| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Clinical Medicine and Dublin Molecular Medicine Centre, Trinity College Dublin, Trinity Centre for Health Sciences1 and Mercer's Institute for Research in Ageing2 , St James's Hospital, James's St., Dublin 8, Ireland
Correspondence
Déirdre Ní Eidhin
dniidhin{at}tcd.ie
Received 17 June 2005
Accepted 23 September 2005
Present address: St Columcille's Hospital, Loughlinstown, Co. Dublin, Ireland. 
The GenBank/EMBL/DDBJ accession number for the slpA and flanking sequences of C. difficile isolates are DQ060625–DQ060643.
Compositional data on SlpA from different PCR ribotypes are available as supplementary material in JMM Online.
 |
INTRODUCTION |
|---|
 TOP INTRODUCTION METHODSRESULTSDISCUSSIONREFERENCES |
|---|
C. difficile makes a crystalline protein surface layer (S-layer), a structural feature of many bacteria. S-layers have been ascribed various roles including nutrient uptake, exclusion of noxious substances, antiphagocytosis and colonization (Sára & Sleytr, 2000). In C. difficile, the S-layer is the predominant surface antigen, and a strong serum IgG response has been found among convalescent patients (Doyle, 2004; Pantosti et al., 1989). There are a number of variant types which are serologically distinct, substantially forming the basis for serotyping of C. difficile strains (Delmée et al., 1986, 1990).
The S-layer is encoded by the slpA gene, the product of which contains a cleavable signal sequence and is further cleaved to give two mature peptides which then associate to form the S-layer (Calabi et al., 2001; Karjalainen et al., 2001). The peptide derived from the N-terminal region of the precursor is smaller and more highly variable, whereas the peptide derived from the C-terminal region is relatively conserved. For convenience, the two products are respectively known as the low molecular weight (LMW) and high molecular weight (HMW) peptides. The LMW peptide appears to be the main serotyping antigen (Poxton et al., 1999). The HMW peptide has sequence similarity to the N-acetylmuramoyl-L-alanine amidase from Bacillus subtilis, and has been shown to possess amidase activity (Calabi et al., 2001). A number of other genes encoding putative amidases, known as slpA paralogues, occur in the vicinity of slpA (Calabi & Fairweather, 2002).
A recent study (Doyle, 2004) described the isolation and typing of C. difficile from a large number of patients who developed diarrhoea while attending the care of the elderly unit at St James's Hospital, Dublin. A total of 14 types were identified within this population by PCR ribotyping, which is based on polymorphism of the intergenic DNA between the regions encoding 16S and 23S rRNA (O'Neill et al., 1996). Since 116 PCR ribotypes of C. difficile have been identified to date (Stubbs et al., 1999), this typing method is considered more discriminatory than serotyping, which only distinguishes 21 types (Delmée et al., 1990). Our primary objective was to sequence the slpA gene from the PCR ribotypes identified and to compare the sequences with published data.
Calabi et al. (2001) reported a large ORF immediately downstream of slpA with strong sequence similarity to the secA gene of other bacterial species. The secA product is an essential component of the general secretory pathway in the Bacteria. It is a large protein which interacts with nascent proteins and with other components of the export pathway and provides energy for translocation by ATP hydrolysis (Schmidt & Kiser, 1999). This multifunctionality is reflected in a high degree of sequence conservation for secA between species. The sequence we obtained from each isolate included 315 bp from the 5' end of this gene. The consistent presence of a short segment of a conserved housekeeping gene in proximity to slpA provided reassurance that we had sequenced the genuine slpA allele in each case and not one of its many paralogues. We also constructed phylogenetic trees for segments of the translated slpA and secA genes and compared their variability between ribotypes.
|
METHODS |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Where possible, more than one isolate was selected, separated by date of isolation or ward or both. Three such well-separated isolates were selected for ribotype 001 (which accounted for approximately half of all isolates) and two each for ribotypes 002, 005 and 012. Single isolates were tested for the remaining ribotypes, either because only one isolate was available or because isolates occurred close together. Cultures were grown in anaerobic jars on Columbia blood agar (Lab M) with 7 % defibrinated horse blood, fastidious anaerobe broth (Lab M) or pre-reduced brain heart infusion broth containing 0·5 % (w/v) thioglycolate (BHI-TG). Stocks were maintained in cooked meat medium (Oxoid), made up in fastidious anaerobe broth.
|
DNA isolation, amplification and sequencing.
