MAIT cell clonal expansion and TCR repertoire shaping in human volunteers challenged with Salmonella Paratyphi A

Mucosal-associated invariant T (MAIT) cells are innate-like T cells that can detect bacteria-derived metabolites presented on MR1. Here we show, using a controlled infection of humans with live Salmonella enterica serovar Paratyphi A, that MAIT cells are activated during infection, an effect maintained even after antibiotic treatment. At the peak of infection MAIT cell T-cell receptor (TCR)β clonotypes that are over-represented prior to infection transiently contract. Select MAIT cell TCRβ clonotypes that expand after infection have stronger TCR-dependent activation than do contracted clonotypes. Our results demonstrate that host exposure to antigen may drive clonal expansion of MAIT cells with increased functional avidity, suggesting a role for specific vaccination strategies to increase the frequency and potency of MAIT cells to optimize effector function.

M ucosal-associated invariant T (MAIT) cells are nonclassical T cells with both innate and adaptive immune properties 1 . These cells respond to a recently identified class of antigens, vitamin B metabolites derived from bacteria, that are presented on the MHC class I-like molecule MR1 2 . This discovery, along with an abundance in humans 3,4 , conservation across species [5][6][7] , and the defective immune response of MR1/ MAIT cell-deficient mice to infection by Francisella tularensis 8 , Klebsiella pneumoniae 9 , and Mycobacterium tuberculosis 10 , has led to the notion that these cells have an important function in controlling bacterial infections.
Most studies of MAIT cells in bacterial infection have been either in vitro 10,16,17 or in vivo mouse models 10,8,18 . These studies have demonstrated that MAIT cells respond to bacteria through upregulation of activation markers and proinflammatory cytokine production, and only in response to bacteria that contain the riboflavin synthesis pathway 16,18 , which provide a source of stimulatory vitamin B metabolites. These findings are important, as MAIT cells can also become activated in an MR1-independent manner through cytokine stimulation 19,20 . Collectively, these studies suggest that the presence of the ligand is important for MAIT cell response to bacterial infection. Most of the few human studies of MAIT cells are cross-sectional, requiring comparison to age-matched healthy controls 16,21,22 . The only longitudinal studies of MAIT cells in response to bacterial infection involve the challenge of human volunteers with live attenuated Shigella 17 , live Salmonella enterica serovar Typhi 23 , and infection of non-human primates with M. tuberculosis 24 . Although these studies describe changes in the frequency and activation of MAIT cells, the TCR repertoire stability and clonal response of MAIT cells is yet to be explored.
MAIT cells are activated during enteric fever. We next examined the activation of circulating MAIT cells in challenged individuals by expression of CD38. There was no increase in the expression of CD38 over time on MAIT cells from individuals who were not diagnosed with enteric fever (Fig. 1d). In contrast, the expression of CD38 was significantly increased on 16.8 ± 4.2% of MAIT cells 9-11 days after challenge in individuals that developed enteric fever (Fig. 1e). This significant upregulation of CD38 was maintained, even following antibiotic treatment where 16.3 ± 2.5% of MAIT cells were still expressing CD38.
To further characterize the MAIT cell phenotype during infection, mass cytometry analysis was performed on blood samples from four diagnosed individuals. We found that MAIT cells expressed the intracellular proliferation marker Ki67 following diagnosis ( Supplementary Fig. 3), similar to observations made in previous bacterial and viral infection Interestingly, very few Ki67 + cells were observed following antibiotic treatment, although 27.7 ± 8.0% of MAIT cells still retained CD38 expression ( Supplementary Fig. 4). Additional analysis of surface markers revealed no substantial differences in the expression of additional activation markers or tissue-homing markers between CD38 + Ki67 + and CD38 − Ki67 − cells at the peak of infection ( Supplementary Fig. 3). Together, the decreased frequency and acquisition of an activated and proliferating phenotype suggests that circulating MAIT cells in individuals diagnosed with enteric fever are responding to the presence of S. Paratyphi A.
