Outbred male specific pathogen-free (SPF) New Zealand White rabbits (3.0–3.5 kg, Charles River Laboratories, Ontario, Canada) with nonreactive syphilis serological tests (VDRL and FTA-ABS) were used for in vivo propagation of T. pallidum subsp. pallidum Nichols strain as previously described (Lukehart and Marra, 2007). All rabbits were fed antibiotic-free food and water, and were housed at 18–20 °C. Animal studies were approved by the local institutional review board under protocol 2020–024 and were conducted in strict accordance with standard accepted principles as set forth by the Canadian Council on Animal Care (CCAC), National Institutes of Health, and the United States Department of Agriculture in facilities accredited by the American Association for the Accreditation of Laboratory Animal Care and the CCAC. Institutional biosafety approval was obtained under biosafety certificate 13,170–010. Nichols strain T. pallidum was cultured in vitro with Sf1Ep cells as previously described (Edmondson et al., 2018) in a HeraCell Vios 160i incubator (Thermo Fisher Scientific, San Jose, CA), with the modification that treponemes were dissociated from Sf1Ep cells using trypsin-free dissociation media (Edmondson and Norris, 2021) for 30 min at 34 °C in 1.5% O₂, 5% CO₂, 93.5% N₂ to maintain the integrity of the T. pallidum outer membrane (Edmondson et al., 2018). Bacteria were quantitated by darkfield microscopy (Nikon Eclipse E600; Nikon Canada, Mississauga, Ontario) using a Petroff-Hauser counting chamber (Hauser Scientific, Horsham, PA).

In vivo grown T. pallidum was used for RNA-seq assays, and in vitro grown T. pallidum was used for western blot and microscopy assays. In vivo Treponema pallidum subsp. pallidum was extracted in Treponema pallidum culture medium 2 (TpCM2) formulated as previously described (Edmondson et al., 2018), or harvested in vitro as described above. Viable T. pallidum (VTP) and infection extract control (IEC) samples were prepared as described previously (Waugh et al., 2023). Briefly, extracts were centrifuged to remove rabbit testicular material and diluted equivalent to working T. pallidum concentrations. The viable treponeme suspension contained in one supernatant [designated viable T. pallidum (VTP)] was kept for endothelial co-incubation analyses as described below. The second supernatant was further processed to remove T. pallidum through sterile filtration with a 0.22 μm cellulose acetate syringe filter (Avantor, Allentown, PA) to create an optimal comparator control (designated infection extract control [IEC]) that contained a level of rabbit testicular or Sf1Ep protein background mirroring the VTP sample. Removal of T. pallidum from the IEC sample was confirmed by darkfield microscopy (Nikon Eclipse E600; Nikon Canada, Mississauga, Ontario) and T. pallidum-specific qPCR, which measured a 49x reduction in T. pallidum FlaA copies in the VTP versus IEC samples.

Human cerebral brain microvascular endothelial cells (hCMEC/d3; Cedarlane, Burlington, ON), an immortalized cell line, also referred to as HBMECs, were grown to 90% confluence in 6-well tissue culture plates (Corning, Corning, NY), at 5% CO₂ in EndoGRO-MV complete culture medium (Millipore, Etobicoke, ON) at 37 °C in 5% CO₂ in a Forma Series II incubator (Thermo Fisher Scientific). Each well was observed via an Olympus CKX41 inverted microscope for successful HBMEC growth prior to incubation at 34 °C in a microaerophilic environment of 1.5% O₂, 5% CO₂, 93.5% N₂ with 3 mL of either in vivo grown VTP (3.0 × 10⁷ T. pallidum per well), or equivalent dilutions of IEC for 45 min, 4, 12, and 24 h. At each timepoint, there were 5 replicates of VTP- and 5 replicates of IEC-exposed HBMEC sample wells. Therefore, individual wells are defined as separate biological replicates. At each timepoint prior to cell lysis and RNA harvesting, T. pallidum organisms were visually observed for motility via darkfield microscopy (Nikon Eclipse E600).

Following treponemal viability assessment, HBMEC cells were quickly washed 3x in cold PBS and incubated in a 1:1 ratio of Qiagen RNAprotect Cell and Qiagen RNAprotect Bacteria (Qiagen, Toronto, ON) for 10 min on ice to inactivate RNases and stabilize RNA. Cells were pelleted at 10,000×g for 10 min, resuspended in 1 mL of fresh RNAprotect cell, stored at −80 °C for downstream processing. RNA was extracted using the Qiagen RNeasy kit according to manufacturer instructions, with the addition of an on-column DNA digestion using RNase-Free DNase (Qiagen). To prevent RNA degradation, 1 μL of SUPERase-In RNase inhibitor was added to the eluted RNA (Qiagen).