DNA was isolated from overnight BHI-TG-grown C. difficile cultures using the Gentra Puregene DNA isolation kit for yeast and Gram-positive bacteria with an additional proteinase K step (400 µg ml–1 added to the lytic buffer, incubation for 1 h at 55 °C followed by 10 min at 80 °C) and omission of RNase treatment. DNA was amplified by PCR using HotStar Taq polymerase (Qiagen), with an initial DNA denaturation step of 15 min at 95 °C followed by 30 cycles of denaturation for 1 min at 94 °C, annealing at 37 °C for 30 s and extension at 68 °C for 2 min 45 s. A final extension was done at 72 °C for 10 min. The primers used were (forward) –107F (5'-ATGGATTATTATAGAGATGTGAG-3') or –27F (5'-AATATAATGTTGGGAGG-3') and (reverse) +562R (5'-ACCTTCACCAGTTTTCAT-3'). Primer –27F was used to amplify DNA from ribotypes 002, 010, 014 and 066. Product purity, size and yield were checked on 0·8 % agarose gels using lambda DNA cut with EcoRI and BamHI or with EcoRI and HindIII as standard. DNA products were cloned in the pBAD ThioTOPO vector and transformed into competent Escherichia coli Top10, supplied as One Shot competent cells [genotype F– mcrA 
(mrr–hsdRMS–mcrBC) 
80lacZM15 lacX74 recA1 deoR araD139 (ara–leu)7697 galU galK rpsL (StrR) endA1 nupG], as recommended by the manufacturer (Invitrogen). Recombinants were selected on LB agar supplemented with ampicillin and checked by restriction digestion and plasmid DNA was isolated for sequencing. Sequencing was also carried out on PCR products directly. DNA was sequenced commercially at the Biochemistry Department, University of Cambridge. We designed the custom primers required to complete sequencing from both strands.
Analysis of slpA sequences. Signal sequences and their cleavage sites were predicted by the SignalP tools (http://www.cbs.dtu.dk/services/SignalP), revised most recently by Bendtsen et al. (2004), which combine two predictors based on neural network and hidden Markov model algorithms. The positions of both cleavage sites on the SlpA precursor protein were also predicted by comparison of deduced sequences with experimentally determined N-terminal amino acid sequences of the mature LMW and HMW peptides (Calabi et al., 2001; Cerquetti et al., 2000). Multiple sequence alignment was done using the CLUSTAL W tool (Thompson et al., 1994) with the Blosum matrix, a gap opening penalty of 10 and a gap extension penalty of 0·2. Secondary structure prediction was based on the consensus from the SOPM (Geourjon & Deléage, 1994), HNN (Guermeur, 1997), DPM (Deléage & Roux, 1987), DSC (King & Sternberg, 1996), GOR IV (Garnier et al., 1996), PHD (Rost & Sander, 1994), PREDATOR (Frishman & Argos, 1996) and SIMPA96 (Levin, 1997) tools, using the NPS interface (Combet et al., 2000). Internal peptide repeats were detected using the RADAR tool (Heger & Holm, 2000). Molecular mass and pI of predicted mature peptides were calculated by the Compute pI/Mw tool of the ExPASy proteomics server of the Swiss Institute of Bioinformatics (http://us.expasy.org/tools/pi_tool.html; Gasteiger et al., 2003). Codon usage was calculated by the CODONFREQUENCY tool from the Wisconsin Package (Accelrys Inc.) and values for relative synonymous codon usage were calculated. Relative synonymous codon usage is defined as the observed occurrence of a given codon divided by the expected occurrence. Values close to 1 are indicative of a lack of bias. Rho-independent terminators were detected by the TERMINATOR tool from the Wisconsin package (Brendel & Trifonov, 1984). To study evolutionary relationships between ribotypes, amino acid sequences were aligned using CLUSTAL W as before. Pairwise Poisson correction distances, which correct for multiple substitutions at the same site, were calculated from the resulting alignment and unrooted neighbour-joining trees were drawn from the resulting distance matrix using MEGA2 software (Kumar et al., 2001). Bootstrap analyses of the phylogeny were performed using 10 000 bootstrap replications.
