Although an initial evaluation of the rBCG-N-hRSV vaccine candidate was performed in a phase I clinical trial (clinicaltrials.gov identifier #NCT03213405), we sought to perform a deeper immunogenicity analysis to further evaluate this vaccine candidate. Briefly, a randomized, double-blind, dose-escalating phase I clinical trial was designed and carried out between 2017-2018 in Santiago, Chile (NCT03213405). The primary and secondary outcomes were to evaluate the safety, tolerability, dissemination, shedding, and immunogenicity of the rBCG-N-hRSV vaccine in healthy adults. The protocol of this study followed the current ethical guidelines, such as Tripartite Guidelines for Good Clinical Practices, the Declaration of Helsinki (18), and local regulations. This study was approved by both the Institutional Ethical Committee (number 15216) and the Chilean Public Health Institute (ISP Chile, number EC819077/16). Inclusion and exclusion criteria and more information about the clinical trial can be found in the initial publication describing this clinical trial, which evaluated safety and tolerability (9).

During the study period, hRSV circulation was reported in Chile, as previously described (9). Particularly, the immunization of the subjects in Cohort A was performed during the peak of hRSV, Cohort B during the end of the hRSV season, and Cohort C during a period of low circulation of hRSV (9). Considering the immunization timeline, seroconversion analyses were performed to determine possible exposure to hRSV from vaccinated volunteers, as indicated in the results section. Further information regarding the outcomes, the ethical approval, inclusion and exclusion criteria, and the recruitment of volunteers can be found in the Supplementary Information (SI) previously published in (9). Briefly, 24 male adults aged 18-50 years were enrolled and randomized into three cohorts of eight subjects. In each cohort, two participants were immunized intradermally in the deltoid area with 2x10⁵ CFU of BCG-WT (Moscow strain). Since the population in Chile was immunized at that time with the Moscow BCG strain at birth, it was used as BCG-WT during this trial. The remaining six subjects were vaccinated with escalating doses of rBCG-N-hRSV (5x10³, 5x10⁴, or 1x10⁵ CFU), which were generated using the BCG Danish strain. The dose of BCG-WT selected is the dose most commonly used in newborns worldwide, and the doses of rBCG-N-hRSV used were escalating doses previously evaluated during pre-clinical studies and shown to be appropriate. Data obtained from subjects immunized with BCG-WT were analyzed together, independently of their cohort. Blood samples were taken before immunization (Day 0), as well as 14-, 30-, 60-, 120-, and 180-days post-immunization. Peripheral blood mononuclear cells (PBMCs) and sera were isolated from these blood samples and stored in liquid nitrogen and at -80°C, respectively, until use.

Secretion of perforin (Perf) and granzyme B (GrzB) by PBMCs was evaluated using the Human Granzyme B/Perforin Double-Color Enzymatic ELISPOT Assay (ImmunoSpot^®). The assay was performed according to the instructions given by the manufacturer, incubating with each stimulus for 48 h at 37°C and 5% CO₂. 2x10⁵ PBMCs were plated and stimulated with either 1.25 µg/mL of N-hRSV (Genscript) or 750 IU/mL of bovine PPD (ThermoScientific) diluted in sterile RPMI 1640 medium (Gibco) supplemented with 2 g/L sodium bicarbonate, 10% FBS (Biological Industries), 2 mM L-Glutamine (Gibco), 0.2% v/v β-mercaptoethanol, and 1X Anti-Anti (Gibco), for 48 h at 37°C and 5% CO₂. Stimulation of PBMCs with 5 µg/mL ConA or 0.5% sterile PBS were used as positive and negative stimulation internal controls, respectively (data not shown). After developing, plates were air-dried face-down on a paper towel for 24 h, and then stored with the underdrain on until spots were counted with an ImmunoSpot S6 CORE Analyzer (Immunospot).

