Functional Identification of Conserved Residues Involved in Lactobacillus rhamnosus Strain GG Sortase Specificity and Pilus Biogenesis

Background: Lactobacillus rhamnosus GG produces pili assembled and anchored by two distinct sortases, SrtC and SrtA, respectively.

Results: Residue substitution within the triple glycine (TG) motif of pilin proteins impacts sortase-mediated polymerization.

Conclusion: The TG motif determines sortase specificity during pilus biogenesis in GG.

Significance: These results help explain the signaling involved in the assembly and anchoring of sortase-dependent pili.

Abstract

In Gram-positive bacteria, sortase-dependent pili mediate the adhesion of bacteria to host epithelial cells and play a pivotal role in colonization, host signaling, and biofilm formation. Lactobacillus rhamnosus strain GG, a well known probiotic bacterium, also displays on its cell surface mucus-binding pilus structures, along with other LPXTG surface proteins, which are processed by sortases upon specific recognition of a highly conserved LPXTG motif. Bioinformatic analysis of all predicted LPXTG proteins encoded by the L. rhamnosus GG genome revealed a remarkable conservation of glycine residues juxtaposed to the canonical LPXTG motif. Here, we investigated and defined the role of this so-called triple glycine (TG) motif in determining sortase specificity during the pilus assembly and anchoring. Mutagenesis of the TG motif resulted in a lack or an alteration of the L. rhamnosus GG pilus structures, indicating that the TG motif is critical in pilus assembly and that they govern the pilin-specific and housekeeping sortase specificity. This allowed us to propose a regulatory model of the L. rhamnosus GG pilus biogenesis. Remarkably, the TG motif was identified in multiple pilus gene clusters of other Gram-positive bacteria, suggesting that similar signaling mechanisms occur in other, mainly pathogenic, species.

Introduction

Pilus structures were first observed on the cell surface of the Gram-positive pathogen Corynebacterium renale using transmission electron microscopy (1). Various other Gram-positive bacteria displaying pili were subsequently described, such as Actinomyces, Clostridia, or Enterococci (2, 3). Until recently, it was acknowledged that pili were a phenotypic trait characteristic to pathogenic bacteria. However, the recent genome sequencing and analysis of Lactobacillus rhamnosus strain GG revealed the presence of a genomic island encoding three structural LPXTG-like pilin proteins called SpaC, SpaB, and SpaA and a pilin-specific sortase SrtC1 (4). L. rhamnosus GG² is globally marketed in probiotic products and has been among the best studied lactic acid bacteria (5–8). In GG, the discovery that mucus-binding pili signal the intestinal cells has generated new insights into the mechanisms involved in host interactions and underlines the potential contribution of pili in probiotic response (9). However, it is not clear what the signaling mechanisms and molecular components are that are involved in this response.

Pili represent a major colonization factor in terms of host interaction, signaling, and also biofilm formation (10, 11). Therefore, it is not surprising that such a phenotypic trait is found in a variety of bacteria colonizing mucosa-associated niches. Typically, the genes encoding the pilin subunits and its dedicated pilin-specific sortase(s) are organized in an operon-like cluster as observed in L. rhamnosus and other Gram-positive bacteria (4, 12). The frequent presence of flanking transposable elements indicates that some bacteria may have acquired pilus gene clusters by horizontal gene transfer events (12). In L. rhamnosus GG, it was even demonstrated that the transposable elements located upstream from the pilus gene cluster constituted a functional promoter and contributed to its expression (13). In Gram-positive bacteria, sortase-dependent pilus structures result from the multimerization of major backbone pilins to which minor pilin subunits are incorporated, i.e. tip and stop pilins (2). In some species, it was shown that the tip and the stop pilins play roles in host interaction and/or pilus assembly (14, 15).

