Cross-serotype borreliacidal activity associated with Bin 1 mAbs

We next wished to determine the functional capacity of the Bin 1 mAbs to promote complement-dependent borreliacidal activity across the different OspA serotypes. To validate the HB19-R1 OspA reporter strains for this purpose, we compared complement-dependent killing of an infectious B. burgdorferi B31-5A4-based reporter strain [34] to the HB19-R1 OspA_ST1 derivative. The assays were performed under conditions optimized for each strain, due to slight differences in growth kinetics and intrinsic complement sensitivities. Cultures were mixed with serial mAb dilutions (0.08-10 nM) in medium containing 10% (B31-5A4) or 2.5% (HB19-R1) guinea pig complement. The following day, IPTG was added to the cultures to induce mScarlet-I expression and MFI was measured 24 h (HB19-R1) or 48 h (B31-5A4) later, as described in the Materials and Methods. Under these conditions, the five Bin 1 mAbs had EC₅₀ values ranging from 0.63 to 1.25 nM against both the B31-5A4 and HB19-R1 OspA_ST1 reporter, while an IgG isotype control (PB10) had no effect on spirochete viability (Fig 2A and 2B). Moreover, the EC₅₀ values for 857-2 and 221–7 obtained using the HB19-R1 OspA_ST1 strain closely match those reported for B. burgdorferi B31 derived using a completely different viability assay [11]. Collectively, these results demonstrate that OspA expression is comparable in both B. burgdorferi genetic backgrounds. Thus, the HB19-R1 reporter strains served as reliable readouts of OspA-mediated complement-dependent borreliacidal activity.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. OspA Bin1 mAbs promote complement-dependent killing of recombinant B. burgdorferi strains expressing OspA ST1-7.

(A) Borreliacidal mAb titration curves and (B) heat map comparing the susceptibilities of an infectious B. burgdorferi B31-5A4 mScarlet-I viability reporter strain and an HB19-R1 strain expressing the same OspA_ST1 (B31) allele to complement-mediated killing by OspA_ST1 mAbs and an isotype IgG1 (PB10) control. Experiments were performed under conditions optimized for each strain as described within the Materials and Methods. Ec₅₀ values represent the mean minimum concentration (nM) of mAb resulting in >50% reduction in MFI relative to controls, as determined from 3-5 independent experiments per strain. Statistical differences in mAb susceptibility were assessed via two-way ANOVA followed by Tukey’s multiple comparison test. No significant differences were observed. (C) Heat map depicting the mean Ec₅₀ (nM) of Bin 1 mAbs against recombinant B. burgdorferi strains expressing OspA ST 1-7. Complement-dependent borreliacidal assays were performed with anti-OspA_ST1 Bin 1 mAbs and a panel of B. burgdorferi HB19-R1 (ospA negative) strains harboring an IPTG-inducible mScarlet-I viability reporter plasmid and expressing ospA ST1-7 under conditions described in the Materials and Methods. Controls included an HB19-R1 strain carrying the IPTG-inducible viability reporter plasmid alone (empty vector, ospA negative) and a mAb with borreliacidal activity restricted to OspA_ST1 (LA-2). EC₅₀ values shown represent the mean lowest concentration of antibody (nM) resulting in >50% reduction in MFI relative to controls. Reporter strains that exhibited resistance to complement-mediated killing by mAbs at 10 nM were later retested at a single higher dose (66.6 nM). The data shown encompasses 3-5 independent experiments per strain with data normalized as described within the Materials and Methods section. Corresponding antibody titration curves and statistical comparisons are shown in S3 Fig.

https://doi.org/10.1371/journal.ppat.1013740.g002

We therefore used the panel of HB19-R1 OspA reporters to assess the ability of the full collection Bin 1 mAbs (8 in total) to confer cross-serotype, complement-dependent borreliacidal activity (Fig 2C). The results reflected the OspA binding profiles in that Class I mAbs (857-2, 221–20, and 221–5) exhibited potent borreliacidal activity against all seven OspA serotypes, with EC₅₀ values ranging from 0.63 to 1.25 nM. Providing further validation of our HB19-R1 reporter system, 857-2’s EC₅₀ against the ST4 reporter strain is consistent with the previously reported value for B. bavariensis strain (PBi), from which the OspA_ST4 sequence was derived [11]. Class II mAbs (221–11, 223–5, 227-2, and 221–7) had reduced potency against OspA ST3 and ST7 (EC₅₀ 2.5-10 nM), while 227-1 (Class III) was completely devoid of ST3 and ST7 killing.

Structural basis of OspA_ST1 recognition by Bin 1 mAbs

To elucidate the epitopes recognized by the three different classes of Bin 1 mAbs, we solved the X-ray crystal structures of FAbs 857-2 (Class I), 221–5 (Class I), 221–11 (Class II), and 227-1 (Class III) in 1:1 stoichiometric complexes with OspA_ST1 at resolutions ranging from 2.2 Å to 3.2 Å (Tables 2 and S3 and Fig 3). Attempts to solve the X-ray crystal structure of 221–20 FAbs in complex with OspA were unsuccessful. After molecular replacement, the resulting phase information was used to calculate electron density maps employed to manually insert the corresponding residues into each model while manually building additional regions within each Fab-OspA model. Refinement of the four Fab-OspA structures revealed molecular models that possessed excellent geometry and that were consistent with the crystallographic data (S4 Table).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Summary of OspA-Fab interactions.

https://doi.org/10.1371/journal.ppat.1013740.t002

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Crystral structures of Fab-OspA complexes reveal conserved epitopes on OspA_ST1.

