Fine‐scale prevalence and genetic diversity of urban small mammal‐borne pathogenic Leptospira in Africa: A spatiotemporal survey within Cotonou, Benin

Abstract Leptospirosis is a zoonotic disease that is caused by spirochete bacteria of the genus Leptospira. Around the world, one million people each year are infected, leading to 60,000 deaths. Infection occurs through contact with environmental pathogens excreted by mammals (notably rodents). Data on Leptospira and leptospirosis in Africa are rather scarce, especially in urban habitats though these appear to be favourable environments for the pathogen circulation and human contamination. Using qPCR, DNA sequencing as well as MST/VNTR approaches, we examined Leptospira occurrence and genetic diversity in 779 commensal small mammals that were sampled over 2 years in the city centre of Cotonou, Benin, from three neighbourhoods with contrasting socio‐environmental conditions. Overall prevalence reached 9.1%. However, very marked variations in both space and time were observed, with local peaks of high prevalence but no clear seasonal pattern. In most sites that could be regularly sampled, Leptospira‐positive rodents were found at least once, thus confirming the widespread circulation of the pathogen within small mammal communities of Cotonou. Interestingly, an unusual diversity of small mammal‐borne Leptospira species and genotypes was retrieved, with up to four species and three different genovars within the same neighbourhood, and even instances of two species and two genovars identified simultaneously within the same household. To our knowledge, such a high genetic diversity has never been described at such a fine scale, a fortiori in Africa and, more generally, within an urban environment. Altogether, our results underline that much remains unknown about leptospirosis as well as the associated infectious risk in African cities where the disease may be massively over‐looked.


