Introduction

Leptospirosis is a zoonotic disease of global importance that has emerged as a major cause of morbidity and mortality among impoverished populations (^{Ko et al., 2009}). Based on global data, more than 500,000 new cases of leptospirosis are reported annually, with mortality rates exceeding 10% (^{WHO, 1999}, ²⁰⁰⁶). Multiple factors, including environmental, demographic, social, and economic factors, have contributed to the emergence of this disease, which affects a broad range of mammalian hosts, including humans, wildlife, and domestic animals (^{Bharti and Nally, 2003}; ^{Lau et al., 2012}).

The precise identification and classification of Leptospira spp. is important for epidemiological and public health surveillance (^{Mohammed et al., 2011}). Leptospires are usually classified by methods based on DNA-DNA hybridization, whereas the cross-agglutination absorption test (CAAT), which uses polyclonal antibodies against lipopolysaccharides (LPSs), has led to the definition of serovars that are today considered to be the basic systematic units of Leptospira spp. (^{Cerqueira and Picardeau, 2009}; ^{Galloway and Levett, 2010}). Serological methods for the characterization of Leptospira species are complex and costly, restricting their worldwide distribution and use (^{Ahmed et al., 2010}).

Many molecular DNA techniques have been applied to identify and classify the species and serovars of Leptospira (^{Ahmed et al., 2012}). These include restriction endonuclease analysis (REA) of chromosomal DNA (^{Marshall et al., 1981}), random amplified polymorphic DNA (RAPD) fingerprinting (^{Ramadass et al., 1997}), DNA-DNA hybridization (^{Yasuda et al., 1987}; ^{Brenner et al., 1999}), arbitrarily primed PCR (^{Ramadass et al., 2002}), pulsed-field gel electrophoresis (PFGE) (^{Galloway and Levett, 2008}) and polymerase chain reaction (PCR) of specific genes followed by restriction fragment length polymorphism analysis (RFLP) (^{Li et al., 2009}). Recently, multilocus sequence typing has been applied as an alternative to immunological methods for the identification and classification of pathogenic leptospires (^{Ahmed et al., 2006}; ^{Pavan et al., 2008}; ^{Leon et al., 2010}; ^{Ahmed et al., 2011}; ^{Boonsilp et al., 2013}). All of these techniques mentioned above have contributed significantly to the current taxonomic classification of the Leptospira genus (^{Morey et al., 2006}; ^{Slack et al., 2009}).

Quantitative DNA-DNA hybridization to measure genetic homology has been used as a reference for the classification of serovars within species (^{Yasuda et al., 1987}; ^{Perolat et al., 1998}, ^{Brenner et al., 1999}). However, this hybridization method is not routinely used for the identification of Leptospira species due its complex and laborious execution, which requires the use of radioactive isotopes and is therefore restricted to reference laboratories (^{Morey et al., 2006}). It has also been observed that some serotypes are more characteristic of a single species, while others contain both pathogenic and nonpathogenic serovars (^{Morey et al., 2006}). Furthermore, little correlation has been shown between serological classification and genotypic systems because a given serogroup can often be found in several species of Leptospira (^{Ahmed et al., 2012}).

In addition to DNA-DNA hybridization and the other molecular methods mentioned above, specific PCR amplification of the 16S rRNA gene has contributed to the molecular characterization of some species of Leptospira (^{Ahmed et al., 2012}). The advantage of this technique is that the use of a DNA template, particularly one designed based on the region that encodes the bacterially conserved 16S rRNA gene, can clearly reveal phylogenetic relationships among species (^{Morey et al., 2006}).

^{La Scola et al. (2006a)} have designed a universal primer pair for the identification of Leptospira species based on the gene encoding the β subunit of RNA polymerase (rpoB). These primers have been used to amplify and sequence the partial rpoB gene from 16 Leptospira species. According to the authors, analysis of the rpoB gene "may be useful as an initial screening test for the serovar identification of a new isolate of Leptospira and the detection or identification of Leptospira in clinical or environmental samples".

