We conducted genome enrichment of the study genomes by using Agilent SureSelect protocol (https://www.agilent.com) before sequencing. Lineage classification using both the 3-gene multilocus sequence typing (MLST) scheme (14) and the core genome lineages generated with PopPUNK (15) confirmed that the strains were part of the TPA subspecies. Most (15/17) of the TPA genomes belonged to the SS14 lineage; the remaining genomes belonged to the Nichols lineage (Table 1; Appendix Figures 1–4). Core genome–based classification outperformed the MLST because it successfully identified the lineages of all 17 samples, whereas MLST failed for 3 samples.

Most genomes from Canada were sampled in Saskatchewan in 2024 during a provincewide syphilis outbreak in the Prairies (Table 1). Those samples were geographically dispersed, likely suggesting multiple independent transmission events; however, detailed sexual contact tracing information is not available. We used the additional 784 TPA genomes from public databases (505 genomes from GenBank, 279 de novo assembled from data in previous studies [Appendix]) to determine the breadth of the mutational landscape of the 16S rRNA gene.

We retrieved reference sequences of 16S rRNA genes from C. acnes (NR_040847.1:1–1486), E. coli (U00096.1:4166659–8200), H. pylori (CP003904.1:1512657–1157), and TPA (Nichols lineage: NC_010741.1:231287–2831 and SS14 lineage: NC_021508.1:231297–2845) from NCBI and aligned them using MAFFT to translate known tetracycline resistance–associated mutation sites (6–8,11) (Table 2) to the TPA coordinate system. Specifically, the mutation at position 1032 in C. acnes corresponds to position 1061 in TPA; the E. coli mutation at position 964 aligns to position 966 and the E. coli mutation at position 1053–1055 aligns to positions 1056–1058 in TPA. Similarly, the H. pylori resistance–associated triplet at positions 965–967 (6,11) maps to 967–969 in TPA (Table 3). The coordinates of the reported mutational triplet in T. pallidum at positions 965–967 appear to be relative to E. coli (11), because the mutational triplet was identified at positions 967–969 in our reference strains. We manually inspected known tetracycline resistance–associated positions (Table 2) for mutations (Appendix).

Alignment of the 16S rRNA genes also revealed that the single-nucleotide polymorphisms (SNPs) associated with doxycycline resistance in C. acnes and E. coli share the same wild-type background as TPA (Table 2). Because a single mutation at those sites can potentially cause doxycycline resistance (7,8), similar mutations could possibly exert comparable effects in TPA. In contrast, resistance-associated mutations in H. pylori arise from a unique wild-type background relative to C. acnes, E. coli, and TPA. If the composition of the nucleotide triplet is key to conferring doxycycline resistance, then 3 of the 4 known resistance alleles will require 2 nucleotide substitutions in TPA. The only H. pylori resistance allele that requires a single mutation in TPA is gGA (T967G) (6).

Of the 784 global TPA genomes, 9 contained a heterozygous G/T allele at position 968 (4). That mutation is not sufficient to cause resistance; however, it does bring the strains within a single SNP of the gtA allele in H. pylori (6). Those 9 TPA genomes were collected during 2013–2019 from the United Kingdom (n = 6), Australia (n = 1), and Hungary (n = 1) and were mostly part of the SS14 lineage (7/9) (Table 4). In contrast, the 17 Canada TPA genomes contained no mutations in the 16S rRNA gene compared with the reference strains.

We called the 16S rRNA variants using a diploid model for TPA to account for the presence of 2 gene copies. To confirm that those findings were not methodological artifacts, we analyzed the 23S rRNA gene and found it to be duplicated. Specifically, we investigated the A2058G and A2059G SNPs in E. coli (13) that correspond to positions 2106 and 2107 in TPA. We detected macrolide resistance in 94% (16/17) (Table 1) of the Canada genomes through a A2106G mutation and in ≈66% (514/784) of the publicly available TPA genomes, supporting the robustness of those AMR findings.

The increasing use of doxycycline as a prophylactic for syphilis presents a growing risk for the emergence of AMR. Given the genetic stability of T. pallidum (10), genomic surveillance programs should prioritize monitoring positions 966–969, 1056–1058, and 1061 (TPA numbering) of the 16S rRNA genes, because those sites could serve as early indicators of emerging doxycycline resistance (6–9,11). Their relationship to doxycycline resistance in TPA remains theoretical and based on comparative genomics (5–7); in vitro phenotypic validation will be essential to determine their functional significance (11). Future analyses must account for the presence of 2 copies of the 16S rRNA gene in TPA, because heterozygous alleles could attenuate the phenotypic expression of doxycycline resistance. Nevertheless, our work establishes a global baseline for 16S rRNA diversity in TPA, simplifying future doxycycline resistance surveillance.