The Not so Good, the Bad and the Ugly: Differential Bacterial Adhesion and Invasion Mediated by Salmonella PagN Allelic Variants.

While advances in genomic sequencing have highlighted significant strain variability between and within Salmonella serovars, only a few protein variants have been directly related to evolutionary adaptation for survival, such as host specificity or differential virulence. The current study investigated whether allelic variation of the Salmonella adhesin/invasin PagN influences bacterial interaction with their receptors. The Salmonella enterica, subspecies enterica serovar Typhi (S. Typhi) allelic variant of PagN was found to bind significantly better to different enterocytes as well as to the extracellular matrix protein laminin than did the major Salmonella enterica, subspecies enterica serovar Typhimurium (S. Typhimurium) allele. The two alleles differed at amino acid residues 49 and 109 in two of the four predicted PagN surface loops, and residue substitution analysis revealed that a glutamic acid at residue 49 increased the adhesive and invasive properties of S. Typhi PagN. PagN sequence comparisons from 542 Salmonella strains for six representative S. enterica serovars and S. diarizonae further supported the role of glutamic acid at residues 49 and 109 in optimizing adhesion to cells and laminin, as well as for cell invasion. In summary, this study characterized unique residues in allelic variants of a virulence factor that participates in the colonization and invasive properties of different Salmonella stains, subspecies and serovars.


Introduction
Salmonella enterica subsp. enterica (S. enterica) is an entero-invasive bacterial pathogen that utilizes a type three secretion system (T3SS) encoded on the Salmonella pathogenicity island 1 (SPI-1) to invade intestinal epithelial cells. T3SS-driven uptake is particularly critical for intestinal S. enterica infections PagN [10] modulates bacterial adhesion and invasion. Here we determined that the S. Typhi allelic variant of PagN provides significantly better bacterial binding and invasive efficiencies for different enterocytes and the extracellular matrix protein laminin, as compared to the major S. Typhimurium allele. To our surprise, PagN from S. enterica subsp. diarizonae was also significantly more adhesive and invasive than the major S. Typhimurium allele. Sequence comparison and functional analysis of substitutions as specific residue positions revealed that amino acids in two of the four PagN surface-exposed loops contributed to these phenotypes. These results further support the role of allelic variation of virulence factors in adjusting the level of pathogenic attributes among Salmonella subspecies and serovars.

Bacterial Strain and Plasmid Constructions
The bacterial strains and plasmids used in this study are described in Table 1 and all PCR primers are listed in Table S1. Unless stated otherwise, all the reagents were from MilliporeSigma (St. Louis, MO, USA). Bacteria were routinely grown in LB-Lennox media unless otherwise indicated. When appropriate, ampicillin (200 µg/mL) or kanamycin (45 µg/mL) was added to the growth medium. For plasmid constructions, the pagN genes were amplified from S. Typhimurium, S. Typhi and S. diarizonae genomic DNA by PCR with the Q5 High-Fidelity DNA Polymerase (New England Biolabs Inc., Ipswich, MA, USA). For PagN expression the allelic genes were cloned into AHT-inducible plasmid pRS1 using Gibson assembly with amplicon prepared from appropriate primers. Three site-directed substitution mutants were prepared by Gibson assembly of pRS1 and amplicons prepared with pagN external and internal primers for S. Typhimurium or S. Typhi, or amplicons prepared with mutagenic primers. A pagN-his fusion construct was prepared by insertion of a NdeI-HindIII restricted amplicon into pET22b. S. Typhimurium SL1344 sipB::aphA-3 mutant was prepared by P22 generalized transduction from strain DMS1507 [27,28] and SL1344 sipB::aphA-3 ∆pagN was constructed by Gibson assembly and allelic exchange as described [29][30][31]. All plasmid and strain constructs were confirmed by PCR and sequencing.

