Although the TprC and TprD proteins are identical in the Nichols, Chicago, and Bal73-1 strains, allelic variants of TprC and TprD exist among syphilis strains and the three Tp subspecies (35, 37). Among the treponemal strains used in this study ( Figure 1 ), four alleles were found at the tprD locus, which include the reference tprD allele (found in the syphilis Nichols, Chicago, and Bal73-1 strains), and the tprD2 allele (found in the syphilis strains MexicoA, Sea81-4, Bal3, and UW249) which encodes the TprD₂ protein (35). Also the subsp. pertenue SamoaD strain and subsp. endemicum IraqB strains harbor a tprD₂ allele in the tprD locus, but their TprD₂ amino acid sequences differ from the subsp. pallidum TprD₂ sequence due to five amino acid substitutions scattered throughout the length of the protein ( Figure 1 ) (35). TprD2 has four unique regions that differentiate it from the reference TprD sequence. These include a large central region of 110 amino acids and three smaller regions toward the COOH-terminal end of the protein ( Figure 1 ) (35). As previously reported, the tprC locus of MexicoA, Sea81-4, and Bal3 encodes a TprC variant with a limited number of amino acid (aa) changes (15 aa for MexicoA, 9 aa for Sea81-4, and 9 aa for Bal3) compared to the reference TprC found in Nichols, Chicago, and Bal73-1 strains ( Figure 1 ) (35). The TprC protein of the pertenue and endemicum strains studied here also shows limited amino acid changes compared to the reference TprC (31 aa for SamoaD and 26 aa for IraqB; Figure 1 ), albeit higher compared to the subsp. pallidum strain (35). We previously reported that TprC and TprD/D₂ sequence variation does not occur randomly, but rather is localized in discrete variable regions (DVRs; Figure 1 ) (35). When TprC and TprD variants are compared (with the exclusion of TprD₂), seven DVRs are found throughout the protein sequence, while 8 DVRs can be identified within the TprD₂ sequences ( Figure 1 ). To obtain predictions of TprC and TprD₂ structures from their amino acid sequences ( Figure 1 ) and map the DVRs on these models, we used the recently developed AlphaFold2 software (https://AlphaFold.ebi.ac.uk/) (38). These new models revealed remarkably similar structures for TprC and TprD₂ and identified these proteins as relatively large β-barrel integral OMPs of 20 β-strands connected by ten external loops (ExLs, protruding toward the extracellular milieu), and nine periplasmic loops ( Figure 2 ). Elevated structural identity of the transmembrane region was also supported by a 0.75 backbone root-mean-square deviation (RMSD) score ( Figure 2A ). The local model quality, indicated by the Predicted Local Distance Difference Test (pLDDT) value was high in the transmembrane and periplasmic loop regions, and slightly lower in the predicted ExLs, suggesting conformational flexibility ( Figure 2B ). Except for two substitutions (aa 407 and 410 mapping to a periplasmic β-turn), all DVRs localized within a subset of the surface-exposed ExLs ( Figure 2C ). More specifically, DVRs were located in ExL1, ExL5-6 and ExL8-10 of the TprC and TprD₂ models; while ExL2-4 harbored conserved loops. ExL6 is also conserved between TprD₂ sequences from various isolates, although its shows only 60% of sequence identity to the ExL6 of other TprC and TprD variants ( Figure 1 ). DALI software (39) and PDB analyses to identify structurally similar porins ( Table S1 ) showed that the highest-scoring structures did not contain the exact number of β-strands predicted by AlphaFold2 for TprC and TprD₂ β-barrels, but slightly higher or slightly lower, but well within the models of integral OMPs with no large periplasmic domains. These results suggest that these Tpr proteins belong to a new family of porins not yet represented in the PDB database.

In silico prediction of B-cell epitopes using BepiPred2.0 (http://www.cbs.dtu.dk/services/BepiPred/), IEDB (https://www.iedb.org/), and BCpreds (http://ailab-projects1.ist.psu.edu:8080/bcpred/data.html) ( Tables S2–S5 and Figure S1 ) showed that the putative TprC and TprD ExLs were also enriched in immunogenic epitopes. Therefore, it is possible that the antigenic variability in the ExLs regions has functional significance in immunity to the T. pallidum subspecies. To validate the B-cell epitope prediction and evaluate the cross-reactivity of these epitopes across species and strains, we performed experimental B-cell epitope mapping of the TprC, and TprD/D₂ proteins with a Geysan pepscan approach based on overlapping synthetic peptides (40) using sera from animals infected with Tp subsp. pallidum, Tp subsp. pertenue, and Tp subsp. endemicum strains. Furthermore, we compared antibody reactivity in sera from infected rabbits with that of sera from rabbits immunized with a subset of full-length refolded recombinant TprC proteins.

