Background

Leishmaniasis is a spectral neglected disease caused by many Leishmania species, primarily transmitted by phlebotomine sand flies [1]. Leishmaniasis is a public health problem in 98 countries, in both rural and urban areas. About 12 million people are currently affected by the disease; in particular, 0.2–0.4 million cases per year of visceral leishmaniasis (VL) and 0.7–1.2 million cases per year of cutaneous leishmaniasis (CL) have been estimated [2]. Leishmania parasites are classified in three different subgenera: Leishmania (Leishmania), Leishmania (Viannia) and Leishmania (Sauroleishmania). Parasites in the subgenera Leishmania and Viannia infect mammals, whereas the Sauroleishmania infect reptiles [3]. Over 30 Leishmania species have been identified, including many that are non-pathogenic or of minor medical importance. The Leishmania donovani complex belongs to the subgenus Leishmania (Leishmania), the etiological agent of VL and CL, and includes L. (L.) infantum and L. (L.) donovani. The species belonging to the subgenus Leishmania (Viannia) are etiological agents of CL and mucocutaneous leishmaniasis.

An accurate diagnostic method that allows the distinction between Leishmania species with overlapping geographical distributions is needed [4]. Since isoenzymatic characterization is laborious and time-expensive, various molecular approaches have been developed based on detection of different Leishmania target sequences [5, 6]. In particular, the conserved region of Leishmania kinetoplast DNA (kDNA) minicircles has been used as a specific target for real-time PCR assays, able to distinguish Old World [7, 8] and New World species [9]. The kDNA is situated at the base of the flagellum and contains thousands of minicircles and dozens of maxicircles concatenated in a giant network [10]. The high number of minicircles (10,000–26,000 copies/parasite) [11,12,13] make them an attractive target for Leishmania detection with high sensitivity [14]. However, the minicircle network is composed of different minicircle subclasses conserved in different Leishmania species [15, 16]. Therefore, most of the available PCR or qPCR assays designed for a single species can potentially amplify more than one Leishmania species [13, 14, 17].

Previous works demonstrated the possibility to discriminate between L. (Leishmania) and L. (Viannia) by SYBR green-based qPCR assays designed on kDNA conserved sequences, followed by melting or high-resolution melt (HRM) analysis [13, 18]. These approaches did not efficiently distinguish L. (L.) infantum from L. (L.) amazonensis, which can coexist in the same geographic area. These species infect both humans and dogs [19], and their differentiation is critical for correct treatment. In fact, both species can present with cutaneous manifestations, but L. (L.) amazonensis can also cause diffuse disease that does not respond well to current treatments [5]. Moreover, the use of immunosuppressant may alter the clinical course of Leishmaniasis (e.g. viscerotropic strains can result in cutaneous lesions, and dermotropic strains can result in visceral disease). These issues are particularly important in South America, where both species are endemic, and deforestation and uncontrolled urbanisation are responsible for the growing incidence of leishmaniasis [20].

The aim of this work is to investigate the kDNA sequence variability among different Leishmania species and to exploit these differences to develop a qPCR-based analysis that differentiates L. (L.) infantum from L. (L.) amazonensis.

Methods

Leishmania strains, clinical samples and DNA extraction

The DNA from promastigotes of Leishmania spp. isolates listed in Table 1 was chelex-purified. Conjunctival swabs (CS) or buffy coat (BC) samples from 21 dogs diagnosed with leishmaniasis were provided by the veterinary clinic “Santa Teresa” (Fano, Italy). The DNA from CS and BC samples was extracted as previously described [21]. The L. amazonensis and L. (Viannia) spp. clinical isolates were provided by Instituto de Patología Experimental (IPE), Facultad de Ciencias de la Salud - Universidad Nacional de Salta- Argentina and typed at the species level at the Institute of Biomedicine and Molecular Immunology, CNR (Palermo, Italy) [13]. The L. infantum clinical isolates have been characterised by multilocus enzyme electrophoresis (MLEE) at Istituto Superiore di Sanità (Rome, Italy).