The pilus biogenesis and anchoring relies on the central role of two functionally conserved sortases, i.e. pilus-specific sortase(s) and housekeeping sortase(s). The pilin subunits contain conserved domains that are required for secretion, anchoring, and pilus assembly (2). Typically, pilin precursors harbor a N-terminal peptide signal that allow secretion of the pilin subunits through the Sec-dependent secretion pathway (2). Following secretion and cleavage of its N-terminal peptide signal, the pilin subunit is retained to the cell membrane by its C-terminal hydrophobic domain (3). They are then covalently coupled to each other by isopeptide bonds that are catalyzed by pilin-specific sortases (12, 14, 16). Sortases specifically recognize the LPTXTG motif within the pilin subunits and cleave it between the threonine residue and the glycine residue (17). The resulting acyl-enzyme intermediate is a covalently bound pilin-sortase complex. Next, the nucleophilic attack of the lysine residue of the pilin on the cysteine residue of the sortase results in oligomerization of pilin subunits (12, 14, 17). Through a similar mechanism, the pilus is elongated and then coupled to the housekeeping sortase. The thioester bond linking the basal pilin subunit of the pilus to the housekeeping sortase is attacked by the amino group of the precursor of peptidoglycan (lipid II). This results in the covalent coupling of the elongated pilus to the bacterial cell wall (17). It is assumed that the termination and attachment of the pilus is initiated once the housekeeping sortase is relaying the pilin-specific sortase. In Corynebacterium diphtheriae, it has been proposed that this sortase switch is dependent on a specific pilin subunit SpaB (18). However, different scenarios also exist; for instance, Bacillus cereus is lacking a stop pilin (19). One central question concerning the pilus completion and this sortase switch still subsists: what are the signals within the pilins that determine the sortase specificity (housekeeping versus pilin-specific sortases) and that allow a concerted and coordinated assembly of the different pilin subunits?

The aim of the present study is to further investigate these regulatory and signaling mechanisms in the pilus biogenesis of L. rhamnosus GG. Over the last few years, the details of the pilus biosynthesis in L. rhamnosus GG have been uncovered (9, 20–23). Based on comparative analysis with other Gram-positive piliated bacteria, it became obvious that the structural organization and functionality of the pilus in L. rhamnosus GG is similar to a number of other bacterial species, such as Enterococcus faecalis and Enterococcus faecium (11, 24). However, there are various differences that may explain their distinctive roles in the human intestinal tract. Using the gene sequences of all predicted LPXTG proteins in L. rhamnosus GG, we identified the conservation of some glycine residues juxtaposed to the canonical LPXTG motif. To examine the role of these residues within what is now called the triple glycine (TG) motif, we constructed a series of pili gene cassette that were recombinantly expressed and analyzed by transmission electron microscopy and immunoblotting. The present study aims at gaining insights into the pilus biogenesis and sortase specificity in L. rhamnosus GG and other piliated Gram-positive bacteria, where the presence of the TG motif potentially reflects similar regulatory mechanisms in pilus biosynthesis.

EXPERIMENTAL PROCEDURES

Bioinformatics Analysis

The C-terminal domain of all predicted LPXTG proteins in L. rhamnosus GG were analyzed using the MEME suite (25). A nonexhaustive number of pilus gene clusters also described in the literature and/or present in sequence databases were retrieved manually, i.e. L. rhamnosus (4), Corynebacterium glutamicum (26), E. faecium (24, 27, 28), B. cereus (29, 30), Streptococcus suis (31–33), Streptococcus agalactiae (2, 34–37), Streptococcus pneumoniae (38–40), E. faecalis (41, 42) and C. diphtheriae (43–45). Their C-terminal domains were subsequently analyzed with the MEME suite (25).

Bacterial Strains and Plasmids

All of the bacterial strains and plasmids used in this study are listed in Table 1. L. rhamnosus strain GG was propagated in MRS broth anaerobically at 37 °C. Lactococcus lactis subsp. cremoris NZ9000 (46) was cultured anaerobically at 30 °C under static conditions in GM17 broth, i.e. M17 broth (Oxoid UK) supplemented with 0.5% (w/v) d-glucose (Sigma). L. lactis transformants were cultured in GM17 agar medium with 5 μg/ml chloramphenicol (Sigma).

TABLE 1.

Strains and plasmids used in this study

Cm^R, chloramphenicol resistance cassette.

