High Resolution Melting Analysis Targeting hsp70 as a Fast and Efficient Method for the Discrimination of Leishmania Species

Background Protozoan parasites of the genus Leishmania cause a large spectrum of clinical manifestations known as Leishmaniases. These diseases are increasingly important public health problems in many countries both within and outside endemic regions. Thus, an accurate differential diagnosis is extremely relevant for understanding epidemiological profiles and for the administration of the best therapeutic protocol. Methods/Principal Findings Exploring the High Resolution Melting (HRM) dissociation profiles of two amplicons using real time polymerase chain reaction (real-time PCR) targeting heat-shock protein 70 coding gene (hsp70) revealed differences that allowed the discrimination of genomic DNA samples of eight Leishmania species found in the Americas, including Leishmania (Leishmania) infantum chagasi, L. (L.) amazonensis, L. (L.) mexicana, L. (Viannia) lainsoni, L. (V.) braziliensis, L. (V.) guyanensis, L. (V.) naiffi and L. (V.) shawi, and three species found in Eurasia and Africa, including L. (L.) tropica, L. (L.) donovani and L. (L.) major. In addition, we tested DNA samples obtained from standard promastigote culture, naturally infected phlebotomines, experimentally infected mice and clinical human samples to validate the proposed protocol. Conclusions/Significance HRM analysis of hsp70 amplicons is a fast and robust strategy that allowed for the detection and discrimination of all Leishmania species responsible for the Leishmaniases in Brazil and Eurasia/Africa with high sensitivity and accuracy. This method could detect less than one parasite per reaction, even in the presence of host DNA.


Introduction
Leishmaniases are a major worldwide public health problem and manifest themselves as a spectrum of diseases that may be exacerbated by other infections, such as human immunodeficiency virus. According to the World Health Organization, these diseases are endemic in 98 countries on 5 continents, with more than 350 million people at risk [1,2]. Clinically, Leishmaniases can be broadly divided as either cutaneous or visceral, but neither form is exclusively linked to a particular species. Although cutaneous manifestations of the diseases are not life threatening, these manifestations can result in obstruction or destruction of the pharynx, larynx and nose in their final stages [2]. The visceral form is the most severe form, characterized by fever, loss of weight, splenomegaly, hepatomegaly, lymphadenopathies and anaemia, often with fatal outcomes if not timely treated [3].
The severity of the disease and its therapeutic responses are variable and depend on the patient's immune response, the Leishmania species and even the parasite strain [4]. In this scenario, the development of optimized protocols for discriminating between the different Leishmania species is extremely useful and important in clinical management and treatment. The ability to evaluate the most appropriate species-specific treatments also supports the elucidation of the mechanisms of action of new drugs and the establishment of new species-specific treatment protocols. Furthermore, the identification of these parasites allows the generation of important data for clinical, epidemiological and ecological studies.
There are very few publications addressing a Leishmaniasis diagnosis using a High Resolution Melting (HRM) analysis, a methodology that detects differences in the nucleotide composition of a specific real-time PCR product. The method is based on thermodynamic differences in the dissociation curve profiles of amplicons generated from real-time PCR. The generated curves are specific signatures that identify polymorphisms due to small differences in nucleotide composition. In spite of the paucity of papers on the HRM method, some workers have already used it to discriminate Leishmania using targets against 7SL RNA [5,6], haspb [7], the rRNA ITS sequence [8,9], the rRNA ITS sequence coupled to hsp70 [10,11] and a FRETbased assay using MPI and 6PGD [12].
In this work, we propose a more efficient protocol using HRM analyses targeting the hsp70 sequence for the discrimination of seven Brazilian Leishmania species, as well as three Eurasian and African species. This methodology was validated with DNA from reference strains, experimental infections in mice, human clinical samples and naturally infected phlebotomine sand flies. shawi (MCEB/BR/84/M8408) were grown at 25°C in M199 medium with 10% fetal bovine serum (Life Technologies, Carlsbad, CA, USA). Procyclic forms of Trypanosoma cruzi (Y strain) and T. brucei (427 strain) were grown at 28°C in liver-infusion-tryptose medium and SDM-79, respectively, with 10% fetal bovine serum (Life Technologies). Human DNA, FMUS-P-IOF-2016, obtained from USP Medical School, was used in specificity tests.

