A randomized, double-blind, placebo-controlled trial was performed at the Centre for Clinical Vaccinology and Tropical Medicine, Churchill Hospital, Oxford, UK (clinicaltrials.gov NCT679172; EudraCT 2011-000381-35) (14). This study was carried out in accordance with the recommendations of the Oxford University Clinical Trials and Research Governance Department, and the protocol approved by NRES South Central––Oxford A (11/SC/0302) and conducted in accordance with the declaration of Helsinki (2008) and the International Conference of Harmonization of Good Clinical Practice guidelines. An independent Data Monitoring and Safety Committee monitored the trial. To perform the assays described, human complement was collected in a parallel, prospective observational study (clinicaltrials.gov NCT01945307; 13/SC/0375). All subjects gave written informed consent in accordance with the Declaration of Helsinki.

The study design and results of the primary study objectives of the clinical trial have been reported previously (14). Briefly, 99 adult participants were randomly assigned in the ratio 1:1:1 to three study arms: (1) vaccine (a single oral dose of M01ZH09 vaccine containing 1 × 10¹⁰ CFU live attenuated S. Typhi M01ZH09 strain), (2) placebo, and (3) comparator (three doses of Ty21a capsules each containing 2 × 10⁹ CFU). Participants were challenged with 10⁴ CFU of wild-type S. Typhi Quailes strain 4 weeks after vaccination (D0) (15). Antibiotic treatment was initiated at typhoid diagnosis (TD) or 14 days after challenge in those participants who were not diagnosed with typhoid fever (nTD). Diagnosis of TD was defined as confirmed S. Typhi bacteremia or by ≥12 h of sustained fever ≥38°C (14, 15) and full protection was defined as not meeting the criteria for diagnosis (nTD).

To prepare a source of human complement, fresh serum from a UK healthy adult donor was depleted of S. Typhi-binding antibodies that may otherwise influence the performance of complement in our assays, as modified from Hart et al. (11, 18). Following three rounds of adsorption with 10¹⁰ CFU of log phase S. Typhi Quailes per 1 mL of serum, significant depletion of anti-Vi and anti-O:LPS antibody titers was achieved and demonstrated by ELISA (Table S1 in Supplementary Material), without change in complement activity, as measured by CH100 and AP100 hemolytic assays (data not shown). To prepare aliquots of log-phase S. Typhi Quailes for the serum bactericidal antibody (SBA) assay, an overnight culture, set up from a single colony in Luria Bertani broth (LB), was diluted 33-fold in LB supplemented with 10% sucrose. This was grown at 37°C shaking to achieve 3.5 × 10⁸–3.5 × 10⁹ CFU/mL and frozen in aliquots ahead of use in SBA.

Serum samples were evaluated in the SBA from a subset of participants from whom serum was available and representing similar numbers across all study arms over multiple time points (Placebo n = 18, Ty21a n = 16, M01ZH09 n = 18, TD n = 30, nTD n = 22). Time points were at pre-vaccination (D28), at challenge (D0), and post-challenge days 28 (D28), 60 (D60), and 180 (D180). Briefly, serum samples were de-complemented by heat-inactivation and diluted before addition of 200 CFU of log phase S. Typhi Quailes in Hanks Buffered Saline Solution (HBSS; Life Technologies Ltd., UK) supplemented with 0.5% Bovine Serum Albumin (BSA; Sigma-Aldrich, UK). S. Typhi-specific antibody-depleted human complement serum was added to a final complement concentration of 25% and bacteria incubated for 1 h at 37°C with shaking before plating on tryptic soya agar (TSA) plates (Oxoid Ltd., UK). In each experiment, complement serum and heat-inactivated complement serum alone were included as controls for bacterial survival; test samples without complement were included to assess any complement-independent killing; and a positive control, from another UK healthy donor, to assess inter-plate variability. Only experimental plates in which the positive control was within twofold of its defined SBA titer and with >60% overlap in CFU counts between the complement only and heat-inactivated complement only controls were accepted. The SBA titer was defined as the reciprocal lowest dilution of test serum to achieve ≤50% bacterial killing relative to complement alone. The slide agglutination test using Salmonella Vi anti-sera (Oxoid Ltd., UK) confirmed that Vi was expressed in the bacterial working stock. In nTD participants bactericidal activity a month after challenge (D28) was not measured, as preliminary experiments confirmed complete bactericidal activity consistent with residual ciprofloxacin in the samples as expected based on its pharmacokinetic profile (D28 coincided with the last day of antibiotic treatment for this subset) (19).