(top row) Ribbon diagrams of OspA_ST1 (green) in complex with Fabs from (A) 857-2 (class I), (B) 221-5 (class I), (C) 221-11 (class II), and (D) 227-1 (class III). Fab heavy chains (V_H and C_H1) are colored cyan; light chains (V_L and C_L) are colored magenta. The OspA N- and C-termini are labelled accordingly, along with selected strand numbers. (middle row) The four OspA (green) ribbon representations aligned in the same orientation with respective Fab contract points colored in blue. (bottom row) The ribbon and surface representations of OspA highlight the Fab-interacting residues, shown in blue. The N and C-termini of OspA are labelled N and C, respectively. PDB ID for each structure are provided in Table 2.

https://doi.org/10.1371/journal.ppat.1013740.g003

In each of the four complexes, the Fabs assumed a canonical structure with two heavy chain immunoglobulin domains (V_H, C_H1) and two light immunoglobulin domains (V_L, C_L) each containing 7–9 β-strands arranged in two β-sheets that folded into a two-layer sandwich with all six CDRs (L1-3; H1-3) on one face of each molecule. OspA was comprised of one antiparallel β-sheet with 21 β-strands (referred to as β-strands 1–21) connecting globular N- and C-terminal domains with a single α-helix (referred to as α-helix A) at the C-terminus (Fig 3). All 21 β-strands were present in their respective electron density maps of OspA complexed with Fabs 227-1 and 857-2. In the OspA-221–11 complex β-strands 1 and 2 were not visible in the electron density maps, while β-strands 17–21 were absent in the electron density maps of the OspA-221–5 complex, likely because of greater flexibility within these regions of OspA when bound to the particular Fabs. The OspA molecules from each of the four complexes were structurally similar to OspA alone (PDB ID: 2G8C), as evidenced by Root Mean Square Deviation (RMSD) of 0.8 Å to 2.4 Å upon superposition (S3 Fig). Thus, the Fabs did not induce any major conformational changes within OspA.

The OspA-Fab structures revealed that 857-2 (Class I), 221–5 (Class I), 221–11 (Class II), and 227-1 (Class III) recognize overlapping but distinct epitopes on OspA with similar modes of engagement (Fig 3). The four Fabs contact OspA β-strands 6–10 and loop regions between β-strands 7–8 (loop 7–8) and 9–10 (loop 9–10), including residues 87, 100, 102, 105, 106, 107–109, 112, 124, 126, 128–132 (Fig 3). All but 857-2 Fab contacts OspA loop 11–12.

Fab 857-2 (Class I; K_D 0.30 nM) and 221–5 (class 1; K_D 0.52 nM) buried 1,397 Ų and 1,423 Ų, respectively, involving CDRs H1, H2, H3, and L2. 857-2 formed seven hydrogen bonds and three salt bridges with OspA, while 221–5 forms 8 hydrogen bonds and five salt bridges with OspA (Table 2). The Fabs each form two salt bridges with OspA residue Lys-107; one via CDR-L2 residue Asp/Glu-55 and the other with CDR-H3 residue Asp-109/110 (Fig 4A-4B). Fab 221–5 makes contacts with Ser-152 in loop 11–12 via framework region (FR) residue Arg-59 but is too distant (4.2 Å) to form an H-bond. 857-2 does not interact with loop 11–12 at all, as Arg-59 is > 9 Å from OspA Ser-152.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Key interactions between each Fab and OspA.

Close-up of notable salt bridges between OspA Lys-107 (green) and Fab residues are shown as yellow dashed lines. Fab heavy chains (V_H/C_H1) are colored cyan, and light chains (V_L/C_L) are colored magenta. (A) Fab 857-2 engages Lys-107 through Glu-55 and Asp-110. (B) Fab 221-5 interacts with Lys-107 via Asp-55 and Asp-109. (C) Fab 221-11 contacts Lys-107 through Asp-109. (D) Fab 227-1 engages Lys-107 through Glu-55. All side chains are drawn as sticks and color coordinated to the main chain color with nitrogen atoms blue and oxygen atoms red.

https://doi.org/10.1371/journal.ppat.1013740.g004

Fab 221–11 (Class II; K_D 0.35 nM) engages with OspA via CDRs H1, H2, H3, and L2 and has a total buried surface area of 1,348 Ų (Table 2). It forms 9 polar contacts with OspA, including a hydrogen bond between CDR-H1 Tyr-32 and OspA Asp-105. Like 857-2 and 221–5, 221–11 also forms a salt bridge between CDR-H3 residue Asp-109 and OspA residue Lys-107 (Fig 4C). However, unlike the other two mAbs, 221–11 light chain FR Glu-55 does not form a second H-bond with Lys-107 (Fig 4C).

Fab 227-1 (Class III; K_D 1.7 nM) stands out from the other three Fabs in that it forms the most extensive interface with OspA, burying a total of 1,704 Ų with contributions from five of the six CDRs (H1-3, L1-2). Fab 227-1 establishes five hydrogen bonds and three salt bridges with OspA, including with Lys-107 (Fig 4D). FR residue Arg-59 forms a hydrogen bond with Ser-152 of OspA, as will be discussed in detail below. 227-1 makes additional contacts with OspA not observed in the other three Fabs, including with loop 11–12 (residues 153–154), β-strand 13, loop 13–14 (residues 171–173), and loop 15–16 (residue 193). These additional interfaces result in 239 Ų of total buried surface area. This larger binding surface likely comes at a cost, considering that 227-1 (Class III) has a restricted OspA serotype profile (i.e., fails to bind OspA_ST3 and OspA_ST7) relative to 857-2 and 221–5, which bind the seven major OspA types.

