C. burnetii field isolate 26QC00015 originated from sheep afterbirth material collected in Thuringia, Germany, in 2009 and was obtained from the Friedrich-Loeffler-Institut strain collection. This isolate was propagated in mouse fibroblasts (L-929) in DMEM (Dulbecco’s Modified Eagle Medium, Lonza Group AG, Basel, Switzerland) supplemented with 5% fetal calf serum (FCS, Th. Geyer Ingredients GmbH & Co. KG, Höxter, Germany) at 37 °C and 5% CO₂ until approximately 80% infectivity was reached. Briefly, 1e+05 L-929 cells/mL were infected with a multiplicity of infection (MOI) of 100, and after 12–14 days, C. burnetii was harvested and quantified via quantitative real-time PCR with icd as target (icd-qPCR) as described previously (Klee et al., 2006). The isolate was subsequently propagated in acidified citrate cysteine medium (ACCM-2, Sunrise Science Products, San Diego, CA, USA) as described elsewhere (Omsland et al., 2011). After 7–10 days, bacteria were harvested by centrifugation (10,000 × g, 20 min, 4 °C) and stored in sucrose–glycerol buffer (270 mM sucrose, 10% glycerol) at −80 °C. Propagation, handling, processing, and storage of C. burnetii were carried out under BSL-3 conditions.

Acetonitrile (ACN), ammonium bicarbonate (NH4CO3), amidosulfobetaine (ASB-14), 5-bromo-4-chloro-3-indolyl phosphate (BCIP), bromophenol blue, chloroform, glycerol, dimethyl sulfoxide (DMSO), iodoacetamide, nitro blue tetrazolium chloride (NBT), phosphoric acid, polyethylene glycol (PEG), sodium dodecyl sulfate (SDS), sucrose, trifluoroacetic acid (TFA), tris(hydroxymethyl)aminomethane (TRIS), and TRIS–HCl were purchased from Merck KGaA, Darmstadt, Germany. Ammonium sulfate [(NH4)₂SO₄], dithiothreitol (DTT), ethanol, glycine, magnesium chloride hexahydrate (MgCl2 × 6H₂O), methanol, non-fat milk powder, sodium chloride (NaCl), thiourea, TRIS base, Triton X-100, Tween 20, and urea were obtained from Carl Roth, Karlsruhe, Germany. Coomassie Brilliant Blue G250 was purchased from SERVA Electrophoresis GmbH, Heidelberg, Germany.

Ruminant sera of Q fever–positive and –negative herds consisting of 2–30 single sera were retained from the serum collection of the National Reference Laboratory for Q fever (Friedrich-Loeffler-Institut, Jena, Germany) and used for immunoproteomic analyses and peptide-epitope screenings using fluorescent peptide antigen ELISA. Each serum pool arose from a single herd of known Q fever positivity or from apparently healthy and Q fever–negative herds. Single serum samples (Supplementary Table S1) and pools of sera (Table 1, Supplementary Table S2) were tested using Q fever (C. burnetii) Ab Test (IDEXX Laboratories, Inc., Westbrook, USA) or PrioCHECK Ruminant Q Fever Ab Plate Kit (Thermo Fisher Scientific) according to the guidelines from the supplier, with a sample-to-positive percentage (S/P %) of ≥40 recorded as positive, values <30 as negative, and values between ≥30 and <40 considered inconclusive.

Total protein extracts from axenic propagated C. burnetii isolates were prepared as described previously (Flores-Ramirez et al., 2014). Briefly, bacterial pellets (10 mg wet weight) were resuspended in lysis buffer (28 mM TRIS–HCl, 22 mM TRIS base, 200 mM DTT, 2% SDS) with protease inhibitor (Halt Protease Inhibitor, Thermo Fisher Scientific Inc., Waltham, MA, USA) and heat-inactivated (110 °C, 15 min) before exiting the BSL-3. Nucleic acids were degraded by adding 12.5 U Benzonase (Merck KGaA) for 12 h at 4 °C. Samples were centrifuged (12,500 × g, 20 min) and proteins precipitated from the supernatant by methanol–chloroform precipitation as described elsewhere (Wessel and Flugge, 1984). Protein pellets were air dried and resuspended in 300–500 μL isoelectric focusing (IEF) buffer (7 M urea, 2 M thiourea, 29 mM TRIS base, 1% ASB-14, 1% Triton X-100). For better solubility, a freeze–thaw step at −80 °C was included. Proteins were quantified using a 660-nm assay (Thermo Fisher Scientific Inc.).

