The Hd, Hj, and Hz66 flagella variants of Salmonella enterica serovar Typhi modify host responses and cellular interactions

Salmonella Typhi, the causative agent of typhoid fever, is a monophyletic, human-restricted bacterium that exhibits limited phenotypic variation. S. Typhi from Indonesia are a notable exception, with circulating strains expressing diverse flagella antigens including Hj, Hd and Hz66. Hypothesizing that S. Typhi flagella plays a key role during infection, we constructed an S. Typhi fliC mutant and otherwise isogenic S. Typhi strains expressing the Hj, Hd, Hz66 flagella antigens. Phenotyping revealed differences in flagellum structure, strain motility and immunogenicity, but not in the ability of flagellated isolates to induce TLR5 activity. Invasion assays using epithelial and macrophage cell lines revealed differences in the ability of these S. Typhi derivatives to invade cells or induce cellular restructuring in the form of ruffles. Notably, the Hj variant induced substantial ruffles that were not fully dependent on the GTPases that contribute to this process. These data highlight important differences in the phenotypic properties of S. Typhi flagella variation and how they impact on the pathogenesis of S. Typhi.

originating in Indonesia harbor pBSSB1, a linear plasmid encoding a fljB analogue that directs the expression of the Hz66 flagellin antigen 12,13 . Plasmid pBSSB1 additionally encodes a repressor of the chromosomal fliC gene; ensuring only one flagella antigen is expressed at a time. A third antigenic variant of flagellin, known as Hj, is also found in some S. Typhi that also originate from Indonesia. Hj is encoded by an allele of fliC gene harboring a 261 bp in-frame deletion in the central region of the Hd fliC coding sequence 14 .
The impact of flagella antigenic variation on pathogenesis and immunity within S. Typhi is not well described. Furthermore, we hypothesized that flagella variation plays a unique role in regulating the immune response to S. Typhi infection in Indonesia, a location where a range of atypical S. Typhi flagella variants circulate. Here, we have engineered an aflagellated S. Typhi DfliC mutant and a set of three otherwise isogenic derivatives of S. Typhi that differ only in the flagella antigen variant expressed on the surface (Hd, Hj and Hz66). These isogenic S. Typhi derivatives were subjected to a range of phenotypic assays including their ability to interact with epithelial cells and macrophages. We show that the flagella type can influence the immune response during typhoid and impact on the ability of S. Typhi to invade host cells.

Results
Construction and primary characterization of isogenic S. Typhi flagella variants. To compare the properties of the Hd, Hj and Hz66 flagella, we used targeted mutagenesis to construct three otherwise isogenic S. Typhi derivatives that differed only in terms of their flagellin gene content (alignments of the flagellin proteins from S. Typhi and S. Typhimurium are shown in Figure 1). For consistency, all gene replacements targeted the fliC locus, i.e. the Hd encoding fliC gene was completely replaced with only the Hj fliC or the Hz66 fljB allele using the native fliC promoter to direct expression. An additional isogenic derivative harboring a null deletion in fliC (DfliC) was constructed to serve as an amotile, aflagellated control. The genetic structure of these flagellated and aflagellated S. Typhi derivatives were confirmed by sequencing and the different derivatives were then screened in agglutination assays with flagella specific antisera and were found to express the appropriate flagella antigens ( Figure 2).
The motility of the S. Typhi flagella variants was measured by assessing their swimming capabilities in soft media over a defined incubation period. The flagellated S. Typhi DfliC derivative was amotile, while each of the flagellated S. Typhi swam between 37 and 58 mm in the agar matrix during the 16-hour incubation period at 37uC (Figure 2e). The Hz66-expressing S. Typhi derivative was consistently the least motile. Despite having the shortest flagella, the Hj S. Typhi derivative migrated significantly further in the soft agar (median; 57 mm, range; 57-58 mm) than the Hd and the Hz66 derivatives (median; 50 mm, range; 49-51 mm, median; 38 mm, range 37-41 mm, respectively) (p,0.0001; 2 sided t-test).