Lipopolysaccharide (O9:LPS) was from Sigma-Aldrich, UK (L2387, >99% free of protein) and Vi capsular polysaccharide was from Sanofi Pasteur, UK (Typhim Vi^®). Flagellin (H) was purified from a flagellin-deficient S. Enteritidis strain that expresses S. Typhi H:d from a plasmid (University of Maryland, USA) (14, 20). Cytolethal distending toxin B (CdtB) and pilus control protein L (PilL) coding sequences lacking transmembrane domains were PCR-amplified from S. Typhi CT18 strain genomic DNA, cloned into plasmid pET28b(+) (Novagen, UK) and expressed in Escherichia coli BL21(DE3)pLysS (Promega, WI, USA). Recombinant His-tagged CdtB and PilL were purified by tagging with nickel-coated agarose beads (Ni-NTA, Invitrogen) and elution from gravity flow columns (Qiagen, Germany). CdtB was renatured in 50-mM sodium phosphate solution and 500-mM NaCl. hlyE was PCR-amplified from Ty21a, cloned into pET21a vector (Novagen, UK), and expressed in an LPS-modified E. coli BL21 strain to produce endotoxin free proteins (Lucigen, ClearCoil BL21 cells). Recombinant His-tagged hemolysin E (HlyE) was purified using HisTrap nickel-affinity column (GE Healthcare) followed by desalting on a HiPrep 26/10 (GE Healthcare).

Immunoglobulin G (IgG), IgA, and IgM isotype responses to O9:LPS, H, HlyE, CdtB, and PilL were measured in serum as previously described (15, 21), with some modifications regarding antigen source and controls. Nominal ELISA units were calculated based on standard curves derived from pooled sera collected from the highest responders to O-antigen following vaccination with M01ZH09 (Emergent BioSolutions, Reading, UK). In addition, Vi-specific IgG were measured at D28, D0, and D28 using a commercial ELISA kit (VaccZyme™ Human anti-S. Typhi Vi IgG Enzyme Immunoassay Kit, The Binding Site Ltd., Birmingham, UK) following the manufacturer’s instructions.

In a subset of samples representing all study arms and both TD and nTD groups, antibodies specific to H or O9:LPS, or total IgA or IgG antibodies were depleted from the serum to assess the contributions of each of these antibodies to the bactericidal activity in sera. Antibodies specific to H or O9:LPS were adsorbed on to 96-well Maxisorp plates (Nunc, Copenhagen, Denmark) pre-coated with 1 µg/mL H or 15 µg/mL O9:LPS antigen in carbonate bicarbonate buffer (Sigma, UK), for eight rounds of 30-min incubations. A plate coated with BSA was used to create mock depleted controls. Total IgA and IgG antibodies were depleted using the Dynabeads Antibody coupling kit (Life Technologies Limited, UK). Briefly, each diluted serum sample was incubated for 1 h with 2-mg magnetic beads coated with either 10 µg of goat polyclonal IgG antibody raised against human IgA or IgG (AbD Serotec, UK), or an goat polyclonal IgG (Santa Cruz Biotechnology, USA). The depletions resulted in approximately 56% (H), 87% (O9:LPS), 72% (IgA), and 93% (IgG) median reduction in the antibody titer, as measured with ELISA.