Non-linear (NL) 24-cm Immobiline DryStrips (pH 3–10, Merck KGaA) were rehydrated overnight with 300 μg C. burnetii protein extracts for colloidal Coomassie staining or 200 μg for Western blotting in 450 μL IEF buffer containing 1.2% Destreak Reagent (Merck KGaA), 1.5% immobilized pH gradient (IPG) buffer 3–10 NL (Merck KGaA), 1% protease inhibitor, 1 μL Lightning Red stain (SERVA Electrophoresis GmbH), and 0.005% bromophenol blue. Isoelectric focusing was conducted on an Ettan IPGphor III unit (GE Healthcare, Chicago, IL, USA) with the following steps: (i) gradient, 200 V, 0:30 h, 50 Vh; (ii) 500 V, 7:00 h, 3,000 Vh; (iii) gradient 1,000 V, 1:00 h, 800 Vh; (iv) gradient 5,000 V, 1:30 h, 4,500 Vh; 5 (v) final step 5,000 V, 5:19 h, 42,000 Vh. Focused strips were equilibrated in equilibration buffer (6 M urea, 4% SDS, 30% glycerol, 50 mM TRIS–HCl, pH 8.8) containing 65 mM DTT, followed by equilibration buffer containing 203 mM iodoacetamide each for 15 min at room temperature. The second dimension was carried out using 12.5% TRIS–HCl polyacrylamide gels in Ettan DALTsix chambers (GE Healthcare). Gels for spot harvesting were stained overnight with colloidal Coomassie (20% ethanol, 1.6% phosphoric acid, 8% ammonium sulfate, 1% Coomassie Brilliant Blue G250) and destained in distilled water.

Parallel gels were blotted onto polyvinylidene fluoride (PVDF) membrane (0.45 μm, Merck KGaA) using TRIS–glycine buffer (25 mM TRIS, 192 mM glycine) with 20% methanol and blocked in TBS-T buffer (20 mM TRIS, 150 mM NaCl, 0.05% Tween 20, pH 7.6) containing 10% non-fat dry milk at 4 °C overnight. Blots were incubated with Q fever–positive serum pool 1 (Table 1; 1:500) or Q fever–negative immune sera from sheep in TBS-T buffer containing 2% non-fat dry milk for 1 h at room temperature. After washing, bound antibodies were traced by alkaline phosphatase–conjugated anti-sheep IgG or anti-goat IgG antibodies (Merck KGaA, 1:5,000) in TBS-T buffer containing 2% non-fat dry milk for 1 h at room temperature. Substrate solution containing NBT (0.33 mg/mL) and BCIP (1.65 mg/mL) in alkaline phosphatase buffer (100 mM NaCl, 100 mM TRIS, 5 mM MgCl₂ × 6H₂O, pH 9.5) was added, and color development was allowed to proceed until visible bands appeared.

Protein spots from colloidal Coomassie–stained gels and –immunostained membranes were matched with Delta2D 4.7 (DECODON Software UG, BioTechnikum Greifswald, Greifswald, Germany) and manually harvested. In-gel trypsin digestion was performed as described elsewhere with adaptations (Shevchenko et al., 2006; Flores-Ramirez et al., 2017). Briefly, spots were excised and destained with 50 mM NH₄CO₃ in 50% (v/v) ACN under gentle agitation. Subsequently, gel spots were dehydrated in 100% ACN, reduced in 10 mM DTT for 30 min at 50 °C, then further dehydrated and alkylated in 50 mM iodoacetamide for 30 min at room temperature in the dark while agitating. After additional dehydration and destaining, 10 ng/μL trypsin (Promega, Madison, WI, USA) in 10 mM NH₄CO₃ containing 10% ACN was added and left for 30 min at 4 °C, followed by incubation at 37 °C for an additional 14–16 h. Peptides were repeatedly extracted with 70% ACN containing 1% TFA, combined, and concentrated.

