Direct Detection and Differentiation of Pathogenic Leptospira Species Using a Multi-Gene Targeted Real Time PCR Approach

Leptospirosis is a growing public and veterinary health concern caused by pathogenic species of Leptospira. Rapid and reliable laboratory tests for the direct detection of leptospiral infections in animals are in high demand not only to improve diagnosis but also for understanding the epidemiology of the disease. In this work we describe a novel and simple TaqMan-based multi-gene targeted real-time PCR approach able to detect and differentiate Leptospira interrogans, L. kirschneri, L. borgpeteresenii and L. noguchii, which constitute the veterinary most relevant pathogenic species of Leptospira. The method uses sets of species-specific probes, and respective flanking primers, designed from ompL1 and secY gene sequences. To monitor the presence of inhibitors, a duplex amplification assay targeting both the mammal β-actin and the leptospiral lipL32 genes was implemented. The analytical sensitivity of all primer and probe sets was estimated to be <10 genome equivalents (GE) in the reaction mixture. Application of the amplification reactions on genomic DNA from a variety of pathogenic and non-pathogenic Leptospira strains and other non-related bacteria revealed a 100% analytical specificity. Additionally, pathogenic leptospires were successfully detected in five out of 29 tissue samples from animals (Mus spp., Rattus spp., Dolichotis patagonum and Sus domesticus). Two samples were infected with L. borgpetersenii, two with L. interrogans and one with L. kirschneri. The possibility to detect and identify these pathogenic agents to the species level in domestic and wildlife animals reinforces the diagnostic information and will enhance our understanding of the epidemiology of leptopirosis.


Introduction
Leptospirosis is a growing and underestimated public health and veterinary concern, caused by pathogenic spirochetes belonging to the family Leptospiracea, genus Leptospira [1,2]. The disease is an important cause of abortion, stillbirths, infertility, poor milk production and death amongst livestock, harboring a significant economic impact [3][4][5]. Its transmission requires circulation of the agents among domestic and wild animal reservoirs, with rodents recognized as the most important sources that establish persistent renal carriage and urinary shedding of Leptospira. Humans are incidental hosts acquiring a systemic infection upon direct or indirect exposure to the urine, blood or tissue of an infected animal. Farmers, veterinarians, sewer workers, pet keepers, rodent catchers and those persons participating in aquatic leisure activities are more prone to acquire the disease.
Conventional classification of Leptospira is based on serological criteria, using the serovar as the basic taxon. To date over 250 pathogenic serovars separated into 25 serogroups are known [6]. The serological classification system is complemented by a genotypic one, in which 21 genetic species are currently recognized, including pathogenic, intermediate and non-pathogenic (or saprophytic) species [7][8][9][10]. Genetic species boundaries hardly correlate with the serological classification [8].
Serological approaches are used commonly for diagnosis of leptospirosis in animals. The reference method is the Microscopic Agglutination Test (MAT), which has the advantage of being specific for serogroups [3] but has several drawbacks of being laborious and requiring a panel of viable Leptospira cultures. Isolation of leptospires, from suspect clinical specimens, constitutes the definitive diagnosis but is also technically demanding, time consuming and subject to contamination and high rates of failure [4]. Isolates are traditionally classified to the serovar level by the Cross Agglutinin Absorption Test (CAAT) [8] which is cumbersome for routine use and is only performed in a few reference laboratories worldwide.
Rapid and reliable laboratory tests for the direct detection of leptospiral infections in animals are in high demand, particularly to support suitable control measures. Serology does not corroborate well with the presence of pathogenic viable leptospires in the kidneys or urine and detection of the agents is necessary to identify healthy animal carriers. Molecular-based assays have been previously described for detecting leptospires in clinical samples. Most approaches are PCR-based and target specific genes or polymorphisms in the genome of pathogenic leptospires. Several real time PCR assays have been described predominantly for use with human samples such as whole-blood, serum or urine [11][12][13][14][15][16][17] but only few have been plentifully validated [18,19]. A few assays were evaluated or used for detecting Leptospira in kidney tissue, blood, urine and other clinical specimens from animals such as sheep [20], dogs [21,22], pigs [5], deer [23], flying foxes [24] and rodents [25,26]. Most assays rely on SYBR green detection chemistry and only differentiate between pathogenic and nonpathogenic leptospires, lacking the ability to distinguish between different species. Nevertheless, speciation of infecting Leptospira from clinical material may be important for determining the clinical significance, the probable source of infection, to distinguish sporadic cases from possible outbreaks and to better access the epidemiology of the disease.
In the present work we have developed a novel and simple TaqMan-based multi-gene targeted real-time PCR approach yielding high sensitivity and specificity for the direct detection and differentiation of the most relevant pathogenic Leptospira species in animal samples, suitable for introduction into the routine diagnostics of veterinary laboratories.