Indonesian Typhoid patients elaborate IgG against Hd, Hz66 and Hj flagellin. Flagellin is highly immunogenic and antibodies (IgG) against S. Typhi flagellin can be measured for a prolonged period after a confirmed typhoid infection 15 . As the isogenic S. Typhi derivatives demonstrated different phenotypic qualities we hypothesized that the organisms expressing these flagella antigens might stimulate different responses from the immune system during natural infection. To test this hypothesis, IgG against Hd, Hj and Hz66 flagellin was measured in a group of typhoid patients. Firstly, the type of flagella genes encoded in S. Typhi isolated during a typhoid case/control study conducted in Jakarta, Indonesia, a region where S. Typhi expressing the Hd, Hj and Hz66 flagella are co-circulating, were assessed by PCR amplification. Thirty S. Typhi isolates, where a corresponding acute serum sample from a typhoid patient was available, were analyzed and 15 were Hd, 4 were Hd:Hz66, 11 were Hj:Hz66, and none were Hj only (Table 1). Available disease metadata was stratified by flagellin variant and there was no significant difference between the three groups and the number of days of fever prior to hospitalization. However, Hz66-positive S. Typhi originated from on average older patients than the isolates expressing Hd alone by a mean of seven years (p50.014; 2 sided t-test).
Serum from the 30 from typhoid patients from which the above S. Typhi flagella variants were isolated and 79 from asymptomatic controls (Table 1) were screened using an ELISA to measure IgG against the three forms of S. Typhi flagellin ( Figure 3). The majority of typhoid patients (25/30), regardless of the S. Typhi flagella variant isolated from their blood, harbored IgG against Hd (Figure 3a). Furthermore, the preponderance of patients (12/15) infected with an Hz66 S. Typhi harbored IgG against Hz66, as did four patients infected with Hd isolates (Figure 3c). Conversely, only three patients demonstrated an anti-Hj IgG response, of which only one was infected with an S. Typhi Hz66:Hj isolate ( Figure 3b). Interestingly, the asymptomatic controls also harbored antibody responses of a similar magnitude to the typhoid fever patients, indicating probable previous exposure. Of the 79 asymptomatic controls, 31 (39.2%) harbored IgG against at least one of the flagellin antigens, with the majority responding to Hd (30/31, 97%), four (5.1%) to Hz66 IgG with only one of these to Hz66 exclusively and also to Hd. Only one control sample harbored measurable anti-Hj IgG and this also harbored IgG against the other two flagellin antigens. We found no significant difference in IgG levels to Hd, Hj, or Hz66 flagellin between the asymptomatic controls or those infected with Hd, Hj or Hz66 S. Typhi (p.0.05 in all pairwise comparisons; 2 sided ttests).

S. Typhi flagella variants induce similar activation of TLR5.
Predicting that the S. Typhi flagella variants may have a differing ability to stimulate Toll Like Receptor 5 (TLR5), we independently co-transfected HEK293 cells expressing TLR5 with the three S. Typhi flagella variants, S. Typhi DfliC or S. Typhimurium SL1344 as a positive control ( Figure 4). All of the S. Typhi derivatives, with notable exception of the DfliC mutant, were able to stimulate TLR5 and produce downstream NFkB signaling with a similar degree of potency.
The interaction of S. Typhi flagella derivatives with macrophages and epithelial cells. The S. Typhi derivatives were inoculated onto THP-1 cells to measure cellular uptake and S. Typhi expressing the Hd flagella were consistently taken up by more efficiently than the Hj and Hz66 expressing derivatives (Figure 5a). The DfliC S. Typhi was reproducibly taken up at a lower frequency than any of the flagellated derivatives. No difference in cytotoxicity was observed between the S. Typhi derivatives using a lactate dehydrogenase assay (data not shown).