Antibody-secreting cell (ASC) responses to LPS, H, and Vi were measured at D28, 7 days after vaccination (D21), before challenge (D0), Day 7 after challenge (D7), and 48 h after the typhoid diagnosis visit (TD + 48) by ELISpot assay, as described in Ref. (14). Spots were photographed using the AID ELISpot optical reader (AID ELR03m, Autoimmun Diagnostika GmbH, Germany), manually counted by two observers, and expressed as spots per 10⁶ PBMCs. Zero ASC counts were given a nominal value of 0.1 and 0.1 added to all other observations.

Daily clinical review and temperature measurements were carried out for 7 days after vaccination and 28 days after challenge. In addition, participants collected symptom data daily (solicited for headache, feeling generally unwell, loss of appetite, abdominal pain, nausea/vomiting, myalgia, arthralgia, cough, diarrhea and constipation, and any unsolicited symptoms) (14). Symptom severity scores were calculated by summing maximum severity values assigned to each of up to 10 symptoms experienced daily during 14 days after challenge and deriving the mean (14).

Blood (10 mL) samples were collected for bacterial culture at each visit. Cultures were performed by the local hospital accredited pathology laboratories according to national standard operating procedures (22–24), and as previously described (14, 15). Quantitative blood culture samples were collected immediately prior to antibiotic treatment by inoculation of 10-mL blood into an ISOLATOR 10 tube (Alere, UK).

Cytokines were quantified in lithium-heparin plasma using a multiplex bead-based ELISA (Milliplex^® Human cytokine/chemokine magnetic bead panel, Merck Millipore, UK), following the manufacturer’s protocol with dilution of beads by twofold. Analytes tested were as follows: EGF, Fractalkine (CX3CL1), GROα (CXCL1), IFNα2, IFNγ, IL1β, IL1RA, IL2, IL6, IL8 (CXCL8), IL10, IL12p40, IL15, IL17A, IP10 (CXCL10), sCD40L, TGFα, TNFα, and VEGF. All samples were run in duplicate and were measured on a Luminex MAGPIX instrument (Luminex, Netherlands). Values were calculated from the average of duplicates which were above cytokine sensitivity limits and whose %CV was <30%.

SBA titers, ELISA antibody levels, and ASC counts were log₁₀-transformed before statistical analysis. Within groups, the log₁₀-transformed data were still skewed so non-parametric tests were used. All reported p-values are based on two-sided tests, with significance assumed at p < 0.05, without multiple testing corrections. Clinical, laboratory, and immunological data were collated using a clinical-trial database (OpenClinica, version 3.1). Data analysis was performed using R version 3.2.1 (25).

Sera collected pre-vaccination in all study participants showed bactericidal activity against wild-type S. Typhi, with a 2–log₁₀ range in titers but without significant differences between vaccine arms (Figure 1A; Table S2 in Supplementary Material). One month after vaccination, at D0, no increase in bactericidal activity was observed in either placebo or Ty21a recipients (Figure 1A). In contrast, a significant increase in bactericidal activity was measured 28 days after M01ZH09 vaccination (p = 0.001). This increase was principally attributable to the subset of M01ZH09 recipients who subsequently went on to be TD after challenge (p = 0.009, Figure 1B).

To identify differences in the immune responses induced by attenuated vaccines and wild-type S. Typhi, the effect of exposure to the challenge agent on levels of bactericidal antibody activity was examined. Sixty days after challenge bactericidal antibody activity was significantly higher than on D0 in the placebo and Ty21a arms in participants with TD after challenge (p = 0.009 and p = 0.035, respectively; Figure 1B). The bactericidal activity at D60 in sera from nTD placebo and nTD Ty21a recipients showed marked variability in inter-individual responses. In contrast, D60 titers in M01ZH09 recipients remained unchanged from D0 in TD participants and waned in nTD participants (p = 0.036; Figure 1C).

