A novel Treponema pallidum subsp. pallidum strain associated with a painful oral lesion is a member of a potentially emerging Nichols-related subgroup

Abstract

Background:

Early syphilitic lesions are typically painless; however, several recent case studies have included patients with tender lesions and no evidence of concurrent infections. Here we present the manifestations and serological and molecular findings of a patient from New York State with a painful tongue lesion.

Methods:

The diagnosis of syphilis was based on a combination of physical examination, serologic, pathologic, and immunohistochemical findings. DNA obtained from a formalin fixed paraffin embedded (FFPE) biopsy was used to characterize the infecting pathogen using PCR, multilocus sequence typing (MLST), and whole genome sequencing (WGS) methods.

Results:

PCR and MLST of the biopsy specimen confirmed infection with Treponema pallidum subsp. pallidum (T. pallidum) of the Nichols cluster. WGS analysis of this strain (herein called NYMC01) showed that it contained 17 unique single nucleotide variations and 4 more complex genetic differences; this novel genotype matched only two specimens, both from a patient in Seattle, Washington, U.S.A. The presence of this rare genotype in two geographically distinct locations suggests the potential emergence and spread of a new subgroup of the Nichols cluster.

Conclusions:

To our knowledge, this is the first genomic sequence obtained from a T. pallidum strain linked to a painful lesion, and the third description of whole genome sequencing of T. pallidum from FFPE tissue. Analysis of additional specimens may reveal that the NYMC01-related genotype represents an emerging T. pallidum subgroup and may also aid in determining whether the painful clinical presentation of primary syphilis is related to specific T. pallidum genotypes.

Summary

A patient diagnosed with syphilis had a painful oral lesion containing a rare genetic variant of the infectious agent Treponema pallidum. This finding raises the question of whether certain genetic variants of this spiral-shaped bacterium may cause unusual, painful lesions.

In 2000, syphilis incidence reached a historic low in the United States of 31,618 cases (11.2 cases per 100,000 population). Since then, the case rate of syphilis has increased almost every year. In 2021, 176,713 cases were reported (53.2 per 100,000), including 53,767 primary and secondary cases (1). Men having sex with men (MSM) have been disproportionately affected, accounting for 53% of all male patients’ primary and secondary cases. However, the syphilis rate among women has increased considerably in recent years, growing 3.25-fold from 2017 to 2021, indicating a heterosexual epidemic (1). The same pattern has been observed in New York State, with a concomitant rise in congenital syphilis diagnoses; 41 cases of congenital syphilis were reported in 2021 as compared to 16 cases in 2017 (1).

Early recognition and prompt treatment of syphilis are crucial for controlling the spread of this infection. Clinicians have been trained to diagnose primary syphilis by recognizing the characteristics of a chancre, which is classically described as an indurated, painless ulcer at the site of Treponema pallidum infection. However, several case series have reported tender chancres, single or multiple, without evidence of coinfections with herpes simplex virus (HSV) or other infectious agents known to cause pain. These painful lesions were attributed to syphilis based on T. pallidum polymerase chain reaction (PCR) testing of the lesions (2–5) or detection of T. pallidum by immunohistochemistry (6). It is reasonable to raise the question of whether the painful chancre phenotype is due to infection caused by a particular strain (or strains) of T. pallidum.

In this study, we describe a patient with a painful oral lesion caused by infection with T. pallidum subsp. pallidum with a novel genotype, as determined by whole genomic sequencing from formalin-fixed, paraffin embedded (FFPE) biopsy tissue.

METHODS

Patient Information and Histologic Procedures

The patient consented to participate in this study, and the patient’s personal health information and identifiers were protected throughout, in accordance with HIPAA regulations.

A biopsy specimen (0.3 × 0.3 × 0.1 cm) of an ulcerated lesion on the tongue was obtained for diagnostic purposes and was fixed in formaldehyde and embedded in paraffin using standard histologic procedures. Sections (4 μm) were stained with hematoxylin and eosin for microscopic analysis. Immunohistochemical staining was performed using purified rabbit IgG against T. pallidum (BIOCARE Medical, Pacheco, CA).

