A total of 12 cycling Droughtmaster heifers, aged approximately 18–24 months with body condition scores ranging from 3 to 4, were selected for this trial. All heifers were in cycle, and this was confirmed by ultrasound scanning for the presence of a corpora luteum (CL). The heifers were sourced from Spyglass Beef Research Facility, Charters Towers, Queensland, Australia, in 2021 and managed as a single group in a paddock at Columba Catholic College (CCC) in Charters Towers (−20° 12′ 33.372” N and 145° 48′ 59.22 E) in North Queensland. Animal Ethics Approval 2021/AE000056 was granted by the Production and Companion Animal (PCA), University of Queensland Animal Ethics Committee.

Six heifers were vaccinated once before oestrous synchronisation using a commercially available vaccine against BGC (Vibrovax^®, Zoetis) following the manufacturer’s instruction, and six were unvaccinated (controls). Nine weeks post-vaccination, both groups were injected with double doses of prostaglandin (PGF2α, Lutalyse^®; Zoetis) at 11 days apart, to ensure heifers were in oestrus for the intravaginal challenge with 3 mL of freshly prepared inoculum of live C. fetus venerealis strain 76223 (Indjein, 2013) bacteria delivering 2.7 × 10⁹ CFU/heifer. The time points for sample collection and the workflow of the vaccine challenge trial are presented in Figure 1.

Freshly grown C. fetus subsp. venerealis bacterial colonies were collected with disposable cotton tips and mixed in 45 mL of phosphate-buffered solution (PBS). The bacterial suspension was diluted and aliquoted to match turbidity of the McFarland standard of 3, corresponding to 9 × 10⁸ CFU/mL. To confirm the CFU/mL count, 10-fold dilutions of the adjusted inoculum was prepared, and 0.1 mL of each dilution was spread onto 5% sheep blood agar plates using a plastic disposable spreader. For each dilution, two blood agar plates were used. The plates were incubated in an AnaeroJar^™ (Thermo Scientific^™, Oxoid^™) at 37°C for 72 h in a micro-aerophilic atmosphere using a CampyGen^™ 2.5 L Sachet (Oxoid^™), and the viable count was determined.

Heifers were assessed once pre-vaccination, twice each week for 9 weeks post-vaccination and then weekly for 6 weeks post-inoculation. Assessments included ovarian ultrasound scanning using a 10 MHz rectal probe transducer (Honda 2,100 V, Honda Electronics, Toyohashi City, Japan) to monitor the stage of the oestrous cycle through recording the presence and size of CLs and large follicles. The oestrous cycle of heifers was confirmed at four time points before challenge and four time points after challenge with reference to Prostaglandin injection and ultrasonography scanning. Heifers were weighed at sample collection, and body condition was assessed (1–5 scale) as described previously (Edmonson et al., 1989). At each time point, vaginal mucus was collected using a Tricamper^™ (Queensland Department of Agriculture & Fisheries, Biosecurity Sciences Laboratory) by inserting the device into the vagina with the leading edge in contact with the dorsal wall of the vagina and moved back and forth in the vagina. After collection, the Tricamper^™ was transferred into 5 mL of chilled PBS. Coccygeal vein blood samples were collected in 8.5 mL Serum SST-II Advance tubes (BD Vacutainer^®) at each time point, allowed to clot, and then chilled overnight. Sera were collected following centrifugation at 1900 × g for 10 min, aliquoted, and stored frozen (−20°C) for subsequent progesterone assay and proteomics analyses.

Serum samples from vaccinated and unvaccinated heifers were pooled separately, except for heifers in dioestrus before challenge. A competitive Bovine Progesterone ELISA (Colorimetric, Novus Biologicals^™) quantified the progesterone (P4) concentration in each sample following the manufacturer’s protocol. In brief, thawed serum samples were mixed thoroughly, pooled (within group and time point), and diluted (10-fold) with 0.05% Tween Buffer. Diluted serum samples were added to a 96-well plate with HRP conjugates and antibodies. The reaction was stopped after adding substrates A and B and then read in a Multiskan^™ FC Microplate Photometer (Thermo Fisher Scientific^®) at 450 nm. The ELISA was conducted in duplicate with standards and blank samples. A standard curve was generated using four parameter logistic (4-PL) curve fit, and P4 concentrations were detected using linear regression analysis in Graph Pad Prism (version 9.5.1).