Prior to cDNA library generation, RNA integrity of each sample was assessed by an Agilent Bioanalyzer 2,100 on an RNA 6000 nanochip (Agilent; Santa Clara, CA), where all RNA samples exceeded an RNA integrity value of 8. Samples then underwent PolyA enrichment using NEBNext Poly(A) mRNA Magnetic Isolation Module (catalog no.: E7409L, NEB; Ipswich, MA). Strand-specific cDNA library preparation was generated at the same time for each sample from polyA-purified RNA with a Roche KAPA HyperPrep Kit (Roche; Basel, Switzerland), followed by the addition of unique 8 bp NGS RNA adaptors to identify samples during multiplexed sequencing (NEXTFLEX; PerkinElmer, Woodbridge, ON), generating cDNA with unique tags for each individual sample. cDNA libraries were amplified using adapter primers, followed by purification beads according to manufacturer instructions (NEXTFLEX; PerkinElmer). Amplified DNA quality was assessed on a bioanalyzer using a high sensitivity DNA chip (Agilent) to assess the fragmentation profile and confirm the absence of adaptor-specific primer dimers. cDNA library concentrations were assessed using Qubit QuantIT HS DNA kit (Thermo Fisher Scientific). Each sample was normalized to 4 nM and pooled. Pooled samples underwent paired-end sequencing with a read length of 150 bp on an Illumina HiSeqX (Illumina; San Diego, CA, USA) at the Michael Smith Genome Sciences Center (BC, Canada). Sequence quality was assessed using FastQC v0.12.1 and MultiQC v1.13 (Ewels et al., 2016). The FASTQ sequence reads were aligned to the hg19 human genome (Ensembl GRCh38.98) using STAR v2.7.10b (Dobin et al., 2013) and mapped to Ensembl GRCh38 transcripts. Read-counts were generated using htseq-count (HTSeq 2.0.2) (Anders et al., 2015). All data processing and subsequent differential gene expression analyses were performed using R version 4.4.0 and DESeq2 (Love et al., 2014) version 1.14.1. Genes with very low counts (with less than 10 counts) were pre-filtered and removed in silico. Differentially expressed genes were identified using paired analysis with the Wald statistics test and filtering for any genes that showed ±1.5-fold-change (FC) with adjusted p-values ≤ 0.05 (cut-off at 5% false discovery rate) as the threshold. Pathway overrepresentation analyses were completed using the Reactome database (Gillespie et al., 2022) annotations and Sigora (v3.1.1) (Foroushani et al., 2013) pathway overrepresentation, and the molecular signature database (Liberzon et al., 2015) (MSigDB) hallmark gene set analysis was performed using PathlinkR (Blimkie et al., 2024). The packages maSigPro (v1.74.0) (Nueda et al., 2014), and the ClusterProfiler R package (v4.10.0) (Yu et al., 2012) were used for figure generation. For Sigora and MSigDB analyses, q-values were set at 0.05 using the Benjamini-Hochberg (BH) correction, and pathways or cellular compartments with a corrected p-value ≤ 0.05 were considered significant. Transcription factor analysis was completed using the Chea3 integrated mean ranking algorithm (Keenan et al., 2019) for each timepoint.

Confluent HBMECs in 6-well plates were exposed to in vitro grown VTP (3×10⁷ T. pallidum per well), equally diluted IEC, or basal TpCM2 media for 12 and 24 h with 3 biological replicates (defined as individual wells) per treatment condition. Cells were lysed for 30 min on ice with gentle agitation in RIPA lysis buffer containing EDTA-free Protease inhibitor cocktail set 3 (Calbiochem, San Diego, CA) and PhosSTOP phosphatase inhibitor (Roche, Mississauga, ON) according to the manufacturer’s instructions. For western blotting, protein concentrations were determined using a BCA assay (Thermo Fisher Scientific). Whole cell lysate (13.5 μg per lane) was subjected to SDS-PAGE using Bolt 12% acrylamide gels (Thermo Fisher Scientific) and transferred to PVDF membrane (Millipore) via wet-transfer at 400 mA for 2 h, and blocked with Intercept TBS blocking buffer (Licor, Lincoln, NE). Rabbit Anti-Snail (CD15D3; 1:1000; Cell Signalling Technology, Danvers, MA; 3879S) and mouse anti-GAPDH (1:1000; Abcam, Ab8245) were used as primary antibodies, while the secondary antibodies used were goat anti-rabbit IgG IRDye 800CW (1:20,000) and goat anti-mouse IgG IR Dye 680RD (1:20,000; Licor). Detection and analysis were completed on a Licor Odyssey CLx using Licor Image Studio version 5.2. Data was tested for normality using a Shapiro-Wilks tests which determined that the data was consistent with normal distribution, and statistical analysis to determine increases in Snail abundance was completed using One-way ANOVA followed by Tukey’s multiple comparisons testing.