Peptides were analyzed by liquid chromatography–mass spectrometry (LC–MS/MS), using nanoAcquity UHPLC and Q-TOF Premier controlled by MassLynx software 4.1 (Waters, GmbH, Eschborn, Germany). Analytes were loaded onto the Symmetry C18 trap column (20 mm length, 180 μm diameter, 5 μm particle size, Waters). After 3 min of desalting/concentration by 1% acetonitrile containing 0.1% formic acid at a flow rate of 10 μL·min⁻¹, samples were passed through a BEH130 C18 analytical column (200 mm length, 75 μm diameter, 1.7 μm particle size, Waters). For quick separation, a 20-min gradient of 5–40% acetonitrile with 0.1% formic acid was used, at a flow rate of 300 nL·min⁻¹. The column outlet was connected to a PicoTip emitter (360 μm outer diameter, 20 μm inner diameter, 10 μm tip diameter, New Objective), and samples were nanosprayed (3.4 kV capillary voltage). Next, the spectra were recorded in the data-independent acquisition MSE (mass spectrometry efficiency) mode using MassLynx software 4.1 (Waters, GmbH, Eschborn, Germany). Alternate scans at low (4 eV) and high (20–40 eV ramp) collision energies were employed to obtain precursors and fragment masses in a single chromatographic run. Ions with 50–1,950 m·z⁻¹ were detected in both channels, although quadrupole mass profile settings allowed efficient deflection of masses <400 m·z⁻¹ in the low energy mode, enabling filtering of contaminating ions. The spectra acquisition scan rate was 0.8 s, with a 0.05-s inter-scan delay. The external mass calibrant Glu1-fibrinopeptide B (500 fmol·mL⁻¹, Sigma-Aldrich) was infused through the reference line at a flow rate of 500 nL·min⁻¹. It was measured every 30 s at 22 eV collision energy and used for the mass correction.

Data were processed using the ProteinLynx Global Server v. 3.0 (Waters, GmbH, Eschborn, Germany). For peak detection, the following thresholds were used: low energy of 140 counts, high energy of 30 counts, and intensity of 1,000 counts. Precursors and fragment ions were coupled using correlations of chromatographic elution profiles in low/high energy traces. The results were searched against Coxiella burnetii RSA 493 full proteome (UniProt accession: UP000002671; 1,812 sequences) downloaded from UniProt. Database searches were performed with carbamidomethylation of cysteine as a fixed modification, oxidation of methionine and deamidation of asparagine and glutamine as variable modifications, a maximum of one missed tryptic cleavage site, and automatic estimation of precursor and fragment mass errors from calibrant peaks. A 4% false discovery rate (FDR) against a randomized database, which was applied at the individual peptide level, was allowed. The accepted peptide score threshold was set to ≥6.00 to ensure a 95% confidence level. Additional quality matching criteria were as follows: (i) a minimum of three consecutive product ions per peptide; (ii) a minimum of seven total product ions per protein; (iii) a minimum of two peptides fitting the protein sequence.

Amino acid sequences of all detected proteins were retrieved from the UniProt database for the C. burnetii reference strain Nine Mile phase I (RSA 493; UniProt accession: UP000002674). Deepnog v1.2.3 (Feldbauer et al., 2021) was used for Clusters of Orthologous Groups (COGs) assignment.

