Surface-accessible residues in the monomeric and assembled forms of a bacterial surface layer protein.

The S-layer protein SbsB of the thermophilic, Gram-positive organism Bacillus stearothermophilus PV72/p2 forms a crystalline, porous array constituting the outermost component of the cell envelope. SbsB has a molecular mass of 98 kDa, and the corresponding S-layer exhibits an oblique lattice symmetry. To investigate the molecular structure and assembly of SbsB, we replaced 75 residues (mainly serine, threonine, and alanine), located throughout the primary sequence, with cysteine, which is not found in the wild-type protein. As determined by electron microscopy, 72 out of 75 mutants formed regularly-structured self-assembly products identical to wild-type, thereby proving that the replacement of most of the selected amino acids by cysteine does not dramatically alter the structure of the protein. The three defective mutants, which showed a greatly reduced ability to self-assemble, were, however, successfully incorporated into S-layers of wild-type protein. Monomeric SbsB mutants and SbsB mutants assembled into S-layers were subjected to a surface accessibility screen by targeted chemical modification with a 5-kDa hydrophilic cysteine-reactive polyethylene glycol conjugate. In the monomeric form of SbsB, 34 of the examined residues were not surface accessible, while 23 were classified as very accessible, and 18 were of intermediate surface accessibility. By contrast, in the assembled S-layers, 57 of the mutated residues were not accessible, six were very accessible, and 12 of intermediate accessibility. Together with other structural information, the results suggest a model for SbsB in which functional domains are segregated along the length of the polypeptide chain.

most cell envelope component of many eubacteria as well as archaea (1)(2)(3). S-layers are composed of identical protein or glycoprotein subunits, ranging in mass from 40 to 200 kDa, which have the ability to self-assemble into crystalline planar arrays. S-layer lattices can possess a variety of plane group symmetries (oblique (p1, p2), square (p4), or hexagonal (p3, p6)) and all contain pores of uniform size and morphology.
The ultrastructure of S-layers has been well characterized by electron microscopy and some aspects of the molecular structures of certain S-layer proteins have been investigated. Most is known about the interaction of S-layer proteins with the underlying cell envelope (Refs. 4 -9, reviewed in Refs. 10 -12, and references therein). Other studies identified structurally and morphologically defined domains (9,13,14). In two cases, the surfaces of S-layers and the constituent proteins were mapped by probing the accessibility to antibodies recognizing either S-layer-specific epitopes (15) or a foreign epitope, which was inserted into the S-layer protein at semirandom positions (16). Despite the relatively large number of studies, little is known about the spatial location of individual residues in assembled S-layers. Knowledge about which residues are located (i) at the external surface of the S-layer lattice, (ii) at the interface between the subunits, and (iii) within the pores ( Fig.  1) is crucial for a better understanding of the assembly and molecular structure of S-layers.
In this report, cysteine scanning mutagenesis combined with targeted chemical modification (17)(18)(19)(20) is used to identify surface accessible residues (21) in SbsB. SbsB is the S-layer protein of Bacillus stearothermophilus PV72/p2, a thermophilic Gram-positive bacterium (22,23). SbsB has a molecular mass of 98 kDa and assembles into a lattice of oblique symmetry. The N terminus of SbsB (residues 34 to 212 in the preprotein sequence) encodes three S-layer homology (SLH) 1 domains (11). SLH domains were originally identified by sequence comparison (14) and have been shown to attach exoproteins of Gram-positive bacteria to the underlying cell wall (see Refs. 10 and 11, and references therein). In most proteins studied so far, SLH domains specifically bind to secondary cell wall polymers (10), and, in the case of SbsB, the polymer is composed mainly of GlcNAc and ManNAc (24). Apart from binding the secondary cell wall polymer (10), the N terminus of SbsB also binds to the peptidoglycan-containing cell wall itself (25). In addition, SbsB contains a second binding region for the secondary cell wall polymer, which was mapped to amino acid positions 271-362 (25). Under native conditions, the N terminus (up to amino acid position 206) of SbsB can be removed by limited proteolysis, while the remainder of the SbsB polypeptide is proteolytically stable. 2 A shortened version of SbsB, beginning with amino acid 200, is sufficient for S-layer formation, as studies with genetically truncated versions of SbsB have demonstrated. 2 Deletions of the C terminus, however, abolish self-assembly. These data suggest the existence of a biochemically defined structural domain, which might correspond to the massive morphological domain, visible in computer-processed transmission electron micrographs of S-layer lattices from B. stearothermophilus PV72/p2 (26).
The goal of the present study was to use targeted chemical modification to identify residues, which are located at the surface of the S-layer protein SbsB. By using monomeric SbsB and SbsB assembled on cell walls, we have been able to discriminate between residues which are (i) located at the surface of the monomer, (ii) positioned on the outer surface of the S-layer lattice or within the lumen of the pores, and (iii) located on the inner surface of the lattice or at the interface between assembled subunits (Fig. 1). In addition, the availability of the extensive set of surface-located cysteines will facilitate further investigations of SbsB and may allow the engineering of Slayers by targeted chemical modification to produce novel biomolecular materials.