Groups of three laboratory rabbits were infected intratesticularly (IT) with one of seven syphilis strains (Nichols, Chicago, Bal73-1, MexicoA, Sea81-4, Bal3, and UW249), one yaws strain (SamoaD), and one bejel strain (IraqB). From these animals, serum samples were obtained at day 30, 60, and 90 post-infection. Pooled sera from animals in each infection group/time point were tested in ELISA to assess reactivity to homologous overlapping synthetic peptides (20-mers overlapping by 10 amino acids) representing the TprC and TprD/D₂ variants previously identified in each strain. The full list of synthetic peptides used in this study, with amino acids encompassing predicted ExLs highlighted in red with yellow text, and percentage amino acid homology among peptides across strains is shown in Table S6 . Peptide nomenclature is explained in Table S6 footnote. Cumulative absorbance data from the three timepoints (sum of the mean absorbance values for day 30, 60, 90 values for each infected rabbit group) are reported in Figures 3A–C . Epitope mapping studies of the NH₂-terminal portion of the protein resulted in the identification of six highly reactive peptide regions ( Figure 3A ) representing sequences shared by all TprC and TprD genes in the studied subspecies pallidum strains: C1-C3, C6, C13-C14, C18, C20, and C25-C29. Based on AlphaFold2 structural predictions, 9 of these 13 peptides had at least 7 amino acids mapping to the predicted external loops of the protein, while only four reside in predicted transmembrane scaffolding and periplasmic loop regions (C1, C6, C20 and C25; Figure 1A and Table S7 ). It is noteworthy that all three B cell epitope prediction programs uniformly predicted all six of the experimentally determined epitope-containing regions of the NH₂-terminal portion of the subspecies pallidum TprC and D proteins ( Tables S2–S5 and Figure S1 ).

Several epitopes were also identified in the COOH-terminal region of these proteins, and corresponded to peptides the same regions in pallidum and non-pallidum subspecies: C46 and C47 homologs from Nichols ( Figure 3A and Table S7 ), SamoaD (S-C46, S-C47; Figure 3B and Table S7 ) and IraqB (I-C46, I-C47; Figure 3B and Table S7 ), C51 homologs from Nichols (N-C51), SamoaD and iraqB (S/I-C51) ( Figures 3A, B and Table S7 ); and C53-C55 homologs from Nichols, Bal3/Sea81-4 ( Figure 3A and Table S7 ), and IraqB ( Figure 3B and Table S7 ). Similarly, the C43, and D45-D47 (ExL8) peptides, mapping to the TprD₂ COOH-terminus were found to contain B-cell epitope(s) ( Figure 3C and Table S7 ). Additional TprD₂ peptides found to be reactive were D33-D35 (ExL6), I-C39, D40-41 (ExL7), C49, and D51 (ExL9). In our 3D models of these proteins, all the reactive peptides in the COOH-terminus fall within predicted ExLs ( Tables S6–S7 and Figure 1 ), except for C43, most of I-C39 (75%), C49, and C53, which are predicted transmembrane scaffolding sequences. Of these “scaffold epitopes”, only one (C43) was predicted by a B-cell prediction program ( Tables S2–S5 and Figure S1 ).

The percentage of immune sera that showed reactivity to many of the peptides was variable. For example, peptides C3, and C13 were recognized by rabbits infected with 28% of the Tp subsp. pallidum strains; peptides C1, and C27 were recognized by rabbits infected with 42% of the strains; peptides C14 and C28 were recognized by rabbits infected with 57% of the strains; C6 was recognized by rabbits infected with 71% of the strains; and peptides C2 and C18 were recognized by rabbits infected with 85% of the Tp subsp. pallidum strains ( Figure 1 ). Overall, based upon the AlphaFold2 models, these results show that humoral reactivity elicited to these Tpr antigens during experimental infection is directed primarily to predicted surface-exposed regions of the TprC/D and TprD₂ proteins.