Table 1 Leishmania strains/isolates used in this study
Full size table

The DNA extracted from isolates and strains was quantified using a Qubit fluorometer (Life Technologies, Carlsbad, USA). Human and canine DNA from uninfected donors, purified using the DNeasy Blood & Tissue kit (Qiagen, Valencia, USA), as well as DNA from Trypanosoma cruzi were used as negative controls.

qPCR assays

An 111 bp KDNA region was amplified by a qPCR assay using primers MLF and MLR (referred as qPCR-ML from hereon) as previously described [13]. Briefly, PCR reactions were carried out in 25 μl volume with 1 μl template DNA (corresponding to 0.1–1 ng parasite DNA) and 24 μl SYBR green PCR master mix (Diatheva srl, Fano, Italy) containing 1 U Taq Polymerase and 200 nM of each primer in a Rotor-Gene 6000 instrument (Corbett Life Science, Mortlake, Australia). The amplification conditions were: 94 °C for 10 min, 40 cycles at 94 °C for 30 s, 60 °C for 20 s and 72 °C for 20 s. At the end of each run, a melting curve analysis was performed from 78 to 92 °C, with a slope of 1 °C/s, and 5 s at each temperature. The reactions were performed in duplicate or triplicate.

To amplify L. (L.) amazonensis DNA minicircles, a new qPCR assay (referred as qPCR-ama from hereon) was designed using the forward primer LMi-amaF (5′-AAA ATG AGT GCA GAA ACC C-3′) with the reverse primer MLR. The LMi-amaF primer was designed specifically for this work exploiting two mismatches (G/A and T/C) among L. (L.) infantum and L. (L.) amazonensis sequences (Fig. 1). The LMi-amaF sequence was tested with BLAST against the non-redundant nucleotide database (organism limited to Leishmania). The position of all primers in the qPCR-ML amplicons is shown in Fig. 1. The qPCR-ama was performed at the same conditions described above. A standard curve for qPCR-ama was established using L. (L.) amazonensis MHOM/BR/00/LTB0016 DNA serial dilutions, ranging from 1.0 to 1 × 10-4 ng. The standard curve concentration was expressed as ng/μl. To evaluate the potential interference of host DNA as background in the qPCR analysis, 30 ng of human DNA were spiked in the reaction tubes. The cycle threshold (Cq) values were determined using the quantitation analysis of the Rotor-Gene 6000 software, setting a threshold to 0.15.

Fig. 1
figure 1

CLUSTAL multiple alignments of qPCR-ML amplicon sequences performed by MUSCLE. The MLF and MLR primer sequences are underlined; the sequence of LMi-amaF primer is in bold. The EcoRI restriction site in L. (L.) amazonensis isolate is boxed. *Sequence of non-digested products after EcoRI restriction

Full size image

High-resolution melt (HRM) analysis

The qPCR-ML amplicons, obtained by all stains/isolates listed in Table 1 and by 62 samples from 21 dogs affected by leishmaniasis, were analysed in duplicate by HRM protocol on a Rotor-Gene 6000 instrument as described previously [13] with slight modifications. Briefly, HRM was carried out over the range from 78 °C to 92 °C, rising at 0.1 °C/s and waiting for 2 s at each temperature. Each sample was run in duplicate or triplicate, and the gain was optimised before melting on all tubes. HRM curve analysis was performed with the derivative of the raw data, after smoothing, with the Rotor-Gene 6000 software. Only samples with Cq values ˂ 30 were considered for subsequent analysis [13, 22].

PCR product sequencing

The qPCR-ML products obtained from the canine clinical samples and the strains/isolates (Table 1, except for IFLA/BR/67/PH8 and MHOM/BR/75/M2904 strains), were purified using the MinElute PCR purification kit (Qiagen) and directly sequenced using both MLF and MLR primers. DNA sequencing was performed using the BigDye Terminator v. 1.1 Cycle Sequencing Kit on ABI PRISM 310 Genetic Analyzer (Applied Biosystems, Foster City, USA). Sequences were manually edited and aligned using default options in MUSCLE [23]. Phylogenetic and nucleotide composition analyses were conducted using MEGA version 6 [24].

