Comparative and Functional Analysis of Sortase-Dependent Proteins in the Predicted Secretome of Lactobacillus salivarius UCC118

Bacterial strains and growth conditions.The strains and plasmids used in this study are listed in Table 1. L. salivarius subsp. salivarius strain UCC118 was cultured at 37°C under microaerophilic conditions (5% CO₂) in de Man-Rogosa-Sharpe (MRS) medium (16) (Oxoid Ltd., Basingstoke, Hampshire, United Kingdom). Lactococcus lactis MG1363 and Lactococcus lactis LL108 were used in this study as plasmid hosts and were cultured without shaking at 30°C in M17 broth (58) (Oxoid Ltd., Basingstoke, Hampshire, United Kingdom) containing 0.5% glucose. When necessary, erythromycin (Em) or chloramphenicol (Cm) was supplemented to a final concentration of 5 μg/ml for L. salivarius UCC118 and lactococcal strains.

DNA manipulations.The primers used in this study were purchased from MWG Biotech (Ebersberg, Germany) and are listed in Table S1 in the supplemental material. For cloning purposes, Pwo polymerase (Roche, Mannheim, Germany) was used for PCR amplification, while for screening purposes, Taq polymerase (Bioline, London, United Kingdom) was used. Restriction endonucleases, T4 DNA ligase, and PCR purification kits were purchased from Roche and used according to the manufacturer's recommendations. Ligation products were precipitated using Pellet Paint (Novagen, United Kingdom) prior to transformation. L. lactis LL108 was used as a host for pORI19 constructs. Plasmid DNA was isolated from L. lactis by using an Escherichia coli plasmid purification kit (QIAGEN, Crawley, United Kingdom) adapted for use with lactococci by the incorporation of 20 mg/ml lysozyme (Sigma, St. Louis, MO). Genomic DNA of L. salivarius UCC118 was isolated as previously described (23).

L. salivarius UCC118 was transformed by using a modified procedure based on that of Serror et al. (51). Briefly, MRS containing 1.9% (wt/vol) glycine was inoculated (1% [vol/vol]) from a freshly grown overnight culture. Cells were harvested at an optical density at 600 nm between 0.3 and 0.6 by centrifugation at 4°C (20 min at 3,000 × g). The pellet was washed twice with buffer (0.5 M sucrose, 7 mM potassium phosphate [pH 7.4], 1 mM MgCl₂) and resuspended in 1/100 the culture volume in the same buffer. Glycerol was added at 1/5 the culture volume and mixed, and 50-μl aliquots of cells were transformed with 1 to 5 μl of DNA (1 μg), using the following parameters: voltage, 1.5 kV; resistance, 400 Ω; and capacitance, 25 μF. Upon transformation, 1 ml recovery buffer (MRS containing 20 mM MgCl₂ and 2 mM CaCl₂) was added, and cells were incubated for 3 hours at 37°C (5% CO₂). Bacteria were plated on selective MRS-agar plates.

Construction of an isogenic sortase mutant.Genomic DNA of L. salivarius UCC118 was used as a template for PCR amplification of the 5′- and 3′-end-flanking regions of the sortase gene (LSL_1606), using primer pairs JP144-JP145 and JP146-JP147. The amplicons were joined by splicing by overlap extension (SOE)-PCR using the primer pair JP144-JP147. The resultant 1.6-kb amplicon was digested using BamHI and EcoRI and cloned into pORI19 digested with the same enzymes. The integrity of the obtained transformants was verified by PCR, using primers ORI47 (located on pORI19) and JP144. The resultant plasmid was named pLS001. Plasmid integrants in L. salivarius UCC118 were constructed as described previously (34), with minor modifications. Briefly, L. salivarius UCC118 containing pVE6007 was transformed with pLS001 and cultured for 24 h at 37°C (5% CO₂) with Em selection. Subsequently, cells were passaged for 50 generations at 42°C, with selection for pLS001 only, thus allowing integration into the chromosome upon loss of pVE6007. Colonies were screened for sensitivity to Cm in 96-well plates. Genomic DNAs were prepared from Em^r Cm^s cultures, and upstream and/or downstream integration was confirmed by PCR, using primer pairs JP166-ORI47 and JP167-ORI48B, respectively. Plasmid integrants upstream, downstream, and where single crossovers occurred both up- and downstream were selected and cultured at 37°C (5% CO₂) without antibiotic selection for at least 50 generations. Ninety-six colonies were randomly selected and screened for an Em^s phenotype. From 18 Em^s cultures, genomic DNA was prepared. The occurrence of a double-crossover event was confirmed for one Em^s culture by PCR amplification using the primer pair JP166-JP167, which flanks the sortase gene.