Strains or plasmids	Relevant characteristics	Reference/source
Strains
L. rhamnosus GG	Type strain	Valio Ltd. culture collection
L. lactis NZ9000	MG1363 containing nisRK genes, srtA+	Ref. 46
L. lactis IL1403 ΔsrtA	Deletion of the housekeeping srtA gene in L. lactis IL1403	Ref. 53
Plasmids
pNZ44	L. lactis expression vector harboring the constitutive promoter P44, CmR	Ref. 47
pAC44	pNZ44 derivative encoding SpaA and SrtC1 as an operon	This study
pSpaA44	pNZ44 derivative encoding SpaA	This study
pQDV44	pAC44 derivative, LPQTGDTV motif	This study
pQGG44	pAC44 derivative, LPQTGGTG motif	This study
pD44	pAC44 derivative, LPHTGDTG motif	This study
pGV44	pAC44 derivative, LPHTGGTV motif	This study
pA44	pAC44 derivative, LPHTGATG motif	This study
pS44	pAC44 derivative, LPHTGSTG motif	This study
pT44	pAC44 derivative, LPHTGTTG motif	This study
pC44	pAC44 derivative, LPHTGCTG motif	This study
pV44	pAC44 derivative, LPHTGVTG motif	This study
pL44	pAC44 derivative, LPHTGLTG motif	This study
pI44	pAC44 derivative, LPHTGITG motif	This study
pM44	pAC44 derivative, LPHTGMTG motif	This study
pP44	pAC44 derivative, LPHTGPTG motif	This study
pF44	pAC44 derivative, LPHTGFTG motif	This study
pY44	pAC44 derivative, LPHTGYTG motif	This study
pW44	pAC44 derivative, LPHTGWTG motif	This study
pE44	pAC44 derivative, LPHTGETG motif	This study
pN44	pAC44 derivative, LPHTGNTG motif	This study
pQ44	pAC44 derivative, LPHTGQTG motif	This study
pH44	pAC44 derivative, LPHTGHTG motif	This study
pK44	pAC44 derivative, LPHTGKTG motif	This study
pR44	pAC44 derivative, LPHTGRTG motif	This study

Construction of a Pili Cassette in L. lactis

Oligonucleotide primers were obtained from Oligomer Oy (Helsinki, Finland). Genomic DNA from L. rhamnosus GG was extracted using a Wizard genomic DNA purification kit (Promega), as previously described (4). Plasmids used in the present study are listed in Table 1. Phusion Hot Start II high fidelity DNA polymerase (Thermo Fischer Scientific), FastDigest restriction enzymes (Thermo Fischer Scientific), and T4 DNA ligase (Promega) were used as recommended by the manufacturers' instructions. The pili cassette pAC44 expresses the L. rhamnosus GG genes spaA and srtC1 under the control of the constitutive promoter P44 (47). The plasmid pAC44 was obtained as follows. The DNA fragment containing spaA and srtC1 genes was amplified by PCR and flanked with the restriction sites NcoI and SpeI. The DNA insert was restricted by NcoI and SpeI for 30 min at 37 °C, and the digested insert was ligated to the similarly cut pNZ44 plasmid (47). The ligation mixture was then electroporated into L. lactis NZ9000, as previously described (48) and plated on GM17 agar plates with 5 μg/ml chloramphenicol. Transformants were screened by colony PCR and then confirmed by restriction mapping and sequencing. Following a similar strategy, the plasmid pSpaA44 expressing the spaA gene was generated. Some of the plasmid constructs were also electroporated into other L. lactis strain derivatives (Table 1), as described above.

Directed Mutagenesis

Directed mutagenesis was performed to substitute amino acid residues located in the TG motif of the spaA gene. Each variant was generated using a similar approach. Briefly, the whole pAC44 plasmid was amplified by PCR using back to back phosphorylated primers, harboring the mutation(s) to introduce. Self-ligation, electroporation, and subsequent screening were performed as detailed above.

Cell Wall Extract Preparation and Immunoblotting Analysis

L. rhamnosus and L. lactis cell walls were isolated following the same procedure as previously described (49) with some changes. Bacterial cultures were adjusted to an A_{600 nm} of 1.0, washed with PBS, and disrupted by bead beating (3 × 30 s) using sterile quartz beads (Merck). Cell lysates were resuspended in 500 μl of PBS and pelleted by centrifugation (10,000 × g) for 30 min. The cell wall-containing pellets were incubated for at least 3 h at 37 °C in a 50-μl solution containing 50 mm Tris-HCl, pH 8.0, 5 mm MgCl₂, 5 mm CaCl₂, 10 mg/ml lysozyme (Sigma), and 150 units/ml mutanolysin (Sigma). Cell wall extracts were then analyzed by immunoblotting using rabbit SpaA antisera as primary antibody, as previously described (13).

Transmission Electron Microscopy (TEM) Analysis

Stationary phase bacterial cells were immunolabeled using rabbit polyclonal anti-SpaA antibodies and fixed onto Formvar carbon-coated copper grids as previously described (21). Grids were examined using JEM-1400 transmission electron microscope (JEOL Ltd., Tokyo, Japan).

