The human study involving Dutch inhabitants of the village of Herpen, NL was reviewed and approved by the Medical Ethical Committee Brabant (Tilburg, Netherlands, NL74801.028.20) and all participants provided written informed consent.

The human study involving Australian veterinary students undergoing routine Q-VAX^® vaccination was reviewed and approved by the University of Queensland Human Research Ethics Committee (Brisbane, Queensland, Australia, protocol 2020001442) and all participants provided written informed consent.

Samples from n=98 de-identified adult blood bank donors from the Amsterdam/Rotterdam area (The Netherlands) with unknown prior exposure to C. burnetii were obtained in November 2020 from the Sanquin Blood bank (Table 1) (24).

Adults with known prior C. burnetii exposure status from earlier studies were recruited from the Dutch village of Herpen and surrounding areas, the epicenter of the 2007-2010 Q fever outbreak (25). All study participants in this cohort had previously been tested between 2014 and 2016 using the Q-detect™ IGRA to determine cellular immunological responses to C. burnetii (24). The majority of the participants enrolled in the present study were also assessed by immunofluorescence assay (IFA) for serological responses in 2014 (26) and followed up in more detail for cellular responses between 2015 and 2017 (27). In the latter study, preference was given to participants with strong responses to whole heat-killed C. burnetii in the IGRA to maximize the potential to detect C. burnetii epitope-specific T-cells. Therefore, the level of cellular responses in this cohort is not representative of all individuals with prior C. burnetii exposure. In total, n=95 Dutch adults with known past exposure status provided written informed consent and were available for blood collection in January 2021, 10-14 years after natural infection. This included n=84 individuals from the initial Q Herpen II study (26, 27) with known (positive or negative) IGRA and serology status, as well as n=11 individuals with known past symptomatic Q fever and positive IGRA status, but unknown prior serology status (Table 1). Of these 95 individuals, n=13 had no history of Q fever disease (26) and scored negative by IFA in 2014 (24) and 2021 (28) as well as by IGRA in 2014, 2015 and 2021 (24), and were designated as unexposed/no prior infection. Amongst the remaining n=82 participants with known positive IGRA in 2014 and 2015, n=81 had available information on clinical symptoms during past infection (either registered (notified) in the national surveillance system, or self-reported) (26, 27) and were subdivided into asymptomatic (n=47) or symptomatic (n=34).

Blood samples were collected from Australian veterinary students before and after undergoing routine Q-VAX^® vaccination between February and May 2021 (Table 1). Briefly, at enrollment student participants were screened for pre-exposure via serological analysis by enzyme-linked immunosorbent assay (ELISA) for phase II IgG, as well as a skin test using a 1/1000^th dose of the Q-VAX^® vaccine by intradermal injection. Three students had low-level positive phase II IgG ELISA results and were therefore additionally assessed but tested negative by a phase II IgG immunofluorescence assay (IFA; titer <1:25) and a complement fixation (CF) assay (titer < 1:2.5). An experienced physician visually assessed skin tests after 7 days. All n=96 students were eligible for Q-VAX^® vaccination based on a negative composite phase II IgG (ELISA, IFA and CF results), and the absence of skin test reactions. The majority of post-vaccination samples were collected within a target window of 28-35 days after vaccination (median 30 days) from n=58 students who attended the follow-up visit. However, short-term lockdowns to contain local clusters of SARS-CoV-2 infections resulted in some visits being outside the target window, with an overall range of 16-68 days for post-vaccination sample collection. Of those vaccinated, n=38 students did not return for the post-vaccination blood draw. Local and systemic reactions within seven days after skin test and vaccination were self-reported by students via an online survey administered through the SurveyMonkey platform (Momentive, Inc.).

