Rapid Leptospira identification by direct sequencing of the diagnostic PCR products in New Caledonia

Background Most of the current knowledge of leptospirosis epidemiology originates from serological results obtained with the reference Microscopic Agglutination Test (MAT). However, inconsistencies and weaknesses of this diagnostic technique are evident. A growing use of PCR has improved the early diagnosis of leptospirosis but a drawback is that it cannot provide information on the infecting Leptospira strain which provides important epidemiologic data. Our work is aimed at evaluating if the sequence polymorphism of diagnostic PCR products could be used to identify the infecting Leptospira strains in the New Caledonian environment. Results Both the lfb1 and secY diagnostic PCR products displayed a sequence polymorphism that could prove useful in presumptively identifying the infecting leptospire. Using both this polymorphism and MLST results with New Caledonian isolates and clinical samples, we confirmed the epidemiological relevance of the sequence-based identification of Leptospira strains. Additionally, we identified one cluster of L. interrogans that contained no reference strain and one cluster of L. borgpetersenii found only in the introduced Rusa deer Cervus timorensis russa that is its probable reservoir. Conclusions The sequence polymorphism of diagnostic PCR products proved useful in presumptively identifying the infecting Leptospira strains. This could contribute to a better understanding of leptospirosis epidemiology by providing epidemiological information that cannot be directly attained from the use of PCR as an early diagnostic test for leptospirosis.


Background
Leptospirosis is recognized as the most widespread zoonosis worldwide [1]. It can be a lethal disease with high endemicity in the tropics. However, epidemics have also been described, most frequently associated with particular meteorological events [2,3].
The epidemiology of leptospirosis has classically been described on the basis of serological data, an indirect biomarker, using the Microscopic Agglutination Test (MAT), a technique regarded so far as the "gold standard" for identifying the infecting serovar from human or animal sera [1,4]. MAT results have provided epidemiologically important data allowing the identification of the infection sources or reservoirs and have largely contributed to the current knowledge of leptospirosis epidemiology. However, MAT is not without weaknesses and was notably shown to be a poor predictor of the infection serovar [5].
The taxonomy of the genus Leptospira has now been clarified from genetics and leptospirosis can now be studied using genetic tools, when isolates are available [6,7]. Similarly, leptospirosis diagnosis increasingly relies on PCR results [3], where a single positive sample provides a certainty diagnosis before serological conversion [4]. This frequently results in the loss of the serologybased identification of the infecting strains, which is epidemiologically important to identify the reservoirs. Therefore, the increased use of PCR has greatly improved the early diagnosis of leptospirosis, but paradoxically restricts data available for epidemiological surveillance. Yet, because the genetic tools implemented provide an insight into the genome of the infecting strain, epidemiologically relevant information might be deduced from sequence polymorphisms of the diagnostic PCR products. This approach was notably suggested and evaluated by Victoria et al. [8] while studying the phylogeny of the S10-spc-a locus: these authors demonstrated that this locus is highly conserved and a useful phylogenic target. They additionally suggested a short 245 bp region of secY as a suitable target for diagnosing leptospirosis by PCR, the sequence of the diagnostic PCR product then being epidemiologically informative. Actually, a diagnostic PCR using this target was later designed, validated according to international guidelines and confirmed to provide an epidemiologically relevant phylogeny [9].
The qPCR technique used for leptospirosis diagnosis in New Caledonia amplifies a 331pb DNA fragment within the lfb1 gene, which sequence polymorphism allows the identification of the species of the infecting Leptospira strain using melting curve analysis [15].
The Multi Locus Sequence Typing (MLST) technique uses sequence polymorphisms of multiple housekeeping genes for isolate characterization and to investigate evolutionary relationships among closely-related bacteria. It is increasingly considered as the gold standard typing method, at least in species where sufficient sequence polymorphisms exists in housekeeping genes, because it relies on sequence data that are exchangeable and independent of the analytical platform [16,17]. This technique, successfully applied to a number of bacterial pathogens, was notably recently applied to the study of leptospires: various typing schemes based on the comparison of 2855-3165 bp concatenated sequences of housekeeping genes were proposed [18][19][20] and evaluated over Leptospira spp. reference strains and isolates.
Because of the limited mammal diversity in New Caledonia, we hypothesized that a limited diversity of pathogenic Leptospira strains would be present and aimed at evaluating if the sequence polymorphism of diagnostic PCR products would allow the identification of the infecting Leptospira. To better investigate this hypothesis and the epidemiology of leptospirosis in New Caledonia, we also performed a MLST study on a collection of isolates and evaluated its direct feasibility using leptospirosis patients' serum DNA extracts. Additionally, extracts from Leptospira-infected deer kidneys contributed to a better description of the Leptospira strains currently involved in leptospirosis in New Caledonia.