Disruption of SDP genes and lacZ by plasmid integration.The primer pairs JP082-JP083, JP090-JP091, JP190-JP191, and JP076-JP081 were used for PCR amplification of internal gene fragments of LSL_0311 (lspA), LSL_1085 (lspB), LSL_1838 (lspD), and LSL_0376 (lacZ), using genomic DNA of L. salivarius UCC118 as the template. PCR amplicons of the internal gene fragments of lspA, lspB, lspD, and lacZ were digested with EcoRI-HindIII, HindIII-EcoRI, BamHI-EcoRI, and BamHI-EcoRI, respectively, and cloned into pORI19, which had been treated with the same respective restriction endonucleases. These recombinant plasmids were designated pLS002, pLS003, pLS004, and pLS005, respectively. Single-crossover plasmid integrants were obtained as described above and confirmed by PCR, using primer pairs JP070-ORI47, JP065-ORI48B, JP164-ORI48B, and JP092-ORI48B for UCC118/pORI19::lspA, UCC118/pORI19::lspB, UCC118/pORI19::lspD, and UCC118/pORI19::lacZ, respectively.

Southern hybridization.Southern hybridization was performed using an ECL hybridization and detection kit (Amersham Biosciences, United Kingdom). Probes were identical to the PCR amplicons of the internal gene fragments described above. The double-crossover deletion was confirmed by using a probe which was complementary to the upstream sequence of the sortase gene. The probe was generated using the primer pair JP144-JP149.

Adhesion assay by viable count method.The colonic cell line C2/bbe1, a differentiated subclone of Caco-2 cells, and the adenocarcinogenic cell line HT29 were used to assess the adhesion abilities of the constructed mutants. C2/bbe1 cells were maintained in Dulbecco's modified Eagle's medium (Sigma, St. Louis, MO) supplemented with 10% (vol/vol) heat-inactivated (10 min at 70°C) bovine serum, nonessential amino acids, and 10 μg/ml human transferrin. HT29 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% (vol/vol) heat-inactivated (10 min at 70°C) bovine serum. Adhesion assays were performed essentially as described previously (28). Briefly, six-well plates were seeded with 1 × 10⁶ cells/well, and upon confluence, C2/bbe1 cells were maintained for 21 days or HT29 cells were maintained for 12 to 15 days to allow the cells to completely differentiate. Prior to the assay, monolayers were washed twice with phosphate-buffered saline (PBS). Bacterial overnight cultures were washed once with PBS, adjusted to an optical density at 600 nm of 1.0, and diluted 10-fold in PBS to reach ∼1 × 10⁸ CFU/ml, as determined by plate counts on MRS-agar. One milliliter of the bacterial suspension was added to the washed monolayers (bacterial cell/epithelial cell ratio of ∼50:1) and incubated for 30 min at 37°C (5% CO₂). Monolayers were washed five times with PBS to remove unbound bacteria. Adherent cells were removed by scraping, serially diluted in PBS, and plated on MRS-agar plates. Adhesion was expressed relative to that of the L. salivarius UCC118 wild-type strain. Adhesion assays were performed in duplicate in three independent experiments.