Protein Expression and Antibody Preparation
The histidine-tagged PagN was expressed from the pET22b construct using the IPTG inducer and isolated by metal chelation chromatography as described previously [35]. A specific polyclonal antiserum against PagN-His was prepared in rabbits by using a conventional immunization protocol (Cocalico Biologicals Inc., Reamstown, PA, USA). The antiserum was adsorbed with E. coli BL21(DE3)/pET22b before use as described previously [36]. Briefly, 1 mL antiserum with 0.06% sodium azide was incubated with bacterial pellets from 10-mL cultures grown overnight for 18 h at 4 • C. After three adsorption cycles, the antiserum was filtered (0.02 µm-pore-size) before use. PagN expression in E. coli AAEC189 for all the binding and invasion studies was induced with AHT by growing the bacteria overnight at 30 • C, or for 2 h at 37 • C, starting with log phase cultures (A 600 = 0.3). Comparable levels and bacterial surface expression of the three cloned PagN alleles were standardized by using various concentrations of inducers (0.005-2 µg/mL AHT) for Western blot analysis and ELISA, as done previously [25]. Outer membrane proteins were prepared as described previously [37].

Bacterial Binding and Invasion Assays
Both epithelial cell cultures grown to confluence in 24-well plates (Corning, CLS3596) were used for the binding assays with recombinant E. coli AAEC189 (E. coli ∆fim) with pRS1 plasmid constructs s expressing a Salmonella PagN allele. Bacteria were grown overnight, diluted 10 −2 in LB broth with inducers (see above), incubated for 16-17 h, washed three times with PBS and diluted in DMEM to inoculate with a multiplicity of infection of 100 bacteria (in 0.25 mL) to 1 enterocyte. Exact bacterial inoculum numbers were checked by standard CFU counts. The culture plates were centrifuged (600 g, 5 min) to initiate contact between the bacteria and the cells and incubated for 1 h at 37 • C in 5% CO 2 . To evaluate cell adhesion, the infected monolayers were washed thrice with PBS to remove non-associated bacteria and treated with 0.5% Triton X-100 to release and count the cell-associated bacteria by CFU enumeration. For the invasion assays, a standard gentamicin protection assay was performed [38]. Following a 1 h infection (see above), cells were incubated with medium containing gentamicin (100 µg/mL) for 90 min at 37 • C in 5% CO 2 . After three washes with PBS, bacteria were released with Triton X-100 and enumerated as described above. All experiments were done in triplicate wells and repeated at least thrice.

Binding to Extracellular Matrix Proteins
Immuno Maxisorb plates with 96 wells (Nunc; Thermo Fisher Scientific, Rochester, NY, USA) were coated with 10 µg/mL of human collagen I, chicken collagen II, human collagen IV, bovine fibronectin, murine laminin or BSA (Sigma-Aldrich) in PBS at 4 • C overnight and then washed with PBS and blocked with PBS plus 1% BSA for 2 h. Binding of bacteria to laminin coated on plates was studied by using PagN-expressing E. coli. Bacteria grown and induced to make PagN were centrifuged, suspended in PBS to 10 7 CFU in 100 µL and added to laminin coated wells. Bacteria harboring empty plasmid pRS1 were used as a control. After incubation for 1 h at 37 • C, unbound bacteria were removed by three washing cycles, anti-PagN antiserum (1:500) was added, followed by wash cycles and incubation with goat anti-rabbit HRP-conjugated antibody (1:2000). After three wash cycles, bound antibodies were detected by using the 1-Step Turbo TMB ELISA substrate (Thermo Fisher Scientific) followed by 2 M sulfuric acid and measuring the absorbance at 450 nm. For the binding inhibition assays, double dilutions of heparin or heparan sulfate (400-0.4 µg/mL) were incubated with bacteria for 1 h and the mixtures were added to the laminin-coated wells for further processing as described above.