hCMEC/d3 endothelial cells were seeded into black wall μClear cell culture-treated bottom 96-well culture plates (Greiner Bio-one, Monroe, NC; 655090) at 24,000 cells/well and grown overnight as described above to achieve 90–95% confluence. In vitro cultivated T. pallidum was cultured as described above, then VTP and IEC samples were generated as described above and diluted to 4.34 × 10⁶ cells/mL in TpCM2 media, or equivalent dilution for IEC. For T. pallidum-HBMEC co-incubation, HBMEC media was removed, then replaced with 100 μL of VTP, IEC, basal TpCM2 media, or 10 nM bovine thrombin in TpCM2 (Thermo Fisher Scientific; RP-43104) in triplicate per condition, resulting in a multiplicity of infection (MOI) of 30 for VTP samples or equivalent dilution for IEC. Incubations were completed for 15 min, 30 min, or 2 h after the addition of T. pallidum, and all incubations were performed at 34 °C in a microaerophilic environment of 1.5% O₂, 5% CO₂, 93.5% N₂ in a HERAcell vios 160i incubator (Thermo Fisher Scientific). At the specified timepoints, T. pallidum-containing supernatants were removed and assessed for motility via darkfield microscopy, where T. pallidum motility remained greater than 80% for all timepoints. Simultaneously, HBMECs were washed 2x in pre-warmed PBS, fixed in 3.7% formaldehyde in PBS without methanol for 10 min at RT, and washed 3x in pre-warmed PBS. Cells were permeabilized in 0.1% Triton 100-X for 5 min, washed 3x in pre-warmed PBS, and blocked in PBS 1% bovine serum albumin (Thermo Fisher Scientific) for 1 h at 37 °C. Alexa fluor 488 Phalloidin (Thermo Fisher Scientific; A12379) stocks were resuspended in DMSO, then diluted to a 1x working solution in PBS with 3 μM DAPI counterstain (4′,6-diamidino-2-phenylindole; Sigma Aldrich). Fifty microliters of staining solution was added to each well, the plates were incubated at RT for 20 min, washed 3x in warm PBS, and 100 μL PBS was added to each well for imaging.

All cell imaging was performed on a Cytation 5 Imaging Reader (Agilent) with a 20x objective. For statistical analyses wells were imaged in the centre of each well in a 3 × 3 field of view square. Images underwent image pre-processing and image deconvolution, after which they were stitched into a single image using linear blending in Gen5 (Agilent; v3.12). All imaging was automated and completed at the same time, and with identical imaging conditions. HBMECs were identified (defined as “objects”) and were counted using the DAPI filter as a primary mask (7–35 μm). To determine HBMEC cellular boundaries, a secondary mask was set by expanding the primary DAPI mask (110 μm) in the GFP (FITC; F-actin) channel using the threshold in mask method (FITC; F-actin); touching objects were split, objects contacting the border of the image were excluded, and gaps in masks were filled. Using the secondary mask, F-actin fluorescence intensity was determined within the boundaries of each object, the fluorescence intensity was calculated for each defined object, and the average object fluorescence intensity was calculated for each well. Significant increases in fluorescence intensity were determined by comparing mean object fluorescence intensity from 3 biological replicates (defined as individual wells) per sample condition. Data was tested for normality using a Shapiro-Wilks tests which determined that the data was consistent with normal distribution, and significance was determined using One-way ANOVA followed by Dunnett’s multiple comparison, using HBMECs exposed to basal media as the baseline for statistical comparison.

To identify HBMEC genes that were differentially expressed (DE) during T. pallidum exposure, RNA-Seq was conducted on five independent wells per treatment condition comparing a viable T. pallidum (VTP) exposure to an infection extract control (IEC) exposure at each timepoint. Briefly, these conditions corresponded to exposure to a VTP sample containing viable T. pallidum (viability confirmed by visual assessment of motile treponemes via darkfield microscopy), alongside exposure to an IEC sample where T. pallidum was removed using a centrifugation and filtration technique that retains background culture contaminants co-purified during T. pallidum harvest (Waugh et al., 2023). Exposing HBMECs to T. pallidum significantly altered gene expression in the brain endothelial cells, where the number of DE genes with adjusted p-value of < 0.05 and a fold-change (FC) cutoff of ±1.5 were 11 at 45-min, 225 at 4-h, 490 at 12-h, and 1,753 at 24-h (Figures 2A–D; Supplementary Table 1). Considerable overlap in DE genes was observed between timepoints, where 311 genes were differentially expressed at both 12 and 24 h, 52 genes were differentially expressed from 4–24 h, and 4 genes were differentially expressed at all timepoints (Figure 2E).