The transcriptome of THP-1 cells infected with the S. Typhi derivatives was measured by DNA microarray analysis of the host mRNA www.nature.com/scientificreports SCIENTIFIC REPORTS | 5 : 7947 | DOI: 10.1038/srep07947 populations. S. Typhi expressing Hd and Hj flagellin induced highly similar transcription patterns in host cells, with higher numbers of differentially expressed genes, compared to the Hz66 or DfliC derivatives ( Table 2). The pathway and gene ontologies of the differential expressed genes groups were determined using InnateDB and could be divided into three main groups with corresponding profiles (Figure 5b). These three main groups were; i) the up-regulation of genes involved in inflammation in cells infected with flagellated bac- www.nature.com/scientificreports teria; ii) the up-regulation of gene involved in gene expression, translation, and protein metabolism in cells infected with either Hd or Hj derivatives and; iii) the proportional down-regulation of genes in the NOTCH pathway, Wnt-mediated gene transcription, and protein kinase activity with flagellated organisms.
Next, the various flagella variants were independently inoculated onto human epithelial-like Hep-2 cells ( Figure 6). The Hd and Hz66 S. Typhi derivatives invade Hep-2 cells with a comparable frequency but the S. Typhi Hj derivative reproducibly demonstrated a significantly higher capability for invasion (0.42% 6 0.15) ( Figure 6a).  Furthermore, S. Typhi DfliC was reproducibly even less invasive and was only marginally more invasive than S. Typhi DinvA. We found no significant difference in the ability of the various flagellated derivatives or DfliC and DinvA S. Typhi to attach to the epithelial cells ( Figure 6b). Similar to the work on THP-1 cells, gene expression profiles were determined using microarray analysis of mRNA populations present in epithelial cells exposed to S. Typhi expressing the different flagellins. Levels of IL8 mRNA and pathways related to cytokine-cytokine receptor interactions and the HIF1a transcription factor network were comparatively over expressed by Hep-2 cells exposed to flagel-lated S. Typhi compared to the DfliC derivative. Unsupervised hierarchical clustering revealed a similar response to S. Typhi expressing either Hd or Hj but this was distinct from S. Typhi Hz66 ( Figure 6c). In fact, the overall host transcriptome response to the Hz66 was more comparable to the response induced by S. Typhi DfliC and DinvA than the Hd and Hj derivatives ( Table 2). This difference was mainly restricted to the up-regulation of genes involved in gene expression, translation and general protein metabolism e.g. genes CDK1, CDKN1B, MNAT1, CCNG1, CCNG2, RPL7, RPL9, RPL14 and MYC (Supplementary information).
During epithelial cell invasion, Salmonella trigger rearrangements of the host cell cytoskeleton including ruffles in the plasma membrane. SEM was used to compare the interaction of S. Typhi expressing Hd, Hj or Hz66 flagellin with epithelial cells during invasion and to gain insight into the different gene expression profiles. All S. Typhi were able to stimulate substantial membrane ruffling but abnormally large ruffles were consistently observed when S. Typhi expressing Hj flagella interacted with epithelial cells (Figure 7a).
Ruffles are triggered in part by effectors secreted through the Salmonella Pathogenicity Island I (SPI1) interacting with host Rho-GTPases, including Rac1, Cdc42, and RhoG [16][17][18] . To assess the contribution of each of these Rho-GTPases to ruffling, siRNAs were generated for each gene and ruffling was observed on Hep-2 cells exposed to individual siRNAs using TEM. siRNAs to RAC1, CDC42 or RHOG completely repressed ruffle formation on Hep-2 cells exposed to S. Typhi expressing the Hd or Hz66 flagella (Figure 7b). Cell ruffling was reduced when Hj S. Typhi were inoculated onto Hep-2 exposed to RAC1 but interestingly not RHOG and CDC42 siRNAs.