Adhesion assay by semiquantitative real-time PCR.Twenty-four-well plates were seeded with 2 × 10⁵ cells/well and maintained until the cells were fully differentiated, as described above. Monolayers and bacterial suspensions were prepared as described above, and 250 μl of bacterial suspension was added per well. Following incubation, washed monolayers were scraped off, resuspended in 250 μl PBS, and combined with 500 μl sterile distilled H₂O and 1 g 0.1-mm zirconia-silica beads (Biospec Products, Bartlesville, OK). Total lysis was achieved by bead beating for 1 min at maximum speed in a mini-bead beater (Stratech, United Kingdom). Cellular debris was pelleted by centrifugation for 2 min at 13,000 × g. Twenty microliters of supernatant was combined with 980 μl sterile distilled H₂O to dilute PCR inhibitors, and 1 μl was used as template DNA for real-time PCR. Real-time PCR was performed using the ABI7000 system (Applied Biosystems, Foster City, CA). PCR master mix was purchased from Biogene (Kimbolton, United Kingdom). A chromosomally located pseudogene (LSL_1319) was chosen as a target for amplification using the primer pair CC01-CC02 (see Table S1 in the supplemental material). A target was chosen from the chromosome instead of any of the resident plasmids to avoid complications that might occur from variations in plasmid copy number. Quantification of adherent bacteria was done by a standard curve method. Adhesion was performed in triplicate in three independent experiments.

Gene expression analysis.To examine the expression of the genes encoding seven identified sortase-dependent proteins in stationary-phase cells compared to that in logarithmic-phase cells, total RNA from cells in both growth phases was isolated using an RNA-easy kit (Ambion, Cambridgeshire, United Kingdom). The Improm-II reverse transcriptase enzyme (Promega, Madison, WI) was used to prepare cDNA according to the manufacturer's recommendations. Primers for real-time PCR were designed using the web-based tool Primer3 (50). For the three pseudogenes, primers were designed to target regions upstream and downstream of the internal stop codon (see Table S1 in the supplemental material). Real-time PCR was performed as described above. Gene expression levels were expressed relative to that of the 16S rRNA gene, as previously described (46). Gene expression analysis of both growth phases was investigated in three independent experiments.

Sequence analysis.The L. salivarius genome sequence (accession no. CP000233 [chromosome] and CP000234 [pMP118 megaplasmid]) has recently been determined (14). The genome sequences of L. plantarum WCFS1 (AL935263) (31), L. johnsonii NCC 533 (AE017198) (48), L. acidophilus NCFM (CP000033) (2), and L. sakei 23K (CR936503) (13) were also analyzed. Sequence analysis was performed with BLAST (3). Signal peptide prediction and cleavage site prediction were performed with SignalP3.0 (6). Transmembrane helices were predicted using the TMHMM server (32). The presence of LysM domains, peptidoglycan-binding domains, and choline-binding domains was determined by screening against the Pfam database (5), and results were filtered using an E value cutoff of <1 × 10⁻⁵. Lipoprotein predictions were performed as previously described (57). Sequences were searched for a WXL-binding domain by using the search string [LI]TW[TS]L, and the results were screened manually to determine the location of the motif within the sequence. Protein or DNA repeats were identified by using the programs Dotter (53) and RADAR (27). Sequences containing repetitive regions were screened manually for the presence of GW residues. Sortase substrates were identified by manual screening and a hidden Markov model (8).

Statistical analysis.Student's t test was employed to investigate statistical differences. Samples with P values of <0.05 were considered statistically different.

Identification of cell wall-anchored proteins.Using the annotated genome sequence of L. salivarius UCC118, a bioinformatic approach was employed to identify secreted proteins, including those predicted to be cell wall anchored. The results were compared to the data from parallel analyses of the genomes of L. plantarum WCFS1, L. acidophilus NCFM, L. johnsonii NCC 533, and L. sakei 23K (Table 2).