Structural basis of cross serotype restriction of Bin 1 mAbs

Based on the structural analysis of the OspA-Fab interactions, we postulate that two distinct molecular interactions account for the reduced binding of 221–11 (Class II) and 227-1 (Class III) to OspA ST3 and ST7. The first interaction involves the Glu_ST1 > Lys_ST3/7 polymorphism at position 131 (Fig 5A and 5B). For example, 227-1 V_H FR3 residue Arg-59 forms a salt bridge with Glu-131 in OspA_ST1, which would be disrupted by the substitution of a Lys at that position. Moreover, a mutation to Lys at position 131 would introduce electrostatic repulsion with Lys-129, which is located ~5 Å away. As a result, the salt bridge between OspA Lys-129 and 227-1 V_H residue Asp-55 would be perturbed. Thus, a single Glu₁₃₁Lys polymorphism would result in the loss of two salt bridges in the case of 221–11 and 227-1. Second, the Ser_ST1 > Asn_ST3/7 polymorphism at position 152 may restrict 221–11 (Class II) and 227-1 (Class III) recognition of OspA ST3 and 7. 221–11 and 227-1 each form H-bonds with OspA_ST1 residue Ser-152 via V_H residue Arg-59 (Fig 5C and 5D). The Asn polymorphism at this position would perturb a corresponding interaction in OspA_ST3 and OspA_ST7.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Influence of OspA primary sequence deviations on antibody function.

(A) Structural basis for Fabs 221-11 and 227-1 cross-reactivity loss in OspA serotypes 3 and 7. Distinct hydrogen bond networks between Fab Arg-59 (cyan) and OspA residues (green) contribute to serotype specificity. (A) In Fab 221-11, Arg-59 forms a salt bridge with OspA Glu-131. (B) The same Glu-131 interaction is observed in Fab 227-1. (C) In 221-11, Arg-59 hydrogen bonds with OspA Ser-152. (D) Similarly, Fab 227-1 also engages Ser-152 through Arg-59. The primary sequence difference distinguishing serotypes (STs) 1, 2, 4, 5, and 6 from STs 3 and 7 involves a Glu-Lys substitution at position 131 and a Ser-Asn at position 152. These two primary sequence differences conceivably disrupt the Arg-59–Glu-131 salt bridge and the Arg-59 hydrogen bond with Ser-152, consequently reducing the cross-reactivity of Fabs 221-11 and 227-1 with ST3 and ST7. All side chains are drawn as sticks and color coordinated to the main chain color with nitrogen atoms blue and oxygen atoms red. Salt bridges are represented as yellow dashes and hydrogen bonds are red dashes.

https://doi.org/10.1371/journal.ppat.1013740.g005

OspA recognition is driven by V_H and V_L germline-encoded residues

All eight of the Bin 1 mAbs, including the four whose structures were solved in complex with OspA (857-2, 221–5, 221–11, 227-1) as well as the previously reported 221–7 (PDB 7JWG), are inferred to utilize the same HV5–51 germline lineage (Table 1 and S6 Fig). In the case of 221–5 (Class I), the V_H is unmutated from the HV5–51 germline, while the V_H elements of the other four mAbs with solved co-complexes have only nominal mutations (4–6 amino acids) that are not associated with direct OspA interactions. Additionally, the H-CDR3 residue involved in the well-conserved salt bridge with Lys-107, namely Asp109 (or Asp110 or 114, depending on CDR3 length), is encoded by the J_H gene segment in five of the six human germline J genes (J_H1 being the exception) (Table 1). Similarly, in six of the mAbs, the V_L is derived from the KV1–13 with germline residues involved in critical contacts, including Glu-55 that forms a salt bridge with OspA Lys-107 (Fig 4). In the case of 221–5 and 221–20 (Class I), which are derived from KV3–11, a single mutation in codon 55 (C > A, resulting in amino acid mutation Ala > Asp) would be sufficient to form a salt bridge with OspA Lys-107. Therefore, BCR combinations with HV5–51 and either KV1–13 or KV3–11 would be expected to have broad affinity for OspA serotypes.

Borreliacidal epitopes are conserved across OspA in diverse Bbsl genospecies

With a detailed structural understanding of the interactions of Bin 1 mAbs with OspA_ST1, we sought to examine epitope conservation across the Bbsl genospecies complex. To accomplish this, we surveyed OspA protein sequence divrersity throughout the Borrelia genus to identify novel OspA types beyond the eight STs and nine ISTs that have been reported [38]. To this end, we constructed an unrooted phylogenetic tree consisting of 135 OspA protein sequences collected from all 23 recognized Bbsl genospecies [39]. To facilitate the identification of clades corresponding to known ST/ISTs, we included 86 OspA reference sequences obtained from the PubMLST OspA IST collection assembled by Lee [38].