Cytokine measurements were performed via the BD™ Cytometric Bead Array (CBA) Human Th1/Th2/Th17 Cytokine Kit (BD Biosciences). Determinations were performed following the instructions of the manufacturer. Briefly, once mixed and equilibrated, beads were mixed with either undiluted sera or PBMC culture supernatant diluted 1:4 in dilution buffer. Samples were acquired in an LSRFortessa X-20 flow cytometer (BD Biosciences). The median fluorescent intensity of PE in each bead population was calculated for every sample. A 4-parameter logistic (4PL) standard curve was constructed using the median fluorescent intensity of PE for each cytokine, subtracting the signal from the blank tube to each sample and standard and constraining the curve with a Bottom parameter = 0. Cytokine concentration was then interpolated in the respective standard curve.

PBMCs stored in liquid nitrogen were thawed, stimulated with antigens, and analyzed as previously described (9). Data were acquired in a BD LSRFortessa X-20™. Gating and analysis were performed using FlowJo software (V10.6.2).

The binding of C1q was measured using ELISA. Plates were coated overnight at 4°C with 50 µL of either 1 µg/mL N-hRSV (Genscript) or 25 µg/mL bovine PPD (ThermoScientific) in 100 mM bicarbonate/carbonate pH 9.5 buffer. Then, plates were washed three times with 200 µL of PBS-Tween 20 0.05% (Wash buffer) and then blocked for two h with 200 µL of PBS-Milk 5% (Blocking solution). Sera samples were diluted 1:30 in Blocking solution and inactivated at 56°C for 30 min. Plates were washed three times, incubated with 50 µL of inactivated sera samples for 1 h at RT, washed three times, incubated for one h with 50 µL of 4 µg/mL purified human C1q (Sigma-Aldrich) diluted in Blocking solution, washed five times, incubated for one h with 50 µL of mouse anti-human C1q (Invitrogen) diluted 1:2,000, washed seven times, incubated for two h with 50 µL of rat anti-mouse IgG1-HRP (Invitrogen) diluted 1:2,000, washed seven times with 200 µL of Wash buffer and washed manually one time with 200 µL of PBS. 50 µL of TMB (ThermoScientific) was added and incubated for 15 min at RT. Then, 50 µL of 2 N sulfuric acid was added and absorbance was immediately measured at 450 nm. A C1q standard curve was incorporated in every plate. To normalize for variable antibody concentration, a C1q Binding Index was defined and calculated as follows:

ELISA plates were coated overnight. at 4°C with 50 µL of 2 µg/mL N-hRSV (Genscript) or 25 µg/mL bovine PPD (Thermo Scientific) in sodium carbonate/bicarbonate 100 mM pH 9.5 buffer. Then, plates were washed three times with 200 µL of PBS-Tween 20 0.05% (Wash buffer). Plates were blocked with PBS-FBS 10% (Blocking solution for IgG1 and IgG2) or PBS-milk 5% (Blocking solution for IgG3) for 2 h at RT, washed three times, incubated with sera diluted 1:5 in Blocking solution for two h (IgG1 and IgG2) or 1:100 (IgG3), washed three times, incubated for 1 h with 50 µL of either rat anti-human IgG1 (BioLegend), rat anti-human IgG2 (BioLegend) or rat anti-human IgG3 diluted 1:500, washed three times, incubated for 1 h with 50 µL of rat anti-mouse IgG1-HRP (Invitrogen) diluted 1:2,000, washed three times with 200 µL of Wash buffer and manually washed one time with 200 µL of PBS. 50 µL of TMB Substrate Reagent (Thermo Scientific) was added to each well and developed for 15 min at RT. 50 µL of 2 N sulfuric acid was added to each well and absorbance at 450 nm was immediately measured. Standard curves for IgG1 or IgG2 were incorporated for every plate.

The evaluation of neutralizing capacity of the antibodies induced upon immunization was measured as previously described (9). Briefly, HEp-2 cells were incubated with 100 plate-forming units (PFU) of a recombinant GFP-hRSV virus. Then, inactivated sera samples were diluted, incubated with the infected cells for 1 h, and then replaced with fresh media (MEM). After incubation for 48 h, GFP⁺ PFUs were visualized and quantified using an epifluorescence microscope. Further details can be found in the previous report (9).