|
RESULTS |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
). Forward primer –107F, based on non-coding sequence starting at position –107 from the slpA gene from strain 630 (ribotype 012), was not successful in amplifying sequences from all ribotypes. Forward primer –27F, used to amplify DNA from ribotypes 002, 010, 014 and 066, was based on a 17 nt stretch beginning 26–27 nt upstream of the slpA gene which was conserved among the remaining ribotypes. A single reverse primer, +562R, based on nt 298–315 of the secA homologue from strain 630 (Calabi et al., 2001), was used throughout. A single product was obtained from each isolate under the PCR conditions chosen. Products varied from 2496 to 2927 bp depending on ribotype (Fig. 1). Each fragment contained a complete ORF of 1830 to 2301 bp, presumed to be slpA, and terminated in a fragment containing 315 bp from the 5' end of another ORF, predicted to be a secA gene based on similarity to the allele in the genome sequence. The intergenic DNA between the two ORFs varied from 202 to 268 bp. The upstream DNA segment was 106–107 bp for products generated from primer –107F and 26–27 bp for products generated from primer –27F.
|
). These differences always translated to amino acid differences, often radical. Fig. 2(a) shows a CLUSTAL W alignment of the translated slpA ORFs from the different ribotypes, with residue differences between pairs of nearly identical peptides underlined. Thus ribotype pairs 005 and 054 differed by 3 nt and 2 amino acid residues and pairs 012 and 046/092, 014 and 066 and 031 and 094 each differed by 1 nt and 1 amino acid residue. From the 14 ribotypes, we thus identified nine classes which varied substantially in sequence.
|
|
), which showed complete alignment and strong conservation (over 62 % identical residues).
Comparison with other published slpA sequences
Sequence differences between slpA genes from the ribotype groups identified among the Dublin isolates are summarized in Table 2 and comparisons are made with published sequences. Many of the strains from which the slpA gene has been sequenced are known by serogroup only and, since there are more ribotypes than serogroups, a serogroup designation alone is often ambiguous. Moreover, not all ribotypes have been assigned to serogroups. Nonetheless, all published slpA sequences from ribotype 012 (or serogroup C) and 017 (or serogroup F) strains are identical to the corresponding sequences from our collection. Indeed, strong similarity was generally found between sequences from strains of a given serogroup. Serogroup A was an exception. This grouping is based primarily on a shared flagellar antigen, and subgroups are based on cross-reactivity of bacteria from which flagella have been removed mechanically (Delmée et al., 1990). It is therefore to be expected that slpA sequences from strains of different serogroup A subgroups may vary. Surprisingly, the sequence from a serogroup A10 strain showed strong similarity to the sequence from serogroup A1 strains. The strain in question, the only A10 strain for which any slpA sequence is available, is peculiar among serogroup A strains in that it has no flagella and does not react with anti-flagellin serum (Delmée et al., 1990). Also surprisingly, we found only 1 nt difference between slpA from ribotype 014 (serogroup H) and ribotype 066 (serogroup A9). We infer that limitations of the serogrouping method may result in discrepancies such as these.