MAIT cell TCRβ usage remains oligoclonal during enteric fever. In vitro experiments to study the activation of MAIT cells isolated from healthy volunteers indicated that activation was partially antigen-dependent as addition of MR1 blocking antibody could partially reduce the MAIT cell response to S. Paratyphi A infection ( Supplementary Fig. 5). The MAIT cell activation response to S. Paratyphi A was also partially IL-12dependent, and a combination of both MR1 and IL-12p40 blocking antibodies prevented MAIT cell activation (Supplementary Fig. 5). This suggested that MAIT cell in vitro response was not only driven by antigen recognition, but also by cytokinedependent activation.
We analyzed the TCRβ repertoire of six individuals challenged with S. Paratyphi A and diagnosed with enteric fever to determine whether the MAIT cell response in vivo was TCR-dependent, and therefore potentially antigen-driven. Consistent with previous findings 11,12 , we found the TRBV usage of MAIT cells was oligoclonal, characterized by the preferential use of TRBV20 and TRBV6 in all diagnosed individuals, with inter-donor variation ranging from 31 to 84% of the total chain usage (Fig. 2a, b). This oligoclonality was maintained during infection and was also evident to a lesser extent in the TRBJ usage (Fig. 2c). This is similar to the TCRβ chain usage observed in healthy (not challenged, control) individuals over time ( Supplementary Fig. 6). These results indicate that the TCRβ usage by MAIT cells is not influenced by the presence of S. Paratyphi A and that if changes were occurring in MAIT TCRβ repertoire that it would be on the clonotypic level.
To determine whether changes were occurring in MAIT cells at the clonotypic level, we analyzed circulating MAIT cell clonotypes based on the CDR3β sequences in six diagnosed individuals. In keeping with the findings of Lepore et al. 12 , we confirmed that the MAIT cell clonotypes were stable over time, shown by the example clonotype scatterplot of a healthy control individual (Fig. 3a). However, MAIT cells in the diagnosed individual preand post-infection scatterplot appear less stable over time.
To explore this further, we analyzed the clonal space homeostasis of MAIT cell clonotypes over time to understand whether the distribution of the clonotypes underwent changes in response to infection. In a healthy control individual, the clonal distribution was relatively stable over time. In contrast, when we analyzed the clonal distribution in a diagnosed individual (donor 1), we observed marked changes, particularly in the proportion of space used by the over-represented MAIT cell clonotypes (Fig. 3b). When comparing the clonal distribution in all diagnosed individuals, there was a significant decrease in the proportion of over-represented clones from 0.17 ± 0.05 at baseline to 0.10 ± 0.04 mid-infection, but there was no significant difference between the baseline and post-infection (Fig. 3c). There were also no significant changes to the MAIT cell clonal distribution in the undiagnosed individuals ( Supplementary  Fig. 7). These results demonstrate that changes to MAIT cell clonotypes in response to infection occur most notably to those that are over-represented.

MAIT cells clonotype composition alters during enteric fever.
To determine whether the over-represented MAIT cell clonotypes following infection were different to those present before infection, we analyzed the circulating MAIT cell clonotype composition based on the CDR3β sequence in six diagnosed individuals. An example clonotype tracking heatmap for a diagnosed individual (donor 1) shows that the clonotypes present at high frequency following infection were not the same clonotypes that were at high frequency before infection (Fig. 4a). A shared clonotype abundance plot with the top five clonotypes displayed for two diagnosed individuals shows clonal expansion and contraction occurring, with a predominating clonotype present postinfection in individual donor 1 (Fig. 4b). Shared clonotype abundance plots for all diagnosed individuals are shown in Supplementary Fig. 8. When comparing the top five clones across all diagnosed individuals, we observed significantly greater proportion of top clonotypes undergoing clonal expansion/contraction (0.53 ± 0.11) compared to healthy controls (0.13 ± 0.07) (Fig. 4c). These results demonstrate that the composition of MAIT cell clonotypes in circulation changes in response to S. Paratyphi A infection.

Expanded MAIT cell TCRβ clonotypes have greater activation.