The resulting phylogenetic tree revealed 33 distinct sequence clusters corresponding to OspA ST 1–8, IST 9–17, and 16 other sequence groups representing previously untyped OspA variants (Fig 6A). The 86 OspA reference sequences grouped into discrete clades consistent with their previously designated ST or IST designations [38]. The one exception was B. valaisiana, which clustered into two distinct clades: four sequences clustering as IST15, and three other sequences sharing close homology with IST16, principally associated with B. turdi (Fig 6A). We also identified a single B. garinii OspA sequence (isolate IPT75) that grouped independently of the eight recognized OspA types found within this genospecies. The 14 other OspA clades identified corresponded to untyped B. burgdorferi sl genospecies, including B. americana, B. andersonii, B. bissettiae, B. californiensis, B. carolinensis, B. chilensis, B. finlandensis, B. japonica, B. kurtenbachii, B. lanei, B. lusitaniae, B. maritima, B. sinica, and B. tanukii (Fig 6A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Phylogenic analysis of OspA within the Borrelia burgdorferi sensu lato (Bbsl) genospecies complex.

(A) Unrooted phylogenetic tree comprised of 135 OspA protein sequences from 23 Bbsl genospecies. Phylogenetic analysis identified 33 distinct OspA clades corresponding to serotypes (ST) 1-8, in silico types (IST) 9-17, and 16 other unclassified OspA IST’s. Clades representing each OspA variant/type are distinguished by color. Enlarged and bolded strain names denote OspA alleles selected for HB19-R1 viability reporter strain construction. Bootstrap values representing approximately maximum likelihood analysis are distinguished by color: > 0.50 = green, > 0.75 = blue, 1.00 = red. Nodes with bootstrap values under 0.50 are not depicted in the figure (gray) (B) Simplified multiple sequence alignment of the Bin1 epitope region of 33 unique OspA types. Representative sequences from each OspA clade were aligned using OspA from B. burgdorferi strain B31 as a reference. Bolded strain names indicate OspA variants expressed by HB19-R1 viability reporter strains. Amino acid residues numbered in white denote conservation within at least two Bin 1 mAb epitopes with solved Fab-OspA crystal structures (857-2, 221-5, 221-11, 221-7, and 227-1). Numbered amino acid residues with colored background form critical interactions with residues within the paratopes of Bin 1 mAbs with solved crystal structures (Lys-107, Glu-31, Ser-152).

https://doi.org/10.1371/journal.ppat.1013740.g006

Genetically diverse OspA types are susceptible to Bin 1 mAbs

To assess epitope conservation across the Bbsl genospecies complex, we generated a multiple sequence alignment (MSA) of representative OspA sequences from each of the 33 clades identified in our phylogenetic analysis. Using OspA ST1 as the reference, the alignment revealed broad conservation of the loop regions between β-strands 7–8 and 9–10, including residues associated with 857-2 and other Bin1 mAb epitopes (Fig 6B and S7 Fig).

To experimentally examine conservation of 857-2’s and other Bin 1 epitopes beyond OspA ST 1–7, we generated 19 additional HB19-R1 viability reporter strains expressing OspA variants from two additional B. bavarienisis types (IST9, IST10) and 17 other Bbsl genospecies. Many of the OspA types selected are occasionally associated with Lyme disease in humans such as B. bavariensis IST9 and IST10, B. spielmanii (IST13), B. mayonii (IST14), B. americana, B. bissettiae, and B. lusitaniae or exhibit limited or uncertain disease potential in humans like B. andersonii, and B. yangtzensis [39–42]. We also generated HB19-R1 viability reporter strains expressing OspA types from non-pathogenic Bbsl genospecies to fully capture Bbsl complex-wide sequence diversity and explore the impact of polymorphisms predicted to disrupt Bin 1 mAb binding. These included B. valaisiana (IST15), B. turdi (IST16), B. sinica, B. finlandensis, B. japonica, B. lanei, B. carolinensis, B. kurtenbachii, B. maritima, and B. tanuki [39,41,43]. The source strains for each OspA type or IST are indicated in bold within the phylogenetic tree (Fig 6A) and corresponding multiple sequence alignment (Fig 6B and S7 Fig).

The Bin 1 mAbs were then evaluated for complement-dependent borreliacidal activity against the 19 additional OspA reporter strains described above. Class I mAbs (221–20, 221–5, 857-2) were the most broadly functional, as exemplified by 221–20, which exhibited potent borreliacidal activity against 23 of the 26 OspA reporter strains within our collection [EC₅₀ 1.25-2.50 nM] (Fig 7). The three resistant strains (EC₅₀ > 60 nM) were B. turdi (IST16), B. lusitaniae, and B. sinica. 221–5 and 857-2 susceptibility profiles were similar to 221–20, with the addition of both having lessened activity against B. valaisiana (IST15), and 857-2 also being ineffective against B. americana (Fig 7). The Class II mAbs (221–11, 227-2 and 223–5) had borreliacidal activity profiles similar to 857-2, albeit with reduced potencies (Fig 7). Interestingly, 223–5 was distinct among this class for its inability to kill B. carolinensis. Finally, 227-1 (Class III) had borreliacidal activity against just 17 of the 26 OspA reporter strains and reduced efficacy (Ec₅₀ 5–10 nM) against OspA variants from B. mayonii (IST14), B. finlandensis, B. japonica, and B. lanei (Fig 7). These results demonstrate that one or more borreliacidal epitopes are conserved across virtually all the major variants within the Bbsl genospecies complex.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Genetically diverse OspA types are susceptible to complement-mediated killing by anti-OspA_ST1 Bin 1 mAbs.