Trypanosomatids DNA
DNA samples from reference strains were purified by a salting-out procedure using an adaptation of the protocol described by Miller et al. 1988 [18]. Approximately 2.5 x 10 9 promastigotes in stationary growth culture were centrifuged at 3000 x g for 10 min at 25°C. The cells were resuspended in 6 mL of lysis buffer (10 mM Tris-HCl, pH 7.4; 400 mM NaCl; 2 mM EDTA) and lysed by the addition of 600 μL of 10% SDS. After overnight digestion with 1 mg of proteinase K at 37°C, 2 mL of saturated NaCl solution was added to lysate, and then, the lysate was vigorously mixed for 15 seconds and centrifuged for 15 minutes at 25°C for the removal of precipitated proteins. Two volumes of cold absolute ethanol were added to the supernatant, and the precipitated DNA was washed with 70% ethanol and resuspended in 1 mL of TE buffer ( Fresh experimentally infected BALB/c mice samples of L. (L.) amazonensis or L. (V.) braziliensis were obtained 6 weeks after infection when the animals were sacrificed and tissues were collected and DNA was obtained as described below; the procedures involving the use of BALB/c mice had the approval of the Ethical Committee for use of Animals of Biomedical Sciences Institute of University of São Paulo (CEUA-ICB-USP), under protocol #145 of October 20 th , 2011, according to Brazilian Federal Law 11.794 of October 8 th 2008. DNA from infected phlebotomines captured in nature were purified using the commercial DNeasy Tissue & Blood kit (QIAGEN, Hilden, Germany), according to the manufacturer´s manual. Paraffin-embedded samples were prepared according to de Lima et al. 2011 [19]. The DNA concentration was measured by spectrophotometry.

PCR assays
Initially, we amplified the hsp70 234 bp fragments for all species analyzed in this study using the primers described by Graça et al. [17]. The alignment of the nucleotide sequence of those fragments was used to design primers for HRM analysis. Oligonucleotides used in the PCR assays to amplify a 144 bp fragment of hsp70 (amplicon 1) were hsp70C reverse, previously described by Graça et al. 2012 [17], and a new forward oligonucleotide designed and named hsp70F2

Cloning and sequencing
The 234 bp hsp70 fragment produced by conventional PCR, as described by Graça et al. 2012 [17], from each Leishmania species used in this study was purified and cloned into a pGEM-T vector using the pGEM-T Easy Vector System (Promega, Madison, WI, USA) and E. coli SURE competent cells. The recombinant plasmids from at least three colonies were purified, and they were sequenced with T7 and SP6 primers and the BigDyeTerminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA). The sequencing was performed on an ABI 3130 XL Platform (Life Technologies).

Target quantification with standard curves
Recombinant plasmids containing the hsp70 target were linearized with ScaI. The plasmid copy number was calculated considering the molar mass concentration, and a serial dilution on a tenth proportion was used to produce standard curves for each quantification test. The quality parameters for the standard curves were obtained by PikoReal Software (Thermo Fischer Scientific) analysis, including the correlation coefficient, linear dynamic range and PCR efficiency. HRM assays were performed at the end of each real-time PCR. The amplicon dissociation analysis was performed by capturing fluorescence signals in 0.2°C intervals and holding for 10 seconds in each range of the melting curve (between 60°C to 95°C). The acquisition of fluorescence data and the construction of dissociation profiles were performed using PikoR-eal96 software. HRM software normalizes melting curves relatively to values from pre-and post-melting point assigned as 100% and 0%, respectively. Then the software determines the normalized difference that means the signal-to-noise ratio difference of each sample versus a user-defined sequence that can be any. The call efficiency is the benchmark measured in percentage of the similarity between two dissociation profiles using fluorescence and Tm values as parameters. The software performs a paired comparison between the profile of the sample of unknown identity and each standard and chooses the standard that has the closest value. The "call" identity refers to the designation allotted to the sample being identified based on that of the closest standard.
The graphs containing the means and standard deviations of the Tm values obtained by the HRM analyses were made in GraphPad PRISM v. 6.02 software. DNA from all Leishmania reference strains analyzed in this study was used as templates in conventional PCR, and the amplicons were cloned and sequenced to confirm the sequences to those deposited in GenBank. The obtained hsp70 amplicon sequences were then aligned, and we chose regions containing polymorphic sites to be used in HRM methodology (Fig 1).

Polymorphic sites on the hsp70 gene
The two pairs of oligonucleotides depicted in the alignment produced the two expected PCR fragments for all Leishmania reference strain DNA used as a template. The 144 bp amplicon 1 is the PCR product used in the amplification of all Leishmania species. The 104 bp amplicon 2 was produced by the oligonucleotide pair designed for species of the L. (Viannia) subgenus (Figs 1, S1 and S2).