RESULTS

Sequence Analysis of the C-terminal Domain of All Predicted LPXTG Proteins in L. rhamnosus GG

The C-terminal regions of all 19 predicted LPXTG proteins in L. rhamnosus GG were retrieved and aligned (Fig. 1). Typically, these sequences showed a similar organization, i.e. a conserved LPXTG motif flanked by a positively charged hydrophobic domain. A closer look at the amino acid residues juxtaposing the canonical LPXTG motif revealed a high conservation of two glycine residues (positions +6 and +8; the residue numbering is based on the LPXTG motif, i.e. the leucine residue is on position +1). Specifically, the conserved glycine residues (positions +6 and +8) are present in SpaA, SpaC, SpaD, and SpaF pilin subunits, whereas the putative stop transfer pilins (SpaB and SpaE) and the other predicted LPXTG proteins have a conserved glutamic acid or aspartic acid residue (position + 6) (Fig. 1). The motif found in SpaA, SpaC, SpaD, and SpaF was termed triple glycine (TG) motif in contrast with the alternative motif called single glycine (SG) motif. SpaA and SpaC pilin subunits are assumed to be processed by the pilin-specific sortase SrtC1 during the elongation of the pilus structure in L. rhamnosus GG. SpaB putative stop pilin and other LPXTG proteins with the SG motif are predicted to be anchored to the peptidoglycan layer by the housekeeping sortase SrtA (21). We therefore hypothesized that secreted proteins with the TG motif may be preferentially recognized by pilin-specific sortases, allowing the multimerization of pilin subunits, whereas proteins with the SG motif would be recognized by the housekeeping sortase. Our observation parallels with the findings of Comfort and Clubb (50), who identified different LPXTG motifs associated with different sortase isoforms.

FIGURE 1.

Identification of two distinct LPXTG signature motifs among the 19 predicted LPXTG proteins of L. rhamnosus strain GG. A shows the SG motif found in 15 LPXTG proteins, including the minor pilin subunits SpaB and SpaE, and the TG motif present in the major and tip pilin subunits, i.e. SpaA, SpaC, SpaF, and SpaD. Asterisks indicate the conserved glycine residues. B shows the two pilus gene operon identified in L. rhamnosus GG (4). Respectively, predicted tip pilins, major backbone pilins, minor pilins, and sortases are indicated by blue, light blue, magenta, and green arrows. The SG (magenta) or TG (blue) motif is also indicated for each gene, and residues of interest are underlined.

Sequence Analysis of the C-terminal Domain of the Pilin Subunits in Other Piliated Gram-positive Bacteria

Further analysis of pilin genes found in eight other Gram-positive species showed that the TG and SG motifs were consistently found within all pilin subunits, although the residues within the SG motif showed a slightly higher diversity than in the TG motif (Fig. 2 and Table 2). Interestingly, as observed in L. rhamnosus GG, the TG motif was mostly found in predicted tip and major backbone pilins. The SG motif was associated with the predicted stop or base minor pilins. The genetic organization of the different pilus gene operons varies significantly, but the SG and TG motifs are mostly conserved, suggesting that similar signaling systems operate, possibly related to sortase recognition and specificity. In an effort to verify our hypothesis, mutagenesis analysis of the TG motif of the GG SpaA major backbone pilin was performed, providing a basis for comprehending the coordinated biosynthesis and assembly mechanism of pili in L. rhamnosus strain GG and other piliated Gram-positive bacteria.

FIGURE 2.

The SG and TG signature motifs among 51 pilin subunits proteins encoded in nine piliated Gram-positive bacterial species: L. rhamnosus, C. glutamicum, E. faecium, B. cereus, S. suis, S. agalactiae, S. pneumoniae, E. faecalis, and C. diphtheriae. A and B, show the SG signature motif (A) and the TG signature motif (B). Asterisks indicate the conserved glycine residues. C illustrates a selected number of the pilus gene operons analyzed that were previously identified in earlier studies, as referred in the main text. The schematics are not representative of the overall orientation of pili clusters within their respective genome. Predicted tip pilin, major backbone pilin, minor pilin, and sortase are indicated by blue, light blue, magenta, and green arrows, respectively. Gray squares correspond to uncharacterized open reading frames located within pili gene clusters. The SG (magenta) or TG (blue) motif is also indicated for each gene, and residues of interest are underlined.

TABLE 2.

Sequences of the LPXTG motif of different pilin genes analyzed in the present study

The LPXTG motif is indicated in red. The residues constituting the SG and TG motifs are highlighted in green and blue, respectively. Sequences were recovered from sequence databases, as described in the main text. In the table, pili gene clusters from S. pneumoniae 19A-6 and S. agalactiae A909 are not shown because of their high similarity with clusters found in S. pneumoniae TIGR4 and S. pneumoniae 2603V/R, respectively. In some clusters, pilin genes are predicted to be nonfunctional, i.e. pseudogenes as shown for example in S. suis P1/7 by Takamatsu et al. (31).

