Introduction

Leptospirosis is a bacterial systemic infection, occasionally fatal, caused by the pathogenic spirochetes of the genus Leptospira. Though claimed as the most widespread zoonosis in the world [1,2], the disease is considered as emerging in many parts of the world. Leptospirosis is most prevalent in tropical and subtropical countries [3,4], presumably because the survival of the bacterium outside the host requires humid and warm conditions that are typical of tropical areas [5]. Rodents are recognized as the main reservoir of pathogenic Leptospira although several animal species are also capable of sustaining biofilm colonization of the renal tubules and shedding the bacteria in their urine [6]. Humans usually get infected through indirect exposure with water or soil contaminated with urine, but direct transmission has also been suggested as important for some species [7].

Traditionally, Leptospira have been classified serologically into 25 serogroups and over 300 serovars using Microscopic Agglutination Test (MAT) and Cross agglutination absorption test (CAAT) analysis, respectively [1,8]. More recently, a genetic classification based on DNA-DNA hybridization complemented by molecular methods and experimental studies have confirmed the existence of at least 22 Leptospira species [9]. The congruence between serological and molecular classifications is poor [1]: one serogroup can be linked to several Leptospira species while serovars can vary within a given clone or lineage, an observation considered as probably resulting from horizontal gene transfer [10]. Among the various molecular tools currently used to genotype Leptospira, multilocus sequence typing (MLST) has emerged as a method of choice as it provides data produced at a local scale that can be further compared to genotypes obtained all over the world and made freely accessible to the scientific community through public databases.

Leptospirosis is endemic in several islands of the southwestern Indian Ocean (SWIO) including the two French overseas departments of Mayotte [11,12] and Reunion Island [13,14] together with the island state of Seychelles [14–16], the latter recording the highest human incidence reported worldwide [17]. On Madagascar, serological evidence of exposure to leptospirosis has been reported in the human community of Moramanga [18], while to our knowledge, a single case of acute leptospirosis infection has been PCR-confirmed on a traveller returning from Madagascar [19]. Only serological evidence of human exposure to pathogenic Leptospira has been reported on the three neighboring islands of Union of the Comoros [20], while no or only scarce data is available for Mauritius [21] and Rodrigues Islands. On Mayotte, improved diagnostic procedures allowed to significantly increase the number of confirmed cases, with about 100 new human cases reported each year since 2009 [12,22]. The MLST analysis of Leptospira isolated from human incident cases on the island revealed a large bacterial diversity of clinical isolates, represented by at least 17 different “sequence types” (STs) and four distinct Leptospira species, namely Leptospira interrogans, Leptospira borgpetersenii, Leptospira kirschneri, and the newly described Leptospira mayottensis [12,23].

Comparatively, information is more abundant on pathogenic Leptospira infecting wild animals from the SWIO. On Mayotte, sequencing of a portion of the 16S rRNA (rrs) gene of leptospires infecting twenty black rats Rattus rattus has identified the same four Leptospira species previously reported in clinical cases from the same island, with a strict identity between the sequences of Leptospira infecting R. rattus and humans [11], designating this rodent species as the probable major reservoir and transmission source of Leptospira to humans on Mayotte [11]. On Comoros and Madagascar, sequencing of the nearly complete 16S rRNA gene of Leptospira infecting kidneys from six bat species (n = 7) has shown the carriage of L. interrogans (Comoros), L. borgpetersenii (Comoros and Madagascar) and other thus far unknown Leptospira species [24]. On Madagascar, partial sequencing of rrs locus has identified a single L. interrogans haplotype among 70 samples from introduced small mammals (rats, shrews and mice) [25]. Only one full five gene-based MLST analysis has been reported so far from the SWIO region, i.e. on Madagascar, where authors identified L. borgpetersenii, L. mayottensis and L. kirschneri from endemic small mammals and bats [26]. This study revealed distinct clustering associated with host type, with no overlap between Leptospira species infecting endemic small mammals versus those infecting introduced ones, a feature which certainly deserves further investigation.

