act as a nonspecific wedgethat effectively nudges the adjacent monomer towardamore acute θ in the heptamer. Thus, in the absence ofthis constraint, PAΔMIL is able to relax θ to achieve theoctameric configuration. The MIL has two functions:


(i) to form the channel in the membrane; and (ii) tocontrol the oligomerization number of pre-channelassemblies.

The relative stability of the PA heptamer andoctamer

To test whether the stability of octameric andheptameric complexes was different, we incubatedmixtures of heptameric and octameric PA at differenttemperatures and pH. Undermildly acidic conditions(pH 5.7), the heptameric form precipitated almostquantitatively as judged by both EMand nanoESI-MS(Fig. 5a; Supplementary Data Table S2); however, theoctameric form maintained its solubility and per-sisted. This difference provided us with the means toisolate the octameric form for crystallization. Sincethese experiments used the PAΔMIL construct, wetested the relative stability of WT PA oligomersformed in the presence of LFN at physiological pH.After assembly under physiological conditions, we

Fig. 5. Octameric and heptameric PA stability and activheptamers and octamers before and after a reduction of pH aconditions in a and b are pH 8 and 0 °C. (a) Heptameric and oct5.7, 7% ethanol, 4 °C. Before exposure, 25% octamer, 75%heptamer, n=14516. (b) Heptameric and octameric WT PA olexposure to pH 7, 37 °C. Before exposure, 28% octamer, 73heptamer, n=1084. Relative percentages are given by the barsEnsemble protein translocation records measured using planarelative translocase activities two types of samples: QPA+LFNhad been purified and shown to contain ∼90% octamer (red).was translocated at the indicatedΔψ andΔpH conditions. Recoy-axes are normalized to the fraction of substrate-blocked changiven as t½ (time for half of the substrate to translocate). L(t½=57 s) and octameric PA (t½=96 s). LFN translocated at Δ(t½=12 s). EF translocated at Δψ=50 mV, ΔpH=1 using htranslocated at Δψ=60 mV using heptameric (t½=137 s)translocations of LFN at 50 mV through either a large (red) or smtranslocation indicate the beginning and end of each tranconductance levels of the large and small channels for compa

found that the sample could be purified by S400 gel-filtration, generating ∼90% pure octamer, as judgedbyEMandnanoESI-MS (Fig. 5b; SupplementaryDataTable S2). Therefore, we conclude that the heptamericform assembled andwas subsequently inactivated byaggregation under physiological conditions; how-ever, the octameric form persisted as a solublecomplex.

Translocase activity of octameric channels

The pre-channel PA oligomer forms the translocasechannel state at acidic pH.10,33 The β-barrel, whichpenetrates the membrane, is comprised of the MILand adjacent β-strands in D2. Our structure revealsthat the contacts made in D1′ (solvent-accessiblesurface area of ∼1100 Å2) and those immediatelyadjacent to it in D2 (∼1400 Å2) are largely identicaland are sufficient to form and maintain stableoligomeric complexes even when the MIL is notpresent. Therefore, the octamer may form channels,

ity. Negatively-stained EM class-average images of PAnd/or change in the temperature and solvent. The initialameric PAΔMIL before (left) and after (right) exposure to pHheptamer, n=10409. After exposure, 89% octamer, 11%igomer complexes with LFN before (left) and after (right)% heptamer, n=8409. After exposure, 91% octamer, 9%at the right for heptamers (black) and octamers (red). (c)r lipid bilayer electrophysiology. The panels compare the, which is N95% heptameric (black); and nPA+LFN, whichFour substrates (LF, LFN, EF, or EFN) were used, and eachrds shown are the average of a set of three repetitions. Thenels that become unblocked due to translocation. Rates areF translocated at Δψ=30 mV, ΔpH=1 using heptamericψ=50 mV using heptameric (t½=16 s) and octameric PAeptameric (t½=50 s) and octameric PA (t½=66 s). EFNand octameric PA (t½=135 s). (D) Left: Single-channelall (black) channel. Black arrowheads on either end of the

slocation. Right: Histogram profiles of portions of therison of the open-channel conductance levels.


and the two Gaussian populations of conductancelevels observed in planar bilayers (Fig. 1b and c) mayreflect octamers and heptamers, which have insertedstably into membranes.We then tested this further and asked whether

octamers and heptamers possess similar translocaseactivity. Here, we compared QPA, which is ∼98%heptameric, to a sample, nPA+LFN,∼90% enriched inoctamer. In an ensemble translocation experiment,channels are first inserted into a membrane at 20 mV;the channels are loaded with LFN, which blocks theconductance; excess LFN is removed by perfusion;and then the LFN is translocated at a higher voltage, orΔψ. We observe that LF, EF, LFN, and EFN translocatethrough the two different PA oligomers with similarefficiencies and rates (Fig. 5c). (Efficiency is themeasured amplitude translocated divided by themaximum theoretical amplitude; the rate is estimatedby the time it takes for half of the protein totranslocate).To further verify that larger and smaller channels

are capable of translocating protein, we performedsingle-channel translocation experiments (Fig. 5d).Here, we formed single PA channels at 20 mV andthen added LFN. Once the channel closed due to LFNbinding, the voltage was raised to 50 mV. Transloca-tion events (n∼10) were recorded until the channelbecame inactive. This procedure was repeated on asecond channel obtained in the same membrane,which had a∼10% larger conductance. Therefore, weconclude that large and small PA channels are

Fig. 6. Heterogeneous assembly mechanism may modulatATR sites and assemble into PA2 and PA4 intermediates. Intermload with EF and/or LF. (In principle, LF and EFmay be involv(Bottom) During extracellular or ATR-independent assemblyPA2LFN and PA4(LFN)2, which then form either PA8(LFN)4 or Ptheoretical model.70 Toxin activity is a combination of the oligoto the formation of inactive oligomeric complexes (*).

functional translocases, and both the octameric andheptameric forms of the PA channel are functional.

Discussion

We suggest that the octameric form of PA has notbeen observed before,4-8 because the standardmethod of PA assembly, which uses an anion-exchange column,10 yields oligomers virtually devoidof octamers (Fig. 2a). Assembly in the presence ofligand, LFN, EFN, LF, EF or dimeric ATR, produces a25–30% population of octameric oligomers in vitro(Figs. 1, 2b, and d). This heterogeneous assemblymechanism is of physiological significance, because asimilar proportion of octameric oligomers is observedon the surface of cells (Fig. 2g).Our crystal structure ofthe octamer is not a regular octagon, but rather iscomposed of four PAdimer pairs arranged in a squareplanar symmetry. This symmetry suggests thatdimeric PA intermediates populate the assemblypathway, and indeed mass spectrometry revealsthat dimeric PA species are general assembly inter-mediates (Fig. 3b). While heptameric and octamericchannels have similar translocase activity, octamericoligomers are more stable under physiological pHand temperature (Fig. 5). We propose that these twodifferent oligomerization states are functionally rele-vant to anthrax pathogenesis; namely, in the twodifferent environments in which the toxin assembles(Fig. 6). (i) On cell surfaces, assembly bottlenecksmay

e toxin activity. (Top) On cells, PA may encounter dimericediates can combine to form either PA8 or PA7, which caned in the mechanism as well to produce similar outcomes.), PA may encounter LF or EF, making the intermediates,A7(LFN)3. Models of LF-PA complexes are derived from amer's stability and translocase activity; instability may lead


be alleviated, allowing for proper endocytosis offunctional complexes by having the two assemblyroutes. (ii) In blood plasma, PA may assemble beforereaching the cell surface and require a more stableoligomeric configuration, since the heptamer isweakly stable under physiological conditions, espe-cially in the absence of its cellular receptor.

