Sample selection and Rattus spp. typing
We sampled 171 Rattus spp. collected across the seven administrative regions of the City of Johannesburg municipality (Figure 1) as part of a plague (Yersinia pestis) surveillance program between January and May 2016. Johannesburg is an inland, high-elevation (~1700m) site that is characterised by cold, dry winters and warm, wet summers (Supplementary materials – S1_Bioclimate). Invasive Rattus spp. were presumptively identified based on morphology (weight and tail to body length ratio). Rattus spp. >250g and with tail/body <1 were classified as R. norvegicus. Animals with equivocal morphological measurements (n=32) and a subset (n=10) of R. norvegicus with unequivocal morphological measurements were further genotyped by cyt b gene characterisation as previously described (6). DNA extractions were performed on kidney samples using the QIAamp DNA Mini kit on a QIAcube system (Qiagen) according to manufacturer’s instructions. For comparison, R. norvegicus samples (n=12) collected during an outbreak of human leptospirosis in a correctional facility in Cape Town (7) also underwent cyt b characterisation. Cyt b PCR products were purified as previously described (6) and sequenced at the core Sanger sequencing facility at the University of Pretoria.
Leptospira detection and prevalence estimation
Leptospira spp. infections were detected using a diagnostic real-time PCR targeting 300bp of the lfb1 gene and incorporating melt curve analysis to identify infecting Leptospira species (10). As the standard lfb1 primers may underestimate the prevalence of L. borgpetersenii infections and L. interrogans-L. borgpetersenii mixed infections may occur (11), samples which initially tested negative, as well as samples (n=12) from Cape Town, eight of which had previously been identified as infected with L. interrogans (7), were retested using a L. borgpetersenii species-specific forward primer (11). All real-time PCR assays were performed as previously described (7,11). Prevalence estimates and logit confidence intervals were estimated using the binom package (https://cran.r-project.org/package=binom) and mapped using the ggmap package (https://cran.r-project.org/package=ggmap). The prevalence estimates from the six regions in which more than 10 animals were sampled were compared using Chi-Square tests. All analyses were performed in R version 3.6.1.
Genotyping of Leptospira spp. infections
To confirm the Leptospira species classification based on melt curve analysis (10) we sequenced a subset (n=37) of the lfb1 amplicons. Although the lfb1 locus has been demonstrated to provide valuable phylogenetic data (12), these initial sequences from samples collected across five regions revealed no sequence polymorphisms. Therefore, we subsequently sequenced additional loci, secY (~450 bp) (n=13) and lipL41 (~500 bp) (n=5), from a subset of samples to increase the resolution of the molecular typing and allow identification of the presumptive serogroup. Similarly, infections in R. norvegicus in Cape Town previously identified as L. interrogans by sequencing of the lfb1 amplicon (7) were further typed by sequencing secY (n=3) and intergenic regions MST1, MST3 and MST9 (n=3) (13) to determine whether further genetic resolution was possible. Primer pairs secYFd/secYR3 and lipL41F3/lipL41R3 were used to amplify secY and lipL41 (14) and MST1, MST3 and MST9 were amplified using published primers (13) on a Techne TC5000 system (Techne Inc). The total reaction volume of 25 µl consisted of 5 µl of DNA extraction, primer concentrations of 0.5 µM, 12.5 µl of MyTaq Red mix 2x (Bioline Reagents Ltd) and 5.5 µl of molecular grade water. A “touchdown” thermal profile comprising initial denaturation at 95°C for 3min, followed by 40 cycles of denaturation at 95°C for 20s, variable annealing for 25s and extension at 72°C for 40s, with a final extension at 72°C for 7min was performed. Annealing temperature was reduced from 60°C to 50°C over the first 10 cycles and then maintained at 46°C. Each PCR run included a negative control (molecular grade water) for every four samples, and a positive control (L. borgpetersenii strain 201501056 for L. borgpetersenii-specific assays and L. interrogans strain 201501067 for all other assays).
Leptospira PCR products were purified using the QIAquick PCR Purification Kit (Qiagen) according to manufacturer’s instructions with a final elution in 35 µl. Purified product was quantified using a Nanodrop ND1000 spectrophotometer (Thermo Scientific) and sequenced by Eurofins Genomics GmbH (Ebersburg, Germany).
Phylogenetic analyses of Rattus spp. cyt b sequences and Leptospira spp. multi-locus sequences
Sequences for each locus were aligned using the ClustalW algorithm and the most appropriate evolutionary model determined using MEGA7 (15). Cyt b sequences from previous studies of Rattus spp. in South Africa (6) were used as reference sequences. Leptospira reference sequences were obtained by querying sequences against the NCBI refseq_genome database using the BLASTn algorithm limited to Leptospira (taxid 171) belonging to the two species (L. interrogans and L. borgpetersenii) identified in this study. Aligned BLAST hits for each locus were linked by NCBI Biosample accession and representative sequences for each Leptospira species and serovar combination selected as reference sequences.
To characterise Leptospira spp. genetic diversity, multi-locus phylogenetic analyses were implemented in BEAST v2.6.0 (16) using each locus as a separate partition with unlinked substitution models and linked clocks (strict) and trees. The most appropriate substitution models as determined by model test in MEGA7 (15) were used for each locus. Multi-locus analyses were run using a chain length of 1x107 and sampled every 1x103 runs with a burn-in of 10%. TRACER v1.7.1 (17) was used to verify that the effective sample size (ESS) was greater than 200 and TREEANNOTATOR v2.6.0 was used to generate a maximum clade credibility tree using mean node heights annotated by posterior probabilities greater than 0.9. Trees were annotated using the R package ggtree (18).