To investigate whether the MAIT cell TCRβ clonotypes that expanded in diagnosed individuals in response to S. Paratyphi A conveyed a functional advantage in their response to antigen, we generated an in vitro model by transducing JRT3 cells with lentiviral vectors containing MAIT cell TCR sequences of interest. We produced lines expressing the most common MAIT TCRα chain TRAV1-2-TRAJ33 ( Supplementary Fig. 9) and the MAIT cell TCRβ clonotypes selected from donor 1 and donor 3. The clonotypes chosen were those that had expanded from pre-to post-infection by a large fold-change and compared this with a clonotype that had contracted by a large fold-change and used the most common TCRβ chain (TRBV20-1). We tested their activation in response to bacterial supernatant and riboflavin metabolite ligands. Details of the TCRβ clonotypes selected are Escherichia coli has been previously shown to produce MAIT cell-activating ligands through the riboflavin synthesis pathway 26 . As S. Paratyphi A also contains the riboflavin synthesis pathway they would be expected to also produce MAIT cell-activating ligands. In response to either S. Paratyphi A or E. coli overnight culture supernatant, the expanded TCRβ clonotypes showed greater activation compared to the contracted clonotypes (Fig. 5a). All lines displayed a dose-dependent response and their activation could be completely blocked with the use of an MR1 blocking antibody ( Supplementary Fig. 12). This suggests that the expanded TCRβ clonotypes have greater activation, but this is not an exclusive response to S. Paratyphi A, as the clonotypes do not convey preferential specificity between the different bacteria.
To investigate whether the extent of Jurkat.MAIT activation to bacterial supernatant by the different TCRβ clonotypes correlated to their activation by stimulatory riboflavin metabolite ligands, we stimulated these cells with the strong precursor MAIT ligand 5amino-6-(D-ribitylamino)uracil (5-A-RU) plus methylglyoxal and the weaker ligand 6,7-dimethylribityl lumazine (RL-6,7-DiMe) (Fig. 5a). Both lines showed a dose-dependent and MR1-driven response to the ligands (Supplementary Fig. 13). However, the expanded TCRβ clonotypes showed a greater activation overall by both ligands when compared to the contracted clonotype, particularly for donor 1 (which was further confirmed by studying the rate of dissociation of 5-(2-oxopropylideneamino)-6-D-ribitylaminouracil (5-OP-RU) MR1 tetramers bound to Jurkat.MAIT lines in the presence of excess anti-MR1 blocking antibody, shown in Supplementary Fig. 14). This pattern of activation reflects what we observed in the bacterial supernatant experiments, suggesting that the riboflavin derivatives were the primary antigen that the Jurkat.MAIT lines were responding to in the bacterial supernatant.
When co-culturing the Jurkat.MAIT lines with C1R.MR1 cells without addition of bacteria or ligand, we observed CD69 expression on the expanded Jurkat.MAIT line from donor 1 was fivefold higher compared to when cultured alone (Fig. 5b). This baseline activation, that can be blocked by MR1 blocking antibody, suggests that this clonotype was activated by components in the cell culture media. This was determined to be due to the vitamin metabolites present in the FBS, as the use of dialyzed FBS, which fails to alter activation of HLA-restricted T cells ( Supplementary Fig. 15), reduced this baseline response (Fig. 5c).
Together, these results suggest that certain MAIT cell clonotypes may expand in response to infection due to their greater functional avidity of their TCR to MAIT cell ligands.