Vaginal mucus samples were collected four time points before challenge and four time points after challenge using the Tricamper^™ sampling tool in PBS. After removal of the tool head, the mucus samples were centrifuged at 800 × g to remove the host pellet, and the collected supernatant was centrifuged at 4000 × g to obtain the microbial pellet for DNA extraction. DNA was extracted using the QIAamp^® DNA Microbiome Kit (QIAGEN^™) following the manufacturer’s protocol. In brief, the microbial pellet was subjected to lysis buffer, benzonase enzyme (to deplete host DNA), and bead beating (to lyse bacteria) before column-based purification. The purified DNA was eluted in AVE buffer. The concentration of DNA was measured using the Qubit Fluorometer (Thermo Fisher Scientific^®), and extracted DNA samples were used for (1) C. fetus qPCR and (2) 16S rRNA gene amplicon sequencing through the University of Queensland’s Australian Centre for Ecogenomics (ACE). DNA was amplified by PCR with primers targeting V5–V8 hypervariable regions of the 16S rRNA genes using 803F (5′-TTAGAKACCCBNGTAGTC-3′) and 1392wR (5′-ACGGGCGGTGWGTRC-3′) primers (Engelbrektson et al., 2010), modified to contain Illumina specific adapter sequence (803F: 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGTTAGAKACCCBNGTAGTC-3′ and 1392wR: 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGACGGGCGGTGWGTRC-3′) in NEBNext^® Ultra^™ II Q5^® Mastermix (New England Biolabs).

An SNP that differentiated C. fetus venerealis and C. fetus fetus isolates was identified in the mraY gene (Ong, 2022). An “in house” TaqMan^™ SNP Genotyping Assay (Thermo Fisher Scientific^®) was developed by labelling C. fetus venerealis and C. fetus fetus-specific probes with VIC^™ and FAM^™ reporter dyes, respectively (Tabor et al., 2024). The AgPath-ID^™ One-Step RT-PCR Reagent (Thermo Fisher Scientific^®) without the Reverse Transcriptase step was provided as a 2x concentrated buffer used as the qPCR mastermix. Each reaction consisted of 1x buffer (5 μL), 900 nM of the forward (5’ AAAATGATGATGAATTGGCGCCATT 3′) and reverse (5’ TGTGATGGAAACCTTATCTGTTATATTGCA 3′) primers, 200 nM of the C. fetus venerealis mraY VIC probe (5’VIC-CGTTTTTTGTGTATTTT 3’MGBNFQ), 200 nM of the C. fetus fetus mraY FAM probe: (5’FAM-CGTTTTTTGCGTATTTT-3’MGBNFQ), and 25X RT-PCR Enzyme Mix (AmpliTaq Gold^™ DNA Polymerase at 0.025 units per reaction). Assays were run in duplicate in the Bio-Rad CFX96 Touch^™ Real-Time PCR Detection System for the two fluorophores under the following conditions: activation at 95°C 10 min, followed by 45 cycles of 95°C 15 s, 69°C 1 min, and a final extension at 69°C for 7 min. Raw amplification data (Cq values and relative fluorescence units (RFU)) were exported for analysis in Excel and RStudio (Tabor et al., 2024).

Serum samples from five vaccinated (V) and five unvaccinated (UV) cycling heifers were selected for proteomics study based on their qPCR results (C. fetus subsp. venerealis positivity) at three different time points: (1) pre-challenge (30.06.2021) when all the heifers were negative for C. fetus venerealis in qPCR; (2) post-challenge (02.07.2021) when all unvaccinated heifers were qPCR positive and vaccinated heifers were negative, and (3) post-recovery 42 days post-challenge (04.08.2021) when both vaccinated and unvaccinated heifers were qPCR negative. One unvaccinated and one vaccinated heifer showing persistent anoestrus were excluded from the study. The serum proteomes were compared pre-challenge (UV0 vs. V0), post-challenge (UV1 vs. V1), and post-recovery (UV2 vs. V2) to characterise the immune response of heifers with and without C. fetus venerealis (see Table 1).

