Anti-Vi IgG titers were measured using a commercial vaccine enzyme-linked immunosorbent assay (ELISA) kit (VaccZyme, Binding Site) according to the manufacturer’s instructions. Anti-IgA titres were measured using a modification to the same commercially available kit, replacing the Typhi Vi-IgG conjugate with a Vi-IgA conjugate. ELISA titres were measured in plasma obtained on day 0 (pre-vaccination), and 28 days post-vaccination and the data has been published previously (21, 22).

2.5 µl of biotinylated 1 mg/ml Vi-antigen (NIBSC) was added to 10 µl of red fluorescent NeutrAvidin™-Labelled microspheres (Thermofisher) per plate and incubated overnight at 4°C, or for two hours at 37°C. Prior to the assay, the beads are washed twice with 0.1%, and 5% BSA, before being resuspended in 1 ml of RPMI per plate.

Plasma samples were first heat inactivated at 56°C for 30 min. Plasma was diluted in RPMI at 1:320 for assessment of antibody-mediated neutrophil phagocytosis. Samples were then incubated for 2 hours with Vi-coupled red fluorescent microspheres (Thermofisher) in 96-well plates.

Leukocytes, including neutrophils were isolated from sodium heparin treated blood from healthy anonymised UK donors using ammonium-chloride-potassium (ACK) lysis buffer (Thermofisher). 5 x 10⁴ leukocytes were added to each well of antibody-opsonised beads. Cells were pelleted and stained with APC-Cy7 anti-CD14 (Clone MyP9, BD Biosciences, Franklin Lakes, USA), Alexa Fluor 700 anti-CD3 (Clone UCHT1700, BD Biosciences, Franklin Lakes, USA) and Pacific Blue anti-CD66b (Clone G10F5, Biolegend, San Diego, USA).

Leukocytes were washed and analysed using a BD LSRFortessa X-20 for flow cytometry. FlowJo v10.6 software was used for gating cell populations. The phagocytic score was calculated as the mean fluorescent intensity (MFI) of red beads within gated neutrophils (CD14, CD3 negative, CD66b positive), multiplied by the percentage of bead-positive gated neutrophils.

Vi-specific IgG was captured using NeutrAvidin plates (Thermofisher), coated for 2 hours at room temperature with 100 µl biotinylated Vi (commercially biotinylated WHO international Standard Vi polysaccharide of C. freundii NIBSC code: 12/244) at 1.25 µg/ml in 0.5X PBS. After washing three times with 200 µl 0.5X PBS, 200 µl of 1:20 diluted (in 0.5X PBS) plasma was incubated for 1 hour at room temperature and then the sample transferred to another Vi-coated blank well for another 1 hour incubation. After each incubation, wells were washed once with 200 µl 0.5X PBS and three times with 200 µl freshly made 25 mM ammonium bicarbonate (Sigma). Following the washes, 50 µl of 100 mM formic acid (Sigma) was added and incubated at room temperature for 5 min with agitation. After transferring the 50 µl eluate to a clean V-bottom 96-well plate, another 50 µl 100 mM formic acid was added, incubated for 5 min with agitation and the 2 eluates from the same sample were pooled.

Samples were dried using a vacuum centrifuge for 2-3 hours at 50/60°C, and shipped to LUMC for processing as below for total IgG. Briefly, samples were resolubilised in 20 μl 50 mM ammonium bicarbonate (Sigma) while shaking for 5 min. Twenty μl of 0.05 μg/μl tosyl phenylalanyl chloromethyl ketone (TPCK)-treated trypsin (Sigma) in ice-cold purified water was added per well, and the samples were incubated at 37°C overnight.

For IgG purification, 2 μl of Protein G Sepharose 4 Fast Flow beads (GE Healthcare) were added per well on an Orochem filter plate and washed three times with PBS. 20 μl of 1:20 diluted plasma (in PBS) was added to each well. Plates were incubated for 1 hour with shaking at 750rpm. Using a vacuum manifold, the samples were washed three times by adding 400 μl PBS, followed by three times 400 μl of purified water. Antibodies were eluted from the beads by addition of 100 μl of 100 mM formic acid (Sigma) incubating for 5 min with shaking, and centrifuging at 100 × g for 2 min. Samples were dried for 2 hours at 60°C in a vacuum centrifuge before resolubilisation in 20 μl 50 mM ammonium bicarbonate (Sigma) while shaking for 5 min. Twenty μl of 0.05 μg/μl tosyl phenylalanyl chloromethyl ketone (TPCK)-treated trypsin (Sigma) in ice-cold purified water was added per well, incubated at 37°C overnight, and stored at -20°C until MS analysis.