Bacterial strains
Eighty five reference strains and clinical and environmental isolates of Leptospira spp. belonging to pathogenic, intermediate and non-pathogenic phylogenetic clades were used in this study (Table 1)

Spiked tissue samples
A sample of kidney tissue from a bovine was used for testing as spiked sample. The kidney was acquired from a local official slaughterhouse (Raporal, Portugal), obtained from a bovine intended for normal human consumption, with no signs of leptospirosis. The bovine was not killed specifically for the purpose of this study. Approximately 200 mg portions of kidney tissue were excised with a sterile scalpel and homogenized with 5 ml of PBS buffer in a sterile plastic bag (Whirl-Pak bags) using a stomacher lab-blender. Kidney samples were individually spiked with the following strains, in order to determine the analytical detection sensitivity: Leptospira interrogans (serovar Autumnalis, strain Akiyami), L. kirschneri (serovar Mozdok, strain Portugal 1990) [28], L. noguchii (serovar Panama, strain CZ 214K) and L. borgpetersenii (serovar Tarassovi, strain Mitis Johnson). All the strains were grown at 29uC and the concentrations of leptospires were determined using a Petroff-Hausser counting chamber and adjusted to 10 8 cells/ml with PBS buffer. For each strain, tenfold serial dilutions from 10 7 to 10 0 cells/ml were prepared in PBS buffer and 0.1 ml aliquots were used to spike 0.9 ml of tissue homogenates. Tissue homogenate spiked with 0.1 ml PBS buffer was used as negative control. DNA extraction was performed as described in the paragraph ''Genomic DNA extraction'' below.

Tissue samples
INIAV IP is the Portuguese Reference Laboratory for animal diseases and provides diagnostic services to national veterinary authorities and private clients. Twenty seven dead wild rodents (25 Mus spp. and 2 Rattus spp.) were sent to the INIAV laboratory during the year 2011 for analysis and further used in this study ( Table 2). The rodents were captured in the Lisbon Zoo under routine operations for rodent population control, by the local veterinary authorities. No animals were sacrificed for the only purposes of research. Additionally, a Patagonian mara (Dolichotis patagonum), also from the zoo, and a swine (Sus domesticus) stillbirth fetus, from a private client, both suspect of dying with leptospirosis, were submitted for analysis to our reference laboratory and later included in this study ( Table 2). On arrival to the laboratory, animals were given a reference number and sent to the pathology where kidney, liver and/or lung tissue samples were collected. Specimens were then analysed using culture-based methods according to the OIE standard procedures for leptospirosis [27]. Briefly, specimens were aseptically collected at necropsy, immediately emulsified in sterile buffered saline solution in a 10% tissue suspension, two to three drops were inoculated in a first tube of medium and two more tubes were similarly inoculated with increasing 10-fold dilutions of the tissue suspension. For the tissue culture, a semisolid Leptospira EMJH medium was used by adding 0.1% agar to commercial EMJH (Difco), to which rabbit serum (0.4%) and 5-Fluorouracil (100 mg/ml) were further added [27].
DNA was extracted directly from tissues homogenates as described below.

Genomic DNA extraction
Genomic DNA was extracted from both bacterial liquid cultures and tissue homogenates using the QIAamp DNA extraction kit according to the manufacturer's instructions (Qiagen, Hilden, Germany), with a final elution volume of 200 ml. The DNA concentration from the pure cultures was estimated spectrophotometrically using a Nanodrop 1000 spectrophotometer (Nanodrop Technologies, Wilmington, DE) and standardized to a concentration of 10 4 genome equivalents (GE)/ml for use in the reactions. The number of GE was estimated using an average Table 1. Leptospira strains used in the present study and results of the real time PCR assays using the species-specific probes and flanking primers.   genome size of 4.6 Mb [29]. Genomic DNA suspensions were stored at 220uC until further use.