Specificity and sensitivity of HRM assays
The average and standard deviation of the melting temperature (Tm) of each amplicon was determined in duplicate from three independent experiments using 50 ng of DNA as a template from each reference species. The melting profiles and obtained Tm values of hsp70 amplicon 1 for all species studied are presented in Figs 2 and 3 and Table 1. For a reliable discrimination, we calculated the dispersion of Tm values and only considered differences in Tm values exceeding 0.3°C (Fig 2).
The standard curves for the quantification assays using the cloned target showed good linear correlations (0.99 for all curves) and efficiencies varying from 92,37 to 97.23% for all tested species, in the range of 10 1 to 10 7 copies (S3 Fig). Moreover, to evaluate the specificity/sensitivity of hsp70 amplicon 1 as a target, HRM assays were performed using genomic DNA from the seven references species of Leishmania in proportions of 1:1 or 1:100 in relation to a human  reference DNA (FMUSP-IOF-2016), and the call identification agreed 100% with the reference samples, even in samples where the call efficiency was approximately 75% (Table 2).
To test if the initial amount of target DNA caused a variation in the Tm, serial dilutions containing 50 ng to 50 fg (DNA amount corresponding to 5.0 x 10 5 to 0.5 of parasite) of Leishmania DNA from reference strains were used as a template to produce both hsp70 amplicon 1 ( Fig 4A) and hsp70 amplicon 2 ( Fig 4B). The Tm variation obtained for both amplicons in each species showed that some species presented a fluctuation of Tm values that overlapped with other species.
In the case of overlapping Tm values for amplicon 1, a sequential discrimination can be performed by HRM analysis of amplicon 2. This amplicon is specific for the L. The Ct values obtained in the amplification curves of amplicon 2, using DNA of all Leishmania studied indicated that the reactions were at least 5 orders of magnitude more specific to Leishmania (Viannia) species than for the L. (Leishmania) species (Figs 5C and S2), confirming that amplicon 2 can be used to discriminate L. (Viannia) from the L. (Leishmania) species.
Moreover, using the information on the geographical origin of the samples associated with the HRM analysis of hsp70 amplicon 2 allowed for the discrimination between L. (L.) donovani Table 1. Tm values obtained in the HRM analysis targeting the hsp70 gene of different Leishmania species. Fifty ng of genomic DNA from each species was tested in duplicate in three independent experiments.     The aligned sequences were then examined for polymorphisms among species as well as among strains of the same species. We then calculated the percentage of similarity and estimated the theoretical Tm value of both amplicons (S1 Table). If we assume that the nucleotide differences that we detected are real polymorphisms and not sequencing errors then we can see from S1 Table that the differences in the theoretical Tm values of each species results in the same discriminatory pattern. Of the 186 strains analyzed, only two strains of L. infantum, MCAN/IR/96/LON-49 and LEM75/zymodeme1, presented a theoretical Tm value whose difference was higher than 0.3°C. We cannot rule out the possibilities that this difference is in fact a real one, due to sequencing errors or reflects different taxa.
In the absence of bona fide samples we also determined the theoretical Tm of amplicons 1 and 2 (S1 Table) Table). The enriettii complex members L. (L.) martiniquensis and L. siamensis presented identical theoretical Tm values.
To validate the HRM protocol for different types of sample preparations, sixteen DNA obtained from real biological samples, like fresh tissue from hamster inoculated with infected sample from human or dog cases; cell culture of the human isolated strain; human fresh tissue; human paraffin-embedded tissue; tissues from experimentally infected BALB/c mice and naturally infected phlebotomines, that had been previously tested in our laboratory by sequencing of SSU rDNA [20] or by discriminatory PCR targeting g6pd [21], were submitted to HRM  analysis. The results obtained presented a correlation with the results obtained with the other targets (Table 4).