ELISA plates were coated overnight at 4°C with 50 µL of 1 µg/mL F-hRSV (A2) (Sino Biological) or G-hRSV (B1) (Sino Biological) in sodium carbonate/bicarbonate 100 mM pH 9.5 buffer. Then, plates were washed three times with 200 µL of PBS-Tween 20 0.05% (Wash buffer). Plates were blocked with PBS-milk 5% (Blocking solution) for 1 h at RT, washed three times, incubated with sera diluted 1:5 in Blocking solution for 2 h, washed three times, and incubated for one h with 50 µL of sera samples serially diluted ranging from 1:100 to 1:3,200 in blocking solution. Then, plates were washed three times, incubated for 1 h with 50 µL of rat anti-human IgG-HRP (Thermo Scientific) diluted 1:2,000 in blocking solution, washed five times with 200 µL of Wash buffer and manually washed one time with 200 µL of PBS. 50 µL of TMB Substrate Reagent (Thermo Scientific) was added to each well and developed for 10 min at RT. 50 µL of 2 N sulfuric acid was added to each well and absorbance at 450 nm was immediately measured.

All graphs and statistical analyses were performed in GraphPad Prism, version 9.1.0. Analyses of standard curves in ELISA assays were performed by a 4-parameter logistic regression (4PL) curve, constrained to a Bottom parameter = 0. Data were then interpolated in those regressions. To calculate fold changes relative to pre-immune conditions, the observed response variable at a given time was divided by that observed at Day 0. A two-way ANOVA for repeated measures with post-hoc Dunnett's test corrected for multiple comparisons against Day 0 was performed to evaluate differences after immunization. Base 10 logarithms of fold changes were calculated before statistical analyses.

One of the advantages of rBCG-N-hRSV, relative to other hRSV vaccine candidates, is the potential induction of both, anti-mycobacterial and anti-hRSV immune responses (8). While CD4⁺ helper T cell responses against N-hRSV and mycobacterial antigens after rBCG-N-hRSV immunization have been previously evaluated (9), the secretion of cytotoxic molecules by CD8⁺ cytotoxic T cells has yet to be assessed. Importantly, a robust CD8⁺ T cell response correlates with protection against hRSV in humans (19), so the induction of an effector CD8⁺ T cell response would be desirable for a vaccine against this virus.

Stimulation of PBMCs with N-hRSV demonstrated a trend to induce an effector cytotoxic molecule response in subjects immunized with 5x10⁴ CFU of rBCG-N-hRSV ( Figure 1 , Supplementary Figures 1 , 2 ). A progressive increase of perforin (Perf) and granzyme B (GrzB) response were observed in PBMCs from subjects of this cohort, with a significant increase in Perf secretion at day 60 post-immunization ( Figure 1 ). It must be considered that the results of the GrzB response from subjects of this cohort showed high variability between subjects at day 30 post-immunization. Interestingly, the GrzB response obtained with 5x10⁴ CFU was similar to 5x10³ CFU of rBCG-N-hRSV ( Figure 1 ). Lastly, immunization with BCG-WT induced a stable secretion of cytotoxic molecules in PBMCs in response to N-hRSV during 30- and 60-days post-immunization, yet not significant.

Stimulation of PBMCs with PPD antigen induced an effector cytotoxic molecule response in subjects immunized with 5x10⁴ CFU of rBCG-N-hRSV, as compared to subjects immunized with other rBCG-N-hRSV doses ( Figure 1 ). However, this response was modest as compared to subjects immunized with 2x10⁵ CFU of BCG-WT. Subjects vaccinated with BCG-WT showed a progressive increase in responsiveness against PPD after immunization ( Figure 1 ). Subjects immunized with 5x10³ CFU of rBCG-N-hRSV showed a low response to PPD, probably due to the low dose of rBCG-N-hRSV in this cohort. Subjects that received 1x10⁵ CFU of rBCG-N-hRSV showed an optimal response to PPD, but lower as compared to subjects immunized with 5x10⁴ CFU of rBCG-N-hRSV ( Figure 1 ). Even though the inhibition of the PPD antigen expression by the recombinant N-hRSV is highly unlikely since mycobacteria constitutively express PPD antigens, it needs to be taken into consideration that the strain used in the dose of BCG-WT and rBCG-N-hRSV are different, and therefore the response to PPD should not be expected to be the same.