In previous studies, the utility of the rpoB gene for spirochete distinction among various bacterial species has been demonstrated (^{Renesto et al., 2000}; ^{Lee et al., 2000}; ^{Khamis et al., 2004}; ^{Balamurugan et al., 2013}). Thus, the aim of this study was to investigate whether the PCR-amplified fragment of rpoB in conjunction with RFLP would allow for the determination of Leptospira serovars.

Material and Methods

Results

In silico analysis of rpoB sequences deposited in GenBank indicated that a combination of four possible restriction enzymes was necessary to distinguish the Leptospira serovars as follows: TaqI, Tru1I, Sau3AI and MslI. Alone, each enzyme was able to identify only one or two different serovars.

Digestion with TaqI resulted in ten different patterns (A to J), which are schematically represented in Table 2 and had the following frequencies: A, 29.4% (20); B, 10.3% (7); C, 7.35% (5); D, 13.2% (9); E, 11.8% (8); F, 4.41% (3); G, 16.2% (11); H, 7.35% (3); I, 1.47% (1); and J, 1.47% (1). Thus, TaqI identified two serovars, Huallaga of L. noguchii (profile I) and Alice of L. santarosai (profile J), as shown in Figure 1. The G profile pattern was observed in almost all L. santarosai serovars, with the exception of the serovar Alexi (profile D), and it was only identified in the Muenchen serovar L. interrogans.

Figure 1 Polyacrylamide gel electrophoresis (6%) of the PCR products resulting from the digestion of the rpoB gene with the restriction endonuclease TaqI. was consistent with the 100-bp molecular weight ladder. 

The Tru1I enzyme also exhibited ten distinct restriction patterns (A to J) with the following frequencies: A, 2.94% (2); B, 23.5% (16); C, 16.2% (11); D, 11.8% (8); E, 22.1% (15); F, 14.7% (10); G, 2.94% (2); H, 2.94% (2); I, 1.47% (1); and J, 1.47% (1). These patterns are summarized in Table 2 and identified the serovars Huallaga of L. noguchii (profile I) and Trinidad of L. santarosai (profile J).

The combination of both enzymes, TaqI and Tru1I, generated 23 distinct patterns with some interesting results as follows: profile A of TaqI and Tru1I (profile AA) was species-specific and was only observed for L. biflexa. Profile AC was displayed by all serovars of L. kirschneri and by serovar Hualien of L. terpstrae; therefore, it is nearly species-specific. Finally, the profiles AG, FG, FE, CE, FD, BE and GE were unique to the serovars Mini, Kaup, Lanka, Szwajizak, Waskurin, Myocastoris and Maru, respectively.

Digestion with the Sau3AI enzyme generated nine distinct restriction patterns, which are summarized in Table 2 with the following frequencies: A, 2.94% (2); B, 22.1% (15); C, 30.9% (21); D, 4.41% (3); E, 7.35% (5); F, 16.2% (11); G, 8.82% (6); H, 5.88% (4); and I, 1.47% (1). Sau3AI digestion identified only the serovar Ranarum to have a serovar-specific profile. However, the combination of all three enzymes generated 30 distinct profiles, including EEE, DFD, ACC, HDD, GBB and AHH, which were specific for the serovars Whitcomb, Icterohaemorrhagiae, Hardjo, Ramisi, Semaranga, Vargonicas and Sarmin, respectively.

Finally, digestion with the enzyme MslI produced only five distinct restriction patterns, which are summarized in Table 2 with the following frequencies: A, 10.3% (7); B, 20.6% (14); C, 57.4% (39); D, 10.3% (7); and E, 1.47% (1). Only the serovar Saopaulo was identified by this enzyme to have a serovar-specific profile.

The combination of the four enzymes TaqI, Tru1I, Sau3AI and MslI generated 35 distinct profiles and identified the serovars Parameles (EFFD) and Celledoni (HDFA). In addition, this combination helped to distinguish the serovars Valbuzzi and Tropica, which had the profiles DFFC and GBCC, respectively (Table 3).

Out of sixty-eight serovars analyzed for RFLP polymorphisms in the region of the coding sequence containing the β-subunit gene of RNA polymerase, 22 serovars from nine species (32%) were identified by digestion with the enzymes TaqI, Tru1I, Sau3AI and MslI (Table 3), and the other 46 strains were clustered into 13 groups with two to seven serovars.

