Immunoproteomic analysis of Borrelia miyamotoi for the identification of serodiagnostic antigens

The tick-borne spirochete, Borrelia miyamotoi, is an emerging pathogen of public health significance. Current B. miyamotoi serodiagnostic testing depends on reactivity against GlpQ which is not highly sensitive on acute phase serum samples. Additionally, anti-B. miyamotoi antibodies can cross-react with C6 antigen testing for B. burgdorferi, the causative agent of Lyme disease, underscoring the need for improved serological assays that produce accurate diagnostic results. We performed an immunoproteomics analysis of B. miyamotoi proteins to identify novel serodiagnostic antigens. Sera from mice infected with B. miyamotoi by subcutaneous inoculation or tick bite were collected for immunoblotting against B. miyamotoi membrane-associated proteins separated by 2-dimensional electrophoresis (2DE). In total, 88 proteins in 40 2DE immunoreactive spots were identified via mass spectrometry. Multiple variable large proteins (Vlps) and a putative lipoprotein were among those identified and analyzed. Reactivity of anti-B. miyamotoi sera against recombinant Vlps and the putative lipoprotein confirmed their immunogenicity. Mouse anti-B. burgdorferi serum was cross-reactive to all recombinant Vlps, but not against the putative lipoprotein by IgG. Furthermore, antibodies against the recombinant putative lipoprotein were present in serum from a B. miyamotoi-infected human patient, but not a Lyme disease patient. Results presented here provide a comprehensive profile of B. miyamotoi antigens that induce the host immune response and identify a putative lipoprotein as a potentially specific antigen for B. miyamotoi serodetection.


Results
B. miyamotoi-infected murine hosts elicit an immune response against multiple proteins regardless of inoculation route. B. miyamotoi membrane-associated proteins were purified, fractionated by 2DE (Fig. 1), and subjected to immunoblotting with antiserum from mice infected by either needle inoculation or tick bite and sacrificed at 8 or 40 days. Proteins corresponding to the immunoreactive spots from the blots were identified by mass spectrometry. Protein identifications for each spot noted below are listed in Table 1 and corresponding raw data are located in Supplemental Table 1.
The largest number of immunoreactive spots was observed with sera collected at 40 dpi (IgG). Spots 1-11 and 14-17, originally detected at 8 dpi, were again observed at this time point including nine Vlps and Vsp1 (Fig. 3). Twenty-three spots were uniquely observed with sera from 40 dpi. Eight of the 23 unique spots (21, 22, 24, 25, 26, 27, 33, and 40) were detected across immunoblots probed with both needle and tick bite inoculated mouse serum of which protein identities were assigned the following gene ontology associations: membrane (multiple Vlps and Vsp1); flagellum associated; phosphoric diester hydrolase activity (GlpQ); ATPase and ATP-binding;  www.nature.com/scientificreports www.nature.com/scientificreports/ translation associated; metabolic processes; transcription associated; polyamine binding; and unassigned (including the previously identified putative lipoprotein) (Fig. 3a,b; Table 1). The remaining 15 immunoreactive spots recognized only in serum from tick bite inoculated mice represented 47 individual proteins, 18 of which had been identified within other spots across previous data points (i.e. 8 dpi needle or tick inoculated and 40 dpi needle inoculated). Twenty-nine proteins were uniquely identified in serum collected 40 days post-tick inoculation and assigned the following gene ontology associations: flagellum associated; ATPase and ATP-binding; catalytic activity; translation associated; proteolysis; transcription associated; protein folding; chemotaxis; and unassigned ( Fig. 3b; Table 1).
The immunoreactivity of selected recombinant proteins was assessed by immunoblotting using mouse anti-B. miyamotoi sera collected at 8 and 40 dpi (Fig. 5). The IgM response in serum collected from the needle inoculated mouse at 8 dpi resulted in strong reactivity against recombinant VlpC2 (r-VlpC2) and was comparatively weak for r-putative lipoprotein, r-GlpQ, r-VlpD9, r-VlpD1, r-VlpD2, r-VlpD10, and r-VlpD8 (Fig. 5a). In contrast, all six r-Vlps were highly IgM immunoreactive with tick bite inoculated mouse antiserum (Fig. 5b). The r-uncharacterized protein and r-Vsp1 displayed no IgM reactivity with antiserum at 8 dpi regardless of inoculation route (Fig. 5a,b). Reactivity against whole cell lysates (WCL) of B. miyamotoi (LB-2001) and B. burgdorferi (B31) was observed at 8 dpi across both needle and tick inoculated sera (Fig. 5a,b).