Bacterial strains
The strains studied were collected from 1989 to 2000 throughout mainland New-Caledonia. Eighteen were isolates from patients' blood received at Institut Pasteur for diagnosis purpose, and 2 were isolated from deer in 1992, kindly provided by the New Caledonian Reference Veterinary Laboratory. Previously studied VNTR (Variable Nucleotide Tandem Repeat) profiles and serological identification of these isolates [13] allowed the selection of isolates from the 4 different serovars identified in our collection. The list of the isolates, their serological and VNTR-based identifications are presented in Table 1.

Clinical specimens
Clinical samples (sera) routinely received at Institut Pasteur in Nouméa, for the diagnosis of leptospirosis were also included in the study. We studied 88 human PCR positive sera collected from January 2008 to February 2010. Twelve PCR-positive deer kidney samples collected in 2010 during a sampling campaign in a slaughterhouse were also included. The 27 human samples used for drawing phylogenic trees are summarized in Table 2.

DNA extraction
For human samples, total DNA from serum (200 μl) was extracted using an automatic method on an EasyMAG apparatus (Biomerieux). For bacterial cultures and animal samples, total DNA from a culture pellet, or kidney (ca. 25 mg) was extracted using the QIAamp DNA minikit (Qiagen) following the manufacturer's instructions.

PCR analysis
The real time PCR routinely used for leptospirosis diagnosis targets the lfb1 gene as described by Mérien et al. [15] and was run on a LightCycler LC 2.0 using the LightCycler FastStart DNA Master SYBR Green I kit (Roche Applied Science, New Zealand).
The secY gene was also amplified using PCR conditions previously described [9,18] or combinations of forward and reverse primers of these 2 techniques. The recently described diagnostic PCR used the cycling conditions described by the authors [9], except that it was performed with the LightCycler FastStart DNA Master SYBR Green I kit on a LightCycler 2.0 apparatus and that the number of amplification cycles was increased to 50.

Detection of PCR products
The amplification products were directly analyzed using the LightCycler software and/or visualized by gel electrophoresis in a 1.2% agarose gel stained with GelRed Nucleic Acid Gel Stain 1× (Biotium).

DNA Sequencing
PCR products were purified using the MinElute PCR Purification Kit or MinElute Gel Extraction Kit (Qiagen) according to the manufacturer's instructions. Purified PCR products were directly sequenced in both forward and reverse directions using the same primers as for PCR using the ABI BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems) with the following modifications: Each 20 μl reaction contained 0.0625× ready reaction premix, 1× BigDye sequencing buffer, 3.2 pmol forward or reverse primer, 5-10 ng DNA and ddH 2 O. Cycle sequencing was performed using initial denaturation at 96°C for 1 minute followed by 60 cycles of 10 seconds at 96°C, 5 seconds at 50°C and 4 minutes at 60°C in a GeneAmp PCR System 9700 (Applied Biosystems). The sequencing products were purified on home-made Sephadex G-50 (Pharmacia) columns in Multiscreen filter plates (Millipore) and sequenced on an ABI 3730 × l automated sequencer.
Assembly, editing and finishing of the sequences using both the forward and reverse reaction results were made using the Staden Package [21]. DNA sequences from reference strains of relevant serovars were retrieved from http://www.mlst.net [20], from LepBank [22] or from GenBank. GenBank accession numbers of the sequences obtained from these isolates are provided as additional file 1 Table S1.