L. salivarius UCC118 possesses the second-largest genome (2.13 Mb) of the fully sequenced lactobacilli and is the only sequenced Lactobacillus strain harboring a megaplasmid (14). Using SignalP3.0, we identified 119 proteins predicted to be secreted, the majority (108) of which are encoded by the chromosome. Eight are encoded by the megaplasmid (pMP118), two are encoded by the 44-kb plasmid pSF118-44, and one is encoded by the 20-kb plasmid pSF118-20. Deduced products of an additional five pseudogenes were predicted to be secreted, with two encoded by the chromosome and three encoded by pMP118. Of the 119 proteins, 44 were predicted to be cleaved by signal peptidases I and 3 were predicted to be cleaved by signal peptidase II, and thus the majority of secreted proteins will remain associated with the cell membrane (Table 2). For other Lactobacillus genomes analyzed, the distributions of identity levels for existing database entries were similar, with a preponderance of proteins with values centrally distributed around identities of 30 to 60% (Table 2). The exception was L. johnsonii, for which a larger proportion of secreted and anchored proteins (see below) displayed significant identity to database entries.

We used a combination of manual inspection of proteins predicted to be secreted and the hidden Markov model of Boekhorst et al. (8) to identify sortase substrates. The hidden Markov model was also used to search the genomes of L. sakei and L. acidophilus, whereas sortase substrates for L. plantarum and L. johnsonii have been described previously (8). We thus identified 10 proteins containing sortase substrates in L. salivarius (Table 3), one of which is encoded by the previously characterized 44-kb plasmid pSF118-44 (22). Four of these proteins are encoded by pMP118, and five are encoded by chromosomal genes. Two SDPs are theoretical products of gene fragments, and four theoretical proteins were derived from pseudogenes caused by interruption with an internal stop codon or a frameshift.

One of the pseudogenes, designated LSL_0152, encodes a protein which shares 30% identity with the mucus-binding protein Mub of L. reuteri (49). LSL_0152 is interrupted by a stop codon (TGA) at nucleotide position 499. Another pseudogene, LSL_1319, shows 21% identity to the R28 protein of Streptococcus pyogenes (54), which is involved in binding to epithelial cells. The DNA sequence of LSL_1319 is interrupted by a stop codon (TAA) at nucleotide position 667. The third pseudogene, LSL_2020b, is located on pSF118-44. The encoded protein shares 25% identity with the collagen adhesin of Staphylococcus aureus (44). LSL_2020b is interrupted after 1,938 base pairs by the stop codon TAA. For these pseudogenes, there is no evidence of a second ribosome binding site with a start codon, which could lead to translation of the distal fragment of the gene. Whereas the other pseudogenes are disrupted by a stop codon, we identified a pseudogene (LSL_1774b) with a frameshift in its sequence which introduced a stop codon. The product of LSL_1774b is homologous (>32% identity) to a 1,480-amino-acid proteinase (PrtR) of the human isolate L. rhamnosus BGT10 (43). Apart from the pseudogenes, two gene fragments harboring a sortase recognition sequence were also identified (Table 3). Both gene fragments are located on pMP118. LSL_1832b is a 2.3-kb gene fragment whose derived amino acid sequence harbors an LPQMG sortase recognition motif. The fragment is homologous (>17% identity) to the C-terminal region of a 1,575-amino-acid salivary agglutinin-binding protein of Streptococcus gordonii (17). The smallest gene fragment harboring a sortase recognition motif is LSL_1902b (147 base pairs). It has no homology with proteins in the nonredundant BLAST database. Apart from the six interrupted/partial genes containing sortase recognition motifs, we identified four predicted sortase-anchored proteins which are intact, designated Lactobacillus surface proteins A, B, C, and D (LspA, LspB, LspC, and LspD, respectively) (Table 3).