These results suggest that the trend to increase the secretion of Perf and GrzB in response to N-hRSV stimulation depends on the amount of N-hRSV in the corresponding rBCG dose ( Figure 1 ). Therefore, among the rBCG doses, the 5x10⁴ CFU of rBCG-N-hRSV seems like an optimal dose for inducing the secretion of cytotoxic molecules by PBMCs in response to either PPD or N-hRSV.

The secretion of various cytokines by PBMCs after stimulation with N-hRSV or PPD ( Figure 2 ) was measured to assess changes in pro- or anti-inflammatory cytokines after immunization. Stimulation of PBMCs with N-hRSV led to increased levels of IFN-γ, IL-6, IL-10, and TNF-α after immunization with 5x10³, 5x10⁴, and 1x10⁵ CFU of rBCG-N-hRSV ( Figure 2 ). Subjects immunized with BCG-WT also showed an upregulation of the same cytokines, still the increase was observed on a single subject. Stimulation with PPD led to increased levels of IFN-γ, TNF-α, IL-2, IL-6, and IL-10 ( Figure 2 ) in subjects immunized with 5x10³ or 5x10⁴ CFU of rBCG-N-hRSV. Subjects immunized with 2x10⁵ CFU of BCG-WT also showed an upregulation of IL-6.

This suggests that immunization with either 5x10³ or 5x10⁴ CFU of rBCG-N-hRSV leads to the secretion of pro-inflammatory cytokines typically associated with immune responses against intracellular pathogens, such as viruses or mycobacteria (20–22).

Additionally, the concentration of several cytokines in sera was evaluated before and after immunization at various time points for subjects immunized with either dose of rBCG-N-hRSV or BCG-WT ( Supplementary Figure 3 ). All subjects exhibited a slight increase in the concentrations of IFN-γ, IL-6, IL-10, and TNF-α in serum, which suggests the lack of severe adverse effects after immunization. However, further secretory cytokine profiling would be needed to confirm this observation. These results suggest that the doses of rBCG-N-hRSV tested do not induce peripheral inflammation, which supports the safety of the rBCG-N-hRSV vaccine candidate in healthy BCG-immunized adults.

Since we detected an antigen-specific T cell response in subjects immunized with rBCG-N-hRSV, we sought to determine whether immunization promotes the expansion of memory T cell subsets (gating on Supplementary Figure 4 ) (23). Albeit non-statistically significant, stimulation of PBMCs with N-hRSV or PPD led to a contraction of the naïve population along with the expansion of the effector memory (T_EM) and the CD45RA-expressing effector memory (T_EMRA) CD4⁺ T cell populations in participants immunized with 5x10⁴ CFU of rBCG-N-hRSV 14 days post-immunization ( Figure 3 ). Interestingly, similar contractions of the naïve population and the expansion of T_EM and T_EMRA was detected in subjects immunized with 1x10⁵ CFU of rBCG-N-hRSV, with an even tighter deviation relative to the 5x10⁴ CFU group ( Figure 3 ). This was also accompanied by a minor increase in the percentage of PPD-stimulated CD4⁺ T cells co-expressing CD25 and CD69 ( Figure 4 ). However, the significant variability between subjects needs to be considered. Interestingly, a significant increase in the percentage of N-hRSV-stimulated CD4⁺ T cells expressing IL-2 could be observed in these participants at day 120 post-immunization in subjects immunized with 5x10⁴ CFU of rBCG-N-hRSV ( Figure 5 ). A slight increase of IFN-γ and TNF-α expression, when stimulated with either PPD or N-hRSV, can be observed in this cohort at day 120 post-immunization ( Figure 5 ). This increase at day 120 post-immunization is similar to the increase observed with BCG-WT, but not with 5x10³ and 1x10⁵ CFU of rBCG-N-hRSV. A mild increase in CD69-expressing CD4⁺ T cells was also observed for PBMCs from participants immunized with the lowest dose of rBCG when stimulated with N-hRSV ( Supplementary Figure 5 ). A slight increase of N-hRSV- or PPD-stimulated CD8⁺ T cells expressing IL-2 or IFN-γ in participants immunized with 5x10⁴ or 1x10⁵ CFU of rBCG-N-hRSV was detected ( Figure 6 ).

