Introduction
The causative agent, Bartonella bacilliformis, is transmitted to humans by sandflies (Lutzomyia spp).
In the acute phase (Oroya fever), B bacilliformis infects human erythrocytes, causing a severe haemolytic anaemia with a case-fatality rate of up to 88% if untreated (decreasing to 0·7–10% if adequately treated). This acute stage is often followed by a chronic phase (verruga peruana) characterised by subcutaneous nodules as a result of vasculoendothelial proliferations.
The seroprevalence of B bacilliformis antibodies is up to 65% in endemic areas of Peru.
The rate of asymptomatic human carriers of B bacilliformis is 37% in post-outbreak areas and 52% in endemic areas; asymptomatic individuals represent the main reservoir of the pathogen.
Microbiological culture methods are limited by the slow growth of B bacilliformis (2–6 weeks), but several PCR-based diagnostic approaches that use peripheral blood have been established.
,
Although PCR-based diagnostics are highly sensitive, they do not allow the detection of past infections, limiting their use for epidemiological surveys and disease control. To overcome this constraint, solid serodiagnostic tests are needed. Only four immunoreactive proteins have been previously described: GroEL (65 kDa heat-shock protein), LysM (lysin domain-containing protein), succinyl-CoA synthetase (subunit SCS-α and SCS-β), and Pap31 (phage-associated protein); Pap31 has been considered as a potentially suitable IgG seromarker indicating past exposure.
Although efforts have been made to apply Pap31 and other candidate proteins (GroEL, SCS-α, and SCS-β) to an ELISA assay format, no approved diagnostic test is available yet.
Additionally, vaccines are not established for this disease, although they are urgently needed.
However, before vaccine development, immunodominant proteins need to be identified.
Evidence before this study
We searched PubMed for articles published in any language with the terms (“Bartonella bacilliformis” OR “Oroya fever”) AND (“ELISA” OR “immunodominant proteins”) from database inception to June 28, 2021. We found 95 articles, of which ten pertained to the evaluation of human serum samples (nine articles) or serum from experimentally infected animals (one article). Four proteins were analysed in more detail for seroreactivity: GroEL (two studies), succinyl-CoA-synthetase (subunits α and β, one study), phage-associated protein (five studies), and lysin motif-containing protein (one study with one patient serum sample). In four studies, characterised serum samples (either from patients who were symptomatic or experimentally infected animals) were included. The use of systematical approaches for analysis of immunodominant targets has not been applied so far, but it is indispensable for defining diagnostic and vaccine proteins.
Added value of this study
To our best knowledge, this study is one of the first that systematically analysed immunoreactive proteins of B bacilliformis. By combining in-silico and in-vitro approaches, we identified 21 putative antigen candidates. Three of these antigens were successfully applied to serodiagnostic tools (line blots and ELISA), and these were evaluated by using serum samples from patients with Carrión’s disease from one of the best characterised serum libraries.
Implications of all the available evidence
We developed serodiagnostic tools for B bacilliformis infections on the basis of genomic analyses, reverse vaccinology, and recombinant protein expression by using qualified patient serum samples. These assays can detect B bacilliformis antibodies and could be used to gain better understanding of the seroreactivity of particular antigens and of the epidemiology of Carrión’s disease in South America. Furthermore, the detailed analysis of the pattern of immunoreactive antigens through line blots represents a first basis for future development of a vaccine.
,
In this study, we aimed to systematically identify immunodominant proteins of B bacilliformis to establish a reliable serodiagnostic tool by combining the analysis of genomic B bacilliformis expression libraries with reverse vaccinology.
Results
In this study, we used serum samples obtained between Dec 23, 1990, and May 5, 2018, from 26 Peruvian patients with B bacilliformis infections and serum samples taken between Aug 28 and Aug 31, 2020, from 96 healthy German blood donors to identify potentially immunodominant proteins of B bacilliforimis.
TableImmunodominant protein candidates of Bartonella bacilliformis
Accession number, designation, functional class, predicted cellular localisation, adhesion probability, number of transmembrane helices, and protein length are given for each protein. EC=extracellular. OM=outer membrane.