Discussion
The establishment of optimized protocols for the detection and identification of the aetiological agents of Leishmaniases are extremely useful tools in a clinical context. Identifying the species can lead to species-specific treatment protocols to promote a better efficacy of treatment, assessing the need for patient follow up as well as the development and understanding of the mode of action of potential new drugs. Several methodologies targeting different genomic or mitochondrial DNA have been described in the past 20 years, and PCR is currently the preferred method in studies involving the detection and identification of Leishmania. These methodologies have been developed by designing primers that exploit species-specific sequence polymorphisms in different targets, such as kDNA [22], the SSU rDNA gene [20,23], the glucose-6-phosphate dehydrogenase gene (g6pd) [21,24], rDNA internal transcribed spacers (ITSs) [25], hsp70 [13][14][15][16][17] and cysteine proteinase B gene (cpb) [7,26]. However, none of these methods represents a gold standard because the targeted polymorphisms were unsuitable for simple and direct identification protocols. These PCR analyses involved the use of multiple targets requiring a combination of several primers creating the need of running more than one reaction to identify a single sample. The multiplex PCR that uses several pair of primers in one reaction and restriction fragment length polymorphism analysis (RFLP) of PCR products both need of a subsequent DNA fractionation by gel electrophoresis. These procedures require experienced operators to interpret the results, besides the risk of laboratory contamination with amplicons, due to the manipulation of PCR product. Table 4. Identification of Leishmania in clinical and experimentally infected and field samples by the HRM analysis targeting hsp70 gene. The hsp70 amplicons 1 or 2 of DNA from each sample were submitted to the HRM analysis. The results were compared with a previous identification performed by SSU rDNA sequencing [17] or g6pd PCR [18]. (a) fresh tissue from hamster inoculated with infected sample; (b) cell culture of the human isolated strain; (c) human fresh tissue; (d) human paraffin-embedded tissue; (e) experimentally infected BALB/c mice; (f) naturally infected Lutzomyia (Lutzomyia) longipalpis; and (g) naturally infected Lu. (Nyssomyia) whitmani; (N/A): not applicable.