We noted a close similarity between sequences from ribotypes 005, 016 and 054. Ribotype 005 has not been assigned to a serogroup, but its slpA sequence closely resembles that of a serogroup A1 strain (Karjalainen et al., 2001) and that of ribotype 054, which has been assigned to serogroup A1. No sequence strongly similar to the slpA from ribotype 002 was found among published slpA sequences.
Prediction of post-translational cleavage sites
The predicted N-terminal amino acid sequence was well conserved for SlpA and showed the hallmarks of a Gram-positive cleavable sequence (Fig. 2a; van Wely et al., 2001). For most ribotypes, both the neural network and hidden Markov models predicted the cleavage site just C-terminal of A24. This prediction concurs with experimental prediction based on N-terminal sequencing of the native SlpA LMW peptide for ribotypes 001, 012 and 017 and possibly 010, 014 and 066 (Calabi et al., 2001; Cerquetti et al., 2000). For ribotypes 002, 005 and 054, there were slight discrepancies between the predictors, with only the neural network predicting the cleavage site after position A24 for ribotype 002 and only the hidden Markov model making a similar prediction for ribotypes 005 and 054. This ambiguity may have been caused by the three consecutive alanine residues at positions 24–26 in these ribotypes (J. D. Bendtsen, personal communication), since alanine is strongly favoured in the position preceding the cleavage site. Position 24 occurs at the end of a highly conserved stretch of 7 amino acid residues with the consensus sequence SAAPVFA, which seems unlikely to be coincidental.
The position of the secondary cleavage site, to cleave the precursor SlpA into the LMW and HMW peptides, was predicted from comparison with published N-terminal amino acid sequence of the mature HMW peptide (Calabi et al., 2001; Cerquetti et al., 2000). For previously unpublished sequences, prediction was based on comparison with published sequences (Fig. 2a). Cleavage is generally predicted to occur N-terminal to an alanine or serine residue and C-terminal to a consensus motif TKS or TYX. Cleavage might actually occur some way upstream of this site, with some residues lost from the N terminus of the peptide during maturation. An absolutely conserved GKR motif occurs close to the predicted C termini of all the LMW peptides, which are otherwise generally dissimilar (Fig. 2a). The GKR motif is generally followed by a leucine residue, or occasionally valine, and is preceded by a conserved tyrosine at position –8, an aromatic residue at –3 and hydrophilic and hydrophobic residues at positions –7 and –4, respectively. The pattern appears in all other published sequences, although, presumably due to the heterogeneity of the LMW peptide, computer-generated alignments do not always recognize it (Calabi & Fairweather, 2002; Karjalainen et al., 2002).
Internal repeats in the HMW peptide
The RADAR program identified an 11 residue motif in the HMW peptide, beginning 40–50 residues from the predicted N terminus and repeated twice at approximately 100–120 residue intervals (Fig. 2a). Four positions were absolutely conserved in all three repeats and a fifth position contained strongly similar residues. Another motif, repeated once, was found N-terminal of the second and third repeats. Both types of repeat occurred in regions of the sequence which were strongly conserved between ribotypes, so that a given repeat showed more similarity between ribotypes than to other repeats within a ribotype. For example, the unvarying DR of the principal motif was followed by the dissimilar residues isoleucine, tyrosine and glutamine in repeats 1, 2 and 3, respectively.
Molecular mass and isoelectric point prediction for the mature peptides
Fig. 3 shows an SDS-PAGE gel of crude SlpA preparations from all ribotypes. SlpA is seen as two dominant bands for all but ribotypes 005 and 054, where a single strong band was observed at approximately 45 000. Calabi & Fairweather (2002) report a similar absence of the second band for strain 167, which is ribotype 016 (Neil Fairweather, personal communication) and has an almost identical slpA sequence to that of ribotype 054. As expected, migration profiles of SlpA peptides were identical for ribotypes with identical or nearly identical slpA sequences. Relative molecular mass (exclusive of glycosylation) and pI were calculated using ExPASy software (Supplementary Table S1 available in JMM Online). The SlpA peptides tended to migrate more slowly than predicted, probably due to the high content of acidic amino acid residues or to post-translational modification. The pI range for the LMW peptides was within the typical range (4–6) for bacterial surface layer proteins (Sleytr et al., 1999) and was consistently more acidic for the HMW peptide (4·46–4·69) than for the LMW peptide (4·83–5·09).
|
311 residues), contained numbers of asparagines close to the mean for all ribotypes, possibly indicating that a minimum number of these residues is essential to maintain the structural integrity of the protein.