Design of TaqMan probes and flanking primers
DNA sequences of representative strains and species of Leptospira were retrieved from NCBI-GenBank and aligned using the ClustalW algorithm implemented in the program MegAlign (vers. 5.03) (DNAStar, USA). Primers and dual labeled hydrolysis probes (TaqMan probes) were designed to target selected speciesspecific genetic polymorphisms of the following pathogenic Leptospira spp.: L. interrogans, L. borgpetersenii, L. kirschneri and L. noguchii (Table 3). Probes and primers specificities were assessed in silico using the BLAST tools from NCBI-GenBank. All probes and primers were synthesized by MWG Biotech (Germany).

Real-time PCR assays
We have implemented the following assay format for testing DNA templates extracted from biological samples: (i) a first duplex amplification step aiming the detection of pathogenic Leptospira spp. (by targeting the leptospiral lipL32 gene; Table 3) and including an internal control to monitor the presence of potential amplification inhibitors (by targeting the mammal b-actin gene; Table 3); (ii) if pathogenic leptospires are detected in the first reaction, these may be further discriminated by testing each of the L. interrogans, L. borgpetersenii, L. kirschneri and L. noguchii targeted probes/primers ( Table 3). The CFX96 real-time PCR detection system (Bio-Rad, USA) was used for all assays. The amplification reactions were optimized individually for all the probes and associated primers using the SsoFast Probes Supermix (Bio-Rad, USA), according to the manufacturer's instructions. Each reaction was conducted in a total volume of 20 ml consisting of 16 SsoFast Probes Supermix, 400 nM of each primer, 150 nM of TaqMan probe, DNase free water (GIBCO) and 5 ml of DNA template solution (extracted from pure cultures or tissues samples). Non-template negative controls (with PCR grade water) were included in each run to rule out the possibility of crosscontamination. The assay thermal conditions were as follows: 95uC for 2 min, followed by 45 cycles of 5 s at 95uC and 15 s at the optimized annealing temperature for each probe ( Table 3). The thermal cycling conditions for the duplex amplification targeting b-actin and lipL32 were 95uC for 2 min, followed by 45 cycles of 5 s at 95uC and 35 s at 60uC. Reproducibility of the assays was assessed by repeating the assays at least twice. Data analyses were performed by the detection system of the real-time PCR equipment, according to the manufacturer's instructions.

Analytical specificity and sensitivity
In order to determine if each set of probe and associated primers was specific for the respective Leptospira target species, the amplification assays were tested on DNA templates extracted from different strains belonging to pathogenic, intermediate and nonpathogenic Leptospira species (Table 1), and from other nonrelated bacteria previously mentioned in ''bacterial strains'' section. The analytical sensitivity of the amplification assays (limits of detection -LODs) were determined using 10-fold serial dilutions of genomic DNA extracted from pure cultures of L. interrogans (serovar Autumnalis, strain Akiyami), L. kirschneri (serovar Mozdok, strain Portugal 1990), L. noguchii (serovar Panama, strain CZ 214K) and L. borgpetersenii (serovar Tarassovi, strain Mitis Johnson). LODs on tissue samples were assessed using DNA extracted from the serially diluted spiked macerates. Each template was tested in triplicate.

Sequencing
Leptospira isolates obtained from tissue samples were identified by comparative sequence analysis of a 245 bp region of the secY gene, as described by Victoria et al. [30]. Briefly, the region of interest was amplified using primers SecYII (59-GAA TTT CTC  TTT TGA TCT TCG-39) and SecYIV (59-GAG TTA GAG  CTC AAA TCT AAG-39), which amplify secY sequences from all pathogenic strains of Leptospira. PCR amplifications were performed on a C1000 thermocycler (Bio-Rad) using the following program: an initial step of denaturation for 5 min at 95uC, followed by 34 cycles consisting of annealing, 45 sec at 54uC, extension, 2 min at 72uC, and denaturation, 30 sec at 94uC. Nucleotide sequences were determined, using the same primers, by commercially available sequencing services. Nucleotide sequence analysis and comparison with other relevant reference sequences were performed using the BLAST suite at NCBI-GenBank and aligned using Clustal X or MEGA software (version 5.0).

