Vitronectin binding protein, BOM1093, confers serum resistance on Borrelia miyamotoi

Borrelia miyamotoi, a member of the tick-borne relapsing fever spirochetes, shows a serum-resistant phenotype in vitro. This ability of B. miyamotoi may contribute to bacterial evasion of the host innate immune system. To investigate the molecular mechanism of serum-resistance, we constructed a membrane protein-encoding gene library of B. miyamotoi using Borrelia garinii strain HT59G, which shows a transformable and serum-susceptible phenotype. By screening the library, we found that bom1093 and bom1515 of B. miyamotoi provided a serum-resistant phenotype to the recipient B. garinii. These B. miyamotoi genes are predicted to encode P35-like antigen genes and are conserved among relapsing fever borreliae. Functional analysis revealed that BOM1093 bound to serum vitronectin and that the C-terminal region of BOM1093 was involved in the vitronectin-binding property. Importantly, the B. garinii transformant was not serum-resistant when the C terminus-truncated BOM1093 was expressed. We also observed that the depletion of vitronectin from human serum enhances the bactericidal activity of BOM1093 expressing B. garinii, and the survival rate of BOM1093 expressing B. garinii in vitronectin-depleted serum is enhanced by the addition of purified vitronectin. Our data suggests that B. miyamotoi utilize BOM1093-mediated binding to vitronectin as a mechanism of serum resistance.


Results
Identification of serum-sensitive B. garinii HT59G which shows a transformable phenotype. We first sought to evaluate the susceptibility of Borrelia strains to human serum in detail using strains isolated from different biological and geographical samples. For this purpose, 17 Borrelia strains of B. garinii and B. bavariensis were examined for serum-sensitivity by determining the survival rate following treatment with 40% Normal human serum (NHS) or Heat-inactivated human serum (HIS) for 16 h (Figure 1). Of these 17 strains, nine strains (B. bavariensis strains J-14, J-16, J-20t, J-32, J-39, J-40, J-41 and B. garinii strains J-21, J-37) obtained from the skin of Lyme disease patients, two B. garinii strains (strains VSBM and VSBP) isolated from cerebrospinal fluid (CSF) of patients, and one B. garinii (strain NT25) isolated from a tick exhibited a serumresistant phenotype. One B. garinii strain (strain VSDA) isolated from patient CSF and four strains of B. garinii (strains Fis01, Far01, Far02, and HT59) isolated from ticks were serum-sensitive. These serum-sensitive strains were selected as candidate hosts for gene library construction of B. miyamotoi. To investigate the transformability of these B. garinii strains, the shuttle vector pBSV2 was electroporated into each serum-susceptible B. garinii strain. Among the five strains tested, transformants were obtained only from B. garinii strain HT59. We therefore picked 10 single colonies of strain HT59 and established 10 clones. Of these 10 clones, clone G also showed a transformable phenotype. When B. garinii strain HT59G was transformed with plasmid pBSV2, an average of 15 transformants was obtained per 1 µg of plasmid DNA ( Table 1).

