A total of 37 asymptomatic thoroughbred mares from farm #1 in central Kentucky with no history of leptospirosis were sampled from December 2022 to February 2023. Blood was collected by jugular venipuncture into serum separator tubes (Vacutainer^®, BD Diagnostics, Franklin Lakes, NJ, United States). Urine for PCR was collected by free catch and chilled on ice. In brief, the tail was wrapped, and sterile water used to remove all visible debris and thoroughly cleaned the vulva and surrounding area. Mid-stream urine was collected into sterile containers and shipped on ice packs for processing by PCR within 24 h. To collect urine for culture, if a mare was PCR-positive, the tail was wrapped and pulled away from the perineal area. Sterile water was used to remove all visible debris and thoroughly clean the vulva and surrounding area. The vulvar region was cleaned lightly with 70% ethyl alcohol and gently patted dry. The process was repeated if any contamination of the site occurred prior to urination. A diuretic was administered by intravenous injection (furosemide 5% injectable, 1 mg/kg based on estimated weight of mare), and midstream urine from successive voids was collected into sterile containers. The urine was passed to an assistant with clean gloves, and a sterile pipette was used to transfer 1 mL of urine from each voided sample to separate conical tubes containing 9 mL of HAN medium containing 5-fluorouracil (100 μg/μL) (30). Remaining urine from this second collection was transported on ice packs for a repeat PCR within 24 h.

Samples from farm #2 in central Kentucky, a farm with a very recent history of abortion by Leptospira as diagnosed by the University of Kentucky Veterinary Diagnostic Laboratory, were collected from a single mare suffering from an abortion in the final trimester of gestation (16 January 2023). Urine, chorioallantoic membrane, amnion, and allantoic fluid were immediately shipped on ice packs for processing within 24 h. A second set of samples collected 2 days later (18 January 2023), including urine, exudate, and a uterine swab, were inoculated directly into HAN medium and immediately shipped for processing within 24 h.

The MAT was performed using a panel of 18 antigens representative of 15 serogroups (Supplementary Table S1), as previously described (32, 33). A titer was considered positive at ≥1:100. Sera were also tested for reactivity in the MAT with the strain H9, an isolate recovered from a mare in this study.

The gene encoding the major outer membrane protein LipL32 discriminates pathogenic species of Leptospira from saprophytes. Before testing equine urine samples by lipL32 rtPCR (34, 35), preliminary studies were performed with spiked equine urine to optimize parameters for equine urine storage and pre-processing. In brief, 1 mL of L. borgpetersenii serovar Tarassovi strain Perepelitsin, at a density of 10⁸ leptospires/mL, was inoculated into 9 mL of freshly collected equine urine. A 1 mL aliquot of this was serially diluted 10-fold to prepare spiked samples containing 10⁷ Leptospira/mL to 10⁰ Leptospira/mL urine. Intact motile leptospires were enumerated by dark-field microscopy as previously described (36). In total, 1 mL of each dilution was then processed for the extraction of DNA after storage in various conditions/time points, including: (A) that same day, (B) that same day after centrifugation at 900 × g for 10 min to remove “sludge,” (C) after storage for 24 h in a Styrofoam container with ice packs, (D) after storage for 24 h in a Styrofoam container with ice packs followed by centrifugation to remove “sludge,” (E) after storage for 48 h in a Styrofoam container with ice packs, (F) after storage for 48 h in a Styrofoam container with ice packs followed by centrifugation to remove “sludge,” (G) after freezing at −20°C for 24 h, and (H) after samples had been centrifuged at 900 × g to remove “sludge” and stored at −20°C for 24 h. Storage treatments A to H correlate with labels are shown in Supplementary Figure S1. Thereafter, urine was centrifuged at 12,000 × g for 30 min. The supernatant was removed, and the pellets were washed twice by resuspending in 1 mL phosphate buffered saline (PBS) and centrifuging at 12,000 × g for 15 min, leaving the final pellets in ~100 μL. DNA was extracted from the urinary pellet using the Maxwell RSC Purefood Purification Pathogen Kit (Promega Corporation, Madison, Wisconsin, United States), following the manufacturer’s instructions, except using a 1 h incubation with 200 μL lysis buffer A and a 100 μL elution volume (37). lipL32 rtPCR was performed using 10 μL of Perfect taq qPCR ToughMix low ROX^™ (Quantabio, Beverly, MA, United States), 400 nmol/L of each primer, 132.5 nmol/L of probe, TaqMan^™ Exogenous Internal Positive Control Reagents: 2 μL of 10X Exo IPC Mix, 0.4 μL of 50X Exo IPC DNA, and 5 μL of DNA extract from culture or sample. rtPCR cycling was conducted on a QuantStudio^™ 7 (Thermo Fisher Scientific, United States) starting with an initial 3 min denaturation at 95°C for Taq polymerase activation, followed by 40 cycles of denaturation at 95°C for 15 s and primer annealing and extension at 60°C for 1 min. PCR of samples was performed in triplicate and considered positive by lipL32 rtPCR when duplicate or triplicates were positive with Ct values <40, as previously described (34, 35). DNA from L. borgpetersenii serovar Tarassovi strain Perepelitsin was used to prepare a standard curve as previously described (38).

