All animal procedures carried out in this study were performed in accordance with the European Union legislation for the protection of animals used for scientific purposes (Directive 2010/63/EU). The ethical terms of the research protocol were approved by the CYROI Institutional Animal Care and Use Committee (Comité d’Ethique du CYROI n° 114, IACUC certified by the French Ministry of Higher Education and Research) under accreditation 03387 (LeptOI) and 03584 (BatMan).

Mammals were trapped on Mayotte during three field sessions carried out from July 2012 to December 2014. Rattus rattus and Tenrec ecaudatus were captured using live traps baited with grilled coconut at eight and five localities, respectively. Sampling sites were selected in order to maximize the species diversity and geographic distribution of mammals in our sample. Trapping localities included a variety of different ecological settings across the island (see S1 Fig). In a local laboratory, R. rattus were sacrificed by cervical dislocation and T. ecaudatus by percussive blow to the head. For each individual, tissue was quickly collected from kidney, lung and spleen, pooled together and immediately stored in dry vials at -80°C. Insectivorous bats, Chaerephon sp. (Family Molossidae), were captured at dusk or at night using mist nets placed either in the vicinity of synanthropic roost sites or across flyways, while frugivorous bats, Pteropus seychellensis comorensis (Family Pteropodidae), were captured at night using mist nets set around fruiting trees. Captured bats were individually placed in clean cloth bags and brought back to a local laboratory. In most cases, animals spontaneously urinated after their removal from holding bags allowing the collection of urine. Bats were subsequently released at dusk at the initial capture site. Dogs (Canis lupus familiaris) were only manipulated and sampled by a local veterinary doctor; kidneys were obtained from animals that were euthanized for medical purposes, when possible urine collected from these same animals and immediately kept in dry vials at -80°C. In addition, urine samples were obtained from non-euthanized dogs through aseptic urethral catheterization. Kidneys from zebus (Bos indica) were purchased at the slaughterhouse of Mayotte (Kaweni) or obtained during traditional butchering. All samples were sent in liquid nitrogen to Reunion Island for analyses.

In addition to the sampling and storage at -80°C of organs and urine mentioned above, a few urine droplets and/or a small piece of freshly sampled kidney (crushed under sterile conditions) were used individually to inoculate three distinct culture media: (i) Ellinghausen-McCullough-Johnson-Harris (EMJH) liquid medium (Difco, Detroit, MI, USA) supplemented with Albumin Fatty Acid Supplement (AFAS; Royal Tropical Institute, Amsterdam, Netherlands) [10] (ii) EMJH liquid medium supplemented with AFAS, rabbit serum and fetal calf serum (1% each); and (iii) semisolid Fletcher medium (Difco, Detroit, MI, USA) supplemented with rabbit serum (8%). All media were supplemented with 5-fluorouracil (5-FU) at a final concentration of 200 μg.mL^-1. Cultures were incubated at 28°C, visually checked for the presence of leptospires using a dark field microscope once a week for four months and positive cultures were further sub-cultured in fresh EMJH liquid medium deprived of 5-FU. DNA was extracted from 1 mL of each positive culture using the EZ1 Biorobot with Qiagen EZ1 DNA Tissue kits (Qiagen, Les Ulis, France).

Approximately 1 mm³ of pooled organs (kidney, lung and spleen) from each R. rattus and T. ecaudatus specimen and only kidney from dogs and zebus were dissected on sterile ice and further processed as previously described [11]. Thirty microliters of urine from bats and dogs were combined with 120 μL of Dulbecco’s modified medium (GIBCO, Grand Island, NY, USA) and 50 μL of ATL buffer (Qiagen, Les Ulis, France). Subsequently, total nucleic acids were extracted from urine or homogenized tissues by using an EZ1 extraction robot and the EZ1 Virus Mini Kit version 2.0. A reverse transcription step was performed on total nucleic acids with GoScript Reverse Transcriptase (Promega, Madison, WI, USA) to obtain cDNA. Detection of a portion of the 16S rRNA gene of pathogenic Leptospira spp. was then carried out on 5 μL of cDNA using a probe-specific real-time Polymerase Chain Reaction system (RT-PCR) [12].