Construction of a Pilus Gene Cassette in L. lactis

Construction of a pilus gene cassette in L. lactis has been previously used as a model to express or study sortase-dependent pili in L. rhamnosus and Streptococcus pyogenes (23, 51). Using the same expression host, we constructed and cloned a minimal operon encoding the SpaA major backbone pilin and its dedicated pilin-specific sortase SrtC1. This operon was placed under the control of the constitutive promoter P44 and expressed in L. lactis subsp. cremoris strain NZ9000 (Figs. 3 and 4). Immunoblotting analysis and electron microscopy observations revealed that the operon was expressed at high level in L. lactis. The cells harbored long pilus structures distributed uniformly on the cell surface, whereas the wild type NZ9000 was pilus-less (Fig. 4). It is noteworthy that L. lactis NZ9000 encodes in its genome a housekeeping sortase SrtA. To exclude the possibility that L. lactis SrtA impacts on the SrtC1-mediated pilus polymerization reaction, the spaA gene only, without the srtC1 gene, was expressed in L. lactis NZ9000. This resulted in pilus-less lactococcal cells, where SpaA proteins were retained as monomers on the cell surface (Figs. 3 and 4). Because this demonstrates the suitability of our expression system, it was used in conjunction with the minimal pilus operon spaA-srtC1 for subsequent mutagenesis and functional analysis.

FIGURE 3.

Construction of a minimal spaA-srtC1 pilus gene operon in L. lactis NZ9000. A shows the operons constructed in the plasmid pNZ44 and introduced into L. lactis NZ9000. B shows the immunoblotting analysis of the cell wall extracts using anti-SpaA serum. Lane 1, NZ9000; lane 2, NZ9000 + pAC44 (coexpression of both spaA and srtC1); lane 3, for NZ9000 + pSpaA44 (expression of spaA only); HMWL, high molecular weight ladder.

FIGURE 4.

Immunogold staining and electron microscopy observations of SpaA pili in L. lactis NZ9000. The bacterial cells were labeled with anti-SpaA antibodies and gold particles (5 nm). A, L. lactis strain NZ9000 (control); B, L. lactis NZ9000 + pSpaA44. C and D, L. lactis NZ9000 + pAC44. Scale is indicated in each panel.

Mutagenesis of the TG Motif and Impact on Pilus Polymerization

The LPXTG motif constitutes a key signal for sortase recognition, allowing the polymerization or anchoring of proteins to the bacterial cell surface. As mentioned above, the sortase motif clearly differs between SpaA/C subunits (LPHTGGXG) and SpaB subunit (LPQTGDTV) (Fig. 1). In an effort to identify what amino acid residues are critical for SrtC1 recognition and pilus assembly, the residues at positions +3, +6, and +8 of the SpaA motif (LPHTGGXG) were replaced by the corresponding residues from the SpaB sortase recognition motif (Fig. 5). The replacement of the whole SpaA sortase recognition motif by the SpaB sortase recognition motif resulted in the lack of pilus structures in L. lactis. Only SpaA monomers or oligomers could be observed on the cell surface, indicating that the replacement of three residues (H3Q, G6D, and G8V) abolished the pilus formation. Subsequently, pilus operons with single substitutions at positions +3, +6, and +8 were constructed to identify what amino acids at these positions were critical for pilus polymerization (Fig. 5). The H3Q substitution had no impact on the pilus assembly. This is consistent with previous studies on the LPXTG motif showing that the residue (X) between the proline and the threonine is interchangeable (52). The G8V substitution had a limited effect, i.e. the presence of slightly shorter pili (Fig. 5). Remarkably, the G6D mutation abolished the pilus assembly completely. No pili could be observed on the cell wall, indicating that the glycine residue on position +6 is essential for pilus assembly and/or pilin-specific sortase recognition. In the L. rhamnosus GG genome, we observed that other predicted LPXTG proteins had a glutamic residue on that same position (+6). These proteins are assumingly anchored to the peptidoglycan layer by the housekeeping sortase SrtA. Indeed, the G6E substitution within the SpaA protein sequence resulted in a similar phenotype as with the G6D mutant, i.e. lack of pili. For both G6E and G6D mutants, transmission electron microscopy showed some SpaA oligomers that could also be observed by immunoblotting analysis (Figs. 5–7). However, no long pilus structures were observed in G6E and G6D mutants. Expanding on this, we introduced the different constructs into L. lactis IL1403 srtA-null mutant (53) to investigate whether the lack of the housekeeping sortase SrtA impacts the pilus phenotype. Four different constructs were introduced into the lactococcal srtA-null mutant and analyzed by TEM (Fig. 8). They all resulted in formation of pili as in contrast with L. lactis NZ9000 harboring SrtA, indicating that the formation of SpaA pili is independent from the housekeeping sortase SrtA. However, it remains unclear how pili were retained onto the cell wall, possibly as SrtC1-pili intermediates. Detached pili were also observed by electron microscopy (data not shown). One construct with the G6D substitution was associated with a different phenotype. Whereas in L. lactis NZ9000, which produces SrtA, the G6D substitution was preventing pilus assembly, this was not the case in L. lactis IL1403 srtA-null mutant (Fig. 8). In the absence of SrtA, SpaA variants with the G6D substitution were processed by SrtC1 and assembled into pili.