EXPERIMENTAL PROCEDURES
Generation of Cysteine Mutants of SbsB-Cysteine mutations of SbsB, the S-layer protein of B. stearothermophilus PV72/p2 (23,27), were made for 75 amino acid residues (31 threonine, 18 serine, 7 alanine, 5 glycine, 2 each of arginine, aspartic acid, valine, and asparagine, and 1 each of isoleucine, tyrosine, phenylalanine, and lysine). The amino acids were dispersed throughout the primary sequence between positions 234 and 913 of the SbsB polypeptide. By using predominantly conservative amino acid changes (threonine, serine, and alanine), negative effects on the ability of SbsB to self-assemble were avoided. Mutation of threonine and serine also favored the identification of surface-located residues, since polar residues, such as serine and threonine, tend to be located on the surfaces of proteins (28,29). All mutants were generated by oligonucleotide-directed mutagenesis using an improved version (30) of the recombination PCR method (31) with 10 PCR cycles. To facilitate the identification of cysteine mutants (see below), an NsiI restriction site (ATGCAT), which encompassed the cysteine codon (underlined), was introduced with each mutation. Isolates for mutants G273C and T617C were subjected to sequencing, which showed that no PCR errors had occurred. For the biochemical assays on the thermal stability of monomeric cysteine mutants in SDS, and on the surface accessibility of monomeric and assembled cysteine mutants by PEG-OPSS modification, two independent isolates for each mutation were used. For the electron microscopic investigation of the ability of the cysteine mutants to self-assemble, one isolate was used, unless otherwise stated. The template for the mutagenesis was pT7sbsBHis. This plasmid harbors the sbsB gene (23), excluding its 31-residue signal sequence, but with an additional sequence coding for a C-terminal 6xHis-tag preceded by a 15-amino acid linker (KPNSADI-HHTGGRSS). The sbsB sequence in pT7sbsBHis had been amplified by PCR from chromosomal DNA of B. stearothermophilus PV72/p2. Sequencing revealed three point mutations compared with the corrected version of the published sequence of sbsB: Thr 210 3 Ser, Ser 402 3 Thr, and Lys 812 3 Glu. The numbers refer to the amino acid positions in the sequence of the preprotein. The corrected version of the published sequence was established after sequencing three independent isolates of the sbsB gene, which was PCR amplified from chromosomal DNA. The corrected version differed from the literature sequence (23) at three positions (Arg 139 3 Ala, Gln 391 3 Glu, and Asp 500 3 Glu).
In Vivo Expression and Recrystallization-All 75 SbsB cysteine mutants were expressed in Escherichia coli, purified with Ni-NTA metalaffinity chromatography, and dialyzed to obtain self-assembly products. Unless otherwise stated, one isolate of each mutant was used. E. coli JM109(DE3) cells (Promega, Madison, WI), which had been made competent by using PEG and dimethyl sulfoxide (32,33), were transformed with plasmid DNA. Bacterial cells grown in terrific broth medium (34) were centrifuged, and pellets (0.35 g wet weight, corresponding to 4.5 ml of bacterial culture) were resuspended in buffer A (1 ml: 8 M urea, 0.03 M Tris, 0.1 M NaH 2 PO 4 , adjusted to pH 8 with NaOH). After three freeze and thaw cycles in dry ice/ethanol the slurry was centrifuged for 20 min at 128,000 ϫ g at room temperature. The supernatant was mixed with buffer A (400 l) and 50% Ni-NTA-agarose (300 l) (Qiagen, number 30210) (pre-equilibrated with buffer A). Following the manufacturers recommendations, the resin was washed and then eluted three times with buffer E (150 l: buffer A plus 250 mM imidazole, pH 8). The three fractions were collected and analyzed by SDS-PAGE and Coomassie Blue staining.
The first two fractions, which contained most of the S-layer protein, were combined in a dialysis membrane with a molecular weight cut off of 15,000 (Spectra/Por Biotech Membranes) and dialyzed for 3 days at 4°C against 10 mM Tris-HCl, pH 7.5, supplemented with 0.1 mM DTT. The dialysis buffer was changed four times. In all cases, a white precipitate formed during the dialysis, which was analyzed by electron microscopy.
Electron Microscopy-Suspensions of all 75 in vivo expressed, dialyzed cysteine mutants of SbsB were examined by negative staining to visualize self-assembly products as described (35). If the first isolate of one mutant did not show regular self-assembly products, two additional isolates were analyzed by electron microscopy. To obtain ultrathin sections of self-assembled Wt-SbsB, samples prepared as described in the section: "Incorporation of In Vitro Expressed SbsB Mutants into Wt-SbsB S-Layers Assembled onto Cell Wall Sacculi" were treated as reported (36).
Expression by Coupled in Vitro Transcription/Translation-SbsB mutants were expressed from plasmid DNA by in vitro transcription and translation (IVTT) using the E. coli S30 T7 IVTT kit (Promega number L1130). The reaction volume contained the amino acid mixture minus methionine or minus cysteine (1.25 l), premix (5 l), and S30 mixture (3.75 l) supplemented with 1 g/ml rifampicin. Either 7 Ci of [ 35 S]methionine (Amersham Pharmacia Biotech, 1,200 Ci/mmol; corresponding to a 15 M solution) (1 l) or 6 Ci of [ 35 S]cysteine (Amersham Pharmacia Biotech, 1,000 Ci/mmol) (1 l) and supercoiled plasmid DNA (500 ng) were added to give a final reaction volume of 12.5 l. The reactions were incubated for 1 h at 37°C and then centrifuged for 5 min at 21,000 ϫ g. The supernatant containing 35 S-labeled SbsB was stored at Ϫ80°C. Before use, the samples were thawed and centrifuged again for 5 min at 21,000 ϫ g.