Codon usage
Codon usage was analysed for the slpA genes of all ribotypes and for the slpA segments encoding predicted mature peptides. Codon usage patterns were compared with those of the same compilation of 148 sequences used for comparison of amino acid composition (Supplementary Fig. S2a). The species bias in favour of codons containing minimal G and C was generally stronger for the slpA gene, which is predicted to be highly expressed. Where codons permitted a choice of A or T in the wobble position, there was an occasional variation in preference. However, the combined frequencies of synonymous codons with a higher AT content generally well exceeded those of codons with a higher GC content.
A comparison of codon usage between the 5' and 3' regions of slpA, encoding the LMW and HMW peptides, respectively (Supplementary Fig. S2b), showed some differences in usage patterns for several amino acids, while maintaining the general bias in favour of A or T in the wobble position. Thus GCA was strongly favoured over GCT for alanine in the LMW peptide, with the reverse being true of the HMW peptide; and GTA strongly favoured over GTT for valine in the LMW peptide, with the reverse being true for the HMW peptide. Analogous differences occurred for serine and leucine, which have six codons each, while maintaining the AT-rich bias. Phenylalanine was an exception, with TTC preferred to TTT for phenylalanine in the HMW peptide, though not the LMW peptide.
Flanking DNA
In strain 630, slpA is flanked upstream by 213 nt of non-coding DNA preceded by an slpA paralogue and downstream by 247 nt of non-coding DNA followed by the secA homologue (Calabi et al., 2001). In our sequences, 106–107 nt of upstream sequence is available for most ribotypes and was found to be identical for ribotypes 012, 046, 092, 017 and 078, which differed by a single nucleotide from ribotypes 031 and 094 and showed greater divergence from ribotypes 001 and 005/054 (Fig. 4a). These ribotypes shared a 22 nt stretch a short distance upstream of the slpA start codon, from which it was possible to design a primer to amplify the relevant sequence from the remaining ribotypes. This shared sequence included a polypurine stretch at positions –9 to –16 (–8 to –15 for ribotypes 005 and 054), containing the motif GGGAGG, strongly suggestive of a ribosome-binding site (Shine–Dalgarno box) in composition and location. No rho-independent terminators were identified in any of the upstream sequences, either the 107 nt available from the fragments sequenced from 10 ribotypes or the 213 nt of intergenic DNA from strain 630 (ribotype 012) known from the Sanger sequencing project. It is tentatively inferred that transcription of the upstream slpA paralogue terminates by a rho-mediated process. Rho-dependent terminators are not readily identifiable in sequences (Henkin, 1996).
|
shows an alignment of the sequences. The slpA gene terminated in TAA for all ribotypes except 078, where it ended in TAG. The first half of the sequence was relatively unconserved, especially for ribotypes 010 and 078, and was considerably shorter in the latter. Sequences encoding potential mRNA stem and loop structures, reminiscent of rho-independent terminators, were found in this region. Rho-independent terminators show similarity between widely distributed bacterial genera (Vermat et al., 2002). Potential regions of dyad symmetry were 10–13 bp, with an unpaired loop of 3–4 nt, and typically contained three or four GC pairs, which would stabilize the structure of the transcript. Two such structures were found in all ribotypes except 010 and 078, which had one each. In rho-independent terminators, the region of dyad symmetry commonly gives way to a 3' run of non-pairing Ts, which is predicted to facilitate the release of the transcript, and the consensus sequence TCTG (Brendel & Trifonov, 1984). The terminators predicted for slpA were generally found to contain either the T-trail, the consensus sequence TATG/TGTG or occasionally both. A region of quite strong conservation was identified (59 % identity) from approximately nt –110 to nt –8 to –11 upstream of secA, ending in a very likely Shine–Dalgarno box for secA (AGGAGG).