Amino acid (aa) sequences from immunogenic proteins described in the literature (Gerlach et al., 2017) or identified by 2D gel electrophoresis and Western blotting were retrieved from UniProt¹ for the C. burnetii reference strain (RSA 493). Protein–protein Basic Local Alignment Search Tool (BLASTp) was used to identify C. burnetii aa sequences. All sequences were downloaded and aligned using Geneious Prime software (Geneious Prime 2021.0.1, Biomatters Ltd., Auckland, New Zealand) to identify protein regions, with a sequence identity of >87% for C. burnetii isolates and <50% for non-Coxiella bacteria.

B-cell epitope prediction for each protein was performed using IUPred3 (Rahman et al., 2015). Peptides of 24–30 aa length or of 18–21 aa length were selected and confirmed using BLASTp to exclude peptides with more than 50–75% aa identity to other ruminant pathogens.

Peptides were synthesized as 2 μmol crude extract with N-terminal biotin SGSG linker and a C-terminal amide group (peptides and elephants, Henningsdorf, Germany). Upon arrival, lyophilized peptides were dissolved in ultrapure water at a concentration of 0.7 μmol/mL, or if not soluble, in 50% dimethylsulfoxide (DMSO) with distilled water (0.35 μmol/mL) or DMSO (0.7 μmol/mL) and stored at −80 °C.

ELISAs were performed as previously described, with modifications (Rahman et al., 2015). Briefly, white flat-bottom 96-well plates with covalently linked streptavidin (Nunc; Fisher Scientific, USA) were used and initially washed one time with 400 μL and two times with 300 μL wash buffer (0.3 M NaCl, 20 mM TRIS–HCl (pH 7.5), 0.1% Tween 20, 0.001% benzalkonium chloride, pH 7.5) for 15 min at room temperature. Peptides were diluted in assay diluent (0.2 M NaCl, 20 mM TRIS–HCl (pH 7.5), 10% chicken serum (Merck KGaA), 0.5% PEG, 0.1% Tween 20, 0.004% benzalkonium chloride, pH 7.5) to final concentrations of 0.75–25 pmol. The binding of peptides was carried out by adding 100 μL peptide solution to each well for 30 min at room temperature with agitation (60 rpm). Unbound peptides were removed by five washes with 300 μL wash buffer, and non-specific binding was blocked with 300 μL blocking buffer per well [0.2 M NaCl, 20 mM TRIS–HCl (pH 7.5), 10% chicken serum, 1% PEG, 0.004% benzalkonium chloride, pH 7.5] for 30 min at room temperature. Serum pools were diluted 1:100 in assay diluent; 100 μL per well was added and incubated for 1 h with agitation (60 rpm). The plate was washed five times with 300 μL of wash buffer, and 100 μL of peroxidase-conjugated anti-goat/sheep IgG monoclonal antibody (Merck KGaA), diluted 1:8,000 in assay diluent, was added for 30 min at room temperature. After five washes with 300 μL of wash buffer, the reaction was traced with QuantaBlu Kit (Thermo Fisher Scientific) as described by the supplier. The relative fluorescence units (RFUs) of each well were recorded, with excitation and emission wavelengths set at 325 nm and 420 nm, respectively.

All serum pools and single serum samples were tested with peptide-coated and -uncoated wells for background correction. A 150% correction factor relative to the signal intensity from the uncoated wells was subtracted from the peptide-coated wells for each serum. Signals above the background indicated peptide reactivity for the respective serum pool or serum (Rahman et al., 2015). Peptides that reacted with Q fever–positive sera but not with negative sera were considered specific peptide epitopes. Pairwise comparison was performed using GraphPad Prism 10, Version 10.6.1, using the two-way ANOVA tool. Reactivity with all 15 peptide epitopes from serum pools derived from true-positive sheep flocks (pools 1–4) or vaccinated animals (pool 9) was compared with serum pools derived from true-negative sheep flocks (pools 5–8) (Supplementary Table S3). p-value ≤ 0.05 were considered statistically significant.

