The MLVA assay presented was easy to perform and analyse, as it consisted of six individual PCR reactions and agarose gel electrophoresis. Selected reference strain isolates were run in tandem during the initial evaluation by different individuals to assess reproducibility, each time they displayed identical fragment sizes as determined by sequencing and agarose gel electrophoresis (data not shown). Dilutions of the template DNA were used to evaluate whether the fragment sizes were template concentration dependant. Dilutions of 1:10 and 1:100 did not effect fragment size but did result in reduced PCR product yield (data not shown). The MLVA assay was proven to be reproducible under varying laboratory conditions. The two limitations of this assay are firstly the use of agarose gel electrophoresis to separate fragments for allelic sizing, due to inherent inaccuracies of this method to size bands of close molecular weights given that the resolution is dependant on agarose composition and concentration and secondly in rounding partial repeat copy numbers to the nearest whole number to make the data analysis easier, isolates that had partial repeats were treated as if they contained whole repeats. As illustrated in Additional file 1, by simplifying the repeat copy numbers we have artificially reduced the resolution of the method and its ability to distinguish between closely related strains that may only vary by a partial repeat at one or more of the loci.
The diversity index calculated for each locus suggests that the loci selected are of highly polymorphic nature and therefore have greater discriminatory power between similar strains than loci with a lower diversity indexes would have. Whilst it has been previously reported in organisms such as Francisella tularensis  and Yersinia pestis  that higher copy repeat numbers may confer higher allelic variability, it was not demonstrated with this study. This may be due to the loci having similar repeat copy sizes (36–47 bp) with the exception of locus V27 with a repeat size of 138 bp. The lack of amplification from loci V8, V27, V29, V36 and V50 from certain reference strains may be due to sequence diversity in the flanking regions up or downstream from the repeat regions or the lack of the VNTR loci all together. Further investigations using isolates of these serovars is needed to determine whether there is sufficient diversity in the remaining loci for it to be valid as a typing method. In the study by Majed et al. , they noted from the dendrogram that the isolates had clustered into distinct global geographical regions. Interestingly in this study, we found that the dendrogram showed no bias towards this global geographical clustering of reference strains.
The different genomospecies of Leptospira were not used in the selection and development of the VNTR loci for this typing scheme. The basis for this decision was that the other genomospecies are considered to be significantly different from L. interrogans based upon DNA-DNA Hybridisation [32, 33] and may not possess the same primer binding sites or indeed the same VNTR loci. In addition due to the use of a serologically based Leptospiral taxonomy system, several serovars belong to more than one genomospecies [32, 33], thus it would be doubtful that the MLVA typing scheme described in this article would be useful in determining the genomospecies of these related serovars due to the significant genetic differences
When the MLVA assay was applied to L. interrogans serovar Australis isolates collected from 1995 to 2004, the six selected loci appeared to show less diversity. This could be due to the fact that the isolates were taken from within a limited geographical area of Far North Queensland. Despite this apparent limited diversity, the phylogenetic analysis revealed several large albeit weakly linked clusters. All of these clusters contained a mixture of hosts, and it would be possible to speculate that the transmission of serovar Australis to humans in that area is via the native rodent population indigenous to that area. Another possible risk whilst not proven in Australia, is that transmission of the organism may occur via canines to humans . Indeed in this study the two strains isolated from canines are genetically similar to two strains isolated from humans. Also of interest, the reference culture for serovar Australis; strain Ballico which was isolated from a patient in North Queensland in 1934, shows homology with LT958 and QHR371A a human and Rattus sordidus isolate respectively, both from the Tully region. Further investigations using MLVA of isolates, could add further detail to the depth of knowledge into the population of serovar Australis and to determine other possible transmission sources to humans.
Further improvements to this method are possible to increase both practicality and discriminatory power for typing of L. interrogans isolates. These improvements could include the use of fluorescently labelled primers and fragment analysis using a DNA sequencer to accurately assign repeat sizes. Multiplexing of the six targets would rationalise the number of PCR reactions needed to complete the MLVA and would also decrease costs in terms of reagents and labour. Improvements that could be introduced to the analysis of the data may include using an allele designation system as described by Lindstedt et al . In addition to improving the method, comparisons between other molecular typing methods such as FAFLP or a sequence based typing scheme, would ultimately determine the validity of the MLVA assay as molecular epidemiology tool for L. interrogans.
The method has potential application in furthering the understanding of Leptospiral molecular epidemiology. As this method can be performed without specialised equipment, a broader range of laboratories including those in developing countries could potentially use this scheme as part of their isolate typing. This method was also easily standardised within our laboratory, with multiple users and different thermal cyclers employed to achieve the same results. This level of standisation at an inter laboratory level would allow the transfer of the method into another laboratory more effectively than that of a method that was operator or equipment specific. The simplification of MLVA data into a concise and portable numerical format as suggested in this article makes it easier to be comprehended by non-technical staff such as public health authorities. In addition, the format of allele data is similar to the allele string that is used for Multiple Locus Sequence Typing (MLST). As an alternative to using Bionumerics, software freely available on the internet such as Sequence Type Analysis and Recombinational Tests (START) http://pubmlst.org/software/analysis/ could be used for the phylogenetic analysis. Bionumerics was ultimately selected over START due to the advanced features of Bionumerics including evolutionary and population modelling.
Further assessment of L. interrogans isolates globally is required to confirm that the selected VNTR loci possess sufficient diversity to be used a typing scheme on an international level.