Discussion
MAIT cells and their bacterial reactivity are conserved across species, indicating their vital role in the host immune response to bacterial infections 6 . In this study, we had a unique opportunity to further our understanding of MAIT cells and their clonal response in bacterial infections as we utilized samples from a controlled human infection model of the invasive bacterial pathogen S. Paratyphi A in healthy volunteers. This has provided We observed that the relative frequency of circulating MAIT cells significantly decreased early in individuals that developed enteric fever. This could indicate that, as early as day 4, the MAIT cells in these individuals are responding to the presence of S. Paratyphi A in the host. This migration of MAIT cells from circulation has been suggested previously by studies reporting lower frequency of circulating MAIT cells in infected patients compared with healthy individuals 16,28 . In addition to this, a decrease in circulating MAIT cell frequencies was observed in trials that administered live S. Typhi 23 and live-attenuated Shigella dysenteriae 1 vaccine 17 . The findings of these previous studies are in line with our observations, and suggest that circulating MAIT cells are responsive to in vivo bacterial infection and may decrease in frequency in the blood as they move to locally inflamed and infected tissues. Accumulation of MAIT cells in infected tissues has been demonstrated by studies using lung infection mouse models 8,18 . We also observed a recovery in MAIT cell frequency following antibiotic treatment, suggesting that MAIT cell redistribution and reduction in circulating numbers is reversible in bacterial infection. This observed MAIT cell recovery is unlike the non-reversible decrease of MAIT cells that is observed in human immunodeficiency virus (HIV) infection 29,30 .
We show that a proportion of circulating MAIT cells in diagnosed individuals were activated and proliferating at the peak of infection, indicating that they were playing a role in the host response to S. Paratyphi A. Similar observations have been made in studies assessing MAIT cell responses in vaccination and live infection trials 17,23,24 . However, the mode of activation of MAIT cells has not been addressed in these human in vivo studies. This is an important point as in vitro studies have shown that MAIT cells can become activated both through TCR stimulation or through TCR-independent cytokine stimulation 19,31 , which we also observed after in vitro S. Paratyphi A infection.
To address whether the activation we observed was due to the MR1−TCR interaction, and thus due to the presence of bacteriaderived ligand, we assessed the MAIT cell TCR repertoire and found that, although the TRBV maintained oligoclonality, the TCR repertoire had undergone reshaping and clonal contraction/ expansion at the clonotypic level during infection. This suggests that the TCR−MR1 interaction may be influencing the circulating MAIT cell population. There are several studies which have indirectly provided evidence that the TCR−MR1 interaction is important for MAIT cell responses to bacterial infection. In a study of critically ill patients, the circulating MAIT cell proportion in patients with streptococcal infections (bacteria which lack the riboflavin synthesis pathway) was higher than those with infections by bacteria which do produce riboflavin metabolites 28 . Also, a recent study using S. enterica serovar Typhimurium in a mouse lung infection model demonstrated that the MAIT cell response to infection required MR1 and the ability for bacteria to produce the riboflavin-derived ligand 18 . These results support the concept that the riboflavin metabolite ligands are important for the MAIT cell response to bacterial infection and that the changes we observed in TCR repertoire could be driven by the riboflavin metabolite antigen.
We observed an increased functional avidity of expanded MAIT cell clonotypes to the riboflavin ligand compared with a contracted clonotype from the same diagnosed individual. The increased functional avidity could, in part, explain the selective clonal expansion we observed in diagnosed individuals. The concept of MAIT cell selectivity based on TCR sequence has been suggested previously by a study assessing MAIT cell TCR repertoire after in vitro infection 14 . However, this study also suggested differences in TRBV chain usage as well as suggesting ligand/pathogen discrimination, features which we did not observe in our study. Our results showed that the Jurkat.MAIT activation level differed between clonotypes, but that this activation was relatively conserved, regardless of whether they were stimulated with S. Paratyphi A, E. coli, or different riboflavin metabolites. Our observations are supported by the findings from a MAIT TCR−MR1 structural biology study that demonstrated that the CDR3β region directly interacts with the MAIT cell ligand and could influence and "fine-tune" the MAIT cell response 15 .
It is likely that the increase in TCR functional avidity for riboflavin metabolites is not exclusively the driving factor for preferential expansion of certain MAIT clonotypes. A driving factor, which we were not able to explore, was the cytokinedependent activation that we demonstrated in the in vitro S. Paratyphi A challenge assay. Recent advances in methodology that can couple TCR sequence with transcriptional profiling of single cells 32 could help elucidate this, along with other factors that could contribute to MAIT cell clonal expansion in response to infection.