FIGURE 5.

Immunogold staining and electron microscopy observations of different L. lactis mutants. A and B, NZ9000 + pQGG44; C and D, NZ9000 + pD44; E and F, NZ9000 + pGV44; G and H, NZ9000 + pQDV44. The TG motif is also indicated in each panel. Scale is indicated in each panel.

FIGURE 6.

Electron microscopy analysis of L. lactis NZ9000 producing SpaA pilus derivatives (coexpression of spaA-srtC1), where the TG motif of SpaA was mutated by substitution of the glycine residue (position +6) by the 19 other amino acid residues. For each mutant, the pictures are labeled according to the corresponding TG motif variant. Scale is indicated in each panel.

FIGURE 7.

Immunoblotting analysis of L. lactis NZ9000 expressing different SpaA pili derivatives (coexpression of spaA-srtC1), where the residue (position +6) of the TG motif of SpaA has been replaced by all other amino acid residues. #, protein marker; X, mutant G6X, where X is the substituted amino acid; , various monomeric forms of SpaA, i.e. full-length SpaA (35.9 kDa), peptide signal cleaved SpaA (∼32 kDa), and peptide signal and LPXTG cleaved SpaA (∼30 kDa); HMWL, high molecular weight ladder. Arrows indicate SpaA n-mers based on an estimation of SpaA n-mers molecular masses. SpaA n-mers associated to the cell wall may therefore correspond to a mass n × ∼30 kDa. Additional post-translational modifications may occur and were not included in the present calculations. Samples were grouped according to their pilus phenotype and/or the properties of the substituted amino acid X within the TG motif (position +6).

FIGURE 8.

Immunogold-staining and electron microscopy observations of SpaA pili expressed in L. lactis IL1403 srtA-null strain. The bacterial cells were labeled with anti-SpaA antibodies and gold particles (5 nm). A and B, L. lactis strain IL1403 srtA-null + pAC44; C and D, L. lactis strain IL1403 srtA-null + pQGG44; E and F, L. lactis strain IL1403 srtA-null + pGV44; G and H, L. lactis strain IL1403 srtA-null + pD44.

Mutagenesis of the Glycine Residue +6 of the TG Motif

To further delineate the contribution of the TG motif, we introduced into L. lactis NZ9000 a spaA-srtC1 operon where residue +6 of the TG motif was substituted by all other possible amino acids. Subsequently, we characterized the resulting lactococcal strains by TEM and immunoblotting analysis (Figs. 6 and 7). As summarized in Table 3, in most cases, the replacement of the glycine residue (position +6) resulted in the loss of pili, and only SpaA monomers or oligomers could be observed on the cell wall. The immunoblotting analysis revealed a series of distinct bands that may represent different SpaA n-mers, as proposed on Fig. 7. Wild type pilus structures were typically detected as a set of slow migrating bands with increasing sizes (ladder pattern), where most SpaA proteins were localized in the higher molecular mass bands. Wild type-like pilus-producing phenotypes were observed only when the glycine residue of the TG motif was replaced by alanine and serine. The substitution with basic residues lysine and arginine resulted in formation of some shorter and sparse pili (Fig. 6) that were detected by immunoblotting analysis only when overexposing the membrane (data not shown). It is noteworthy that in two cases, i.e. G6M and G6Q substitution, a low proportion of bacterial cells harbored few and short pili and were therefore not detectable by immunoblotting analysis. For other mutants analyzed, the pilosotype was homogeneous within the overall population and did show any pili formation, as summarized in Table 3.

TABLE 3.

Pilus phenotype of the different SpaA pilus derivatives produced in L. lactis NZ9000

For the G6K and G6R mutants, some short and sparse pili were observed by electron microscopy but could be detected by immunoblotting only when overexposing the blotting membrane, suggesting that only a low proportion of the bacterial population displayed short pili. +, pili formation; −, no pili; +/−, scarce pili.