Antibodies present at 40 dpi (IgG) recognized the entire panel of r-Vlps, independent of infection route. Reactivity against r-GlpQ was present in antiserum from needle inoculated mice, while background antibody recognition was observed in tick inoculation-derived serum (Fig. 5c,d). The r-putative lipoprotein was recognized by antiserum from both needle-and tick-inoculation. Neither r-Vsp1 nor the r-uncharacterized protein were recognized by serum at the 40 dpi time point (Fig. 5c,d). Additionally, the B. miyamotoi WCL was highly reactive against antibodies in both groups of serum samples. Recognition was observed, to a lesser extent, in the B. burgdorferi WCL (Fig. 5c,d).
Evaluation of cross-reactive anti-B. burgdorferi serum antibodies against recombinant B. miyamotoi antigens. Anti-B. burgdorferi sera from mice infected by tick bite (14 dpi) demonstrated strong cross-reactivity (IgM) with regards to all six Vlps B. miyamotoi recombinants and, to a lesser extent, r-putative lipoprotein, GlpQ, and Vsp1. No cross-reactivity was observed for the r-uncharacterized protein (Fig. 6a). IgG cross-reactivity was observed against all six recombinant Vlps with weak reactivity against Vsp1 (Fig. 6b). Both r-putative lipoprotein and r-GlpQ were negative for IgG cross-reactivity against antibodies present in the B. burgdorferi-infected murine sera (Fig. 6b). Murine sera antibodies recognized multiple proteins in the

Reactivity of human BMD and Lyme disease patient serum against recombinant B. miyamotoi antigens.
A BMD patient serum sample blotted against recombinant antigens demonstrated IgG reactivity against all r-Vlps (VlpD9, VlpC2, VlpD1, VlpD2, VlpD10, and VlpD8), the r-putative lipoprotein, and r-GlpQ, but no reactivity against the r-uncharacterized protein or r-Vsp1 (Fig. 7a). Multi-protein reactivity was observed in the B. miyamotoi WCL but was comparatively weaker in the B. burgdorferi WCL. IgG serum antibodies from a LD patient demonstrated cross-reactivity against all r-Vlps but was absent for r-uncharacterized protein, r-GlpQ, and r-Vsp1. Significantly, there was also no cross-reactivity against the r-putative lipoprotein. Both B. burgdorferi and B. miyamotoi WCLs demonstrated reactive proteins (Fig. 7b).

Discussion
Accurate diagnosis of BMD continues to warrant greater attention as the impact on human disease burden increases. Since its emergence as a human pathogen in 2011, cases have been described across multiple continents where ticks of the Ixodes genera are present, however, approaches to generate diagnostics to distinguish B. miyamotoi infections from other borrelioses are only beginning to be investigated 5,7,13,[19][20][21][22] . The present work took an immunoproteomic approach to compare host acute and convalescent antibody responses against B. miyamotoi infection by either needle or tick vector inoculation for the identification of novel serodiagnostic targets. Infection in murine hosts produced differential antigen recognition patterns dependent upon infection duration (8 and 40 dpi) and inoculation route (needle vs. tick bite). IgM antibodies (8 dpi) were reactive against a number of proteins previously identified during B. miyamotoi infection including: GlpQ, Vsp1, and 9 Vlps (VlpC2, D2, D1, D8, D9, D10, 5, 15/16, and one undesignated Vlp) 11,14,23 . Antibody responses to multiple Vlps suggests common epitope recognition, tandem expression, or borrelial serotype switching during acute infection. Additionally, this could also be indicative of the clonality, or lack thereof, concerning strains used for infection. Immunogenic proteins spots (number 1-17) were excised from a corresponding silver stain 2DE and identified by mass spectrometry. All proteins were consistently numbered across immunoblot and time point and resulting identities from individual spots listed in Table 1 and Supplemental Table 1. Molecular weight of proteins are expressed in kDa. Dashed lines demarcate where the gel has been cropped to exclude molecular weight markers. Uncropped blots can be viewed in Supplemental Fig. 2.
Borrelia miyamotoi-infected I. scapularis used in this study originated from field collection 24 . Consequently, murine infection may represent a non-clonal strain of B. miyamotoi, contributing to the expression of multiple Vlps, which may have important implications on observed antibody response.