In summary, we have determined the timing and dynamics of the MAIT cell response to S. Paratyphi A infection. We have clearly shown that MAIT cells undergo an adaptive clonal response during infection, yet are still driven by their specificity to a narrow set of vitamin metabolite antigens. These findings demonstrate how MAIT cells sit directly at the interface between innate and adaptive immunity and that lifetime exposure to different microbial infections could impact and shape the host's MAIT cell TCR repertoire. To understand the functional implications of this, future experiments should assess whether select MAIT cell clonotypes that expand in response to S. Paratyphi A infection may play a protective role in subsequent infections with S. enterica or other bacterial species.

Methods
Participants and human infection model. The controlled human S. Paratyphi A infection model utilized has been previously described 25 . Briefly, adult participants (aged 18-60) ingested a single dose of 1-5×10 3 colony-forming units (CFU) of S. Paratyphi A strain NVGH308 suspended in NaHCO 3 (0.53 g/30 mL) solution. The following 14 days involved daily follow-up and blood monitoring. A diagnosis of enteric fever was defined as S. Paratyphi A bacteremia and/or persistent fever (>38°C ) lasting >12 h. Participants were treated with ciprofloxacin (500 mg, twice daily for 14 days) at the time of diagnosis, or at day 14 if they remained asymptomatic and were not diagnosed. Healthy adult blood donor controls were from laboratory volunteers or leukocyte cones obtained from NHS blood and transplant service. Peripheral blood mononuclear cells (PBMCs) were isolated from all blood samples and leukocyte cones by gradient centrifugation using Lymphoprep™ (AxisShield).
The C1R.MR1 cell line was produced by cloning the MR1 gene into a pHR-IRES-GFP vector which was then co-transfected into 293T cells with the HIV gagpol and VSV-G expression plasmids using FuGENE 6 Transfection Reagent (Roche) according to the manufacturer's instructions. The supernatant from this culture containing the lentiviral particles was then used to transduce C1R cells that were then sorted based on high expression of GFP. MR1 expression was then confirmed by MR1 antibody staining and flow cytometry.
The Jurkat.MAIT cell lines were produced by cloning into a pHR-IRES vector the TRAV1-2-TRAJ33 (CDR3α: CAVMDSNYQLIW) MAIT cell TCRα chain and then cloning in the TCRβ chain of interest. The vector was then co-transfected into 293T cells with the HIV gag-pol and VSV-G expression plasmids using X-tremeGENE™ 9 DNA Transfection Reagent (Sigma) according to the manufacturer's instructions. The supernatant from this culture containing the lentiviral particles was then used to transduce JRT3 cells (which lack an endogenous TCRβ chain). TCR expression and pairing with MAIT cell TRAV1-2 (Vα7.2) chain was confirmed by flow cytometry and cells sorted based on CD3 and Vα7.2 expression.
Preparation of bacterial stocks. All work involving S. Paratyphi A was carried out in a CAT3 facility. S. Paratyphi A strain NVGH308 and E. coli strain DH5α were plated on Lysogeny broth (LB) agar plates overnight at 37°C. A single colony was picked, plated, and incubated overnight at 37°C to ensure single origin colony. A single colony was chosen and grown in LB+10% sucrose overnight. LB+10% sucrose was inoculated with the overnight culture 1:100 and grown for 3 h (to logphase growth point). Frozen aliquots were stored at −80°C and thawed just prior to use in in vitro experiments. The CFU for each bacterial stock was determined and used for all multiplicity of infection calculations. Tetramer staining and decay assay. The MR1 5-OP-RU and 6-formylpterin (6-FP) PE-conjugated tetramers were provided by the NIH Tetramer Core Facility. For staining PBMCs, cells were stained with LIVE/DEAD ® Fixable Aqua Dead Cell Stain Kit, followed by incubation with the MR1 tetramer (5 μg/mL) for 40 min at room temperature followed by surface antibody staining for 20 min at 4°C. For the tetramer decay assay, Jurkat.MAIT lines were stained with LIVE/DEAD ® Fixable Aqua Dead Cell Stain Kit followed by incubation with 5-OP-RU MR1 tetramer (2.5 μg/mL) for 40 min at room temperature. Excess tetramer was washed off and cells re-suspended in buffer with or without 20 μg/mL anti-MR1 antibody (clone 26.5, Biolegend). Samples were left at 37°C and periodic samples taken, washed and fixed immediately. Once all samples were collected, CD3 surface staining was performed.