| INTRODUC TI ON
Leptospirosis infects one million and kills 60,000 people each year (Costa et al., 2015). The etiologic agents, spirochetes bacteria of the genus Leptospira, multiply in renal tubules of animal reservoirs, especially rodents, and are subsequently excreted in the environment (reviewed in Haake & Levett, 2015). Leptospira gathers nine currently recognized human-pathogenic species and 350 serologically distinguishable serovars (review in Karpagam & Ganesh, 2020). Animals and humans become infected following contact with contaminated soils and waters Thus, breeding and water-related activities and events (e.g. rice and market garden production, water-based recreational activities, flooding episodes) are important risk factors for infection (Mwachui et al., 2015;review in Karpagam & Ganesh, 2020).
Poor and environmentally degraded urban areas such as slums have also been associated with high leptospiral prevalence, with the case of Pau da Lima favela in Salvador, Brazil, being particularly well documented (e.g. Ko et al., 1999;Reis et al., 2008;Cassanovas-Massana et al., 2018;reviewed in Cornwall, 2016). However, in this latter case, the infectious system is relatively simple, with one single rodent reservoir, Rattus norvegicus, shedding the Leptospira interrogans serovar Copenhageni that is responsible for most human leptospirosis cases (Kô et al., 1999;Tucundava da Faria et al., 2008;Costa et al., 2014). In Rio de Janeiro, Brazil, brown rats are also considered the key reservoir of the urban cycle of L. interrogans serovar Icterohaemorrhagiae that is dominant in the city (Martins & Lilenbaum, 2013). In the same manner, only one clonal genotype was found in rats from Johannesburg, South Africa (L. borgpetersenii Javanica; Moseley et al., 2020). However, in other settings, several Leptospira species and serovars may coexist, thus making the ecology of the disease probably much more complex, and the investigation of Leptospira species and serovars crucial (Guernier et al., 2017).
For instance, in the city of Lyon, France, one single species (L. interrogans) but two serovars (Icterohaemorragiae and Copenhageni) were found in Rattus norvegicus from eight different sites of the city (Ayral et al., 2015). In Malaysian towns, three species of Rattus (R. rattus, R. norvegicus and R. exulans) were found to harbour two Leptospira phylogenetic lineages and two serovars (L. borgpetersenii serovar Javanica and L. interrogans serovar Bataviae) (Benacer et al., 2016).
In the same manner, L. interrogans and L. kirschneri were identified in only three house mice, brown and black rats in Kibera slum, Nairobi, Kenya (Halliday et al., 2013). The dynamics of leptospires circulation in the environment may even rendered even more complex by coinfection events: in Madagascar, several cases of co-infections by up to three different Leptospira species (L. interrogans, L. borgpetersenii and L. mayottensis) were observed in various animal reservoirs, with a probable spill-over of L. mayottensis from endemic rodents to the invasive black rat (Moseley et al., 2018).
The coastal region of West Africa along the Guinea Gulf is experiencing an accelerated urbanization (i.e. the so-called Abidjan-Lagos corridor; Choplin, 2019), where hundreds of thousands of urban dwellers live in socio-environmentally degraded conditions. In this region, leptospiral burden, though under-documented, is expected to be important (Dobigny et al., 2018). Contrary to what is usually observed in European and American towns where only rats and mice coexist, the small mammal communities from the coastal West African city centres still harbour native species due to the ongoing but incomplete bio-invasion of cosmopolitan rodents (e.g. Garba et al., 2014). As such, south Benin cities are characterized by (i) the co-occurrence of invasive black and Norway rats with the native Mastomys natalensis as well as, to a lesser extent, Cricetomys gambianus and Praomys derooi; and (ii) the significative prevalence of the native shrew Crocidura olivieri in small mammal assemblages (>20% of all captures in Cotonou) (Hima et al., 2019;Houéménou et al., 2019). We recently provided an overview of rodent-and shrewborne leptospires diversity in this part of the country . It was demonstrated that several Leptospira species (L. interrogans, L. kirschneri and L. borgpetersenii) were circulating in both the native and invasive mammalian hosts. It was also suggested that some kind of host-specificity may exist and that local spatiotemporal variations in prevalence was important. In the current study, we build on these results by conducting a long term study, sampling at a finer scale: we sampled in three neighbourhoods of Cotonou, investigating small mammal-borne leptospires diversity and ecology at both the species and serovar levels during a period of 2 years. To our knowledge, the present study is the first one to provide data on Leptospira genotypes in Benin.
Agla is a relatively recent district with poorly sanitation (NB: it Impacts • Pathogenic Leptospira were monitored in small mammals from three socio-environmentally contrasted neighbourhoods of Cotonou city centre, Benin, during a 2 year-long survey.
• Small mammal-borne pathogenic Leptospira prevalence varies greatly in both space and time within Cotonou city, Benin, with some high local peaks but no obvious seasonal patterns.
• Unusually elevated Leptospira species and genovar diversities were observed at the city-, neighbourhoodand even household-scales, thus representing a rare example of Leptospira species and genovar coexistence at a very fine scale. currently benefits from various urban management actions that were started after the present study). Extensive parts of this wide shallows get flooded at the beginning of the great rainy season (May-July) due to rainfall accumulation. Ladji is a very poor, densely populated and insalubrious neighbourhood, which is located along the edge of Lake Nokoué. This area gets partly flooded during the second half of the great rainy season (June and July) until the end of the short rainy season (September-October) following an increase of the lake level. Saint-Jean is a formal district of colonial origin with a mix of hard-built and precarious habitats. It is not prone to flooding per se, but rainfall may create large ephemeral ponds.
Explicit oral agreements were systematically obtained with local traditional (e.g. family and household heads, shop, firm and garden owners) as well as administrative (i.e. Cotonou City Hall services, urban district chiefs) authorities before rodent trapping. None of the rodent species captured for this study has protected status according to IUCN/CITES. All animals were treated in a human manner in accordance with the guidelines of the American Society of Mammalogists (Sikes & Gannon, 2011). Field work in Benin was conducted under a research agreement signed between Republic of Benin and the French Institute of Research for Development (30th, September 2010). According to the Nagoya protocol, access and sharing of benefits was agreed by the government of the Republic of Benin (file 608/DGEFC/DCPRN/PF-APA/SA). Note that no ethical committee agreement is required in Benin to conduct research on pest animals such as those described in the present study. 18.14 while both Sherman (8 × 9 × 23 cm; ©Sherman) and locally made wire-mesh traps (10 × 10 × 25 cm) were used together for all subsequent sessions (March 2017 to June 2018). The use of these two types of traps was important since different species may be preferentially captured depending on the trap type (Garba et al., 2014).
Baits were made of a mixture of peanut butter and fish. Traps were set for three consecutive nights and checked each morning. When a trap captured an individual, it was replaced by a new one, while empty traps were rebaited.
Rodents were brought alive to the laboratory where they were always processed on the day of capture. They were euthanized using diethyl-ether and cervical dislocation. General morphology and external measurements were used to provide a preliminary taxonomic identification. Many African rodent genera can sometimes be difficult to distinguish on morphological grounds alone (Granjon & Duplantier, 2009). For this reason, a particular attention was paid to species-specific identification. To this end, all Rattus individuals were also identified using a genotyping method as well as cytochrome b DNA sequencing (Dobigny et al., 2008 andBerthier et al., 2016). In addition, all Praomys individuals, at least one Mastomys individual per site per session, the single house mouse captured (Mus musculus), as well as one giant African rat (Cricetomys) and one shrew (Crocidura) per district were unambiguously identified through cytochrome b gene sequencing from kidney-or spleen-extracted DNA (data are not shown, but see Dobigny et al., 2008, andBerthier et al., 2016, for DNA sequencing and genotyping procedures). Reference data- One kidney and spleen fragments were collected on all animals, 96° ethanol-preserved andall deposited in the CBGP sample collections, Montpellier, France (https://doi.org/10.15454/ WWNUPO), with duplicated 96° ethanol-preserved spleens also deposited at the URIB, Abomey-Calavi University, Cotonou, Benin.