Data were analyzed using Excel (Microsoft Office Professional Plus Version 1808) and GraphPad Prism 10. Receiver operating characteristic (ROC) curves were generated with MedCalc Version 14.8.1 (MedCalc Software Ltd., Ostend, Belgium). For a six-peptide combination, a composite score was calculated for each serum sample with the corresponding Youden Index set as the individual cutoff. This score, ranging from 0 to 6, was used for ROC analysis in comparison with the commercial ELISA. Individual serum samples were classified as positive based on ELISA results (S/P% > 40), qPCR confirmation of the infection, and/or herd status. Vaccinated animals and incomplete data sets were excluded. Sensitivity (Se) was calculated as TP/(TP + FN) and specificity (Sp) as TN/(TN + FP) with TP, true positive; TN, true negative; FP, false positive; and FN, false negative for selected peptide panels. Youden Index was calculated as follows: Se + Sp - 1 (Schisterman et al., 2008). A Youden Index of 1 describes a perfect test with no false-positive or false-negative results, a value above 0.5 describes an acceptable test, and 0 indicates a test with no discriminatory power.

A total of 156 seroreactive proteins were detected from C. burnetii 26QC00015 protein extracts using 2D gel electrophoresis and Western blotting with a pool of sheep sera collected during a Q fever outbreak (pool 1, Table 1). Representative images of the colloidal Coomassie–stained gel, Western blotting, and spot matching are displayed in Figure 1. Additional images of Coomassie-stained gels, Western blot spot matching, and identified immunoreactive proteins are provided in Supplementary Figures S1–S4 and Supplementary Table S4.

Detected proteins were mostly involved in general metabolism (46.15%), falling into COG categories for energy production and conversion (COG C), coenzyme (COG H), lipid (COG I), and amino acid (COG E) transport and metabolism (Figure 2). Another 32.05% were involved in cellular processes and signaling, while 17.95% were associated with information storage and processing. Notably, several proteins belonged to the cell wall/membrane/envelope biogenesis (COG M, 11.54%), including the potential virulence factor enhanced entry protein EnhA.1. Of the most frequently detected and previously described seroreactive proteins, eight highly conserved cytosolic proteins involved in metabolism, cellular processing, and signaling, as well as information storage and processing, were detected. This includes the well-characterized membrane-bound or outer membrane–bound proteins Com1 and CBU_0482, as well as the secreted protein and potential virulence factor Mip.

The majority of the detected proteins (n = 130) were identified in only one or two experiments, while 26 proteins were reproducibly detected three or four times and are listed in Table 2. Most of these frequently detected proteins were involved in general metabolism (7.69%), primarily in energy production and conversion (COG C) and lipid transport and metabolism (COG I). Other frequently detected proteins were associated with information storage and processing (5.13%) and cellular processes and signaling (3.85%) (Figure 2). Only two membrane-bound or periplasmic proteins, CBU_1538 and TolB, were identified as seroreactive in this study.

Based on a previously published literature review of C. burnetii immunoreactive proteins, nine proteins were selected for epitope prediction based on the highest publication frequency and the highest number of sera used in these studies (Gerlach et al., 2017). These included outer membrane proteins Com1 (CBU1910) and OmpH (CBU0612), periplasmic protein YbgF (CBU_0092), peripheral cell membrane protein DnaK (CBU_1290), cytosolic proteins GroEL (CBU_1718), Tuf-2 (CBU_0236), and SucB (CBU_1398), secreted protein Mip (CBU_0630), and one protein of unknown function and location (CBU_0937). Subsequently, based on detection frequency in 2D gel electrophoresis and Western blot analyses, four additional proteins were selected: the peripheral cell membrane protein Coq7 (CBU_1870), the cytosolic protein ScpB (CBU_1060), and two proteins of unknown location, CBU_0974 and CBU_0943 (Supplementary Table S4).