A 45 mL aliquot of urine from each asymptomatic mare was centrifuged at 900 × g for 10 min at 4°C, and the supernatant was transferred to a clean 50 mL conical tube. Urine was then centrifuged at 12,000 × g for 30 min at 4°C. The supernatant was removed, and the pellet washed twice by resuspending in 1 mL of PBS and centrifuging at 12,000 × g for 10 min at 4°C. DNA was extracted as described above, and lipL32 rtPCR was performed as previously described (34, 35).

A 1 mL aliquot of freshly collected voided urine was immediately inoculated into 9 mL of HAN medium (30) and transported to the National Animal Disease Center (NADC), Ames, for culture of Leptospira, as previously described (37). Two 10-fold serial dilutions were made from the initial inoculum (i.e., 500 μL) into 5 mL of liquid HAN medium and incubated at 37°C in 5% CO₂. Vulvar exudate and chorioallantois samples were vortexed with 9 mL of HAN medium and two 10-fold serial dilutions were made (i.e., 500 μL) into 5 mL of liquid HAN at 37°C in 5% CO₂. Inoculated tubes were examined daily by darkfield microscopy for the first 2 weeks and then periodically for 6 months.

DNA was extracted from a 5 mL culture of the isolated strains using the Maxwell RSC Purefood Purification Pathogen Kit (Promega Corporation, Madison, WI), following the manufacturer’s instructions. The genomic DNA concentration was determined by Qubit (Qubit dsDNA Broad Range Assay Kit, Qubit 3.0 fluorometer, Invitrogen, Carlsbad, CA, United States). Illumina whole-genome sequence (WGS) was obtained (Nextera XT DNA Library Preparation Kit and the MiSeq Sequencer, 2 × 250 v2 paired-end chemistry, Illumina, San Diego, CA, United States), according to the manufacturer’s instructions. Illumina WGS reads were taxonomically identified using Kraken 2 version 2.1 (39). Reads were assembled with SPAdes 3.13 (40) and verified by comparing the expected genome size with the actual assembly size and verifying contigs as Leptospira by BLASTN (41) against the NCBI nucleotide (NT) database.

The serogroup of strains H9 and H44 were determined by the MAT method using a panel of polyclonal rabbit reference antisera representing 13 serogroups (Supplementary Table S2). The serovar of strains H9 and H44 was determined by performing MAT with panels of monoclonal antibodies (mAbs) that characteristically agglutinate serovars from the serogroups Australis and Pomona, respectively, as previously described (42).

DNA from two Leptospira PCR-positive culture-negative urine samples (designated as H2 and H8) was subjected to pan-pathogenic Leptospira DNA capture and enrichment, as previously described (43). The two samples were processed differently because DNA from sample H8 was highly fragmented and displayed low concentrations of nucleic acids (~0.5 ng/μL), whereas sample H2 had DNA that was more intact (average fragment size >6,000 bp) and concentrated (~4 ng/μL). For sample H2, the DNA was diluted to ~2 ng/μL in a volume of 40 μL and sonicated to an average size of 228 bp using a Q800R2 sonicator (QSonica, Newtown, CT, United States). For sample H8, 40 μL of undiluted DNA was subjected to brief sonication, and the final average fragment size was 103 bp. Short-read next-generation libraries were prepared separately using Agilent Sure-Select methodology. The libraries were then pooled together in equimolar amounts and were subjected to one round of DNA capture and enrichment and then sequenced on an Illumina MiSeq instrument using a MiSeq v3 600 cycle kit (2 × 300bp reads).

To estimate the percentage of Leptospira reads in the enriched sequences, reads were mapped against the standard Kraken database with Kraken v2.1.2 (39). Reads assigned as Leptospira were then extracted and assembled using SPAdes v3.13.0 (40) with default settings; assemblies were also generated for reads that were generated from isolates for samples H9 and H44. Assemblies for H2, H8, H9, and H44 were placed into a genus dendrogram containing 66 Leptospira reference genomes with Mashtree v1.2.046 (44) to confirm species identification. GenBank accession numbers for each genome are presented in figures.