Next, we completed Transcription factor (TF) enrichment and co-regulatory analyses (Keenan et al., 2019) on each of the study timepoints to identify TFs that may be responsible for the observed changes in gene expression in HBMECs during T. pallidum exposure. Using an integrated mean ranking algorithm (Keenan et al., 2019), local TF co-regulatory networks were generated for the top 10 TFs predicted to be most influential in regulating the expression of the DE genes at each timepoint (Table 1; Figures 3A–D; Supplementary Tables 2–5). These analyses identified highly connected co-regulatory TF networks at each timepoint, all of which featured AP-1 (Activating Protein-1) subunits such as Fos, FosB, and Jun (Table 1; Figures 3A–D). Notably, the transcripts for FOS, FOSB, JUN, JUNB, and JUND were upregulated at one or more timepoints in this study (Table 1; Supplementary Table 1). NR4A1 (Nuclear Receptor subfamily 4 group A member 1) and NR4A3 were also enriched at the 12- and 24-h timepoints, and transcripts encoding these TFs were significantly upregulated at all timepoints in the study (Table 1). Transcription factors involved in the cellular differentiation processes of EndMT and Epithelial to mesenchymal transition (EMT) were also enriched at all timepoints, including HEY1 (Hes-related family bHLH transcription factor with YRPW motif 1), HEY2, BHLHE40 (Basic Helix–Loop–Helix Family Member e40), EGR1 (Early Growth Response 1), and EGR2 (Table 1). Snail, the primary TF driving EndMT (Kokudo et al., 2008; Derada Troletti et al., 2019; Ma et al., 2021) ranked among the top 10 most enriched TFs for all timepoints (Supplementary Tables 2–5). Significantly increased SNAI1 transcripts were observed in VTP-exposed HBMECs from 12- to 24-h (3.08, 3.11 FC, respectively), contrasting with minimal alteration of SNAI1 transcripts in IEC-exposed HBMECs over the same period (Table 1; Figure 4A). Western blot validation confirmed T. pallidum exposure induced Snail protein upregulation, with significantly higher protein abundance at 24 h and increased protein abundance at 12 h (VTP-versus IEC-exposed cells; Figures 4B,C). Collectively, these findings identify the early and sustained alteration of EndMT-associated TFs and their influence on HBMEC transcriptional responses to T. pallidum exposure.

We next performed Sigora pathway overrepresentation analysis (Foroushani et al., 2013) on up- and down-regulated genes at each timepoint to investigate pathways that are significantly overrepresented in T. pallidum-exposed HBMECs. Sigora uses Reactome database annotations (Gillespie et al., 2022) and contextualizes pathway overrepresentation by weighting DE gene pairs that occur uniquely in a single pathway. The number of overrepresented pathways increased with time, with 1, 15, 19, and 46 significantly overrepresented pathways at 45-min, 4-, 12-, and 24-h, respectively. These investigations identified overrepresented cellular pathways within the categories of gene expression, extracellular matrix (ECM) organization, immune system, programmed cell death, disease, and signal transduction, and key pathways from these categories are highlighted below (Figures 5A–D; Supplementary Tables 6–9).

Within the category of gene expression, we observed overrepresentation of pathways related to p53-mediated cell death and p53-mediated transcriptional regulation at the 4- to 24-h timepoints (Figures 5B–D; Supplementary Tables 7–9). At the 12-h timepoint, pathways related to EndMT were upregulated, such as “SMAD2/SMAD3: SMAD4 transcriptional regulation” and “regulation of PTEN gene expression” (Figure 5B; Supplementary Table 7). We also observed dysregulation of pathways involved in ECM organization, with “integrin signaling” first downregulated at 4-h, followed by upregulation of additional ECM pathways (“non-integrin membrane-ECM interactions,” “molecules associated with elastic fibres,” and “signaling by receptor tyrosine kinases”) at 24-h (Figures 4D, 5B; Supplementary Tables 7, 9). Within the category of immunity, we observed downregulation of pathways in “chemokine receptors bind chemokines” and TNF and IFN signaling (4- and 12-h), suggesting inflammatory pathway downregulation (Figures 5B,C). While “chemokine receptor signaling” remained downregulated at 24-h, pathways involving TNF and IFN signaling were upregulated, indicating differential temporal regulation of these immune pathways in HBMECs exposed to T. pallidum (Figure 5D).