| DNA extraction and qPCR-based detection of pathogenic Leptospira
Genomic DNA was extracted from one fragment of ethanolpreserved kidney tissue samples using 96-Well Plate Animal Genomic DNA Mini-Preps Kit (Biobasics © ) according to the manufacturer's recommendations. DNA was eluted using 150 μl elution buffer. Detection of pathogenic Leptospira was performed following a probe-based qPCR approach that targets a fragment of the LipL32 gene, using a LightCycler® 480 (Roche Diagnostics, France) and according to previously described protocols (Dobigny et al., 2015). All experiments were conducted in duplicate in 384-well micro-titre plates with 2 μl DNA in a 7 μl final volume for each reaction. On each qPCR plate, a standard curve was made using Leptospira interrogans serovar Icterohaemorragiae strain Verdun DNA extracted from culture (Spirochete Laboratory, Pasteur Institute of Paris, National Reference Center for Leptospirosis and WHO Collaborative Centre).
Extraction (no animal tissue) and qPCR reaction (i.e. molecular biology grade water instead of DNA) negative controls were systematically added to each plate.

| Leptospira molecular characterization
Leptospira species identification relied on 16S rDNA barcoding. To do so, amplification was performed following published protocols with slight modification (Mérien et al., 1992) Leptospira interrogans serovar Copenhageni strain Fiocruz DNA were included as negative and positive controls, respectively. Amplified products were sent for sequencing (@Genoscreen) and Leptospira sequences were aligned using genomic tools freely available on the PRABI-Gerland bioinformatic platform website (https://npsa-prabi. ibcp.fr/cgi-bin/npsa_autom at.pl?page=/NPSA/npsa_multa linan. html), and identified using Nucleotide BLAST search implemented in NCBI (http://blast.ncbi.nlm.nih.gov) as well as genetic similarity with reference Leptospira DNA sequences available in the VetAgro USC1223 Lab (see Merien et al., 1992).
Molecular typing was performed using two complementary approaches, namely the multi-locus Variable Number Tandem Repeat (VNTR) and Multi-spacer Sequence Typing (MST) methods. First, VNTR analysis was conducted on loci VNTR4, VNTR7, VNTR10 and VNTR Lb4 as described in Salaün et al. (2006). In brief, 5-7 μl of DNA were added to 45 μl of PCR mix ( © Qiagen) for PCR amplifications with 40-45 cycles. Annealing temperatures were 52°C for VNTR4 and VNTR10, 54°C for VNTR7 and 53°C for VNTRLb4. Second, samples identified as belonging to the L. interrogans serogroup Icterohaemorrhagiae by the VNTR analysis were further characterized using the discriminating MST3 locus. We relied on the protocol previously described by Zilber et al. (2014) with an annealing temperature of 50°C. Amplification products were sent for sequencing ( © Genoscreen), and the resulting sequences were compared to the MST reference database (Zilber et al., 2014). VNTR and MST3 profiles were used in combination to deduce Leptospira genotype and putative serovar assignation.
Pearson's χ 2 tests were performed to compare prevalence between host species, districts and trapping sessions. Seasonal prevalence was also compared using a Pearson's χ 2 test, with February/ March, June and October/November sessions of years 1 and 2 being pooled, but they were further investigated through a Kruskal-Wallis comparison of the prevalence measured in Agla and Ladji households between seasons (with years 1 and 2 pooled). All statistics were conducted under R v.3.5.0 (R consortium, 2018).

| Small mammal diversity and distribution
In total, 779 small mammals were collected in Agla, Ladji and Saint-Jean during six campaigns that took place between November 2016 and June 2018. Among them, Rattus rattus was the most dominant species (N = 456), followed by the shrew Crocidura olivieri (N = 164), then Mastomys natalensis (N = 88), Rattus norvegicus (N = 42), Cricetomys gambianus (N = 17), Praomys derooi (N = 11) and one single house mouse (Mus musculus). Rattus rattus was found to be abundant and widespread in all three districts, with black rats trapped at least once in 31 out of 35 (88.6%) of the prospected households (Table 1 and Table S1). In the same manner, shrews were quite numerous and present at least once in 29 (82.9%) houses from all three districts.
The native commensal Mastomys natalensis was mainly captured in Agla (70.5% of all Mastomys natalensis captures) with most individuals (79%) captured in two neighbouring households (see sites S-AGL-5 and S-AGL-5′ in Table S1). Due to the inadequate size of In the same manner, total numbers of animals captured within each district were similarly high (N = 224, N = 249 and N = 306 in Saint Jean, Ladji and Agla, respectively; Table 1). The trapping results detailed by district, session and species are provided in Table 1. Details by trapping sites (i.e. households) are provided in the Table S1.