Single nucleotide polymorphisms (SNPs) were identified among two enriched genomes (H2 and H8), two genomes from cultured isolates (H9 and H44), and 103 publicly available Leptospira genomes (GenBank accession numbers for each genome are presented in figures) by aligning reads against reference genomes Leptospira interrogans serogroup Canicola serovar Canicola strain LJ178 (GCA_008831445.1), Leptospira interrogans serogroup Icterohaemorrhagiae serovar Copenhageni strain Fiocruz_L1-130 (GCA_000007685.1), or Leptospira kirschneri serogroup Grippotyphosa serovar Grippotyphosa strain RedPanda1 (GCA_027563495.1) using minimap2 v2.22 (45) and calling SNPs from the BAM file with GATK v4.2.2 (46) using a depth of coverage ≥3× and a read proportion of 0.9. Based on a reference self-alignment with NUCmer v3.1 (47), SNPs that fell within duplicated regions were filtered from downstream analyses. All of these methods were wrapped by NASP v1.2.1 (48). Maximum likelihood phylogenies were then inferred on the concatenated SNP alignments using IQ-TREE v2.2.0.3 with default parameters (49), 1,000 bootstrap replicates, and the integrated ModelFinder method (50); the phylogenies were rooted with either Leptospira interrogans serovar Canicola strain LJ178, Leptospira interrogans serovar Copenhageni strain Fiocruz_L1-130, or Leptospira kirschneri serovar Grippotyphosa strain RedPanda1, as appropriate. To determine breadth of coverage for the enriched genomes, reads were aligned against Leptospira interrogans serovar Copenhageni strain Fiocruz_L1-130 for sample H2 and Leptospira kirschneri serovar Grippotyphosa strain RedPanda1 for sample H8 with minimap2, and the per base depth of coverage was calculated with Samtools v1.6 (51).

All animal experimentation was conducted in accordance with protocols as reviewed and approved by the Animal Care and Use Committee at the NADC and USDA institutional guidelines. Strains H9 and H44 were propagated in liquid HAN medium at 29°C and were evaluated for virulence by intraperitoneal injection of 10⁸ leptospires in 500 μL into groups (n = 4 per group) of golden Syrian hamsters (Mesocricetus auratus). A negative control group (n = 4) received HAN medium alone. After 3 weeks of inoculation, animals were euthanized and exsanguinated by cardiac puncture, and whole blood smears were evaluated. Dried slides were submitted to the National Animal Disease Center Microscopy Services laboratory for Giemsa staining as reported previously (52), and Giemsa-stained slides were evaluated by counting the first 100 white blood cells to identify the portion of foamy macrophages. Differential cell counts were evaluated in R (53), where a linear regression model fitting challenge was used as a fixed effect to generate least square means and standard errors. Kidney and liver tissues were harvested for culture. qPCR for lipL32 was performed on liver and kidney samples, as previously described (54), except using DNA from L. interrogans serovar Canicola strain Hond Utrecht IV for a standard curve. Sera were collected for MAT, which was performed according to World Organisation for Animal Health (WOAH) guidelines (32), using a panel of 18 antigens representative of 15 serogroups (Supplementary Table S1), as well as strains H9 and H44. A titer was considered positive at ≥ 1:100.

In addition to strains H9 and H44, L. interrogans serogroup Icterohemorrhagiae serovar Copenhageni strain Fiocruz_L1-130 (55, 56) and L. interrogans serogroup Australis serovar Bratislava strain PigK151 (57) were propagated in HAN medium until mid-late log phase and harvested by centrifugation (10,000 × g, 4°C, 30 min), washed twice with PBS, and processed for one-dimensional (1-D) sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) on 12% acrylamide gels (Bio-Rad, Hercules, CA, United States), according to the manufacturer’s guidelines. Proteins were visualized by staining with SYPRO Ruby (Invitrogen), and LPS was visualized by staining with Pro-Q Emerald 300 (Invitrogen), according to the manufacturer’s guidelines. For immunoblotting, samples were transferred by semi-dry transfer (Amersham TE77 PWR) to an Immobilon-P transfer membrane (Millipore, 220 Bedford, MA, United States) and blocked overnight at 4°C with Starting Block (PBS) blocking buffer (Thermo Fisher). Membranes were individually incubated with indicated antisera [anti-LipL32, anti-Loa22 or anti-LipL21 (58, 59)] or reference antisera [anti-Pomona, anti-Bratislava, anti-Copenhageni, or anti-Ramisi (NVSL, USDA, Ames, IA)] diluted in blocking buffer followed by incubation with horseradish-peroxidase anti-rabbit immunoglobulin G (Sigma, St. Louis, MO, United States) conjugate diluted 1:4,000 in blocking buffer. Bound conjugates were detected using Clarity Western ECL substrate (Bio-Rad), and images were acquired using a Bio-Rad ChemiDoc MP imaging system.