Leptospira spp. in samples testing positive by RT-PCR and/or culture were genotyped using a previously described MLST scheme encompassing six genes: secY, adk, rrs2, icdA, lipL32 and lipL41 [13], and recently optimized to improve the amplification of SWIO lineages [4]. RT-PCR is more sensitive than any MLST PCR (likely because of the very short length of RT-PCR amplicon), and some samples that were positive by the former detection system were MLST negative. In an attempt to further characterize Leptospira spp. from these particular samples, we used an alternative primer set (LA-LB) targeting a shorter piece of rrs2 [14]. The amplification of each marker was realized with GoTaq Hot Start Green Master Mix 2X (Promega, Madison, WI, USA) and further sequenced on both strands by direct Sanger sequencing (Genoscreen, Lille, France) using the same amplification primer sets. All sequences were deposited on GenBank under the following accession numbers: KT338823-KT338942, KT725237-KT725243 and KX427207-KX427227.

Accessible sequences from clinical Leptospira isolates from Mayotte were included in the study [15]. Since icdA sequences were not provided for these clinical isolates, we used five out of the six markers of the MLST scheme in our analyses. In addition, we added bacterial sequences obtained from endemic terrestrial small mammals from Madagascar, Microgale cowani, Microgale dobsoni, Microgale majori, Microgale longicaudata and Microgale principula (Family Tenrecidae, Subfamily Oryzorictinae). The complete MLST of these Leptospira strains were sequenced in a previous study [4] and from cultures realized during field missions carried out in the Central Highlands of Madagascar, Réserve Spéciale d’Ambohitantely, in March and October 2014 [16].

Phylogenetic analyses were performed on each gene separately and subsequently on concatenated sequences using the best model of sequence evolution determined by jModelTest v.0.1.1[17] for each dataset. Bayesian analyses were performed with MrBayes 3.1.2 [18] and consisted of two independent runs of four incrementally heated Metropolis Coupled Markov Chain Monte Carlo (MCMCMC) starting from a random tree. MCMCMC was run for 10,000,000 generations with trees and associated model parameters sampled every 100 generations. The convergence level of each phylogeny was validated by an average standard deviation of split frequencies inferior to 0.05. The initial 10% of trees from each run were discarded as burn-in and the consensus phylogeny along with posterior probabilities were obtained from the remaining trees. Bayesian trees were visualized and rooted to midpoint with FigTree v.1.3.1 (Andrew Rambaut, Institute of Evolutionary Biology, University of Edinburgh, 2006 to 2009; http://tree.bio.ed.ac.uk/).

Genetic diversity was compared between the identified clades in the multilocus phylogeny by estimating the nucleotide diversity (π) within each clade from the concatenated sequences using DNASP v.5.10.01[19].

Pathogenic Leptospira spp. were detected by RT-PCR in 27.0% of T. ecaudatus, 15.9% of R. rattus, 13.2% of dogs, 10.0% of P. seychellensis comorensis and 5.6% of zebus. All urine samples from Chaerephon sp. tested negative by RT-PCR. Bacterial cultures were attempted using 117 freshly sampled kidney and 94 urine samples. Overall, leptospires culture was successful for samples testing positive through RT-PCR only. Among the 81 R. rattus tested by culture inoculated with kidney, 14 were positive by RT-PCR, of which eight allowed positive cultures (57.1%). Similarly, eight of the 10 RT-PCR-positive samples (80.0%) from T. ecaudatus allowed isolation of leptospires by culture. Two of them (MDI295 and MDI321) yielded positive cultures from both urine and kidney samples. Lastly, no isolate grew in urine cultures from RT-PCR positive P. seychellensis comorensis samples. No culture was attempted when sampling dogs or zebus.