On Reunion Island, human leptospirosis was first reported in 1953 but the first outbreak most likely occurred in 1868 [27]. Since 1953, several studies have been conducted to assess the burden of leptospirosis on the island. From 2008 to 2012, the mean annual incidence was estimated at 8.2 cases per 100,000 inhabitants with a fatality rate around 4%, and Icterohaemorrhagiae identified as the major serogroup in severe forms [13]. Lower prevalences of serogroups Canicola, Grippotyphosa and Australis have also been reported [28]. Pagès and colleagues [13] have highlighted the following population groups as being at highest risk: farmers and green space workers, people under 20 years old and participating in aquatic activities, people between 20 and 30 years old that fish, and people between 50 and 60 year old gardening at home. The seroprevalence of infection and/or Leptospira carriage in potential reservoirs, have been explored in wild small mammals (rodents, shrews, tenrecs and bats) and domestic animals (dogs, cats, cattle, goats, swine, rusa deers and horses) [14,29–32]. Since then, the black rat, Rattus rattus, abundant in most regions and biotopes of the island, has been considered to be the primary reservoir and transmitter of Leptospira spp., a conclusion mainly based on the observation that Icterohaemorrhagiae is by far the main serogroup found in rats and clinical cases [29], though the same serogroup has also been identified in dogs, pigs, rusa deer and tenrecs [29]. Recently, the first molecular study from Reunion Island identified that patients with acute leptospirosis (n = 42) were all infected by L. interrogans [33]. Although the typing method, High Resolution Melting (HRM) analysis of two VNTR sequences, allows a rapid diagnosis on clinical samples together with the characterization at the serovar and species levels, it is likely not resolutive enough as to trace the source(s) leading to human infection and illness.

The aim of our study was to determine the wild and/or domestic animal species that not only serve as reservoir hosts for pathogenic Leptospira species, but also are primary sources for human infection on Reunion Island, an area considered as endemic for human leptospirosis.

Materials and Methods

Animal samples

We trapped 799 rodents (562 Rattus rattus, 170 Rattus norvegicus, 67 Mus musculus domesticus), 171 shrews (Suncus murinus) and 25 tenrecs (Tenrec ecaudatus), those species representing all terrestrial small mammal diversity occurring on Reunion Island. Trapping occurred in 2012–2014 during dry and rainy seasons on twenty sampling sites, fifteen of them selected along two altitudinal transects on wet windward and dry leeward coasts (Fig 1). Details on trapping of wild animals and study sites have been published elsewhere (see [34]). Kidneys from cows (n = 33) and pigs (n = 22) were directly collected at the unique slaughterhouse of the island, which was not able to provide information regarding the geographic origin of the samples. Kidneys from stray dogs (n = 45) were collected just after the animals were euthanized at the community pound of Saint André (eastern coast). The only cat sample included in the study was urine collected by a veterinarian.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Sampling sites along the two altitudinal transects on western and eastern coasts, together with additional sampling sites in the north and west coast of Reunion Island.

https://doi.org/10.1371/journal.pntd.0004733.g001

Leptospira genotyping

Three major MLST schemes exist for Leptospira spp. typing worldwide, all supported by the website database http://pubmlst.org/leptospira/. In our study, scheme #3 [40] was preferred to the other two [10,41] as several molecular data from the SWIO have been made freely available using this same scheme, hence allowing comparison of local STs with those from SWIO islands [12,26]. MLST was attempted for adk, icdA, lipL32, rrs2, secY, and lipL41 loci as previously described [40] and recently optimized [26]. All positive domestic animal samples were first submitted to secY sequencing, revealing a nearly clonal population among infected rats (see Results). Hence, a randomly selected subgroup of positive wild animal samples was selected from eleven sites (n = 17; Table 1) and submitted to full MLST genotyping.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Leptospirosis testing results by host species.