RFLP analysis

The RFLP analysis was performed on qPCR-ML products with EcoRI restriction enzyme (Roche Life Sciences, Indianapolis, USA), following the manufacturer’s protocol. The EcoRI digestion was performed at 37 °C for 2 h in a 20 μl mixture containing 7 U of enzyme and 50–100 ng of PCR products. The digestion mixtures were analysed on a 3% high-resolution MetaPhor (Cambrex, East Rutherford, USA) agarose gel, and the selected products were excised, purified with MinElute PCR purification kit (Qiagen) and sequenced as described above.

Statistical analysis

Statistical analysis was performed with GraphPad InStat version 3.06 (GraphPad Software, San Diego, CA). Differences among Tm values were evaluated using an Unpaired t-test with Welch’s correction. Normality distribution was assessed by D’Agostino & Pearson omnibus normality test (alpha = 0.05).

Results

Inter- and intra-specific genetic variability of kDNA minicircles

To investigate the genetic variability in the kDNA minicircle sequences amplified by qPCR-ML assay, the PCR products from all strains/isolates reported in Table 1 (except L. amazonensis IFLA/BR/67/PH8 and L. braziliensis MHOM/BR/75/M2904) and 6 CS samples from 3 dogs with leishmaniasis were directly and bidirectionally sequenced. The electropherograms showed superimposed peaks in some region, suggesting a further kDNA sequence heterogeneity. Therefore, sequences were manually edited before alignment. Numerous polymorphic loci were found (Fig. 1). The Leishmania (Viannia) subgenus isolates had significantly lower GC content respect to Leishmania (Leishmania) isolates (47.4% ± 0.2% and 51.3% ± 0.5%, respectively) (Mann–Whitney test, U = 3.00, P = 0.015), explaining melting differences found previously [13]. The sequences of Leishmania (Viannia), L. (L.) amazonensis isolates/strains and L. (L.) infantum strains/clinical samples, clustered separately (Fig. 2).

Fig. 2
figure 2

Maximum likelihood phylogenetic tree of qPCR-ML amplicons. The tree was constructed with MEGA 6 software using Tamura-Nei model. *Sequence of non-digested product after EcoRI restriction 

Full size image

A restriction site for EcoRI was evidenced in the L. (L.) amazonensis isolate, but it was not confirmed in L. (L.) amazonensis MHOM/BR/00/LTB0016 strain (Fig. 1), making it not feasible to develop a PCR-RFLP assay to distinguish L. (L.) amazonensis from L. (L.) infantum. Moreover, the RFLP analysis of the L. (L.) amazonensis isolate revealed a partial digestion of the PCR product (Additional file 1: Figure S1). The non-cleaved band was extracted from the gel and directly sequenced. The lack of a EcoRI restriction site, as well as many similarities with the L. (L.) infantum sequences (Figs. 1 and 2), showed the presence of different minicircle classes amplified by the qPCR-ML assay.

HRM analysis alone does not fully discriminate between L. (L.) infantum and L. (L.) amazonensis

