Identification of Immunogenic Linear B-Cell Epitopes in C. burnetii Outer Membrane Proteins Using Immunoinformatics Approaches Reveals Potential Targets of Persistent Infections

Coxiella burnetii is a global, highly infectious intracellular bacterium, able to infect a wide range of hosts and to persist for months in the environment. It is the etiological agent of Q fever—a zoonosis of global priority. Currently, there are no national surveillance data on C. burnetii’s seroprevalence for any South American country, reinforcing the necessity of developing novel and inexpensive serological tools to monitor the prevalence of infections among humans and animals—especially cattle, goats, and sheep. In this study, we used immunoinformatics and computational biology tools to predict specific linear B-cell epitopes in three C. burnetii outer membrane proteins: OMP-H (CBU_0612), Com-1 (CBU_1910), and OMP-P1 (CBU_0311). Furthermore, predicted epitopes were tested by ELISA, as synthetic peptides, against samples of patients reactive to C. burnetii in indirect immunofluorescence assay, in order to evaluate their natural immunogenicity. In this way, two linear B-cell epitopes were identified in each studied protein (OMP-H(51–59), OMP-H(91–106), Com-1(57–76), Com-1(191–206), OMP-P1(197–209), and OMP-P1(215–227)); all of them were confirmed as naturally immunogenic by the presence of specific antibodies in 77% of studied patients against at least one of the identified epitopes. Remarkably, a higher frequency of endocarditis cases was observed among patients who presented an intense humoral response to OMP-H and Com-1 epitopes. These data confirm that immunoinformatics applied to the identification of specific B-cell epitopes can be an effective strategy to improve and accelerate the development of surveillance tools against neglected diseases.


Introduction
Coxiella burnetii is a polymorphic and obligate intracellular Gram-negative bacterium that is highly infectious and can persist in the environment for months. It is the causative agent of Q fever in humans and coxiellosis in animals-a worldwide disease that is 1 of 13 global priority zoonoses [1][2][3]. In addition, C. burnetii is classified as a potential agent of bioterrorism, due to its remarkable resistance to environmental stress, extremely low infectious dose, and ease of dissemination [4][5][6]. A wide range of animals is known to to diagnosis based on whole antigens, presents low-cost production, higher specificity and reproducibility with no batch-to-batch variation, and large-scale production [56][57][58]; however, it still depends on the accurate identification of more immunodominant epitopes. Therefore, this study aimed to identify B-cell epitopes from C. burnetii outer membrane proteins (OMP-P1, OMP-H, and Com-1) in silico, and to validate them against samples from Brazilian exposed individuals.

Prediction of Linear B-Cell Epitopes in C. burnetii OMPs
In order to improve the accuracy of in silico identification of epitopes, we used a combination of three prediction algorithms: BepiPred 1.0, ABCpred, and ESA. Firstly, 10 sequences were predicted as linear B-cell epitopes in studied proteins-2 sequences in OMP-H, and 4 sequences each in Com-1 and OMP-P1. Among all 10 sequences, 4 were predicted by all algorithms, while 3 were predicted by Bepipred and ABCpred, 2 by ABCpred and ESA, and 1 potential epitope by Bepipred and ESA (Table 1).
As shown in Table 4, 77% of studied individuals (n = 20) presented antibodies against at least one of the identified epitopes. Regarding the reactivity against each studied protein epitope, 18 patients (69%) responded to at least one of the OMP-H epitopes, 17 patients were reactive to at least one of the Com-1 epitopes, while 15 individuals presented antibodies against at least one of the OMP-P1 epitopes.