Complement-dependent bactericidal assays were performed as described in Materials and Methods using anti-OspA Bin 1 mAbs and a panel of B. burgdorferi HB19-R1 mScarlet-I viability reporter strains expressing 19 OspA in silico types (ISTs). The heat map summarizes mean EC₅₀ values for complement-mediated killing by Bin 1 mAbs from Classes 1-3. EC₅₀ values represent the lowest mean antibody concentration (nM) producing >50% reduction in mScarlet-I fluorescence relative to untreated controls and were derived from ≥3 independent experiments. Susceptibility profiles of reporter strains expressing OspA ST1-7 are included for comparison. Strains that exhibited resistance to mAbs at 10 nM were retested at 66.6 nM. Corresponding antibody titration curves and statistical comparisons between the HB19-R1 OspA ST1 reporter strain and strains expressing OspA ST2-7 are shown in S8 Fig.

https://doi.org/10.1371/journal.ppat.1013740.g007

Lys-107 polymorphism as a determinant of susceptibility to nearly all Bin 1 mAbs

As noted above, HB19-R1 strains expressing OspA from B. turdi (IST16) and B. lusitaniae PoHL-1 were resistant to all the Bin 1 mAbs tested, while B. americana and B. valaisiana (IST15) were resistant to a subset of Class I mAbs and all the Class II and Class III mAbs (Fig 7). Examination of the OspA MSA (Fig 6B) suggested that polymorphisms at OspA_ST1 Lys-107 may account for all or part of the observed resistance. In the case of B. turdi (IST16), for example, there is a Thr residue rather than a Lys at position 107, while OspA from B. lusitaniae has a deletion at the equivalent position (Fig 6B). Polymorphisms at Lys-107 would preclude 857-2 and other mAbs from forming critical salt bridges with OspA (Fig 4).

To investigate the importance of Lys-107 in the context of OspA recognition by the Bin 1 mAbs, we generated HB19-R1 viability reporter strains expressing OspA_ST1 variants harboring a K107T substitution or Lys-107 deletion (ΔK107). Complment-dependent killing assays revealed that the OspA_ST1 K107T and ΔK107 reporter strains were completely resistant to Class II (227-2, 223–5, 221–11) and Class III mAbs (227-1), even at the highest antibody concentrations tested (66.6 nM) (Fig 8). Similarly, both modifications were detrimental to Class I mAbs 857-2 and 221–5’s borreliacidal activity, as evidenced by 16- fold increases in EC₅₀ values (Fig 8). For 857-2, the OspA_ST1 ΔK107 deletion was slightly more deleterious than the K107T substitution (EC₅₀ 20 nM vs 5 nM), while 221–5’s resistance was comparable between the two (10 nM). Conversely, the OspA_ST1 K107T and ΔK107 reporter strains remained sensitive to killing by 221–20, demonstrating the unique nature of 221–20’s paratope-epitope interaction and differentiating 221–20 from the other two Bin 1 mAbs.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. Lys-107 polymorphisms are a determinant of susceptibility to nearly all Bin 1 mAbs Complement-dependent killing assays were performed as described in the Materials and Methods with anti-OspA Bin1 mAbs and B. burgdorferi HB19-R1 viability reporter strains expressing “WT” OspA variants from B. burgdorferi B31 (ST1), B. valaisiana VS116 (IST15), and B. turdi TPT2017 (IST16), or mutated derivatives containing deletions, insertions, or substitutions at residue 107.

Heat maps (A) and radar plots (B) depicting the Ec₅₀ profiles of WT and mutated OspA variants. Ec₅₀ values denote the lowest antibody concentration (nM) capable of reducing mScarlet-I fluorescence by >50% relative to untreated controls and represent the mean of > 3 independent experiments. Reporter strains exhibiting resistance to killing by mAbs at 20 nM were re-examined at 40 nM and 66.6 nM doses. (C) Heat maps graphically depicting significant differences in mAb susceptibility between HB19-R1 reporter strains expressing WT and mutated OspA variants as determined via two-way ANOVA followed by Tukey’s multiple comparisons test. Adjusted P-values reflecting significant differences between strains are shown in shades of purple or pink. Antibody titration curves summarizing borreliacidal activity against each strain are provided in S9 Fig.

https://doi.org/10.1371/journal.ppat.1013740.g008

In light of these results, we examined whether introduction of Lys-107 in an otherwise resistant OspA variant from B. valaisiana (IST15) was sufficient to render cells susceptible to killing by the Bin 1 mAbs. In accordance with that hypothesis, the OspA_IST15_G106insK reporter strain was susceptible to all Bin 1 mAbs with EC₅₀ values comparable to OspA_ST1 (EC₅₀ range 0.63- 2.50 nM), thereby demonstrating that antibody recognition is transformed by the addition of a single Lys residue at position 107. We also generated a reporter strain with a B. turdi OspA (IST16) allele encoding a T107K substitution to investigate the role of Lys-107 in the context of polymorphisms elsewhere in 857-2 and other Bin 1 mAb epitopes (e.g., residues 128–132). In fact, the T107K substitution was sufficient to render B. turdi OspA at least partially susceptible to the Class I mAbs (857-2, 221–5, 221–20) and one Class II mAb (223–5) with EC₅₀ values ranging from 10-40 nM (Fig 8). Collectively, these experiments demonstrate that Lys-107 is a key susceptibility determinant for virtually all Bin 1 mAbs.

Cloning and expression of OspA and anti-OspA mAbs

Recombinant human monoclonal IgG1 antibodies were purified via Protein A affinity column from Expi293 cells that had been transiently transfected with V_H and V_L plasmids, as described [11]. Recombinant OspA_ST1 (residues 18–273) from B. burgdorferi B31 (NCBI:txid224326) was expressed in Escherichia coli BL21-DE3 and purified as described [12]. Purified OspA was maintained in 20 mM HEPES, 150 mM NaCl, 20 mM imidazole, pH 7.50. Recombinant non-lipidated OspA serotypes 1–7 purified from E.coli were kindly provided by Dr. Meredith Finn (Moderna) and described elsewhere [34].