Data-dependent acquisition (DDA) results were processed using ProteinPilot software (SCIEX^® 5.02) and searched against all bovine proteins in UniProtKB (downloaded on 22 February 2022; 47,124 total entries) with settings described by Raza et al. (2021). False discovery rate (FDR) analysis was conducted with limits of 99% confidence and 1% local FDR. To quantify the abundance of peptide in each sample, peak-view analysis was performed using PeakView 2.1 (SCIEX^®) with specific settings, including allowing shared peptides, a peptide confidence threshold of 99%, FDR set to 1%, XIC extraction window of 6 min, and XIC width of 75 ppm. A linear mixed model using MSstats (version 2.4) in R was conducted to identify differentially abundant proteins with Benjamini and Hochberg corrections to adjust for multiple comparisons and a significance threshold of p < 0.05 (adjusted p-value) (Kerr et al., 2019). Before conducting the statistical analysis, a Python script was used to remove data that did not pass FDR (significance threshold). The MSstats package was used for normalisation, data verification, and statistical testing to identify differentially abundant proteins. MSstats default setting was used as follows: “equalizeMedians” (default) which represents constant normalisation (equalising the medians) based on reference signals is performed. Variables included in the linear model were the Protein Name, Peptide Sequence, Precursor Charge, Fragment Ion, Product Charge, Isotope Label Type, Condition, Bio Replicate, Run, and Intensity (Choi et al., 2014). All significant proteins were included in subsequent analyses.

Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) was used to identify protein–protein interaction and characterisation for gene ontology (GO) terms for biological processes (BPs) and Kyoto Encyclopedia Genes and Genomes Pathways (KEGG) using Uniprot accession identifiers of significantly differentially abundant (DA) proteins as a target list (Szklarczyk et al., 2019). The Bos taurus genome was used as background in the STRING analysis with the following basic settings: meaning of network edges as evidence; active interaction sources included were text trimming, experiments, databases, co-expression, neighbourhood, gene expression, and co-occurrence; high confidence (0.700) for the minimum required interaction score, and k-means clustering with the number of clusters set at 3. Volcano plots were made with differentially abundant proteins (adjusted p < 0.05 and log2FC > 0.2) using ggplot2 and tidyverse in R (version 4.1.1). Functional enrichment analysis was performed with clusterProfiller (version 4.0), and graphs were produced with dot plot and category network plot functions from this package in RStudio.

As cycling can influence the microbiome content, oestrous and dioestrous phases were confirmed by rectal palpation, ultrasonography, and the progesterone ELISA assay. Follicular dynamics of each individual heifer and progesterone concentration are provided in Supplementary Figure S1.1: Follicular dynamics; Supplementary Figure S1.2: Progesterone (P4) ELISA curve; and Supplementary Table S1.1: Concentration of P4 in samples. Three vaccinated (200,639, 200,668, and 200,681) and one unvaccinated (200578) heifer did not express oestrous synchrony before challenge, while after challenge all heifers followed the expected pattern except for one unvaccinated heifer (200578) which was found consistently in anoestrus and was excluded from further analyses.

Successful challenge with C. fetus venerealis was confirmed by qPCR. Prior to challenge, all heifers were C. fetus venerealis qPCR negative at days 2, 7, 14, 21, and 28. After challenge, 67, 50, 33, 17, and 17% of unvaccinated heifers were qPCR positive, compared to 0, 33, 67, 33, and 0% of vaccinated heifers. Vaccinated heifers eliminated the infection sooner than the unvaccinated heifers at 4 weeks versus 5 weeks, respectively (Figure 2 and Supplementary material S2).

A total of 210 unique serum proteins were identified by ProteinPilot software (SCIEX^® 5.02) (Supplementary Table S4.1). The relative abundance of each protein within each group was quantified with an FDR cutoff value of 1%, and a total of 130 unique proteins was identified after Peakview 2.1 (SCIEX^®) analysis (Supplementary Tables S5.1–S5.7). The serum proteomes of unvaccinated and vaccinated heifers were compared at three time points, and significant proteins (adjusted p < 0.05) with log2FC more than 0.2 were considered as significantly differentially abundant proteins (DAPs) in each group for further analysis. The total numbers of DAPs found in pre-challenge (UV0 vs. V0), post-challenge (UV1 vs. V1), and post-recovery (UV2 vs. V2) groups were 45, 51, and 58, respectively, in which 28, 29, and 25 proteins were highly abundant in unvaccinated heifers (Figures 8, 9A–C and Supplementary Tables S6.1–S6.3).