Whole lithium-heparin anti-coagulated blood was stimulated on the day of blood collection with C. burnetii antigen (heat killed strain Cb02629), following the Q-detect™ protocol (24). The Cb02629 strain, isolated from the placenta of a goat infected during the Dutch outbreak, carries the phase I LPS variant. Cb02629 antigen was prepared from a master cell bank by a cell-free culture method using acidified citrate cysteine medium at the Central Veterinary Institute, Wageningen Bioveterinary Research, Lelystad, The Netherlands (29, 30). Recovered cells were washed repeatedly to remove cell media and then heat-killed. For the stimulation, microtubes were pre-coated with heat-killed whole cell C. burnetii Cb2629 antigen or a positive control, namely Staphylococcus enterotoxin B (Sigma, Cat. No. S4881). Both antigens were deposited in a sucrose matrix. The negative control stimulus for the assay was the sucrose matrix on its own. Stimulation was carried out by adding 250µL blood per stimulation tube. After 21-23 hours of incubation at 37°C, whole blood plasma supernatants were collected and IFNγ concentrations assessed using a fully validated in-house ready-to-use human IFNγ ELISA. All samples were assayed in a 4-fold dilution. Concentrations were calculated using a standard curve obtained by four parameter logistic curve fitting. Negative control responses that were too low to be calculated were assigned a concentration of 0.6 pg/mL, which is the limit of quantification of the ELISA. Concentrations above the range of the standard curve were assigned 105% of the concentration of the highest standard (500 pg/mL) multiplied by the dilution factor, resulting in an upper limit of quantification of 2100 pg/mL. The negative control cutoff was set at >40 pg/mL and positive control cutoff was set at <40 pg/mL. A subject was scored positive if negative and positive controls met the cut-offs, the C. burnetii-induced IFNγ production was ≥10 pg/mL above background and the stimulation index (SI; IFNγ conc.^{(C. burnetii )}:IFNγ conc.^{(negative control)}) of the C. burnetii-specific response was SI ≥10, while those with a SI ≥3 but <10 were judged as borderline as previously described (24). 11/347 individuals had borderline IGRA results and one individual amongst these was positive for phase II IgG by IFA.

In this specific study, IGRA results were not deemed inconclusive when the negative (n=7) (Supplementary Figure 1A) or positive (n=8) control (Supplementary Figure 1B) cut-offs of IFNγ production were not met. Instead, innate responses in the same C. burnetii and negative control stimulated samples, as determined by V-plex, were used as an indicator of sample viability. Samples were then judged as IGRA positive or negative solely based on C. burnetii-induced IFNγ level and SI criteria.

For the Dutch cohorts, phase II IgG antibody titers were determined according to the manufacturers’ instructions by IFA (Focus Diagnostics) at the Jeroen Bosch Hospital, ‘s-Hertogenbosch, the Netherlands. Phase II IgG titers of ≥1:64 were interpreted as IFA-positive, consistent with the original Q Herpen II study (26, 28).

The pre-vaccination screening of the Australian veterinary student cohort was conducted using a NATA (31) accredited in-house indirect phase II IgG ELISA with microwells coated with the Nine Mile strain (ATCC 616-VR) and manufactured by Vircell S.L (Spain). In addition, pre- and post-vaccination serum samples underwent extended serological testing in parallel to assess phase II IgG and IgM responses by ELISA, phase I and phase II IgG, IgM and IgA by IFA using Vircell S.L (Spain) manufactured slides coated with C. burnetii, Nine Mile strain (ATCC 616-VR), and for CF antibodies to both phase I and II using the Virion/Serion (Germany) Complement Fixation Test system. IFA dilutions for IgG commenced at 1:100 while IgA and IgM commenced at 1:25. CF responses commenced at doubling dilutions from 1:2.5 as recommended for pre-vaccination screening. For the purpose of calling participants seropositive or negative by phase II IgG IFA, titers of ≥1:100 were interpreted as IFA-positive. However, in the n=3 cases when phase II IgG IFA was conducted to confirm or refute ELISA results during pre-vaccination screening, a cutoff of ≥1:25 was used.

Dimension reduction was performed on background corrected cytokine data after arcsinh transformation using Uniform Manifold Approximation and Projection (UMAP) (32) with the package umap-learn. Subsequently spectral clustering was performed on UMAP embeddings using the package sklearn. The optimal number of clusters was identified using the kneed algorithm as implemented in the package kneed. All analyses were done using Python version 3.8. Figures were created using GraphPad Prism v9 (San Diego, CA, US) and using packages Matplotlib (Version 3.5.1) and Seaborn on Python 3.8.

Statistical analysis was performed using GraphPad Prism v9 (San Diego, CA, US).

In order to study the induction of immune responses upon Q-VAX^® vaccination, we assessed whole blood cytokine responses to heat-killed C. burnetii as well as serological responses to phase I and phase II C. burnetii antigens both pre- and post-vaccination in a cohort of Australian veterinary students. Cellular responses were determined by IGRA and whole blood stimulation supernatants were additionally assessed for release of eight other innate and adaptive cytokines and chemokines.