In vitro infection.
Mass cytometry sample preparation. Frozen PBMCs (2-3 million) from challenged volunteers were thawed and washed with RPMI-1640 media supplemented with 1% FBS, 1% penicillin/streptomycin, 2 mM L-glutamine, 1% 1 M HEPES and 50 μM β-mercaptoethanol. Cells were washed twice in cold PBS, and incubated on ice with 200 μM cisplatin for 5 min. Cells were then washed with CyFACS buffer (PBS+4% FBS+0.05% sodium azide) and stained with α-GalCer-Eu153 tetramer (that was produced from α-GalCer monomers that were tetramerized using streptavidin conjugated with 153Eu) for 30 min at room temperature in the presence of 10 μM free biotin. This was followed by a 30-min incubation in primary antibody mix (Supplementary Table 1) on ice. Cells were then washed twice in CyFACS and stained with metal-tagged surface antibodies (Supplementary Table 2). After 30 min, the cells were washed twice with CyFACS and incubated in Foxp3 Fixation/Permeabilization buffer (eBioscience) on ice for 30 min to enable intranuclear staining. Cells were then washed twice in Permeabilization (perm) buffer (Biolegend) (as this is the permeabilization buffer optimized for our CyTOF protocol) and stained with metal-tagged intracellular/intranuclear antibodies (Supplementary Table 3) for 30 min on ice. After washing twice in perm buffer, cells were incubated on ice with metal-tagged streptavidin-155 for 10 min. Cells were then washed twice in perm buffer, once in PBS, and fixed in 2% paraformaldehyde (PFA) at 4°C. The following day, cells were washed twice with perm buffer and once with PBS.
Mass cytometry sample barcoding. We next barcoded stained samples to eliminate doublets and be able to run multiple samples simultaneously. Cells were incubated with cellular barcodes for 30 min as previously described 33 . Pt-102, Rh-103, Pd-104, -105, -106, -108, and -110 were used for this experiment. 113 was not used due to interference with CD57 on mass 115, based on our observations from previous experiments. A small degree of interference was observed between 110 and 112/114 but this did not affect de-barcoding as we could easily gate on 110 positive cells without including 112/114 positive cells. In addition, since CD14positive cells (112 and 114 positive) were excluded for analysis, interference between 110 and 112/114 was not an issue.
Mass cytometry sample batching and acquisition. Cells were then washed once with perm buffer and incubated in CyFACS for 10 min on ice. Cellular DNA was labeled at room temperature with 250 nM iridium interchelator (Fluidigm) diluted in PBS with 2% PFA. After cellular DNA labeling cells were washed twice with CyFACS and kept in a 96-well U-bottom plate.
To accommodate the required numbers of samples, two barcoded batches were prepared, ensuring each batch included all time-points for a given volunteer. A single healthy donor's PBMCs, that was prepared and stained in parallel with volunteer samples, was included in each batch as a control for batch-to-batch variation.
For acquisition, small aliquots of cells from each batch were filtered through a 0.35 μM cell strainer into a 5 mL polystyrene tube. The batch aliquot was then washed twice in water before being resuspended in water at 500,000 cells/mL. EQ Four Element Calibration Beads (Fluidigm) were added at a final concentration of 1% prior to sample acquisition. The same pooling steps were repeated each time the acquisition of an aliquoted batch was completed. This approach was taken to minimize the duration the stained cells were kept in water (1-2 h maximum).
Both batches of samples were acquired on subsequent days to minimize batch effects which can occur if stained cells are stored in CyFACS long term. Batch effect was analyzed by performing a tSNE plot to compare the PBMCs from the healthy donor stained and acquired with the two distinct batches. Cells acquisition was performed on a CyTOF2 (Fluidigm) and the total number of live cells acquired for each sample was between 30,000 and 100,000.