Bacterial Genomes and PagN Sequences
Salmonella genomic sequences from 497 previously studied S. enterica from 6 serovars with two biovars (serovars Typhi, Dublin, Choleraesuis, Typhimurium, Enteritidis, Newport and Gallinarum, with biovars Gallinarum and Pullorum for the latter serovar) and from 45 confirmed S. diarizonae (Table S2) were obtained from NCBI RefSeq database or assembled from NCBI SRA, EBI ENA and Wellcome Sanger Institute repositories as described [10]. The genomes were used to determine their encoded PagN sequences. The genomes of 13 incorrectly serotyped S. diarizonae strains were detected with SISTR (their corrected serovar attribution was added to Table S2) and not analyzed for PagN [39]. Protein alignments were done with Megalign, DNASTAR Lasergene (Madison, WI, USA).

PagN Structure Analysis
The 3D-structural model of PagN was predicted by using the corresponding sequence from S. Typhimurium LT2 and I-TASSER [40,41]. Among the five best predicted models, the 3rd model was chosen to be shown in Figure 1B (Protean 3D, DNASTAR Lasergene), with an overall ERRAT quality factor of 84.4 [42], in agreement with a published model for S. Typhi PagN [43]. The five best I-TASSER predicted S. arizonae PagN structures essentially overlapped with the ones of S. Typhimurium and S. Typhi.

Results and Discussion
PagN of S. Typhimurium and S. Typhi had similar sequences with two to three substitutions in predicted surface-exposed loops 1 and 2 ( Figure 1). Since PagN of both strains act as adhesins and invasins for human intestinal epithelial cells, we wondered whether these allelic variants differentially impacted these properties. To ensure consistent and comparable expression of PagN, we cloned each gene into inducible expression plasmids to make pDMS1973 and pDMS1974 for the expression of PagN from S.Typhi (pagN Ty ) and S. Typhimurium (pagN Tm ) respectively. Western blot analysis of outer membrane preparations from E. coli AAEC89 carrying one of these two plasmid constructs detected bands specific for each PagN allele ( Figure 2A).  (Table S3) [10] each representing N strains from one or more serovars, or clusters within serovars or subspecies with the same sequence in the surface loops of mature PagN; S. Typhimurium_LT2 (n = 68) with same sequence for S. Enteritidis Color codes are for the membrane spanning beta-barrel (yellow), the periplasmic loops (grey), surface-exposed loop 1 (purple) with residue 49 (turquoise), loop 2 (red) with residue 109 (turquoise), loop 3 (green) and loop 4 (dark blue).  (Table S3) [10] each representing N strains from one or more serovars, or clusters within serovars or subspecies with the same sequence in the surface loops of mature PagN; S. Typhimurium_LT2  Moreover, both alleles of PagN were detectable on the bacterial surface by immunofluorescence ( Figure 2B, panel a to f) and ELISA (not shown). Not surprisingly, the strongest signal was observed for the S. Typhimurium PagN allele, as this was the immunogen used to prepare the antiserum. Moreover, both alleles of PagN were detectable on the bacterial surface by immunofluorescence ( Figure 2B, panel a to f) and ELISA (not shown). Not surprisingly, the strongest signal was observed for the S. Typhimurium PagN allele, as this was the immunogen used to prepare the antiserum.