Samples were thawed and thoroughly mixed. Two hundred nl of total IgG or 5 μl of Vi-specific IgG glycopeptide samples were injected into an Ultimate 3000 RSLCnano system (Dionex/Thermo Scientific, Breda, the Netherlands) coupled to a quadrupole-TOF-MS (MaXis HD; Bruker Daltonics, Bremen, Germany). The LC system was equipped with an Acclaim PepMap 100 trap column (particle size 5 μm, pore size 100 Å, 100 μm × 20 mm, Dionex/Thermo Scientific) and an Acclaim PepMap C18 nano analytical column (particle size 2 μm, pore size 100 Å, 75 μm × 150 mm, Dionex/Thermo Scientific). Glycopeptides were identified based on their m/z, retention time and literature (23).

Data from this study was analysed using R version 4.00. Phagocytosis data were first tested for normality using Shapiro-Wilk normality test. The Wilcoxon paired test used for comparisons between non-parametric groups, including comparisons between time points and the Wilcoxon test (unpaired) was used to test for comparisons between study cohorts. Correlation analyses of glycosylation data used Spearman’s r correlation, with 95% CI. Fold change in ADNP from baseline to D28 was calculated using the formula: F o l d   C h a n g e = ( m e a n   p h a g o c y t i c   s core [ P o s t - V a c ] - m e a n   p h a g o c y t i c   s c o r e [ P r e - V a c ] ) m e a n   p h a g o c y t i c   s c o r e [ P r e - V a c ]

To assess the Vi-specific ADNP response in children vaccinated with Vi-TCV in Nepal, plasma samples from children vaccinated in the TyVAC study were analysed for their capacity to induce neutrophil phagocytosis at baseline and 28 days post-vaccination (n = 77). These data show that ADNP significantly increased 28 days following Vi-conjugate vaccination (p <0.0001) ( Figure 1 ), with an average 3.6-fold increase in ADNP from baseline.

In order to compare anti-Vi ADNP responses between an adult population vaccinated in the UK and children vaccinated in Nepal, we measured ADNP for a subset of both Nepalese and UK vaccinated volunteers within the same assay (n = 17, n =16, respectively). UK participants were selected who had received the Vi-TCV vaccine and who were not diagnosed with typhoid fever in the human challenge study. Participants from the TyVAC study were selected randomly from the initial ADNP cohort. As previous functional antibody assays had been published using serum that was not collected in the TyVAC Nepal study, we reran the ADNP assay using UK participant plasma, alongside the Nepal samples.

Twenty eight days following Vi-TCV vaccination, ADNP increased in both groups when compared with baseline (p<0.0001, p < 0.001) ( Figure 2 ). When comparing ADNP responses between cohorts there was no significant difference between the median ADNP responses at D28 in Nepalese children when compared with UK adults ( Figure 2 ). In addition, whilst there were higher baseline ADNP levels in the UK adults compared with Nepalese children (p=0.02), the fold changes between baseline ADNP and D28 were not significantly different between study cohorts. Therefore, ADNP is induced by plasma antibodies from children vaccinated in Nepal in a similar magnitude as seen in UK adult vaccinees.

When measuring anti Vi-IgG and Vi-IgA titres at baseline, there was no significant difference between our subset of UK and Nepal based vaccinees; however the titres were largely at the lower limit of detection of the ELISA assay (data not shown). Twenty eight days following vaccination, Vi-IgG titres were significantly higher in the Nepalese cohort, while no difference was observed in Vi-IgA titres ( Supplementary Figure 1 ).

To better understand ADNP responses at both baseline and 28 days following vaccination in Nepalese children, we investigated if anti-Vi ADNP responses differed by age. We compared phagocytosis scores prior to, and 28-days after Vi-TCV vaccination in children between 0-4, 5-9 and 10-15 years of age ( Figure 3 ) who received typhoid vaccination. At baseline, prior to vaccination, ADNP activity was higher along with increasing participant age ( Figure 3A ); the oldest children, 10-15 years of age, had significantly higher levels of ADNP when compared with children aged 5-9 (p<0.05) and children aged 0-4 (p<0.0001). These children also had higher levels of Vi-IgG at baseline when compared with younger children aged 5-9 ( Supplementary Figure 2A ). Twenty eight days following vaccination, however, plasma from all children had a similar capacity to induce neutrophil phagocytic activity, and there was no significant difference in ADNP between age groups ( Figure 3B ). Subsequently, due to the lower baseline in the younger children, the fold change in ADNP between baseline and D28 also differed between age groups ( Figure 3C ).

Higher levels of serum Vi-antibodies are associated with increased exposure to S. Typhi (24, 25). We therefore postulated that higher Vi-IgG or Vi-IgA titres could be driving the higher pre-vaccination ADNP levels observed in children as they age. For Vi-IgG, we did measure higher antibody titres in the older age group (10-15 years) when compared with younger children (5-9 years) at baseline ( Supplementary Figure 2A ). However, the correlation between Vi-IgG titres and ADNP responses pre-vaccination was weak (R =0.24, p=0.044) ( Figure 4A ). Additionally we did not observe a significant difference in baseline Vi-IgA titres between children as they aged in our cohort of children ( Supplementary Figure 2B ). No correlation was observed between baseline ADNP and Vi-IgA titres (R =-0.075, p=0.54) ( Figure 4B ).