PCR of equine urine samples that were centrifuged at 900 × g for 10 min to remove “sludge” and processed on the same day or within 24 h of storage at 4°C provided the lowest limits of detection at 10 Leptospira/mL of urine (Supplementary Figure S1). Storage of samples at 4°C for 48 h, freezing, or not removing “sludge” by centrifugation at 900 × g for 10 min increased the limit of detection to 10² Leptospira/mL of urine. Freezing samples without removing “sludge” increased the limit of detection to 10³ Leptospira/mL of urine.

Urine from 37 asymptomatic mares, designated as H1 to H37, was tested by rtPCR for lipL32 using optimized protocols as described above. Three urine samples (8%), designated as H2, H8, and H9 from three different pregnant mares, were PCR-positive with Ct values of 37, 38.2, and 36.9, respectively (Table 1). Repeat testing of urine collected 4 days later was PCR-positive for mares H8 and H9 but negative for mare H2. Urine from mare H9 was culture-positive for Leptospira. Complete data on all equine samples are shown in Supplementary Table S3.

Urine, chorioallantoic membrane, amnion, and allantoic fluid collected from a mare (designated as H44) showing clinical disease, and in the process of aborting at the time of sample collection, were positive by rtPCR for lipL32 with Ct values of 26, 31.8, 29.9, and 25.4, respectively (Supplementary Table S3). Chorioallantoic membrane was culture-positive for Leptospira. Samples collected from mare H44 after 2 days post-abortion, including urine, vulvar exudate, and uterine swabs, were all positive by rtPCR for lipL32, but all cultures were heavily contaminated (Supplementary Table S3).

An isolate of Leptospira cultured from urine of asymptomatic mare H9, designated as strain H9, was genotyped as L. kirschneri (Figure 1; Supplementary Figure S2). Serotyping of strain H9 with reference antisera identified the isolate as belonging to serogroup Australis (Supplementary Table S3). Additional serotyping with monoclonal antibodies to identify serovar confirmed that strain H9 was most similar to serovar Rushan due to similar reactivity patterns with the Rushan reference strain within the serogroup Australis (Supplementary Figure S3). Strain H9 is classified as L. kirschneri serogroup Australis serovar Rushan.

An isolate of Leptospira cultured from chorioallantois of mare H44 presenting with abortion was genotyped as L. interrogans (Figure 2; Supplementary Figure S2). Serotyping of strain H44 with reference antisera identified the isolate as belonging to serogroup Pomona (Supplementary Table S3). Additional serotyping with monoclonal antibodies to identify serovar confirmed that strain H44 was most similar to serovar Pomona due to similar reactivity patterns with the Pomona reference strain within the serogroup Pomona (Supplementary Figure S4). Strain H44 is classified as L. interrogans serogroup Pomona serovar Pomona.

Given the fastidious growth requirements of pathogenic Leptospira and the inherent difficulties with their culture from urine of large animal species, urine samples that were PCR-positive for lipL32 but culture-negative for Leptospira were processed using a culture-independent DNA capture and enrichment system to obtain Leptospira genomic information directly from urine samples H2 and H8. Enriched samples H2 and H8 had very high and moderate proportions, respectively, of sequencing reads that assigned to Leptospira: 98.36% for H2 (1,529,676 of 1,555,190 reads) and 60.08% for H8 (562,885 of 936,958 total reads). Sample H2 revealed a breadth of coverage of 99.95% with an average depth of 77.5× (72–78×) against reference genome L. interrogans serogroup Icterohemorrhagiae serovar Copenhageni strain Fiocruz_L1-130. Sample H8 displayed a breadth and depth of coverage of 92.72% and 32.5× (3–176×) when aligned against Leptospira kirschneri serogroup Grippotyphosa serovar Grippotyphosa strain RedPanda1. The whole genome Leptospira dendrogram placed enriched and isolate genomes for strains H8 and H9 in the L. kirschneri clade, whereas enriched and isolate genomes for strains H2 and H44 grouped with L. interrogans (Supplementary Figure S2). To obtain higher resolution within L. interrogans, a core genome phylogeny was constructed containing enriched genome H2 and isolate genome H44 plus 51 diverse L. interrogans genomes (Supplementary Figure S5). This analysis placed H44 within the “serovar Pomona” clade and H2 within the “serovar Copenhageni” clade.

