We conducted trapping and sample collection under permits issued by the Madagascar Ministry of Environment and Forests (no. 154/13/ MEF/SG/DGF/DCB.SAP/SCB; no. 312/13/MEF/SG/DGF/DCB.SAP/SCB; no. 178/14/MEF/SG/DGF/DCB.SAP/SCB). We conducted this study in accordance with Institut Pasteur animal use guidelines (https://www.pasteur.fr/en/file/2626/download? token=YgOq4QW7); the study was approved by a committee of the Institut Pasteur de Madagascar.

During 2013–2015, we sampled small mammal hosts at 11 sites in Moramanga District, eastern Madagascar. Two sites were within an uninhabited humid forest, and the remaining sites included areas of human habitation and heterogeneous land use. We identified host species on the basis of phenotypic characteristics, external measurements, and craniodental measurements (when appropriate) (12). We euthanized and dissected animals and stored kidneys in 95% ethanol.

We extracted DNA from 0.04 g of kidney tissue with DNeasy Blood and Tissue Kit (QIAGEN, Valencia, CA, USA) using the manufacturer’s instructions and an elution volume of 100 µL. We detected Leptospira with a TaqMan qPCR assay targeting the 16S rRNA gene (13) using the StepOne Real-Time PCR System (Life Technologies, Waltham, MA, USA). Any sample with a cycle threshold <36 in 1 assay or <40 in 2 replicate assays was classified as Leptospira positive.

Initial genotyping of positive samples was achieved by amplification and sequencing of ≈300 bp of the lfb1 gene on an Eco-Illumina qPCR System (Illumina Inc., San Diego, CA, USA) (6). To characterize mixed infections, we designed forward primers targeting the lfb1 locus of 4 Leptospira species (L. interrogans 5′-CCTCTTACGCACAGATCRGTC-3′, L. borgpetersenii 5′-CCAACACTCCCTCCTCTATCAGC-3′, L. mayottensis 5′-CGCAGACTAGCAGCCCAACC-3′, and L. kirschneri 5′-GACCGCTTACGCACAGATCG-3′) and paired them with the standard lfb1 reverse primer using the same thermal profile. After sequencing, we retested samples with redesigned primers targeting Leptospira spp. not previously identified and sequenced those products (GenBank accession nos. MG759567–664). Each assay included a negative control (sterile water) for every 4 samples and a positive control. We purified PCR products using the QIAquick PCR Purification Kit (QIAGEN) and sent them to Eurofins Genomics GmbH (Ebersburg, Germany) for sequencing. We calculated prevalence and logit CIs using the binom package in R version 3.2.2 (https://cran.r-project.org/package=binom).

We captured 2,847 small mammals across 11 sites; 5 invasive species (R. rattus and R. norvegicus rats, Mus musculus mice, and Suncus murinus and S. etruscus shrews) accounted for 93% (2,653/2,847) of the captures. Of these, we captured R. rattus rats most frequently (n = 2,312) and at all sites, including forest sites. Although we found endemic hosts at all sites, 56% (107/190) were captured at forest sites. We tested 723 captured animals (43–102 animals/site) for Leptospira. We tested all endemic host samples and a subset of introduced host samples for each site. Overall prevalence of infection was 26%, ranging from 11% in Microgale spp. tenrecs to 48% in M. musculus mice (Table).

Figure

Figure. Proportion of Leptospira-positive Rattus rattus rat samples (n = 63) infected with L. borgpetersenii, L. interrogans, or L. mayottensis before and after characterizing mixed infections, Madagascar, 2013–2015. We initially genotyped R....

We genotyped 93 Leptospira-positive samples; the prevalence of mixed infections was 19% (95% CI 13%–29%). This value is still likely an underestimate, considering that cross-amplification of Leptospira species within the same taxonomic subgroup occurred. Mixed infections comprised L. interrogans and either L. borgpetersenii (n = 14) or L. mayottensis (n = 3); 1 animal was infected with all 3 species. All mixed infections were detected in rodents (order Rodentia): 78% (14/18) in R. rattus rats, 11% (2/18) in R. norvegicus rats, and the remaining 2 in endemic Nesomys rufus mice and Eliurus minor rats. After characterizing mixed infections, the proportion of R. rattus rats infected with L. borgpetersenii nearly doubled (Figure), and the number of L. mayottensis–infected R. rattus rats equaled the number of L. mayottensis–infected endemic hosts (n = 4). All of the L. mayottensis–infected R. rattus rats were captured at sites with human habitation; 75% (3/4) of L. mayottensis–infected endemic hosts were captured at forest sites. The L. interrogans lfb1 genotype most commonly identified was identical to the lfb1 sequence obtained from a human with a case of leptospirosis contracted in Madagascar (9).

We present definitive molecular evidence that small mammal hosts carry mixed infections of pathogenic Leptospira spp. The characterization of mixed infections and testing of sympatric endemic and invasive reservoir hosts has altered our understanding of leptospirosis epidemiology. Previously, only L. interrogans was detected in Rattus spp. rats in Madagascar (11). We show that, when mixed infections are characterized, the prevalence of L. borgpetersenii and L. interrogans in R. rattus rats is similar. Similar to findings from Mayotte (14), R. rattus rats are a potential source of human infection for 3 of the 4 Leptospira species present in Madagascar. Because of the high abundance and widespread distribution of these rats, they could act as a key reservoir for Leptospira, including for L. mayottensis, which might occur as spillover infections from endemic species.

The high prevalence of mixed Leptospira infections also provides a potential explanation for the genetic and serologic diversity of pathogenic Leptospira in the region (5,7), considering horizontal genetic transfer has been implicated in Leptospira evolution, including evolution of the locus responsible for serologic classification (rfb) (3). Further work is needed to better characterize the evolutionary and landscape-scale epidemiologic consequences of mixed infections. Moreover, the prevalence of infection and the identification of an lfb1 genotype in Rattus spp. rats identical to that in a human case (9) suggests that rodentborne transmission of Leptospira might be an underreported health problem in Madagascar. Studies on human exposure are urgently needed.