Size Exclusion Chromatography of in Vitro-expressed SbsB-An IVTT mixture containing [ 35 S]methionine Wt-SbsB-His (70 l) was analyzed with a size exclusion column (Superdex 200 HR 10/30, Amersham Pharmacia Biotech) on an Ä KTA purifier (Amersham Pharmacia Biotech). For column washing, loading, and elution, 50 mM Na phosphate buffer, pH 7.0, 50 mM NaCl was used. The absorbance of the fractions was monitored at 220 nm. The molecular weight standards used for calibration were blue dextran (M r 2,000,000), thyroglobulin (M r 669,000), ferritin (M r 440,000), aldolase (M r 158,000) (Amersham Pharmacia Biotech, gel filtration HMW calibration kit 17-0441-01), albumin (M r 66,000) (Sigma, A-4503), and cytochrome c (M r 12,400) (Sigma, C-7752). The protein in each fraction (0.5 ml) was precipitated with trichloroacetic acid, taken up in Laemmli buffer, boiled, and analyzed by SDS-PAGE followed by autoradiography or PhosphorImager analysis (37). 2 Y. Wang and H. Bayley, manuscript in preparation. Incorporation of in Vitro-expressed SbsB Mutants into Wt-SbsB S-Layers Assembled onto Cell Wall Sacculi-Seventy-five [ 35 S]methionine-labeled cysteine mutants (two independent isolates of each) were incorporated into S-layers of Wt-SbsB from B. stearothermophilus PV72/p2 by self-assembly on peptidoglycan-containing cell wall sacculi of B. stearothermophilus of PV72/p2 (27). Cysteine mutants were affinity purified by mixing IVTT mixture (12.5 l) with 50% Ni-NTA-agarose (15 l) (Qiagen, number 30210) and buffer A (see above), and then by following the manufacturers recommendations. Two fractions (22 l each) were eluted with buffer E. To prepare the self-assembly products, peptidoglycan-containing cell wall sacculi (100 g dry weight) and Wt-SbsB (25 g, dry weight) isolated from B. stearothermophilus PV72/p2 (25) were taken up in buffer E (10 l; buffer E contains urea) and mixed with the combined eluates of the metal affinity column (approximately 5 ng of [ 35 S]methionine-labeled cysteine mutant as determined by SDS-PAGE and PhosphorImager analysis). The sample was transferred into a dialysis membrane with a molecular weight cut off of 15,000 (Spectra/Por Biotech Membranes) and dialyzed for 2 days at 4°C against 10 mM Tris-HCl, pH 7.5, supplemented with 0.3 mM DTT. The dialysis buffer was changed four times. To remove excess DTT and unincorporated soluble protein, the dialysate was centrifuged for 5 min at 100 ϫ g and the pellet washed once with 50 mM Tris-HCl, pH 7.5 (100 l). The supernatants of the washing steps were pooled and retained for further analysis. The pellet was resuspended in 50 mM Tris-HCl, pH 7.5 (50 l), and used for PEG-OPSS modification as described in the section "PEG-OPSS Modification of Assembled SbsB Mutants." For the visualization of self-assembly products by electron microscopy, samples were prepared as described above, except that the IVTT mixture was supplemented with unlabeled 15 M methionine (1 l) instead of radiolabeled methionine. The samples were examined as described in the section "Electron Microscopy." To determine the extent of incorporation of each cysteine mutant into self-assembly products, the protein content of the supernatant and pellet fractions from two isolates each of all 75 mutants were analyzed by SDS-PAGE and autoradiography. The pooled washing solutions were centrifuged for 5 min at 15,000 rpm (21,000 ϫ g) and the supernatant was carefully withdrawn. Protein in the supernatant was precipitated by the addition of trichloroacetic acid to 5%, taken up in Laemmli buffer, boiled, and analyzed along with samples of the pellet fraction by SDS-PAGE and PhosphorImager quantification.
Assay of the Thermal Stability of Monomeric SbsB Mutants in SDS-To assay the thermal stability of monomeric SbsB in SDS, [ 35 S]methionine IVTT mixture (3 l) was added to 6 ϫ Laemmli buffer (0.35 M Tris-HCl, pH 6.8, 10.3% SDS, 30% glycerol) containing 0.1 M DTT (21 l). Three aliquots of the solution were taken, one was left at room temperature (25°C), and the others were incubated at 50°C (5 min) or at 100°C (5 min), cooled to 25°C and loaded onto SDS-PAGE gels. The gels were run at 25°C and 100 V, dried, and analyzed by autoradiography. This assay was performed on two independent isolates for each mutation.
PEG-OPSS Modification of Monomeric SbsB Mutants-Seventy-five [ 35 S]methionine-labeled cysteine mutants (two independent isolates of each) were modified with methoxypoly(ethylene glycol) orthopyridyl disulfide, M r 5,000 (PEG-OPSS) (Shearwater Polymers, Huntsville, AL). IVTT mixture (3 l) was diluted with 0.5 M Tris-HCl, pH 7.5 (18 l), followed by the addition of 2.5 mM DTT (3 l). Three aliquots (3 ϫ 7 l) of this mixture were taken and incubated for 10 min at room temperature. Two aliquots were chilled in an ice-water bath for 10 min, and mixed with a prechilled aqueous solution of 50 mM PEG-OPSS (2 l). After 10 and 20 s, unreacted cysteine residues were blocked by the addition of N-ethylmaleimide (2.8 M NEM in ethanol) (2 l). For SDS-PAGE a portion of the blocked reaction mixture (3 l) was mixed with 6 ϫ Laemmli buffer (8 l) (devoid of DTT or ␤-mercaptoethanol) and heated for 5 min at 100°C. To the third aliquot, 50 mM PEG-OPSS (2 l) was added and the mixture was incubated for 10 min at 25°C. A portion of this unblocked reaction mixture (3 l) was mixed with 6 ϫ Laemmli buffer (8 l) (devoid of DTT or ␤-mercaptoethanol), heated for 5 min at 100°C. The mixture was then supplemented with 0.5 l of the ethanolic solution of 2.8 M NEM and heated again for 5 min at 100°C. Samples (5 l) were separated by SDS-PAGE for autoradiography and PhosphorImager analysis.

PEG-OPSS Modification of Assembled SbsB Mutants-[ 35 S]
Methionine-labeled cysteine mutants which had been incorporated into Slayers of Wt-SbsB from B. stearothermophilus PV72/p2 and assembled onto peptidoglycan-containing cell wall sacculi of B. stearothermophilus of PV72/p2 (see section "Incorporation of In Vitro Expressed SbsB Mutants into Wt-SbsB S-Layers Assembled onto Cell Wall Sacculi") were modified with PEG-OPSS. The assay was performed on two inde-pendent isolates for all mutations. A freshly prepared suspension of cell wall sacculi with bound Wt-SbsB S-layer doped with a radioactively labeled mutant (30 l) was mixed with 0.5 M Tris-HCl, pH 7.5 (9 l), and 2.5 mM DTT (3 l). Three aliquots (3 ϫ 14 l) of the mixture were prepared, and treated and analyzed as described in the section, "PEG-OPSS Modification of Monomeric SbsB Mutants."