A dendrogram obtained from a matrix constructed with the results from the fragments generated by PCR-RFLP with the four restriction endonucleases (Figure 2) showed clustering of the sixty-eight reference serovars. Several of the tested strains appeared to be distant from others of the same species in relation to the current taxonomic classification. The serovar Kaup (L. inadai) was grouped with the serovar Waskurin (L. interrogans); the serovar Ranarum (L. meyeri) was similar to the nonpathogenic L. biflexa; the serovar Muenchen (L. interrogans) clustered with those of L. santarosai; the serovar Nona (L. borgpetersenii) was closer to the serovar Mangus (L. inadai); the serovar Hualien (L. terpstrae) grouped with the those of L. kirschneri; the Huallaga and Myocastoris serovars (L. noguchii) were located in different branches; the serovar Tonkini (L. borgpetersenii) grouped with the majority of those of L. interrogans; the serovar Ramisi (L. kirschneri) was closer to those of L. borgpetersenii; and the serovar Alexi (L. santarosai) was grouped with those of Djasiman, Pyrogenes and Sentot (L. interrogans), Malaya (L. inadai) and Sejroe (L. borgpetersenii).

Figure 2 Dendrogram constructed by joint analysis of the bands generated by the restriction endonucleases TaqI, Tru1I, Sau3AI, and MslI. 

Discussion

The correlation between the serological and genotypic classifications of leptospires is low, and identification is complicated because the same serovar can be distributed among different species (^{Ahmed et al., 2012}; ^{Balamurugan et al., 2013}). It is assumed that this lack of correlation between species and serovars is the result of horizontal transference between species of the genes that determine serotypes, but the basis of this transference, which is responsible for exchanging genetic determinants, is still unknown (^{Cerqueira and Picardeau, 2009}). A single base difference differentiated many strains of L. interrogans and L. kirschneri; therefore, phylogenetic representation may be less meaningful than sequence identities at variable positions (^{Cerqueira and Picardeau, 2009}).

The aim of this work was to identify Leptospira strains at the serovar level by performing PCR-RFLP to amplify a 600-bp fragment of the coding sequence of the β subunit of the RNA polymerase gene. The rpoB gene has been widely studied in other organisms and is considered by many researchers to be more useful than the 16S ribosomal RNA gene for the differentiation of bacterial species (^{La Scola et al., 2006a}; ^{Ahmed et al., 2006}; ^{Macheras et al., 2011}; ^{Ahmed et al., 2012}). In addition, twenty-five sequences of the rpoB gene of Leptospira are already available in databases, thereby facilitating access and minimizing project costs.

In a previous report, ^{La Scola et al. (2006a)} have compared similarities in the rrs and rpoB genes between different Leptospira serovars. Using the rpoB gene, they were able to effectively distinguish 11 of 19 serovars tested, differentiating them from other species and showing greater numbers of polymorphisms in both genes, leading to the conclusion that the rpoB gene could distinguish species with a higher number of differences between base pairs.

In this study, 68 Leptospira serovars were analyzed for polymorphisms in a specific region of the rpoB gene using the endonucleases TaqI, Tru1I, Sau3AI and MslI. These enzymes were selected after in silico restriction digestion of the rpoB sequences deposited in GenBank. We were able to identify 22 strains from nine species at the serovar level (32%). The rpoB gene has been widely used as an alternative tool in the phylogeny and identification of different species of bacteria, such as Coxiella burnetii (^{Mollet et al., 1998}), Afipia (^{Khamis et al., 2003}), Mycoplasma (^{Kim et al., 2003}), Corynebacterium (^{Khamis et al., 2004}), Acinetobacter (^{La Scola et al., 2006b}), Mycobacterium (^{Adekambi et al., 2006}; ^{Ben et al., 2008}), Halobacterium (^{Minegishi et al., 2010}) and Cyanobacteria (^{Gaget et al., 2011}).

In a recent study, the rpoB gene has been successfully used to identify or detect Leptospira species in animals and humans in India (^{Balamurugan et al., 2013}). Because each Leptospira serovar is associated with specific host symptoms, their identification is essential for the development of epidemiological studies (^{Cerqueira and Picardeau, 2009}, ^{Li et al., 2009}).