PCR results on clinical isolates
All 7 PCRs described for the MLST scheme by Thaipadungpanit et al. [20] successfully amplified a product of the expected size with DNA from all isolates belonging to the species L. interrogans. However, for some isolates, the annealing temperature for amplifying mreA had to GenBank accession numbers of the sequences obtained from these specimens are provided in additional file 1 Table S2.
be lowered down to 45°C to obtain a successful amplification. For L. borgpetersenii isolates, only pntA and glmU could successfully be amplified. The secY product used by Ahmed et al. [18] was successfully amplified from all isolates, either L. interrogans or L. borgpetersenii. Using the diagnostic PCRs, lfb1 was amplified with extracts from human sera or deer kidney with leptospires concentration equal to or lower than 50 per ml or per mg, respectively. The secY diagnostic PCR product could be amplified from clinical samples containing down to ca. 60 leptospires/ml on our qPCR platform. glmU and pntA were successfully amplified from clinical specimens containing ≥ ca. 200 leptospires per ml using either DNA polymerase tested.

Diagnostic PCR product-deduced phylogeny
We aimed at evaluating if the direct sequencing of a diagnostic PCR product could also allow the putative identification of the infecting strain. Early diagnosis of human leptospirosis in New Caledonia relies on the lfb1 PCR [15]. Therefore, the lfb1 diagnostic PCR products of the collection isolates, from patients recruited between January 2008 and February 2010 and from deer kidneys sampled in 2010 were directly sequenced. lfb1 sequences of reference strains retrieved from GenBank were also included and aligned. This allowed the construction of an lfb1-based phylogeny, supported by a 222 bp sequence. This allowed the distinction of 2 clusters among New Caledonian L. borgpeterseniiinfected clinical samples, one including references sequences of the serovars Sejroe and Castellonis, the other including the sequence of the reference strain of Hardjo-bovis respectively. These results are summarized in Figure 1 and Table 2 and 4. Among L. interrogans-infected clinical samples, five clusters were evidenced: one cluster included the reference strains of the serovars Icterohaemorragiae, Copenhageni and Pyrogenes (later named cluster L. interrogans 1), one cluster included reference strains of the serovars Lai, Australis and Autumnalis (named cluster L. interrogans 2), one cluster included the reference strain of the serovar Bataviae (cluster L. interrogans 3), one cluster included reference strains of the serovars Canicola and Pomona (cluster L. interrogans 4); lastly, one cluster included no reference sequence of any known serovar (later named L. interrogans 5).
We also evaluated if the direct sequencing of the secY diagnostic product [9] could confirm the existence of the different clusters identified using lfb1 polymorphism (Figure 2). The 202 bp PCR product could successfully be amplified and sequenced from DNA extracted from all isolates. Using DNA from clinical specimens, samples from both lfb1-deduced clusters of L. borgpetersenii were successfully amplified and sequenced, but only samples from 3 out of the 5 lfb1-deduced clusters of L. interrogans could be amplified (clusters L. interrogans 1, 4 and 5). However, samples from the two remaining clusters (clusters L. interrogans 2 and 3) were scarce (see Table 4) and had low Leptospira concentrations (see Table 2). secY products using DNA from these clinical specimens could not be generated, even using combinations of primers used for the MLST study [18] and for diagnosis [9]. However, the phylogeny deduced from a 174 bp alignment of the diagnostic secY product confirmed the clusters identified by both the MLST and lfb1 typing schemes. Strains from cluster L. interrogans 5 had sequences 100% identical to L. interrogans Hardjo-prajitno (strain Hardjoprajitno) and to L. meyeri serovar Perameles strain Bandicoot, a strain recently re-assigned to the species L. interrogans [25]. GenBank accession numbers of the sequences generated and used in this study are provided as additional file 1 Tables S1 and S2.

MLST-deduced phylogeny
DNA sequences retrieved from databases or sequenced from products successfully amplified were concatenated and allowed drawing a phylogeny of the New Caledonian isolates, together with reference strains (Figure 3). GenBank accession numbers of the sequences generated and used in this study are provided as additional file 1 Tables S1 and S2. Because the DNA sequences of the same reference strains were not available, secY sequences were not concatenated with the 7 genes used by Thaipadungpanit et al. [20]. The phylogeny was established independently for L. interrogans strains and isolates (7 genes providing a concatenate sequence of 3155 bp) and for L. borgpetersenii (2 genes for a total concatenate sequence of 968 bp). Both phylogenies are presented in Figure 3a and 3b respectively. These results evidenced three clusters among the L. interrogans New Caledonian isolates and two clusters among L. borgpetersenii isolates. Based on sequences of reference isolates available in databases, these clusters could putatively be assigned to a few serogroups. Among L. interrogans isolates, one cluster could correspond to serovars Pomona, Canicola, Pyrogenes or Hebdomadis, another one to the serovar Icterohaemorragiae or Copenhageni. Lastly, one L. interrogans cluster did not match to any known reference strain. Among L. borgpetersenii isolates, one clustered with L. borgpetersenii Hardjo-bovis JB197, whereas four other isolates clustered together, but no publicly available sequence allowed putatively identifying this cluster.