RESULTS
The properties of 75 cysteine mutants of SbsB are reported here. With the exception of mutant R79C, the mutations were dispersed throughout the primary sequence between amino acid positions 234 and 913 of the 920-residue SbsB polypeptide. The numbers refer to the amino acid positions in the sequence of the preprotein (23). All mutants were derived from Wt-SbsB-His, which lacks the 31-amino acid signal peptide, but carries a 6xHis tag at the C terminus attached through a 15-amino acid linker. The sequence of Wt-SbsB-His differs from the natural sequence at three codons altered by PCR errors: Thr 210 3 Ser, Ser 402 3 Thr, Lys 812 3 Glu. Electron microscopy showed that neither the C-terminal 6xHis tag nor the three point mutations changed the ability of the S-layer protein to self-assemble (Fig.  3B). In the following, Wt-SbsB-His refers to this sequence. The S-layer protein SbsB isolated from B. stearothermophilus PV72/p2 is referred to as Wt-SbsB.
For the scanning mutagenesis, the residues alanine, serine, and threonine were the primary choices for the mutation to cysteine, thereby producing conservative changes that would not disrupt the native structure of SbsB. This selection was also intended to increase the likelihood of finding surface accessible residues, since polar residues, such as serine and threonine, tend to be located at the surface of proteins (28,29).
All 75 mutants were generated by site-directed mutagenesis using a PCR-based protocol featuring a low number of PCR cycles. Sequencing of single isolates of two different mutants showed that no PCR errors had occurred. To ease the identification of positive clones, an NsiI restriction site (ATGCAT), which encompassed the cysteine codon (underlined), was introduced with each mutation.
For all mutants at least two independent isolates were generated. For the biochemical assays on the thermal stability of monomeric cysteine mutants in SDS, and on the surface accessibility of monomeric and assembled cysteine mutants by PEG-OPSS modification, two independent isolates for each mutation were used. For the assay on the ability of the cysteine mutants to self-assemble as determined by negative staining and electron microscopy, one isolate was used, unless otherwise stated.
In Vivo Expression and Self-assembly of SbsB Cysteine Mutants-SDS-PAGE confirmed that His-tagged SbsB mutants were expressed at high levels in vivo ( Fig. 2A, lane 1) and could be quantitatively removed from the cell lysate by binding to Ni-NTA resin, as exemplified by mutant G273C (Fig. 2A, lane  2). All mutants were expressed and purified, yielding similar protein concentrations in the eluates from Ni-NTA resin, as shown for 13 different mutants (Fig. 2B). The average total protein yield after three elutions was approximately 8 mg of SbsB protein from 4.5 ml of bacterial culture. The purified mutants were all dialyzed under the same conditions. During dialysis a white precipitate formed. To test whether self-assembly products had been formed, the precipitates of all 75 mutants were analyzed by electron microscopy following negative staining. For example, the electron micrograph of the precipitate from mutant T315C (Fig. 3A) shows a regular pattern with oblique lattice symmetry characteristic of Wt-SbsB-His selfassembly products (Fig. 3B). In summary, 72 out of 75 mutants had the ability to form self-assembly products like Wt-SbsB-His (Table I). Therefore, in most cases, the mutation does not dramatically alter the structure of the protein, even though the substitution to cysteine is often accompanied by a second, con-servative 3 amino acid replacement to allow the introduction of an NsiI restriction site (ATGCAT; cysteine codon underlined). In some cases non-conservative amino acid replacements were deliberately generated to probe the conformational tolerance of SbsB (e.g. E837C/Q838I). In summary, among the 48 cysteine mutants, which were accompanied by a second amino acid change, 29% (14/48) had one conservative and one non-conser-vative change, 2% (1/48) two non-conservative changes while the rest, 69% (33/48), had two conservative amino acid changes. Among the 18 mutants with amino acid changes in three positions, 6% (1/18) were conservative in all three positions, 55% (10/18) conservative in two positions, and 39% (7/18) conservative in one position. 56% (5/9) of cysteine mutations not accompanied by a second change were non-conservative. The three isolates that did not show self-assembly products with regular lattices were: T240C/V241I; D723E/G724C, and T816C/ G817I. These three isolates did, however, form a white precipitate upon dialysis. To exclude the possibility of additional mutations stemming from defective mutagenic primers, we sequenced the corresponding region and found no errors. The electron microscopic examination of second and third isolates confirmed that the three mutants were assembly-compromised, and this eliminates the possibility that mutations stemming from PCR errors might have caused the assembly-compromised phenotype. It was, however, observed, that in one out of two separate dialysates of one isolate of mutant T240C/V241I traces of S-layer formed, suggesting that the ability to assemble had not been completely eliminated in this mutant.
In Vitro Expression of Cysteine Mutants-Selected SbsB cysteine mutants 4 were expressed in vitro with radiolabeled cysteine to demonstrate the successful mutational change to cysteine. For example, G273C was expressed in vitro with [ 35 S]cysteine and analyzed by SDS-PAGE and autoradiography (Fig. 4). As expected, G273C (Fig. 4, lane 3) but not Wt-SbsB-His (Fig. 4, lane 1), which does not harbor a cysteine codon, gave a positive signal. With [ 35 S]methionine both Wt-SbsB-His (Fig. 4, lane 2) and G273C (Fig. 4, lane 4) were labeled. (Wt-SbsB-His contains 5 methionine codons.) The major protein band migrating at 100 kDa (indicated by an arrow in Fig. 4) corresponds to full-length SbsB. The less intense and faster migrating bands most likely result from premature termination of transcription or translation, since they did not bind to Ni-NTA resin (data not shown), suggesting that they did not contain the C-terminal 6xHis tag.