Sample Source HRM Identification Previous Diagnosis Method
Another way to exploit DNA polymorphisms is the determination of the C+G composition of PCR products from conserved regions by calculating the Tm of the amplicon in a melting curve. HRM methodology has been successfully used for Leishmania identification using different targets, such as the 7SL RNA gene that discriminated L. tropica, L. major and species that cause visceral Leishmaniases in clinical samples [5,6]. Additionally, using the same target, researchers determined that rodent Ctenodactylus gundi is a potential host of L. tropica in Tunisia [5]. Polymorphisms on haspb (Hydrophilic Acylated Surface Protein B gene) analyzed by HRM allowed the differentiation of strains of L. (L.) donovani from distinct regions of East Africa [7]. In Southeastern Iran, the rRNA ITS sequence incriminated Phebotomus sergenti as a natural vector of L. (L.) tropica [10], or the discrimination between L. (L.) tropica and L. (L.) infantum in Turkey [9]. HRM analysis of the ITS-1 rRNA region discriminated L. (L.) major, L. (.L) tropica, L. (L.) aethiopica and L. (L.) infantum in samples from Middle East, Asia, Africa and Europe [8]. The combination of two targets, hsp70 and the rRNA ITS1 sequence, using the absolute HRM values allowed for the discrimination of six American Leishmania species [11] and MPI/6PGD-FRET PCR distinguished L. (V.) braziliensis from L. (V.) peruviana [12].
Here, we described an algorithm using HRM methodology for the rapid detection and discrimination of Leishmania species circulating in Brazil and Eurasia/Africa (Fig 6). We used the sequence coding for hsp70, but in order to obtain a discriminatory PCR product, we designed the primers to encompass a region that was no larger than 144 bp and that had relevant polymorphisms for HRM analysis, that is, shifts of AT base pairs to CG or vice-versa. Moreover, to be effective, the total amount of polymorphisms was taken into account, and compensatory changes were avoided. Using these criteria, we obtained two PCR products: amplicon 1 and amplicon 2. Using the algorithm described in Fig 6,   infantum, but these two species cannot be discriminated from each other (Fig 2).
The occurrence of an overlap in the Tm value for the Brazilian species L. (L.) amazonensis and L. (L.) lainsoni after a positive reaction of amplicon 1 can be solved by a positive reaction of amplicon 2. This amplicon sequence is specific for Leishmania (Viannia) species, so L. (L.) amazonensis will not be amplified and L. (V.) lainsoni will present the corresponding Tm value (Fig 5). The occurrence of an overlap in the Tm value for the American species L. (L.) amazonensis and L. (L.) mexicana can be solved by amplicon 1 sequencing because this amplicon is not identical, but there are two mismatches (position 82 A to G and position 100 G to T in L. (L.) amazonensis and L. (L.) mexicana, respectively (Fig 1), that are compensatory in the melting profile. It is interesting to note that these two species are very closely related. Uliana et al. [23] distinguished L. (L.) amazonensis from L. (L.) mexicana by SSU rDNA, but Castilho et al. [21] also failed to distinguish these species by g6pd because the region of the g6pd sequence that was used is identical in the two species. It is also interesting that Hernandez et al. [11], using a larger amplicon (337 bp) of hsp70, succeeded in differentiating L. (L.) mexicana from L. (L.) amazonensis; however, Fraga et al. [13] failed to distinguish these two species using RFLP in another region of hsp70. However, when the complete nucleotide sequence of the hsp70 PCR fragment of 1268 bp is used, the discrimination between the two species can be achieved [27]. These problems once again emphasize that one gene or a particular sequence of a gene is not reliable to define a species or plot its phylogeny. Recently, Real et al. [28] showed that L. (L.) mexicana and L. (L.) major had, respectively, 5 and 7 species-specific orthologous gene families, while L. (L.) amazonensis had 23 different gene families. Moreover, the geographical parameter can also be used; Uliana et al. used SSU rDNA polymorphism to show that these species present a characteristic distribution in Latin America that correlates to monoclonal antibody profiles [29].
The Tm overlap for Eurasian species occurred for L. (L.) donovani and L. (L.) infantum, which presented identical sequences for amplicon 1. Again, the geographical origin of the sample can be used because L. (L.) donovani is more frequently found in India and East Africa and presents anthroponotic behavior. L. (L.) infantum is found in Africa, China and the Mediterranean and shows zoonotic behavior [30]. However, the two species can be discriminated by multilocus enzyme electrophoresis (MLEE) or multilocus microsatellite typing (MLMT) [30]. Recently, the haspb coding region was initially used in a classical PCR coupled to RFLP [31], while the gene coding for cpb was used as a target in conventional PCR [7]. We propose to use the latter in case of doubt between the two species (Fig 6).
The in silico analysis of amplicon 1 and 2 from other Leishmania species from America or from Eurasia/Africa, also indicated the potentiality of the hsp70 HRM protocol to discriminate We also noticed that the initial amount of template DNA influenced the Tm determination (Fig 4). This Tm variation could be important in cases where the Tm values are in the same range and can lead to a misidentification if the reference sample is at a different concentration. This is the case for L. (L.) amazonensis and L. (L.) lainsoni. However, as has been previously explained, the use of hsp70 amplicon 2 allowed for the discrimination between these two species. The two other species that presented an overlapping Tm range depending on the initial amount of DNA were L. (V.) braziliensis and L. (V.) guyanensis, which could be discriminated by the use of an HRM analysis on the same amplicon 2.
In fact, when we applied the protocol described here to other Leishmania isolates, the obtained "call" (the identification of the problem sample in relation to the reference samples) presented a 100% correlation with the reference strains ( Table 3).
The test of sixteen samples consisting of fresh hamster tissue from animals injected with human or dog biopsy macerates, fresh or paraffin embedded human biopsies, tissues of experimentally infected BALB/c mice or even naturally infected phebotominae, produced identification "calls" comparable to the identification results using SSU rDNA sequencing or g6pd PCR (Table 4), showing that the source of the sample as well as its conservation do not interfere in the HRM protocol. Moreover, the use of HRM protocol is easier than the use of SSU rDNA and/or g6pd PCR, since those methods require either sequencing of the product or three or more distinct PCRs followed by gel electrophoresis analysis.
Overall, the hsp70 HRM protocol described herein accurately and sensitively identified Leishmania species that are important in the majority of cases of Leishmaniases in the Brazil and Eurasia. The test is simple and rapid, and its use in the clinic or in research samples has many advantages, such as a lower total cost for the identification of a sample and other characteristics that facilitate its application. There is no need for sequencing or gel fractionation to analyze the product, thus avoiding laboratory contamination with PCR products because these products are discarded without being manipulated. It also reduces the need for trained personnel to analyze the fractionation profile of an electrophoretic gel or sequencing data to provide a result. Also the HRM assay presents a possibility of quantifying parasites present in samples because it is a real-time PCR-based technique. Moreover, the whole process can be automated because the analyzer software will produce the "call" result by comparing the tested samples to the reference sample identities, which must always be included in the reactions.
In conclusion, the protocol described herein is a low cost, reliable, easy to apply, potentially automated procedure that is a good alternative for the detection, quantification and identification of Leishmania species in biological and clinical samples.  Table. Polymorphisms detection by in silico analysis of Leishmania hsp70 sequences. The hsp70 regions compassing amplicon 1 or amplicon 2 were retrieved from GenBank Database [32] using the sentence "heat shock protein 70 kDa" as descriptor words in "search" field. The obtained sequences were formatted as FASTA files and aligned on BioEdit Sequence Alignment Editor v.7.1.8 [33]. The identity indexes were obtained by pairwise alignments on BioEdit software. Only sequences encompassing the whole amplicon were analyzed. Theoretical melting temperatures of hypothetic amplicons were calculated using OligoCalc oligonucleotide properties on-line calculator [34]. (DOCX)