Previously, we demonstrated that HRM analysis of qPCR-ML products can discriminate between the Leishmania subgenera Leishmania and Viannia, and can differentiate the L. (L.) infantum MHOM/TN/80/IPT1 reference strain (two melting peaks) from L. (L.) amazonensis (one melting peak) [13]. However, HRM analysis performed on several clinical specimens from dogs infected by L. (L.) infantum, showed two peaks comparable to L. (L.) infantum MHOM/TN/80/IPT1 or a profile with a single melting peak, suggesting a kDNA minicircle parasite variability [13]. To confirm and further extend these data, we performed HRM analysis on qPCR-ML products obtained by amplification of L. (L.) infantum strains/isolates listed in Table 1 and 62 canine clinical samples from 21 dogs infected by L. (L.) infantum. The results showed that 22% and 56% of the L. (L.) infantum amplicons presented a double melting peak and a single melting peak, respectively. The remaining 22% of samples was undetermined due to lack of reproducibility between replicates. L. (L.) amazonensis amplicons always had a single melting peak. Tm values between L. (L.) amazonensis isolate, and MHOM/BR/00/LTB0016 strain was not significantly different (unpaired t-test with Welch’s correction, t = 1.23, P = 0.246). Tm values of strain IFLA/BR/67/PH8 were not considered due to late amplification (Cq ˃ 30) and consequent lack of reproducible Tm profiles [22]. The mean Tm of L. (L.) amazonensis amplicons was significantly different from the mean Tm of L. (L.) infantum amplicons showing a single peak (unpaired t-test with Welch’s correction, t = 8.77, P ˂ 0.001) (Table 2). However, the Tm value distribution overlapped (Fig. 3), therefore samples having Tm values ˂ 84.4 °C appeared in a range of uncertainty, as well as samples undetermined due to lack of concordance between replicates or due to late amplification (i.e. Cq ˃ 30). Taken together, HRM analysis did not allow discrimination between L. (L.) amazonensis and L. (L.) infantum in 53.4% of cases. Therefore, an approach based on a second qPCR was needed.

Table 2 Tm comparison of L. (L.) amazonensis and L. (L.) infantum amplicons showing a single HRM peak
Full size table
Fig. 3
figure 3

Tm values distribution of qPCR-ML amplicons of L. (L.) amazonensis and L. (L.) infantum with single melting peak. a Box-and-whisker plot is representing qPCR-ML amplicons Tm distribution. The chart shows the distribution of data into quartiles, the smallest and the largest observation, and the mean (x). No outliers were identified. b Examples of overlapping melting curves obtained from L. (L.) amazonensis MHOM/BR/00/LTB0016 (Tm 84.10–84.15 °C) (upper panel) and L. (L.) infantum MHOM/IT/86/ISS218 (Tm 84.15–84.20 °C), clinical isolate 791 (Tm 84.10–84.15 °C), clinical isolate 10816 (Tm 83.93 °C) (lower panel). These melting curves, showed in duplicates, represent an example of lack of discrimination between L. (L.) infantum and L. (L.) amazonensis

Full size image

qPCR assay for amplification of L. (L.) amazonensis

A qPCR assay which amplified an 88 bp sequence from L. (L.) amazonensis (qPCR-ama) was designed using a new forward primer (LMi-amaF) and the reverse primer MLR. The sequence of the LMi-amaF primer was first evaluated in silico using BLAST. We found 100% identity in 10 out of 16 (62.5%) sequences of L. (L.) amazonensis kDNA minicircle present in GenBank (Additional file 2: Figure S2). Notably, no sequence of L. (L.) infantum matched with the LMi-amaF primer. Moreover, only one sequence out of 101 (1%) of L. (L.) donovani, 3 out of 11 (27%) of L. (L.) Mexicana, and 7 out of 24 (29%) of L. (V.) braziliensis matched with LMi-amaF sequence, accounting for the largest presence of the minicircle subclass matching LMi-amaF primer in L. (L.) amazonensis.

The efficiency and detection limit of the qPCR-ama assay were evaluated using 10-fold serial dilutions of L. (L.) amazonensis MHOM/BR/00/LTB0016 DNA (from 1.0 to 1 × 10-4 ng) in 2 independent experiments in duplicate or triplicate. There was a linear correlation between the log of DNA concentration and Cq value (slope =-3.39, R 2 = 0.98) with a reaction efficiency of 97% (Fig. 4). To verify the interference of host DNA, the standard curve was also spiked with 30 ng of purified human DNA. The presence of host DNA delayed the Cq values in the standard curve, while the assay efficiency remained similar (Fig. 4). The detection limit was 0.1 pg of parasite DNA in the spiked samples, corresponding roughly to 1.3 parasites/PCR tube. This value was determined by the genome size of L. (L.) amazonensis (29.6 Mb, 64.8 fg for the diploid genome) [25], adding 9.7 fg (15%) kDNA [26]. The calculated total DNA for a single parasite was 74.5 fg.