Definition of the Different Substrates Used for PEG-OPSS Modification-The assembly states of in vitro expressed SbsB and Wt-SbsB self-assembly products doped with in vitro expressed SbsB were examined by size exclusion chromatography and electron microscopy, respectively.
To determine whether in vitro-expressed SbsB was mono-  The properties of in vivo expressed mutants (self-assembly) and in vitro expressed mutants (thermal stability in SDS and surface accessibility) are tabulated. The assays for in vitro expressed mutants were performed for two independent isolates, and the assay for in vivo expressed mutants on one isolate, unless otherwise stated. S-layer formation was determined by electron microscopy. The assay of the thermal stability in SDS is a tool to probe the conformation of the monomeric mutant based on the structural stability of an SDS-refractile folded motif. Modification with PEG-OPSS was performed for monomeric cysteine mutants and for SbsB cysteine mutants incorporated into self-assembly products of SbsB from B. stearothermophilus PV72/p2 bound to peptidoglycan-containing cell wall sacculi of B. stearothermophilus PV72/p2. Key: self-assembly: ϩ, self-assembly products similar to Wt-SbsB-His; Ϫ, greatly reduced ability to form self-assembly products (confirmed with three independent isolates). Thermal stability in SDS: ϩϩϩ, at 25°C, 80-kDa SbsB band and at 50°C, 80-and 100-kDa bands at similar intensities (i.e. like Wt-SbsB-His); ϩ, at 25°C, 80-kDa band and at 50°C, 100-kDa band predominates; Ϫ, 100 kDa band at 25 and 50°C. Modification: ϩϩϩ, more than 50% modification after 20 s reaction; ϩ, between 50 and 10% modification after 20 s; Ϫ, less than 10% modification after 20 s. 100% modification is the extent of modification for the denatured protein.
In vivo expressed, Self-assembly Wt-SbsB-His All 75 in vitro-expressed SbsB mutants (two independent isolates each) were incorporated into S-layers by dialyzing Histag purified [ 35 S]methionine-labeled SbsB out of 8 M urea in the presence of a 5000-fold molar excess of in vivo expressed Wt-SbsB and cell wall sacculi of B. stearothermophilus PV72/p2. Self-assembly products of Wt-SbsB bind with their inner surface against the peptidoglycan-containing cell wall sacculi leaving the outer side of the lattice surface exposed and accessible (24) (Fig. 1). In the present study, cell wall-bound self-assembled mutants were used to identify residues located on the outer surface of the S-layer. It was undesirable to use S-layers assembled in the absence of cell wall sacculi, because free S-layers tend to form double layers in suspension. An electron micrograph shows that Wt-SbsB assembled onto peptidoglycan-containing cell wall sacculi as sheet-like S-layers (Fig. 5). Rigorous examination of the ultrathin sections of embedded sacculi confirmed that the preparations did not contain unbound S-layers. Free S-layer monolayers and multilayers were found, however, when the ratio of Wt-SbsB protein over cellwall sacculi was increased (not shown).
Since 72 in vivo-expressed SbsB cysteine mutants were able to crystallize into self-assembly products like Wt-SbsB-His (see above), it was very likely that in vitro expressed SbsB cysteine mutants would crystallize with Wt-SbsB to form S-layers. To determine how much in vitro expressed SbsB was incorporated into self-assembly products, we determined the amount of radiolabeled SbsB in the supernatant and the pellet after dialysis for all mutants. Analysis by SDS-PAGE and PhosphorImager quantification demonstrated that more than 90% of the in vitro expressed SbsB became incorporated into self-assembly products with Wt-SbsB (data not shown). The fraction of incorporated in vitro expressed protein was very similar for all 75 mutants tested. This indicates that the in vitro expressed mutants T240C, G724C, and T816C were incorporated into the lattice of Wt-SbsB. By contrast, the in vivo expressed mutant polypeptides had a greatly reduced ability to form crystalline self-assembly products in the absence of Wt-SbsB. Hence, it is likely that SbsB mutants T240C, G724C, and T816C adopt the assembly-competent conformation in the presence of a large excess of wild-type protein. In the following, T240C, G724C, and T816C were included in the assays designed to probe the structure of SbsB.
Assay of the Thermal Stability of Monomeric Cysteine Mutants in SDS-This assay was performed to probe the conformation of the [ 35 S]methionine-labeled monomeric form of the mutants and carried out for all 75 mutants with two independent isolates. Wt-SbsB-His (like Wt-SbsB) remains partly folded in SDS at 25°C. Upon SDS-PAGE, the unheated protein migrates at an apparent molecular mass of 80 kDa (Fig. 6A, lane 1) compared with 100 kDa for heated unfolded SbsB (Fig. 6A,  lane 3). After treatment at 50°C, both the partly folded, faster migrating form as well as the unfolded form were found at similar band intensities (Fig. 6A, lane 2). As an example, this analysis is shown for mutants T440C (Fig. 6A, lanes 4 -6), S640C (Fig. 6A, lanes 7-9), and N645C (Fig. 6A, lanes 10 -12) with three different incubation temperatures (25, 50, and 100°C). At an incubation temperature of 25°C, mutant S640C (Fig. 6A, lane 7) and Wt-SbsB-His (Fig. 6A, lane 1) showed one major protein band migrating at an apparent molecular mass of 80 kDa corresponding to an partly unfolded polypeptide chain. When the incubation temperature was changed to 50°C, S640C (Fig. 6A, lane 8) and Wt-SbsB-His (Fig. 6A, lane 2) exhibited two major protein bands migrating at apparent molecular masses of 80 and 100 kDa. At 100°C, the major protein band of S640C (Fig. 6A, lane 9) and Wt-SbsB-His (Fig. 6A, lane  3) migrated at 100 kDa. The similarities in the temperaturedependent migration of S640C and Wt-SbsB-His are consistent with the notion that the mutations in S640C do not destabilize the structural motif in SDS. In contrast, the major protein band of N645C runs at 100 kDa at 25°C and 50°C (Fig. 6A, lanes 10  and 11, respectively) indicating that this mutant is more readily unfolded in SDS than Wt-SbsB-His. T440C exhibited a 80-kDa band at 25°C and only one band migrating at 100 kDa at 50°C (Fig. 6A, lanes 4 and 5, respectively). This suggests that T440C is less readily unfolded in SDS than N645C, but more readily unfolded than Wt-SbsB-His. In summary, 51 out of 75 mutants showed a behavior similar to Wt-SbsB-His. Six mutants had a strongly destabilized phenotype (only a 100-kDa band independent of the incubation temperature) and 18 mutations were moderately destabilized (80-kDa band at 25°C and a 100-kDa band at 50°C) (Table I).