Leptospira prevalence after real-time PCR diagnosis; species identification, either by 16S rRNA or secY sequencing and rate of success; and secY or MLST amplification and sequencing success.

https://doi.org/10.1371/journal.pntd.0004733.t001

For species identification of other positive wild animal samples, including two culture isolates, the secY gene was amplified with the primers secY F/R [40] or alternatively G1/G2 [42]. When secY PCR amplification failed, a 16S rRNA gene sequencing was alternatively used with previously published primers, i.e. LA/LB [43] or LA/R2 [44].

Obtained amplicons were directly sent for direct Sanger sequencing on both strands (Genoscreen, Lille, France), or alternatively cloned into the pGEM-T easy vector (Promega, Madison, WI, USA) when the quantity was insufficient for direct sequencing. Obtained clones (five per sample) were subsequently sequenced using M13 universal primers for Sanger sequencing. The overall analysis strategy is presented in Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Flow chart of the molecular methods performed to detect and identify Leptospira species and haplotypes infecting samples from Reunion Island.

At each step, the number (N) of tested animal samples // clinical samples is indicated. ^a The real time PCR diagnosis was attempted on either DNA extracted from clinical samples, or cDNA obtained after a reverse transcription step for animal samples. ^b Six-loci MLST was attempted only for a selected subgroup of the 198 animal samples successfully amplified on secY locus using MLST scheme #3, i.e. eight domestic animals and 17 randomly selected wild small mammals (all Rattus spp).

https://doi.org/10.1371/journal.pntd.0004733.g002

Results

Leptospira diversity within clinical samples

For 44 of the 66 clinical samples, we successfully amplified all six MLST loci (Table 1). For twelve clinical samples, only one to five MLST loci could be successfully amplified, even when using degenerated primers [26] (see S2 Table for details). For the last seven clinical samples, we failed to obtain successful PCR amplification at any of the six MLST loci. When these seven samples were further re-tested through an alternative real time PCR [37], only two samples tested positive at Ct values exceeding 42, whereas five sera provided negative amplification, suggesting that these were either false positive samples or positive samples that degraded during transport or conservation. Those seven human negative samples were discarded from the analysis and were no longer considered (Fig 2). The twelve partially sequenced samples allowed only identification of the infecting Leptospira at the species level; samples for which secY PCR product could be amplified were included in the MST analyses. In fine, of the 59 clinical samples successfully amplified at one or more locus, 57 were assigned to L. interrogans and two to L. borgpetersenii (Table 1). The secY-based MST (Fig 3), the neighbor-joining trees based on secY and rrs2 (Fig 4), and the MLST analysis all showed two L. interrogans clusters. As the two L. borgpetersenii-positive clinical samples allowed PCR amplification on two or three loci only (i.e. lipL32 and rrs2 for both samples, lipL41 for one sample only) but not on secY locus, they do not appear in the secY-based MST or the MLST-deduced phylogeny.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Minimum spanning trees of Leptospira based on secY gene (501 bp sequence), including our sequences from Reunion Island and twelve bat sequences previously published [46].

A) Minimum spanning tree of Leptospira interrogans (n = 257) and B) Leptospira borgpetersenii (n = 22). We used the goeBURST Full MST algorithm. The alleles are identified by a number (secY-1 to secY-54, see http://www.pubmlst.org/leptospira/) or an identifier; the circle size reflects their abundance in the data set. Group founders are in light green and common nodes are in light blue, except when overlapping with alleles from our sample, with specific colours referring to hosts (see legend). Links between the elements uses a grayscale, with lighter gray links showing more differences; the number of differences is indicated on the links. For two incomplete secY sequences (< 501 bp), the numbers of differences, shown in parentheses, were calculated on the overlapping nucleotides only, i.e. 443 for KU183602, and 491 for KJ607951.

https://doi.org/10.1371/journal.pntd.0004733.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Neighbour-joining phylogenetic trees (Kimura’s 2-parameter distances, 500 replicates) inferred from (A) secY (501 bp sequence) and (B) rrs2 (450 bp sequence) genes.

