Longitudinal accumulation of in vivo and in vitro-grown Treponema pallidum subsp. pallidum TprK variants in the presence and absence of immune pressure

Immune evasion by Treponema pallidum subspecies pallidum (T. pallidum) has been attributed to antigenic variation of its putative outer-membrane protein TprK. In TprK, amino acid diversity is confined to seven variable (V) regions, and generation of sequence diversity within the V regions occurs via a non-reciprocal segmental gene conversion mechanism where donor cassettes recombine into the tprK expression site. Although previous studies have shown the significant role of immune selection in driving accumulation of TprK variants, the contribution of baseline gene conversion activity to variant diversity is less clear. Here, combining longitudinal tprK deep sequencing of near clonal Chicago C from immunocompetent and immunosuppressed rabbits along with the newly developed in vitro cultivation system for T. pallidum, we directly characterized TprK alleles in the presence and absence of immune selection. Our data confirm significantly greater sequence diversity over time within the V6 region during syphilis infection in immunocompetent rabbits compared to immunosuppressed rabbits, consistent with previous studies on the role of TprK in evasion of the host immune response. Compared to strains grown in immunocompetent rabbits, strains passaged in vitro displayed low level changes in allele frequencies of TprK variable region sequences similar to that of strains passaged in immunosuppressed rabbits. Notably, we found significantly increased rates of V6 allele generation relative to other variable regions in in vitro cultivated T, pallidum strains, illustrating that the diversity within these hypervariable regions occurs in the complete absence of immune selection. Together, our results demonstrate antigenic variation in T. pallidum can be studied in vitro and occurs even in the complete absence of immune pressure, allowing the T. pallidum population to continuously evade the immune system of the infected host. Author Summary Syphilis continues to be a disease of global and public health concern, even though the infection can be easily diagnosed and effectively treated with penicillin. Although infected individuals often develop a strong immunity to the pathogen, repeated infection with the syphilis agent, Treponema pallidum subspecies pallidum (T. pallidum), is possible. Several studies point at antigenic variation of the T. pallidum TprK protein as the mechanism responsible for evasion of the immunity that develops during infection, pathogen persistence, and re-infection. Past studies have highlighted the importance of immune clearance of dominant variants that, in turn, allows less represented variants to emerge. The contribution of an immunity-independent baseline generation of variability in the tprK gene is less clear. Here, we used deep sequencing to profile tprK variants using a laboratory-isolated T. pallidum strain nearly isogenic for tprK that was propagated over time in vitro, where no immune pressure is exerted on the pathogen, as well as in samples obtained from immunosuppressed and immunocompetent rabbits infected with the same strain. We confirmed that tprK accumulates significantly more diversity under immune pressure, and demonstrated a low but discernible basal rate of gene conversion in complete absence of immune pressure.


Introduction
achieved by weekly intramuscular (IM) injections of 20 mg methylprednisolone acetate (Sicor,134 Irvine, CA). Treatment was started 3 days before experimental infection, and doses were 135 administrated weekly for a total of six injections. A lesion biopsy (4-mm punch biopsies taken 136 under local lidocaine anesthesia) was harvested from each rabbit weekly for a period of 5 137 weeks; these samples were used for DNA extraction and tprK profiling in the present study as 138 shown in Fig 1A schematic. passaged in vitro at Week -1 and again at Week 0, at which point treponemes were 1) 148 inoculated intradermally into a control rabbit, and 2) seeded into two sets of weekly-passaged 149 cultured samples for 7 weeks. Samples were collected weekly from punch biopsies of the rabbit 150 and from cell culture, then analyzed for treponemal load and tprK sequence. 151 152 153