The E. coli strain AAEC189 lacks adhesive type 1 fimbriae, providing us with a bacterial context in which to study PagN-mediated bacterial binding and cellular uptake free of known Salmonella adhesins and invasins. Using both RKO and IPEC-J2 enterocytes, we found that both PagN Ty and PagN Tm mediated bacterial binding and invasion, as previously reported for ovarian hamster epithelial-like CHO-K1 cells and human colonic HT-29 cells [12,23]. More importantly, comparisons of the two alleles indicated that PagN Ty was more efficient than PagN Tm for both cell adhesion and invasion ( Figure 3A), supporting an allelic variant effect on a virulence property. Additional studies with a S. Typhimurium pagN and sipB (essential SPI-1 translocon subunit) deletion mutant complemented with different PagN-expressing plasmids did not show significant different levels of adhesion/invasion of RKO cells (data not shown), suggesting a dominant phenotype due to the expression of one or more other Salmonella adhesin(s) and invasin(s) under the used growth conditions. Thus, the use of E. coli with a controlled expression system allowed us to bypass the masking effect of additional Salmonella adhesins/invasins differentially expressed in different serovars and environmental growth conditions and identify functional differences due only to sequence variations of PagN alleles [44,45]. The E. coli strain AAEC189 lacks adhesive type 1 fimbriae, providing us with a bacterial context in which to study PagN-mediated bacterial binding and cellular uptake free of known Salmonella adhesins and invasins. Using both RKO and IPEC-J2 enterocytes, we found that both PagNTy and PagNTm mediated bacterial binding and invasion, as previously reported for ovarian hamster epithelial-like CHO-K1 cells and human colonic HT-29 cells [12,23]. More importantly, comparisons of the two alleles indicated that PagNTy was more efficient than PagNTm for both cell adhesion and invasion ( Figure 3A), supporting an allelic variant effect on a virulence property. Additional studies with a S. Typhimurium pagN and sipB (essential SPI-1 translocon subunit) deletion mutant complemented with different PagN-expressing plasmids did not show significant different levels of adhesion/invasion of RKO cells (data not shown), suggesting a dominant phenotype due to the expression of one or more other Salmonella adhesin(s) and invasin(s) under the used growth conditions. Thus, the use of E. coli with a controlled expression system allowed us to bypass the masking effect of additional Salmonella adhesins/invasins differentially expressed in different serovars and environmental growth conditions and identify functional differences due only to sequence variations of PagN alleles [44,45].  pRS1 (no pagN), pDMS1973 or pDMS1974 to express PagNTy or PagNTm respectively, to human (RKO) and porcine (IPEC-J2) was analyzed by incubating 3 × 10 7 CFU/mL bacteria at a MOI of 100 for 60 min for adherence and another 90 min for invasion assays. Bacteria expressing PagNTy bound and invaded significantly better than the bacteria expressing PagNTm (p < 0.01-0.001). (B) Adherence of E. coli AAEC189 expressing PagNTm to RKO cells was significantly enhanced (p < 0.05) when PagNTm was mutated from aspartate to glutamate (at site 49, D49E) with pDMS2081 or to glutamine (at site 109, D109Q) with pDMS2082, whereas invasion of epithelial cells was affected only by the D49E substitution (p < 0.001), but not by the D109Q substitution. Data represent one of three separate and reproducible experiments each with triplicate data expressed as mean ± SEM (ns is for not significant, * p < 0.05, ** p < 0.01, *** p < 0.001 and **** p < 0.0001).
PagN sequence alignments for 52-75 strains for each of the seven S. enterica serovars, including S. Typhimurium [10], highlighted that all PagN had aspartic acid residues at position 49 and 109, with the exceptions of S. Typhi that had a glutamic acid at position 49, and a group of S. Newport strains that had a glutamine at position 109 ( Figure 1A). Since the allelic PagN proteins of all strains of S. Typhi and S. Typhimurium varied at these two positions and were each predicted to be located in a different bacterial surface loop ( Figure 1B), we investigated their relative involvement in bacterial adhesion and invasion. For this, we generated PagNTm with a D49E or a D109Q substitution and coli AAEC189 with empty vector pRS1 (no pagN), pDMS1973 or pDMS1974 to express PagN Ty or PagN Tm respectively, to human (RKO) and porcine (IPEC-J2) was analyzed by incubating 3 × 10 7 CFU/mL bacteria at a MOI of 100 for 60 min for adherence and another 90 min for invasion assays. Bacteria expressing PagN Ty bound and invaded significantly better than the bacteria expressing PagN Tm (p < 0.01-0.001). (B) Adherence of E. coli AAEC189 expressing PagN Tm to RKO cells was significantly enhanced (p < 0.05) when PagN Tm was mutated from aspartate to glutamate (at site 49, D49E) with pDMS2081 or to glutamine (at site 109, D109Q) with pDMS2082, whereas invasion of epithelial cells was affected only by the D49E substitution (p < 0.001), but not by the D109Q substitution. Data represent one of three separate and reproducible experiments each with triplicate data expressed as mean ± SEM (ns is for not significant, * p < 0.05, ** p < 0.01, *** p < 0.001 and **** p < 0.0001).