DNA Extraction, Analysis, and Sequencing

DNA was extracted from two 10 μM microtome sections of the FFPE block using a QIAamp^® DNA FFPE Advanced UNG kit (Qiagen), which reduces formalin-induced DNA crosslinking and removes deaminated cytosine residues by uracil-N-glycosylase treatment (7, 8). One μl of the resulting DNA preparation was used for nested PCR of TP0136, TP0548, TP0705, and rRNA genes as described by Grillová, et al. (9). The products were purified by agarose electrophoresis and Sanger sequenced (Azenta/GENEWIZ) for multilocus sequence typing (MLST). For whole genome sequencing (WGS), DNA from the microtome sections was further purified using AMPure bead clean-up technology (Beckman Coulter). Library preparation (NEBNext^® Ultra^™ DNA Library Prep Kit, New England BioLabs) and Illumina sequencing (Illumina HiSeq in the 2 × 150-bp Paired End configuration) were performed by Azenta Life Sciences.

Genome Assembly and Data Analysis

DNA sequencing reads were preprocessed using Cutadapt v2.3 (10). Standard Illumina sequencing adapters were removed, and paired-end reads were aligned to the T. pallidum Nichols strain genome sequence (NCBI NC021490.2), SS14 strain sequence (NC021508.1), T. denticola strain ATCC 35405 sequence (NC002967.9), T. medium strain ATCC 700293 sequence (GCA_000413035.1), T. vincentii strain F0403 sequence (GCA_000412995.1), Haemophilus ducreyi sequence (CP015425.1), Human Herpesvirus I strain 17 sequence (GCA_000859985.2), or Human Herpesvirus II strain HG52 sequence (GCA_000858385.2) using the default parameters of Bowtie2 v2.3.4.1 and a seed substring length of 18 nt (11). The resulting Sequence Alignment Map (SAM) files for the Nichols strain was converted to Binary Alignment Map (BAM) format and sorted using Samtools (12). Polymorphisms were identified using Snippy 3.2 (13). The genome assembly was then refined through an iterative process using MiniMap2 (14) and template modification to resolve discrepancies, as described previously (15). Annotation was performed by NCBI using the Prokaryotic Genome Annotation Pipeline (PGAP) Version 6.5. The resulting NYMC01 genome sequence and related data were deposited in the NCBI nucleotide database (Accession Number CP125219, BioProject PRJNA962037, and BioSample SAMN34377415). The Type Strain Genome Server (16, 17) was used to create a BLAST Distance Phylogeny (GBDP) tree using default parameters.

Preliminary alignments indicated that the CP124560 and CP124561 genome sequences were most similar to NYMC01, as compared to the Nichols strain and SS14 strain sequences. Revised assembly (RA) analyses of the corresponding specimens P-22–20198 and R-22–10139 (18) were performed using the Illumina sequencing reads downloaded from the NCBI Short Read Archive (Biosamples SAMN34387050 and SAMN34387051, respectively). Assemblies were then produced using MiniMap2 (14) and the NYMC01 genomic sequence as template, followed by iterative refinements using refined templates to resolve discrepancies (15). Sequence ambiguities with <80% nucleotide agreement at a given position were designated as N. The number of N’s in the NYCM01, P-22–20198_RA, and R-22–10139_RA assemblies was 120, 180, and 213, respectively, with most of these being in the highly variable tprK gene.

RESULTS

Clinical Case

In November 2021, a 39-year-old heterosexual, HIV-negative woman presented to an otolaryngology clinic in New York State for the evaluation of a painful tongue lesion of one month’s duration. The lesion was severely tender, and the tongue was erythematous, swollen, and friable (Fig. 1A). No other lesions or symptoms were found or reported at that time. To exclude a possible malignancy, a biopsy was performed. The patient was treated with clindamycin for a presumed bacterial superinfection. After seven days of treatment, there was no improvement and a 7-day course of doxycycline was then prescribed. Changing the treatment from clindamycin to doxycycline significantly improved the tongue lesion (Fig. 1B).

Figure 1. Photography of tongue and lesion before and after seven days of treatment with oral doxycycline.

(A) Before treatment. The area of the biopsy is apparent in the middle of the tongue. (B) After treatment, showing resolution of the lesion.