GlpQ, Vsp1, and multiple Vlps were again recognized at 40 dpi, suggesting their utility as serodiagnostic targets with both IgM and IgG antibody responses. A greater number of antigens were recognized by the IgG response at 40 dpi relative to the IgM response at 8 dpi indicating the maturity of the humoral response with time. Redundancy in protein identification from these spots was prevalent; however, unique identifications included proteins grouped with cellular components, molecular functions, biological processes, and several hypothetical proteins with unclassified function (listed in Table 1).
Immunoreactive spots were more abundant in 2DE immunoblots probed with antisera from tick bite inoculated mice compared to needle inoculated mice indicative of differential protein expression profile. Alternatively, immunoreactivity against needle inoculated B. miyamotoi LB-2001 could reflect multiple passages through SCID mice prior to stable in vitro cultivation [25][26][27] . Although there are no direct data regarding B. miyamotoi plasmid loss or gene rearrangement upon serial passage, it has been posited as an underlying reason for gene loss in at least one study 28 . Plasmid content representative of the strains utilized in this study was not compared but may also play a role in the observed results. However, it has been firmly established that environmental factors such as temperature and pH, which differ in the tick versus in vitro cultivation, impact B. burgdorferi gene expression profiles providing rationale for investigations into gene regulation during tick transmission [29][30][31][32][33] . For example, the spots present in the immunoblots probed with antisera from tick bite inoculated mice but absent in needle inoculated mice may offer clues to determine borrelial gene products essential for survival, maintenance, and dissemination in ticks.
Assessment of various Vmps in conjunction with GlpQ to improve serodiagnosis of BMD suggests that multi-antigenic detection yields higher sensitivity and specificity 14  . For both sets of immunoblots an acidic (pH 4-7) and basic (6-11) pH range were utilized. Sera were diluted 1:200 for immunoblotting. Immunogenic protein spots (numbers 1-11 and 14-40) were excised from a corresponding silver stain 2DE and identified by mass spectrometry. All proteins were consistently numbered across immunoblot and time point and resulting identities from individual spots listed in Table 1  www.nature.com/scientificreports www.nature.com/scientificreports/ human BMD antiserum. However, the Vlps were similarly recognized with antisera from B. burgdorferi-infected mice and the human LD patient suggesting that serological utility of these antigens may be compromised as a reliable diagnostic tool. The observed cross-reactivity is possibly due to orthology to B. burgdorferi proteins VlsE and OspC, and is especially notable as the 6 th invariant region (IVR6) of VlsE comprises the C6 antigen that is used extensively in diagnostic testing for LD 16,17 . It is noteworthy that both the BMD and LD patient serum used in this study were C6-positive. Indeed, recent reports have indicated that BMD patient serum samples react positively in C6 antigen testing 12,34 . Thus, previous or active infections with BMD or LD cannot be excluded for patient sera used in this or future studies where observable reactivity against Vlps or alternative antigens is noted.
Recombinant Vsp1 and the uncharacterized protein were not reactive against the mouse antisera used in the 2DE immunoblot discovery and, therefore, were not confirmed as immunogens in our mouse model. This result may be due to the presence of multiple overlapping non-immunogenic proteins identified within a single immunoreactive spot. The uncharacterized protein was most closely homologous to the B. hermsii Alp protein 18 . Like the B. miyamotoi uncharacterized protein, Alp did not elicit an antibody response in infected mice and was postulated to be associated with the tick environment. Curiously, the lack of recognition of our recombinant Vsp1 with mouse antiserum conflicted with the findings of Wagemakers et al. who described Vsp1 as an immunodominant antigen from their B. miyamotoi strain LB-2001 needle inoculated mice 11 . However, those investigators injected their mice with a dose of organisms 3 logs higher than that used in our study (10 7 vs 10 4 ) perhaps accounting for the observed differences. However, these researchers did observe Vmp switching from dominant expression of Vsp1 to VlpC2 in acute infection 11 . Our infection model resulted in dominant, acute expression of VlpC2, possibly due to Vmp switching brought about by differences in cultivation conditions or infection kinetics between studies. Similarly, we did not observe an anti-Vsp1 response from tick bite inoculated mice whereas the Wagemakers et al. study found that only 2 of 9 human BMD (presumably from tick bite infection) patient serum samples were positive for Vsp1 11 . Our preliminary results demonstrated no Vsp1 reactivity in BMD or LD human sera. Though recent studies have shown its utility as a specific diagnostic marker, there could exist disparity among B. miyamotoi strains originating in North American versus Europe 13 . These results illustrate the complexity of differential Vmp gene expression and the corresponding role of host seroconversion dependent on inoculation route which awaits further study. Interestingly, reactivity against r-GlpQ was seen by IgG in serum from needle inoculated mice at 40 dpi, and was weakly recognized in serum from mice infected by tick inoculation. This result suggests differential upregulation between inoculation routes and/or differences in antibody responses between reservoir vs. incidental (human) hosts that warrants further investigation.