ELISA

Immulon 4HBX plates (ThermoFisher, Waltham, MA) were coated overnight at 4°C with 1 µg/mL of the seven OspA serotypes (non-lipidated) and the next morning blocked for 2 h with 2% goat serum in PBS supplemented with 0.1% Tween-20 (PBST). Serial two-fold dilutions of each antibody starting from 1 µg/mL were made in separate dilution plates, and transferred to the serotype coated plates for 1 h. Plates were then washed three times with PBST, and a goat anti-human HRP conjugated secondary antibody (SouthernBiotech) was added for 1 h. Plates were then washed three times with PBST, developed with SureBlue TMB substrate (Seracare, Milford, MA), and quenched with 1M phosphoric acid. Absorption at 450 nm was read by a SpectraMax iD5 spectrophotometer using Softmax Pro v7.1 (Molecular Devices, San Jose, CA).

Affinity measurements with BioLayer interferometry

Certain monoclonal antibody binding kinetics first reported by Haque and colleagues [32] have been revised to reflect refined and optimized BLI protocols using an Octet RED96e (Sartorius AG, Gottingen, Germany). AHC sensors (#18–5060, Sartorius) were used to capture mAbs at 0.4 µg/mL for 5 min, to a loading level of ~0.5 nm, then dipped into wells containing a three-fold dilution series ofnon-lipidated OspA (ranging from 0.6 µg/mL, 21.58 nM to 0.0025 µg/mL, 0.089 nM) for 10 min. Finally, sensors were dipped into buffer alone for 30 minutes to measure dissociation. Buffer used in all wells was 0.1% BSA in PBS, with 0.01% Tween20. Sensors were regenerated before the first mAb and between each mAb with 0.2 M glycine, pH 2.2. Traces were background corrected with a reference sensor, in which mAb was loaded but not exposed to OspA, and a reference sample, in which a bare sensor was exposed to the highest concentration of OspA. Traces for each mAb were fit to a global 1:1 model. k_obs vs concentration was linear for each system.

Construction of B. burgdorferi strains expressing OspA serotypes

The IPTG-inducible mscarlet-I viability reporter plasmid, pGW189 [34] was modified to facilitate the expression of ospA variants of interest under the regulatory control of the P_ospAB promoter from B. burgdorferi strain B31. To accomplish this, transcription terminators were first introduced into pGW189 to prevent read through transcription from the lacI and mscarlet-I ORFs. The lambda_T0_rrnBT1 and rrnbT2 transcription terminators were amplified from plasmids pTN7 [50] and puc18_mtn7-lacZ_gentR[51] using NEB’s Q5 DNA polymerase and tailed primers listed in S5 Table. Amplified DNA fragments were then assembled into pGW189 immediately downstream of the lacI and mscarlet-I ORFs using the SacI/HindIII restriction sites and NEB’s HiFi Assembly kit, creating pGW206.

The ospA promoter and ORF from B. burgdorferi B31 were next amplified in two overlapping DNA fragments with tailed primers designed to introduce a silent SphI restriction site ~50 bases into the ospA ORF and permit assembly into pGW206 via the PvuI and PvuII restriction sites. The resulting viability reporter/OspA ST1 expression plasmid, pGW217, subsequently served as the basis for creating additional OspA expression plasmids. To accomplish this, pGW217 was first digested with SphI and PvuII to remove the ospA_ST1 ORF immediately following the conserved lipoprotein insertion sequence (amino acid residue 17). Gene fragments from additional ospA serotypes/Bbsl genospecies were then synthesized (IDT) and assembled directly into the vector using HiFi DNA Assembly kit (NEB).

Silent mutations were introduced into ospA gene sequences when necessary to reduce sequence complexity for in vitro synthesis. Mutations were introduced at sites that resulted in a shift to the most common or second most common codon utilized by B. burgdorferi B31. The codon adaptive index for each ospA cDNA was assessed prior to synthesis using JCAT [52] to ensure that the introduced silent mutations did not reduce translation efficiency in B. burgdorferi.

Gene fragments were also synthesized to create B. burgdorferi B31, B. valaisiana, and B. turdi reporter plasmids with insertions, deletions, or substitutions at position 107 within the ospA ORF. To permit direct comparison with B31 point mutants, a second “wild type” OspA ST1 reporter plasmid was also created containing the same silent mutations that were introduced to increase GC content/reduce sequence complexity for in vitro synthesis. The sequences of primers and synthesized ospA gene fragments used in this study are provided in S5 and S6 Tables.

OspA reporter plasmids were sequenced and transformed into B. burgdorferi HB19-R1 [37], a high-passage derivative of the human blood isolate HB19 that lacks linear plasmid 54 (ospAB⁻), using previously described methods [53]. Transformants were selected in liquid BSKII supplemented with gentamicin (100 µg/ml) and individual clones were isolated through serial dilution [53]

To evaluate native plasmid content in transformants, the plasmid profiles of B. burgdorferi HB19 and HB19-R1 were characterized using NEB’s Taq Quickload 2 × PCR MM with established B. burgdorferi plasmid profiling primer sets [54]. Additional primer sets targeting cp26, lp21, lp28–1, lp5, lp28–5, and lp28–6 were also included in the analysis [54,55].Since plasmid content was uniform among clones transformed with mScarlet-I reporter plasmids carrying ospA ST1–7 (pGW217–pGW223, pGW250), plasmid profiling was not performed on HB19-R1 reporter strains encoding OspA ISTs due to the labor-intensive nature of the screening process. All Bbsl strains used or constructed in this study are listed in S1 and S2 Tables.