Direct MLST from clinical specimens
To further confirm the existence of the 5 L. interrogans clusters identified with lfb1 polymorphism on clinical samples, we tried to amplify and sequence glmU and pntA from these clinical samples, using the MLST primers and PCR conditions. Actually, these 2 genes are correctly amplified from isolates belonging to both L. interrogans and L. borgpetersenii species and their polymorphism allows discriminating the same clusters within New Caledonian L. interrogans isolates as the 7 genes do (data not shown). When using L. interrogansinfected clinical specimens, these two genes were successfully amplified from samples infected with ≥ ca. 200 leptospires per ml.

Discussion
While studying the sequence polymorphism of our diagnostic lfb1 qPCR product [15] in clinical specimens and a collection of isolates, we identified 2 L. borgpetersenii clusters and 5 L. interrogans clusters (Figure 1). Interestingly, one L. interrogans cluster (cluster 5) contained only sequences from human clinical specimens and did not include any known sequence of a reference strain, even after extensive searches in public databases. In order to confirm these presumptive identifications and to try to identify the cluster 5, we then conducted a MLST study using a collection of Leptospira isolated in the 1989-2000 period from human leptospirosis cases in New Caledonia, together with two isolates from deer kindly provided by the Veterinary Reference Laboratory. This MLST study similarly evidenced (Figure 3) three clusters of L. interrogans (corresponding to isolates grouped in L. interrogans clusters 1, 4 and 5). The clustering of isolates was in agreement with the lfb1-derived phylogeny. This result suggests that in the New Caledonian context, these lfb1-derived L. interrogans clusters are monophyletic and probably each correspond to a single serovar. Again, L. interrogans cluster 5 did not contain any sequence of a known reference isolate, suggesting that it might correspond to a serovar not yet described, or at least not included in public sequence databases. Though the MLST phylogeny suggests that strains from this latter cluster could be related to the serovar Australis, seroconversions observed in New Caledonian patients infected with this strain merely point to Pyrogenes, a serogroup regarded as serologically related to Australis (data not shown). Whether this cluster corresponds to a serovar not yet described or to a serovar described but which corresponding gene sequences have not been published remains to be studied.
To further identify L. interrogans clusters 2 and 3 and to evaluate the feasibility of direct MLST from clinical specimen DNA extracts, we then tried to evaluate the sequence polymorphism of the MLST targets using these clinical samples. Unfortunately, though both glmU and pntA could successfully be amplified and sequenced from extracts of patients containing ca. 200 leptospires per serum ml or more, none of the patients identified in these 2 clusters had leptospiraemia higher than 50 leptospires per ml. Interestingly, none of the isolate of our collection had lfb1 sequences identical to any of these two clusters. Because our isolate collection contains only strains collected until the year 2000, it cannot be known whether strains from these clusters were present in New Caledonia before 2001. They most probably already represented a limited part of the human cases during this earlier period, as suggested by their low incidence over more than 2 years from 2008-february 2010 (see Table 4). It can also be hypothesized that strains from these clusters are of limited virulence to humans, therefore only associated with low leptospiraemia and would therefore seldom be evidenced, either by cultures (before 2001) or PCR (after 2001).
Within L. borgpetersenii isolates, only two of the seven genes used in the MLST study of L. interrogans could be amplified. Actually, the set of primers used here was described by Thaipadungpanit et al [20] for use in L. interrogans isolates and was not supposed to amplify these genes in isolates from other species. Other MLST schemes have been used over a wider range of Leptospira species [18,19]. These could have allowed a better typing of New Caledonian L. borgpetersenii isolates or clinical specimens. An ongoing program aimed at sequencing the complete genomes of a very large number of pathogenic Leptospira isolates (Vinetz J., com. pers.) will allow the selection of the most appropriate targets and to design primers for MLST studies addressing other Leptospira species. The phylogeny deduced from the sequence of these 2 genes evidenced two clusters of L. borgpetersenii, one including the fullysequenced L. borgpetersenii serovar Hardjo-bovis [26], the other one containing no reference sequence. Again, these clusters were in agreement with the clusters derived from the lfb1-based phylogeny. Interestingly, sequences from the cluster containing the Hardjo-bovis reference strain were found only in deer and none of the 88 human clinical samples evidenced this sequence. This suggests that the introduced deer C. timorensis russa might be a reservoir for this Leptospira strain.
Other gene phylogenies have been studied, demonstrating that these genes might be sequenced to more precisely identify Leptospira strains, notably ligB [27], rpoB [28] and secY [8,9,18]. However, though they might prove useful in MLST or other phylogeny studies, most of them can currently only be used when sufficient amounts of DNA of the infecting strain is available, because no high-sensitivity diagnostic PCR was validated using these gene targets. However, a secY-based diagnostic PCR was recently described [9] and the sequence polymorphism of the gene segment amplified was validated as a relevant phylogenic tool [8,9]. Therefore, we evaluated if the phylogeny of clinical specimens using this target would confirm the ones obtained with both MLST and the lfb1 sequence polymorphism, and notably confirm and provide a more precise identification of L. interrogans clusters 2 and 3. The secY-derived phylogeny was in agreement with both the MLST and the lfb1-derived phylogenies and identified the same clusters ( Figure 2). However, L. interrogans clusters 2 and 3 that were only evidenced by lfb1 polymorphism from clinical specimens could not be confirmed because no secY PCR product could be amplified from any of these specimens. Whether this was due to the low leptospiraemia of the corresponding patients (see Table 2) and using a different qPCR platform and different PCR reagents from the ones described by Ahmed et al. [9] or to primer mismatch in the corresponding DNAs remains unknown. Interestingly, L. interrogans cluster 5 had a secY sequence identical to L. meyeri serovar Perameles strain Bandicoot (a strain recently reassigned to the species L. interrogans [25]) and L. interrogans serovar Hardjo strain Hardjoprajitno. However, this identity was not confirmed by MLST or lfb1 sequences.