Epitope mapping using short peptides based on the TprC and TprD/D₂ sequences from multiple Tp strains and sera from infected animals also allowed us to investigate cross-reactivity to non-homologous peptides to determine the fine specificity of the antibody response to these antigens. Such analyses focused on peptides mapping to the proteins` COOH-terminal regions, due to the higher sequence variability in this region, compared to the more conserved NH₂-terminal region ( Figure 1 ). Major variable regions include peptides C46 - C47 (mapping to the predicted ExL8), C51 (ExL9), and C55 (ExL10) ( Figure 1 and Tables S6–S7 ). Four distinct variants of each of the C46, C47, and C55 peptides, and three variants of C51, representing all sequences found in the strains studied here, were tested against all nine pools of immune sera obtained at day-90 post-experimental infection.

As shown in Figures 4A–D , very few sera were reactive only to their homologous peptide. For example, the Bal73-1and SamoaD antisera were primarily reactive only to their own C46 sequences ( Figure 4A ), although the Bal73-1 antiserum showed a very modest reactivity to the IraqB peptide variant ( Figure 4A ). When reactivity against the C47 peptide was analyzed, the Chicago, Bal73-1 sera reacted only to their own peptide, the IraqB antiserum reacted to all variants but the SamoaD peptide, and the Sea81-4 serum only saw the Nichols C46 variant, but not its homologous peptide. ( Figure 4B ). Only Chicago, Bal73-1, and SamoaD sera showed complete strain-specificity for the C51 peptides ( Figure 4C ), while none of the sera reactive to C55 showed complete strain-specificity ( Figure 4D ). When tested against TprD2 peptides, most antisera did not show any reactivity. There were, however, two exceptions, as the Chicago sera cumulatively showed reactivity to the D34 and D47 peptides, with OD values of 3.6 and 6.6, respectively. However, only the D47 peptide was consistently recognized at all time points, while D34 was recognized only at day 60. Overall, these data indicate that cross reactivity is possible, and perhaps suggest that immunization with a given sequence might generate cross-reactive antibodies able to overcome the obstacle of sequence diversity among TprC epitopes, a feature that is desirable in vaccine development as they may be broadly opsonic or neutralizing.

Refolded antigens, analyzed using circular dichroism (CD), were found to have a β-pleated sheet component of about 48% for all three antigen variants. Random coil was also found to be 48% of the protein structure, while only 4% was identified as alpha helices. Epitopes recognized following immunization with any of three recombinant full-length TprC variants from Tp subsp. pallidum strains (Nichols/Chicago/Bal73-1, Sea81-4/Bal3, and MexicoA) were also identified to evaluate differences with infection-induced antibody responses. Results showed that sera from animals immunized with the Nichols TprC sequence were highly reactive to peptides C1-C3, C6 and C47, and moderately reactive to peptides C5, C9, C16-18, C28, C32, C34-C35, C53 and C55 ( Figure 5A ). Of these 16 peptides, six mapped almost exclusively to putative surface-exposed loop regions (C28, C32, C34, C35, C47, and C55), five (C1, C5-C6, C16, and C53) mapped to predicted transmembrane scaffolding sequences, while five peptides (C2-C3, C9 and C17-C18) contained both surface-exposed loops and scaffold regions. Sequences of these peptides and location in the predicted protein models are reported in Table S8 . When tested against non-homologous peptides ( Figure 5B ), the Nichols TprC-immunized sera strongly recognized the SamoaD/IraqB C2-C3 variants, and all three heterologous C47 variants (SamoaD, Iraq B, and Sea81-4), while modest reactivity was seen towards the MexicoA/UW249 C55 peptide variant, the SamoaD/IraqB C34, and both C26 variants from SamoaD and IraqB ( Figure 5B ). Immunization with the Bal3 variant of TprC elicited high reactivity to peptides C1-3, C6, and C13, and moderate reactivity to peptides C7, C16-18, C20, C43, C47, and C49 ( Figure 5C and Table S8 ). Of these thirteen peptides, six (C2-C3, C13, C17-C18, and C47) mapped predominantly to ExLs, and seven (C1, C6, C7, C16, C20, C43, and C49) predominantly to the protein transmembrane scaffolding ( Table S8 ). Cross-reactivity to non-homologous peptides was seen predominantly to the SamoaD/IraqB C2 and C3, IraqB C22, and all variants of C47 and C51 ( Figure 5D ). Antisera from rabbits immunized with the MexicoA TprC variants primarily recognized homologous peptides C1-3 and, secondarily, C5, C6, and C28 ( Figure 5E ). Of these six, one peptide mapped to the predicted ExL6 (C28), three mapped only to the transmembrane scaffolding (C1, C5-C6) and two mapped to a peptide predicted to contain portions of both (C2-C3) ( Table S8 ). Cross-reactivity to the non-homologous SamoaD/IraqB C2 and C3 was also detected ( Figure 5F ). Overall, these data show that, as seen in infection-induced antibody responses, the humoral response following immunization with full-length TprC variants is mainly elicited by predicted surface-exposed sequences, rather than sequences mapping to the β-barrel transmembrane scaffolding, and that cross-reactivity to non-homologous peptides is possible.