Discussion
Most C. burnetii infections are asymptomatic or associated with nonspecific symptoms. Based on this, seroprevalence surveys arise as a potential way to evaluate the real prevalence of C. burnetii [19]. However, despite studies in the Netherlands [59,60], Australia [61], the USA [62], Bhutan [63], Northern Ireland [64], Cyprus [65], and South America [66,67], the true global prevalence of C. burnetii infections is still unknown due to Pathogens 2021, 10, 1250 9 of 18 the dearth of representative national seroprevalence surveys in the world. In this scenario, the identification of linear B-cell epitopes arises as a potential approach to improve and accelerate the development of novel diagnostic tests that could allow effective C. burnetii seroprevalence surveys.
In this study, we identified B-cell epitopes in antigen candidates for the development of serological tests for C. burnetii, and showed their immunogenicity in individuals naturally exposed to infection. In this way, we explored three C. burnetii outer membrane proteins (OMPs): OMP-H (CBU_0612), an outer membrane chaperon protein highlighted as a very strong immunodominant marker for both acute and persistent forms of Q fever [35,41]; Com-1 (CBU_1910), a strongly immunogenic protein [38,40] that is considered a reliable Q fever serodiagnosis marker [42,43]; and OMP-P1, a porin, highly expressed in large-cell variants (LCVs)-the replicative form of the bacterium in the cell hosts-and downregulated in small-cell variants (SCVs), the metabolically inactive and resistant form found in the environment [44][45][46]. However, the use of whole antigens in serodiagnosis may result in cross-reactivity with other related proteobacteria. Based on this, the B-cell epitopes' identification arises as a promising alternative to improve the specificity of serological tests for C. burnetii.
In Brazil, although the presence of C. burnetii has been demonstrated in animals [20][21][22][23], humans [21,[23][24][25][26][27] and, most recently, in artisanal cheese [28] and unpasteurized milk [29], Q fever remains poorly understood and reported. Moreover, despite the possibility of mistaking Q fever for influenza, dengue, malaria, leptospirosis, and other infectious diseases [30], in Brazil, only suspected rickettsiosis cases are investigated as Q fever, corroborating the underreporting of this zoonosis, and explaining the limited number of C. burnetii-reactive samples in this study (n = 26). Remarkably, similar-sized samples were previously used to successfully define C. burnetii proteins as antigens to serological tests [37,[84][85][86], indicating that the number of patients used in our study was sufficient to prove the natural immunogenicity of the identified epitopes.
Regarding the response against each epitope, we observed that endocarditis seems to be more frequent in patients who are HR to the OMP-H and Com-1 epitopes (Figure 3), suggesting that an intense humoral response to the Com-1 and OMP-H epitopes may be associated with chronic Q fever complications. This hypothesis corroborates the potential of recombinant OMP-H as an antigen for the serological diagnosis of C. burnetii endocarditis cases [37]; however, more studies involving a higher number of patients and complications are still necessary to prove this hypothesis. Moreover, we observed low hospitalization rates in high responders to Com-1 epitopes, while we observed a low frequency of prostration reports in non-responders to OMP-H epitopes when compared to LR and HR to OMP-H epitopes. Finally, the humoral response against OMP-P1 seems not to be associated with clinical features, given that OMP-P1 (197)(198)(199)(200)(201)(202)(203)(204)(205)(206)(207)(208)(209) was the epitope with the lowest recognition rate, and did not present high responders, while we observed no differences in clinical features among HR, LR, and NR to OMP-P1 (215)(216)(217)(218)(219)(220)(221)(222)(223)(224)(225)(226)(227) . These data corroborate the study of Qingfeng Li et al., which showed that recombinant OMP-P1 induced a low antibody level, but a high cellular immune response, in immunized mice [92]. Here, we observed that OMP-P1 linear B-cell epitopes are poorly immunogenic, but more studies aiming to identify its TCD4 and TCD8 epitopes are still necessary, and may be used to improve vaccine development [93].
In conclusion, despite the limitations in sample size and study design, six linear B-cell epitopes were identified by a combination of in silico tools and, subsequently, confirmed as naturally immunogenic in individuals exposed to C. burnetii. This was the first study evaluating the natural immunogenicity of the predicted linear B-cell epitopes in the C. burnetii outer membrane proteins OMP-H, Com-1, and OMP-P1 using samples of human patients. Therefore, more serosurveys regarding these epitopes are encouraged, and are essential in order to estimate their real specificity and sensitivity. Currently, synthetic peptides representing B-cell epitope sequences are considered a valuable tool as novel molecules for the reliable and rapid diagnosis of infectious diseases [58,94], with promising results for Q fever/coxiellosis [49,55], as corroborated by this study. When compared to whole proteins, the use of synthetic peptides as diagnostic antigens allows a higher specificity and reproducibility with no batch-to-batch variation, along with easy and low-cost production. In this context, we believe that the identification of B-cell epitopes may be an effective strategy to improve and accelerate the development of surveillance tools for Q fever and other neglected diseases.