Fig. 4
figure 4

qPCR-ama standard curves constructed with serial dilutions of L. (L.) amazonensis MHOM/BR/00/LTB0016 DNA. The standard curves were obtained with serial dilutions ranging from 1.0 to 0.0001 ng/tube, spiked (upper curve, y = -3.3720x + 24.4850; R 2 = 0.9915) or non-spiked (lower curve, y =-3.3933x + 17.9160; R 2 = 0.9803) with 30 ng human DNA. Each point derived from a duplicate of two independent experiments

Full size image

The specificity of the primers was first tested using two strains and one isolate of L. (L.) amazonensis, Trypanosoma cruzi and host DNA (human or canine). No amplification was observed when using T. cruzi, human or canine DNA, while all L. (L.) amazonensis strains were positive on qPCR (Fig. 5). Cq values were much lower compared to Cq obtained with qPCR-ML using the same template amount (Table 3). The qPCR-ama was then tested with strains of L. (V.) guyaniensis, L. (V.) panamensis, L. (V.) braziliensis and L. (L.) infantum. The results showed that L. (V.) guyaniensis, L. (V.) panamensis, L. (V.) braziliensis also amplified. However, these species could be distinguished from Leishmania (L.) species by qPCR-ML and HRM analysis. DNA from L. (L.) infantum isolates/strains did amplify, but Cq values were dramatically increased compared to qPCR-ML. If template DNA was diluted to have Cq ˃ 25 in qPCR-ML, there was no detectable amplification from qPCR-ama (Table 3). As evidence of this, the canine clinical samples tested with qPCR-ML (with Cq ˃ 25) had negative results with qPCR-ama, confirming infection by L. (L.) infantum (Table 4). Combined evaluation of qPCR-ML and qPCR-ama Cq values provided discrimination between L. (L.) infantum and L. (L.) amazonensis in all tested samples.

Fig. 5
figure 5

Electrophoretic analysis of qPCR-ama products. Lanes 1–5: L. (L.) amazonensis MHOM/BR/00/LTB0016 serial dilution ranging from 1.0 to 0.0001 ng DNA/reaction tube; Lane 6: L. (L.) amazonensis isolate; Lane 7: L. (L.) amazonensis IFLA/BR/67/PH8; Lane 8: Trypanosoma cruzi; Lane 9: human DNA; Lane 10 canine DNA. Abbreviations: m, marker 9; NTC, no template control

Full size image
Table 3 qPCR-ML and qPCR-ama results
Full size table
Table 4 qPCR-ML and qPCR-ama results in canine clinical samples
Full size table

Discussion

Previous work has shown that the Leishmania (Leishmania) and Leishmania (Viannia) subgenera could be distinguished by exploiting kDNA variability through HRM analysis of qPCR-ML amplicons [13]. Here, several qPCR-ML amplicons from different Leishmania species were sequenced, indicating the presence of different subclasses of minicircles amplified from the same species. Notably, the presence of over 100 minicircles classes encoding different gRNAs in a strain of L. tarentolae has been reported and quantified using NGS technology, indicating the heterogeneity of the mitochondrial genomes in this parasite [27]. The EcoRI partial digestion of the amplicon obtained from L. (L.) amazonensis isolate confirmed the presence of different sequences in the same PCR product. In fact, the sequence of the non-digested product evidenced a subclass of molecules without the EcoRI site, clustering with L. (L.) infantum sequences. The fact that the sequence of this product was not visible in the first sequence analysis can be explained by a relatively low presence of this minicircle subclass.