Cell-surface assembly and endocytosis

The current model for anthrax toxin assemblyproposes that PA assembles into a heptamer on cellsurfaces expressing ATRs (Supplementary Data Fig.S1). Initially, PA binds a cell-surface receptor, ATR1 orATR2, is cleaved by a furin-type protease, and thenbegins to assemble into the ring-shaped oligomer.While the currentmodel predicts that these oligomerswill be heptameric,1 the oligomeric states populatedon cell surfaces have not been reported.We addressedthis question specifically by extracting oligomers fromcell surfaces and analyzing the distribution ofoligomeric states by EM. We find that PA formsboth a heptamer and the novel octamer in ratio of∼2:1, using cell lines expressing either ATR1 or ATR2.What selection pressures on the toxin might maintainthis dual-oligomerization-state mechanism? It isthought that cell-surface assembly and endocytosismay be coupled processes. For example, the rate ofcellular internalization through endocytosis for eitherfree ATR or PA-boundATR is slow; and this basal rateis accelerated only if the PA is pre-assembled into PAheptamers or PA-boundATR subunits are aggregatedby antibody cross-linking.22 Therefore, PA assemblyand the corresponding aggregation of ATRs triggersendocytosis.Our experiments extend the current understanding

of assembly and now we can improve the model.First, we assume that ATR1 and ATR2 are effectivelydimeric on cell surfaces, because ATR-mediatedassembly in solution promotes only octameric assem-bly when the receptor is dimeric (Fig. 2b and c). Adimeric receptor model agrees with previous studies,which examined the aggregation state of ATR1'stransmembrane helix.24 Therefore, we propose thatthe antibody cross-linking experiment reported byAbrami et al.22 more likely involves a higher-orderaggregation of PA subunits beyond the formation ofPAdimers.We anticipate this, because it is known thatLF and EF promote the dimerization of PA (Fig. 3);21

and an unintended consequence of this dimerizationmay be that PA2LF or PA2EF ternary complexes areendocytosed improperly before assembly into func-tional ring-shaped oligomeric complexes. Thus, anefficient coupling of cell-surface assembly and endo-cytosis would limit premature endocytosis of non-functional dimers and tetramers.A key function of the dual assembly mechanism

may be to allow assembly under a range ofconcentrations of the PA monomer. If we considerthe two extreme cases of either low or highconcentrations of the PA monomer, assembly wouldbe hindered if only one oligomeric state was possible.On one hand, supposing only the octameric assembly

pathwaywas possible and the concentration of the PAmonomer was low, then the dimeric ATR sites maynot be saturated, and assembly would be inhibited atthe critical oligomerization step, allowing unas-sembled complexes to be endocytosed. On the otherhand, supposing only the heptameric form waspossible and the concentration of the PA monomerwas high, the dimeric ATR sites would be fullysaturated with PA, and assembly would be inhibiteduntil one PA could dissociate from its ATR bindingsite, allowing for an odd number of subunits toassemble immediately before endocytosis. This latterscenario is especially prohibited by knowing that thedissociation lifetime for the PA–ATR2 interaction is ofthe order of days.35 In summary, the mixed-oligomer-izationmechanism alleviates these potential assemblybottlenecks, allowing the toxin components to fullyassemble and remain efficacious under the widedynamic range of extracellular concentrations of PA.

Extracellular assembly in blood plasma

Anthrax toxin complexes were first identified in theblood of B. anthracis-infected guinea pigs; and thisblood (after sterilization with antibiotic treatment)could be administered to a second uninfected animalto impart its lethal affect.26 Throughout the later stagesof infection, both the proteolytically activated form ofPA (nPA) and LF are found in the sera of infectedanimal models.28 This form of PA, of course, may co-assemble with LF; however, the earlier study did nottest this possibility.We demonstrate here, using nativegel electrophoresis, that PA can be activated proteo-lytically in bovine blood plasma, and stable toxincomplexes can form (Supplementary Data Fig. S4).Under aqueous conditions, we have shown thatoctamers form in the presence of LFN or EFN (Fig. 2dand e). Thus, octamer formation is not limited to areceptor-dependent mechanism, and octamers mayassemble extracellularly in blood plasma in an LF- orin an EF-dependent manner (Fig. 6).

Toxin stability

Recent work in vitro revealed that heptameric pre-channel complexes are unstable under physiologicalconditions, readily converting to the channel state.8

Here, we demonstrate that physiological tempera-tures and pH lead to irreversible aggregation of theheptameric form, such that the octameric form is leftbehind in stable, isolable complexes (Fig. 5a and b).We think this effect may be linked to pH-dependentdifferences in the octameric pre-channel to channeltransition or inherent differences in the stability ofoctameric toxin complexes. ATRs have been shownpreviously to stabilize the pre-channel conformationunder physiological conditions.7,33,36,37 Toxin com-plexes assembled in the bloodstream, however, donotbenefit from ATR stabilization. Thus octameric com-plexes may represent a more stable toxin configura-tion that may persist in the blood of infected animals,serving a primary role of maintaining cytotoxicityunder physiological conditions.


Octamer structure

Our crystal structure shows that the octamer is aring of eight PA monomers, consisting of a square-planar arrangement of four PA dimers (Fig. 4c and d).This structure is in contrast to the near-perfect, regularheptagon model of the heptamer.7 Our EM andcrystallographic studies of these toxin complexessuggest that the octamer is composed of four pairsof PA subunits that are conformational heterodimers;and the heptamer may actually consist of three PAheterodimers and one asymmetric monomer (Fig. 2).Our analysis of the precise symmetry of the heptamerby EM is limited by the resolution of the technique,and further crystallographic studies are required.Nonetheless, the octamer and its tetrameric arrange-ment of PA pairs suggests that dimeric PA inter-mediates produced by interactions with LF, EF or adimeric ATR subsequently assemble in a pairwisefashion to form the octamer (Figs. 2 and 6).On the basis of the crystal structure, we conclude

that the WT octamers should bury 6100 Å2 ofadditional surface area per oligomer relative to theheptamer, indicating that octamers may possessgreater inherent stability, preventing disassemblyand/or premature conversion to the channel state.Each PA heterodimer, of course, is poised to bind anLF/EF molecule (or a pair of ATRs), affording theoctamer four binding sites and the heptamer three(Figs. 3 and 6). The improved stability of octamerictoxin complexes may result from full occupancy offour heterodimeric, hydrophobic, LF-binding sites.For the heptamer, however, only three such sites canbe occupied, leaving one PA subunit exposed. More-over, additional interfaces between adjacent LFsubunits in the octameric complex may create anovel ring of interfaces between adjacent LF subunits,adding stability that cannot be attained in theheptameric complex (Fig. 6). Future structural studieswill elucidate how these two lethal toxin configura-tions may differ. We conclude that the added surfaceburial and structural symmetry of the even-numberedoctameric configuration of toxin complexes mayexplain the improved stability over the heptamericconfiguration.

Toxin activity

We propose that the general paradigm for theaggregate physiological anthrax toxin activity is aproduct of the catalytic rate of translocation and theinherent stability of the two possible oligomericconfigurations: the heptamer and the octamer (Fig.6). A secondary effect relates to the fact that theoctameric configuration provides an additional LF-and EF-binding site per complex (Figs. 3a and 6),increasing the potential toxicity of saturated octa-meric toxin complexes. We find in our planar lipidbilayer translocation assays that the two oligomerstranslocate LF and EF at similar rates (Fig. 5c).Therefore, we propose differences in toxin activitywill likely result from intrinsic differences in oligomerstability. Without ATR stabilization, heptameric toxin

complexes may be inactivated under physiologicalconditions (Fig. 5a and b), whereas octameric toxincomplexes may remain soluble and fully functional.We hypothesize that these physiological conditionsare present when the toxin assembles in an ATR-independent manner (e.g., in the blood, lymph, orphagolysosomal compartment25), and these condi-tions may favor the octameric form.

Staphylococcal α-hemolysin

The pathogenic factor produced by Staphylococcusaureus, called α-hemolysin, is comprised of multiplecopies of a 293 residue polypeptide.38 The assembledtoxin ultimately forms a circular, β-barrel-type, ion-conducting porewith amushroom-like architecture.12

Interestingly, this pore-forming toxin is believed tohave multiple oligomeric states. EM,39-41 atomic forcemicroscopy42 studies, and electrophysiology studies43

determined that the α-hemolysin can be hexameric.However, chemical cross-linking studies,44 crystal-lographic studies,12,44 single-molecule, photo-bleach-ing fluorescence methods,45 and electrophysiologystudies43 suggest the toxin forms a heptamer. How-ever, the relative populations of the two oligomericstates and the conditions affecting these relativedistributions have not been reported.

Pathogenesis

Whymay PA assemble into amixture of oligomers?B. anthracis optimizes its lifecycle and proliferate in itshost by the secretion of a toxin, which at the initialstages of pathogenesis, may first help germinatedbacteria escape from macrophages,25 and secondallow a small population of bacteria to selectivelysuppress the immune system. However, these toxinconditions are nonlethal. During infection,B. anthracissecretes low, nearly undetectable concentrations ofthe toxin components; at the latest stages of infection,100 μg/mL PA is detectable in the blood of infectedanimals.28 This large increase in the concentration ofPA immediately precedes death in animal models.27

Such a wide dynamic range in toxin levels suggests arelationship between toxin assembly and cytotoxicity.We propose that assembly of octameric oligomers onthe cell surface alleviates a potential assembly bottle-neck imposed by ATR dimerization, while in thebloodstream, lymph, or phagolysosome25 the octa-meric toxin complexes may function as a more stablespecies. Thus, heterogeneous assembly may functionin two key contexts: the known cell-surface assemblypathway (SupplementaryData Fig. S1) and a putativeATR-independent assembly pathway (Fig. 6; Supple-mentary Data Fig. S4).PA assembly pathways are further complicated by

dimerization of ATRs,23,24 and induced dimerizationof PA subunits by LFN and EFN. The heptamericassembly route may be efficient at low concentrations,where PA likely assembles predominantly frommonomers. On the other hand, PA assembly wouldbe attenuated at high concentrations, where PAassembles via dimeric intermediates, in the absence


of the octameric assembly route. Endocytosis coincideswith PA assembly,22 and a loss in toxin activity couldoccur if partially assembled complexes were endocy-tosed. Therefore, we reason that the observed hetero-geneity (Fig. 2g) may serve to alleviate these potentialassembly bottlenecks incurredduring oligomerization.At later stages in anthrax infection, when the

concentrations of PA and LF are high, a significantproportion of toxin complexes may assemble beforeencountering cell surface ATRs. In fact, the blood ofinfected animals contains PA that is almost exclu-sively proteolytically activated.28 Here, we demon-strate that PA, LF, and EF assemble into lethal andedema toxin complexes in bovine blood (Supplemen-tary Data Fig. S4). Assembly in the bloodstreamshould produce a proportion of octameric toxincomplexes similar to that we observe in vitro (Fig. 2).However, we expect that octameric complexes couldpersist in these conditions, while heptameric com-plexes may form inactive aggregates. Therefore,octamer formation could provide a mechanism ofovercoming these harsh, attenuating conditionsencountered extracellularly (in the absence of anATR). The proposed two-oligomer assembly modelsuggests that the anthrax toxin can modulate itsaggregate activity through assembly due to thesedifferences in toxin stability. Further structure/func-tion studies of heptameric and octameric oligomerswill clarify this molecular mechanism.

Materials and Methods

Proteins

PA

Recombinant WT PA, carboxy-terminally His6-taggedPA,46 and all other PA mutants described here were over-expressed in the periplasmofEscherichia coliBL21(DE3), andthey were purified as 83 kDa monomers as described.35 Amodified QuikChange procedure47 using Pfu Turbo poly-merase (Agilent Technologies, Santa Clara, CA) wasimplemented to make a deletion construct (PAΔMIL) fromthe PA expression vector, PA83 pET22b+ (EMD Chemicals,Gibbstown,NJ).13 In PAΔMIL, residues 305–324were deletedand two point mutations (V303P and H304G) wereintroduced simultaneously, leaving a type II turn in placeof the MIL (the residue numbering is that used in 1ACC.)5

Dimeric PA

The QuikChange procedure was used to engineer anS170C point mutation into WT PA. PA S170C, purifiedunder oxidizing conditions, and judged to be dimeric bySDS-PAGE.

Soluble dimeric ATR2

Recombinant soluble anthrax toxin receptor domain(sATR2), residues 40–217,35 was expressed from a pGEXvector (GE Healthcare) as a glutathione-S-transferase (GST)fusion protein, affinity-purified on glutathione Sepharose asdescribed.35 The GST-sATR was shown to be fully dimericby nanoESI-MS and is called dsATR2.

Soluble monomeric ATR2

A monomeric version of sATR2 (msATR) was made bysubcloning residues 40–217 of ATR2 into pET15b (EMDChemicals) via the NdeI and BamHI restriction sites, using apGEX expression clone.35 This construct has an amino-terminal His6 tag as described.34 The protein was similarlypurified on a His6 affinity column and judged to be pure bySDS-PAGE, and it was confirmed to be monomeric bynanoESI-MS.

LFN and EFN

Recombinant LFN (LF residues 1–263) and EFN (residues1–254 of EF)were over-expressed frompET15b constructs,48

and then purified from the cytosol using His6 affinitychromatography. LFN was further processed by incubatingwith bovine α-thrombin to remove the amino-terminal His6tag and purified over Q-Sepharose anion-exchange resin asdescribed.19 EFN was used with its His6 tag and notprocessed any further. Each protein sample was verifiedby matrix-assisted laser desorption/ionization mass spec-trometry (MALDI-MS).

PA oligomerization on Q-Sepharose

PA monomers were first treated with trypsin at a 1:1000mass ratio for 15min at room temperature,making nPA. Thetrypsin was blocked by the addition of a 1:100 mass ratio ofsoybean trypsin inhibitor and 1 mM phenylmethylsulpho-nyl fluoride (PMSF). Then the small, 20 kDa fragment,releasedby trypsinizationwas separated byanion-exchangechromatography, using Q-Sepharose High Performanceresin (GE Healthcare), such that the remaining ∼60 kDaportion oligomerized as described.10,35 This crude, oligo-meric PA mixture (QPA) was used throughout and containsmixtures of both heptameric and octameric complexes asjudged by EM and nanoESI-MS.

nPA oligomerization in the presence of LFN/EFN orms/dsATR2

nPA was prepared as described above and then dilutedinto an appropriate buffer containing stoichiometricamounts of LFN/EFN or ms/dsATR2. LFN/EFN assemblyexperiments were done by diluting nPA to ∼1 mg/mL in20 mM phosphate, 150 mMNaCl, pH 7.5, containing excessLFN/EFN. The assembly reactionwas allowed to equilibratefor 2 h at room temperature to afford complete oligomeriza-tion, as assessed by native gel. The dsATR2- and msATR2-assembly experiments were carried out by diluting nPA to∼2 mg/mL in 20 mM cacodylate buffer, 150 mM NaCl,1 mM CaCl2, pH 7.5, containing the determined stoichio-metric amount of msATR2 or dsATR2. The assemblyreaction was allowed to equilibrate for 10 min at roomtemperature to afford complete oligomerization (as assessedby native gel electrophoresis). Assembly reactions werepurified on an S200 gel-filtration column equilibrated inassembly buffer to remove unassembled components.msATR2 does not promote oligomerization of nPA, so itwas assembled as described above for QPA.

Isolation of WT octameric PA

The nPA+LFN complexes were prepared as describedabove, dialyzed against 10 mM Tris, pH 8, and diluted to2 mg/mL in 0.1 M cacodylate buffer, pH 7. The sample was


incubated for 5 min at 37 °C, concentrated, and purified onan S400 gel-filtration column. The purity and oligomerichomogeneity of the resulting complexes were assessed byEM and MS.

Electrophysiology

An Axopatch 200B amplifier (Molecular Devices Corp.,Sunnyvale, CA) was used in the voltage clamp, capacitorfeedback mode. The amplifier was interfaced to a Cyber-Amp 320 signal conditioner (Molecular Devices), whichtypically filtered the data at 200 Hz via a low-pass, 4-pole,Bessel section. The filtered, analog signal was typicallyrecorded by computer at 400 Hz using a Digidata 1440Aanalog-to-digital converter (Molecular Devices) and AXO-CLAMP software (Molecular Devices). Most data analysis,post-acquisition filtering, and curve fitting used a combina-tion of CLAMPFIT (Molecular Devices), ORIGIN6.1 (Origi-nLab Corp., Northampton, MA), and custom Perl scripts.Relative macroscopic membrane insertion activities for

WT PA and mutant forms were assayed as follows. Planarlipidbilayerswerepainted49 using a 3% (w/v) solutionof thelipid, 1,2-diphytanoyl-sn-glycerol-3-phosphocholine(DPhPC; Avanti Polar Lipids, Alabaster, AL), in n-decaneas solvent. The bilayer was painted inside either a 100 μm ora 200μmapertureof a 1mL,white delrin cupwhile bathed inaqueous buffer S (100mMKCl, 1mMEDTA, 10mMsuccinicacid, pH 6.6). The cis (the side to which the PA oligomer isadded) and the trans compartments were bathed insymmetric buffer S. For macroscopic current measurements,PA oligomer (25 pM) was added to the cis compartment,which was held at a Δψ of +20 mV. (Δψ, the membranepotential, is definedasΔψ=Δψcis–ψtrans,whereψtrans≡0mV.)Channel insertion increased over a period of minutes andstabilized after 20–30 min. Macroscopic currents were thenmeasured at this point or the channel-inserted membranewas used in translocation experiments as described below.Single-channel measurements were obtained using

DPhPC/decane films formed on a 100 μm aperture in awhite delrin cup. Single-channel conductance measure-ments were carried out at a Δψ of +20 mV in symmetricbuffer S. Single-channel current recordingsweredeterminedby adding a dilute solution (∼10-14 M) of PA oligomer to thecis chamber, stirring briefly, and thenwaiting 5–30min untildiscrete steps in current were observed. The clampingvoltage and current responses were acquired at 400 Hzunder low-pass filtering at 200 Hz. To observe subtle, slow-timescale fluctuations in the single-channel current records,we implemented a Gaussian-filter algorithm; this filter doesnot cause ”overshoot” during channel opening and closingtransitions. The filter's Gaussian kernel was defined with σhaving a width of 100 data points. A Perl script was writtento apply the Gaussian filter to our datasets.To calculate the mean unitary conductance levels, time

courses not treated with the Gaussian filter were analyzedby fitting Gaussians to 0.003–0.03 pA binned histograms ofthe current responses, using either ORIGIN6.1 or CLAMP-FIT. Some records containedmultiple, but readily separable,current steps up to as many as 10 single channels. Themeans, μ, from Gaussian curve fits were then subtractedfrom similar fits to the noise observed at “zero” current toobtain each single-channel current i. The single-channelconductanceγ is calculated from the single-channel current iand voltage V by:

g = i=V

Errors from μ were propagated to establish errors in γ foreach measurement, which were ∼0.5 pS.

Single-channel recordings of the nPA+LFN sample

We also analyzed the nPA+LFN sample, which wasformed by taking 0.1 mg/ml of nPA and adding astoichiometric equivalent of LFN. The mixture was incu-bated at room temperature for 1 h. The assembled complex(as judged by native gel electrophoresis, gel-filtrationchromatography, and EM) was applied to planar lipidbilayers membranes at ∼10-14 M. We initially observedchannel insertion at a Δψ of –20 mV to preclude the LFNmoiety from binding within the channel and blocking theconductance. Note that because the PA sample was diluted109-fold, newly inserted channels could be shifted back to aΔψ of +20 mV to record their currents once the LFNdissociated from the channel.

Electrophysiology-based translocation assay

For translocation experiments, a universal bilayer buffer(UBB) was used (10 mM oxalic acid, 10 mM Mes, 10 mMphosphoric acid, 1 mM EDTA, 100 mMKCl). The pH of theUBB is either 5.6 or 6.6, depending onwhether aΔpHwill beformed during the translocation assay. Once a membranewas formed, QPA+LFN or nPA+LFN was added to the ciscompartment (at pH 5.6); the cis compartment was held at aΔψ of±20mVwith respect to the trans compartment. (i.e., inthe case of nPA+LFN, the Δψ was held at –20 mV to allowfor channel insertion to be observed, because at +20mV,LFNwould block the channel). When nPA+LFN was used, theexcess LFN was perfused away, and residual bound LFNwas then translocated at 80 mV to clear the channels beforeinitiating further translocation experiments. After theensemble channel population was established, LF, LFN, EF,or EFN were added to the cis compartment. The progress ofsubstrate binding to PA63 channels was monitored by thecontinuous fall in conductance. After the conductance blockof PA channelswas complete, excess substratewas removedfrom the cis compartment with perfusion with UBB, using apush–pull, hand cranked 10 mL syringe pump. The ciscompartmentwas perfusedwith 10ml of UBB at a flow rateof 2 ml/min and Δψ held at +20 mV. Translocation of LF,LFN, EF, and EFN were initiated by jumping the Δψ to ahigher positive voltage and/or jumping theΔpH to a higherpositive value where:

DpHupHtrans � pHcis

The ΔpH is induced by adding predetermined amounts of0.4 M phosphoric acid to the cis chamber. The final pH isdetermined using a pH meter following the completion ofthe translocation recording.

Electron microscopy

All the sampleswereprepared for EM in a similarmanner.The PA oligomer at 20–30 nM monomer was incubated inbuffer E (20 mM Tris, 100–250 mM NaCl, pH 8) for 5 min.The 400-mesh copper grids were covered successively by aholey carbon film and a continuous carbon film. A 4 μl ofsample was applied to a freshly glow-discharged supportgrid for 30 s and then stained with five successive drops(75 μl each) of either 1% (w/v) uranyl formate (StructureProbe, Inc., West Chester, PA) or 2% (w/v) uranyl acetate(Sigma-Aldrich, St. Louis, MO).Negatively stained EM images were recorded with a

Tecnai 12 (FEI Company, Hillsboro, OR) operated at 100 kVor at 120 kVat a magnification of either 49,000× or 50,000×.In some cases, data were collected on Kodak SO163 films


(EastmanKodak,Rochester,NY) at 600–800 nmunderfocus.Film images were digitized with a Nikon Super Coolscan8000 (Nikon USA, Melville, NY) at a 12.7 μm pixel size,resulting in 2.54 Å/pixel at the specimen scale. In othercases, data were collected directly on a CCD camera,resulting in 2.13 Å/pixel at the specimen scale. Particleimages were selected for each data set by automatic ormanual particle picking with boxer in EMAN.50

Reference-free processing was done with the softwarepackage IMAGIC (Image Science Software, Berlin, Ger-many) or SPIDER.51 Images were subjected to threesuccessive cycles of multi-reference alignment, multivariatestatistical analysis, and classification.52,53 The last classifica-tion was done using only the lowest order eigenvectors asdescribed,54 to separate the data by size and the heptamericand octameric oligomerization states.A secondmethod of image processingwas usedwhereby

reference images were made from 2D projections of low-resolution density maps generated from the crystal struc-tures of the PA heptameric,7 and octameric pre-channels,using SPIDER.51 Boxed images were then subjected toreference-based alignment and classification using thelowest order eigenvectors as stated above. Final class-average images were inspected manually for their oligomernumber, designated as either heptamer or octamer, andtabulated to produce the final percentages of heptamers andoctamers. Each method of classification, reference-free orcrystal structure-referenced, produced similar results (Sup-plementary Data Table S2).

Extraction of PA oligomers from CHO cells

C-CHO and T-CHO cells were a kind gift from ArthurFrankel. The cell line was created from a spontaneous ATR-deficient CHO cell mutant line (PR230-CHO). The C-CHOand T-CHO lines were derived from stable transfectionswith the human ATR2 and ATR1 expression clones,respectively.55 The cell lines were grown to confluence inHam's F12 medium (Invitrogen), 10% (v/v) fetal bovineserum (Invitrogen), 100 units/mL of penicillin, 100 μg/mLof streptomycin (Sigma-Aldrich) in a humid 5% CO2atmosphere at 37 °C, as described.32 Confluent cells weretreated for 1 hwith 50mMammonium chloride, 50mM4(2-hydroxyethyl)-1-piperazineethanesulfonic acid (Hepes), pH7.3 in Ham's medium to inhibit endosomal acidification,preventing conversion of PA to the channel state.56 His6-PAmonomers (WT PAwith a carboxy-terminal His6 tag) wereapplied to cells in the same medium at 100 μg/mL. Cellswere incubated with His6-PA for 1 h at 4 °C, washed withfive volumes of ice-cold PBS to remove unbound PA, andlysed in buffer L (20 mM Tris, 0.35 M NaCl, 10 mMimidazole, 1% (v/v)Nonidet P-40 (or IPEGAL), 0.25% (w/v)deoxycholic acid, 1 mM PMSF, pH 8) as described.55 Celldebris was removed by centrifugation and the supernatantwas incubated with 0.5 mL of His6-affinity resin (Ni-NTASuperflow, Qiagen, Valencia, CA) overnight at 4 °C withstirring. The resinwaswashedwith five volumes of buffer Land eluted in buffer L supplemented with 300 mMimidazole. The elution was applied to electron microscopygrids for analysis.

Mass spectrometry

Mass spectra of the protein complexes were acquiredusing a quadrupole time-of-flight (Q-TOF) mass spectro-meter equipped with a Z-spray ion source (Q-Tof Premier,Waters, Milford, MA). Ions were formed using a nanoelec-trospray (nanoESI) emitters prepared bypulling borosilicate

capillaries (1.0 mm O.D./0.78 mm I.D., Sutter Instruments,Novato CA) to a tip I.D. of∼1 μm with a Flaming/Brownmicropipette puller (model P-87, Sutter). The instrumentwas calibrated with CsI clusters formed by nanoESI using a24 mg/mL solution of CsI in 70:30 (v/v) Milli-Q water:2-propanol before mass measurement. The protein solutionfor the stoichiometry determinations was prepared asdescribed above and then concentrated to 10 μM followedby dialysis into 10 mM ammonium bicarbonate, pH 7.8.Immediately before mass analysis, the solution was diluted1:1 (v/v) with 200 mM ammonium acetate, pH 7.8. Aplatinum wire (0.127 mm diam., Sigma, St. Louis, MO) wasinserted through the capillary into the solution andelectrospray was initiated and maintained by applying 1–1.3 kV to the wire (relative to instrument ground). The rawdatawere smoothed three times using theWatersMassLynxsoftware mean smoothing algorithm with a window of50 m/z (mass/charge ratio).The reactions for assembly kinetics were initiated by

mixing 10 mM ammonium bicarbonate solutions ofpurified nPA and LFN monomer in a 1:1 (v/v) ratio toinitiate oligomerization, and mass spectra were acquiredcontinuously for 50 min. Variation in the voltage appliedto the nanospray capillary and new capillaries at ∼19 minand ∼30 min were required to maintain ion current. Massspectra were averaged for 5 min intervals and smoothedthree times using the Waters MassLynx software meansmoothing algorithm with a window of 50 m/z. Eachpeak in a given charge-state distribution was integratedand the peak areas summed to give an absoluteabundance for the corresponding analyte. Relative abun-dances were calculated as a fraction of the totalabundances of the four analytes of interest monitoredduring the experiment.

Protein crystallization

Purified QPAΔMIL oligomer (judged to be rich in octa-

meric complexes by EM) was prepared in buffer X, whichcontained 74 mM sodium acetate, 7 mM Tris, 0.62 MNaCl,37 mM tetrabutylammonium bromide, 7% (v/v) ethanol,0.07% (w/v) n-dodecyl-β-D-maltopyranoside, pH 5.7,centrifuged to remove precipitated protein, and concen-trated to 13 mg/mL. Initial crystallization conditions wereestablished by sparse-matrix crystallization screens,57

except our screens contained ∼1000 unique conditions. AMosquito nanoliter, liquid-handling robot (TTP Labtech,Cambridge, MA) was used to form 200 nL drops in 96-wellformat at 18 °C, using the hanging-drop, vapor-diffusionmethod.58 Diffraction-quality crystals were formed with1 μL hanging drops using the Mosquito (at 18 °C),containing a 1:1 (v/v) mixture of 13 mg/ml QPA

ΔMIL

oligomer in buffer X with the reservoir solutions (18–30%(v/v) t-butanol, 0.1 M Tris, pH 7.5–8.5). Often irregular,rectangular-prism-shaped crystals formed overnight andgrew as large as ∼200 μm×200 μm×75 μm. Crystals wereharvested in a 30% (v/v) polyethylene glycol (PEG) 400cryoprotectant, where the PEG only replaced the water inthe mother liquor, and immediately flash-frozen in liquidnitrogen.

X-raydiffractiondata collection, solution and refinement

X-ray diffraction data were collected at the AdvancedLight Source in Lawrence Berkeley National Lab, Beamline8.3.1,59 using a Quantum 315r CCD area detector (ADSC,Poway, CA). The crystals diffracted to 3.2 Å in the triclinic


space group P1 with unit cell dimensions of a=125.60 Å,b=125.67 Å, c=125.82 Å, α=106.64°, β=110.82°, andγ=110.98° (Supplementary Data Table S3). The diffractiondata were indexed and scaled in HKL2000.60 The scaleddataset was 98.7% complete to 3.2 Å. The self-rotationfunction in the CCP4 suite61 revealed strong peaks at a χangles of 45°, 90° and180°.Molecular replacementwasdonewith PHASER62 in CCP4, where the search model was aloop-stripped chain A from 1TZO.7 This molecular replace-ment solution placed eight PA monomer chains in theasymmetric unit. The molecular replacement solution wasrefined with rigid-body constraints defined by the knowndomain boundaries in PA, using PHENIX.63 Noncrystallo-graphic symmetry was established by first making struc-tural alignments of individual monomers in CHIMERA,64

and then calculating the center of mass of each chain tocompute the geometric arrangement of the chains about theoligomeric ring. We found the oligomer was an irregularoctagon, and pairs of monomers of two different conforma-tions, A and B formed the sides of a regular, square, planartetramer. Subsequent rounds of model building in COOT65

were followed by coordinate and B-factor refinement, with4-fold noncrystallographic symmetry restraints, usingPHENIX. 2Fo – Fc and Fo – Fc omit electron density mapswere recalculated after iterations of model building andrefinement, where Fo and Fc are the observed and calculatedstructure factors, respectively. MOLPROBITY66 andPROCHECK67 were used to validate the structure'sgeometry and stereochemistry during model building.Surface burial calculat ions were made usingGETAREA1.1.68 Molecular graphics renderings were com-puted using CHIMERA.64

Protein Data Bank accession code

Coordinates and structure factors for the PA octamerhave been deposited in the Protein Data Bank (PDB) withaccession code 3HVD.

Acknowledgements

K.L.T did the single-channel studies. A.F.K., I.I.T.,and T.T.Z. did the EM studies. G.K.F. preparednanoESI-MS complexes. H.J.S. did the nanoESI-MS.A.F.K. and K.C.D. crystallized the octamer; A.F.K.and B.A.K. refined and analyzed the octamer withguidance from J.M.B. A.F.K. did the ensembletranslocations. B.A.K, A.F.K., J.M.B, E.R.W., andH.J.S. wrote the manuscript. We thank J. Mogridge,K. Bradley, and R.J. Collier for discussions; E. Nogalesand P. Grob for EM studies and advice; J. Liphardt foradvice on single-molecule data; A. Frankel for the giftof theCHOcell lines; J. Fraser and S. Stephensen for X-ray diffraction screening, data collection and structurerefinement; S. Greenberg and J. Yassif for purifyingPAΔMIL; R. Zalpuri at the Robert D. Ogg ElectronMicroscope Laboratory; and A.T. Iavarone at theQB3/Chemistry Mass Spectrometry Facility for help-ful discussions. This work was supported by Uni-versity of California start-up funds and NIH researchgrants R01-AI077703 (to B.A.K.) and R01-GM064712(to E.R.W.).

Supplementary Data

Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.jmb.2009.07.037

References

1. Young, J. A. & Collier, R. J. (2007). Anthrax toxin:receptor binding, internalization, pore formation, andtranslocation. Annu. Rev. Biochem. 76, 243–265.

2. Bradley, K. A., Mogridge, J., Mourez, M., Collier, R. J.& Young, J. A. (2001). Identification of the cellularreceptor for anthrax toxin. Nature, 414, 225–229.

3. Scobie, H. M., Rainey, G. J. A., Bradley, K. A. & Young,J. A. (2003). Human capillary morphogenesis protein 2functions as an anthrax toxin receptor. Proc. Natl Acad.Sci. USA, 100, 5170–5174.

4. Milne, J. C., Furlong, D., Hanna, P. C., Wall, J. S. &Collier, R. J. (1994). Anthrax protective antigenforms oligomers during intoxication of mammaliancells. J. Biol. Chem. 269, 20607–20612.

5. Petosa, C., Collier, R. J., Klimpel, K. R., Leppla, S. H. &Liddington, R. C. (1997). Crystal structure of theanthrax toxin protective antigen. Nature, 385, 833–838.

6. Mogridge, J., Cunningham, K. & Collier, R. J. (2002).Stoichiometry of anthrax toxin complexes. Biochemistry,41, 1079–1082.

7. Lacy, D. B., Wigelsworth, D. J., Melnyk, R. A.,Harrison, S. C. & Collier, R. J. (2004). Structure ofheptameric protective antigen bound to an anthraxtoxin receptor: a role for receptor in pH-dependentpore formation. Proc. Natl Acad. Sci. USA, 101,13147–13151.

8. Katayama, H., Janowiak, B. E., Brzozowski, M.,Juryck, J., Falke, S., Gogol, E. P., Fisher, M. T. et al.(2008). GroEL as a molecular scaffold for structuralanalysis of the anthrax toxin pore. Nature Struct. Mol.Biol. 15, 754–760.

9. Friedlander, A. M. (1986). Macrophages are sensitiveto anthrax lethal toxin through an acid-dependentprocess. J. Biol. Chem. 261, 7123–7126.

10. Blaustein, R. O., Koehler, T. M., Collier, R. J. &Finkelstein, A. (1989). Anthrax toxin: channel-formingactivity of protective antigen in planar phospholipidbilayers. Proc. Natl Acad. Sci. USA, 86, 2209–2213.

11. Krantz, B. A., Finkelstein, A. & Collier, R. J. (2006).Protein translocation through the anthrax toxintransmembrane pore is driven by a proton gradient.J. Mol. Biol. 355, 968–979.

12. Song, L., Hobaugh, M. R., Shustak, C., Cheley, S.,Bayley, H. & Gouaux, J. E. (1996). Structure ofstaphylococcal α-hemolysin, a heptameric transmem-brane pore. Science, 274, 1859–1866.

13. Benson, E. L., Huynh, P. D., Finkelstein, A. & Collier,R. J. (1998). Identification of residues lining theanthrax protective antigen channel. Biochemistry, 37,3941–3948.