Three separate phylogenies were constructed based on the core genomes of: (1) enriched genome H8, isolate genome H9, and 36 L. kirschneri reference genomes (Figure 1); (2) isolate genome H44 and 16 L. interrogans reference genomes that assign to the “serovar Pomona” clade (Figure 2); and (3) enriched genome H2 and 9 L. interrogans reference genomes that fall within the “serovar Copenhageni” clade (Figure 3). Genomes for strains H8 and H9 clustered together in a clade within, yet distinct from, other L. kirschneri genomes (Figure 1). Genome H44 grouped with other genomes of L. interrogans serovar Pomona from US isolates of Leptospira derived from a fox, sea lions, and a human (Figure 2). Finally, enriched genome H2 was most like other L. interrogans serovar Copenhageni genomes (Supplementary Figure S5), yet still distinct by displaying 376 SNP differences to separate it from the “serovar Copenhageni” clade (Figure 3).

Sera from 37 asymptomatic mares were tested by MAT. Of these, 35 (94.6%) were seropositive (titer ≥1:100). Equivalent highest titers were observed for more than one serogroup in 10 samples (Supplementary Table S3). The most frequent highest-reacting MAT titer was with serogroup Australis (31.4%), followed by serogroups Pomona (20%), Djasiman (5.7%), Grippotyphosa (5.7%), and Icterohaemorrhagiae (2.8%). All equine sera were seronegative when tested by the MAT using strain H9.

Serum from the clinically infected mare (H44) presenting with abortion had a high MAT titer to serogroup Pomona (1:102,400).

Experimentally inoculated hamsters did not show any clinical signs of infection or weight loss after intraperitoneal inoculation with L. kirschneri serogroup Australis strain H9. Similarly, hamsters inoculated with L. interrogans serogroup Pomona strain H44 did not show clinical signs of disease, except for one hamster that was losing weight and euthanized at day 14 post-inoculation. After 3 weeks of infection, all remaining hamsters were euthanized. All kidney samples from each group tested positive by lipL32 qPCR (Table 2 and Figure 4). Livers from hamsters inoculated with strain H44 were also PCR-positive but livers from hamsters inoculated with strain H9 were PCR-negative. Kidneys and livers from hamsters inoculated with strain H44 were all culture-positive, but only two of four hamsters inoculated with strain H9 were kidney-culture positive (Table 2). Hamsters inoculated with L. kirschneri serogroup Australis strain H9 had positive MAT titers against strain H9 and reference strain L. interrogans serogroup Australis serovar Bratislava strain Jez Bratislava but were seronegative when tested against other reference antigens (Table 2). Hamsters inoculated with L. interrogans serogroup Pomona strain H44 had positive MAT titers against strain H44 and reference strain L. interrogans serogroup Pomona serovar Pomona strain Pomona but were seronegative when tested against other reference antigens (Table 2). Manual blood count differentials determined that all hamsters challenged with both H9 and H44 produced circulating foamy macrophages (1.50 ± 0.66 and 1.55 ± 0.66, respectively, foamy macrophages per 100 white blood cells evaluated), a marker associated with virulence and disease severity in the hamster model of leptospirosis (52, 60).

Total protein profiles of L. kirschneri serogroup Australis strain H9 and L. interrogans serogroup Pomona strain H44 were compared with those of L. interrogans serogroup Icterohemorrhagiae serovar Copenhageni strain Fiocruz_L1-130 and L. interrogans serogroup Australis serovar Bratislava strain PigK151 (Figure 5A). As expected for pathogenic Leptospira, similar protein profiles were detected across different species and serovars, and all strains were confirmed to express the pathogen-associated protein LipL32 and known virulence factors Loa22 and LipL21 (Figures 5B,C).

Total lipopolysaccharide (LPS) profiles of each strain were also compared (Figure 6A). The results confirm the unusual and atypical LPS profile of pathogenic Leptospira compared with that of E. coli and confirm different LPS profiles between serovars of Leptospira from different serogroups. Antigenic differences of LPS between serogroups and different serovars within the same serogroup were confirmed by immunoblotting with reference antisera specific for (1) serogroup Australis serovar Ramisi, (2) serogroup Pomona serovar Pomona, (3) serogroup Icterohemorrhagiae serovar Copenhageni, and (4) serogroup Australis serovar Bratislava (Figures 6B–E).

Sequencing data and accession numbers are available at NCBI under BioProject ID PRJNA994138.