Amino acid properties (position 6)	TG/SG motif	Amino acid substitution (position 6) in other analyzed species	Pili formation	Predicted SpaA n-merization state
Small nonpolar side chain	LPHTGGQG	Yes	+	High molecular mass polymer
	LPHTGAQG	Yes	+	High molecular mass polymer
	LPQTGGQG	Yes	+	High molecular mass polymer
	LPHTGGQV	Yes	+	High molecular mass polymer
Charged side chain (acidic)	LPQTGDQV	Yes	−	Monomers
	LPHTGDQG	Yes	−	Oligomers, up to ∼220 kDa
	LPHTGEQG	Yes	−	Oligomers, up to ∼220 kDa
Polar side chain	LPHTGNQG	Yes	−	Oligomers, up to ∼220 kDa
	LPHTGQQG	Yes	−	Oligomers, up to ∼220 kDa
	LPHTGSQG	Yes	+	High molecular mass polymer
	LPHTGTQG	No	−	Oligomers, up to ∼90 kDa
	LPHTGCQG	No	−	Oligomers, up to ∼120 kDa
	LPHTGYQG	No	−	Oligomers, up to ∼90 kDa
Charged side chain (basic)	LPHTGKQG	No	+/−	Oligomers, scarce short pili SpaA Oligomers, scarce short pili
	LPHTGRQG	No	+/−	Oligomers, up to ∼120 kDa; SpaA monomers/dimers
	LPHTGHQG	No	−	Monomers/dimers
Hydrophobic side chain	LPHTGVQG	Yes	−	Monomers
	LPHTGLQG	No	−	Monomers/dimers
	LPHTGIQG	No	−	Monomers/dimers
	LPHTGMQG	Yes	−	Monomers/dimers
Conformationally special hydrophobic side chain	LPHTGPQG	No	−	Oligomers, up to ∼90 kDa
Aromatic (and hydrophobic) side chain	LPHTGFQG	No	−
	LPHTGWQG	No	−

DISCUSSION

The recent discovery of mucus-binding pili constituted a milestone in understanding how L. rhamnosus strain GG interacts with and signals its host (4, 9, 21, 23). The genetic organization of L. rhamnosus GG pilus cluster is quite similar to some other piliated Gram-positive bacteria, such as E. faecalis (11). Therefore, a deeper comprehension of the mechanisms involved in the pilus biogenesis would not only benefit our general understanding of L. rhamnosus GG colonization and persistence in the gut but also provide insights into pilus assembly occurring in other Gram-positive bacteria.

In L. rhamnosus GG, all predicted LPXTG proteins possess either the TG or SG motifs (Fig. 1), allowing us to hypothesize on the role of these two conserved motifs in sortase specificity and recognition. This expands on the work by Comfort and Clubb that identified prevalent LPXTG motif for each sortase isoform (50). To verify our hypothesis, we established a L. lactis model encoding a minimal pilus cassette and a series of relevant mutated derivatives. The coexpression of SpaA and SrtC1 resulted in SpaA pilus structures on the cell wall of L. lactis, indicating that these two proteins are sufficient to generate the pilus backbone. The single production of SpaA pilins in the presence of SrtA did not yield to any pili, confirming that the SrtA function is restricted to protein anchoring only. In L. rhamnosus GG, SpaB and SpaC pilin subunits are also incorporated into the backbone (21) but most likely involved in other biological functions (4). The tip pilin SpaC has a TG motif and the putative stop pilin SpaB a SG motif, suggesting that SpaC might be similarly recognized by SrtC1 as SpaA.

In our L. lactis pilus model, the substitution of the glycine residues (position +6) in the TG motif by acidic residues D or E (mimicking the SG motif) significantly altered pilus formation. Only SpaA oligomers were observed on the cell wall by TEM and immunoblotting analyses. Interestingly, in the absence of SrtA, the G6D SpaA pilins were polymerized by SrtC1. This showed that both G6D and G6E SpaA pilins may be preferentially recognized by the housekeeping SrtA but can still be processed by the pilin-specific SrtC1. This result may explain the lack of pili and the detection of G6D/G6E SpaA pilin aggregates/oligomers in the presence of both SrtA and SrtC1. In L. rhamnosus GG pilus biogenesis, SpaA and SpaC (TG motif) would be processed by SrtC1, whereas SpaB (SG motif) would be preferentially processed by SrtA, terminating the pilus polymerization upon incorporation of SpaB subunits into the pilus backbone (Fig. 9). The incorporation of SpaB within the backbone remains nevertheless possible, because both sortases appear to show certain substrate recognition flexibility. This is in agreement with previous observations where SpaB proteins were found in L. rhamnosus GG pilus backbone (21). Within the TG motif, the conservation of the three glycine residues clearly has an impact on the efficiency of the pilus polymerization reaction. In most cases, the substitution of the second glycine residues (position +6) by other amino acid groups prevented the pili assembly. Only SpaA pilin subunits with an altered TG motif consisting of an Ala or Ser residue could result in the production of wild type-like pilus structures.