Parallel in vivo and in vitro cultivation of T. pallidum 154
One day prior to in vitro inoculation of the Chicago C strain of T. pallidum, 10 5 rabbit Sf1 155 epithelial cells were plated into each well of a 6-well plate and incubated at 37ºC overnight in a 156 5% CO2 atmosphere within a HeraCell 150 incubator (Thermo Fisher Scientific, Waltham, MA). 157 At the same time, treponemal culture medium (TpCM2) was prepared according to the 158 Edmondson et al. protocol [25], and incubated overnight at 34ºC in a microaerophilic 159 atmosphere (1.5% O 2 , 5% CO 2 and 93.5% N 2 ) to deplete the medium of oxygen. 160 The subsequent day, Sf1 epithelial cells were rinsed with equilibrated TpCM2 medium 161 and then equilibrated in 5 mL of media in the microaerophilic incubator for three hours. A frozen 162 glycerol stock of T. pallidum strain Chicago C was thawed at room temperature and bacteria 163 were enumerated on a Leica DM2500 darkfield microscope (Leica, Wetzlar, Germany) to 164 inoculate 10 6 bacteria into each well of the 6-well plate. The plate was then returned to the 165 microaerophilic incubator for seven days. Residual inoculum bacteria were collected via 166 centrifugation and resuspended in 100 μ L of buffer ATL (Qiagen, Hilden, Germany) for DNA 167 extraction. After seven days, the culture was retrieved, and supernatants were discarded. 168 Treponemes were then released from the epithelial cell layer via trypsin digestion, enumerated 169 as described above and passaged ( Figure 1B). Starting with the week 0 culture, inocula were 170 increased to 10 7 cells per well and the bacteria were seeded into two 6-well culture dishes. At every passage starting from Week 0, cultured treponemes were harvested from matching wells 172 from both plates and pelleted via centrifugation. The remaining wells were used to seed two 173 additional plates. In parallel, the same treponemes in Week 0 were used to infect a New 174 Zealand white rabbit. Prior to rabbit infection, treponemes were enumerated once more on a 175 Nikon OptiPhot2 darkfield microscope (Nikon, Tokyo, Japan) to ensure viability and correct 176 density. After verifying viability, 10 6 bacteria per site were administered intradermally at eight 177 sites on the shaved back of the rabbit. The rabbit was monitored briefly and then returned to its 178 cage. After this point, infected rabbits were shaved daily to allow monitoring of lesion 179 development and facilitate weekly biopsy collection. Starting a week after infection (Week 1), a 180 single lesion developing on the back of the rabbit was harvested using a 4-mm dermal biopsy 181 punch. Tissue biopsies harvested from week 1 to week 7 were minced using sterile scalpels 182 and further homogenized into 400 primers (S1 Table)  week. PCR products were run on a 1% agarose gel to confirm the presence of a band at 219 approximately 1.6 kb and cleaned with 0.6x Ampure beads (Beckman Coulter). Two nanograms 220 of PCR product was used in a two-fifths Nextera (Illumina) library prep with 15 cycles of 221 amplification. Low molecular weight products were removed with 0.6x Ampure beads, yielding final libraries with a minimum size of approximately 300 bp. Libraries were quantified, pooled, 223 and run on a 1x192 single end run on an Illumina Miseq. A second replicate of each sample was 224 prepared as described above to control for potential polymerase error. Samples yielding a 225 minimum of 1000 mean coverage mapping to tprK were kept for downstream analysis, with raw 226 reads ranging from 15461 -368148 reads. 227 228 Sequencing analysis 229 tprK reads were analyzed with custom python and R scripts. Briefly, raw reads were 230 adapter-trimmed and filtered for quality scores above Q20 using trimmomatic v0.39 [26]. internal repeats [13]. We used blastn to extract exact matches greater than or equal to 5 241 nucleotides in length within the 12.5-kb locus containing the tprD gene, using only matches with 242 at least 10 reads of support in any given sample. We then used these segments to recursively 243 build on each other to recreate each unique tprK allele, allowing exact or overlapping 1-2 244 internal repeats between segments. We allowed for up to 5% of no coverage in each variant in 245 order to account for the possibility of segments that are <4 bp long -below the minimum length 246 requirement for blastn -and for differences in our variable region boundary definitions 247 compared to what was seen when tprK anatomy was previously characterized [13,15,28]. 248

Data Availability 250
Reads from tprK sequencing of the samples used in this study are available under the 251 NCBI BioProject number PRJNA734645. All code used for analysis is publicly available on 252 GitHub (https://github.com/greninger-lab/longitudinal_tprk).