Discussion
We performed a series of experiments using a combination of typhoid patients and in vitro assays to assess the impact of flagella type on host cell-pathogen interactions involving S. Typhi. To facilitate these studies we constructed a novel series of carefully engineered isogenic S. Typhi derivatives differing only in the antigenic structure of their flagella. These data are of interest as the majority of global S. Typhi isolates are monophasic and express only the classical Hd flagella yet novel S. Typhi are originating in the Indonesian archipelago that can express alternative flagellin either from an allelic variant of Hd known as Hj or from a novel fljB gene encoded on a linear plasmid pBSSB1, known as Hz66 13,14 . Here we show that the different flagellin have distinct structural features that directly impinge on the motility of the bacteria and their pathogenic potential. Firstly, we found that Hz66 flagella were measurably thicker than Hj or Hd flagella and that this property translated into poorer motility when comparing S.    Typhi Hz66 to Hd and Hj-positive derivatives. Correspondingly, S. Typhi Hj produced shorter and thinner flagella structures and swam faster than both Hd and Hz66 S. Typhi derivatives. During laboratory observation the Hj flagella were found to be more fragile that the other flagella variants and were detected mainly in the culture medium whereas the other flagella were predominantly attached to the bacterial body. This data are broadly in keeping with previous observations on S. Typhimurium flagellin genes with deletions approximately the same size and location as Hj 19 . However, our results are somewhat different some previously published observations, in which non-isogenic Hd strains were reported to have a higher motility 20 but here isogenic derivatives were used to compare these phenotypic characteristics. Data collected using Hep-2 or THP-1 cells indicated that the type of expressed S. Typhi flagella can dramatically influence host/pathogen interactions. These results were supported by our unpublished observation that S. Typhi DflgK derivatives that produce and secrete unpolymerized flagellin also exhibit a reduced ability to invade Hep-2 cells, in keeping with the impact of flagellin in S. Typhimurium pathogenesis 21,22 . The increased capacity of the S. Typhi Hj to invade Hep-2 cells appears to correlate with the induction of larger cellular actin ruffles than those induced by either Hd or Hz66 S. Typhi derivatives. The formation of these ruffles is associated with invasion, triggered when Salmonella have intimate contact with a non-phagocytic cell and involves a number of Salmonella Pathogenicity Island associated effector proteins 23 . These effectors interact with Rho GTPases within the host cell, stimulating a rearrangement of actin in the host cell cytoskeleton [16][17][18] . To investigate the mechanism of ruffle formation we performed a number of iRNA experiments to suppress the expression of Rho GTPases, which are known to interact with Salmonella effectors (RHOG, RAC1, CDC42). The resulting data demonstrated that the dramatic ruffle formation induced by S. Typhi Hj occurred independently of CDC42 and RAC1. These data predict the existence of an alternative signaling cascade resulting in bacterial internalization, which is activated through contact with bacterial flagellin. In support of this hypothesis, it has been shown that S. Typhimurium flagellin is injected into the host cell cytosol in part through SPI1 24 , and that flagellin may be acting as an 'effector protein' during cellular invasion and uptake.
As with our observations in epithelial cells, the presence of flagella was required for the efficient uptake of S. Typhi into macrophages. Here, we also observed differences in invasion rates between the S. Typhi flagellated derivatives but here the S. Typhi Hd derivative had a greater capacity for internalization than the S. Typhi Hj. This differential interaction between the Hd and Hj derivatives with macrophages may, in part, be one of the factors influencing why Hd S. Typhi are successfully globally whereas Hj derivatives predominantly restricted to the islands of Indonesia. However, additional factors such as host genetics and environmental conditions may play an even greater role in this atypical geographic restriction. Others have suggested a role for the predatory protozoan species carried by Indonesians 25 . A role for immune invasion in the evolution of the novel Indonesian S. Typhi remains unproven, although the apparent poor immunogenicity of Hj may be contributing to local selection on S. Typhi. Here, we speculate that the Hz66 variant is moving into a niche in individuals who have been exposed to Hd S. Typhi.