PagN sequence alignments for 52-75 strains for each of the seven S. enterica serovars, including S. Typhimurium [10], highlighted that all PagN had aspartic acid residues at position 49 and 109, with the exceptions of S. Typhi that had a glutamic acid at position 49, and a group of S. Newport strains that had a glutamine at position 109 ( Figure 1A). Since the allelic PagN proteins of all strains of S. Typhi and S. Typhimurium varied at these two positions and were each predicted to be located in a different bacterial surface loop ( Figure 1B), we investigated their relative involvement in bacterial adhesion and invasion. For this, we generated PagN Tm with a D49E or a D109Q substitution and expressed them in E. coli strain AAEC189. Although both mutated PagN mediated better bacterial binding to RKO cells than did PagN Tm , only the D49E substitution in PagN had a significantly stronger effect on bacterial invasion ( Figure 3B). This result highlighted the contribution of the S. Typhi glutamic acid at position 49 for the improved interaction of PagN with a host cell receptor to promote bacterial uptake. This result is consistent with a role for the PagN allele in the increased pathogenicity of S. Typhi relative to S. Typhimurium pathogenesis following human infection, in agreement with the former serovar's hematogenous bacterial spreading in humans resulting frequently in sepsis, in contrast to the latter serovar and its pathogenesis that is usually contained in the gastro-intestinal organs.
Other than subsp. enterica, most subsp. of S. enterica are associated with cold-blooded animals. Therefore, we expected that a construct expressing the PagN allele of these non-enterica subspecies would neither bind nor efficiently invade human RKO cells and therefore would serve as a negative control. Thus, we cloned the S. diarizonae PagN (PagN di ) in the same inducible expression plasmid used for the two other PagN alleles to make pDMS2062 and confirmed protein production ( Figure 2A) and surface expression ( Figure 2B, panels g and h). To our surprise, E. coli making PagN di were significantly more adhesive and invasive than the ones expressing PagN Tm ( Figure 4A).
Microorganisms 2019, 7, x FOR PEER REVIEW 9 of 14 expressed them in E. coli strain AAEC189. Although both mutated PagN mediated better bacterial binding to RKO cells than did PagNTm, only the D49E substitution in PagN had a significantly stronger effect on bacterial invasion ( Figure 3B). This result highlighted the contribution of the S. Typhi glutamic acid at position 49 for the improved interaction of PagN with a host cell receptor to promote bacterial uptake. This result is consistent with a role for the PagN allele in the increased pathogenicity of S. Typhi relative to S. Typhimurium pathogenesis following human infection, in agreement with the former serovar's hematogenous bacterial spreading in humans resulting frequently in sepsis, in contrast to the latter serovar and its pathogenesis that is usually contained in the gastro-intestinal organs.
Other than subsp. enterica, most subsp. of S. enterica are associated with cold-blooded animals. Therefore, we expected that a construct expressing the PagN allele of these non-enterica subspecies would neither bind nor efficiently invade human RKO cells and therefore would serve as a negative control. Thus, we cloned the S. diarizonae PagN (PagNdi) in the same inducible expression plasmid used for the two other PagN alleles to make pDMS2062 and confirmed protein production ( Figure  2A) and surface expression ( Figure 2B, panels g and h). To our surprise, E. coli making PagNdi were significantly more adhesive and invasive than the ones expressing PagNTm ( Figure 4A). and S. Typhimurium PagN (PagNTm) to RKO cells showed the bacteria expressing PagNdi binds significantly better than PagNTm (p < 0.01) and was slightly more invasive (p < 0.05). (B) A substitution of aspartate to glutamate at site 109 in S. Typhimurium PagN (D109E) increased both binding and invasion of E. coli into RKO cells as compared to the bacteria expressing PagNTm (p < 0.01). Data represent one of three separate and reproducible experiments, each with triplicate data expressed as mean ± SEM (* p < 0.05, ** p < 0.01, *** p < 0.001 and **** p < 0.0001).