Construction of plasmid archives for B. garinii HT59G transformation.
At the time of this study, the genome of B. miyamotoi strain MYK3 was not available. Therefore, candidate genes encoding membrane proteins were selected from the genome sequence of B. miyamotoi strain FR64b, which is published in GenBank (Acc. Nos. CP004218-CP004266). From this database, 649 open reading frames (ORFs) that were predicted to be non-chromosomal encoding were extracted. Of these 649 ORFs, 90 ORFs were predicted to be displayed on the bacterial surface of B. miyamotoi using SignalP or LipoP analysis. For each of these 90 ORFs, specific PCR primers were used for DNA amplification. All ORF PCR products were detected from template genomic DNA of B. miyamotoi strain MYK3. The shuttle vector was created for the 90 ORFs by combining linearized pBSV2 and each PCR fragment using the In-Fusion procedure according to the manufacturer's instructions (see "Materials and methods"****). Of these 90 ORFs, 84 ORFs were isolated from E. coli DH5α. Of these 84 ORFs, transformation of B. garinii HT59G was successful with 76 ORFs (coverage ratio: 84.4%).   Fig. 3, both P83/100 and OspA were digested by proteinase K treatment in a dose-dependent manner. BOM1093 was also accessible to proteinase K. In contrast, flagellin protein, which reacted with monoclonal antibody H9724, was resistant to proteinase K treatment.
Human serum Vn co-precipitated with His-tagged BOM1093. Serum resistance phenotype of Borrelia is due to its ability to bind complement regulator (for e.g., FH). Therefore, we examined the binding of B. garinii HT59G/pBOM1093 to complement regulators FH, factor H-like protein 1 (FHL-1), factor I, properdin, carboxy peptidase N, C4b binding protein (C4BP), C1 inhibitor, complement factor H-related protein 1, clusterin and vitronectin (Vn). B. garinii HT59G/pBOM1093 (6xHis-tagged on C terminus) incubated in 20% NHS was tested by a pull-down assay using Ni-NTA magnetic beads. The only complement regulator that bound to BOM1093 was Vn, which is reported to inhibit the terminal pathway of the complement system (e.g. C9 polymerization). In contrast, Vn was not detected when B. garinii HT59G/pBSV2 was used as a prey antigen (Fig. 4). Based on this data, we conducted further analyses for Vn and BOM1093.

B. garinii HT59G/pBOM1093 binds purified-recombinant Vn in a dose-dependent manner.
To confirm the results of the pull-down assay, we performed an ELISA assay to examine the binding of purified recombinant Vn to B. garinii HT59G/pBOM1093. A Vn-binding ELISA assay found dose-dependent and saturation binding of recombinant Vn for B. garinii HT59G/pBOM1093. In contrast, Vn binding was not detected when B. garinii HT59G/pBSV2 was tested. The results are shown in Fig. 5. The K D value for the interaction was estimated to be 10.8 nM (95% Confidential interval range 8.5-13.0 nM).

C-terminal region of BOM1093 is required for Vn-binding.
To define the Vn-binding facilities of BOM1093, C-terminal-truncation mutants were subjected to a pull-down assay. Although intact 6xHis tagged-BOM1093 co-precipitated with Vn, none of the truncated proteins (6xHis tagged-BOM1093 1-208 , 6xHis tagged-BOM1093 1-158 , 6xHis tagged-BOM1093 1-108 , and 6xHis tagged-BOM1093 1-58 ) co-precipitated with Vn ( Fig. 7A,B). Moreover, an ELISA assay revealed that B. garinii HT59G/pBOM1093 1-208 , in which the C-terminal    www.nature.com/scientificreports/ residues 209-308 were deleted, showed reduced binding compared to B. garinii HT59G/pBOM1093 (Fig. 7B). Furthermore, Vn-binding was not detected in the other mutants due to deletion of the C terminus of BOM1093. These results suggest that the 209-308 region of the C-terminal amino acids of the BOM1093 protein is essential in enabling B. garinii HT59G/pBOM1093 to bind to Vn in vitro.

Depletion of Vn enhanced bactericidal activity of serum-resistant B. garinii HT59G/ pBOM1093.
It is well characterized that NHS-derived Vn inhibits the complement system. Therefore, removal of Vn from NHS enhances bactericidal activity. In this study, we prepared human serum depleted of Vn (HSΔVn) and used it in the serum susceptibility assay of B. garinii transformants. In this study, Vn depletion was confirmed by western blotting with clusterin (CLU), as a positive control (Fig. 8A). The Vn-depleted serum was subjected to bactericidal assays using B. garinii HT59G/pBOM1093 and the mock control (B. garinii HT59G/ pBSV2). B. garinii HT59G/pBOM1093 showed a significant increase in serum susceptibility when incubated with HSΔVn (Fig. 8B). However, when 1 mM of purified recombinant Vn was added to the HSΔVn, serumresistance was observed for B. garinii HT59G/pBOM1093.