Tissue histopathology revealed an intense lymphohistiocytic inflammatory reaction (Fig. 2A), and T. pallidum immunocytochemistry revealed a high density of spirochetes in the subepithelial connective tissue (Fig. 2B). Subsequent serologic testing revealed a reactive anti-T. pallidum IgG EIA and an RPR titer of 1:8. The diagnosis and appropriate treatment were delayed because syphilis was not initially suspected. Following infectious disease consultation, the patient was treated with 2.4 million units of benzathine penicillin administered intramuscularly, with complete lesion resolution. The patient was referred to the local health department for contact tracing. The patient reported having provided oral sex to a male partner a few weeks before the occurrence of the lesion. The male partner reported no symptoms, but had a positive T. pallidum EIA with an RPR titer of 1:64. Based on this test result, he was treated for syphilis with benzathine penicillin.

Figure 2. Photomicrograph of the tongue biopsy.

(A) A dense lymphocyte, histiocyte, and reactive plasma cell infiltration was present (Hematoxylin-Eosin). (B) A high prevalence of spirochetes (red) was detected by immunostaining using anti-T. pallidum antiserum with a Giemsa counterstain (blue).

Genetic Typing of T. pallidum

Given the atypical presentation of primary syphilis with a single painful oral lesion, further testing to confirm T. pallidum infection was performed. DNA purified from two microtome sections of the FFPE tongue biopsy was used as a template for PCR amplification of three T. pallidum genes. TP0136, TP0548, and TP0705 were chosen for amplification and sequencing to allow for MLST analysis (9). The genes encoding T. pallidum 23S rRNA were also amplified and sequenced to examine for possible macrolide resistance.

All three T. pallidum genes amplified well, confirming that the spirochetes visualized in the biopsy by immunocytochemistry corresponded to T. pallidum. Sequencing of the PCR amplicons revealed that the DNA corresponded to T. pallidum strains of the Nichols cluster with the MLST type 9.16.3 (9, 19). Each of the three MLST genes had single nucleotide polymorphisms (SNVs) that differed from the canonical Nichols strain sequence (GenBank NC_021490.2). In the T. pallidum MLST database, a single strain identified as Philadelphia-2 had the same pattern; it had been given the sequence type (ST) designation ST-109 (20). In addition, the 23S rDNA amplified from the sample contained an A→G mutation at the position corresponding to A2058 in the Escherichia coli 23S rRNA gene that confers macrolide resistance (16, 17).

T. pallidum Genome Analysis

DNA obtained from two microtome sections of the FFPE biopsy sample was prepared in a manner that reduces the effects of formalin-induced crosslinking and cytosine deamination (7, 8) and then subjected to random Illumina sequencing. No enrichment of T. pallidum DNA was performed, so the vast majority of the reads represented human DNA sequences. Of the 319,467,467 total high-quality paired-end reads obtained, 616,104 reads (~0.19 %) were clearly identifiable as T. pallidum DNA, based on preliminary alignments with Nichols and SS14 genomic sequences. This number of reads was sufficient to permit the assembly of a genomic sequence for the newly identified strain with an average paired read coverage of 40.5-fold. The assembled genome contained 1,139,649 bp with an overall percentage (G+C) of 52.8%; these values are very close to those found for other T. pallidum strains, consistent with the high overall sequence identity and gene synteny found within this species. We observed only two regions of sequence ambiguity, a site with low read coverage at nt 329,123 (in gene QH729_01580) and the expected sequence variation observed in the variable regions within the gene encoding the antigenic variation protein TprK (21, 22). We named the infecting strain NYMC01 because of its initial diagnosis by investigators associated with the New York Medical College (NYMC).

Comparison of the genetic differences between the NYMC01 strain and the Nichols and SS14 strains revealed that NYMC01 was more closely related to the Nichols strain (Fig. 3, Table S1). Compared to the Nichols strain, the NYMC01 strain had 88 differences including changes within 66 genes, corresponding to differences in 6.3% of the 1041 open reading frames (ORFs) in the T. pallidum genome. These included insertions-deletions, homopolymeric tract length variations (HTLVs), complex (multi-nucleotide) substitutions, and single nucleotide variations (SNVs). The majority of the differences (69 of the 88) were SNVs, which resulted in amino acid changes in 60 genes (Fig. 3, Table S1). When the NYMC01 genome was compared to the SS14 strain (representative of the SS14 cluster of strains), 267 genome differences were observed (Fig. 3), consistent with the NYMC01 strain being more similar in sequence to the Nichols strain than to the SS14 strain.