| Leptospira prevalence in small mammals
Overall pathogenic Leptospira prevalence in small mammal was 9.1% (74/779). In 13 cases (out of 74), only one Ct value (for 'Cycle threshold', sometimes referred to as Cp, the 'Crossing point'; that is the number of amplification cycles above which fluorescence exceeds background signal) could be obtained, while Ct values could be cal-  Pearson's χ 2 test, p = 3.66e-8). Differences in the two most prevalent areas, namely Agla and Ladji, were not significant (χ 2 test, p = .14).

| DISCUSS ION
We has already been shown that mean prevalence of rodent-borne leptospires along the extensively urbanizing coastal corridor was 12.9% , with an overall prevalence in the city of Cotonou between 12% and 13% (Houéménou et al., 2013.
However, important variations in both space and time were noted, suggesting that local conditions were critical to explain Leptospira circulation in domestic and peri-domestic rodents from this area .
Our 2 year-long survey at a finer scale clearly confirms such variations within Cotonou, thus demonstrating spatial inequalities in terms of infectious risks between poor districts of the same city centre. Indeed, the non-flood-prone area of Saint-Jean appeared almost free of rodent-borne leptospires, while the two insalubrious districts of Agla and Ladji were characterized by significant levels of rodentborne Leptospira. There, the observed prevalence was moderate (mean values of 11.4% and 14.1% in Agla and Ladji, respectively; 13.2% when the two districts are pooled). However, in all sites that could be sampled during all six sessions (N = 12), Leptospira-positive rodents were found at least once, thus confirming widespread circulation of the pathogens within small mammal communities of these two districts. In addition, small mammals in these two neighbourhoods are abundant and omnipresent: overall trapping success reached 18.4% in these two areas while 98.4% (64 out of 65 sites*session) and 85.3% (58 out of 68) of households were found rodent-infested in Agla and Ladji, respectively (data are not shown).
Keeping in mind that a few dozen infected rats may shed thousands of millions of leptospires per day (Costa et al., 2015), our results in Cotonou show that, in fine, small mammals represent a potentially massive source of excreted leptospires into the urban environment.
Beyond inter-neighbourhood differences in prevalence, very local variations were observed between sometimes very close households, too. For instance, comparing only sites for which the six sampling sessions were available (i.e. trapping could be performed and small mammals were captured at each of the six sessions), the 2 year-long site-specific prevalence ranged from 5.7% to 25% in Agla, and from 7.3% to 20.6% in Ladji (in addition to one site sampled only three times and reaching a prevalence of 53.3% in the latter locality) (see Table S1). During the same session, site-specific prevalence could vary from 0% to 100%, although this was usually based on very low sample sizes (Table S1). Nevertheless, these results are in agreement with previous studies suggesting that leptospiral risks strongly depend on local fine-scale socio-environmental conditions (Maciel et al., 2008), including in southern Benin .
In some urban settings, R. norvegicus was found associated with very high Leptospira prevalence (e.g. 80.3% in Salvador, Brazil: Tucunduva de Faria et al., 2008). Accordingly, we found that Norway rats showed the highest species-specific prevalence (45.2%) in Cotonou. Furthermore, several sites where high small mammalborne Leptospira prevalence were observed also harboured R. norvegicus (e.g. S-AGL-8 or S-LAD-8; Table S1). However, the species is often associated with shallows and sewers (Dossou et al., 2015; our own observations), thus making it difficult to distinguish between the role of habitat-vs. host-related determinants in favouring the presence of rodent-borne pathogenic leptospires . Interestingly, other species were also trapped in many of the sites with positive Norway rats, and proved to be Leptospirapositive as well. In addition, Norway rats were not, or very rarely trapped in some other sites where prevalence was quite elevated essentially due to the presence many positive black rats (e.g. S-AGL-9 and S-LAD-9). Even shrews (that were present in all but two households of Agla and Ladji that could be sampled at least twice) were sometimes found Leptospira-positive in the absence of Norway rats.
In the same manner, though limited in numbers, our data show that three out of the four serovars identified were shared by Rattus rattus and R. norvegicus, and that L. interrogans serovar Icterohemorragiae was found in the two rat species as well as in Mastomys natalensis and Crocidura olivieri (Table 2b). Altogether, these patterns suggest that all small mammals share leptospires widely, and that they may all play a role in Leptospira ecology within the Cotonou urban environment. This weakens the hypothesis of partial rodent host-specificity of Leptospira species in South Benin that we recently proposed . Our new results rather point towards the importance of local environmental conditions in determining the possible presence of rodent-borne leptospires at the household's level . Landscape and health ecology approaches will be useful to further scrutinize these aspects and identify the urban landscape elements that favour Leptospira circulation within small mammal assemblages.
Part of the leptospire life cycle takes place in free or soilassociated waters where they can infect vertebrate hosts (Haake & Levett, 2015). Thus, it seems reasonable to hypothesize that their prevalence in rodents is associated with rainfall and standing waters. Accordingly, some observations were made at the scale of cities and villages of South Benin where prevalence peaks indeed seem to occur during, or 1 month after moderate (100-200 mm) monthly rainfall . Though based on a limited amount of leptospire-positive water samples collected in Cotonou, isotopic signatures also suggested that leptospires were essentially detected in pond waters formed at the beginning of the rainy season following low to moderate rainfall events (Houéménou et al., 2021). In the present study, very high prevalence was observed during March 2018, where rainfall was moderate though atypically precocious during this particular year (Table 1 and Figure 2). However, no obvious pattern was observed for the other sessions/seasons (Figure 2).
In other words, we are unable to confirm conclusively any association between leptospire circulation and (instantaneous or delayed) monthly rainfall.
An interesting result concerns the specific and genotype diversities that were quite high within limited spaces, that is the same neighbourhoods, and even within the same households. Indeed, four different leptospire species and four different genotypes were identified in Agla and Ladji -more than were previously described at the Cotonou and the whole southern Benin spatial scales (Houéménou et al., 2013 (Mgodé et al., 2015). Here, we combined MST and VNTR profiles, which proved to be quite efficient in obtaining well-discriminated genotypes. However, their complete taxonomic or serologic identification through comparison to already known profiles is very limited due to the absence of reference data from West Africa.
Second, the coexistence of bacterial strains within the same environment may lead to within-host coinfections that may be difficult to detect. First, one may imagine that different strains preferentially colonize distinct areas in the host's kidney. To our knowledge, such intra-host and tissue-level ecology of Leptospira has not been documented, thus precluding addressing this issue further. Second, we found one single L. borgpetersenii/L. interrogans coinfection in the same Mastomys natalensis individual, which echoes the seventeen double and one triple infection recently described in Madagascar (Moseley et al., 2018). However, as in many other studies, we may have under-detected such coinfection events due to our use of qPCR-and sequence-based approaches, which probably favour the quantitatively dominant strain, and fail to efficiently detect mixed infections. These caveats may represent important biases for studies on the determinants of leptospire species distribution. Yet, we believe that this aspect deserves to be investigated further due to the importance of genetic and immunologic diversity of Leptospira in the implementation of any vaccine strategy as well as their probable impact on the aetiology of the infection in both humans and animals.
The use of high throughput sequencing may be a promising way to investigate these aspects.