Within the category of signal transduction, we observed the pathway “negative regulation of MAPK (Mitogen-Activated Protein Kinase)” to be upregulated at 4- and 12-h, and “RAF/MAP kinase cascade” to be downregulated at 12-h, indicating the involvement of MAPK signaling in the endothelial response to T. pallidum. At the 24-h timepoint, overrepresented pathways within the category of signal transduction included “downregulation of TGFβ receptor signaling,” and upregulation of “NOTCH4 intracellular domain regulates transcription,” “signaling by receptor tyrosine kinases,” and “diseases of signal transduction,” indicating the potential for T. pallidum to modulate a range of endothelial signaling pathways during exposure (Figure 5D). Further, pathways related to Rho GTPase activity were downregulated at 4- and 24-h, however the downregulated genes in these pathways are primarily GTPase-activating proteins (GAPs) which negatively regulate Rho GTPases (Supplementary Tables 7, 9). Relatedly, the Rho GTPase-mediated pathway “Sema4D induced cell migration and growth cone collapse” and “Rho GTPase activates NADPH oxidases” were upregulated at 12- and 24-h, respectively (Figures 5C,D).

To identify temporal connections between overrepresented pathways, we next performed pathway network analysis at each timepoint, to identify pathway clusters and connections based on overlapping DE genes (Figures 6A–D). Through this analysis we determined that the relationships between overrepresented pathways became more connected with longer T. pallidum exposures (Figures 6A–D). At earlier timepoints, pathways formed distinct clusters which were separated into categories such as immune responses and signal transduction (Figures 6A–C). The 24-h network showed the most connections within and between clusters of pathways, including connections between clusters belonging to different categories (Figure 6D), suggesting a converging transcriptional trajectory in T. pallidum-exposed HBMECs.

To identify higher-order overrepresented transcriptional profiles in HBMECs exposed to T. pallidum, we performed hallmark molecular signature enrichment analysis using the MSigDB database (Liberzon et al., 2015) on each study timepoint. Hallmark gene sets are curated and contextual, whereby gene function is considered for direction of regulation. Due to this context, and the higher-order approach to pathway analysis, hallmark gene sets can be both up- and downregulated at single timepoints. Through these analyses, we identified hallmark gene set overrepresentation within the categories of development, immune, signaling, and stress (Figures 7A–D), supporting the results of the Sigora pathway overrepresentation analysis. Consistent with the pathway analysis, hallmark gene set enrichment also identified inflammatory gene sets to be downregulated. Additionally, the gene set “TNF-α signaling through NF-κB” was simultaneously up- and downregulated at all timepoints, resulting from concurrent differential regulation of genes belonging to these pathways (Figures 7A–D). Further, the “apoptosis” gene set was overrepresented at all timepoints, although the direction of regulation varied (Figures 7B–D). We also observed overrepresented gene sets within the signaling category, including downregulation of “KRAS signaling UP” from 4- to 24-h, and upregulation of both “Wnt β-catenin signaling” and “NOTCH signaling” at 24-h (Figures 7B–D).

In support of the overrepresented TGFβ and SMAD signaling pathways identified by Sigora pathway analysis, in the hallmark gene set “TGFβ signaling” was also upregulated from 45-min to 12-h (Figures 7A–C). Notably, the “epithelial to mesenchymal transition (EMT)” gene set was both up- and downregulated from 4- to 24-h, with upregulation being most significant at these timepoints (Figures 7A–D). This bi-directional regulation reflects the dynamic regulation of this cellular process. Currently, there is no hallmark gene set for EndMT; however, EMT and EndMT have shared regulatory pathways (Saito, 2013), and thus this analysis provides insight into the regulatory factors common to both pathways. Additional EMT- and EndMT-related hallmark gene sets were overrepresented at the 24-h timepoint, such as “myogenesis” and “angiogenesis” (Figure 7D; Supplementary Table 13). These data reinforce that EndMT-inducing pathways and gene sets, such as TGFβ, SMAD, and NOTCH signaling, and SNAI1 transcription/Snail protein expression were induced in HBMECs exposed to T. pallidum.

We observed that many genes involved in EndMT-related processes were DE, including upregulation of the SMAD negative regulator PMEPA1 (Prostate transmembrane Protein Androgen Induced 1) and inhibitory SMAD7, as well as downregulation of activating SMAD3. Genes associated with SNAI1 overexpression (Pinto et al., 2018) were also DE, such as upregulation of the EndMT-promoting ECM component FBLN5 (Fibulin-5), and downregulation of adhesion molecules (Li et al., 2021) ICAM-1 (Intercellular Adhesion Molecule 1), VCAM-1 (Vascular Cell Adhesion Protein 1), as well as genes encoding junctional proteins, including CLDN1 (Claudin-1), OCLN (Occludin), and ZO-2 [Zona Occludens-2 (also known as TJP2)] (Table 2; Supplementary Table 1). vWF (von Willebrand Factor), a secreted factor important for endothelial inflammation, angiogenesis, endothelial dysfunction, and adhesion of platelets and leukocytes to endothelial cells (Randi and Laffan, 2017), was upregulated at the 12- and 24-h timepoints (Table 2; Supplementary Table 1). Mesenchymal marker transcripts (Piera-Velazquez and Jimenez, 2019) were increased, including upregulation of SM22-α [Smooth Muscle Protein 22 Alpha (also called TAGLN)], α-SMA [Smooth Muscle Actin Alpha 2 (also called ACTA2)], Integrin-β3, ACTC1 (Actin Alpha Cardiac Muscle 1), COL3A1 (Collagen type III A1), COL4A1, COL4A2, POSTN (Periostin), and FN1 (Fibronectin) (Table 2; Supplementary Table 1). These data demonstrate that HBMECs exposed to T. pallidum undergo significant transcriptional alterations consistent with the process of EndMT, and identified the involvement of Snail in the host endothelial response to T. pallidum.