Within the student cohort, 25% (24/96) were positive and 4% were borderline (4/96) by IGRA prior to vaccination, however none of the participants were positive for phase II IgG by IFA. Amongst the cytokines assessed, the adaptive cytokines IL-2, IFNγ and the IFNγ-induced protein-10 (IP-10) were significantly elevated four weeks post-vaccination (Figure 1A). Innate cytokines and other chemokine responses (IL-8, MIP-1α, IL-1β, IL-6, IL-10 and TNFα), in contrast, did not change significantly after vaccination (Supplementary Figure 2). Notably, the relative increase particularly for IFNγ (and IP-10) was greater and more consistent in participants who were IGRA negative prior to vaccination (Supplementary Figure 3). 97% (34/37) of participants who were IGRA negative pre-vaccination had a 3-fold increase in IFNγ response after vaccination. However only 38% (8/21) of participants who were IGRA positive pre-vaccination had a three-fold increase in IFNγ response after vaccination. The six participants with the highest pre-vaccination C. burnetii-specific IFNγ response (60-240 pg/mL) showed either no clear increase or a decrease in IFNγ responses post-vaccination. Phase II IgG, IgM IFA titers and CF titers were significantly elevated post-vaccination, while phase II IgA IFA titers remained low at 4 weeks post-vaccination (Figure 1B). No IgG, IgA or CF responses were seen against phase I C. burnetii. although 2/58 participants had detectable phase I IgM IFA titers (1:50 and 1:200) at four weeks post-vaccination (data not shown). Correlation analysis between cytokines, antibodies, and days post-vaccination (Q-VAX^®) revealed a positive correlation for IL-2 responses with IFNγ release, phase II IgG, IgM (IFA and ELISA) and CF antibody titers and the length of the interval between vaccination and sample collection for response assessment (days after Q-VAX^®) (Supplementary Figure 4).

Next we compared how cytokine response patterns to C. burnetii differ between study participants that were exposed to inactivated C. burnetii by Q-VAX^® vaccination versus those that experienced a natural infection with viable C. burnetii. In addition to the vaccination cohort of Australian veterinary students, we also assessed C. burnetii-specific responses in 193 Dutch adults from the Sanquin blood bank cohort and the Dutch village cohort with different degrees of natural exposure to C. burnetii. In the Sanquin blood bank cohort where prior exposure to C. burnetii is unknown, n=20 (20% of cohort) were positive by IGRA (5/98 IGRA borderline and 15/98 IGRA positive) and n=3 had positive IgG antibody titers for phase II C. burnetii (3% of cohort). This was comparable to the Australian veterinary students with unknown exposure status prior to vaccination (29% IGRA positive and all sero-negative for Phase II IgG). The distribution and degree of C. burnetii-specific IFNγ responses were highly similar between the Dutch Sanquin blood bank cohort and the Australian students pre-vaccination cohort (Supplementary Figure 1C), consistent with minimal prior clinically-relevant Coxiella exposure in both groups, as reflected by the absence of positive skin tests in the students. Post-vaccination, n=43 participants tested positive by IGRA (74% of cohort) and n=48 had positive IgG antibody titers for phase II C. burnetii (83% of cohort). Only two of the vaccinees (3% of cohort) remained negative by both IGRA and phase II IgG IFA post-vaccination. In the Dutch village cohort, which includes a large proportion of individuals with known past exposure to C. burnetii, n=71 participants (75% of cohort) were IGRA positive and n=49 had measurable IgG antibody titers for phase II C. burnetii (52% of cohort). This distribution of cellular and humoral responses reflects the varying prior exposure status within each cohort (Figure 2A).

Since IgG antibody titers for phase II C. burnetii are a widely accepted diagnostic marker for prior C. burnetii infection and recent studies have demonstrated the utility of the IGRA in identifying past C. burnetii exposure (23, 24, 33), we decided to re-group the individuals from the Sanquin blood bank cohort, the Dutch village cohort and the Australian pre-vaccination cohort into three groups based on IGRA outcome and phase II IgG IFA titer (Table 2). These groups are referred to as exposure status here after. Dutch blood bank donors and study participants from the Dutch village and Australian veterinary cohorts (pre-vaccination) who were scored negative by IGRA and had a negative phase II IgG IFA titer were categorized as Unexposed. Dutch and pre-vaccination Australian participants who had a positive or borderline IGRA and/or a positive IgG antibody titer for phase II C. burnetii were categorized as Naturally Exposed. The Australian post-vaccination samples were treated as an independent sample set and categorized as Vaccinated. The distribution of exposure status within each clinical cohort is shown in Figure 2B. In vitro IL-2, IFNγ and IP-10 responses to C. burnetii were significantly elevated in exposed compared to unexposed participants, regardless of whether exposure occurred through infection or vaccination (Figure 3). Notably, in naturally exposed participants from the Dutch village cohort assessed more than a decade past infection, adaptive cytokine responses were as high (IL-2) or higher (IFNγ and IP-10) than those observed 4-5 weeks after vaccination (Figure 3). While some innate cytokine responses (IL-8, IL-10 and MIP1α) did not differ with exposure status, IL-1beta, IL-6 and TNFalpha responses were slightly but significantly higher in naturally exposed individuals (Supplementary Figure 5).