Figure 3. Frequencies of genetic differences between the NYMC01 strain and the Nichols (A) and SS14 (B) strains.

(A) A total of 88 genetic differences were observed between genomes of NYMC01 and the Nichols strain. (B) The distribution of 267 differences between NYMC01 and SS14 strains is shown. (C and D) Relatively few differences were found between the NYMC01 genome and the sequences P-22–20198_RA and R-22–10139_RA, derived from pharyngeal and rectal swab specimens of a Seattle patient (18). (E) Only 4 differences were observed between the P-22–20198_RA and R-22–10139_RA. tprK sequence differences related to antigenic variation as well as indeterminant regions (N’s) were excluded from these analyses.

NYMC01 has a Near-Unique Genotype and is Part of a Novel Nichols Subgroup

To determine how common the NYMC01 genotype was, we performed nucleotide-based BLAST searches of each of the sequence differences between NYMC01 and the Nichols strain against the ~480 available T. pallidum genome sequences available in the NCBI database at the time of this study. The results are shown as values in parentheses in the column labeled “NYMC01 Sequence” in Table S1. Some of these differences were common in other T. pallidum strains, whereas others occurred in less than 50% of the strains. In the latter group, a subset of 21 genetic ‘markers’ were present in only two other available genome sequences, with the GenBank accession numbers CP124560 and CP124561. These genome sequences were derived from DNA from two swab specimens obtained in 2022 from the same human subject in the Seattle, Washington USA area; CP124560 (specimen P-22–20198) was from the throat, and CP124561 (specimen R-22–10139) was from the rectum (18). Of the 21 shared genotypic markers, 17 were SNVs, one was a single nucleotide insertion, and three have more complex genetic changes involving replacement of two or more nucleotides in a short region. The fact that these genotypic markers are present in NYMC01 and the two Seattle strains and are missing in the other available genome sequences indicates that these strains almost certainly represent a unique subgroup arising from a single common ancestor.

At the time of the submission of this article, the CP124560 and CP124561 sequences contained large regions of unresolved sequences (N’s) due to a masking process used to facilitate the processing of a large number of genome sequences (N.A.P. Lieberman and A.L. Greninger, personal communication). To provide a more complete comparison of these specimens with NYMC01, we obtained the Illumina reads used for these assemblies from the NCBI Short Read Archive (SRA) and performed our own assemblies using a previously described procedure (15). For clarity, these revised assemblies (RA) are referred to by their specimen numbers P-22–20198_RA and R-22–10139_RA rather than their corresponding GenBank numbers. Alignment of the three sequences verified that they were nearly identical to one another, with only 4 differences between P-22–20198_RA and R-22–10139_RA and 6 and 8 differences, respectively, between these specimens and NYMC01 (excluding N regions and tprK heterogeneities) (Fig. 3; Table S2). The high degree of genome identity among these three samples further verifies that they share a common origin.

Phylogenetic Analysis

To further analyze the relationship between NYMC01 and other T. pallidum strains, phylogenetic analysis was performed. The resulting Genome BLAST Distance Phylogeny (GBDP) showed that NYMC01, P-22–20198_RA, and R-22–10139_RA formed a distinct subgroup within the Nichols cluster (Fig. 4). This result provided further evidence that these three strains arose from a common ancestor within the T. pallidum Nichols cluster.

Figure 4. Phylogenetic analysis of NYMC01 and T. pallidum strains.

A Genome BLAST Distance Phylogeny (GBDP) tree was created using the Type Strain Genome Server (16, 17). The NYMC01 strain (marked with a red dot) clustered closely with P-22–20198_RA and R-22–10139_RA.

The sexually transmitted pathogen Haemophilus ducreyi is known to produce painful ulcers called chancroids (23). To rule out the possibility of H. ducreyi coinfection, an alignment of the FFPE biopsy Illumina reads with the H. ducreyi genome (CP015425.1) was performed. None of the read pairs aligned with the H. ducreyi genome, suggesting that the patient’s painful lesion phenotype was not caused by H. ducreyi infection. Likewise, alignments with the FFPE biopsy reads and the genomes of the type strains of the three common treponemes found in the human oral cavity were performed (24). None of the reads aligned with the genomes of T. denticola ATCC 35405, T. medium ATCC 700293, or T. vincentii F0403. Additionally, there was 0% alignment between the Illumina reads from the FFPE biopsy sample and representative strains of Human Herpesvirus I or Human Herpesvirus II, viruses that can cause painful sores in the human oral cavity (25). Together, these results suggest that the patient’s painful chancre was the result of T. pallidum infection and not the result of a co-infection with other pathogens commonly found in the human oral cavity.