Depending on the log2FC values, the most abundant proteins found in unvaccinated heifers before challenge were albumin (B0JYQ0; log2FC = 2.7), apolipoprotein D (APOD; log2FC = 2.1), and immunoglobulin (Ig) alpha 2 heavy chain (A5D7Q2_IgM; log2FC: 1.8). Kininogen-II (P01045; KNG2) and two complement components (A5D9D2; C4BPA and EIB805; C3) were also significantly higher in abundance (log2FC = 0.9, 0.9, 0.8, respectively) in the pre-challenge unvaccinated group. While kininogen-1 (KNG1; log2FC = 3.7), Ig-like domain-containing protein (F1MLW8; log2FC = 1.5), globin C1 (GLNC1; log2FC = 1.5), conglutinin (CGN1; log2FC = 1.3), and vitamin D-binding GC protein (Q3MHN5; log2FC = 0.4) were significantly higher in vaccinated pre-challenge heifers.

In the post-challenge comparison, albumin (ALB; log2FC = 1.4), KNG2 (log2FC = 1.4), and complement components, namely, C8G, C4BPA, C6, and C3, were highly abundant proteins in unvaccinated heifers. Haptoglobin homologs (HP-20 and HP-25) also had a significant abundance in post-challenge unvaccinated heifers. In the vaccinated post-challenge, serotransferrin (Q29443; log2FC = 1.8) was identified as the top high abundant significant protein along with immunoglobulin lambda-1 light chain (F1MLW8; log2FC = 1.4), endopin 2C (Q32T06; log2FC = 1.0), and apolipoprotein D (F1MS32; log2FC = 0.9). The abundance of IgG was higher in the unvaccinated heifers after challenge and increased in the vaccinated heifers after recovery. Globin C1, globinA1, and conglutinin were highly abundant proteins in pre-challenge and post-challenge vaccinated heifers while identified as highly abundant proteins in post-recovery unvaccinated heifers. Unvaccinated heifers also showed higher abundances of zinc alpha 2 glycoprotein (AZGPI; log2FC = 2.3), clusterin (CLU; log2FC = 1.1), alpha 2-macroglobulin (A2M; log2FC = 0.6), and CD5 antigen-like (CD5L; log2FC = 0.6) proteins after recovery.

When comparing the relative abundance of proteins within groups before and after challenge, the number of DAPs was lower in vaccinated (V0 vs. V1; n = 25) than the unvaccinated (UV0 vs. UV1; n = 43) (Figures 10A,C and Supplementary Tables S6.4, S6.5). Within UV0 vs. UV1 comparison, kininogen-1 (log2FC = 4.7), fetuin B (FETUB; log2FC = 1.4), C8B (log2FC = 0.8), HP homolog 25 (log2FC = 0.4), and uncharacterised proteins “identified as IgG after BLAST search” (A0A6B9SF17, A0A6B9SDW5, A0A3Q1M1Z4; log2FC = 15, 1.2, 1.2) increased in response to challenge. In the vaccinated group, IgG (A5D7Q2; log2FC = 1.4), albumin (log2FC = 1.1), clusterin (CLU; log2FC = 1.0), apolipoprotein C II (log2FC = 0.7), serotransferrin (log2FC = 0.7), and coagulation factor-5 (F5; log2FC = 0.7) increased in response to challenge. Comparing post-challenge with post-recovery, a total of 68 and 49 differentially abundant proteins were identified in the unvaccinated (UV1 vs. UV2) and vaccinated (V1 vs. V2) groups, respectively (Figures 10B,D; Supplementary Tables S6.6, S6.7). In unvaccinated heifers (UV1 vs. UV2), the abundances of complement components (C8B, CF1, C7, C6, C8G, C5, and C4BPA; log2FC = 2.0, 1, 1, 0.9, 0.5, 0.5, 0.5) and coagulation factor (F5; log2FC = 0.9) significantly decreased in post-recovery heifers. In vaccinated heifers, the abundances of KNG1, CLU, C5, and F5 were also reduced in post-recovery heifers, while C6, C8G, and F2 increased in abundance.