UMAP was used to perform dimension reduction of all individual samples in this study based on their individual cytokine signatures. Subsequently, the embeddings were clustered using spectral clustering. Four clusters of unique cytokine signatures were identified (Figure 4A). Unexposed individuals were mostly confined to clusters 1 and 2, while clusters 3 and 4 were largely populated by pre-exposed individuals (by infection or vaccination). The largest proportion of all vaccinees was found in cluster 4, and cluster 3 comprised solely of individuals with prior exposure to C. burnetii whether through natural exposure or vaccination. (Figures 4B, C). Within the unexposed group of individuals (Figure 4B), those from the Dutch and Australian cohorts completely overlapped (data not shown). Strong MIP1α and IL-6 responses were seen in all clusters. Although IL-8, IL-10, IL-1β and TNFα had similar response magnitudes across all clusters, the individual samples with the highest response for these four cytokines were seen in cluster 2. IP-10 responses were the main driver of separation of cluster 2 (IP-10 low) from 1, 3 and 4 (IP-10 high) (Figure 4D), while higher IL-2 and IFNγ responses most clearly distinguished naturally exposed and vaccinated individuals (cluster 3) from unexposed ones (Figures 4B, D).

It is important to note that this difference in IP-10 responses arises from relatively high baseline production of IP-10 in combination with low production of IP-10 upon whole blood stimulation with C. burnetii by individuals in cluster 2. Individuals in cluster 1, on the other hand, had similar concentrations for IP-10 in the baseline unstimulated sample and in C. burnetii stimulated sample, while individuals in cluster 3 and 4 showed increased production of IP-10 with C. burnetii stimulation (Supplementary Figure 6). This pattern was seen regardless of whether individuals within each cluster were unexposed, naturally exposed or vaccinated (data not shown).

Median concentrations of IFNγ, IL-2 and IP-10 were highest in cluster 3 (Figure 5). Serological responses were compared between clusters for vaccinated Australian subjects only. While statistically significant differences were not seen in phase II IgG IFA titers across the clusters, mean phase II IgG levels by ELISA were higher in cluster 3 and 4. In addition, phase II antibody levels (IgG and IgM) determined by IFA, ELISA and CF were highest in cluster 3 (Figure 6).

Finally, we investigated whether there was any obvious relation between cellular responses to C. burnetii and clinical symptoms experienced upon Q-VAX^® skin test and vaccination or during natural infection. To this end, pre-vaccination cellular responses were compared prospectively to self-reported side effects upon skin test and vaccination in the Australian veterinary student cohort, while cellular responses in the Dutch village cohort were retrospectively related to previously self-reported symptoms experienced during infection. In the Australian student cohort, a digital online questionnaire was filled out by 67/96 and 54/96 students post-skin test and post-vaccination, respectively. Post-skin test, 31% reported redness, 67% pain, 36% swelling and 19% itchiness at the site of injection. Of note, none of these reactions were observed at the injection site on the 7^th day post-skin test when the attending physician scored the skin test. All of the participants were deemed skin test negative due to the lack of induration in the skin following palpation and thus received Q-VAX^® vaccination. Following Q-VAX^® vaccination, vaccinees commonly reported local adverse effects such as pain (74%), redness (57%) and swelling (56%). Systemic responses reported were less common and included fever (2%), headache (24%), lethargy (19%) and joint pain (19%). Naturally exposed individuals in the Dutch village cohort were categorized as symptomatic or asymptomatic based on self-reported symptoms, recorded either in 2014 (26) or earlier if individuals were officially registered (notified) through their treating physician in the Dutch national surveillance system. We found no association between skin test (Figure 7A) or vaccination related side effects (Figure 7B) and cytokine response signatures in the Australian veterinary student cohort with a comparable proportion of individuals reporting symptoms or not within each cluster. Similarly, a comparable proportion of individuals of the Dutch village cohort within each cytokine cluster had or had not experienced and re-collected symptoms during natural infection (Figure 7C). There were also no definitive associations between these self-reported symptoms and individual cytokine responses, including IFNγ (Supplementary Figure 7, data not shown for other cytokines).