DISCUSSION

Despite several published reports of painful syphilitic chancres, clinicians continue to rely on the classic painless presentation when diagnosing patients with syphilis. Larsen and Larsen (2) reported 12 patients with confirmed syphilitic chancres who presented with multiple painful herpetiform-like penile ulcers. Only one of those cases was diagnosed with primary syphilis at the first visit, and one had a coinfection with T. pallidum and HSV. Towns et al. reported a series of 183 men with T. pallidum PCR-positive primary anogenital lesions, with 49.2% reported as painful and 37.7% as multiple lesions. Only 2.7% of the anogenital lesions were associated with HSV and, of the 37 men with both painful and multiple primary lesions, only three had concurrent HSV (3).

Another sexually transmitted pathogen that presents with a painful lesion is H. ducreyi (23). In the currently reported case, high throughput sequencing of the FFPE sample did not detect any H. ducreyi sequences, confirming that this pathogen was not responsible for the painful lesion. Likewise, sequences from T. denticola, T. medium, and T. vincentii, three common species of pathogenic treponemes found in the human oral cavity (24), were not detected in the FFPE sample. Human Herpesvirus I and Human Herpesvirus II sequences were also not detected, suggesting that the patient’s painful tongue lesion was not the result of these common viruses (25).

In addition to pain, the patient’s lesion had other atypical features that delayed the diagnosis, including diffuse intense tongue inflammation with friable tissue that mimicked a neoplastic lesion. Especially in cases of oral syphilis, lesions may deviate from the classical presentation (26). There are case reports of chancres confused with neoplasms (6), which suggests that, while also excluding malignancy, providers should test for syphilis on patients with a non-resolving oral ulcer.

It could be argued that the diagnosis of the primary stage of syphilis cannot be made unequivocally in this case. Mucous patches of secondary syphilis are associated with pain, and there is no definitive test to distinguish between primary and secondary oral lesions. However, several lesion characteristics support the presence of a chancre in the case described in this report: a single, large ulcer with heaped borders, no patch present, a presumed short incubation period per the patient’s history, no signs of systemic symptoms, and no lesions outside of the mouth. The patient’s relatively low RPR titer of 1:8 did not preclude a diagnosis of secondary syphilis, but made it less likely.

FFPE tissue samples are challenging samples for DNA or RNA extraction, often resulting in low yields and decreased performance in the later processing steps. We used a commercially available kit that reduces crosslinking and removes deaminated cytosine residues using uracil-N-glycosylase (UNG) (7, 8). This process increases the amount of DNA available for sequence analysis and decreases sequencing artifacts caused by the deamination of cytosine to uracil, which results in C→T conversions during amplification and sequencing reactions. We are aware of only two other reports of T. pallidum genomic sequences obtained from FFPE tissues. The first was a recent study in which sufficient DNA was recovered from a FFPE fine needle aspirate of a cervical lymph node from a syphilis patient to provide a nearly complete genomic sequence (27). A second, as yet unpublished report, was regarding the genome sequence obtained from FFPE material from a congenital syphilis case that occurred in 1947 (GenBank Accession Number NZ_CP115658.1). This sequence assembly had relatively low coverage (13.99X) and 118 gaps; the longest contiguous sequence was 23,447 bp. This incomplete coverage may have been due to the age of the specimen. In comparison, our genome assembly had 40.5X coverage and only one gap that could not be resolved. The tprK region of the NYMC01 sequence was ambiguous due to the expected sequence heterogeneity resulting from antigenic variation (21, 22).

MLST analysis using three target genes (9) demonstrated that NYMC01 had the sequence type ST-109, in which all three genes had SNVs that differed from those of the Nichols strain. Philadelphia-2 (USL-Phil_2, isolated in 1987) was the only other strain with ST-109 described in the T. pallidum MLST database (https://pubmlst.org/organisms/treponema-pallidum, accessed 5 January, 2024) (9, 19). No clinical information or genome sequence is available for Philadelphia-2; therefore, at this time we cannot link this strain with an atypical clinical presentation.