Results 254
Longitudinal analysis of tprK in multiple model systems 255 We profiled longitudinal evolution of tprK in treponemes obtained from multiple model 256 systems over two experiments. In the first experiment, longitudinal samples were obtained from 257 skin lesions of five pharmacologically immunosuppressed rabbits and five immunocompetent 258 rabbits infected with the tprK-isogenic Chicago C strain (Fig 1A). In the second experiment, we 259 propagated the Chicago C strain in culture and also infected one additional control animal, 260 harvesting samples from each source every week for seven weeks after initial inoculation (Fig  261   1B). 262 The quantity of T. pallidum in each sample at the time of harvest was determined using 263 quantitative PCR targeting the T. pallidum tp0574 gene. A median tp0574/CFTR copy number 264 ratio of 0.64 was measured in the immunosuppressed vs. immunocompetent rabbit experiment, 265 while the median copy number ratio was 6.75 the in vivo vs. in vitro experiment, mostly due to 266 treponemal enrichment in the in vitro culture. As expected, there was a noticeable but non-267 significant decrease in copy number over time in immunocompetent rabbits compared to 268 immunosuppressed rabbits (Fig 2A, S1 Fig)  tp0574/CFTR copy number ratio. In contrast, at week 7 of the second experiment, all samples 273 harvested from the immunocompetent rabbit had greater than 3.0 tp47/CFTR copy number ratio 274 ( Fig 2B). 275 We then amplified the tprK gene and performed next-generation sequencing on samples 276 with recoverable tprK PCR product. Sequencing data for all seven variable regions at varying 277 timepoints were successfully obtained for 23 immunosuppressed rabbit samples and 19 278 immunocompetent rabbit samples (Fig 2A), as well as 8 samples from the control rabbit, 8 samples each from the two replicate culture sets, and 2 early passage culture specimens 280 (labeled "Inoculum", Fig 2B).  Table). Similarly, across the 10-week period of the in vivo vs. preparations to call high-confidence sequences at depths that allowed better capture of the true 316 diversity of tprK variants. An additional metric, Pielou's evenness, was used to quantify diversity 317 as the extent of equitability in the distribution of observed variants in each variable region. We 318 confirmed a significant difference in Pielou's evenness score in V6 of immunocompetent rabbits 319 versus immunosuppressed rabbits beginning at Week 3 (p<0.0001) (Fig 3A, S2 Table), but we did not observe any significant difference in mean diversity between these two groups of rabbits 321 in any other variable region. 322 We then assessed the longitudinal TprK evolution in the novel culture system by 323 comparing tprK sequences from samples passaged in a single immunocompetent rabbit and 324 samples passaged in vitro. Similar to the results from immunocompetent vs. immunosuppressed 325 rabbits, we observed a 1.7-fold increase in Pielou's evenness score of V6 in samples passaged 326 in vivo compared to in vitro starting at week four post-infection (Fig 3B). Using an unpaired two-327 sample t-test, we also found a significantly greater number of unique tprK V6 alleles in the 328 control rabbit than cultured samples (p=0.015). In contrast to the previous experiment, using 329 deep sequencing, we additionally observed a significant accumulation of variable region 330 diversity over time in V2, V3, V4, and V7 in the control rabbit compared to in vitro samples. 331 Overall, these data demonstrate a significant increase of TprK variants, especially in V6, in 332 immunocompetent rabbits compared to both immunosuppressed rabbits and in vitro culture, 333 confirming that immune selection plays a significant role in the appearance of TprK variants. We further investigated accumulation of V region alleles in immunosuppressed rabbits 345 and in culture, where variants were generated at similar rates (Fig 4, Table 1). Interestingly, in 346 both types of passaging, V6 generated significantly more unique variants over time than in the 347 other variable regions (p<0.0001), suggesting that greater V6 accumulation of new variants is 348 its in he is not solely a function of immune selection (Table 1). To estimate the baseline rate of sequence 349 change, we examined the weekly decrease of the most common sequence for each variable 350 region because of the extremely low frequency levels of the many newly generated sequences. 351 We observed a mean weekly decrease of 2.40% relative frequency in the most highly abundant 352 sequence in V6 in the immunosuppressed rabbits compared to a mean weekly decrease 353 ranging between 0.37% and 1.04% for the other variable regions (S3 Table). In the cultured 354 treponemes, we similarly observed a mean weekly decrease of 2.25% in the most highly 355 abundant sequence in V6 compared to a mean weekly decrease ranging between 0.  variable region is compared to V1 for both experiments using T-Tests following  Roger's method [29]. Significance is denoted with asterisks (0***, 0.001**, 0.01*). 369 370 To better understand how these variants are being generated over time, we compared 371 the generation of V region sequences between our two sets of samples propagated in culture. In 372 a purely stochastic model, we would expect to see emergence of several sequences present in 373 one set that are absent in the other. However, we observed that the relative frequencies of tprK 374 alleles were remarkably reproducible across our two culture replicates for all variable regions (r 2 375 = 0.9998) (Fig 5A). We did not observe any variable region sequences above 0.62% relative 376 frequency present in one culture replicate that was not present in the other (Fig 5B). In addition, 377 there were a total of 128 unique alleles over all variable regions generated in culture that were 378 not present in the inoculum above a relative frequency of 0.1% in both library preparations.
These novel alleles stayed at low frequencies, reaching a maximum of 0.79% relative frequency 380 ( Fig 5B). Predictably, at high relative frequencies (≥70%), the clonal sequence decreased in 381 frequency from a mean of 95.46% to 90.67% across all variable regions over time (Fig 5C) as 382 lower-level V region alleles increased in number. At lower relative frequencies (0.5 -4%), the 383 pattern became less clear, though over time relative frequencies of alleles in both culture 384 replicates trended towards closer shared identity (Fig 5D and Fig 6), and often increased in 385 frequency together. conclusions. In addition, many sequences, including the V1 clonal sequence in vitro, are unable 526 to be constructed with the previously described internal repeats [13]. More investigation on tprK 527 anatomy and potentially addition of new internal repeats is needed to account for these 528 sequences. Understanding how these new sequences are being generated would better inform 529 Gilead, unrelated to the work performed here.