LspA (LSL_0311) is a 1,209-amino-acid protein which contains seven repeats of 79 amino acids (R1 to R7) (Fig. 1). R1 and R7 are the least conserved repeats, sharing 73% identity, whereas R2 to R6 are more conserved, sharing >92% identity. Pfam analysis revealed that each of these repeats is similar to mucus-binding domains (PF06458), with E values ranging from 10⁻¹ to 10⁻⁶ but with all scores being above the gathering threshold. BLAST-NR searches did not reveal homology to a functionally characterized protein, since the closest homologue is a hypothetical protein of Streptococcus suis (ZP_00874951), as shown in Fig. 1. LspB (LSL_1085) is an 827-amino-acid protein (Fig. 1) containing an LPQMG cleavage motif. Three 13-amino-acid repeats were identified at the C-terminal end of the protein. The repeats are 100% identical, and Pfam analysis revealed no predicted function. The top BLAST hit for LspB is an enterococcal surface protein (Esp) of Enterococcus faecium (AAQ89938) which has no assigned function (Fig. 1). LspC (LSL_1335) is 785 amino acids in size and has four repeats of 97 amino acids (Fig. 1). There is over 98% identity among these repeats, and their sequences are similar to those of mucus-binding domains, as predicted by Pfam analysis, with E values ranging between 10⁻³ and 10⁻⁴ but with all scores being above the gathering threshold. It is homologous to the 3,269-amino-acid mucus-binding protein (Mub [AAF25576]) previously characterized in L. reuteri (49). This protein has two types of repeats. One set of repeats is divergent, with 15 to 85% identity, whereas the second set of repeats is conserved, displaying >91% identity. Both types have been shown to be involved in binding to mucin components (49). The four repeats of LspC show a higher sequence identity to the diverse repeats of Mub (13% identity), whereas there is very low identity (5% identity) to the conserved repeats of Mub. LspD (LSL_1838) is encoded by pMP118 and consists of 493 amino acids (Fig. 1). No repeats were identified, and the top BLAST hit is a hypothetical protein of the fungus Magnaporthe grisea (15.4% identity). Similar homology was noted for a sortase-dependent hypothetical protein of Streptococcus agalactiae (NP_735436). Since LspD is plasmid encoded, it is noteworthy that there is 15% homology to PrgA, a hypothetical surface exclusion protein of Enterococcus faecalis (45). Surface exclusion proteins block the conjugative transfer of plasmids to cells bearing identical or closely related plasmids (15).

The hidden Markov model was also used to search the genomes of L. sakei and L. acidophilus, but no additional sortase substrates were identified (see Table S5 in the supplemental material).

We identified three lipoprotein sequences in L. salivarius UCC118 (see Table S5 in the supplemental material). One of the proteins (LSL_0953) is a hypothetical protein with no clear function; it has homology to a phage-like protein but is not clustered in any of the four phage-associated regions (Sal1 through Sal4) in L. salivarius UCC118 (60). The second lipoprotein, LSL_0969, is also a hypothetical protein, whereas we annotated LSL_1445 as a glutamine-binding protein (see Table S5 in the supplemental material). Twenty-five lipoproteins were previously identified in L. plantarum (31), but when we applied a refined search string for the identification of lipoproteins in gram-positive bacteria, as described previously by Sutcliffe and Harrington (57), only three sequences were identified as lipoproteins. Five lipoproteins were identified in L. acidophilus, two were identified in L. sakei, and L. johnsonii appeared to have one lipoprotein (see Table S5 in the supplemental material).

We searched the Pfam database to identify proteins which can be anchored to the cell surface by choline-anchoring domains (25). Results were filtered using an E value cutoff of <1 × 10⁻⁵. No sequences in the screened Lactobacillus genomes were found to contain a choline-binding domain with an E value below the set cutoff value. Similarly, no proteins in L. salivarius UCC118 were found to contain a peptidoglycan- binding domain with a value below the cutoff. Based on the set cutoff, L. johnsonii and L. plantarum each have a single protein harboring a peptidoglycan-binding domain (see Table S5 in the supplemental material).

GW repeats have been shown to mediate binding to the cell envelope (38). Four proteins were identified in L. salivarius UCC118 that contain repetitive sequences containing GW residues (see Table S5 in the supplemental material). Two proteins with GW repeats are phage related: LSL_0295 is a protein with a hypothetical function and is part of Sal2, an inducible phage in L. salivarius UCC118 (60), and LSL_0783 was annotated as a phage terminase and was part of a phage remnant (60). LSL_0982 is predicted to encode a glycosyltransferase in exopolysaccharide gene cluster 1 (14). LSL_1266 was annotated as RNase BN. Comparison with the Cluster of Orthologous Genes database indicated that LSL_1266 is a membrane protein (see Table S5 in the supplemental material). Meanwhile, 11 proteins with GW repeats were identified in L. plantarum, 1 was identified in L. johnsonii, and 3 each were identified in L. acidophilus and L. sakei (see Table S5 in the supplemental material).