PagN alignments of 48 available strains of S. diarizonae highlighted that they all had an aspartic acid at position 49, like most evaluated subsp. enterica serovars (Typhi being the exception, Figure  1A). In contrast, unlike the subsp. enterica serovars, all PagNdi carried a glutamic acid at position 109, suggesting that a D109E substitution at this position would be functionally significant. In support, a D109E substitution in the S. Typhimurium PagN (pDMS 2083) improved not only bacterial binding, but also bacterial uptake by RKO cells (Figure 4B), indicating that glutamic acid in this position  PagN (PagN Tm ) to RKO cells showed the bacteria expressing PagN di binds significantly better than PagN Tm (p < 0.01) and was slightly more invasive (p < 0.05). (B) A substitution of aspartate to glutamate at site 109 in S. Typhimurium PagN (D109E) increased both binding and invasion of E. coli into RKO cells as compared to the bacteria expressing PagN Tm (p < 0.01). Data represent one of three separate and reproducible experiments, each with triplicate data expressed as mean ± SEM (* p < 0.05, ** p < 0.01, *** p < 0.001 and **** p < 0.0001).
PagN alignments of 48 available strains of S. diarizonae highlighted that they all had an aspartic acid at position 49, like most evaluated subsp. enterica serovars (Typhi being the exception, Figure 1A).
In contrast, unlike the subsp. enterica serovars, all PagN di carried a glutamic acid at position 109, suggesting that a D109E substitution at this position would be functionally significant. In support, a D109E substitution in the S. Typhimurium PagN (pDMS 2083) improved not only bacterial binding, but also bacterial uptake by RKO cells (Figure 4B), indicating that glutamic acid in this position optimizes the function of PagN as both an adhesin and invasin. Whether other substitutions in PagN di , such as the arginine or lysine at position 150 and 154 in loop 3 ( Figure 1) modulate bacterial invasion remains to be determined. Like S. Typhi, S. diarizonae have type IVB pili involved in intestinal cell invasion, raising the possibility that the concomitant increased binding property of the S. diarizonae PagN allele contributes to the reported increased bacterial virulence in humans under specific conditions [5]. Taken together, the results showed that PagN alleles with a glutamic acid residue at either 49 or 109 improves bacterial binding and/or invasiveness, possibly due to the more ionizable and long chain characteristics of glutamic acid relative to aspartic acid and glutamine.
In addition to promoting bacterial attachment to cells, many bacterial adhesins interact with various host glycoprotein of the extracellular matrix, as exemplified by the ability of PagN Ty to bind to laminin [23,[46][47][48]. S. enterica may encounter laminin either on intestinal areas denuded of epithelial cells (e.g., extrusion zones at villi tips) or in the subepithelial space after invading enterocytes and crossing the intestinal epithelial layer. Notably, the low pH and magnesium concentrations in Salmonella-containing vacuoles (SCV) of invaded cells induce the PhoPQ two-component system responsible for PagN expression [49][50][51]. Thus, whereas the S. enterica SPI-1 system is induced in the intestinal environment and used for enterocyte invasion, expression of PagN in SCV (where SPI-1 is largely repressed) may prepare Salmonella for future cycles of cell invasion after escape from intestinal epithelial cells. Our side-by side comparison of PagN Ty -, PagN Tm -and PagN di -expressing E. coli showed that all three bacteria adhered significantly to laminin, albeit the former better than the latter two ( Figure 5A). None of the bacteria bound to collagen I, II or IV or fibronectin (not shown).  [5]. Taken together, the results showed that PagN alleles with a glutamic acid residue at either 49 or 109 improves bacterial binding and/or invasiveness, possibly due to the more ionizable and long chain characteristics of glutamic acid relative to aspartic acid and glutamine.