FIGURE 9.

Proposed regulatory model of the pilin subunits and their dedicated sortases in L. rhamnosus strain GG.

The TG/SG motif was not only found in L. rhamnosus GG but also in a number of other Gram-positive bacteria, such as C. diphtheriae, E. faecalis, S. pneumoniae, or B. cereus (Fig. 2 and Table 2). The genetic organization of the different pilus gene clusters greatly varies. However, we consistently detected the TG motif to be present in the predicted tip and major backbone pilin, whereas the SG motif was mostly found in the predicted stop/minor pilin subunit. In most cases, the TG motif is highly conserved among these tip and major backbone pilins, indicating that this TG motif may be required for the recognition of the tip and major pilin by the pilin-specific sortase, as shown in L. rhamnosus GG. It is noteworthy that the TG motif still allows some flexibility. As demonstrated, the replacement of the glycine residue (position +6) by a serine did not impact on the pilus phenotype, which would explain why the SpaABC pili of C. diphtheriae NCTC 13129 is functional, although the SpaC pilin has the following TG motif: LPLTGSNG. Such a motif may be still recognized by the pilin-specific sortase. In contrast, the SG motif shows a higher diversity among species. Rather than being a truly conserved motif, the SG motif is a non-TG motif. The role of the TG motif in the concerted assembly of the pilin subunits is possibly based on the distinct affinity of the different sortase isoforms involved. As shown in the present study, one TG/SG motif is favored by one sortase isoform (SrtC/SrtA) but does not necessarily prevent its recognition by another. Our model may explain also the stoichiometry of the different pilins incorporated into the pilin backbone. In L. rhamnosus GG, the putative stop pilin SpaB has a SG motif, which is preferentially recognized by the housekeeping sortase SrtA. However, SpaB is found in a low proportion in the backbone (21), which may illustrate its weak but possible processing by the pilin-specific sortase SrtC1.

The remarkable conservation of the TG motifs in different pilus gene clusters from various Gram-positive bacterial species also suggests that similar regulatory pathways may intervene in terminating the pilus assembly by switching from pilin-specific sortases (polymerization) to housekeeping sortase (anchoring). In the present study, we illustrated the TG/SG motif through a number of 18 pilus gene clusters from nine species. However, it remains difficult to establish the extent of this motif across piliated Gram-positive species, and there are exceptions to this general pattern, as observed in some S. pyogenes strains for example. This does not necessarily exclude the possibility that a similar model occurs based on a distinct extended sortase recognition motif. The interaction between the pilin substrate, especially the TG/SG motif and the sortase, is functionally important but still remains unclear, because the sortase residues involved in substrate specificity have not been clearly identified. The identification of a flexible lid (DPF motif) located in the catalytic pocket of pilin-specific sortases was shown to be involved in regulating the catalytic activity of the enzyme (54–56) but may not be involved in the substrate specificity. In the present study, the pilin-specific sortase SrtC1 of L. rhamnosus GG does not possess this conserved lid motif as observed in some other pilin-specific sortases. The remarkable conservation of the TG motif within pilin subunits indicates that other sortase residues possibly located in the vicinity of the catalytic site may control the substrate specificity. One cannot exclude the possibility that this specificity is due to the presence of the three glycine residues of the TG motif. Such motif would ensure greater chain flexibility nearby the LPXTG motif, promoting a better access to the sortase catalytic pocket. In contrast, the presence of the SG motif, which makes the chain more rigid, may be preferentially recognized by the housekeeping sortase. Elegant work by Suree et al. (57) using NMR spectroscopy data revealed the intimate interaction between the active site residues of S. aureus SrtA and the LPXTG motif and also proposed a mechanism of binding to lipid II, resulting in important data on the transpeptidation reaction. It is, however, not possible to predict the amino acid interactions outside the canonical LPXTG motif. Hence, a similar structural approach using the TG/SG motif would further contribute to understanding the docking details of sortases and their dedicated substrates in L. rhamnosus GG and other Gram-positive species. A recent report indicated that the structure of the SpaA pilin of L. rhamnosus GG is soon to be solved, bringing to light important data to support the present regulatory model (58).