Proteins can be anchored to the cell envelope by LysM domains, which bind to the peptidoglycan in the bacterial cell wall (55). We identified nine proteins in L. salivarius UCC118 with such a domain (see Table S5 in the supplemental material). Two proteins, LSL_0304 and LSL_0805, are phage related and were both annotated as lysozyme. LSL_0304 belongs to Sal2, whereas LSL_0805 belongs to Sal1, a phage remnant (60). Three LysM-type proteins were annotated as hypothetical proteins (LSL_0090, LSL_0901, and LSL_1267), and three proteins were annotated as peptidoglycan binding proteins (LSL_1034, LSL_1036, and LSL_1371). One protein harboring a LysM domain has sequence similarity to a teichoic acid translocation ATP-binding protein (TagH; LSL_0373). The sequence identity with TagH is mainly at the N-terminal end of LSL_0373, whereas the LysM domain is located at the C-terminal end (see Table S5 in the supplemental material). Furthermore, we identified 11 proteins with a LysM domain in L. plantarum, 1 each in L. johnsonii and L. acidophilus, and 4 in L. sakei (see Table S5 in the supplemental material).

Kleerebezem and coworkers (31) identified a novel C-terminal WXL domain which they proposed could be a binding domain for the cell envelope, and 19 proteins containing this domain were identified in L. plantarum. Chaillou et al. (13) identified 15 proteins containing a YXXT(L/I)TW(T/S)L motif in L. sakei. However, when we applied this search motif, no proteins with this motif could be identified in L. sakei. The motif was modified to [LI]TW[TS]L, and this search returned nine proteins for L. sakei. No C-terminal WXL motif could be identified in the L. sakei proteins LSA0611 and LSA1731 (see Table S5 in the supplemental material). An [LI]TW[TS]L motif was identified in one protein in L. salivarius (LSL_1295) (see Table S5 in the supplemental material), which is homologous to a neopullulanase, whereas the motif was identified in six proteins of L. plantarum. No proteins with this motif were identified in L. acidophilus or L. johnsonii.

Construction of an isogenic sortase mutant.Previous studies targeting the sortase gene have shown that sortase-dependent proteins play a role in adhesion and virulence in a range of organisms (7, 9, 29, 30, 33). In order to investigate whether a sortase-dependent protein(s) in L. salivarius UCC118 is involved in adhesion, we constructed a mutant strain lacking the sortase gene (LSL_1606). The small size of sortase did not allow us to disrupt the gene by plasmid integration, and we therefore opted for a gene deletion, using a double-crossover strategy. Upstream and downstream flanking regions of 772 bp and 818 bp, respectively, were amplified. The upstream flanking amplicon includes the first 13 codons of the sortase gene, whereas the downstream flanking amplicon includes the last 3 codons. Both flanking amplicons were joined by SOE-PCR and cloned into pORI19. The resultant recombinant plasmid, pLS001, was transformed into L. salivarius UCC118 harboring pVE6007, and a double-crossover mutant was obtained as described in Materials and Methods. The deletion of the sortase gene in strain UCC118 was verified by Southern hybridization (Fig. 2). A PCR using wild-type genomic DNA with the primer pair JP144-JP149 resulted in a 1.1-kb amplicon which was used as a probe. Genomic DNAs of both the wild-type strain and the sortase mutant were digested with XhoI. The hybridization patterns showed bands of 5.8 kb and 5.1 kb for the wild-type and mutant strains, respectively (Fig. 2). An XbaI-XhoI double digest produced bands of 3.6 kb and 2.9 kb for the wild-type and mutant strains, respectively, confirming the deletion of sortase. The strain lacking the sortase gene was designated UCC118ΔsrtA.