We found that HBMECs exposed to T. pallidum demonstrated transcriptional alteration indicative of TGFβ and SMAD signaling, as well as pathways that modify TGFβ signaling such as NOTCH1, Wnt/β-catenin, and receptor tyrosine kinase (RTK) signaling, all of which contribute to EndMT (Figures 5, 7; Supplementary Tables 6–13). Importantly, TGFβ signaling is controlled through feedback loops, particularly negative feedback, where genes induced by TGFβ signal activation both attenuate the activity of TGFβ signaling proteins and repress transcription of genes involved in TGFβ signaling (Yan et al., 2018). The findings in the current study are consistent with the activation and downstream regulation of TGFβ signaling. Specifically, Sigora pathway analysis identified upregulation of the pathway “SMAD2/SMAD3: SMAD4 heterotrimer regulates transcription” at 12-h, which is the primary transcriptional response induced by TGFβ signal activation (Yan et al., 2018) (Figure 6C). Negative feedback was also evident through upregulation of the pathway “downregulation of TGFβ receptor signaling” at 24-h, where the DE genes in this pathway are upregulated as a negative feedback response to TGFβ signal activation (Figure 7D). Further, hallmark gene set analysis, which accounts for the context and function of DE genes, identified upregulation of TGFβ signaling from 45-min to 12-h (Figures 7A–C). These data indicate the activation and regulation of TGFβ signaling pathways in the endothelial response to T. pallidum.

To further investigate these findings, we assessed the transcript level changes of specific growth factor genes involved in EndMT-activating and regulatory pathways. Specific DE genes within TGFβ signaling include the downregulation of TGFBR (TGFβ Receptor)-2 and −3 at 12- and 24-h, and SMAD3, a prominent activating TF downstream of TGFβ signaling, at the 4- and 12-h timepoints (Table 2; Supplementary Table 1). Expression of TGFBRs is negatively regulated by genes induced by TGFβ signaling, such as the inhibitory SMAD7, which is induced by the SMAD2/SMAD3: SMAD4 dimer as a negative feedback mechanism (Yan and Chen, 2011; Yan et al., 2018) and was upregulated in this study (Table 2; Supplementary Table 1). Upregulation of additional TGFβ-induced negative regulators (Yan et al., 2018) was observed, including BAMBI (BMP and Activin Membrane Bound Inhibitor) at 24-h, and PMEPA1 at 12- and 24-h (Table 2; Supplementary Table 1). Additional upregulated TGFβ-induced genes (Watanabe et al., 2010; Ma et al., 2021) include ID (Inhibitor of DNA binding)-1, −2, and −3, and fibrosis regulator (Nakerakanti et al., 2011) CCN2 (Cellular Communication Network Factor 2) (Table 2; Supplementary Table 1). Upregulated genes that enhance TGFβ and EndMT include EDN1 (Endothelin-1) at 4- to 12-h, and POSTN (Periostin) that is involved in TGFβ-induced EndMT/EMT through integrin-β3 signaling (Yue et al., 2021), which was upregulated at 24-h (Table 2). These data demonstrate the activation and regulation of TGFβ signaling in HBMECs exposed to T. pallidum.