Leishmania (L.) infantum and L. (L.) amazonensis infect both humans and dogs, and they can coexist in the same geographical area; therefore, their differentiation is critical for correct treatment. Notably, Sanches et al. [28] showed a high percentage of L. (L.) amazonesis infection in naturally infected dogs in an endemic area, underlining the necessity to discriminate these two species not only in human but also in veterinary medicine. Since HRM analysis resulted inconclusive for discrimination of L. (L.) infantum and L. (L.) amazonensis in about 53% of samples, a new SYBR-green qPCR assay (qPCR-ama) was designed to amplify a minicircle subclass preponderant in L. (L.) amazonensis, rather than targeting a hypothetical species-specific sequence. In fact, several PCR assays designed on minicircles also reported amplification of non-intended species [17, 29]. Moreover, Gomes et al. [30] showed that non-target organisms such as T. cruzi could be amplified, even if Cq values are ˃ 30. Therefore, the performance of different qPCR assays in series or the standardisation of multiplex qPCR appeared to be the only way of properly identifying different Leishmania species [31]. Due to the difficulties in designing species-specific primers, we based our qPCR assay on the evaluation of relative abundance of minicircle subclasses. We observed that qPCR-ML amplified several Leishmania species including L. (L.) amazonensis but to a much less sensitivity (higher Cq). The opposite was found for qPCR-ama: in this case, L. (L.) amazonensis was amplified much more efficiently (lower Cq) compared to the other tested species, because primers were designed on a minicircle sequence that was more abundant in this species. Therefore, we have been focusing on Cq rather than presence or absence of an amplification curve. Hence, a minicircle subclass most represented in L. (L.) amazonensis was used as a target to differentiate L. (L.) amazonensis and L. (L.) infantum. The template DNA can be diluted to make undetectable the low-represented minicircle subclasses. The data in Table 3 show clearly this point: if DNA template from L. (L.) infantum was diluted to have Cq ˃ 25 in qPCR-ML (considering 1:10 dilution equivalent to ~ 3.3 cycles increase), the qPCR-ama turned negative. On the contrary, under appropriate dilution conditions, the qPCR-ama will give positive amplification only for L. (L.) amazonensis species (Table 3). These conditions could also be met by clinical samples, in which pathogen DNA is low represented respect to host DNA and the Cq can be delayed (as evidenced in the qPCR-ama standard curves).

The results of in silico analysis accounted for the possibility to amplify more L. (L.) amazonensis strains other those tested in this work, despite the database does not give information about the relative abundance of the kDNA minicircle sequences into the different strains. For example, the kDNA minicircle sequence of strain IFLA/BR/67/PH8 (GenBank M21325) shows a mismatch in the LMi-amaF sequence (Additional file 2: Figure S2) but it was efficiently amplified in our assay (Table 3), probably due to the presence of another minicircle subclass matching the primer sequence.

Taken together, a new diagnostic approach for Leishmania species discrimination from clinical samples or isolates has been developed: first, a qPCR-ML followed by HRM analysis is performed. In the case of lack of amplification, or if HRM results are inconclusive, the qPCR-ama assay is performed, and the evaluation of Cq values for both assays will allow the discrimination of the two species (Fig. 6). This approach is affordable for any molecular diagnostic laboratory equipped with an HRM qPCR instrument since it is based on an SYBR green chemistry and requires only three primers. Moreover, since the cost per reaction is relatively low (less than 1 €) and the PCR conditions are the same, the two assays could be run simultaneously to speed up the diagnostic process.

Fig. 6
figure 6

A possible diagnostic approach based on qPCR and HRM analysis. The qPCR-ML followed by HRM analysis is performed. In the case of lack of amplification, or if HRM results are inconclusive, the qPCR-ama can also be performed. The evaluation of Cq values for both assays will allow discriminating between L. (L.) infantum and L. (L.) amazonensis

Full size image

Conclusions

The new approach to differentiate L. (L.) amazonensis and L. (L.) infantum infections in geographical regions in which these species co-exist could be a useful diagnosis tool. Due to qPCR sensitivity, this approach will allow species differentiation from isolated parasites and from DNA extracted from clinical samples. This study represents a proof of concept: more samples should be analysed, and accuracy studies should be performed in clinical environments to validate the real utility of this method. Moreover, more species could be tested to extend the feasibility of this approach to other species or complex.