Although the assay of the thermal stability of cysteine mu- tant monomers indicates that some mutations affect the conformational stability in SDS, all except three mutants retained their ability to assemble into crystalline arrays under native conditions (see above). Mutant T816C was both compromised in assembly and destabilized in SDS. However, two of the three mutants compromised in assembly did not have a destabilized phenotype in SDS (T240C and G724C). Therefore, a destabilized phenotype is not required to interfere with self-assembly. The increasing incubation temperature is indicated by a triangle, with 25°C at the left, 50°C in the middle, and 100°C at the right end of the triangle. The bands of faster migrating, partly unfolded, and slower migrating, unfolded SbsB are indicated by arrows. B and C, PEG-OPSS modification of in vitro expressed SbsB cysteine mutants to determine the surface accessibility of the cysteine sulfhydryls in B, the SbsB monomer, and in C, SbsB incorporated into self-assembly products of Wt-SbsB from B. stearothermophilus PV72/p2 bound to peptidoglycan-containing sacculi of B. stearothermophilus PV72/p2. In vitro expressed cysteine mutants T683C, T617C, S580C, and S855C (either monomeric or incorporated into self-assembly products were mixed with the sulfhydryl-reactive reagent PEG-OPSS and incubated for 10 s (lanes 1, 4, 7, and 10) or 20 s (lanes 2, 5, 8, and 11) at 0°C before the reaction was terminated by the addition of N-ethylmaleimide. The extent of modification of unfolded protein was obtained by boiling a reaction mixture for 5 min in the presence of SDS before the addition of NEM (lanes 3, 6, 9, and 12). in vitro expressed [ 35 S]methionine-labeled SbsB cysteine mutants (two independent isolates of each) were subjected to the sulfhydryl-specific PEG-OPSS modification. The modification reagent PEG-OPSS has a molecular mass of approximately 5 kDa and is therefore expected to react preferentially with sterically unhindered, surface-located cysteines, whereas the reaction with residues buried within the SbsB protein would be restricted due to steric exclusion. PEGylation of SbsB results in an increase of the molecular mass, which can be monitored by SDS-PAGE. Interestingly, modification with PEG-OPSS with a molecular mass of 5 kDa resulted in a gel shift corresponding to an increase of approximately 15 kDa in apparent molecular mass. An example of this type of analysis is displayed in Fig. 6B for mutants T683C (Fig. 6B, lanes 1-3), T617C (Fig. 6B, lanes  4 -6), S580C (Fig. 6B, lanes 7-9), and S855C (Fig. 6B, lanes  10 -12). The extent of modification was determined for reaction times of 10 s (Fig. 6B, lanes 1, 4, 7, and 10) and 20 s (Fig. 6B,  lanes 2, 5, 8, and 11) under nondenaturing conditions at 0°C, and for 5 min in the presence of SDS at 100°C (Fig. 6B, lanes  3, 6, 9, and 12). The reactions for 10 and 20 s were performed at 0°C to detect differences in the rate of modification; the reaction at room temperature results in almost instant modification (data not shown). The modification reaction was stopped by the addition of sulfhydryl-reactive NEM in 56 molar excess over PEG-OPSS. When mutant T683C was allowed to react for 10 s (Fig. 6B, lane 1) and 20 s (Fig. 6B, lane 2) with the reagent, the major protein band migrated at 100 kDa, showing that no modification had occurred. However, under denaturing conditions the major gel band was upshifted (Fig. 6B, lane 3) due to the covalent modification with PEG. The modification pattern for mutant T683C under native and denaturing conditions suggests that the cysteine residue at position 683 is not surface accessible in the native state. In contrast, mutants S580C and S855C exhibited both an unmodified and an upshifted protein band after the 10-s reaction time under native conditions (Fig. 6B, lanes 7 and 10, respectively). After 20 s, the extent of modification was increased (Fig. 6B, lanes 8 and 11). Under denaturing conditions, mutants S580C and S855C became completely modified and migrated as upshifted band (Fig.  6B, lanes 9 and 12). The patterns of modification indicate that cysteine residues at positions S580C and S855C are surface accessible. Mutant T617C exhibited a reactivity pattern, which was consistent with an intermediate surface accessibility (Fig.  6B, lanes 4 -6). In summary, 34 mutants were not surface accessible, 23 were very accessible, and 18 were of intermediate surface accessibility as probed by the modification with PEG-OPSS (Table I). The upshift of PEGylated SbsB could be reversed by the addition of DTT, which cleaves the disulfide bridge between the protein and PEG (data not shown). This proved that the SbsB mutants were specifically PEGylated at cysteine.