A side-by-side comparison of the infection- vs. immunization-induced humoral response to peptides is shown in Figure 6 . For this comparison, the mean value of the cumulative reactivity seen in sera at day 30, 60, and 90 sera post-experimental infection is shown for each peptide. Sera from immunized animals were obtained three weeks after the last immunization. All sera were tested at the same dilution. In general, immunization-induced reactivity to most peptides appeared to be higher than that elicited by experimental infection; specific examples include C1-C3, C5, C9, C16-C17, C28, C32, C34-C35, N-C47, and C53 peptides ( Figure 6A ). For Nichols-clade T. pallidum strains ( Figure 6A ), which contain identical tprC and tprD loci, this was most noticeable for epitopes located in the NH₂- and COOH-terminal regions of the protein. In contrast, infection-induced antibody responses to epitopes in the central part of the protein were comparable to or higher than those induced by immunization. For TprD2-containing subsp. pallidum strains ( Figures 6B, C ), immunization-induced responses were limited to the NH₂-terminal portion of the protein (including ExL1-3) and virtually no immunization-induced antibodies were detected for epitopes in the central and COOH-terminal regions, although these were recognized by infection-induced responses. Overall, these data support that, in most cases, immunization elicits a higher reactivity to TprC B-cell epitopes compared to experimental infection, particularly for those epitopes located in the NH₂-terminal portion of the protein. These data support the preferential use of the amino portion of TprC, which contains multiple conserved ExLs, for vaccine studies.

Outbred adult male New Zealand White rabbits ranging from 3.0-4.0 Kg were obtained from R&R Rabbitry (Stanwood, WA). Prior to entry into the study, serum from each animal was tested with both a treponemal (FTA-ABS) and a non-treponemal (VDRL; BD, Franklin Lakes, NJ) test to rule out infection with the rabbit syphilis agent Treponema paraluiscuniculi. Only rabbits seronegative in both tests were used for either propagation or experimental infection for sample collection. Tp strains were propagated by intratesticular (IT) inoculation and harvested at peak orchitis as previously described (60). For experimental infections, groups of three rabbits were infected IT with a total of 5x10⁷ Tp cells per testis. In total, nine Tp isolates (one isolate per rabbit group) were used: seven Tp subsp. pallidum isolates (Nichols, Chicago, Bal73-1, Sea81-4, Bal3, MexicoA, and UW249), one Tp subsp. endemicum (IraqB) and one Tp subsp. pertenue (SamoaD) ( Table S9 ). Briefly, on the day of infection bacteria were extracted from rabbit testes in sterile saline containing 10% normal rabbit serum (NRS), and testicular extract was collected in sterile 15-ml tubes. Extracts were centrifuged twice at 1,000 rpm (180 x g) for 10 minutes in an Eppendorf 5810R centrifuge (Eppendorf, Hauppauge, NY) to remove gross rabbit cellular debris. Treponemes were enumerated under a dark-field microscope (DFM) and percentage of motile organisms was recorded. Extracts were then diluted in serum-saline to the desired concentration (5x10⁷/ml). Following IT injection, treponemal motility was assessed again to ensure that the time elapsed before injection into the new host did not affect pathogen viability. After IT inoculation, establishment of infection was assessed by monitoring development of orchitis during the following three weeks as well as by performing FTA-ABS and VDRL tests on sera collected at day 30 post-inoculation. Immune sera were collected from the animals at day 30, 60, and 90 post-infection. Animals were then euthanized. Extracted sera were heat-inactivated at 56°C for 30 min and stored at -20°C until use for ELISAs.