The bovine vaginal microbiome is known to be dominated by bacteria from the Firmicutes, Bacteroidota, and Proteobacteria phyla (Laguardia-Nascimento et al., 2015). The abundances of these bacterial phyla are associated with different factors including oestrous cycle, pregnancy, days of post-partum and post-partum uterine diseases, breed, age, and environment (Wang et al., 2018b; Ault et al., 2019a,b; Quereda et al., 2020; Gohil et al., 2022). The bovine reproductive tract microbiome is a reflection of environmental organisms presented in soil, skin, and water-confirmed in metagenomics studies published in Scientific Reports in 2022 (Ong et al., 2022a). In our study, vaccination against C. fetus venerealis did not affect the phyla of the vaginal microbiome of heifers with both vaccinated and unvaccinated heifers dominated by Proteobacteria, Firmicutes, Actinobacteriota, and Bacteroidota before challenge. Proteobacteria and Firmicutes were identified as the highest abundant bacteria in healthy heifers which is similar to studies reported by Tasara et al. (2023). Giannattasio-Ferraz et al. (2019) found Firmicutes as the dominant bacterial phylum followed by Bacteroidota, Proteobacteria, and Actinobacteria when characterising the vaginal microbiomes of healthy Gyr and Nellore cattle breeds.

Here, we have shown that the microbiome changes after a live challenge, with C. fetus venerealis. The mean abundance of Proteobacteria decreased, while Firmicutes and Campylobacterota increased. An increased abundance of Firmicutes is known to help to prevent the growth of pathogens by reducing the pH of vaginal mucus (Vieco-Saiz et al., 2019). Interestingly, we did not find an increase in Bacteroidota, Fusobacteria, and Porphyromonas in vaccinated heifers after challenge even though these phyla have been synergistically associated with post-partum clinical infections such as metritis, endometritis, and cervicitis (Wang et al., 2018a; Galvão et al., 2019, reviewed by Ong et al., 2021).

Firmicutes, especially Streptococcus spp., increased during oestrus over dioestrus after challenge in both vaccinated and unvaccinated heifers. Streptococcus spp. was more abundant in vaccinated than unvaccinated individuals. The presence of Streptococcus spp. has been shown to promote uterine health and is negatively correlated with uterine inflammation and pathogenic bacteria such as Trueperella pyogenes (Williams et al., 2005; Gilbert and Santos, 2016). From previous studies, it was observed that the bovine vaginal microbiome is hormone-dependent, and the heterogenicity in microbial communities observed during luteal and follicular phases impacts the vaginal microbiome of heifers (Laguardia-Nascimento et al., 2015; Quereda et al., 2020; Gohil et al., 2022). It has been observed that microbial phyla such as Firmicutes, Actinobacteria, and Bacteroidota are found to be prominent during oestrus in some cattle/buffalo, with a reduction in the Proteobacteria, which is high during dioestrus (Mahalingam et al., 2019). An increased abundance of Corynebacterium (Actinobacteriota) was found in the vaccinated heifers after challenge compared to unvaccinated in our study. Corynebacterium spp. (non-diphtheria) can produce bacteriocin, an antimicrobial agent which regulates vaginal microbiocenosis in humans and may prevent the growth of opportunistic bacteria (Gladysheva et al., 2022).

In this study, the challenge with C. fetus venerealis resulted in a shift in the heifer vaginal microbiomes, no clinical disease, and a return to oestrus. The mean abundance of Campylobacter spp. increased after challenge in both vaccinated and unvaccinated heifers. Although the abundance of Campylobacter was higher in unvaccinated heifers than vaccinated, the vaccinated heifers cleared the infection earlier as confirmed using qPCR. A recent study on the cervicovaginal microbiota of female beef cattle harbouring C. fetus venerealis showed that positive and negative heifers had similar bacterial composition and that the presence of C. fetus venerealis did not affect cyclicity which is similar to our study (De Carli et al., 2023). Normally, female cows infected with C. fetus venerealis can clear the infection and gain mucosal immunity within 3–5 months of infection. However, immunity is short-lived, and cows with compromised immunity often fail to return to oestrus (Bondurant, 2005).