TGFβ can also signal independently of SMADs, including through pathways involving MAPK cascades, GTPases, and Wnt/β-catenin (Piera-Velazquez and Jimenez, 2019). In the current study, we observed significant overrepresentation of these signaling pathways at one or more timepoints, as determined by both Sigora and hallmark gene set overrepresentation analyses (Figures 5, 7; Supplementary Tables 6–13). Among these pathways, DUSP2 (Dual Specificity Phosphatase 2) and DUSP8 were among the most upregulated genes, indicating negative feedback of MAPK signaling (Table 2). Other regulators in the EndMT and non-canonical TGFβ pathways include the upregulation of NECTIN4 and IL-11, the latter of which contributes to fibrosis and EndMT and was upregulated at all timepoints (Table 2) (Schafer et al., 2017; Milara et al., 2022). Correspondingly, NOX4 (NADPH oxidase 4), which mediates fibrosis and cellular differentiation downstream of IL-11 and TGFβ (Cucoranu et al., 2005), was upregulated from 4- to 12-h (Table 2). Additionally, NEDD9 (neural precursor cell expressed, developmentally down-regulated 9), a focal adhesion scaffold protein involved in transducing cell adhesion and integrin signaling through RTKs, as well as promoting EMT and cancer metastasis (Jin et al., 2014), was significantly upregulated at 12 and 24 h (Table 2). Other growth factor signaling genes that contribute to EndMT were also upregulated, including PDGFB (Platelet-Derived Growth Factor B), VEGFA (Vascular Endothelial Growth Factor A), PGF (Placental Growth Factor), and FLT1, which encodes VEGF Receptor 1 (VEGFR1) (Table 2). In alignment with the differential expression of growth factors and overrepresentation of growth factor signaling pathways and hallmark gene sets, RTK signaling pathways were significantly overrepresented at the 24-h timepoint (Figure 5D). Collectively, these data demonstrate the alteration and regulation of growth-factor signaling genes and pathways in HBMECs during T. pallidum contact.

In the current study, we found T. pallidum-exposed HBMECs exhibited significant transcriptional alterations of ECM constituents and regulatory proteins, accompanied by overrepresentation of ECM organization and signaling pathways at several timepoints (Figures 5B,D; Table 3; Supplementary Tables 7, 9). ECM components that were DE included upregulation of collagen subtypes 3, 4, 5, 7 and 27, and downregulation of COL5A3 (Table 3; Supplementary Table 1). The collagen receptor and EMT positive regulator (Walsh et al., 2011), DDR2 (Discoidin Domain Receptor 2), was also upregulated at 24-h, as were the ECM components LAMA1 (Laminin subunit alpha 1) and FN1 (Fibronectin) that have previously been shown to facilitate T. pallidum attachment to the endothelium (Cameron, 2003; Cameron et al., 2004; Brinkman et al., 2008) (Table 3; Supplementary Table 1).

We also observed genes encoding ECM modifying proteins to be both up- and downregulated, indicating the dysregulation of endothelial ECM homeostasis during T. pallidum exposure. These genes include those belonging to the primary ECM modifying protein families such as MMP (Matrix Metalloproteinase), ADAM (A Disintegrin And Metalloproteinases), ADAMTS (A Disintegrin And Metalloproteinases with ThromboSpondin Motifs), and ADAMTSL (ADAMTS-Like proteinases) (Table 3; Supplementary Table 1). We also observed dysregulation of the plasmin regulation, including upregulation of the endothelial plasminogen activator PLAT [Tissue-type plasminogen activator (also known as tPA)]. Other related genes included downregulation of the inhibitor of tPA SERPINB2 [Plasminogen activator inhibitor-2 (also known as PAI-2)], and upregulation of SERPINE1 [plasminogen activator inhibitor-1 (also known as PAI-1)] (Table 3; Supplementary Table 1). Notably, SERPINE1 (PAI-1) expression is induced by TGFβ and Rho/ROCK activation (Samarakoon et al., 2008). Conversely, SERPINB2 is downregulated in response to TGFβ (Elsafadi et al., 2017). These findings demonstrate that T. pallidum exposure induces significant alterations in transcript levels of key ECM components, regulators of ECM deposition and degradation, and activators/inhibitors of the blood plasminogen/plasmin system. Further, these data highlight the importance of the ECM as a signaling and regulatory interface during T. pallidum-host interactions.

In our dataset we observed that pathways involving Rho GTPase activity, as well as those that interact with or modulate Rho signaling, such as RTK, ECM, and integrin signaling, were significantly overrepresented at one or more timepoints (Figures 5, 7D; Supplementary Tables 6–9, 13). Consistent with the overrepresented Rho signaling pathways, Rho GTPases RHOB, RHOD, RND1 (Rho family GTPase 1) were upregulated, and RHOJ was downregulated (Table 4; Supplementary Table 1). Cdc42 regulators were also DE, including upregulation of the Cdc42-activating protein CDC42BPG (Cdc42 binding protein kinase gamma). Several integrins were DE, including upregulation of ITGΒ3 at 12- and 24-h, and downregulation of ITGAX and ITGΒ8 (Table 4; Supplementary Table 1). These findings highlight the dysregulation of Rho GTPases, integrins, and regulators of endothelial junctional dynamics during T. pallidum contact.