The sortase mutant has reduced adhesion to epithelial cells.Following the construction of UCC118ΔsrtA, we tested the strain for adhesion to intestinal epithelial cells. UCC118ΔsrtA adhered significantly less to HT29 cells (P = 0.04) than the wild-type strain did (Fig. 3A). We also employed a semiquantitative real-time PCR method to validate the viable count method (Fig. 3B). The adhesion of the UCC118ΔsrtA mutant was also significantly reduced (P = 0.04) as measured by this method, at 61% of the level of the wild-type strain. The adhesion of UCC118ΔsrtA to Caco C2 cells was also reduced significantly (68%; P = 0.007) compared to that of the wild-type strain, as determined by real-time PCR, but this reduction was less than that observed for the sortase gene mutant grown on HT29 cells. Collectively, these data indicate that one or more sortase-dependent proteins are involved in adhesion to human epithelial cells.

Transcriptional analysis of sortase-dependent proteins.We employed endpoint reverse transcription-PCR (RT-PCR) to test if genes encoding sortase-dependent proteins in L. salivarius UCC118 were expressed in vitro when the strain was cultured in MRS broth. RNA was prepared from stationary-phase cells and reverse transcribed. For each target gene, internal primers were designed, and for the three pseudogenes, primers were designed upstream and downstream of the internal stop codon (see Table S1 in the supplemental material). After 50 cycles of PCR, no gene expression was detected for the chromosomally located gene lspC and the pseudogene LSL_2020b, which is carried on the previously described 44-kb plasmid pSF118-44 (22). The remainder of the sortase-dependent proteins were expressed (Fig. 4).

Previously, it was reported that the adhesion of L. salivarius UCC118 to HT29 cells is growth phase dependent (18), with a significant increase upon entry into stationary phase. We therefore investigated whether there was differential gene expression of sortase-dependent proteins in the two growth phases by performing real-time PCR, using RNA isolated from cells in the respective growth phases. All of the genes except the pseudogene LSL_1319 were transcribed at higher levels in stationary phase than in logarithmic phase (Table 4). The pseudogene LSL_0152 was transcribed at 2- to 2.5-fold higher levels in stationary phase than in logarithmic phase, while the transcription of lspB and lspD increased dramatically (Table 4). No gene expression was detected for lspC and LSL_2020b in logarithmic-phase cells (data not shown).

Role in adherence of LspA, LspB, and LspD.Transcriptional analysis showed that the expression of lspC was not detected after 50 cycles of PCR, and we therefore omitted this gene as a target for disruption. Internal gene fragments of lspA, lspB, and lspD were cloned into pORI19, and the resulting constructs were designated pLS002, pLS003, and pLS004, respectively. The genes for lspA, lspB, and lspD were thus disrupted by single-crossover plasmid integration, as described in Materials and Methods. Gene disruption was verified by PCR analysis, using genomic DNA as the template, with primers internal to the Em gene of the integrated plasmid and primers located upstream or downstream of the target gene (data not shown). Integrants of the individual target genes were analyzed by Southern hybridization to confirm the genomic arrangement (data not shown). The gene encoding LspD is located on the 242-kb megaplasmid pMP118 in L. salivarius UCC118, and it has been reported that this plasmid is present at four or five copies per cell (14). Southern hybridization confirmed the disruption of lspD in all copies of pMP118 (data not shown).

L. salivarius UCC118/pORI19::lspA adhered significantly less to HT29 cells (P = 0.001) than the wild-type strain did (Fig. 3A). A control lacZ mutant constructed by pORI19 integration adhered at 97.1% ± 4.4% of the level of the wild-type strain, showing that growth in Em alone was not responsible for lowered adhesion (data not shown). A reduction in adherence was also recorded for L. salivarius UCC118/pORI19::lspD (Fig. 3), but it was not significant (P = 0.23). No significant difference in adhesion was recorded for UCC118/pORI19::lspB from that of the wild-type strain (P = 0.56). These data were corroborated by the semiquantitative PCR assay (Fig. 3B), by which statistical significance was detected only for the adhesion reduction of the lspA knockout strain. Adhesion to Caco C2 cells was also significantly reduced for strain UCC118/pORI19::lspA (77%; P = 0.009) but was not significantly reduced for the lspB and lspD mutants (92% and 94%, respectively, as determined by real-time PCR [data not shown]).