Sequences for the tprC gene of Tp isolates (Nichols, Sea81-4, and MexicoA) were previously cloned (35). For expression, the tprC sequences were sub-cloned into the pET23b+ vector (Life Technologies) between BamHI and HindIII using the primers C-S (5’-cgggatccgatgg gcgtactcactccgca) and C-As (5’-gcaagcttccatgtcactttcattccac). For sub-cloning, the tprC ORF was amplified in a 100-μl final volume using 0.4 units of GoTaq polymerase (Promega) with approximately 10 ng of DNA template. MgCl₂ and dNTP final concentrations were 1.5 mM and 200 μM, respectively. Initial denaturation and final extension (72°C) were for 10 min each. Denaturation (94°C), annealing (60°C), and extension (72°C) were carried out for 1 min each for a total of 35 cycles. Amplicons were purified, digested, and ligated into the pET23b+ vector. As a result of cloning into pET23b+, 28 additional amino acids were added to the TprC ORFs (14 NH₂-terminal and 14 COOH-terminal amino acids), including the COOH-terminal 6×His tag for affinity purification. Ligation products were used to transform OneShot TOP10 chemically competent E. coli cells (Life Technologies) according to the provided protocol. Transformations were plated on LB-Ampicillin (100 μg/ml) agar plates for selection. For each cloning reaction, individual colonies were screened for the presence of insert-containing plasmids using primers annealing upstream and downstream of the pET23b+ vector poly-linker (T7 promoter and terminator primers). Positive plasmids were extracted from overnight liquid cultures obtained from replica colonies by using the Plasmid Mini kit (Qiagen, Germantown, MD), and two to five clones for each strain were sequenced to ensure sequence fidelity to the previously cloned templates (35). For expression of recombinant antigens, a suitable clone for each tprC gene variant was used to transform E. coli Rosetta (DE3) competent cells (Life Technologies).

E. coli cells were grown overnight in LB media supplemented with ampicillin (100 μg/ml). The following day, multiple flasks containing 200 ml of auto-inducing media (61), were inoculated with 20 ml of overnight culture in a 2-liter baffled flask and grown at room temperature for 72 h at 175 rpm in a shaking incubator. Expression of recombinant antigens in induced and un-induced controls was assessed by immunoblot using a monoclonal anti-poly-histidine antibody (Millipore-Sigma, diluted 1:2000) after SDS-PAGE. Prior to purification, presence of the recombinant protein in the soluble and insoluble cellular fractions was evaluated by SDS-PAGE and immunoblot. Recombinant TprC purification was carried on under denaturing conditions. Briefly, E. coli cell pellets were resuspended in 5 ml/g of dry culture weight of binding buffer (50 mM NaH₂PO₄, 10 mM imidazole, pH 8.0) w/o denaturing agent, and the suspension was sonicated in ice with 100 pulses of 6 s each, with each pulse being separated by 10-s intervals. Insoluble components (containing the desired products) were precipitated by centrifugation and resuspended in 5 ml/g of culture weight of binding buffer (50 mM NaH₂PO₄, 5 mM imidazole, pH 8.0) containing 6M Guanidine-HCl denaturing agent and sonicated again as above. Insoluble components were precipitated again by centrifugation and the supernates were saved. For affinity chromatography, 5.0 ml of nickel-agarose (Ni-NTA agarose, Qiagen) was packaged into a 1.5x14 cm column (Bio-Rad, Carlsbad, CA) and washed with 3 column volumes of molecular-grade H₂O and 6 column volumes of binding buffer + denaturing agent. Cell lysate was then loaded, and the flow was adjusted to 1 ml/min. Unbound proteins were washed using 10 bed volumes of binding buffer, followed by 6 column volumes of wash buffer (50 mM NaH₂PO₄, 20 mM imidazole, pH 8.0) containing denaturing agent. Washing continued until the A ₂₈₀ of the flow through was <0.01 AU. Recombinant TprC was eluted with 15 ml of elution buffer (50 mM NaH₂PO₄, 300 mM imidazole, pH 8.0) containing denaturing agent. Eluted fractions devoid of visible contaminants by SDS-PAGE and Coomassie staining were pooled, and protein concentration was assessed by micro-bicinchoninic (BCA) assay (Thermo-Fisher). Pooled fractions were then dialyzed in PBS using a 10 kDa MWCO Slide-A-Lyzer dialysis cassette (Thermo-Fisher) over 12 hours, ensuring PBS change every ~4 hours. Precipitated protein, resulting from elimination of Guanidine-HCl during dialysis was transferred into microcentrifuge tubes and spun down at full speed. After removing the supernate, the pellet was resuspended in a volume of PBS containing 6M urea suitable to achieve a protein concentration of ~4 mg/ml, and protein concentration was then reassessed using the micro-BCA assay kit (Thermo-Fisher). Prior to immunizations, urea was eliminated using Profoldin (Hudson, MA) M7 renaturing columns for membrane proteins, which were used according to the manufacturer’s protocol. M7 renaturing columns were found to provide the best yield when screened along with 19 other conditions offered by Profoldin. Lipid composition of the elute buffer included lysophosphatidylcholine (∼5 mM), arginine (∼150 mM), glycerol (∼10%), dodecyl maltoside (0.7 mM), and Tris-HCl (0.1 mM), pH 7.5). Following buffer exchange, soluble protein concentration was evaluated using micro-BCA assay and analyzed by circular dichroism (CD) to evaluate percentage of β-sheet, alpha-helix, and random coil. CD spectra (190 to 260 nm) were acquired in triplicate at room temperature using 0.5 mg/ml of recombinant refolded TprC in a Jasco-1500 high-performance CD spectrometer. CD spectra were analyzed using the online platform Dichroweb (http://dichroweb.cryst.bbk.ac.uk/html/home.shtml) (62) and the spectra from buffer alone for background subtraction.