Flow cytometry-based antibody binding to strains

To analyze the ability of Bin 1 mAbs to bind to native OspA ST1–7 on the bacterial surface of primary isolates, Borrelia strains B. burgdorferi B31 (ST1), B. afzelii Pko (ST2), B. garinii PBr (ST3), B. bavariensis PBi (ST4), B. garinii PHei (ST5), B. garinii TN (ST6), and B. garinii T25 (ST7) (kindly provide by Dr. John Leong, Tufts University) were cultured at 33°C with 5% CO₂ until mid-log phase. Bacterial cells were stored at -80ºC in fresh media containing 20% glycerol, and preparation, incubations, and flow cytometry analysis was performed as described [47]. To analyze the ability of Bin 1 mAbs to bind to OspA ST1–7 on the bacterial surface of reporter strains HB19-R1, (GGW1072–1079; S2 Table), bacterial cells were cultured at 33°C with 5% CO₂ in BSKII media, minus gelatin and containing gentamycin 50 µg/ml, until mid-log phase. Bacterial cell preparation, incubations, and flow cytometry analysis was performed as described [56]. The Salmonella-specific mAb, Sal4 IgG, was used as an isotype control [57].

Complement-dependent borreliacidal assays

Complement-dependent bactericidal assays were performed as previously described [34] with several modifications. Because guinea pig and human-derived complement perform equivalently in this assay with respect to sensitivity and reproducibility (S11 Fig), all experiments were conducted using guinea pig complement due to financial constraints. Frozen aliquots of OspA-expressing HB19-R1 reporter strains were thawed at room temperature and 500 µl from each stock (equivalent to 5x10⁷ spirochetes) were transferred to 50 ml conical tubes containing 45 ml of in gelatin-free BSK II supplemented with 50 µg/ml of gentamicin. Tubes were then sealed, and the cultures were incubated at 33ºC under static growth conditions. Three days later, spirochetes were collected via centrifugation (4,000 x g), the supernatant was removed, and cell pellets were resuspended in phenol red-free BSK II with gentamicin (50 µg/ml) at a density of 3 x 10⁷ spirochetes/ml.

Spirochetes were next seeded into white bottom 96 well microtiter assay plates (Costar) and mixed 1:1 with serial dilutions of anti-OspA_ST1 mAbs prepared in phenol red-free BSKII containing 5% guinea pig complement (Sigma-Aldrich) and gentamicin (50 µg/ml), resulting in 3x10⁶ spirochetes per well with a final complement concentration of 2.5% and antibody concentrations between 10.00 nM - 0.08 nM, or 20 nM - 0.16 nM (depending on strain/experiment. Eight untreated (antibody-free) control reactions per strain were also included on each assay plate to determine baseline and peak fluorescence for data normalization.

Following antibody addition, assay plates were incubated overnight at 37ºC with 5% CO₂ without agitation. The next day (18–20 h later), 1 mM IPTG was added to appropriate wells to induce expression of mScarlet in surviving spirochetes. Assay plates were then returned to the incubator for another 24 h period. The following day, median fluorescence intensity (MFI) was measured three times per plate at 569 nm (excitation) and 611 nm (emission) using a Spectromax ID3 plate reader (Molecular Biosystems) and Softmax pro version 7.1 software.

Mean MFI data was independently normalized for each assay plate using the untreated/non-induced and untreated/IPTG-induced controls to establish baseline (0) and peak (1) MFI. Graphpad Prism version 9.5.1 was then used to generate killing curves and heatmaps, while radar plots were constructed in Microsoft Excel. Ec₅₀ values were assigned to the lowest antibody dilution resulting in >50% reduction in MFI relative to untreated controls. Ec₅₀ values reported in heatmaps and radar plots reflect the mean calculated using normalized MFI data from three to five independent experiments. Strains that exhibited resistance to anti-OspA Bin 1 mAbs at the highest concentration tested (10 or 20 nM) were subjected to additional bactericidal assays with 10 µg/ml of the antibodies in question. Resistance at this dosage is shown as “>60 nm” in heat maps and radar plots.

Cloning, expression and purification of OspA and anti-OspA Fabs

The PCR amplicon encoding B. burgdorferi OspA residues 18–273 was subcloned in frame into the pSUMO expression vector with an N-terminal deca-histidine and SUMO tag. All cloning was performed using a standard ligase independent cloning protocol. OspA was expressed in E. coli strain BL21 (DE3). The transformed bacteria were grown at 37°C in TB medium and induced at 20°C with 0.1 mM (IPTG) at an OD₆₀₀ of 0.6 for ~16 hours at 20°C. After induction, cells were harvested and resuspended in 20 mM Hepes pH 7.5 and 150 mM NaCl. The cell suspension was sonicated and centrifuged at 30,000 g for 30 minutes. After centrifugation, the protein-containing supernatant was purified by nickel-affinity and size-exclusion chromatography on an AKTAxpress system (GE Healthcare), which consisted of a 1mL nickel affinity column followed by a Superdex 200 16/60 gel filtration column. The elution buffer consisted of 0.5M imidazole in binding buffer, and the gel filtration buffer consisted of 20mM Hepes pH 7.5, 150mM NaCl, and 20mM imidazole. Fractions containing OspA (18–273) was pooled and subject to TEV protease cleavage (1:10 weight ratio) for 3 hours at room temperature in order to remove their respective fusion protein tags. The cleaved protein was passed over a 1mL Ni-NTA agarose (Qiagen) gravity column to remove the added TEV protease, cleaved residues, and uncleaved fusion protein. Each IgG was subjected to papain digestion followed by affinity depletion of the Fc fragment by Protein A FPLC chromatography to generate each Fab. After purification, each Fab and OspA were mixed in a 1:1 stoichiometry to form a stable complex concentrated to a final concentration of 10 mg/ml for all crystallization trials.