We found a wide variation in gene expression patterns between cells exposed to the S. Typhi flagella derivatives. For example, the S. Typhi Hz66 derivative consistently induced transcriptome profile more similar to DfliC and DinvA than the Hd and Hj S. Typhi, which were in turn more similar. All the flagellated bacteria, however, induced a robust and measurable acute inflammatory response, in keeping with the lack of detected difference in signaling through TLR5. Indonesia is still endemic for typhoid fever, with an estimated incidence of 810/100,000 cases per year 26 . The number of individuals within the community with a substantial IgG response to S. Typhi flagellin presumably reflects this high prevalence of typhoid in Indonesia. The serum from several individuals (both typhoid fever patients and community controls) exhibited signatures correlating with multiple infections with S. Typhi expressing different flagella antigens. This theory of multiple infections is supported by the difference in median ages, since those infected with Hj or Hz66 S. Typhi were significantly (for Hz66) older than those with Hd infections. These data suggest that Hz66 S. Typhi may be more 'opportunistic' than S. Typhi Hd, exploiting a niche after a previous infection/exposure to S. Typhi Hd. An additional observation was that both Hd and Hz66 flagellin appears to induce a more robust IgG response than Hj. It is noteworthy that when mice were immunized with Hj flagellin they mounted a much weaker antibody response compared to similar mice immunized with Hz66 or Hd flagellin (our unpublished observations). This reduced immunogenicity of Hj flagella is perhaps not surprising, since the single dominant B cell epitope of Hd is centered at residues 229-230, within in the section missing in Hj (aa 224-310) 27 .
In summary, the data presented here indicates the important, active role of flagella in host pathogen interactions during S. Typhi infection, engaging both innate and adaptive branches of the immune response. The differences in invasion and immunogenicity observed between flagellin variants suggests an almost opportunistic behavior of the less widespread variants (Hj and Hz66), taking advantage of preexisting anti-Hd immunity; an issue that should be taken into account when developing novel whole-cell flagellated Salmonella vaccines.

Experimental procedures
Bacterial and genetic manipulation. The attenuated S. Typhi Ty2 derivative BRD948 (Hd), harboring deletions in the aroA, aroC and htrA genes, was used for all experiments to avoid safety issues related to genetically engineering a containment level three organism 28 . S. Typhi BRD948 and derivatives are approved for use in a containment level two laboratory in the United Kingdom 29 . All genetic manipulations were performed using Luria-Bertani (LB) media supplemented with 40 mg L 21 of lphenylalanine and l-tryptophan, and 10 mg L 21 of p-aminobenzoic acid and 2,3dihydroxybenzoic acid (aro mix). When required, media was supplemented with chloramphenicol, ampicillin or kanamycin and growth temperatures were adjusted (37uC or 42uC), according to the requirements for the elimination of plasmids through temperature sensitive replication. Isogenic S. Typhi flagella variants (Hd, Hj and Hz66) were constructed using the lambda red recombinase method 30 . Firstly, a fliC (non-motile) mutant was constructed. PCR amplicons were designed and produced to remove fliC in its entirety but leaving the upstream and downstream regions intact. Plasmid pKD3 was used as a template for PCR amplifications. PCR amplicons were electrotransformed in S. Typhi BRD948 pKD46 as previously described 29 . A non-flagellated mutant derivative, S. Typhi BRD948 DfliC, was generated after selection with chloramphenicol and screened for a lack of motility using 0.3% LB swim agar plates with appropriate supplementation. The pKD3 insertion, including the antimicrobial resistance cassette, was removed using pCP20, as previously described [2]. Hj and Hz66 flagellin variants were constructed by PCR amplification of the Hj and Hz66 loci (and 200 bp of upstream and downstream flanking sequence) from strain S. Typhi Ty404 31 . PCR amplicons were electrotransformed in S. Typhi BRD948 DfliC containing pKD46 as before. Organisms 'complemented in situ' with alternative flagellin genes were selected on the basis of their ability to swim in 0.3% LB swim agar plates with appropriate supplementation. Motile organisms were subcultured to ensure purity and screened by PCR amplification and sequencing to ensure the Hj and Hz66 encoding genes were inserted correctly.
Cellular invasion assays. Hep-2 cells were grown in Dulbecco's Modified Eagle Medium (DMEM, Sigma), 10% fetal calf serum (FCS, Sigma) and 2 mM L-glutamine (Sigma), at 37uC in 5% CO 2 . The day before the infection assays were performed, cells were seeded at 1 3 10 5 cells/well in 24-well plates and incubated overnight. Bacteria were added at a multiplicity of infection (MOI) of 10:1 (bacteria:cell). Plates were then centrifuged at 600xG for 5 minutes to ensure contact with cells. Cells were incubated for 2 hours, washed with PBS and fresh medium supplemented with 100 mg mL 21 gentamycin was added. Cells were further incubated for an additional three hours. After washing, cells were lysed with 100 ml/well of 1% Triton X-100. Serial dilutions were performed and plated on agar plates for enumeration after overnight incubation.