The assembled mutant T683C remained unmodified after 10 and 20 s (Fig. 6C, lanes 1 and 2, respectively), but could be modified under denaturing conditions (Fig. 6C, lane 3). This suggests that cysteine in position 683 is not surface accessible in the assembled form; the cysteine in monomeric T683C was not surface accessible either (Fig. 6B, lanes 1-3). Interestingly, the assembled mutant T683C could not be completely modified under denaturing conditions (Fig. 6C, lane 3), which is in contrast to the complete modification of T683C monomers (Fig.  6B, lane 3). The incomplete modification of the assembled mutant might indicate that irreversible oxidation of the sulfhydryl group of cysteine occurred in the course of the 2-day-long dialysis of the samples, even though dialysis was performed under reducing conditions and DTT was added prior to modification. Incomplete modification under denaturing conditions was also observed for all other assembled mutants (e.g. Fig. 6C, lanes 3,  6, 9, and 12). Therefore, the extent of modification under native conditions was normalized with respect to the extent of modification obtained under denaturing conditions. Assembled mutant S580C showed modification after 10 s (Fig. 6C, lane 7) and reached maximal modification after 20 s (Fig. 6C, compare lane 8 with lane 9). This indicates that the cysteine at position 580, which was very surface accessible in the monomer (Fig. 6B,  lanes 7-9), retains its high surface accessibility when assembled into S-layers. In contrast, the cysteine in mutant S855C, which was very surface accessible in the monomer (Fig. 6B, lanes 10 -12) was not modified in the assembled form (Fig. 6C,  lanes 10 -12). Similarly, the cysteine residue in mutant T617C was occluded in the self-assembly product (Fig. 6C, lanes 4 -6), while it exhibited intermediate surface accessibility in the monomeric form (Fig. 6B, lanes 4 -6). In summary, 57 assembled mutants were not surface accessible, six were very surface accessible, and 12 were of intermediate surface accessibility as probed by the modification with PEG-OPSS (Table I). DISCUSSION To better understand the molecular structure and assembly of S-layer proteins, it is important to know which residues are located (i) at the external surface of the S-layer lattice, (ii) at the interface between the subunits, and (iii) within the pores (Fig. 1). In the present study, the differential accessibility of monomeric and assembled SbsB, the S-layer protein of B. stearothermophilus PV72/p2, was investigated. Seventy-five single-cysteine mutants were generated and analyzed by targeted chemical modification to probe the surface accessibility of the cysteine residues. Seventy-four mutations were evenly dispersed throughout the primary sequence between amino acid positions 234 and 913 of the 920-residue SbsB polypeptide (Fig.  7A, black bars). The N-terminal 200 amino acids were not included in the scanning mutagenesis, because this portion of SbsB can be deleted without affecting the formation of selfassembly products 2 (Fig. 7B).
To test whether point mutations to cysteine affect the conformation of SbsB, we first examined the formation of selfassembly products by in vivo expressed protein. Electron microscopic examination revealed that mutants T240C, G724C, and T816C had a greatly reduced ability to form regular lattices under conditions where the other mutants self-assembled. The positions of these three mutations are indicated in Fig. 7A (blue bars).
Next, we probed the thermal stability of an SDS-refractile structural motif in in vitro expressed monomeric protein, which causes SbsB to migrate at an apparent molecular mass of 80 kDa upon SDS-PAGE (Fig. 6A). The unidentified structural motif is most likely a ␤-sheet, since other ␤-sheet-rich proteins often exhibit a similar behavior (38 -40). Apparently, the structural motif includes the C terminus of SbsB, as C-terminal truncation mutants 2 and prematurely terminated in vitro expressed SbsB (Fig. 6A, lanes 1-3) do not exhibit temperaturedependent gel migration. In summary, of 75 mutants analyzed, 51 showed a behavior similar to Wt-SbsB-His (Table I and Fig.  7A). Six mutants had a strongly destabilized phenotype (only a 100-kDa band independent of the incubation temperature) and 18 mutants were moderately destabilized (80-kDa band at 25°C and a 100-kDa band at 50°C). All except one (T816C) destabilized mutants formed self-assembly products, indicating that the mutants folded correctly but were unfolded in the presence of SDS. Correlation of the observed phenotype with the severity of the mutation (conservative or non-conservative mutations; single, double, or triple mutations) revealed a bias toward the destabilized phenotype when two or more amino acids were changed. Among triple mutations 17% (3 out of 18) were strongly destabilizing and 39% (7/18) moderately destabilizing, as compared with 6% (3/48) and 21% (10/48) of double mutations, and 0% and 11% (1/9) of single mutations. The positions of the different classes of mutations were, however, scattered throughout the primary sequence of SbsB (Table I and Fig. 7A). This allowed us to consider structure-destabilizing mutations with regard to their position in the primary sequence. All six mutants with a strongly decreased stability in SDS were clustered toward the C terminus of SbsB: N645C, G769C, G796C, T816C, E837C, and S846C. Moreover, the remaining destabilized mutants were mainly clustered into two groups positioned from amino acid 408 to 486, and at the C terminus from amino acid 762 to 827 (Fig. 7A, small green bars). In summary, assembly-compromised and destabilizing mutations are clustered toward the C terminus (positions 580 to 846) and between residues 408 and 486 of SbsB. Since the SDS-refractile structural motif involves the far C terminus, it can be inferred that mutations with a destabilizing effect on folding interact directly or indirectly with the C terminus. Hence, we propose that the sequence from residue 580 to 846 forms a structural domain including the C terminus. We suggest further that the sequence between amino acids 408 and 486 is either part of or interacts with the C-terminal domain. With the exception of mutants E837C and G796C, the most strongly destabilizing residues (Table I and Fig. 7A, large green bars) are not surface accessible (Table I) and therefore most likely interact with other residues at sites buried within the C-terminal domain.