From these 13 proteins, 30 peptides were selected for synthesis (Supplementary Table S5). The selection criteria were a disordered score (IUpred3) > 0.5 and high sequence specificity: >87% identity to published C. burnetii isolates and ≤87% to Coxiella-like endosymbionts. To minimize cross-reactivity, we confirmed that these peptides shared <73% sequence identity with homologous proteins from other common bacterial pathogens, including Staphylococcus spp., Nocardia spp., Mycobacterium spp., Escherichia spp., Leptospira spp., Salmonella spp., Enterococcus spp., and Brucella spp.

All 30 peptides were screened in triplicate against nine field serum pools, comprising 90 individual serum samples from sheep flocks positive or negative for Q fever, as well as from vaccinated animals. Notably, serum pool 9 consisted of two individual samples (14QC1941 and 14QC1945) that tested negative in a commercial ELISA when analyzed separately. However, when combined into this pool, they yielded a positive signal (Table 1, Supplementary Table S2). These samples were originally collected during an outbreak in North Rhine-Westphalia (Pool 1). A reaction was considered positive if it exceeded 150% of the background signal in at least two replicates.

Fifteen peptides were excluded from further analysis due to a lack of reactivity or inconsistent (i.e, non-reproducible reactivity across independent experiments) results (data not shown). The remaining 15 peptides reacted with at least one Q fever–positive serum pool and showed no reaction with Q fever–negative serum pools (Table 3; Supplementary Figure S5 and Supplementary Table S2). The majority of these peptides reacted with two (n = 3) or three (n = 7) positive serum pools. Only peptide no. 22 reacted with all five positive serum pools. No peptide reacted with the negative serum pools, except peptide 20, which showed a single positive reaction in one of three independent experiments. The serum pool from vaccinated animals reacted with five peptides, all of which were also recognized by Q fever–positive serum pools.

To further evaluate the diagnostic potential of selected peptides for applications requiring high sensitivity (e.g., prevalence studies) or high specificity (confirmatory testing of a suspected or clinical case), they were tested against individual serum samples from sheep of known Q fever status. Of the 90 sera used to create the initial pools, 79 were selected for ELISA analysis. The remaining 11 samples were unavailable due to their insufficient volume.

All serum samples were grouped as originating from “infected,” “uninfected,” or “vaccinated” animals.

Reactivity with peptides was detected in all groups. In the infected group, all serum samples (n = 60) showed reactivity to at least one peptide; in some cases, up to 10 peptides were recognized. Individual serum samples (11/13) from the vaccinated group, which tested negative in pool analysis, showed reactivity with up to seven peptides (see also Supplementary Table S1).

To determine the diagnostic potential of each peptide, ROC curves were generated and compared with the commercial ELISA. In Figures 3A–C, the ROC curves are displayed for six individual peptides with the highest discriminatory power. The AUC values for individual peptides ranged between 0.5 and 0.7, indicating poor to fair discriminatory power (Supplementary Figure S6). In contrast, the commercial ELISA shows near-perfect discrimination, likely caused by the selection of true-positive and true-negative serum samples.

To evaluate whether peptide combinations improve the diagnostic performance, Se and Sp were calculated for all possible peptide combinations using the rule that a sample was considered “overall positive” if it reacted to at least one of the selected peptides (Supplementary Table S6). The highest Youden Index of 0.55, indicating an acceptable discriminatory power, was achieved for peptide combinations comprising six, seven, or eight peptides (Table 4). The generated ROC analysis (Figure 3D) reveals that by using a combination of peptides, the discriminatory power improved. However, the plateau of the curve indicates that using more than six peptides provides limited further benefit.

Taken together, these results show that the selected peptides allowed the detection of antibodies specific to C. burnetii in serum samples of both infected and vaccinated animals. Although reactivity was observed in serum samples from the uninfected group, the overall signal intensity and number of reactive peptides were lower than in the infected group. The best-performing combination reached a Se of 80% and Sp of 75%. This indicates that the approach attempted here to select and use peptides for Q fever diagnosis is promising; however, achieving robust diagnostic performance will require epitope prediction and validation on a much larger scale.