B-Cell Epitope Prediction
Firstly, based on information from the UniProt database, transmembrane regions, signal peptides, and cytoplasmic regions of the studied proteins were excluded from Bcell epitope predictions. The identification of linear B-cell epitopes in the extracellular regions of the studied proteins was performed based on the combination of three different algorithms: Bepipred 1.0 (threshold: 0.35), Emini Surface Accessibility prediction (ESA) (threshold 1.0), and ABCpred (threshold 0.75). Sequences predicted by at least two of the used algorithms were considered predicted B-cell epitopes, and were evaluated for antigenicity using VaxiJen (http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html accessed on 1 June 2021) (threshold 0.4).
Briefly, Bepipred takes a single sequence in FASTA format input, and each amino acid receives a prediction score based on hidden Markov model profiles of known antigens and incorporates propensity scale methods based on hydrophilicity and secondary structure prediction. For each input sequence, the server outputs a prediction score [95]. Meanwhile, ESA calculates the surface accessibility of hexapeptides, with values greater than 1.0 indicating an increased probability of being found on the surface [96]. Moreover, ABCpred is the first server developed based on a recurrent neural network (machine-based technique), using fixed-length patterns to predict B-cell epitopes in an antigen sequence, with 65.93% accuracy [97]. VaxiJen is the first server for alignment-independent prediction of protective antigens; it was developed to allow classification of antigens based solely on the physicochemical properties of proteins, without recourse to sequence alignment. A bacterial protein dataset, threshold 0.4, was used to derive models for the prediction of whole-protein antigenicity, showing prediction accuracy from 70% to 89% [98,99]. All prediction algorithms were accessed in March 2019.

Evaluation of B-Cell Epitopes' Degree of Conservation
To evaluate the degree of conservation of the predicted epitopes, the identified sequences were compared for similarity with bacteria from the phylum Proteobacteria (Francisella tularensis, Legionella Pneumophila, Escherichia coli, Pseudomonas Aeruginosa, Rickettsia rickettsii, Ehrlichia chaffeensis, Bartonella henselae, Brucella melitensis, Afipia felis, and Campylobacter jejuni) and microbiota bacteria (Enterobacter spp., Klebsiella spp., Citrobacter spp., and Enterococcus coli) using BLASTp (https://blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE=Proteins accessed on 21 June 2021). When compared to this group of organisms, the sequences that presented E-values higher than 1 were considered to be non-conserved B-cell epitopes.

Three-Dimensional (3D) Structure Modeling
The 3D structures of the studied OMPs were modeled using the Robetta server (http: //new.robetta.org/ accessed on 21 June 2021); this server is continually evaluated through continuous automated model evaluation (CAMEO), and its primary service is to predict the 3D structure of a protein given its amino acid sequence [100][101][102][103]. The PDB formats reported from this server were visualized using PyMOL V1 viewer software [104,105].

Evaluation of Epitopes' Exposure in the Protein Quaternary Structures
Considering that the OMPs generally form oligomers, the modeled 3D structures of the studied proteins were submitted to the GalaxyHomomer server (http://galaxy.seoklab. org/cgi-bin/submit.cgi?type=HOMOMER accessed on 21 June 2021), which performs comparative modeling of quaternary structures [106,107]. In this study, it was defined that the sequences that had at least 30% of their amino acids interacting with other chains were considered hidden in the quaternary structure of the protein, and were excluded from further analyses. Finally, sequences predicted as antigenic and linear B-cell epitopes-nonconserved among proteobacteria, and exposed in quaternary structures-were selected for experimental validation. The schematic abstract of the prediction of B-cell epitopes is represented in Figure 4.

Peptide Synthesis
Sequences predicted as antigenic linear B-cell epitopes were chemically synthesized by the company GenOne Biotechnologies, Brazil. Analytical chromatography of the peptides demonstrated a purity of approximately > 95%, and mass spectrometry analysis of the peptides indicated their expected mass.