Evaluation of bactericidal activity in sera from TD individuals over a longer time course confirmed that exposure to the challenge strain-induced bactericidal antibodies only in placebo and Ty21a recipients (Figure 1B). In both groups, titers peaked at D28 after challenge and returned to near pre-vaccination baseline levels by D180. M01ZH09-vaccinated TD participants had bactericidal titers persistently higher than baseline for at least 60 days after challenge, despite the apparent absence of any boosting effect by exposure to/infection with wild-type challenge.

We hypothesized that the presence of circulating bactericidal antibodies on the day of challenge may have a protective effect on challenge outcome. However, comparison of the bactericidal antibody titers between TD and nTD participants at D0 demonstrated no significant difference in protection (Figure 1D), even when assessed by vaccine allocation (Figure S1A in Supplementary Material). The similar titer ranges between TD and nTD groups indicated that serum bactericidal activity at D0, resulting either from preexisting exposure or vaccine induction, did not protect against TD.

The D0 bactericidal antibody titers in TD participants were positively correlated, however, with time to typhoid diagnosis after challenge, when vaccination arms were combined (Figure 2A). Furthermore, D0 bactericidal antibody titer in these participants negatively correlated with quantitative bacterial blood count at diagnosis and with cumulative symptom severity score following challenge, but not with temperature (Figures 2B,C; Table 1). We found no significant correlations when placebo TD was similarly analyzed (Table 1). Taken together, these observations suggest that higher bactericidal activity delayed the onset of infection and reduced disease severity in vaccinated participants.

The concentrations of various plasma cytokines associated with the acute phase of infection after S. Typhi challenge (26) were measured and analyzed to evaluate the impact of bactericidal activity on other events in disease pathogenesis. TNFα, IL6 and IL8 were negatively correlated with the D0 bactericidal titers in the vaccinated participants (Figures 2D–F; Table 1). However, CRP did not correlate with D0 titers (Table 1). There was no association between D0 titers and these cytokines for the placebo group (Table 1). These data corroborate an effect of bactericidal antibody activity in reducing the inflammatory response during infection in live oral vaccine recipients. This was further supported by a vaccine subgroup analysis, which demonstrated stronger correlations in the M01ZH09 arm in whom the highest D0 bactericidal titers were observed (Table 1).

The concentrations of IgG and IgA to assorted S. Typhi antigens were measured to elucidate the antigen specificity of functional antibodies. Other than O9:LPS, flagellin, and Vi polysaccharide that are known key S. Typhi surface epitopes, we also tested responses to hemolysin E (HlyE), cytolethal distending protein subunit B (CdtB) and pilus control protein L (PilL). In antigen array experiments these antigens were previously found to be the dominant targets of circulating antibodies in patients with acute typhoid fever (27, 28). Negligible anti-Vi titers were observed in almost all tested samples (Figure S2A in Supplementary Material). In a mixed-effects model, we identified O9:LPS as being the most likely antigen targeted by bactericidal antibody with some activity also shown to flagellin (Table 2).

For confirmation, we depleted sera of antigen-specific antibodies prior to assessing bactericidal activity. As shown in Figure 3, depletion of O9:LPS antibodies resulted in significant reduction in bactericidal titers from samples at baseline or at D28 (both p = 0.009), independent of subset. In contrast, depleting antibodies specific to flagellin, or total IgA antibodies, did not affect activity (Figure 3).

Consistent with these bactericidal activity data, the numbers of IgM, IgA, or IgG antibody-secreting cells specific to O9:LPS present in circulation 7 days after vaccination were not associated with protection from typhoid fever (Figures S1B–D in Supplementary Material). Similarly, the D0 IgM, IgA, or IgG antibody titers against O9:LPS or flagellin were not associated with protection against development of disease (Figures S1E–G and S2B–D in Supplementary Material). Anti-O9:LPS antibody titers showed a similar longitudinal profile to bactericidal activity (Figure 4). A notable exception was that anti-O9:LPS IgG significantly increased in Ty21a TD participants following vaccination (p = 0.009 at D0) but not in response to challenge.