Design of probes and primers
Species-specific sets of primers and probes targeting L. interrogans, L. borgpetersenii, L. kirschneri and L. noguchii are listed in Table 3. As shown in Figures S1, S2, S3 and S4 in File S1, these sets of probes and primers contained sufficient polymorphisms to warrant 'in silico' species specific amplification.

Analytical specificity and sensitivity
Execution of the PCRs on DNA extracted from various bacteria, revealed a highly specific amplification from any of the pathogenic strains belonging to the respective target Leptospira spp., i.e. L. interrogans, L. kirschneri, L borgpetersenii and L. noguchii. None of the other strains yielded a positive amplification reaction (Table 1; Fig. 1A). The analytical sensitivity (LOD) of the amplification assays were found to be between 1 and 10 genome copies in the PCR mixture for each probe and primer set.

Spiked tissue samples
The LOD of the PCRs on spiked tissue samples was similar for all probe/primers sets targeting the respective target species, and estimated to be 10 3 leptospires/ml of tissue homogenate (< per 20 mg of tissue) (Fig. 1B). Furthermore, the same LOD was estimated for the lipL32-targeted probe/primers when used in duplex amplification reactions with the mammal b-actin probe (not shown).

Clinical tissue samples
DNA extracted from 27 kidney samples of wild rodents were analysed with the lipL32 and mammal b-actin targeted duplex assay ( Table 2; Fig. 2A). Leptospiral DNA was detected in three samples, as demonstrated by a positive amplification of the lipL32 gene region (Table 2; Fig. 2A). Furthermore, the partial b-actin gene was amplified from all samples, showing that the PCR reactions were not significantly inhibited by potential contaminants. When tested with each of the L. interrogans, L. borgpetersenii, L. kirschneri and L. noguchii targeted probes/ primers, only these three samples showed amplification (Table 2; Fig. 2B). Two of these DNA samples were identified as L. borgpetersenii and one sample as L. interrogans. Testing a pooled sample of kidney and liver tissues from a Patagonian mara, and a lung sample from an aborted swine fetus with the duplex PCR revealed a positive amplification for both samples (Table 2). Subsequent testing with the species-specific sets of probes and Table 2. Results of the bacteriological culture and of the real time amplification assays for the tissue samples analyzed in the present study.                           primers showed that the Patagonian mara was infected with L. interrogans and the swine fetus with L. kirschneri. Leptospira isolates were only cultured from the samples that also yielded PCR-positive results, thus confirming the presence of viable leptospires (Table 2).
Molecular speciation through analysis of the partial sequences of the secY gene was in concordance with the results obtained by the species-specific PCRs. Two isolates were identified as L. borgpetersenii (from wild rodents; GenBank accession numbers KM066006 and KM066007), one as L. kirschneri (from the swine fetus; accession number KM066009) and two as L. interrogans (from a wild rodent and the Patagonian mara; accession numbers KM066008 and KM066010, respectively).