Groups of three rabbits each were immunized with one of the purified, refolded recombinant TprC variants. Rabbits were injected with 125 μg of refolded protein every 3 weeks for a total of three immunizations. Prior to injection, antigen was mixed with an equal volume of in Titermax Gold Adjuvant (Millipore-Sigma), a water-in oil emulsion containing squalene, the block co-polymer CRL-8300, and a microparticle stabilizers to obtain a final volume of 1 ml. Immunogen-adjuvant preparation was performed according to the manufacturer`s instruction, and immunizations were performed via four 250 μl injections (each containing 31.25 μg of protein) into 4 intramuscular sites. Three weeks after the last boost, immunized animals were deeply anaesthetized, bled through cardiac puncture, and then euthanized.

Overlapping synthetic peptides (20-mers overlapping by 10 aa) were designed to represent the sequences of all TprC and TprD/D₂ loci present in each of the seven strains examined in this study starting after the predicted signal peptide (AA 1-22; Figure 1 and Figure 2 ). Only the C56 peptide and its variants ( Table S6 ), which represent the proteins’ COOH-terminus, were synthesized as 26-mers. A total of 120 peptides ( Table S6 ) were produced by Genscript (Piscataway, NJ). Upon receipt, lyophilized peptides were rehydrated in sterile PBS to a stock solution of 200 μg/ml. Solubility of hydrophobic peptides was increased by adding up to 4% (v/v) DMSO per manufacturer’s instruction when needed (peptides C1, C4-7, C10, C15-16, C20, C25, C38-C39, C43-44, C53; Table S6 ). Reconstituted peptides were stored at -20°C until use. For ELISA, peptides were further diluted to 10 μg/ml in PBS, and 50 μl of working dilution (500 ng total) were used to coat the wells of a 96-well Microwell Maxisorp flat-bottom plate (Thermo-Fisher, Waltham, MA) as previously described (44). Absorbance was measured at OD₄₀₅ using a Molecular Devices SpectraMax Plus microplate reader (Molecular Devices, San Jose, CA). A micro-BCA protein assay (Thermo Fisher) was performed in plates coated with Ag and washed to demonstrate that all peptides bound to the well surfaces in the plates (data not shown). For each serum from each group, the value of each replicate experimental wells minus background reactivity (i.e., three times the mean of the wells tested with pooled uninfected rabbit serum) was calculated and plotted. If residual value for the No-antigen control wells was present after subtraction, statistical significance was calculated with one-way ANOVA with the Bonferroni correction of multiple comparisons or t-test, with significance set at p<0.05. Except for figures showing cumulative absorbance, graphs represent the mean ± SEM for triplicate wells tested with pooled sera from the 3 rabbits in each group after background subtraction.

We used the ColabFold interface (63) to construct Multiple Sequence Alignments (MSA) for the TprC and TprD₂ query sequences by searching UniRef30 (64), Mgnify (65) and ColabFold sequence databases with MMSeq2 (66). The MSA was used as input for structure prediction with AlphaFold2 (38) using the default settings (template=False, amber_relax=False, 3 recycles). Visualization was performed using PyMol software (https://pymol.org ) (67).