Phylogenetic analysis
Evolutionary relationships between ribotypes based on the slpA genes were examined by aligning the HMW peptides using CLUSTAL W and constructing neighbour-joining trees (Fig. 5a). Alignment gaps were omitted from the phylogenetic analysis. The LMW peptide was not included in the analysis, as there is very little overall sequence conservation between ribotypes in this region and two of the ribotypes contain a much shorter peptide than the others. The same analysis was carried out with the available N-terminal 105 residues of sequence from the SecA homologue (Fig. 5b), so that variation in SlpA, presumed to be an antigen under evolutionary pressure to diversify, could be compared with variation in SecA, an essential protein which is not surface exposed.
|
|
DISCUSSION |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
We show a strong relatedness between the slpA sequences from ribotypes 005, 016 and 054, between ribotypes 012, 046 and 092, between ribotypes 014 and 066 and between ribotypes 031 and 094. It is useful to have this sequence information on slpA, which is strongly related to serogroup designation and complements ribotyping, which is based on non-coding DNA that is subject to different evolutionary pressure. Since there are more ribotypes than serogroups, it is not surprising that different ribotypes should have similar serogrouping antigens. However, the sequence data we have acquired indicate how very alike some slpA sequences from different ribotypes are, at both the nucleotide and amino acid sequence levels. This near-identity occurs alongside important strain differences, such as toxin production. Thus ribotype 031, which is deficient in both A and B toxins, has an slpA with one nucleotide difference compared with ribotype 094, which produces both toxins. Similarly, slpAs from ribotypes 066 (non-toxigenic) and 014 (toxin A+ and B+) differ by a single nucleotide. The sequence from ribotype 002 strains (equivalent serogroup A2 according to Stubbs et al., 1999) does not appear to resemble closely that published for any other strain.
An international effort is being made to coordinate typing methods in order to correlate the dominant types in outbreaks of C. difficile disease around the world. Serogrouping, which is largely based on differences between SlpA variants, is complicated by the existence of flagellated strains, since flagellar antigens may cross-react and production of flagella can vary with culture age and conditions. Serogroup A strains share serologically cross-reactive flagellins and subgroups were originally distinguished by SDS-PAGE profiles of major peptides, probably SlpA, in whole-cell preparations (Delmée et al., 1986). The role of flagellar antigens was appreciated later (Delmée et al., 1990) and it became possible to distinguish the A subgroups by slide-agglutination on removal of the flagella by sonicating the bacteria. However, it was also noted that strains of some other serogroups were flagellated and that cross-reactive flagellins occurred widely. Although it was claimed that non-serogroup-A strains had fewer flagella, and did not cross-react with group A strains in slide-agglutination, it is conceivable that mistakes could occur with heavily flagellated non-group-A strains.
Although some of the present work highlights the diversity of SlpA in different ribotypes and serogroups, some features appear consistently, notably the conserved stretch of sequence in the leader peptide, the GKRL/V motif and adjacent conserved residues in the vicinity of the secondary cleavage site and the five blocks of conserved sequence in the HMW peptide, interspersed with stretches of variable length and sequence. We also detected the DR-containing repeat motif, originally identified by Calabi & Fairweather (2002) in a comparison of SlpA molecules from six strains, in all of our sequences.