Discussion
In previous studies, several transformable Borrelia strains have been used as surrogate strains for serum susceptibility analysis. B. burgdorferi strain B313, which is a derivative of strain B31, is one of the transformable and serum susceptible strains 20 . The strain is convenient for genetic analysis because it forms colonies on a semi-solid BSK agar plate. B. garinii strains G1 and 50.97 have also been used for genetic analyses of serum susceptibility because these strains are transformable and susceptible to human serum 21,22 . However, transformants of B. garinii strains were isolated using a limiting dilution technique in liquid BSK medium. This process for isolating clonal transformants is an intricate procedure that requires several weeks. Furthermore, more than 10 µg of plasmid DNA is required for efficient transformation for these strains. In this study, we established another transformable and serum-susceptible B. garinii HT59G. Transformation efficacy of the strain was 15 transformants/µg plasmid DNA when pBSV2 plasmid was used, and transformants were isolated from semi solid agar plates. These results suggest that the B. garinii strain HT59G used in our study may be a useful and convenient tool to investigate serum resistance mechanisms of borreliae. Bacterial pathogens display proteins on their surface, some of which bind complement regulators and inhibit the host complement system. Several reports have indicated that borreliae have the ability to evade the innate immune system. Lyme disease borreliae produce the protein p43 which binds C4BP to the bacterial surface, thereby regulating the classical and lectin pathways through degradation of C4b 23 . Most Lyme disease borreliae produce several complement regulator-acquiring surface proteins (CRASPs): CRASP-1 to CRASP-5 (CspA, CspZ, and Erps). CspA and CspZ are ligands for FH and/or FHL-1. These proteins can also bind C3b and then promote C3b degradation on the bacterial surface to regulate the complement pathway 24 . Erps (ErpP, ErpC, and ErpA) can also bind to FH and contribute to serum resistance of Lyme disease borreliae; however, the biological significance of these interactions remains unclear. RF borreliae also express complement regulator-binding proteins on their surface 25 29,30 . A CD59-like protein of B. burgdorferi that binds to C9 has also been suspected; however, the borrelial factor was not identified 31 . These interactions contribute to the inhibition of the membrane attack-complex, thereby preventing bacteriolysis. Complement regulation by Vn binding, however, has not been reported in Borrelia.
Vn was discovered in 1967 32 and initially called S-protein, but was later renamed by Hayman et al. 33 . Vn binds to the membrane attachment site of the C5b-7 complex, thereby blocking insertion into the target membrane and inhibiting C9 polymerization and attachment 18,34 . Bacterial pathogens thereby evade bactericidal action by the complement system through binding host Vn to their surfaces 35 . Vn-dependent serum resistance has been well studied for several pathogens. The serum resistance mechanism by Vn binding has been reported in LcpA of Leptospira species, Lpd of Pseudomonas aeruginosa, PE of non-typable Haemophilus influenzae, and UspA2 of Moraxella catarrhalis [36][37][38][39] . In this study, we revealed that B. miyamotoi BOM1093 acts as a virulence factor contributing to human serum resistance by binding to serum Vn.
In previous studies, B. miyamotoi has shown a serum-resistant phenotype 16,40 , and only CbiA (locus tag BOM1283, Acc. No. AHH05826) has been identified as a factor responsible for this phenotype 17 . Briefly, it was shown that CbiA binds to FH and interacts to complement components (C3, C3b, C4, C4b, C5, and C9), thereby potentially blocking the alternative, classical, and terminal pathways of the complement system. However, it has not been shown whether this multi-functional protein is also capable of binding Vn. To our knowledge, this is the first study to report that Vn-binding to a borrelial factor promoted serum resistance of B. miyamotoi.
BOM1093 was identified as the Vn-binding protein in this study. Using Protein Basic Local Alignment Search Tool (BLASTP) analysis, BOM1093 and BOM1515 proteins were identified as related to antigen P35 41 in Lyme disease borreliae. In addition, the amino acid residue of BOM1093, which was suggested to be an important region for Vn binding, had a homology of 94.0% identity with the sequence of BOM1515. BOM1515 is also expected to confer serum resistance to B. miyamotoi by Vn-binding. From BLASTP analysis, we found that BOM1093 was conserved in RF borreliae including hard-tick-borne RF borreliae. Sequence similarity of BOM1093 ranged from 59 to 100% in B. miyamotoi and from 56 to 68% in RF borreliae. The bom1093 lineage possessed by RF borreliae may have a function similar to that of B. miyamotoi bom1093. In this study, we conclude that the C-terminal region of BOM1093 is involved in Vn-binding. However, it is also possible that the C-terminal region of BOM1093 is required for structural stability of the protein and that its removal results in a protein with disrupted structure and function. To resolve this question, further investigation is required.