Although the concentration of anti-N-hRSV antibodies in the serum only significantly changed after immunization with 1x10⁵ CFU of rBCG-N-hRSV ( Supplementary Figure 6 ) (9), it is relevant to evaluate the relative proportions of anti-N-hRSV IgG subclasses in the samples of all subjects and the potential deposition of complement proteins mediated by these antibodies to rule out potential lung damage that could be driven by anti-N IgG in the absence of a robust neutralizing response in infants (17). Since lung damage has been associated with excessive complement system activation, the ability of anti-N-hRSV antibodies to bind exogenously-added C1q was determined as a functional measurement for the induced anti-N-hRSV IgG subclasses (24).

Exogenous C1q binding to anti-N-hRSV or anti-PPD antibodies in sera was quantified for each serum sample ( Figure 7 ). Both anti-N-hRSV and anti-PPD antibodies from subjects immunized with 5x10⁴ and 1x10⁵ CFU of rBCG-N-hRSV were found to bind similar, or less complement after immunization as compared to day 0 post-immunization, which was the day with highest binding to IgG. These trends were maintained when C1q binding was normalized by total anti-N or anti-PPD IgG titers (C1q Binding Index). However, C1q binding for both anti-N-hRSV and anti-PPD antibodies from subjects immunized with 5x10³ CFU of rBCG-N-hRSV increased after immunization. Fold changes compared to the pre-immune condition showed similar trends ( Supplementary Figure 7 ).

Although the amount of C1q binding to purified antigen-specific IgG subclasses was not evaluated, we did evaluate antigen-specific IgG isotypes in sera that bound to N-hRSV or PPD as a proxy of this function. Interestingly, immunization with rBCG-N-hRSV did not change the concentration of IgG subclasses, neither for anti-N-hRSV nor for anti-PPD antibodies ( Figure 8 ). As expected, IgG1 titers for anti-N-hRSV were higher than IgG2 titers. To further confirm that there was no induction of potentially harmful, non-neutralizing, and highly complement-binding anti-N-hRSV IgG, we measured anti-N-hRSV IgG3 levels. As expected, we found no quantifiable antigen-specific IgG3 signals, further supporting the lack of harmful complement-binding responses ( Supplementary Figure 8 ).

Although participants reported no symptoms of a common cold during the study, two of them presented serological evidence of hRSV exposure (>two-fold increase in anti-F-hRSV and anti-G-hRSV IgG titers) ( Supplementary Table 1 , Supplementary Figure 9 ). Both subjects were part of the cohort immunized with the lowest tested dose of rBCG-N-hRSV, and their data were excluded from the global analysis of the study.

One participant was probably exposed to hRSV between 14 and 30 dpi and exhibited increased serum cytokines, slightly increased C1q binding to anti-N-hRSV antibodies, no increase in anti-N-hRSV IgG subclasses, increased Perf, GrzB, TNF-α, and IFN-γ secretion by PBMCs upon N-hRSV stimulation, and no changes in T cell memory phenotypes or expression of activation markers by 30 dpi ( Supplementary Figure 9 ). This participant also showed modest cellular responses to PPD stimulation, marked by delayed Perf and GrzB secretion and early TNF-α and IFN-γ secretion by PBMCs ( Supplementary Figure 10 ).

The other participant was probably exposed before 14 dpi and showed no systemic cytokine responses, a slight induction of C1q binding to anti-N-hRSV antibodies and anti-N-hRSV IgG1, no Perf or GrzB responses to N-hRSV or changes in memory phenotypes, increased activation of CD8⁺ T cells (as measured by the expansion of the CD69-expressing population), and a marked increase in the secretion of cytokines by PBMCs stimulated with N-hRSV ( Supplementary Figure 9 ). This subject also showed poor responsiveness to PPD, characterized only by the secretion of cytokines by PBMCs, but no Perf or GrzB responses ( Supplementary Figure 10 ).