Clinical and animal samples from Reunion Island are written in black, using GenBank accession numbers (for one published bat sequence) or identifiers accompanied by the host name in the case of animal samples (RR: Rattus rattus; RN: Rattus norvegicus). Corresponding GenBank accession are reported on S1 Table (supporting information). Clinical samples from Mayotte are written in blue, using GenBank accession numbers and corresponding Leptospira species. Li: Leptospira interrogans; Lb: Leptospira borgpetersenii; Lk: Leptospira kirschneri; Lmay: Leptospira mayottensis.

https://doi.org/10.1371/journal.pntd.0004733.g004

Discussion

Although the mean prevalence of renal carriage among rats on Reunion Island (36.3%) was close to that reported in R. rattus on Mayotte (29.8%) [11], the genetic data reported in the present study reveal a striking contrast between the two islands: the rich Leptospira genetic diversity (at least 4 species with rather balanced representation) reported in rats from Mayotte, also found in human cases, contrasts with the low Leptospira genetic diversity reported herein in humans, rats and shrews on Reunion Island: though we identified three Leptospira species, L. interrogans represents 96.6% of clinical samples, 95.8% of positive animal samples and 100% of rat samples.

Other studies have reported a low genetic diversity among locally circulating Leptospira infecting local rat populations [25,51–53]. In contrast to earlier studies in New Caledonia [54], New Zealand [55] or Argentina [56], showing the carriage of L. interrogans or L. borgpetersenii in black rats (R. rattus), and the absence of L. borgpetersenii in Norway rats (R. norvegicus), L. borgpetersenii was absent in rats or shrews in our large sample of wild small mammals but was identified in a single mouse. The role of mice as maintenance hosts for L. borgpetersenii serogroup Ballum is largely recognized [1,3]. The perfect identity between lipL32 fragments from a single mouse and one of the two clinical samples positive for L. borgpetersenii highlights mice as a possible reservoir of pathogenic Leptospira at risk for humans on Reunion Island. However, this result should be interpreted cautiously as a single locus might not be resolutive enough as to infer Leptospira species; indeed, LipL32 sequences of Leptospira borgpetersenii and Leptospira weilii have been shown to be indistinguishable [57]. Further investigations targeting other loci with higher nucleotide polymorphism are needed to ascertain this point.

A different L. borgpetersenii infecting the second clinical case suggested the existence of a second potential reservoir host. A bat species endemic to Reunion Island, Mormopterus francoismoutoui, has been demonstrated as a L. borgpetersenii carrier in urine [50]. However, the comparison of rrs2 sequences from our two clinical samples infected with L. borgpetersenii to GenBank sequences obtained from this endemic bat species (n = 12) showed a low degree of genetic relatedness (Fig 3), thus excluding bats as a source of contamination leading to overt clinical leptospirosis in humans. A serological survey conducted in 2009 showed that up to 32% of cattle were seropositive for leptospirosis in Reunion Island, with Sejroe reported as the main circulating serogroup [30]. Our investigation finally identified cattle as a potential reservoir of L. borgpetersenii at risk for humans, as one L. borgpetersenii infecting a cow was found related to the L. borgpetersenii infecting a clinical case (perfect identity between 434 bp lipL32 fragments). As for mice, further investigation is needed in order to verify a potential transmission between cattle and human, possibly through an environmental maintenance of pathogenic leptospires [58].

Apart from the two L. borgpetersenii alleles common to human cases of leptospirosis and one cow or one mouse, at least two additional L. borgpetersenii haplotypes were identified in cows only, indicating a higher Leptospira diversity in cattle than in any other animal investigated on the island, which was already evidenced in other settings [59]. Our experience and previous published work [60,61] have highlighted recurrent failure to detect and amplify specifically L. borgpetersenii strains. Consequently, acute clinical cases related to this species might be underestimated.