Mass cytometry data analysis. After mass cytometry acquisition, the data was exported in flow cytometry file format, normalized 34 and events with parameters having zero values were randomized using a uniform distribution of values between −1 and 0. Each sample containing a unique combination of two metal barcodes was de-convoluted using manual gating in FlowJo to select cells stained with two (and only two) barcoding channels. MAIT cells were identified as CD14 − CD19 − CD3 + Vα7.2 + CD161 + and marker expression assessed using Cytobank (MRC) software. The total number of MAIT cells identified in each sample ranged from 83 to 346.
Activation assay for assessing dialyzed FBS. 100,000 CD4 + T cells specific for the S. enterica protein CdtB were stimulated with 3 μg/mL CdtB 105-125 peptide for 18 h in culture media containing either normal or dialyzed FBS (Gibco). Cells were then stained for flow cytometry using anti-CD69 (A488)-conjugated antibody.
TCRβ repertoire library preparation. Cells for TCR repertoire analysis were sorted directly into lysis buffer from RNAqueous ® -Micro Total RNA Isolation Kit (Ambion). Sorted cell numbers ranged from 5000 to 10,000 cells per sample. RNA was then extracted using the RNAqueous ® -Micro Total RNA Isolation Kit following the manufacturer's instructions. The unbiased TCR amplification method was used as previously described 35 . Briefly, template-switch cDNA was produced using SMARTScribe™ Reverse Transcriptase (Clontech) with switch_oligo and BC1R primers described in Table 2. First and second PCR steps were conducted using Phusion ® High-Fidelity DNA Polymerase (New England Biolabs) and primers described in Table 2. Samples were pooled and NEBNext ® Ultra™ DNA Library Prep Kit for Illumina ® (New England Biolabs) was used following the manufacturer's instructions. Samples were sequenced on Illumina MiSeq platform using MiSeq reagent kit V2 300 cycle (Illumina).
TCRα repertoire library preparation. cDNA was synthesized using the oligo-dT primer method of the RETROscript ® Kit (Ambion) following the manufacturer's instructions. The TCRα chain was amplified by PCR using Phusion ® master mix and primers specific for TRAV1-2 (TRAV1-2_FW) and the TRAC region (AC1R_RV) primer details outlined in Table 2. The PCR products were used for a second nested PCR using primers Smart_TRAV1-2_FW and AC2R_RV. PCR product was used as a template for a third nested PCR using primers Step_1_FW and Hum_bcj_RV containing a barcoded sequence. Samples were then pooled for MiSeq analysis as described in TCRβ library preparation.
TCR repertoire analysis. The sequencing files generated by the MiSeq were split based on their barcoded sequence using Jemultiplexer (GBCS) and then run through the MiTCR software 36 , aligning TCRα/β sequences using default parameters. VDJ Tools 37 was then used to filter out the non-functional sequences prior to post-analysis visualization. Clonotypes were defined based on their amino acid sequence. Frequencies were determined by number of assigned sequence reads. Default strict sample intersection rules were applied to VDJ tools analysis.
Statistical analysis. Statistical analysis was performed using Prism software (Graphpad). For multiple comparisons, one-way ANOVA was used with Dunnett's test to correct for multiple comparisons. For comparison between two groups, either Student's paired two-tailed t-test or unpaired two-tailed t-test was used. Significant result is indicated where *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Sample sizes were limited by the availability of volunteer samples but adequate to detect differences between sample groups, as indicated by the reproducibility and variability of each experiment.
Ethics. Written informed consent was obtained for all study volunteers in accordance with the Declaration of Helsinki and ethical approval was received from Oxford Research Ethics Committee (14/SC/0004 and 14/SC/1204). Samples obtained from the controlled human S. Paratyphi A infection model were part of registered clinical trials (NCT02100397 and NCT02192008) and conformed to the Good Clinical Practice (GCP) guidelines.
Data availability. Sequence data that support the findings of this study have been deposited in NCBI Sequence Read Archive with the primary accession code PRJNA412739.