Whole genome sequencing further demonstrated the presence of 88 genome differences with the Nichols strain and 279 differences with the SS14 strain, respectively. In comparison, there were 20 SNVs between the Chicago and Nichols genomes, which are also members of the Nichols cluster (28). Although the NYMC01 strain has many differences from the Nichols strain (Table S1; Figs. 3 and 4), it is still clearly a member of the Nichols group. These data suggest that the NYMC01 strain has diverged significantly from other Nichols cluster strains and raises the question of whether these genotypic differences may be related to the unusual, painful chancre phenotype in the patient described in this study.

We were surprised to find that the NYMC01 genome contained 23 genetic differences not found in most other T. pallidum strains. Two of these (in tprD) were unique to NYMC01, and 21 were shared with two specimens from Seattle, Washington, U.S.A. (Table S1). These two Seattle swab specimens (P-22–20198 [pharyngeal] and R-22–10139 [rectal]) represent a rare example of T. pallidum genomic sequencing from two different sites in the same individual. They were part of a larger epidemiological study by Lieberman et al. (18) involving 24 individuals attending the Seattle Sexual Health Clinic in 2021–2022. The draft sequences from these samples were augmented by our own revised assemblies using the Illumina reads provided by the Seattle group. The combination of the unique genotype of this subgroup and the close identity of the genomic sequences from NYMC01, P-22–20198, and R-22–10139 strongly indicates that they all arose from a single clonal population. The fact that the Philadelphia-2 strain shares an MLST type with these specimens suggests that the subgroup may have been in the United States as early as 1987. It is anticipated that additional T. pallidum genome sequences with the NYMC01-related genotype will be identified, which will provide insights into the geographical distribution and spread of this subgroup. Interestingly, the presence of painful lesions (or syphilitic lesions of any type) was not noted in the medical record of the individual from whom the P-22–20198 and R-22–10139 specimens were obtained (M. Golden, N. A. P. Lieberman, and A. L. Greninger, personal communication). Additional cases with T. pallidum genome information are needed to resolve the question of whether this genotype is associated with specific pathological patterns.

A number of NYMC01-related genetic differences are predicted to change the amino acid sequences of encoded proteins and hence may affect protein function and, potentially, human pathogenesis (28) (Table S1). T. pallidum is a highly motile organism that disseminates rapidly (29). Changes in motility and chemotaxis could potentially result in reduced or abnormal patterns of pathogenesis, as has been observed in the spirochetes Leptospira interrogans and Borrelia burgdorferi (30, 31). In the NYMC01 strain, predicted single amino acid changes or truncations were present in genes encoding the flagellar motor switch protein (FliM), the flagellar hook length control protein (FliK), methyl-accepting chemotaxis proteins (Mcp’s) important in sensing chemoattractants and repellents, and the chemotaxis methyltransferase CheR (Table S1). Sequence differences were also found in genes encoding proteins with a variety of cellular activities such as the V-type ATP synthase, a PBP1 family protein involved in cell wall synthesis, members of the Tpr major sheath protein (MSP) porin family, and even the 50S ribosomal subunit protein RplA and the sigma factor RpoD. Currently, the effects of these differences may have on T. pallidum cellular processes and pathogenesis are unknown.

Syphilis is known as the great mimicker, since its presentation can take many forms and has challenged clinicians for centuries. Given the ongoing syphilis epidemic in the USA, clinicians should learn about atypical presentations such as the recently reported cases of painful lesions. Whether such painful lesions are associated with certain T. pallidum genotypes remains to be determined. Therefore, the first genomic sequence of a T. pallidum strain recovered from a patient with a painful oral syphilitic lesion is important for future comparative analyses. Additional studies characterizing T. pallidum genomic information in the context of associated clinical manifestations may provide insight into the potential relationship between spirochetal genetic differences and the clinical course of syphilis.

Supplementary Material

Table S2

Table S2. Limited number of nucleotide differences between T. pallidum subsp. pallidum strains NYMC01, P-22–20198, and R-22–10139.

Table S1

Table S1. Nucleotide polymorphisms identified in T. pallidum subsp. pallidum NYMC01.