For the generation of recombinant B bacilliformis expression libraries, genomic DNA fragments were expressed in three different reading frames (E coli DH5α, average insert size of 1900 bp), resulting in approximately 24 700 recombinant clones (pET28a, about 9100 clones; pET28b, about 5400 clones; and pET28c, about 10 200 clones), covering more than 99% of the B bacilliformis KC583 genome. We used a pooled plasmid preparation for each reading frame to create three expression libraries in E coli BL21 (DE3). Analysis of protein production by anti-T7 tag immunoblotting revealed an expression rate higher than 96% (321 of 332 colonies, according to the analysis of one colony blot agar plate). GroEL and flagellin were characterised as immunoreactive by screening approximately 2000 colonies with rabbit-anti B bacilliformis serum, confirming the functional capability of the libraries.

Figure 1Immunoblot analysis of target proteins
All target proteins were produced in Escherichia coli BL21 (DE3) and subjected to immunoblotting (left lanes show induced protein expression, right lanes show uninduced protein expression). Heterologous protein production was confirmed by T7 tag immunoblotting (anti-T7, technical control). For analysis of immunoreactivity, serum of a Bartonella bacilliformis-immunised rabbit and a pool of five IFA-positive and immunoblot-positive serum samples of Peruvian patients were used. Reactivity is given individually per target protein (bottom rows, indicated as a plus sign [reactive] and a negative sign [non-reactive]). IFA=immunofluorescence assay.

Figure 2Line blot analysis of serum samples from Peruvian patients and German controls using B bacilliformis seroreactive proteins
(A) Design of the line blot: polyhistidine-tag purified antigens were printed on nitrocellulose membranes. (B) Line blot and densitometric analysis: line blots (left) were incubated either with human patient serum (IFA titre ≥1:320, six samples) or control serum (six samples) of German healthy blood donors; densitometric analysis of reactivity is depicted by heat maps (right); numbers refer to the position of each protein on the line blot; the most reactive proteins are labelled. (C) Analysis of antigen reactivity in line blots: results for the most reactive antigens (BbadA, BbadB, Pap31, hypothetical protein B, porin B, and autotransporter E) are presented for serum samples with high anti-Bartonella bacilliformis IFA titres from patients with B bacilliformis infection (n=12; titres ≥1:320), samples with IFA titres lower than 1:320 (n=14), and control samples (n=96); values higher than a band intensity of 100 units are considered positive; respective p values between sample categories are given. IFA=immunofluorescence assay.

Figure 3Analysis of serum samples from Peruvian patients and German controls using Bartonella bacilliformis ELISA
For the ELISA analysis (A, B), we used serum samples with high anti-B bacilliformis IFA titres from patients with B bacilliformis infection (n=12; ≥1:320), samples with an IFA titre lower than 1:320 (n=14), and control samples (n=96). OD450 values higher than 0·29 (A) or 0·34 (B) were considered positive. Respective p values between sample categories are given. We drew ROC curves using all serum samples from Peruvian patients (n=26; C, D) and using samples of Peruvian patients with an IFA titre of 1:320 or higher (n=12; E, F). Values are depicted in a ROC diagram according to sensitivity (true-positive rate) and 1 − specificity (false-positive rate). IFA=immunofluorescence assay. OD450=optical density of 450 nm. ROC=receiver operating characteristic.
Discussion
By using a combination of an in-silico reverse vaccinology approach and large-scale immunoscreening of expression libraries, we identified 22 potentially immunodominant B bacilliformis proteins, of which 21 were recombinantly expressed. Seven proteins were not immunoreactive with serum samples of Peruvian patients. The remaining 14 seroreactive antigens were analysed with line blots. Autotransporter E, porin B, and hypothetical protein B were the best candidates for a diagnostic ELISA with a calculated sensitivity of 77% and specificity of 94% (combination of all three proteins) and sensitivity of 81% and specificity of 95% (combination of porin B and autotransporter E) for all Peruvian patient samples. These data suggest that the combination of porin B and autotransporter E is sufficient to detect human anti-B bacilliformis IgG-antibodies with ELISA.
,
For unclear reasons, B bacilliformis FlgE and flagellin were not recognised by patient serum samples and FlgH showed only weak reactivity; therefore, they are not suitable serodiagnostic markers. Autotransporter proteins are the largest family of secreted proteins in Gram-negative bacteria and are known as key immunodominant vaccine components (eg, pertactin of Bordetella pertussis).
,
In our study, autotransporter E showed the strongest serum reactivity among all tested proteins, suggesting its potential use in serodiagnostics and as a vaccine candidate. Trimeric autotransporter adhesins (TAAs) are often immunodominant and considered as promising vaccine candidates (eg, Acinetobacter trimeric autotransporter [Ata] and Haemophilus influenzae adhesin [Hia]).
