Completion of the L. major genome has allowed the generation of arrays containing 60-mer oligonucleotide probes designed by NimbleGen Systems 5556 and in this work, we present the generation of a full genome DNA microarray composed of 70-mer oligonucleotide probes suitable for both L. major and L. infantum analysis (see Materials and methods for a full description of the arrays). These full genome arrays were used for deciphering how Leishmania resists the antifolate model drug MTX. Two MTX resistant mutants, L. major MTX60.4, which has previously been studied with small targeted arrays 29, and L. infantum MTX20.5, were studied using the full-genome microarrays. Mutants of both species are highly resistant to MTX (Figure 1a), and since they were selected in a stepwise fashion, it is likely that multiple resistance mechanisms may exist in these mutants and could thus be uncovered by these arrays. The resistant cells had a similar generation time as the wild-type parent cells.

The DNA microarrays were first validated by hybridizing fluorescently labeled digested DNA of wild-type L. major and L. infantum cells. The arrays were found to yield uniform and reproducible results (not shown) and were deemed appropriate for RNA expression profiling experiments. Total RNAs were thus purified for both wild-type and mutant strains, used to synthesize fluorescent probes, and hybridized to the microarrays as described in Materials and methods. Scanning and normalization led to expression data that were first represented as scatter plots. As evident from these plots (inserts in Figure 2a,b), most genes in both species are equally expressed between the sensitive and resistant strains. Indeed, the bulk of expression (RNA level) ratios between sensitive and resistant strains were close to 1. Nonetheless, there were notable differences. First, the RNA levels of a total of 61 genes were found to be modulated (cut-off of 2, p < 0.05) in the L. infantum MTX20.5 mutant compared to the wild-type strain (Figure 2a; