Surface architecture of endospores of the Bacilluscereus/anthracis/thuringiensis family at thesubnanometer scaleLekshmi Kailasa, Cassandra Terrya,1, Nicholas Abbotta, Robert Taylora, Nic Mullinb, Svetomir B. Tzokova, Sarah J. Todda,B. A. Wallacec, Jamie K. Hobbsb, Anne Moira, and Per A. Bullougha,2

aKrebs Institute for Biomolecular Research, Department of Molecular Biology and Biotechnology, University of Sheffield, Sheffield S10 2TN, United Kingdom;bDepartment of Physics and Astronomy, University of Sheffield, Sheffield S3 7RH, United Kingdom; and cDepartment of Crystallography, Institute ofStructural and Molecular Biology, Birkbeck College, University of London, London WC1E 7HX, United Kingdom

Edited by Richard M. Losick, Harvard University, Cambridge, MA, and approved August 5, 2011 (received for review June 14, 2011)

Bacteria of the Bacillus cereus family form highly resistant spores,which in the case of the pathogen B. anthracis act as the agents ofinfection. The outermost layer, the exosporium, enveloping sporesof theB. cereus family aswell as a number ofClostridia, plays roles inspore adhesion, dissemination, targeting, and germination control.We have analyzed two naturally crystalline layers associated withthe exosporium, one representing the “basal” layer to which theoutermost spore layer (“hairy nap”) is attached, and the other likelyrepresenting a subsurface (“parasporal”) layer. We have used elec-tron cryomicroscopy at a resolution of 0.8–0.6 nm and circular di-chroism spectroscopic measurements to reveal a highly α-helicalstructure for both layers. The helices are assembled into 2D arraysof “cups” or “crowns.” High-resolution atomic force microscopy ofthe outermost layer showed that the open ends of these cups facethe external environment and the highly immunogenic collagen-likefibrils of the hairy nap (BclA) are attached to this surface. Based onourfindings,wepresent amolecularmodel for the spore surface andpropose how this surface can act as a semipermeable barrier anda matrix for binding of molecules involved in defense, germinationcontrol, and other interactions of the spore with the environment.

electron crystallography | cryoelectron microscopy | two-dimensional crystal

Bacterial endospores are uniquely resistant survival structuresthat result from a complex differentiation process. The

dormancy, resistance (1), and widespread dispersal of bacterialendospores mean that they are ubiquitous in the environment.Spores of a few species can be very effective infectious agents,remaining dormant in the environment until the opportunity forinfection arises, in addition to being resistant to host defensesafter infection. Formation of spores involves a tightly controlledsequence of steps of gene expression and cellular morphologicalchange. Although these steps have been well-studied (2), thestructure of the fully assembled spore remains poorly understoodat the molecular scale. Coupled with previous genetic andmorphological studies, our recent work on electron crystallog-raphy of spore proteins suggests that spores may present an at-tractive model system for exploring the molecular mechanismsdriving the formation of complex supramolecular structures (3).Spores contain a relatively dehydrated cellular core that is

membrane-bound and surrounded by a peptidoglycan cortex andan outer protective coat containing multiple protein layers (4–6).Some species, including, but by no means limited to, pathogenssuch as Bacillus cereus and Clostridium botulinum (food-bornepathogens), B. thuringiensis (an insect pathogen), and B. anthracis(animal and human pathogen), possess a further outermost layercalled the “exosporium” (7), a loosely fitting shell surrounding thecoat. The volume between the spore coat and exosporium, the“interspace,” contains proteins but has not been well-studied. InB. cereus and the closely related B. anthracis and B. thuringiensis,the exosporium is made up of a “basal” layer supporting a fila-mentous “hairy nap” (8) and is composed of proteins, lipid, and

carbohydrate (9). Proteins of the exosporium are likely to haveintimate interactions with the environment and are also potentialcandidates for vaccines and ligands for spore detection (10, 11).Two exosporium glycoproteins, BclA and BclB, have been

identified in B. anthracis (8, 12, 13). These contain a central tri-meric collagen-like domain, a processed N-terminal domain an-choring the protein to the basal layer, and a C-terminal headdomain whose structure has been elucidated to atomic resolutionin the case of BclA (14). BclA filaments are themajor componentsof the hairy nap (8, 13). ExsFA/BxpB and its homolog ExsFB arerequired for attachment of BclA to the basal layer (15, 16). BclA,ExsFA, and ExsY form high-molecular weight complexes, ExsYbeing required for the complete assembly of the exosporium (17–19). The exosporium is likely to contain a number of other struc-tural proteins (many reviewed in ref. 4) as well as loosely boundproteins such as immune inhibitor A (20) and arginase (21).The exosporium is likely to play multiple roles in the interaction

of the spore with its environment. For example, in B. anthracis,BclA–integrin interaction promotes spore uptake by macrophagesin vitro (22). The exosporium contributes protection againstmacrophage-mediated damage (21), whereas enzymes in theexosporium (23) moderate responses to germinants. However,although the exosporium may have a role in natural anthraxinfections, a BclA-containing exosporium is not required for an-thrax infection in animal models (24). The adherent propertiesexhibited by the exosporium (25) point to a role in attachmentto surfaces; this has important implications for development ofdecontamination protocols.A substantial part of the basal layer of the exosporium from

members of the B. cereus family takes the form of a natural 2Dcrystal (7). Ball et al. have isolated fragments of this layer (type IIcrystals) from B. cereus, B. anthracis, and B. thuringiensis, de-scribing the 3D structure to ∼30 Å resolution (3). The basicarchitecture of the crystalline basal layer, including unit celldimensions, is identical in all three species and the followingdescription applies to all of them. The layer forms a network of“crown”-like structures, which were cautiously proposed to betrimeric although the density showed strong indications of highersymmetry. The open ends of the crowns are linked together to

Author contributions: L.K., C.T., B.A.W., A.M., and P.A.B. designed research; L.K., C.T.,N.A., R.T., N.M., S.B.T., S.J.T., B.A.W., and P.A.B. performed research; L.K., C.T., N.A.,R.T., N.M., S.B.T., B.A.W., J.K.H., A.M., and P.A.B. analyzed data; and L.K., B.A.W.,A.M., and P.A.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1Present address: Medical Research Council Prion Unit and Department of Neurodegen-erative Disease, University College London Institute of Neurology, Queen Square, Lon-don WC1N 3BG, United Kingdom.

2To whom correspondence should be addressed. E-mail: p.bullough@sheffield.ac.uk.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1109419108/-/DCSupplemental.

16014e16019 | PNAS | September 20, 2011 | vol. 108 | no. 38 www.pnas.org/cgi/doi/10.1073/pnas.1109419108

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

form tunnels connecting one surface of the basal layer with theother (figure 5 in ref. 3). In some spores, there is evidence ofanother crystalline layer, a “parasporal” layer (type I crystals),located within the interspace of the spore. This layer does notappear to envelop the spore completely; it follows the same ar-chitectural theme as the basal layer, but the crowns appearsmaller and more closely packed (3). The parasporal layer mayrepresent a different conformation of the same complex found inthe basal layer, and/or it may be an intermediate assembly stateon the way to forming a mature exosporium basal layer.Our aim is to move on to the next stage in visualizing the

detailed molecular architecture of the exosporium, to mapidentified proteins onto the structure, and to understand howthese various proteins determine function and architecture. Afew key questions are: (i) Which surface revealed in the B. cereusfamily basal layer structure of Ball et al. (3) corresponds to thespore surface exposed to the environment? (ii) Where and how isthe hairy nap attached to this outer surface? (iii) Where are the“core” structural proteins located in the exosporium? These arelikely to be the proteins resistant to all washing regimes andmaking up the crystalline lattice. (iv) Where are the more easilyremoved “accessory” proteins such as enzymes located? (v) Whatis the relation between the crystalline basal and parasporal lay-ers? To address these questions, we need higher resolution andto exploit noncrystallographic techniques to visualize the lessregular components of the spore surface. In this study, we haveused electron cryomicroscopy (cryo-EM) to reveal the internalstructural details of the crystalline layers and circular dichroism(CD) spectroscopy to confirm the high helical content; atomicforce microscopy (AFM) has revealed the fine details of thesurfaces facing both the environment and the spore interior.

ResultsParasporal Layer Crystals: Projection Structure at 6 Å Resolution.Electron micrographs of exosporium fragments from B. thur-ingiensis kurstaki HD1 embedded in glucose were recorded andsix of them were processed. A representative Fourier transformis displayed in Fig. S1A. Phases were consistent with p3 symmetry(Table S1). After phase origin alignment and averaging, themean phase error was significantly less than that expected forrandom phases (90°) to a resolution of 6 Å, and we have trun-cated all subsequent calculations to this resolution (Table S2).Individual images were corrected for fall-off in the envelope ofthe contrast transfer function (26); temperature factors for in-dividual films ranged from 158 to 521 Å2. The projection mapcalculated from the averaged Fourier terms is shown in Fig. 1A.The unit cell has dimensions of a = b = ∼67 Å. By comparingthis map with the 3D map in negative stain (3) and also with

maps calculated from crystals embedded in a glucose/uranylformate mixture, we have been able to superimpose the 6-Å maponto a projection of the negatively stained protein assembly,whose approximate envelope is indicated by the filled gray circles(Fig. 1A). This allowed us to identify which threefold symmetryaxis corresponds to the central symmetry axis of the trimericcrown. The envelope of the trimeric crown encloses a number ofvery dense circular features, which have the appearance of pro-jected α-helices; we have circled the six densest features in anarbitrary cluster in magenta with their trimeric partners in blue.

Basal Layer Crystals: Projection Structure at 8 Å Resolution. Electronmicrographs of exosporium fragments from B. cereus 10876 em-bedded in ice were recorded and 16 of them were processed. Arepresentative Fourier transform computed after unbending isshown in Fig. S1B. Fourier phases were consistent with p6 planegroup symmetry (Table S3). Table S4 shows that the mean phaseerror is significantly less than that expected for random phases(45°) to a resolution of 8 Å, and we have truncated all subsequentcalculations to this resolution. Temperature factors applied toindividual films ranged from 546 to 1,000 Å2. The calculatedprojectionmap is shown in Fig. 1B. The unit cell has dimensions ofa= b= ∼80 Å. There is only one unique sixfold symmetry axis, sowe can match this to the previously determined structure in neg-ative stain (3), which has its sixfold axis centered on the crown(although only p3 symmetry had been imposed in this previousstudy). The approximate maximum extent of the projected densityof a negatively stained crown is indicated by filled gray circles inFig. 1B. The six highest unique density features have been circledin magenta with their symmetry partners in blue; some or all ofthese could correspond to projections of α-helices.

CD Spectroscopy of Exosporium. The CD spectrum recorded froman exosporium preparation of B. cereus (Fig. 2) had the appear-ance of a protein with a high α-helical content, with a positive peakat 193 nm and approximately equal magnitude negative peaks at210 and 224 nm. Secondary structural analyses indicated a helicalcontent of 75% with no β-sheet, 20% “other” (random coil), and∼5% collagen. The low normalized root-mean-square deviation(NRMSD) value (0.07) indicated a strong correspondence be-tween the data and the calculated secondary structure.

AFM of Basal Layer Crystals. The height and phase images froma folded fragment of B. cereus 10876 exosporium (Fig. 3 A and B)show the two surfaces. Fig. 3B shows patches of ordered crys-talline lattice on one side (indicated by the dashed white arrows)and disordered surface (indicated by the solid white arrows) onthe other side of the folded exosporium. The crystalline surfaceappears to be made up of an array of “dome”-shaped structures.

Fig. 1. Projection maps for the two crystal types. (A) Projection map with p3 symmetry at 6 Å resolution for parasporal layer crystals. Contours are at intervals∼0.3× root-mean-square density. Contours below average density are not shown. The approximate projection of the envelope of an individual trimeras determined by Ball et al. (3) is indicated by the filled gray circle. The six densest features are circled in magenta with their symmetry mates in blue. a = b =∼67 Å. (B) Projection map with p6 symmetry at 8 Å resolution for basal layer crystals. Features are as described above. a = b = ∼80 Å.

Kailas et al. PNAS | September 20, 2011 | vol. 108 | no. 38 | 16015

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

Additional representative AFM images can be seen in Fig. S2.The ordered surface is more clearly illustrated in an ultra-high-resolution image of B. cereus exosporium obtained using tor-sional tapping that reveals a studded surface of hexagonallyclose-packed units of ∼8-nm diameter (Fig. 3 C and D). Theexosporium in B. thuringiensis 4D11 strain retains the same basallayer architecture as B. cereus, but has little or no nap; therefore,the outer surface of the basal layer can be visualized by AFMwithout obstruction from the hairy nap (Fig. 4). The heightimages from the B. thuringiensis 4D11 (Fig. 4C) exosporiumfragments show that the crystalline lattice on one side of the B.thuringiensis exosporium appears to be made up of hexagonallyclose-packed concave “cups” forming a honeycomb-like pattern.Fig. 4B shows that the B. thuringiensis exosporium has two dif-ferent crystalline surfaces on opposite sides of the basal layer

with no apparent disorder on either side. An additional imagewith a wider field of view showing the honeycomb pattern can beseen in Fig. S3.

DiscussionOne of the main advantages of the B. cereus family exosporiumfor structural studies is its in vivo crystalline nature, which allowsanalysis by electron crystallography. Moreover, the basic archi-tecture of the exosporium in B. cereus, B. anthracis, and B.thuringiensis is identical (3), meaning that we can directly com-pare different strains for variations in structural detail and drawgeneral conclusions for the whole B. cereus family. In this study,we report exceptionally high resolution (6–8 Å) electron crys-tallographic projection maps and have thus taken the first step inbuilding up a high-resolution 3D map of the exosporium. Bothparasporal and basal layer crystals show density features (circledin Fig. 1 A and B) that are entirely consistent with end-on pro-jection views of α-helices and appear remarkably similar to he-lices seen in projection maps calculated from integral membraneproteins (27); these features have diameters of ∼10 Å and arepacked with minimum center-to-center spacings on the order of10 Å, as expected for close-packed helical bundles. The CDspectrum of exosporium from B. cereus (Fig. 2) indicated a veryhigh α-helical content of ∼75%, which is qualitatively consistentwith the observations by EM. Notably, the CD spectrum alsoindicated ∼5% polyproline (collagen-like) helix content; this islikely to arise from BclA and possibly other proteins of the hairynap. We are unsure of the precise molecular composition of thetwo crystal types, but a very approximate molecular mass of 100kDa was estimated for the hexameric crown found in basal layercrystals (3). If we assume a 75% helical content, then each crownwould contain ∼650 amino acids in a helical conformationequivalent to an ∼100-nm length of helix, which could span thethickness of the crystal (as determined by AFM measurements;see below) about 20–25 times, equivalent to roughly three to four

Fig. 2. CD spectrum of exosporium fragments from B. cereus.

A

C D

B

Fig. 3. AFM height (A and C) and phase (B and D) images showing a foldedfragment of B. cereus exosporium. [Scale bars, 200 nm (A and B) and 50 nm(C andD).] Dark tobright variation in color represents a change inheightof 100nm (A) and 10 nm (C) and a change in phase of 40° (B) and 8° (D). ImagesA andB were taken in normal tapping mode using high-resolution diamond-likecarbon tips, and C andDwere taken in torsional tapping mode using T-shapedcantilevers. The ordered surface is indicated by the dashed white arrows, andthe disordered surface is indicated by the solid white arrows in B.

BA

C D

Fig. 4. AFM height images of exosporium fragments from B. thuringiensis4D11. [Scale bars, 500 nm (A), 50 nm (B), and 10 nm (C and D).] Dark to brightvariation in color represents a change in height of 80 nm (A) and 15 nm (B).B is a magnified image of the square area depicted in A. C denotes the areaindicated by the white arrow, and D represents the area indicated bythe black arrow in B. Images were taken in torsional tapping mode usingT-shaped cantilevers.

16016 | www.pnas.org/cgi/doi/10.1073/pnas.1109419108 Kailas et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

crystal-spanning helices per subunit. If we examine Fig. 1B, we seeone exceptionally dense feature (marked with an asterisk) withother less dense features (circled) clustered nearby at a center–center distance of ∼10 Å. Other, less dense (circled) featuresnearby could also correspond to shorter, tilted, or more mobilehelices. Similar clusters of dense helix-like features can be seen inthe projection of parasporal layer crystals, but we have not beenable to find any broad match of density features in the two maps.Ball et al. (3) had proposed that the type I structure originated

from a parasporal layer located in the interspace between theexosporium and the spore coat. It was suggested that this layermight represent an immature or nascent form of the type IIcrystal, which makes up the true exosporium basal layer. Theunambiguous threefold and sixfold symmetries now determinedfor the parasporal and basal layer crowns, respectively, show thatat the very least, the two forms must represent different assemblystates. Although we can show that the two forms have features incommon such as the overall crown-like architecture and proba-ble clusters of α-helices traversing the layer, there are clearlydifferences in detail. These differences may arise from bothconformational and compositional differences within the sub-units. A further notable feature is that the density in parasporallayer crystals appears much more tightly packed than in the basallayer crystals, consistent with its smaller unit cell. Of more cer-tain origin are the type II crystals, which make up a substantialpart of the exosporium basal layer to which the hairy nap is at-tached. Because one surface of the basal layer must be involvedin the interaction with the hairy nap and the other surface islikely to interact with proteins deeper within the spore, the as-signment of the relative orientation of the crystalline basal latticeis essential if we are to model a 3D structure of the spore and tounderstand how other spore proteins and factors from the ex-ternal environment interact with the crystalline layers. We foundfrom negative-stain electron micrographs that B. cereus 10876exosporium has a very clear hairy nap whereas B. thuringiensis4D11 has little or no nap. It should be noted, however, that thebasic basal layer architecture is identical for all B. cereus, B.anthracis, and B. thuringienis strains examined, and in general B.thuringiensis strains do have a hairy nap (3). The AFM imagesfrom the B. cereus samples (Fig. 3 and Fig. S2) suggest that oneside of the basal layer is made up of a well-ordered array of

dome-shaped structures and the other side is highly disordered.This disordered side is most likely to represent the hairy napcollapsed onto the external surface of the spore. Therefore, theordered array of domes must correspond to the interior-facingside of the basal layer. The domes appear to have a diameterranging between ∼7.5 and 9 nm that fits the unit cell dimensionsof the basal layer crystals determined by EM. In contrast to B.cereus, the AFM images from the B. thuringiensis 4D11 samplesshow that both surfaces of the basal layer are well-ordered (Fig.4), with one surface showing the same dome structures as seen inB. cereus. However, the other surface shows a honeycomb-likepattern with concave cups. This cup view must correspond to theviews into the crowns described by Ball et al. (3) for both B.cereus and B. thuringiensis strains and also inferred for B.anthracis. Because this B. thuringiensis strain is found to havelittle or no nap and the AFM images from the B. cereus strainshow the inner surface to be made of domes, the side having theconcave cups must correspond to the external face of the basallayer. From this, we can infer that the crowns described in ref. 3open out toward the external face of the spore.The use of high-resolution AFM to complement the EM work

has provided insights into the detailed architecture of the exo-sporium. AFM studies had previously been done on spore sur-faces (28–30), but the exosporium appeared to be featureless(28) or to have a granular texture (30), as the hairy nap presenton the exosporium would not have revealed the underlying or-dered structure of the basal layer. Because we had fragmentedthe exosporium as well as looked at B. cereus family strains withand without the hairy nap (but with identical basal layer archi-tecture), we were able to image both the outer and inner side ofthe crystalline basal layer and this has revealed the well-orderedcrystalline lattice. In our studies with fragmented exosporium, wehave calculated the thickness of the exosporium in B. cereus to be∼10 nm, whereas that in B. thuringiensis 4D11 is ∼4 nm. As thisB. thuringiensis sample was apparently devoid of the hairy nap,we can conclude that the crystalline basal lattice is ∼4-nm thickand the collapsed hairy nap is contributing to ∼6 nm in height onthe B. cereus basal layer.Fig. 5 shows a proposed model of the exosporium of the Ba-

cillus cereus family, bringing together data from negative-stainelectron microscopy, cryo-EM, and AFM. One primary struc-

Fig. 5. A schematic diagram illustrating a possible model for the exosporium of the B. cereus family. This outermost layer of the spore contains a 2D lattice ofcups opening out to the environment on the convex outer face of the layer. On the concave inner face, the cups may or may not be completely sealed, butthere are pores or channels between them. The collagen-like fibrils of the nap are shown schematically attached to the convex face, terminating in theglobular tumor necrosis factor-like C-terminal domain. The lengths of the fibrils vary between strains, and those shown here are not necessarily to scale.(Inset) A higher-magnification schematic view of a small segment of the exosporium. The crystalline basal layer is represented by the 3D surface determinedfor B. cereus by Ball et al. (3). The crowns are indicated schematically to have a high α-helical content, and we have shown the fibers of BclA to be docked inthe vicinity of the threefold symmetric linkers between the crowns. This may be via ExsFA/ExsFB, shown as small pillars anchoring BclA to the crystallinematrix. However, the docking sites may be elsewhere on the crown perimeter.

Kailas et al. PNAS | September 20, 2011 | vol. 108 | no. 38 | 16017

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

tural role for the basal layer is to act as a platform to which thehairy nap is attached. Our comparative AFM analysis of a B.cereus strain with nap and a very closely related B. thuringiensisstrain without nap has established that the nap is attached to thesurface corresponding to the crown openings. The positions ofthe nap anchor points on the surface are unknown, but are mostlikely to be on the outer perimeter of the crowns. In Fig. 5, wehave placed BclA fibrils attached near the threefold connectors,but they could be elsewhere on this outer surface. This could bean attractive hypothesis, because these connections have three-fold symmetry and could then dock specifically onto the three-stranded collagen-like fibers found in the proteins of the napsuch as BclA (14) in a specific fashion. However, there is evi-dence that one BclA trimer is linked to only one EsxFA/BxpBmonomer (16), which would break such symmetry. At the base ofthe hairy nap fibrils there is likely to be a complex of ExsF/BxpBand ExsY proteins, which would make up part of the observedbasal layer crystal structure. There are likely to be other lessabundant glycoprotein fibers, such as BclB, attached to the sameface; these have not been shown.What are the protein–protein interactions that hold the

crowns together in the lattice? This is an intriguing question,because the exosporium is very flexible. ExsFA and BclA areclearly not essential to hold the lattice together, because exo-sporium from ΔexsFA and ΔbclA, mutants maintains the crystallattice (8, 15). We speculate that ExsY and CotY could be goodcandidates for major components of the lattice because a ΔexsYstrain of B. cereus makes only a small terminal cap of exosporiumand a ΔexsY ΔcotY strain is completely devoid of exosporium (17,18, 31). It is possible that the main bulk of the exosporiumcontains ExsY whereas the cap contains CotY, which is 85%identical to ExsY (17). CotY could confer subtly differentstructural properties on the cap compared with the rest of theexosporium. The projection structure of the basal layer crystalsshows that the crystalline part of the basal layer is made up ofhexameric assemblies linked together. The putative helicesclustered around the sixfold symmetry axis (Fig. 1B) could haveroles in lining the wall of the crown cavity, whereas the putativeperipheral helices could have roles in defining pore size andalso linking individual crowns together through helix–helix andother interactions.The exosporium clearly forms a semipermeable barrier, be-

cause it fully encloses the spore. The projection map shown inFig. 1B shows relatively low density at the threefold crystallo-graphic symmetry positions, which coincide with the “linkers”revealed in the 3D negative-stain map. With the caveat thatthere is always additional uncertainty in interpretation of fea-tures on symmetry axes due to the enhanced noise level, this isconsistent with the large stain-filled volume that is seen to becapped by the threefold linker (3). Centered on these threefoldpositions in the unstained projection (Fig. 1B), we have drawna circle (dotted in red) whose circumference just grazes the threeprominent symmetry-related putative helices. This circle is lessthan 20 Å across, suggesting a minimum diameter for the cavitycapped by the linker; this cavity is connected to the three poresdistributed around the threefold axis (Fig. 5). A 20-Å constric-tion is entirely consistent with pore sizes estimated from per-meation experiments (32). A passage of this size is clearly largeenough to allow entry of the endospore germinants alanine orinosine (33) but not degradative enzymes or antibodies; equally,it would act as a barrier to the loss of proteins from the in-terspace between coat and exosporium.Fig. 5 summarizes the main conclusions of this paper, although

it should be noted that in different strains the basal layerthickness may vary depending on whether the layer has a singletier or is made up of multiple layers stuck together: (i) Theexosporium forms a semipermeable barrier enclosing the sporeand breached by passages as narrow as ∼20 Å in diameter and

small enough to exclude typical proteolytic enzymes but largeenough to allow entry of spore germinants. The barrier may notbe identically structured around the whole spore surface becausea specialized “cap” structure directs the escape of the germi-nating cell enclosed within the exosporium (31). (ii) The exo-sporium is made up of interlinked cups opening to the externalenvironment of the spore. (iii) These cups are constructed inlarge part from α-helices traversing some or all of the thicknessof the crystalline part of the basal layer; the bottom of the cupscould be closed off or might have a small depression or opening,which would be revealed only with higher-resolution data. (iv)The threefold symmetric linkers between the cups form opencaps over the relatively empty volume between crowns, and therecould be docking sites for the hairy nap fibers in the vicinity ofthese or elsewhere on the crown perimeter. The docking sitesmay be formed by a complex between BclA (and/or potentiallyother glycoprotein homologs), ExsFA/BxpB, and ExsY and/ortheir cognate homologs ExsFB and CotY. (v) The cups could actas a matrix to sequester various additional proteins associatedwith the exosporium such as arginase, alanine racemase, and soforth. (vi) The tapering of the cups toward the interior of thespore would facilitate the expected concave curvature of theinner surface of the exosporium as it wraps around the spore.(vii) The exosporium is highly deformable, which may allowa larger surface-to-surface contact area than would be achievedwith the more tightly wrapped spore coat. This could be an ad-vantage for attachment and spore dispersal in the environment.This deformability may come partly through flexible cross-linksbetween cups and/or partly through weak boundaries or defectsbetween crystalline domains that would allow bending andfolding. To test the structural aspects of this model, we requirehigher-resolution structural information in three dimensions.

Materials and MethodsExosporium Preparation. Crystalline fragments of exosporium were preparedas described in ref. 23, which included a French press step and salt and de-tergent washes. A detailed description of the spore preparation and exo-sporium isolation is given in SI Materials and Methods. Basal layer crystalswere from B. cereus American Type Culture Collection (ATCC) 10876 andB. thuringiensis 4D11. Parasporal layer crystals were from B. thuringiensiskurstaki HD1.

Electron Microscopy. Crystals from B. cereus ATCC 10876 were embedded invitreous ice using a Vitrobot freeze plunger (FEI) and those from B. thur-ingiensis kurstaki HD1 were embedded in 1% glucose or 1% glucose/1%uranyl formate on carbon-coated 400-mesh EM grids. Samples were moun-ted on a liquid nitrogen-cooled Oxford Instruments CT3500 Cryotrans coldstage in a Philips CM200 FEG transmission electron microscope operating at200 kV. Low-dose images (27) were collected at a nominal magnification of66,000 with an exposure of ∼10 electrons/Å2 in 1 s and recorded on KodakSO-163 film.

Image Processing. The developed micrographs were digitized on a Zeiss SCAIscanner. Digitized images were imported into 2dx_Image (34) and subjectedto rounds of crystal unbending (27). For each image, possible plane groupsymmetries were determined using ALLSPACE within the 2dx suite. Fourieramplitudes and phases generated for individual images from 2dx_Imagewere refined to a common phase origin using ORIGTILTK within the MRCsoftware suite (27). SCALEIMAMP was used to determine B-factors foramplitudes (26), and processed images were merged using the MRC suite ofprograms. Merged phases and amplitudes were averaged with AVRGAMPHSand used to generate figure-of-merit weighted projection maps (35).

CD Spectroscopy of Exosporium. A B. cereus exosporium suspension was ex-amined in a demountable Suprasil quartz cell (Hellma) with a path length of0.01 cm. Three repeats of the spectrum were measured at 10 °C over thewavelength range 190–280 nm with a step size of 1 nm using an Aviv 62dsspectropolarimeter with a detector acceptance angle of $90° (essential forsamples such as exosporium that scatter light). The data were processedusing CDtool software (36). Sample scans were averaged and the baselinescans of buffer were subtracted. The resulting spectrum was smoothed using

16018 | www.pnas.org/cgi/doi/10.1073/pnas.1109419108 Kailas et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021

a Savitsky–Golay filter. The protein concentration was determined using theminimum NRMSD method (37). Secondary structure content was analyzedusing Selmat3 analysis software with the SP175 reference dataset (38).

Atomic Force Microscopy. An exosporium suspension was applied to an ∼200-Å-thick carbon film supported on a copper Maxtaform finder grid (AgarScientific). The grid was imaged under ambient conditions in Tapping modeusing a Dimension 3100 AFM with Nanoscope IV controller (Veeco Instru-ments). High-resolution carbon tips (∼1-nm tip radius/DP15/HI9Res-C; Mik-roMasch) and silicon tips (<10-nm tip radius; Olympus) of typical springconstant 40 N/m and resonant frequency of ∼300 KHz were used for imag-ing. Additionally, “torsional tapping” measurements were carried out usinga Veeco Multimode AFM with a Nanoscope IIIa controller, extender elec-tronics, and signal access module as described in ref. 39. The T-shaped can-tilevers (TL01 HiRes-C; MikroMasch) used had a torsional spring constant of

∼15–20 N/m and an ∼1-nm tip radius. Height and phase images were col-lected simultaneously. Phase images were generated by recording the phaselag of the cantilever oscillation relative to the drive signal given to the piezodriver and given contrast that is indicative of mechanical and adhesiveproperties. Under normal imaging conditions, dark areas are soft and/orsticky, whereas bright areas are stiff and less sticky.

ACKNOWLEDGMENTS. We thank Prof. Pete Artymiuk, Dr. Pu Qian, Dr. JohnOlsen, and Mr. Chris Glover (University of Sheffield) for helpful discussionsand suggestions with the drawing of the schematic diagram, and Dr. ChristaWeber (University of Sheffield) for assistance with the use of high-resolutionAFM tips. The authors (L.K., C.T., R.T., S.B.T., S.J.T., B.A.W., A.M., and P.A.B.)thank the Biotechnology and Biological Sciences Research Council and (N.M.and J.K.H.) the Medical Research Council and Engineering and PhysicalSciences Research Council for funding.

1. Nicholson WL, Munakata N, Horneck G, Melosh HJ, Setlow P (2000) Resistance ofBacillus endospores to extreme terrestrial and extraterrestrial environments. Micro-biol Mol Biol Rev 64:548e572.

2. Errington J (2003) Regulation of endospore formation in Bacillus subtilis. Nat RevMicrobiol 1:117e126.

3. Ball DA, et al. (2008) Structure of the exosporium and sublayers of spores of theBacillus cereus family revealed by electron crystallography. Mol Microbiol 68:947e958.

4. Henriques AO, Moran CP, Jr. (2007) Structure, assembly, and function of the sporesurface layers. Annu Rev Microbiol 61:555e588.

5. Driks A (1999) Bacillus subtilis spore coat. Microbiol Mol Biol Rev 63:1e20.6. Setlow P (2006) Spores of Bacillus subtilis: Their resistance to and killing by radiation,

heat and chemicals. J Appl Microbiol 101:514e525.7. Gerhardt P, Ribi E (1964) Ultrastructure of the exosporium enveloping spores of Ba-

cillus cereus. J Bacteriol 88:1774e1789.8. Sylvestre P, Couture-Tosi E, Mock M (2002) A collagen-like surface glycoprotein is

a structural component of the Bacillus anthracis exosporium. Mol Microbiol 45:169e178.

9. Beaman TC, Pankratz HS, Gerhardt P (1971) Paracrystalline sheets reaggregated fromsolubilized exosporium of Bacillus cereus. J Bacteriol 107:320e324.

10. Cybulski RJ, Jr., Sanz P, O’Brien AD (2009) Anthrax vaccination strategies. Mol AspectsMed 30:490e502.

11. Swiecki MK, Lisanby MW, Shu F, Turnbough CL, Jr., Kearney JF (2006) Monoclonalantibodies for Bacillus anthracis spore detection and functional analyses of sporegermination and outgrowth. J Immunol 176:6076e6084.

12. Thompson BM, Stewart GC (2008) Targeting of the BclA and BclB proteins to theBacillus anthracis spore surface. Mol Microbiol 70:421e434.

13. Boydston JA, Chen P, Steichen CT, Turnbough CL, Jr. (2005) Orientation within theexosporium and structural stability of the collagen-like glycoprotein BclA of Bacillusanthracis. J Bacteriol 187:5310e5317.

14. Réty S, et al. (2005) The crystal structure of the Bacillus anthracis spore surface proteinBclA shows remarkable similarity tomammalian proteins. J Biol Chem 280:43073e43078.

15. Sylvestre P, Couture-Tosi E, Mock M (2005) Contribution of ExsFA and ExsFB proteinsto the localization of BclA on the spore surface and to the stability of the Bacillusanthracis exosporium. J Bacteriol 187:5122e5128.

16. Tan L, Li M, Turnbough CL, Jr. (2011) An unusual mechanism of isopeptide bondformation attaches the collagenlike glycoprotein BclA to the exosporium of Bacillusanthracis. mBio 2:e00084-11.

17. Boydston JA, Yue L, Kearney JF, Turnbough CL, Jr. (2006) The ExsY protein is requiredfor complete formation of the exosporium of Bacillus anthracis. J Bacteriol 188:7440e7448.

18. Johnson MJ, et al. (2006) ExsY and CotY are required for the correct assembly of theexosporium and spore coat of Bacillus cereus. J Bacteriol 188:7905e7913.

19. Redmond C, Baillie LWJ, Hibbs S, Moir AJG, Moir A (2004) Identification of proteins inthe exosporium of Bacillus anthracis. Microbiology 150:355e363.

20. Charlton S, Moir AJG, Baillie L, Moir A (1999) Characterization of the exosporium ofBacillus cereus. J Appl Microbiol 87:241e245.

21. Weaver J, et al. (2007) Protective role of Bacillus anthracis exosporium in macro-phage-mediated killing by nitric oxide. Infect Immun 75:3894e3901.

22. Oliva C, Turnbough CL, Jr., Kearney JF (2009) CD14-Mac-1 interactions in Bacillusanthracis spore internalization by macrophages. Proc Natl Acad Sci USA 106:13957e13962.

23. Todd SJ, Moir AJG, Johnson MJ, Moir A (2003) Genes of Bacillus cereus and Bacillusanthracis encoding proteins of the exosporium. J Bacteriol 185:3373e3378.

24. Bozue J, Cote CK, Moody KL, Welkos SL (2007) Fully virulent Bacillus anthracis doesnot require the immunodominant protein BclA for pathogenesis. Infect Immun 75:508e511.

25. Brahmbhatt TN, et al. (2007) Bacillus anthracis exosporium protein BclA affects sporegermination, interaction with extracellular matrix proteins, and hydrophobicity. In-fect Immun 75:5233e5239.

26. Havelka WA, Henderson R, Oesterhelt D (1995) Three-dimensional structure of hal-orhodopsin at 7 Å resolution. J Mol Biol 247:726e738.

27. Henderson R, Baldwin JM, Downing KH, Lepault J, Zemlin F (1986) Structure of purplemembrane from Halobacterium halobium: Recording, measurement and evaluationof electron micrographs at 3.5Å resolution. Ultramicroscopy 19:147e178.

28. Chada VGR, Sanstad EA, Wang R, Driks A (2003) Morphogenesis of Bacillus sporesurfaces. J Bacteriol 185:6255e6261.

29. Zaman MS, et al. (2005) Imaging and analysis of Bacillus anthracis spore germination.Microsc Res Tech 66:307e311.

30. PlompM, Leighton TJ, Wheeler KE, Malkin AJ (2005) Architecture and high-resolutionstructure of Bacillus thuringiensis and Bacillus cereus spore coat surfaces. Langmuir21:7892e7898.

31. Steichen CT, Kearney JF, Turnbough CL, Jr. (2007) Non-uniform assembly of the Ba-cillus anthracis exosporium and a bottle cap model for spore germination and out-growth. Mol Microbiol 64:359e367.

32. Gerhardt P, Scherrer R, Black SH (1972) in Spores V: Papers Presented at the FifthInternational Spore Conference, Fontana, Wisconsin, 8–10 October 1971, edsHalvorson HO, Hanson R, Campbell LL (Am Soc Microbiol, Washington, DC), pp 68e74.

33. Barlass PJ, Houston CW, Clements MO, Moir A (2002) Germination of Bacillus cereusspores in response to L-alanine and to inosine: The roles of gerL and gerQ operons.

Microbiology 148:2089e2095.34. Gipson B, Zeng X, Zhang ZY, Stahlberg H (2007) 2dx—User-friendly image processing

for 2D crystals. J Struct Biol 157:64e72.35. Crowther RA, Henderson R, Smith JM (1996) MRC image processing programs. J Struct

Biol 116:9e16.36. Lees JG, Smith BR, Wien F, Miles AJ, Wallace BA (2004) CDtool—An integrated soft-

ware package for circular dichroism spectroscopic data processing, analysis, and ar-chiving. Anal Biochem 332:285e289.

37. Miles AJ, Whitmore L, Wallace BA (2005) Spectral magnitude effects on the analysesof secondary structure from circular dichroism spectroscopic data. Protein Sci 14:

368e374.38. Lees JG, Miles AJ, Wien F, Wallace BA (2006) A reference database for circular di-

chroism spectroscopy covering fold and secondary structure space. Bioinformatics 22:

1955e1962.39. Mullin N, et al. (2009) “Torsional tapping” atomic force microscopy using T-shaped

cantilevers. Appl Phys Lett 94:173109.

Kailas et al. PNAS | September 20, 2011 | vol. 108 | no. 38 | 16019

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

021