Although 16S amplicon sequencing can provide an insight into microbial dysbiosis, it has some limitations such as the inability to identify to the species level and the biased amplification of 16S rRNA due to the variability of 16S genomic copy numbers in different species. Metagenomics by adaptive sampling is a long-read sequencing method that depletes host DNA to provide more accurate and unbiased profiling (Ong et al., 2022a,b). A further limitation of this study is the small number of animals used for the trial. It would be of interest to validate the results in a larger field study, utilising metagenomic adaptive sampling to further determine the association of BGC immunity with the vaginal metagenome.

To identify vaccination biomarkers, we compared the serum proteomic profile to identify differentially abundant proteins between the unvaccinated and vaccinated heifers before challenge. Conglutinin was highly abundant in the vaccinated group post-challenge. Conglutinin is an important immune response protein particularly associated with innate immunity. It plays a crucial role in recognising and eliminating pathogens by binding to the sugar surface of microorganisms and facilitates their removal by iC3b opsonin, a degradation product of the complement system, which is deposited on microbial cell surfaces (Friis-Christiansen et al., 1990; Dec et al., 2012). The comparison of serum proteomes of unvaccinated heifers following challenge (UV0 vs. UV1) identified several interesting biomarker candidates including kininogen I, fetuin B, complement components (C6, C3, C8B, C8G, and CF1), and HP-25 homologs in response to C. fetus venerealis infection. Fetuin B (a cystatin) inhibits ovastacin, which has been shown to be involved in zona pellucida hardening, thus interfering with fertilisation (Dietzel et al., 2013). Furthermore, fetuin B inhibits Meprin, which has been shown to modulate the immune system by processing and activating pro-inflammatory cytokines and chemokines that in turn induce the migration of leukocytes to sites of injury or infection (Herzog et al., 2019).

To characterise the immune response to C. fetus venerealis infection, we compared the serum proteomes of post-vaccination with those post-challenge from unvaccinated and vaccinated heifers (UV0 vs. V0 and UV1 vs. V1). Vitamin D-binding protein (DBP) was in high abundance in vaccinated cattle before challenge (UV0 vs. V0) and in unvaccinated cattle after challenge (UV1 vs. V1). Vitamin D-binding protein (DBP) can bind to proteoglycans present in the membrane of immune cells and enhance complement C5a-stimulated chemotactic activity of activated neutrophils (Metcalf et al., 1991). This protein has previously been identified as a biomarker for specific inflammatory diseases, such as cirrhosis (Lisowska-Myjak et al., 2020). Furthermore, DBP acts as an anti-tuberculosis agent, leading to increased bacterial killing by the induction of a group of antimicrobial peptides (Cathelicidins) through the triggering of Toll-like receptors (Liu et al., 2006) and thus has been identified as a potential biomarker for tuberculosis and paratuberculosis in human (Seth et al., 2009).

Protein HP-20 and HP-25 homologs were also observed as important highly abundant proteins when comparing UV1 and V1. The log fold change of these HP homologs increased in unvaccinated heifers after challenge. HP homolog proteins are acute-phase response proteins and are actively associated with phagocytosis and immune responses by inhibiting iron uptake of microbes. These proteins have been identified as important protein biomarkers for both clinical and subclinical endometritis and metritis in cattle (Miller et al., 2019; Menta et al., 2023). Clusterin is another important protein observed in post-challenge unvaccinated and vaccinated heifers in response to C. fetus venerealis infection. Log fold change of clusterin was higher in vaccinated (1.7) than unvaccinated (0.5) heifers. Clusterin is a glycoprotein which has diverse functions including stimulation of inflammatory cytokines, complement inhibition, immunity modulation, and cell invasion (Wilson and Zoubeidi, 2017). Research suggests that there is an association between CLU and pregnancy-related conditions such as intrauterine growth restriction and recurrent pregnancy loss in humans (Mohanty et al., 2020). The identified proteins appear to be part of a complex immune response to C. fetus venerealis infection. Their differential abundance could serve as potential biomarkers to monitor the infection or immune status of the host.