14. Nassi, S., Collier, R. J. & Finkelstein, A. (2002). PA63

channel of anthrax toxin: an extended β-barrel.Biochemistry, 41, 1445–1450.

15. Blaustein, R. O. & Finkelstein, A. (1990). Diffusionlimitation in the block by symmetric tetraalkylammo-nium ions of anthrax toxin channels in planarphospholipid bilayer membranes. J. Gen. Physiol. 96,943–957.

16. Wesche, J., Elliott, J. L., Falnes, P. O., Olsnes, S. & Collier,R. J. (1998). Characterization ofmembrane translocation




by anthrax protective antigen. Biochemistry, 37,15737–15746.

17. Zhang, S., Udho, E., Wu, Z., Collier, R. J. & Finkelstein,A. (2004). Protein translocation through anthrax toxinchannels formed in planar lipid bilayers. Biophys. J. 87,3842–3849.

18. Krantz, B. A., Trivedi, A. D., Cunningham, K.,Christensen, K. A. & Collier, R. J. (2004). Acid-inducedunfolding of the amino-terminal domains of the lethaland edema factors of anthrax toxin. J. Mol. Biol. 344,739–756.

19. Krantz, B. A., Melnyk, R. A., Zhang, S., Juris, S. J.,Lacy, D. B., Wu, Z. et al. (2005). A phenylalanine clampcatalyzes protein translocation through the anthraxtoxin pore. Science, 309, 777–781.

20. Mogridge, J., Mourez, M. & Collier, R. J. (2001).Involvement of domain 3 in oligomerization by theprotective antigen moiety of anthrax toxin. J. Bacteriol.183, 2111–2116.

21. Cunningham, K., Lacy, D. B., Mogridge, J. & Collier,R. J. (2002). Mapping the lethal factor and edemafactor binding sites on oligomeric anthrax protectiveantigen. Proc. Natl Acad. Sci. USA, 99, 7049–7053.

22. Abrami, L., Liu, S., Cosson, P., Leppla, S. H. & van derGoot, F. G. (2003). Anthrax toxin triggers endocytosisof its receptor via a lipid raft-mediated clathrin-dependent process. J. Cell Biol. 160, 321–328.

23. Go, M. Y., Chow, E. M. & Mogridge, J. (2009). Thecytoplasmic domain of anthrax toxin receptor 1 affectsbinding of the protective antigen. Infect. Immun. 77,52–59.

24. Go, M. Y., Kim, S., Partridge, A. W., Melnyk, R. A.,Rath, A., Deber, C. M. & Mogridge, J. (2006). Self-association of the transmembrane domain of ananthrax toxin receptor. J. Mol. Biol. 360, 145–156.

25. Banks, D. J., Barnajian, M., Maldonado-Arocho, F. J.,Sanchez, A. M. & Bradley, K. A. (2005). Anthrax toxinreceptor 2 mediates Bacillus anthracis killing of macro-phages following spore challenge. Cell Microbiol. 7,1173–1185.

26. Smith, H., Keppie, J. & Stanley, J. L. (1954). Observa-tions on the cause of death in experimental anthrax.Lancet, 267, 474–476.

27. Ezzell, J. W., Abshire, T. G., Panchal, R., Chabot, D.,Bavari, S., Leffel, E. K., Purcell, B., Friedlander, A. M.& Ribot, W. J. (2009). Association of Bacillus anthraciscapsule with lethal toxin during experimental infec-tion. Infect. Immun. 77, 749–755.

28. Mabry, R., Brasky, K., Geiger, R., Carrion, R., Jr,Hubbard, G. B., Leppla, S. et al. (2006). Detection ofanthrax toxin in the serum of animals infected withBacillus anthracis by using engineered immunoassays.Clin. Vaccine Immunol. 13, 671–677.

29. Fish, D. C. & Lincoln, R. E. (1968). In vivo-producedanthrax toxin. J. Bacteriol. 95, 919–924.

30. Blaustein, R. O., Lea, E. J. & Finkelstein, A. (1990).Voltage-dependent block of anthrax toxin channels inplanar phospholipid bilayer membranes by sym-metric tetraalkylammonium ions. Single-channel ana-lysis. J. Gen. Physiol. 96, 921–942.

31. Hille, B. (1968). Pharmacological modifications of thesodium channels of frog nerve. J. Gen. Physiol. 51,199–219.

32. Young, J. J., Bromberg-White, J. L., Zylstra, C., Church,J. T., Boguslawski, E. & Resau, J. H. (2007). LRP5 andLRP6 are not required for protective antigen-mediatedinternalization or lethality of anthrax lethal toxin.PLoS Pathogen, 3, e27.

33. Miller, C. J., Elliott, J. L. & Collier, R. J. (1999). Anthrax

protective antigen: prepore-to-pore conversion.Biochemistry, 38, 10432–10441.

34. Santelli, E., Bankston, L. A., Leppla, S. H. &Liddington, R. C. (2004). Crystal structure of acomplex between anthrax toxin and its host cellreceptor. Nature, 430, 905–908.

35. Wigelsworth, D. J., Krantz, B. A., Christensen, K. A.,Lacy, D. B., Juris, S. J. & Collier, R. J. (2004). Bindingstoichiometry and kinetics of the interaction of ahuman anthrax toxin receptor, CMG2, with protectiveantigen. J. Biol. Chem. 279, 23349–23356.

36. Novak, J. M., Stein, M. P., Little, S. F., Leppla, S. H. &Friedlander, A. M. (1992). Functional characterizationof protease-treated Bacillus anthracis protective anti-gen. J. Biol. Chem. 267, 17186–17193.

37. Leppla, S. H. (1991). The anthrax toxin complex. In(Alouf, J. E. & Freer, J. H., eds), pp. 277–302, AcademicPress, London.

38. Bhakdi, S., Bayley, H., Valeva, A., Walev, I., Walker, B.,Kehoe, M. & Palmer, M. (1996). Staphylococcal alpha-toxin, streptolysin-O, and Escherichia coli hemolysin:prototypes of pore-forming bacterial cytolysins. Arch.Microbio. 165, 73–79.

39. Arbuthnott, J. P., Freer, J. H. & Bernheimer, A. W.(1967). Physical states of staphylococcal alpha-toxin.J. Bacteriol. 94, 1170–1177.

40. Ward, R. J. & Leonard, K. (1992). The Staphylococcusaureus alpha-toxin channel complex and the effect ofCa2+ ions on its interaction with lipid layers. J. Struct.Biol. 109, 129–141.

41. Olofsson, A., Kaveus, U., Thelestam, M. & Hebert, H.(1988). The projection structure of alpha-toxin fromStaphylococcus aureus in human platelet membranesas analyzed by electron microscopy and imageprocessing. J. Ultrastruct. Mol. Struct. Res. 100,194–200.

42. Czajkowsky, D. M., Sheng, S. & Shao, Z. (1998).Staphylococcal alpha-hemolysin can form hexamersin phospholipid bilayers. J. Mol. Biol. 276, 325–330.

43. Furini, S., Domene, C., Rossi, M., Tartagni, M. &Cavalcanti, S. (2008). Model-based prediction of thealpha-hemolysin structure in the hexameric state.Biophys. J. 95, 2265–2274.