The GKRL/V motif is intriguing, given that GKR represents the longest stretch of absolutely conserved sequence in the LMW peptide and occurs near the precursor cleavage site. A considerable number of proteins from eukaryotes and viruses are cleaved post-translationally C-terminal to a pair of basic residues. These include mammalian growth factors and receptor proteins and glycoproteins from HIV and varicella zoster virus (reviewed by Seidah & Chrétien, 1999). Some bacterial toxins are activated by cleavage at pairs of basic residues by host proteases, but cleavage of bacterial proteins by bacterial proteases at such sites has not been widely reported. It is possible that the conservation of the site we have identified has more to do with binding of the relevant protease or transient association with the plasma membrane to facilitate cleavage at the distal sites already predicted experimentally.
There are several reports (Calabi et al., 2001; Cerquetti et al., 2000) that SlpA is glycosylated, probably more extensively on the HMW peptide, but the nature and extent of glycosylation are unknown. The widespread occurrence of glycosylation among bacterial proteins, including S-layers, has only recently been accepted and its potential role in virulence appreciated (reviewed by Schäffer & Messner, 2001; Schmidt et al., 2003; Spiro, 2002). Clostridial S-layers were among the first bacterial glycoproteins to be described (Sleytr & Thorne, 1976). In bacteria, glycopeptide linkages may be N-glycosyl on asparagine residues or, more often, O-glycosyl on serine, threonine and tyrosine residues. The sequence and structural requirements for glycosylation sites seem less well defined in prokaryotes than in eukaryotes, making their prediction from primary sequence virtually impossible, at least for the moment (Schäffer & Messner, 2001). Although it is known that the toxins mediate their activity by glycosylation of a critical threonine residue on the host Rho family of GTPases (reviewed by Spiro, 2002), at least two toxin-negative strains are reported to have a glycosylated SlpA (Cerquetti et al., 2000), implying the existence of a separate mechanism for SlpA glycosylation.
Some striking differences were noticed in amino acid composition between SlpA and the compilation published for C. difficile proteins. Some of these, e.g. the unusually high content of acidic amino acids and the absence of cysteine, are features of S-layers in general. There is a certain bias in favour of amino acids with smaller side-groups where these are functionally similar, i.e. aspartate is favoured over glutamate, asparagine over glutamine and alanine and valine over other hydrophobic amino acids, possibly because of the need for economy or possibly because of constraints imposed by the crystalline structure. Conversely, some amino acids, although present at a low level, may be critically important. Thus arginine is rare in SlpA, but where it occurs it is often in a relatively conserved region. Of the seven to ten arginines found in the HMW peptide, four positions are absolutely conserved and three of these are associated with the repeated DR motif. In the LMW peptide, which has two to five arginines, one is always found in the GKR motif.
S-layers composed of two peptides have been described for a few other organisms, but the SlpA from C. difficile is the only reported incidence of a two-component S-layer derived from a single precursor peptide. The reason for this phenomenon is not clear. Although it is known that the LMW and HMW peptides have different lattice structures (Cerquetti et al., 2000), detailed tertiary structure comparisons are not possible at this time. Secondary structure analyses (not shown) indicated that both peptides were composed predominantly of random coil (44–49 %) interspersed with alpha helical regions (28–34 %) and extended strand (19–22 %) and were not generally informative. We therefore compared the amino acid composition of the LMW and HMW peptides. This generally reflected the overall composition of the translated ORF, with the greatest difference noted for threonine, which was quite abundant in the LMW peptide at an average of just over 11 %, compared with 5·8 and 5·7 %, respectively, for the HMW peptide and the global average. A survey of published eubacterial S-layer sequences shows that a high threonine content (>9 %) is quite common, with the highest content (18·2 %) for Caulobacter crescentus, for reasons which have not been explained. The LMW peptide also had a higher content of aromatic amino acids, although SlpA had a rather low content of aromatic amino acids overall. We noted that the LMW peptide had two absolutely conserved tyrosines in close proximity (Fig. 2a), one absolutely conserved phenylalanine and one position which always contained either residue. It is tempting to suppose that the aromatic amino acids might be involved in binding of the LMW to a host ligand via a carbohydrate receptor or that they play an essential role in stabilizing the crystalline lattice structure.