For adhesion/invasion assays, Hep-2 cells were incubated at 4uC for 20 minutes prior to infection. Bacteria were resuspended in ice-cold DMEM and added at a MOI of 100:1. Plates were incubated for 1 hour at 4uC to allow attachment but not invasion; cells were then washed and lysed as described above. For the invasion assays, plates were washed and fresh pre-warmed DMEM was added to the wells. Cells were then incubated for one hour at 37uC. The media was then changed to DMEM with gentamycin and cells were incubated for another hour, before washing and cell lysis as described above.
THP-1 cells were cultured in RPMI-1640 medium (Sigma), containing 10% FCS, 2 mM L-glutamine, at 37uC and in 5% CO 2 . A week prior to infection, THP-1 cells were seeded onto 24-well plates, 1 3 10 5 cells/well, and differentiated into macrophages with 50 ng mL 21 of phorbol 12-myristate 13-acetate (PMA, Sigma). Before adding the bacteria, cells were washed with PBS and fresh RPMI was added. Infections performed as described above. Cells were incubated for 30 minutes; then media was changed to RPMI supplemented with gentamycin. The cells were then incubated further for selected periods of time and then washed and lysed as described for Hep-2 cells.
Electron microscopy. For microscopic analysis of infected cells, cells were seeded onto glass coverslips and infections were performed as described above. Samples were prepared for scanning electron microscopy (SEM) as previously described 32 and for transmission electron microscopy (TEM) negative staining as previously described 33 . Images were collected using a 120 kV FEI Spirit Biotwin with a Tietz F4.15 CCD camera and the flagellum dimensions calculated using version 3 of TEM Tecnai software.
Ruffling signaling cascade. A day prior to transfection, 2 3 10 4 Hep-2 cells per well were seeded into 24-well plates. Immediately prior to transfection, fresh media was added to the wells. Transfections were performed using Lipofectamine TM RNAiMAX After successful transfection, cells were infected with S. Typhi as described above for a period of one hour. After infection, cells were washed with PBS and processed for Western blotting and SEM. For Western blotting, cells were lysed with 100 mL cells lysis buffer (10 mM Tris-Cl pH 7.6, 5 mM EDTA, 150 mM NaCl, 0.5% Triton X-100, complete mini EDTA-free protease inhibitor (Roche)) and then centrifuged for 10 minutes at 14,000 rpm at 4uC. The supernatant was recovered and mixed with loading buffer. SDS-PAGE was performed using 15% acrylamide gels. Membranes were blocked in 0.05% PBS and 5% Tween BSA. Anti-RAC1 (1/1,000, mouse, Millipore); anti-CDC42 (1/1,000, rabbit, Cell Signalling Technologies); anti-RHOG (1/250, rabbit, Santa Cruz Biotechnologies) and anti-actin (1/10,000, rabbit, Sigma) were used as primary antibodies.
Microarray analysis. Both THP-1 and Hep-2 cells were seeded and infected with S. Typhi as described above for a period of one hour. The strains used for infection were BRD948, BRD948-Hj, BRD948-Hz66, BRD948-DfliC and, in the case of Hep-2 cells, BRD948-DinvA. An uninfected control well was also included. After infection, the cells were washed and RNA was purified using the RNeasy Mini Kit (Qiagen), as per manufacturer's instructions. RNA samples were then amplified and labeled using Illumina TotalPrep 96 kit (Ambion, Austin, TX, USA) and hybridized onto Illumina TM Human HT-12_V4 Beadchips (Illumina, San Diego, CA, USA). The chips were scanned on an Illumina BeadArray Reader and raw intensities were extracted using Illumina BeadStudio Gene Expression Module. Microarray data are available in the ArrayExpress database (www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-2395.