44. Gouaux, J. E., Braha, O., Hobaugh, M. R., Song, L.,Cheley, S., Shustak, C. & Bayley, H. (1994). Subunitstoichiometry of staphylococcal alpha-hemolysin incrystals and on membranes: a heptameric transmem-brane pore. Proc. Natl Acad. Sci. USA, 91,12828–12831.

45. Das, S. K., Darshi, M., Cheley, S., Wallace, M. I. &Bayley, H. (2007). Membrane protein stoichiometrydetermined from the step-wise photobleaching of dye-labelled subunits. Chembiochem, 8, 994–999.

46. Sun, J., Vernier, G., Wigelsworth, D. J. & Collier, R. J.(2007). Insertion of anthrax protective antigen intoliposomal membranes: effects of a receptor. J. Biol.Chem. 282, 1059–1065.

47. Zheng, L., Baumann, U. & Reymond, J. L. (2004). Anefficient one-step site-directed and site-saturationmutagenesis protocol. Nucleic Acids Res. 32, e115.

48. Lacy, D. B., Mourez, M., Fouassier, A. & Collier, R. J.(2002). Mapping the anthrax protective antigen bind-ing site on the lethal and edema factors. J. Biol. Chem.277, 3006–3010.

49. Mueller, P., Rudin, D. O., Tien, H. T. & Westcott, W. C.(1963). Methods for the formation of single bimole-cular lipid membranes in aqueous solution. J. Phys.Chem. 67, 534–535.

50. Ludtke, S. J., Baldwin, P. R. & Chiu, W. (1999). EMAN:


semiautomated software for high-resolution single-particle reconstructions. J. Struct. Biol. 128, 82–97.

51. Frank, J., Radermacher, M., Penczek, P., Zhu, J., Li, Y.,Ladjadj, M. & Leith, A. (1996). SPIDER and WEB:processing and visualization of images in 3D electronmicroscopy and related fields. J. Struct. Biol. 116,190–199.

52. Stark, H., Mueller, F., Orlova, E. V., Schatz, M., Dube,P., Erdemir, T., Zemlin, F., Brimacombe, R. & van Heel,M. (1995). The 70S Escherichia coli ribosome at 23 Aresolution: fitting the ribosomal RNA. Structure, 3,815–821.

53. van Heel, M., Harauz, G., Orlova, E. V., Schmidt, R. &Schatz, M. (1996). A new generation of the IMAGICimage processing system. J. Struct. Biol. 116, 17–24.

54. White, H. E., Saibil, H. R., Ignatiou, A. & Orlova, E. V.(2004). Recognition and separation of single particleswith size variation by statistical analysis of theirimages. J. Mol. Biol. 336, 453–460.

55. Liu, S. & Leppla, S. H. (2003). Cell surface tumorendothelium marker 8 cytoplasmic tail-independentanthrax toxin binding, proteolytic processing, oligo-mer formation, and internalization. J. Biol. Chem. 278,5227–5234.

56. Rainey, G. J., Wigelsworth, D. J., Ryan, P. L., Scobie,H. M., Collier, R. J. & Young, J. A. (2005). Receptor-specific requirements for anthrax toxin delivery intocells. Proc. Natl Acad. Sci. USA, 102, 13278–13283.

57. Jancarik, J. & Kim, S. H. (1991). Sparse matrixsampling: A screening method for crystallization ofproteins. J. Appl. Cryst. 24, 409–411.

58. McPherson, A., Jr. (1976). The growth and prelimin-ary investigation of protein and nucleic acid crystalsfor X-ray diffraction analysis. Methods Biochem. Anal.23, 249–345.

59. MacDowell, A. A., Celestre, R. S., Howells, M.,McKinney, W., Krupnick, J., Cambie, D., Domning,E. E., Duarte, R. M., Kelez, N., Plate, D. W., Cork,C. W., Earnest, T. N., Dickert, J., Meigs, G., Ralston,C., Holton, J. M., Alber, T., Berger, J. M., Agard, D. A.& Padmore, H. A. (2004). Suite of three proteincrystallography beamlines with single superconduct-ing bend magnet as the source. J. Synchrotron Rad. 11,447–455.

60. Otwinowski, Z. & Minor, W. (1997). Processing ofX-ray diffraction data collected in oscillation mode.In (Carter, C. W. & Sweet, R. M., eds), Methods in

Enzymology, Vol. 276, pp. 307–326Academic Press,Inc., New York; Macromolecular Crystallography,part A.

61. CollaborativeComputational Project,Number 4. (1994).The CCP4 suite: programs for protein crystallography.Acta Crystallogr. D Biol. Crystallogr. 50, 760–763.

62. Storoni, L. C., McCoy, A. J. & Read, R. J. (2004).Likelihood-enhanced fast rotation functions. ActaCrystallogr. D Biol. Crystallogr. 60, 432–438.

63. Adams, P. D., Gopal, K., Grosse-Kunstleve, R. W.,Hung, L. W., Ioerger, T. R., McCoy, A. J., Moriarty,N. W., Pai, R. K., Read, R. J., Romo, T. D., Sacchettini,J. C., Sauter, N. K., Storoni, L. C. & Terwilliger, T. C.(2004). Recent developments in the PHENIX soft-ware for automated crystallographic structure deter-mination. J. Synchrotron Rad. 11, 53–55.

64. Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch,G. S., Greenblatt, D. M., Meng, E. C. & Ferrin, T. E.(2004). UCSF Chimera—a visualization system forexploratory research and analysis. J. Comput. Chem. 25,1605–1612.

65. Emsley, P. & Cowtan, K. (2004). Coot: model-buildingtools for molecular graphics. Acta Crystallogr. D Biol.Crystallogr. 60, 2126–2132.

66. Davis, I. W., Leaver-Fay, A., Chen, V. B., Block, J. N.,Kapral, G. J., Wang, X., Murray, L. W., Arendall, W. B.,3rd, Snoeyink, J., Richardson, J. S. & Richardson, D. C.(2007). MolProbity: all-atom contacts and structurevalidation for proteins and nucleic acids.Nucleic AcidsRes. 35, W375–W383.

67. Laskowski, R. A., MacArthur, M. W., Moss, D. S. &Thornton, J. M. (1993). PROCHECK: a program tocheck the stereochemical quality of protein structures.J. Appl. Cryst. 26, 283–291.

68. Fraczkiewicz, R. & Braun, W. (1998). Exact andefficient analytical calculation of the accessible surfaceareas and their gradients for macromolecules. J. Comp.Chem. 19, 319–333.

69. McKay, A.R., Ruotolo, B. T., Ilag, L. L.&Robinson, C. V.(2006). Mass measurements of increased accuracyresolve heterogeneous populations of intact ribosomes.J. Am. Chem. Soc. 128, 11433–11442.

70. Lacy, D. B., Lin, H. C., Melnyk, R. A., Schueler-Furman, O., Reither, L., Cunningham, K., Baker, D. &Collier, R. J. (2005). A model of anthrax toxin lethalfactor bound to protective antigen. Proc. Natl Acad. Sci.USA, 102, 16409–16414.

The Protective Antigen Component of Anthrax Toxin Forms...

Documents

Transcript of The Protective Antigen Component of Anthrax Toxin Forms...