Conclusion
In conclusion, using a newly established transformable B. garinii strain, we revealed that B. miyamotoi has the ability to bind Vn through the membrane protein BOM1093. We hypothesize that Vn-binding may contribute to pathogenicity of B. miyamotoi in humans by allowing it to evade the serum complement system. This is the first study to report that Vn-binding is associated with serum resistance of Borrelia.

Materials and methods
Bacterial strains and culture conditions. Borrelial strains used in this study are listed in Table 2 B. garinii HT59G was isolated from strain HT59 by subsurface colony formation 42 . These Borrelia strains were grown at 34 °C in Barbour-Stoenner-Kelly (BSK)-M medium 43 supplemented with 7% rabbit serum with or without selectable antibiotics. For antibiotic selection of shuttle vector transformants, kanamycin (200 µg/ml) was used. Escherichia coli DH5α was used for the preparation of plasmids for electroporation into B. garinii HT59G.

Preparation of NHS and HIS. NHS from a human blood donor without history of a borrelial infection
was used. The serum was confirmed serologically negative by the absence of IgG and IgM antibodies against Borrelia spp. A serum diagnostic test was performed for Lyme disease by immunoblotting using a commercial kit, recomLine Borrelia IgM/IgG (Mikrogen GmbH, Neuried, Germany). We obtained ethical approval for the use of human serum (The details are provided in the Medical Ethics section). HIS was prepared by incubating NHS at 56 °C for 30 min.
Screening assay for serum sensitivity. The serum sensitivity of each borrelial strain was assessed using cells harvested from mid-log phase cultures. The cells (~ 10 7 cells/ml) were incubated in 40% NHS or HIS at 37 °C. After incubation for 16 h, cell viability was assessed using dark-field microscopic counts of moving cells in 10 fields under 300× magnification. Data are presented as percent survival, calculated as follows: (average number of moving cells/numbers of morphologically collapsed cells per ten 300× magnification fields) × 100. In each assay, the B. burgdorferi strain 297 and B. garinii VSDA were used as survival control and serum-sensitive control, respectively 35,49 . Electroporation of Borrelia strains. Electroporation of serum-sensitive Borrelia strains was performed as described previously 50 Table S1.