We were able to PCR amplify all six loci of the MLST scheme using DNA extracted from all 16 Leptospira cultures, obtained from eight R. rattus and eight T. ecaudatus samples. Interestingly, for T. ecaudatus, we failed to amplify the lipL41 gene with the conventional primers from Ahmed et al.[13], and used alternative primers designed by Dietrich et al. [4]. As expected, PCR targeting MLST loci were more arduous when using DNA extracted from tissue and urine of PCR-positive samples for which culture had failed. For instance, no successful PCR targeting MLST loci was recorded for the two RT-PCR positive T. ecaudatus failing to yield positive culture. Using the MLST scheme, only rrs2 locus was amplified from four PCR-positive R. rattus, while two loci (secY and rrs2) were amplified from two other R. rattus. Similarly, only rrs2 was amplified from six out of the seven RT-PCR-positive dog samples and from one out of the two RT-PCR-positive P. seychellensis comorensis samples. Lastly, no sequence could be obtained from the single zebu sample testing positive by RT-PCR. In addition, for seven PCR-positive R. rattus samples failing to produce any data using the MLST scheme, we successfully amplified and sequenced the rrs2 gene using an alternative primer set (LA-LB). Altogether, we could identify leptospiral diversity at the species level for 21 R. rattus, eight T. ecaudatus, six dogs and one frugivorous bat from Mayotte (see Table 1). Rattus rattus-borne Leptospira spp. appeared genetically diverse as L. borgpetersenii, L. kirschneri and L. interrogans were identified for thirteen, 5 and 3 samples, respectively. The sequencing of rrs2 locus in dog samples identified the infecting Leptospira spp. as L. borgpetersenii (n = 3) and L. kirschneri (n = 3). We also identified one of the two PCR-positive frugivorous bats as infected by L. kirschneri. Most importantly, all Leptospira strains isolated and/or genotyped from T. ecaudatus samples were identified as L. mayottensis.

We constructed a concatenated phylogeny (2215 bp) by using bacterial sequences from the 16 animal Leptospira isolates from Mayotte of the present study, 17 previously described clinical isolates from Mayotte [20] and nine Leptospira strains infecting Malagasy endemic Tenrecidae (six previously published [4] and three obtained from three distinct animals sampled in Réserve Spéciale d’Ambohitantely). The phylogeny presented on Fig 1 shows that all bacterial sequences obtained from T. ecaudatus clustered into the monophyletic L. mayottensis clade that includes two clinical L. mayottensis isolates (str. 200901116 and str. 200901122). This clade also contains Leptospira sequence types (STs) obtained from Microgale cowani and M. dobsoni, these latter species being tenrecids endemic to Madagascar. Out of the eight sequences obtained from T. ecaudatus in Mayotte, seven were identical to each others and similar to clinical L. mayottensis str. 200901122 (100% of pairwise identity based on 2215 bp) and the last one (MDI272) was closely related to the clinical L. mayottensis str. 200901116 (99.70% of pairwise identity based on 2215 bp). Of note, the couple of kidney/urine cultures obtained from two T. ecaudatus (MDI295 and MDI321) yielded identical sequence. Leptospira mayottensis infecting Malagasy Microgale spp. displayed a 27-fold higher nucleotide diversity (π = 0.00298) compared to that infecting T. ecaudatus from Mayotte in four different sampling sites (π = 0.00011, S1 Fig). The phylogeny confirms R. rattus as carriers of L. interrogans and L. borgpetersenii. For L. interrogans, the same ST is shared by one R. rattus (MDI219) and the L. interrogans clinical isolate str. 200901482 (100% of pairwise identity based on 2215 bp). For L. borgpetersenii, one clade of this bacterial species included seven sequences obtained from R. rattus that were almost identical (99.99% pairwise identity, 2215 bp) to five previously described clinical L. borgpetersenii isolates. Noteworthy, sequences embedded in this clade were clearly distinct from L. borgpetersenii sequences obtained from Malagasy tenrecids. The nucleotide diversity of L. borgpetersenii from Malagasy tenrecids (π = 0.00181) was higher than that found in R. rattus from Mayotte (π = 0, one single ST).

Although full MLST was not successful for bats, dogs and for some of the R. rattus, we still used rrs2 sequences alone in order to disentangle the role of these animals in human leptospirosis on Mayotte. For instance, the single leptospiral rrs2 (obtained using MLST scheme primers) sequence obtained from a frugivorous bat (P. seychellensis comorensis) was genetically related to clinical L. kirschneri isolates (98.70% pairwise identity, based on 452 bp) although nucleotide divergence at this locus together with PCR failure on all other loci suggested that L. kirschneri detected in frugivorous bats was actually divergent from clinical samples (see S2 Fig). Sequences obtained from dogs using the same primers revealed perfect identity with L. borgpetersenii (n = 3) and L. kirschneri (n = 2) clinical isolates from Mayotte. Lastly, as these rrs2 primers failed to produce amplification on some rat samples, we used an alternative primer set (LA-LB) that allowed amplification of a shorter rrs2 sequence, revealing five identical L. kirschneri sequences that, importantly, diverged from clinical isolates (3 mismatches out of 245 bp), thus in favour of a predominant role of dogs in L. kirschneri transmission to humans.