The L. kirschneri found in two mice was not identified in any of our clinical samples on Reunion Island, whereas this species has been reported in clinical infection in Mayotte [11,12] as well as in other countries [2]. The reasons that may account for this difference are unknown. As mild infections not requiring hospitalisation or spontaneously healing cases are probably being underdiagnosed and underreported, it would be worth exploring whether these cases are associated with infection with other Leptospira species.

At an infraspecific level, we identified two clusters of L. interrogans, referring to sequence types ST02 and ST34 (http://www.pubmlst.org/leptospira/). International isolates previously published on the pubmlst.org database showed ST02 to be isolated from Belgium and Brazil, while ST34 was isolated from Jamaica and India. The second (minor) cluster, ST34, identified in six clinical samples, could not be detected in any of the five wild small mammal species targeted by our study. Additional sampling of domestic animals interestingly highlighted stray dogs as possible carriers of both L. interrogans STs found in human acute cases. Dogs have been repeatedly pinpointed as involved directly or indirectly in human epidemiology (contamination from dogs to humans, or from a common environmental source) [62–66]. For instance, a serosurvey conducted on Reunion Island in 2009 identified Canicola and Icterohaemorrhagiae as the major serogroups (43.5% and 21.7% respectively) in seropositive stray dogs, while Sejroe, Panama, Tarassovi and Ballum serogroups were also identified [29]. On the other hand, up to 17% of clinical cases were found to be infected with serogroup Canicola [28], suggesting that dogs, as the reservoir for this serogroup, are the source of contamination on Reunion Island. The importance of dogs in the epidemiology of the disease, if confirmed on Reunion Island, might be related to the lack of mammals diversity locally, as suggested on the island of Barbados [67].

An epidemiological survey conducted by CIRE-OI (Regional Office of French Institute for Public Health Surveillance in Indian Ocean, see [13]) indicates that, regarding the six patients infected with L. interrogans ST34, (i) five were actually living in the eastern, windward humid coast of the island and (ii) for those five cases, the disease was associated to leisure activities in the eastern side rivers and waterfalls. Noteworthy, of 56 R. rattus positive for Leptospira trapped in sites near rivers in the eastern transect (Fig 1), none of the 53 available sequences corresponded to L. interrogans ST34. Hence, it seems very unlikely that R. rattus acts as the reservoir of L. interrogans ST34.

In the end and as expected, our analyses stress black and Norway rats as reservoirs of Leptospira lineages of medical importance on the island. However, we were able to identify other animal species likely playing a role in the local epidemiology of human leptospirosis: mice, cattle and stray dogs. Noteworthy, the very low infra-specific diversity of L. interrogans infecting our infected samples did not enable us to assess the respective contribution of rats versus dogs to acute human infection with L. interrogans. Leptospira isolation from animals and humans would help to better assess the epidemiological links between humans and animals. Our conclusions are weakened by PCR and/or sequencing failures, likely resulting from different primer annealing efficiencies depending on the Leptospira species used as template (supported by our incomplete results on cattle and mice samples infected with L. borgpetersenii), or from low renal carriage, as suggested by high Ct values (e.g. Ct >30 in rats or >37 in dogs with undetermined infecting Leptospira, see S3 Table). Although time-consuming and fastidious, Leptospira culture needs to be carried out systematically, whenever possible, as previously suggested by other authors [68].

The high infection prevalence of cows with L. borgpetersenii raises another issue: the available human vaccine against leptospirosis currently proposed on the island target only L. interrogans species serogroup Icterohaemorrhagiae, while Sejroe was reported to be the main serogroup circulating in cattle in Reunion Island in 2009 [29]. This means that the vaccination of the ranchers’ population, at risk for leptospirosis for being in close contact with cows, might be inefficient against any other serogroup likely carried by livestock, in particular the one associated with L. borgpetersenii.