TAAs of B bacilliformis were already considered as potential vaccine targets due to their homology to Neisseria meningitidis NadA, which is included in the Bexsero meningococcal group B vaccine (GlaxoSmithKline, Brentford, UK).
,
However, the reactivity of BbadA and BbadB with approximately 10% of the control serum samples excludes their potential use as specific immunoreactive proteins. Porins are not used in human vaccines, but their potential use was shown by porins from Salmonella enterica serotype Typhi that induced a strong antibody response in mice.
The Vaxign software predicted several porins (porin A, B, and C and Pap31) as immunodominant targets. Moreover, porin A, porin B, and Pap31 were recognised by serum samples of Peruvian patients and a detailed analysis underlined that porin B might represent a valuable serological marker and potential vaccine target. From the remaining outer membrane proteins (TonB-dependent receptor, LysM, DUF1561, and hypothetical protein B), only hypothetical protein B showed a significant difference in reactivity to control samples, and was thus included in our ELISA.
,
,
The implementation of serodiagnostic tools allowing broad epidemiological monitoring of Carrión’s disease and the availability of a vaccine would be highly desirable; however, only a few immunogenic targets have been analysed.
,
Novel sequencing techniques and bioinformatics have revolutionised clinical microbiology and vaccine development.
Recent approaches combine reverse vaccinology with comparative multigenome analyses to identify species-conserved antigens.
Accordingly, we included five genomic sequences of B bacilliformis in our strategy, revealing several genomic discrepancies (appendix 3 pp 9–10). In particular, the loss of the TAA BbadB in strain KC584 and the diversity of Pap31 restrict their use as diagnostic or vaccine targets.
To minimise risks of missing antigens, we combined the in-silico Vaxign prediction with an immunoscreening of B bacilliformis expression libraries,
resulting in the finding of immunodominant autotransporter E. Screening of the expression libraries with rabbit serum retrieved two further immunodominant antigens (GroEL and flagellin). The reasons for these discrepancies in antigen prediction might be caused by the search algorithms of the Vaxign software and by the different exposure routes to the pathogen (natural infection in humans, and immunisation with inactivated bacteria in rabbits).
,
Despite studies evaluating its serodiagnostic use, GroEL was not predicted by the Vaxign software nor was it detected with B bacilliformis genomic expression libraries.
,
LysM was shown to react with serum samples of patients diagnosed with bartonellosis.
This protein was also identified as a potential target protein by the Vaxign software and exhibited a good reactivity with patient serum samples (figure 1). However, LysM was also highly reactive with healthy donor serum samples (92%). Because of its reactivity with IgM and IgG antibodies, Pap31 was suggested as a potential seromarker for both acute and past B bacilliformis infections.
,
Surprisingly, our data did not confirm these previous findings when using linear epitopes in line blots because only two of 26 patient samples were reactive (figure 2), suggesting that Pap31 might be more reactive as a conformational epitope. An in-silico study published in 2020 described a high genetic variability of the pap31 gene in B bacilliformis strains, which was supported by our findings (appendix 3 pp 9–10).
This variability, potentially involved in evading the host’s immune response, restricts the use of Pap31 for further serodiagnostic and vaccination strategies. Finally, SCS-α and SCS-β subunits of the succinyl-CoA synthetase were previously identified as immunodominant proteins, but showed only moderate suitability in an ELISA-based approach.
Further work will focus on the production of readily usable ELISA kits through a professional manufacturing platform, so that they can be used for seroepidemiological analyses in Peru and other South American countries. The results from such surveys will offer a much better epidemiological understanding of B bacilliformis infections and will lead to a solid knowledge base of which proteins might be included in a vaccine against B bacilliformis able to confer long-term immunity against Oroya fever.
VAJK conceived the study. AAD, AW, WB, ES, AL, PT, and VAJK did experimental laboratory work, data collection and analysis, data interpretation, and writing. TGS and PT did genome sequence analyses. PV was involved in the collection of patient serum samples. HGA and CU-G contributed with data interpretation and critical review of the manuscript. All authors had full access to all the data in the study and had final responsibility for the decision to submit for publication. AAD, AW, TGS, WB, and VAJK verified the underlying data of the study. All authors approved the final manuscript.
Read more here: Source link