In addition to promoting bacterial attachment to cells, many bacterial adhesins interact with various host glycoprotein of the extracellular matrix, as exemplified by the ability of PagNTy to bind to laminin [23,[46][47][48]. S. enterica may encounter laminin either on intestinal areas denuded of epithelial cells (e.g., extrusion zones at villi tips) or in the subepithelial space after invading enterocytes and crossing the intestinal epithelial layer. Notably, the low pH and magnesium concentrations in Salmonella-containing vacuoles (SCV) of invaded cells induce the PhoPQ twocomponent system responsible for PagN expression [49][50][51]. Thus, whereas the S. enterica SPI-1 system is induced in the intestinal environment and used for enterocyte invasion, expression of PagN in SCV (where SPI-1 is largely repressed) may prepare Salmonella for future cycles of cell invasion after escape from intestinal epithelial cells. Our side-by side comparison of PagNTy-, PagNTm-and PagNdi-expressing E. coli showed that all three bacteria adhered significantly to laminin, albeit the former better than the latter two ( Figure 5A). None of the bacteria bound to collagen I, II or IV or fibronectin (not shown). To determine the potential role of PagN residue 49 and 109 in PagNTy binding to laminin, we tested the three mutants described above. As for the intestinal epithelial cell binding results, bacteria with a glutamic acid at position 49 of PagNTm (D49E) bound significantly better to laminin than did bacteria with PagNTm ( Figure 5B). Glutamic acid or glutamine at position 109 of PagNTm (D109E and D109Q) also increased binding, albeit less efficiently.
Since both PagN and heparin bind to laminin [52], we next determined whether heparin or heparan sulfate could inhibit PagN-mediated bacterial binding to laminin. In contrast to the reported inhibitory effect of heparin on PagN-mediated invasion of CHO-K1 cells [53], neither heparin nor heparan sulfate at concentrations as high as 400 µg/mL interfered with the binding of PagNTyexpressing bacteria to laminin (data not shown). These combined results suggested that PagN and heparin bind to different laminin sites and that the invasion of CHO-K1 cells by PagN-expressing To determine the potential role of PagN residue 49 and 109 in PagN Ty binding to laminin, we tested the three mutants described above. As for the intestinal epithelial cell binding results, bacteria with a glutamic acid at position 49 of PagN Tm (D49E) bound significantly better to laminin than did bacteria with PagN Tm (Figure 5B). Glutamic acid or glutamine at position 109 of PagN Tm (D109E and D109Q) also increased binding, albeit less efficiently.
Since both PagN and heparin bind to laminin [52], we next determined whether heparin or heparan sulfate could inhibit PagN-mediated bacterial binding to laminin. In contrast to the reported inhibitory effect of heparin on PagN-mediated invasion of CHO-K1 cells [53], neither heparin nor heparan sulfate at concentrations as high as 400 µg/mL interfered with the binding of PagN Ty -expressing bacteria to laminin (data not shown). These combined results suggested that PagN and heparin bind to different laminin sites and that the invasion of CHO-K1 cells by PagN-expressing bacteria is laminin-independent. Thus, our studies highlight the independent adhesive properties of PagN in binding to either intestinal cells or the extracellular matrix protein laminin, two relevant targets for Salmonella host invasion.
In summary and together with our previous studies on variants of fimbrial adhesins [25,26,54], this study on PagN, a Salmonella adhesin/invasin, further supports the importance of protein sequence allelic variants in virulence properties [55], including pathogenic properties such as adhesion and invasion [4].