To further investigate the activation of Rho GTPase signaling in endothelial cells during T. pallidum exposure, we performed fluorescent F-actin staining of T. pallidum-exposed HBMECs (Figure 8). A subset of HBMECs exposed to T. pallidum for 15 and 30 min exhibited a contracted cellular morphology, featuring prominent cortical actin rings and slight membrane blebbing (Figures 8A,B). Mean F-actin fluorescence intensity per cell was significantly higher in HBMECs exposed to T. pallidum compared to those exposed to basal media. Similarly, HBMECs treated with a biologically relevant concentration of thrombin showed a trend toward increased fluorescence intensity and displayed a contracted morphology with cortical actin ring formation, resembling the response observed in T. pallidum-exposed cells (Figures 8A–D). HBMECs exposed to IEC exhibited moderate contraction and a trend toward increased fluorescence intensity, indicating a cellular response to background components in the T. pallidum culture media. These findings highlight the pronounced response of HBMECs to viable T. pallidum. These alterations had dissipated at the 2-h timepoint, suggesting a temporal and reversible cellular response to T. pallidum engagement (Figure 8C). Additionally, the observed changes in cellular morphology precede alterations in transcription of cell-structure related genes and pathways, which was first observed at the 4-h timepoint. In summary, our transcriptomic analyses identified integrin/non-integrin, ECM, RTK, and Rho GTPase signaling pathways as significantly overrepresented in T. pallidum-exposed HBMECs. Microscopy analyses supported these findings by demonstrating HBMEC F-actin contraction during T. pallidum exposure, with viable T. pallidum inducing endothelial actin cytoskeletal alterations within minutes of contact.

We identified significant dysregulation of immune signaling pathways based on both Sigora and hallmark gene set analyses. Sigora pathway analysis identified the downregulation of chemokine receptor signaling from 4 to 12 h and the downregulation of IFN gamma and TNF signaling pathways at the 4- and 12-h timepoints, followed by upregulation of TNF and IFN alpha/beta signaling at 24 h (Figures 5A–D). Additionally, hallmark gene set analysis identified dysregulation of inflammatory responses from 4- to 24-h (Figures 7B–D). Downregulated genes within these pathways included monocyte recruiting and activating cytokines/chemokines, such as monocyte chemoattractant protein 1 (MCP-1 or CCL2), CCL20 (Chemokine C-C motif 20), and colony-stimulating factors (CSF)-1, CSF2, and CSF3 (Figures 5B–D). IFN immune signaling genes included downregulation of IRF1 (Interferon regulatory factor 1), and upregulation of IFNB1 (IFN-β) (Table 4; Supplementary Table 1).

TNF hallmark gene sets were differentially regulated at all timepoints in the present study, where the “TNF-ɑ signaling through NF-κB” gene set demonstrated both significant up- and downregulation, reflecting the diverse roles of differentially expressed genes within these pathways (Figures 5, 7; Supplementary Tables 6–13). Indeed, genes encoding TNF receptors (TNFRs) were downregulated from 4- to 12-h and were both up- and downregulated at 24-h (Table 4; Supplementary Table 1). TNFR adaptors such as the TRAF (TNF-receptor-associated factor) family exhibited differential expression, primarily at the 24-h timepoint (Table 4; Supplementary Table 1). Immune signaling genes downstream of both IFN and TNF, such as the NF-κB factors TCIM (Transcriptional and Immune Response Regulator), NFKB1, NFKB2, and RelB, were upregulated, while the NF-κB inhibitor IKBKE (Inhibitor of nuclear factor Kappa-B Kinase subunit Epsilon) was downregulated (Table 4; Supplementary Table 1). Collectively these findings indicate T. pallidum exposure alters TNF, IFN signaling, NF-κB signaling and the endothelial immune response to T. pallidum.

In this study, we also found cell death signaling pathways to be overrepresented in both Sigora and hallmark gene set analyses. The apoptosis hallmark gene set was downregulated at 4- and 12-h and was simultaneously up- and downregulated at 24-h, though upregulation was more significant (Figures 7B–D). Many pro-apoptotic factors within these pathways were DE, including upregulation of pro-apoptotic genes PMAIP1 (Phorbol-12-Myristate-13-Acetate-Induced Protein 1 [also known as Noxa]), BBC3 (Bcl-2 Binding Component 3 [also known as Puma]), and BCL2L11 (Bcl-2 like 11 [also known as BIM]), and downregulation of the anti-apoptotic factor BCL2A1 (Table 4; Supplementary Table 1). Sigora pathway analysis identified overrepresentation of pathways related to cell death signaling through p53 at the 24-h timepoint, where DE genes involved in p53-mediated cell death include upregulation of pro-apoptotic genes TP73 (Tumor Protein 73), and TP53INP1 (Tumor Protein p53 Inducible Nuclear Protein 1). Finally, key apoptosis activators CASP8 (Caspase 8) and CASP1 were downregulated at 24-h (Table 4). These findings demonstrate the dysregulation of apoptotic signaling through intrinsic cell death pathways, such as p53, BH3-only, and through interference with CASP8-regulated cell death signaling.