Crystallization and data collection

Fab-OspA crystals were grown by sitting drop vapor diffusion at 4°C using a protein to reservoir volume ratio of 1:1 with total drop volumes of 0.2 μl. Crystals of each OspA-Fab complex were produced using the crystallization solutions shown in S3 Table. All crystals were flash frozen in liquid nitrogen after a short soak in the appropriate crystallization buffers supplemented with 25% ethylene glycol. Data were collected at the 24-ID-E beamline at the Advanced Photon Source, Argonne National Labs. Data resolution limit was determined based on the I/s criterion which was 2.0 in the highest resolution shell. All data was indexed, merged, and scaled using HKL2000 [58] then converted to structure factor amplitudes using CCP4 [59].

Structure determination and refinement

The structure of each Fab-OspA complex was solved by molecular replacement using Phaser [58]. Molecular replacement calculations were performed using the Fab coordinates from PDB code 2XTJ as the search model for Fab 221–5, the Fab heavy chain coordinates from PDB code 4RIR along with the Fab light chain coordinates from PDB code 4M6O were independently used as search models for Fabs 221–11 and 857-2. The Fab coordinates from PDB code 4M6O were also used as the search model for Fab 227-1. OspA coordinates (PDB code 1FJ1) were used as the search model for OspA in all four Fab-OspA complex structures solved. The resulting phase information from molecular replacement was used for manual model building of the Fab-OspA model using the graphics program COOT [60] and structural refinement employing the PHENIX package [61]. Data collection and refinement statistics are listed in S4 Table. Molecular graphics were prepared using PyMOL (Schrodinger) (DeLano Scientific LLC, Palo Alto, CA). The structures generated in this study were deposited in the Protein Data Bank (PDB; http://www.rcsb.org/pdb/) under accession numbers shown in Table 2.

Phylogenetic tree construction

A collection of 135 OspA protein sequences containing representatives from 23 recognized Bbsl genospecies was compiled from borreliabase.org, NCBI’s OspA IST typing database (PubMLST), and Bacterial and Viral Bioinformatics Resource Center (BV-BRC.org). MUSCLE (accessed via EMBL) was then used to generate a multiple sequence alignment of the collection. The resulting MSA file was then used to construct an unrooted approximately maximum likelihood phylogenetic tree using Fasttree and LG modelling (accessed via BV-BRC.ORG). The resulting phylogenetic tree was then formatted and visualized using ITOL version 7 [62]. Files associated with phylogenetic tree construction are provided in S1 File.

OspA multiple sequence alignment (MSA)

A MSA was created containing one representative sequence from each of the 33 OspA clades/types identified through phylogenetic analysis. OspA variants selected for the alignment included OspA sequences utilized in viability reporter strain construction and OspA sequences from type or reference strains (when available). Sequence alignment was performed using the MAFFT algorithm with OspA from B. burgdorferi B31 as a reference. Following alignment, the Bin 1 epitope region (β strands 6–10, residues 82–146) was visualized using the MSA viewer interface on BV-BRC.org. A graphic of the alignment was then created with the Taylor color scheme and the sequence logo displayed to emphasize conservation between OspA types/clades.

Microsphere immunoassay (MIA)

ST 1–7 and OspB were coupled to Magplex-C microspheres (5mg antigen/ 1x10⁶ microspheres) using a xMap Antibody Coupling Kit as recommended by the manufacturer (Luminex Corporation, Austin, TX). Successful coupling of ST 1–7 and OspB was confirmed using LA-2 and H6831, respectively. Beads were protected from light and stored 2–8°C in xMAP AbC Wash Buffer (5x10⁶ microspheres/mL) until use.

OspA Bin 1 mAbs were serially diluted (1:2) in assay buffer (1 x PBS, 2% BSA, pH 7.4) in round bottom, non-treated plates (Costar, Kennebunk, Maine) with starting concentrations of 2.5 mg/mL (223–5), 0.625 mg/mL (857-2, 221–5, 221–20, 221–11, 227-1), or 0.3125 mg/mL (227-2). Coupled microsphere stocks were diluted (1:50) in assay buffer and added (50 μL/well) to black, clear-bottomed, non-binding, chimney 96-well plates (Greiner Bio-One, Monroe, North Carolina). The mAb dilutions (50 μL) were combined with the microspheres and incubated at room temperature for 1 hr in a tabletop shaker (600 rpm). Plates were placed on a magnetic separator and washed three times using wash buffer (1 x PBS, 2% BSA, 0.02% TWEEN-20, 0.05% Sodium azide, pH 7.4). PE labeled goat anti-Human IgG Fc, eBioscience (Invitrogen, Carlsbad, California) secondary antibody diluted 1:500 in assay buffer was added (100 μL/well) and incubated at room temperature for 30 min in a tabletop shaker (600 rpm). Plates were washed as previously stated. The microspheres were resuspended in 100 μL of wash buffer and placed back on the tabletop shaker (600 rpm) for 5 minutes prior to analysis using a FlexMap 3D (Luminex Corporation).