Discussion
In this work we present a two step real-time PCR strategy to infer the presence of pathogenic leptospires in clinical and veterinary samples. In the first step, we assess if an animal tissue sample is infected with a pathogenic leptospire by targeting its lipL32 gene. The lipL32 gene encodes an outer membrane lipoprotein that is confined to pathogenic Leptospira species [16]. The second step identifies the four most common and veterinary relevant pathogenic Leptospira species, L. interrogans, L. borgpetersenii, L. kirschneri and L. noguchii using dedicated sets of probes and primers.
Probes and flanking primers were developed by in silico analysis and further tested for their practical utility on DNA extracted from cultured bacteria, spiked tissues and clinical specimens. The amplification assays have proved to be specific to the respective targeted species, with no cross-reactions when non-pathogenic leptospires or other pathogens were tested. The amplification of the b-actin gene was included in the initial lipL32-based PCR to assess the presence of amplification inhibitors in tissue samples [31]. However, the abundant presence of b-actin gene copies in DNA samples extracted from tissues may ensure some amplification even when low levels of potential inhibitors are present (but amplification curves are usually weaker and anomalous). The analytical sensitivity deduced for the amplification assays, i.e. 1 to 10 GE on DNA extracted from cultured leptospires and 10 3 leptospires/ml tissue homogenate, were similar to the ones of other previous studies concerning the molecular detection of leptospires [15][16][17]19,22].
The panel of species-specific probes and flanking primers may be extended with the design of novel oligonucleotides, e.g. for use in regions where the occurrence of additional species of pathogenic leptospires is common. As far as we know, this is the first report describing a strategy capable of clearly identify four most frequently found pathogenic Leptospira species based on the use of TaqMan probes.
From 27 kidney samples of wild rodents, and samples from a Patagonian mara and a porcine fetus suspected of leptospirosis, three rodent samples and the samples from the Patagonian mara and fetus all yielded a positive PCR test for the presence of pathogenic leptospires. In concordance, these samples were also positive by culture. Culture provides proof of infection and thus is an ideal reference standard. Consequently, these results are consistent with a 100% clinical sensitivity and specificity of the PCR. Subsequent prospective analysis of a larger sample set would allow substantiating this conclusion.  Phylogenetic identification of the cultures also allowed supporting the findings obtained with the species-specific PCRs. Indeed, speciation by phylogeny was in all cases in concordance with the results obtained via the PCR method.
Initially, we anticipated that more samples would be positive by the real time PCR assay than by culture [5,[32][33][34]. Recently, Fornazari et al. [20] reported that quantitative PCR presented the highest sensitivity among several techniques to detect leptospires in tissues samples, the bacteriological culture being the least sensitive. Apparently, our procedure of culturing, using macerated fresh tissue has been highly effective. Alternatively, it cannot be excluded that the bacterial load of the tissues might have been very high. Nevertheless, the low rate of positive animals (11%) is not too discrepant from the prevalence values found in other studies where leptospiral DNA was detected in rodents tissues by PCR-based assays, which ranged from 13% to 20% [25,35,36]. Furthermore, as far as we know, the region of Lisbon, where the rodents were captured, is not usually regarded as having major leptospirosis problems [2], which may also reflect a lower prevalence of the agent in reservoirs such as wild rodents. We anticipate that our assays may be useful in studies inferring the prevalence of pathogenic leptospires in wild rodents and other animals, with the advantage of differentiating the infecting Leptospira species. The amplification assays described were able to detect pathogenic leptospires in samples of animal tissues, such as kidney or lung. Although the analysis of this kind of samples is not essential for an early diagnosis of leptospirosis, it has a great value in situations such as epidemiological and post-mortem investigations. The last situation is very well illustrated in this work with the detection of pathogenic leptospires in tissues of a Patagonian mara and a swine fetus. Both animals were suspect of having leptospirosis, which was confirmed by this study. The porcine fetus was infected with a strain belonging to L. kirschneri. Pigs may be infected by several Leptospira species (and serovars) that may cause infertility, fetal death and abortion. Leptospira kirschneri has been reported but seems to be less frequently found in pigs in Portugal than other species [37]. The Patagonian mara, a relatively large rodent that lived in the local zoo, was found to be infected with L. interrogans. To our knowledge, this is the first report describing the molecular detection or the isolation of a pathogenic leptospire from that rodent, which proved to have died of leptospirosis. Zoos are often infested with rats that are notorious reservoirs of L. interrogans. We hypothesise that this Patagonian mara has been infected by rats as the primary infection reservoir, which would support the potential hazard of rodents in zoos for both (exotics) animals and public.
The amplification assay described in this work is able to indentify the four most relevant pathogenic species of Leptospira infecting farm and wild animals. While the approach can be extended to other Leptospira species, it is important to continually evaluate the specificity of previously designed probes and primers and, if necessary, modify and improve the sequences, in order to ensure an effective and specific detection and identification of the circulating Leptospira species.

Conclusions
The molecular assays presented in this work allow the detection and identification of four relevant pathogenic species of Leptospira, directly from animal tissues. The assays proved to be specific and sensitive, and much faster than the bacteriological culture, reducing the time for confirmatory leptospirosis diagnosis. The assays are amenable to future automation possibilities and will reinforce the diagnostic information and enhance our knowledge about the epidemiology of leptopirosis.

Supporting Information
File S1 Sequence alignments showing the complementary targets of the species-specific Leptospira interrogans, L. kirschneri, L. noguchii and L. borgpetersenii probes and respective flanking primers.