Given the high rate of protein synthesis needed to maintain the integrity of the S-layer during growth and division (Sleytr & Messner, 1983), it is not surprising that codon usage in slpA reflects the strong AT-rich bias of C. difficile. The differences in codon usage between the LMW and HMW peptides (albeit within this bias) may reflect different evolutionary origins or perhaps an influence of DNA secondary structure.
A short stretch of presumably non-coding DNA upstream of slpA was fairly well conserved among 10 ribotypes, but differed sufficiently to prevent the use of the original left-hand primer in ribotypes 002, 010, 014 and 066. The intergenic region from slpA to secA varied substantially in sequence, particularly the moiety closest to the slpA gene. The fragment from ribotype 078, which had an upstream sequence similar or identical to those of nine other ribotypes, seems truncated in the region immediately downstream of slpA, lacking a predicted terminator and even employing a different termination codon. A recombination event in this region might explain why ribotype 078 is found on different clades in the bootstrap analyses of SlpA and SecA (Fig. 5). In ribotype 010, for which we have very little sequence upstream of slpA, the slpA–secA intergenic region, which is relatively long, also appears to encode a single terminator. This ribotype is found on a fairly strong clade with ribotypes 014 and 066 for SecA in the bootstrap analyses, but is not strongly associated with any other ribotype for SlpA. There is little variation in the DNA sequence either upstream or downstream from slpA in ribotypes 012, 046/092, 017, 031 and 094, and all of these are quite tightly linked on both trees. However, ribotype 001, which shows small to moderate differences in upstream and downstream flanking sequences, is part of this clade for SlpA, but not for SecA. A strain that acquires a novel surface layer protein by recombination, thereby generating a new strain, presumably has greater ability to evade a host's immune response. Toxin production, resulting in profuse diarrhoea, is an effective means of dispersing spores and spreading the disease. This may be the case with ribotype 001, which appears to have acquired a novel slpA gene by recombination with another clade and is responsible for approximately half of C. difficile infections in the UK (Brazier, 1998) and in St James's Hospital (Doyle, 2004).
A sequence similarity search of the published genome for C. difficile 630 has revealed the presence of a second secA allele located approximately 1 Mb from the gene flanking slpA and in the opposite orientation. Although uncommon, two secA alleles have been found in a number of genera, including Mycobacterium and Listeria, where the second SecA (SecA2) appears to have a specific role in the export of virulence-related proteins (Pallen et al., 2003). The product of the gene flanking slpA in C. difficile 630 (genome sequence) more closely resembles the SecA2 of other species, being the smaller of the two proteins and lacking a region predicted to interact with the chaperone SecB (Fekkes et al., 1997). Although it may play a subsidiary role, the predicted SecA2 does show strong similarity to its counterpart in other species, notably Listeria monocytogenes and Streptococcus parasanguinis (approx. 40 % identical residues), indicating a high degree of conservation between genera. Since C. difficile is not known to grow outside a human or mammalian host in nature, if secA2 is required for colonization, it is de facto an essential gene. The location of a gene for a major transport protein immediately downstream from slpA is unlikely to be coincidental, and it is known that S-layer expression in Aeromonas salmonicida depends on an ATP-dependent transport protein encoded by a downstream gene (Chu & Trust, 1993). Interestingly, in Listeria monocytogenes, SecA2 is believed to be responsible for the secretion of two autolysins, NamA and p60, and deletion of either secA2 or either of the structural genes leads to accelerated clearance of the organism from spleens and livers of infected mice (Lenz et al., 2003). The HMW peptide of C. difficile SlpA has N-acetylmuramoyl amidase activity, and slpA is located among a cluster of genes encoding similar domains (Calabi et al., 2001). Although C. difficile autolysins are predicted to be involved in wall remodelling during cell growth and division (Calabi et al., 2001), it is possible that they may have additional roles in virulence.
|
ACKNOWLEDGEMENTS |
|---|
|
REFERENCES |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