Following vaccination, we investigated the relationship between Vi-antibody titre and ADNP. Twenty eight days after vaccination there is both an increase in Vi-antibody titres and phagocytosis scores ( Supplementary Figure 3 ). Vi-IgG titre also increased with age ( Supplementary Figure 2D ). Yet as with pre-vaccination ADNP, there was no correlation between ADNP and Vi-IgG (R =0.018, p=0.89) and Vi-IgA (R =0.17, p=0.16) antibody titres ( Figures 4C, D ). Therefore, differences in ADNP magnitude at baseline and following vaccination are not solely due to antibody quantity and may be driven by other changes in antibody quality such as Fc receptor binding or modulation by Fc glycosylation.

Glycosylation profiles including galactosylation, sialyation and fucosylation, were measured in a randomly selected subset (n =24) of plasma samples from Nepalese children aged 9 months – 15 years. Due to the similarities in amino acid sequence in the Fc region proximal to the asparagine at position 297, IgG2 and IgG3 cannot be distinguished from each other, and were therefore recorded together. LC-MS was used to measure Fc glycopeptides, providing IgG subclass-specific glycosylation profiles for both Vi-specific antibodies and the total IgG antibody compartment (26).

We first assessed the correlation between Fc glycan traits of Vi-specific IgG1 and Vi-IgG2/3 at D28 with ADNP fold change from baseline ( Figures 5A, B ). We observed that lower percentages of sialylated Vi-IgG1 antibodies correlated with higher levels of ADNP induced at D28 from baseline (R = 0.47, p=0.021), ( Figure 5C ). A lower percentage of terminal galactose residues, on the Fc domain of Vi-IgG1 28 days post-vaccination, also correlated with ADNP at D28 (R = 0.38, p=0.067) ( Figure 5D ), however this was not statistically significant. Moreover, when breaking down galactosylation levels into the percentage of agalactosylated (G0), monogalactosylated (G1), and digalactosylated (G2) Vi-IgG1 antibodies; a higher percentage of monogalactosylated Vi-IgG1 correlated with levels of ADNP at D28 from baseline (R = 0.48, p=0.028), whereas the percentage of digalactosylated Vi-IgG1 inversely correlated with ADNP (R = -0.047, p=0.021) ( Supplementary Figures 4B, C ).

When measuring Fc glycosylation of Vi-IgG2 and 3 at D28, we observed similar patterns as seen with Vi-IgG1 and ADNP, whereby lower percentages of sialylation and galactosylation were associated with an increase in ADNP from baseline, however these were non-significant ( Figure 5B ). Furthermore, when investigating the associations of G0, G1 and G2 galactosylation levels with ADNP, monogalactosylation of Vi-IgG2/3 also correlated with ADNP fold change (R= 0.43, p=0.038) ( Supplementary Figure 4E ).

As sialyation biosynthetically occurs following galactosylation, and the two have been shown to correlate in previous studies (27), we investigated if sialyation and galactosylation of Vi-specific antibodies correlated. We observed a strong correlation between the percentage of galactosylation and sialyation of Vi-IgG1 (p<0.0001) ( Figure 5A ), confirming that the two processes are linked. To understand if there was a sialyation specific effect on ADNP, we correlated the number of sialic acid residues per galactose (S/G) with ADNP fold change from baseline. For Vi-IgG1, we observed a significant correlation between lower percentages of sialyation per galactose and ADNP at D28 (R = 0.43, p=0.035) ( Figure 5E ). Thus, suggesting, that for Vi-IgG1, sialyation of the Fc region may have a specific effect on the capacity for antibodies to induce ADNP following vaccination, in conjunction with galactosylation.

To confirm that these associations with ADNP were driven by Vi-specific antibodies and not by glycosylation profiles of all non-Vi specific IgG antibodies, we correlated total IgG glycosylation profiles at D28 with ADNP. No associations with antibody-induced neutrophil function were observed (data not shown). Furthermore, we did not correlate fucosylation levels of the Fc domain 28 days following vaccination with ADNP due to the consistently high fucosylation levels across participants ( Supplementary Figure 5 ). We also examined if age was a factor in driving glycosylation states at D28 and associations with ADNP, as glycosylation states at baseline have been known to vary with the age of children and adolescents (28). Therefore, we tested for interactions between age and Vi-IgG1 galactosylation and sialylation in a linear model to predict ADNP fold change. This model showed that the interaction between the two glycosylation profiles with age, respectively, was non-significant (p=0.95, p=0.64).