Plasmid construction and transformation of recipient B. garinii.
The outline of plasmid construction is summarized in Fig. 9. The shuttle vector pBSV2 was used for transformation of B. garinii HT59G. Plasmid pBSV2 was originally constructed by Stewart et al. 52 . The flagellin gene (flaB) promoter (pflaB) was used for gene expression of the borrelial gene in the recipient Borrelia strain. DNA fragment of pflaB on plasmid pTM61 55 was amplified by PCR using a set of primers: pTM61_pflaB_R and pTM61_pflaB_F + tag (Supplemental Table S1). Plasmids for borrelial transformation were constructed using the In-Fusion HD Cloning System (Clontech Laboratories, Mountain View, CA, USA). Briefly, plasmid pBSV2 was linearized by digestion with restriction enzymes (Hind III and Xba I, Takara Bio, Shiga, Japan). DNA fragments of each ORF of B. miyamotoi were amplified by PCR using specific primers for each ORF. Concatenation of pflaB, each amplified fragment from the genomic DNA of B. miyamotoi strain MYK3 and linearized pBSV2 was performed using the In-Fusion system according to manufacturer's instructions. The primers used for amplification of DNA fragments from B. miyamotoi are listed in the Supplemental Table S2. The constructed plasmid was propagated using E. coli DH5α and purified with the Qiaprep Spin Miniprep kit (QIAGEN, Calif, USA). The purified plasmids were subjected to nucleotide sequencing to ensure no mutations introduced during the cloning process. The oligonucleotide primer pair (T7 and AS-T, Supplemental Table S1), which amplified the DNA fragment of multi-cloning site of pBSV2, was used for PCR amplification and sequencing. Each plasmid was used to transform B. garinii HT59G. The transformed B. garinii HT59G was picked up from the BSK-M plate containing 200 µg/ml kanamycin. Transformation with plasmid was confirmed by kanR-PCR and PCR using DNA primers T7 and AS-T.  56 . The samples corresponding to 1.6 × 10 6 whole cell equivalents were separated on a 12.5% SDS-PAGE gel, transferred to a Sequi-Blot PVDF membrane (Bio-Rad Laboratories, Hercules, CA, USA), and western blotting was performed as previously described 57 . Monoclonal antibodies (mAb) against borrelial OspA (H5332) 58 , flagellin (H9724) 59 , p100 (Mab958) (Merck, Darmstadt, Germany), or anti-BOM1093 rabbit serum (prepared in this study) were used. To detect these antibodies, Horseradish peroxidase (HRP)conjugated goat anti-mouse IgG (Jackson ImmunoResearch, West Grove, PA, USA) or rabbit IgG (Merck) were used. HRP-conjugated antibodies were detected by chemiluminescence using the electrogenerated chemiluminescence (ECL) Prime detection reagent (GE Healthcare Bioscience, Piscataway, NJ, USA).
Pull down assay. Borrelia garinii HT59G cells expressed as 6xHis-tagged to C-terminal of BOM1093 (10 8 cells/reaction) were incubated with NHS (20%) for 1 h at 37 °C and washed 3 times with TBS buffer (20 mM Tris-HCl pH 7.0, 0.25 M NaCl) containing 5 mM Pefabloc SC. Hexa-His-tagged BOM1093 obtained from sonicated cells was purified using HisPur Ni-NTA Magnetic Beads (Thermo Fisher Scientific, Waltham, MA, USA). The sample was separated by 10% SDS-PAGE, transferred to a PVDF membrane, and analyzed using either mAb or polyclonal antibody (pAb) (1:1000) to detect complement regulators for 1 h followed by the HRPconjugated monoclonal antibody (1:5000) for 1 h. The blot was developed using the ECL Prime detection reagent. For the detection of complement regulators, antibodies were used as follows; Anti-Vn, Anti-clusterin, Anti-FHL-1, Anti-factor I and Anti-complement factor H-related protein 1 mAbs were purchased from R&D Systems (Minneapolis, MN, USA), Anti-C4BP mAb was from Santa Cruz Biotechnology (Dallas, TX, USA), Anti-properdin mAb was from Abcam, Anti-carboxy peptidase N mAb was from Bioss (Woburn, MA, USA), Anti-FH pAb was from Merck, Anti-C1 inhibitor pAb was from Complement Technology (Tyler, TX, USA).   Medical ethics. The normal human serum (NHS) was obtained from healthy Japanese blood donors. The blood collection was carried out in accordance with international guideline and regulations (Declaration of Helsinki, 1964). All experimental protocols used human serum, the procedure of blood collection, and documented informed consent were approved by the Ethical Committee of the National Institute of Infectious Diseases for medical research using human subjects (Approval No. 791 on June 26, 2017). All volunteers provided informed consent.
Statistical analysis. Results were assessed using the Student's t test for paired data. A value of p ≤ 0.01*** was considered statistically significant.

Data availability
Materials established in this study are available from the corresponding author on reasonable request.