| CON CLUS ION
The present study confirms the moderate prevalence of pathogenic Leptospira among urban small mammals of South Benin.
However, taking into account the strong abundance and density of these reservoir hosts within cities, especially slums, as well as the recurrent flooding episodes that characterized this region, it is expected that inhabitants are at high risk of exposure for leptospirosis (Dobigny et al., 2018). Our longitudinal monitoring documented one of the highest levels of genetic diversity of small mammalborne leptospire species and genogroups that has ever been observed at such a fine-scale, with the co-occurrence of several genetic Leptospira species and strains within the same neighbourhoods and even households. This raises several issues in terms of evolutionary ecology of the pathogen (e.g. host specificity, intrahost competition, ecological preferences of each Leptospira species/strain outside the mammalian host, etc.) as well as disease management (e.g. Leptospira species-or strain-specific exposome; design of a locally adapted vaccine). We recommend that raising awareness of the public health authorities, at both the national and local levels, is done urgently, so that studies in humans (including strain typing) are organized, and treatment of leptospirosis is made possible and accessible in Benin.

ACK N OWLED G EM ENTS
This work is part of the activities conducted by the 'observatory of small mammals as indicators of environmental changes in West Africa' (ObsMICE) funded by the French Institute of Research for Sustainable Development (IRD). We are thankful to Serena Dool