The putative lipoprotein reacted strongly against the anti-B. miyamotoi antibodies raised in both the needle and tick inoculated mice, emerging as a novel candidate serodiagnostic antigen. The putative lipoprotein has a calculated molecular mass of 35 kDa and is encoded by a gene localized to the lpD plasmid of B. miyamotoi strain LB-2001 26 . The homologous protein in B. miyamotoi CT13-2396 has 80% amino acid identity 35    in our singular LD human serum. Importantly, antibody response directed against the putative lipoprotein was dominant in the BMD convalescent human serum sample, thereby, providing preliminary evidence for serological distinction between BMD and LD. Antigenic cross-reactivity with the putative lipoprotein against antibodies directed to other RFB remains to be determined. A limitation of current work includes the number of human serum samples, especially those which are of similar infection status across disease groups (i.e. LD vs. BMD) and will be the subject of future studies. Additional human serum samples from BMD, LD, and soft tick-borne RFB patients will be necessary in the appraisal of this putative lipoprotein as a serodiagnostic target.
In conclusion, we used an immunoproteomic approach to identify B. miyamotoi antigens that induce the host antibody response post-infection. Our results indicated that although Vmps are immunodominant, agreeing with other reports, antigenic cross-reactivity with anti-B. burgdorferi antibodies may be problematic. The putative lipoprotein has emerged as a potential serodiagnostic candidate to augment BMD testing with GlpQ and Vmps. Multiple proteins revealed as immunogenic in this study may aid in the development of new serodiagnostic tests for BMD.

Methods and Materials
Borrelia miyamotoi culture and mouse infections. Low passage (≤4) B. miyamotoi strain LB-2001 36 , (kindly supplied by Joppe Hovius, Center for Experimental and Molecular Medicine, Amsterdam, The Netherlands) was cultivated at 34 °C with 5% CO 2 in modified Kelly-Pettenkofer medium (MKP-F) as previously described 27,37 . Spirochetes were visualized using dark field microscopy and enumerated with a Cellometer (Nexcelom, Lawrence, MA). Outbred, female CD-1 mice (n = 5) (Charles River, Wilmington, MA) 6-8 weeks of age were needle inoculated subcutaneously with a 100 µl suspension of B. miyamotoi (1 × 10 4 ) in MKP-F. A separate cohort of CD-1 mice (n = 3-5) were infested with B. miyamotoi-infected I. scapularis immatures (i.e. larvae or nymphs) that originated from naturally infected female ticks collected in Minnesota that passed infection to their offspring. The tick feeding protocol was previously described 38 followed by serum collection at 8 or 40 dpi. Serum from two CD-1 mice infested with nymphal B. burgdorferi (strain B31)-infected I. scapularis was collected 14 dpi and pooled. All murine infections were performed once. Animal experiments were approved and   . Identities were accepted with a >90% peptide and >99% protein probability threshold and at least two unique peptides per protein identity (Supplemental Table 1). Mass spectrometry was performed for all spots at least once, with some identified twice.
Production of recombinant proteins and immunoblotting. Genes were amplified by PCR from B.
miyamotoi genomic DNA (gDNA) using primers listed in Table 3, cloned into pETite N-His vector (Lucigen, Middleton, WI), and transformed into E. coli BL21 (DE3) according to the manufacturer's instructions. Recombinant proteins were expressed and purified using a QiaExpress Ni-NTA FastStart kit (Qiagen, Valencia, CA) as described previously 39 . Proteins (100 ng/individual protein) were transferred to PVDF membranes for immunoblotting with mouse or human serum samples (1:200), followed by incubation with alkaline phosphatase conjugated goat anti-mouse or goat anti-human IgM or IgG (1:5000) with development by NBT/BCIP with   Table 3. Primers used for gene amplification for pETite expression vector cloning. Vsp = variable small protein; Vlp = variable large protein; underlined nucleotides denote sequences used for expression vector insertion.