Pigs are another potential reservoir that deserves attention. Leptospirosis is a common disease of swine throughout the world and can be a significant cause of reproductive loss [69]. None of the 22 pig samples included in the study tested positive to Leptospira, while a serological survey conducted in 2009 on Reunion Island showed that up to 47.2% of pigs were seropositive for leptospirosis [29]. These two observations are not contradictory. First, the acute infection of an animal with leptospires does not prejudge its ability to develop a bacterial biofilm in the renal tubules and to shed the bacteria into the environment. Hence, serological data and PCR data on kidneys may be discrepant. Second, the kidney samples used in our study were collected from growing-finishing pigs under one year old, a potential age bias that might have affected our findings. Indeed, a proper assessment of pigs as reservoirs of leptospires at risk for human health requires the screening of older animals sampled from different farms as well as from family breedings throughout the entire island.

It is hypothesized that Leptospira were introduced onto islands with their animal hosts, and that a variable number of introduced strains have adapted to the new local environment and available hosts [14]. On Reunion Island, the fact that L. interrogans is almost clonal in rats is in favour of a recent leptospiral introduction on the island concomitant with the introduction of rats, back to the 16^th century. Leptospires detected in dogs, mice and cows show higher genetic diversity, which could result from multiple introduction events of infected animals of different species (for cows, there has been multiple importations from France of both Holstein Frisonne and Montbéliarde until 2008) or from distinct geographical origin, carrying different Leptospira lineages. We suggest that in contrast to Mayotte where the high diversity of Leptospira in humans and animal reservoirs is most likely related to (older) introduction events of infected hosts, potentially from countries with high host endemicity associated with Leptospira diversity [26], on Reunion Island, the narrow diversity of Leptospira might reflect more recent and/or fewer host introduction events. This is partly supported by the fact that fifteen of the seventeen STs from Mayotte strains have not been previously identified [12], contrary to STs from Reunion Island clinical cases that have already been identified worldwide. Previous studies have suggested that Leptospira from Mayotte and Madagascar are genetically closely related, and that Leptospira identified in Mayotte are probably mainly derived from Leptospira from Madagascar, possibly introduced in Mayotte via their hosts [14]. Although no genetic information is available so far on pathogenic Leptospira prevalent on Comoros, the serological profiles in humans from Comoros are comparable to those depicted in Mayotte, and noteworthy, antibodies from the serogroup Icterohaemorrhagiae are not detectable [20]. These findings contrast with those from human leptospirosis in Reunion Island and the Seychelles, where the Icterohaemorrhagiae serogroup is most common [70]. Hence there is compelling evidence that leptospirosis epidemiology might be similar along the Comoros, Mayotte and Madagascar axis and associated to indigenous or endemic Leptospira lineages, but different from that prevalent on Reunion Island and potentially Seychelles and Mauritius, where Icterohaemorrhagiae serogroup is dominant in clinical cases [15,16,21]. The near clonality reported herein in rat-borne Leptospira on Reunion Island is highly suggestive of recent pathogen introduction. This has to be addressed in the future by similar molecular analyses of Leptospira isolates from the other SWIO islands for which almost no molecular data are available so far, i.e. Mauritius and Seychelles.

Supporting Information

Acknowledgments

The authors thank the Direction de l’Environnement, de l’Aménagement et du Logement (DEAL Reunion) for issuing permits, the Office National des Forêts (ONF), the Parc National de La Reunion and the Réserve Naturelle Nationale de l’Etang St Paul, which gave authorization for the trapping of Tenrec ecaudatus (DIR/I/2013/008) and allowed us to sample inside protected areas. We also acknowledge Jean-Paul Dussud from Services Vétérinaires de La Réunion for providing cow samples, and CIREST and intercommunal pound of St André for allowing the dog sampling. We thank Amélie Desvars for providing four cow samples collected in 2009 during her PhD and Jean-Sébastien Dehecq for providing Reunion Island maps. This paper is dedicated to the memory of our late colleague Alain Michault, who disappeared the 5^th August 2015 during the preparation of this manuscript.