Studied Population
A total of 57 samples of suspected Q fever cases were provided by the Brazilian National Reference Center for Rickettsioses-Laboratory of Hantaviruses and Rickettsiosis-Oswaldo Cruz Foundation. Serum samples were previously tested by IFA for the presence of immunoglobulin G antibodies (IgG) against Coxiella burnetii, using a commercial kit from Scimedx ® (Dover, NJ, USA) [108]. Firstly, suspected samples were screened at a 1:64 dilution, and those that were considered positive were further diluted to determine the IFA s end titers. Negative and positive controls were included for each test run. The study was reviewed and approved by the Oswaldo Cruz Foundation Ethical Committee and the National Ethical Committee of Brazil (number CAAE: 39056120.6.0000.5248). were selected for experimental validation. The schematic abstract of the prediction of Bcell epitopes is represented in Figure 4.

Peptide Synthesis
Sequences predicted as antigenic linear B-cell epitopes were chemically synthesized by the company GenOne Biotechnologies, Brazil. Analytical chromatography of the peptides demonstrated a purity of approximately > 95%, and mass spectrometry analysis of the peptides indicated their expected mass.

Studied Population
A total of 57 samples of suspected Q fever cases were provided by the Brazilian National Reference Center for Rickettsioses-Laboratory of Hantaviruses and Rickettsiosis-Oswaldo Cruz Foundation. Serum samples were previously tested by IFA for the presence of immunoglobulin G antibodies (IgG) against Coxiella burnetii, using a commercial kit from Scimedx ® (Dover, NJ, USA) [108]. Firstly, suspected samples were screened at a 1:64 dilution, and those that were considered positive were further diluted to determine the IFA´s end titers. Negative and positive controls were included for each test run. The study was reviewed and approved by the Oswaldo Cruz Foundation Ethical Committee and the National Ethical Committee of Brazil (number CAAE: 39056120.6.0000.5248).

Experimental Confirmation of the Antigenicity of the Predicted Epitopes
Samples of suspected and confirmed Q fever cases were screened for the presence of naturally acquired antibodies against the synthetic peptides via ELISA. Briefly, MaxiSorp 96-well plates (Nunc, Rochester, NY, USA) were coated with 100 µg/mL of a peptide. After overnight incubation at 4 • C, plates were washed with phosphate-buffered saline (PBS) and blocked with PBS-containing 5% non-fat dry milk (PBS-M) for 1 h at 37 • C. Individual plasma samples diluted 1:100 on PBS-M were added in duplicate wells, and the plates were incubated at 37 • C for 2 h. After three washes with PBS-Tween20 (0.05%), bound antibodies were detected with peroxidase-conjugated goat anti-human IgG (Sigma St. Louis, MO, USA), followed by TMB (3,3',5,5'-tetramethylbenzidine). The absorbance was read at 450 nm using an xMark™ microplate absorbance spectrophotometer (Bio-Rad, Hercules, CA, USA). The results for total IgG were expressed as the reactivity index (RI)-the ratio between the mean optical density (OD) of tested samples and the mean OD of 31 negative controls plus 2 standard deviations (SD). Subjects were considered IgG responders to a particular antigen if the RI was higher than 1.

Statistical Analysis
All statistical analyses were carried out using Prism 5.0 for Windows (GraphPad Software, Inc., San Diego, CA, USA). The one-sample Kolmogorov-Smirnoff test was used to determine whether a variable was normally distributed. Differences in frequencies of IgG responders to synthetic peptides were evaluated using Fisher's exact test. The Mann-Whitney test was used to compare reactivity indices against synthetic peptides between responders to each epitope. A two-sided p-value < 0.05 was considered significant.

Conclusions
Our data corroborate the use of immunoinformatics approaches to identify epitopes as targets of humoral response that can be further explored in the development of novel surveillance tools for Q fever and other neglected diseases. Here, exploring linear B-cell epitopes in C. burnetii-OMP-H, Com-1, and OMP-P1-we experimentally confirmed the natural immunogenicity of two epitopes in each studied protein. Moreover, our data suggest that an intense humoral response to OMP-H and Com-1 epitopes may be related to Q fever complications, such as endocarditis.

Institutional Review Board Statement:
This study was conducted in accordance with the guidelines of the Declaration of Helsinki, and approved by the Oswaldo Cruz Foundation Ethical Committee and the National Ethical Committee of Brazil (number CAAE: 39056120.6.0000.5248).

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
Publicly available datasets were analyzed in this study. This data can be found here: https://www.uniprot.org/ accessed on 16 August 2021.

Conflicts of Interest:
The authors declare no conflict of interest, and the funders had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.