Conclusions
Using a combination of MLST and other sequence polymorphisms, we evidenced 7 different Leptospira genovars belonging to both L. interrogans and L. borgpetersenii. They would correspond to at least 7 strains currently circulating in New Caledonia, should two or more strains not be discriminated by this typing scheme. Within these 7 putative strains, one was presumptively identified as L. borgpetersenii Hardjo-bovis and could be found only in deer, which might constitute its reservoir. Because deer hunting is a highly frequent practice in New Caledonia both for leisure and subsistence and it can be assumed that hundreds of people are exposed to deer kidneys weekly (frequently bare foot and with no protective gloves), this suggests that this strain is either poorly transmitted, as discussed in light of its genome reduction [26], or of low virulence to humans. We also identified a L. interrogans strain (cluster 5) that could not be related to any known reference strain. Though its secY sequence suggests that it could be related to known reference strains (L. interrogans -formerly L. meyeri-sv. Perameles strain Bandicoot and L. interrogans sv. Hardjo strain Hardjoprajitno), the more precise MLST sequence polymorphism contradicts this identification. These strains could therefore correspond to a serovar not yet described. We directly amplified two genes of the MLST scheme using extracts from human clinical specimens with leptospiraemia of 200 leptospires per ml or higher. It might therefore be possible to conduct MLST studies directly from clinical specimens if selecting samples with leptospiraemia equal to or higher than 200/ml. Lastly, we demonstrated that the polymorphism of our lfb1 diagnostic PCR target is able to provide epidemiologically relevant information, at least in a simple mammal biodiversity context as in New Caledonia. This approach was already proposed using another diagnostic PCR target, namely secY [9] that we also evaluated in our study. Using direct sequencing of leptospirosis diagnostic PCR products would partly offset the loss of epidemiological information resulting from the increased use of PCR in the early diagnosis of leptospirosis. This direct typing is currently used in New Caledonia, to better identify the different reservoirs of these Leptospira strains. The major mammal species are currently being sampled, in order to better decipher the circulation schemes and reservoirs and adapt prevention measures.

Additional material
Additional file 1: Tables S1 and S2. GenBank Accession Numbers of the nucleotide sequences used in this study.