To identify residues located at the surface of the SbsB protein, we probed the reactivity of cysteines in SbsB monomers with a hydrophilic PEG conjugate (M r 5,000). Due to its size and hydrophilicity, the reagent most likely reacts preferentially with solvent-exposed residues of the protein, while the reactivity of buried residues is slowed down or strongly diminished. The attachment of PEG to SbsB results in an increase in molecular mass, which can be readily monitored by electrophoresis. In initial studies, we used a small (M r 500) charged molecule (19,21), but the gel shift of the modified SbsB protein was too small to observe by SDS-PAGE. Surprisingly, in SDS- FIG. 7. Graphical summary of the most important features of single-cysteine substitution mutants of SbsB correlated with other structural information. A, chart displaying the results of the analysis of in vivo and in vitro expressed SbsB cysteine mutants. Key: Blue bars, reduced ability to form assembly products as assayed by electron microscopy. Small and large green bars, respectively, moderately and strongly reduced stability of in vitro expressed SbsB monomers in SDS. Small and large red bars, respectively, very high surface accessibility of cysteine residues in monomers and in self-assembly products bound to cell wall sacculi. Black bars, cysteine mutants analyzed in this work. Small bars, single mutation; medium bar, double mutation; large bars, triple mutation. The positions of the cysteine residues refer to their positions in the unprocessed SbsB polypeptide carrying the signal sequence. B, effect of N-and C-terminal deletions of recombinantly expressed SbsB on the formation of self-assembly products as assayed by electron microscopy. 2 C, schematic drawing of SbsB, showing the location of the S-layer homology (SLH) domain, and the peptidoglycan and secondary cell wall polymer (SCWP)-binding domains. The SLH domains were found by sequence comparison (11), and the cell wall-binding domains were derived from proteolysis protection assays, and binding assays with proteolytically and genetically truncated fragments of SbsB. (Amino acid positions are given in parentheses and refer to the preprotein polypeptide sequence.) Adapted from Refs. 10 and 25. polyacrylamide gels, the PEG-SbsB conjugates migrated with an apparent molecular mass of 115-120 kDa, which is higher than the sum of the molecular masses of the protein and the polymer. A strongly decreased mobility in SDS gels was also reported for other proteins modified with PEG (41)(42)(43). Out of 75 mutants tested 23 (31%) were very accessible to the modification and 18 (24%) moderately accessible in the monomer. The proportion of at least partly exposed residues (55%) is very high, compared with the predicted fraction (33%) of at least partly exposed residues (44,45). The high number of surface accessible residues is, however, not surprising as 79% of the residues mutated to cysteine were polar or charged (28,29).
To identify residues positioned on the outer surface of the S-layer lattice and/or within the lumen of the pores (Fig. 1), we also probed the accessibility of the same 75 cysteines in assembled SbsB lattices, which were bound to cell wall sacculi through their inner surfaces. A PEG polymer of 5 kDa has a hydrodynamic diameter of approximately 5.2 nm assuming a spherical shape (46). Given its size, it is likely, that PEG-OPSS is occluded from the pores of the S-layer lattice. Two-dimensional image-reconstruction using transmission electron micrographs of WT-SbsB lattices reveals several depressions, interpreted as pores, ranging in diameter from approximately 2.5 to 6 nm (26). It is therefore likely, that PEG-OPSS is primarily reactive toward residues located at the outer surface, while residues lining the pores of the S-layer lattice are, except for the largest pores, not reactive. Residues positioned at the inner surface of the S-layer lattice (Fig. 1) are not accessible, as it is very unlikely that OPSS-PEG can permeate the cell-wall sacculi to gain access to cysteine residues. The hydrodynamic diameter of a PEG molecule of 5 kDa (5.2 nm) (46) is higher than the exclusion threshold for isolated cell walls from B. megaterium (2.2 nm) (46), and B. subtilis (4.2 nm) and E. coli (4.1 nm) (47). While the determination of the thresholds was based on the uptake and permeation of test molecules (e.g. PEG and dextran) over the course of up to 2 h, it was shown that in the first 3 min after mixing test molecules the size of the cell wall pores poorly permeate the cell wall (47). It is therefore very unlikely that OPSS-PEG penetrated the peptidoglycan meshwork during modification experiments, which were performed for a maximum of 20 s at 0°C. Upon incorporation of SbsB monomers into the S-layer lattice, the number of accessible residues was reduced. In summary, 57 mutants were not surface accessible, six very surface accessible and 12 of intermediate surface accessibility. While the total number of very and moderately accessible residues was decreased by 56% from 41 to 18, the number of very accessible residues dropped by 74%. This drastic reduction is plausible given that (i) the inner surface of the S-layer lattice must be occluded by the cell wall and that (ii) part of the monomer surface is shielded at the subunit-subunit interface in the lattice. A decreased reactivity of cysteines might also be a result of the formation of disulfide bonds at domain-domain interfaces. However, this is not likely since the reducing agent DTT was added prior to modification with PEG-OPSS. Furthermore, no protein bands corresponding to SbsB dimers were observed upon SDS-PAGE and autoradiography of the modification mixtures. (No reducing agents were used for the SDS-PAGE.) Among the six very accessible residues in the lattice, five were accessible in the monomer too (T433C, S480C, S580C, D781C, and E837C), while one was of intermediate accessibility in the monomer (S651C). The increased reactivity of residue S651C upon incorporation into S-layers is significant, since it argues against the possibility that the presence of cell wall sacculi causes a general reduction in the reactivity of the protein. Residues, which were very surface accessible in the SbsB monomer were clustered into three loosely defined regions (Fig.  7A, small red bars): cluster one, amino acid 240 -281; cluster two, amino acids 433-629; cluster three, amino acids 744 -913. It can be deduced that, in the native structure of SbsB, these sequences are located in regions with high surface to volume ratios. Conversely, sequences with no or very few surface accessible residues (281-433 and 629 -744) might belong to globular regions with low surface to volume ratios (28). The high surface to volume "domains" may correspond to the low-mass spikes and bridges in the reconstructed image of Wt-SbsB (26). To reinforce this interpretation, greater cysteine coverage or more independent clues would be needed.
In summary, the examination of 75 single-cysteine mutants confirms and extends the current domain model of SbsB. Two N-terminal domains (residues 31 to 212 and 271 to 362) bind SbsB to a secondary cell wall polymer and the peptidoglycan (10, 25) (Fig. 7C). In our studies, the location of the second binding domain is confirmed by the occlusion of surface accessible residues (residues 240 to 347) upon assembly of SbsB onto cell wall sacculi. The C terminus of SbsB (residues 580 to 846) forms a structural domain, which is likely to correspond to the massive morphological domain visible in computer averaged electron micrographs (26). Amino residues 408 to 486 most likely interact with or are part of the C-terminal domain.