Clustering analysis of the results of this study correctly grouped 22 serovars by species. Considerable similarities in the analyzed genomic region were observed among all serovars. Analysis of dendrograms constructed from the results of each restriction enzyme and from the collective results for all of the enzymes showed the formation of clusters, for which serovars of various species had identical profiles. The groups formed by the rpoB gene profiles showed varying degrees of similarity and clade formation. Based on this, similar banding patterns were observed among the serovars Mangus, Nona, Alexi, Pyrogenes, Sentot, Malaya and Sejroe, despite the fact that they belonged to different species. These findings are in accordance with similar dendrogram analyses reported previously (^{Perolat et al., 1998}; ^{Morey et al., 2006}; ^{Cerqueira and Picardeau, 2009}; ^{Balamurugan et al., 2013}), showing similar cluster formations and variations in serovar-species grouping.

The addition of new enzymes for the production of additional profiles should clarify the positions of other serovars. Still, these results suggest that the use of this technique to assess gene sequences may reveal a precise sensu stricto classification of these serovars.

Molecular techniques have been used for the characterization of Leptospira isolates; however, most can only make identifications to the species level (^{Galloway and Levett, 2010}), such as 16S rRNA sequence analysis (^{Morey et al., 2006}), RFLP (^{Li et al., 2009}) and MLST (^{Boonsilp et al., 2013}). PFGE has demonstrated the reliable and reproducible identification of Leptospira at the serovar level (^{Galloway and Levett, 2010}). These approaches have greatly contributed to a revolution in both Leptospira detection and characterization (^{Ahmed et al., 2012}). On the other hand, the molecular tools described so far for the characterization of Leptospira suffer from significant drawbacks. For example, PFGE, RFLP, and REA require large quantities of purified DNA, have low levels of discrimination, produce data that is difficult to interpret, suffer from a lack of reproducibility and require specialized equipment (^{Ahmed et al., 2006}).

Notably, the 16S rRNA gene has been considered the gold standard in molecular surveys of bacterial and archaeal diversity, but it has several disadvantages as follows: it is often present in multiple copies, has little resolution below the species level and cannot be readily interpreted in an evolutionary framework (^{Vos et al., 2012}).

The main advantages of the use of the rpoB gene over the 16S rRNA gene are as follows: (i) it is universally present in all prokaryotes; (ii) it typically occurs in a single-copy, essential protein-encoding gene, and sequence errors can be readily identified and removed if they introduce disruptions in the reading frame; (iii) it possesses both slowly and quickly evolving regions, enabling the design of probes and primers of differing specificities; (iv) it has a housekeeping function, making it less susceptible to some forms of lateral gene transfer; and (v) it is large enough in size to contain phylogenetic information, even after the removal of regions that are difficult to align (^{Case et al., 2007}; ^{Vos et al., 2012}).

Our findings "in vitro" indicate that the PCR-RFLP technique is a powerful and reproducible test that may be used as a complement or alternative tool to assess the distribution of Leptospira strains within species. Additionally, we recommend the use of PCR-RFLP with in silico digestion of the polymorphic sequences of other conserved genes already deposited in GenBank as a promising technique for the genomic classification of the Leptospira genus.

Conclusion

This study demonstrated that PCR-RFLP is practical and efficient, enabling the differentiation of species and serovars with good discriminatory power, reproducibility and easily interpretable results. In addition, this method is cost-effective for most research laboratories. This technique has also been shown to be suitable for phylogenetic studies and the classifications of species, serovars and strains. The selected 600-bp polymorphic sequence of the rpoB gene produced restriction profiles that allowed for the accurate and timely identification of 32% of the 68 tested strains. We demonstrated that this approach achieves the stated purpose and that serological typing is unreliable for the classification of pathogenic Leptospira. However, additional studies should be undertaken to reclassify these serovars within the species with which they have greater genotypic affinities based on analysis of hypervariable regions of multiple housekeeping genes and especially to investigate whether the clinical leptospirosis symptoms induced by these serovars are presented according to the species with which they are most phylogenetically related.