Normalization and data analysis of the microarrays were performed using GeneSpring X software (Agilent Technologies). A quantile normalization using a baseline correction from the median of all samples was performed. For each comparison, differentially expressed genes were defined as those exhibiting a fold change $ 2 and a FDR (false discovery rate) corrected p-value # 0.05. Adjusted p-values were calculated using the Benjamini and Hochberg method 34 . Pathway, gene ontology (GO), and interaction analysis was performed using InnateDB (www.innatedb.ca). Over-represented pathways or GO terms were deemed significant if having a FDR corrected p-value # 0.05. TLR5 signaling. TLR5 signaling was measured by using HEK293 cells (human embryonic kidney cells, ATCC number CRL-1573) transfected with a TLR5 expression vector, an NFkB-luciferase reporter (firefly luciferase, Stratagene) a Renilla-luciferase reporter as a transfection control (Promega), and a 'filler' plasmid pEF-BOS, using Fugene 6 (Roche). Twenty-four hours after transfection, the cells were incubated with serially diluted overnight cultures of the relevant bacteria (BRD948-Hd, BRD948-Hj, BRD948-Hz66,BRD948-DfliC and Salmonella Typhimurium UK1 (positive control)). After six hours infection, the cells were washed and incubated with firefly and TK-Renilla luciferase substrates. Cells were then infected with Luciferase activity was measured using the Luciferase Assay System (Promega) as per manufacturer's instructions. The optimal bacterial dilution was defined as the dilution at which the greatest differential in the expression of the two luciferases for the positive control was provided. The same bacterial concentrations for the different mutants were then compared for pNFkB-luciferase expression; results are presented as NFkB-luciferase activity relative to Renilla-luciferase activity.
www.nature.com/scientificreports SCIENTIFIC REPORTS | 5 : 7947 | DOI: 10.1038/srep07947 Anti-flagellin ELISA. The metadata associated with the available serum samples for serology are as previously described 26,31,35 . Briefly, the serum from the typhoid cases with a range of S. Typhi variants was collected as part of a case/control study conducted by Vollaard et al. 26 , all these patients had positive blood-culture containing S. Typhi. Community controls were selected from the local community at random and as previously described 35 . Strains isolated from blood cultures of typhoid patients were assessed by PCR amplification to detect the nature of the native flagellin gene(s) 31 . For this work, thirty typhoid cases with known flagella variants and 79 randomly chosen community controls were selected for serological analysis.
For the ELISA assays, Nunc MaxiSorp (Thermo Scientific) or Microlon (Greiner) 96-well plates were coated with 2 mg mL 21 of S. Typhi flagella antigens (Hd, Hj or Hz66, purified as described previously) in phosphate buffer pH 9.5 and incubated overnight at 4uC. The plates were blocked with 1% BSA (Sigma) in PBS-0.05% Tween (Sigma) for 1 hour at 37uC. Sera were added in serial dilutions in PBS-0.05% Tween-0.1% BSA and incubated for two hours at 37uC. The secondary antibodies (rabbit anti-human IgG-HRP conjugated (Dako), mouse anti-human IgG1-biotin conjugated (Sigma) or mouse anti-human IgG2-biotin conjugated (Sigma)) were added at a 1/1,000 dilution in PBS-0.05% Tween-0.1% BSA and incubated for two hours at 37uC. Plates were developed with OPD (SIGMAFAST TM OPD, Sigma) at room temperature for 10 minutes, following the manufacturers' instructions. The reaction was stopped with 20 ml/well 3 M H 2 SO 4 . Plates were read at 490 nm using an ELISA plate reader.
Statistical analysis. Statistical tests were applied to determine differences between the strains with respect to flagella morphology, motility, and cellular invasion. The nature of these tests is outlined in the results with the corresponding p-value. ANOVA and 2sided t-tests were performed using GraphPad Prism (GraphPad Software Inc.), for multiple of pairwise comparisons, respectively. No correction for multiple testing was applied. p values of # 0.05 were considered to be statistically significant.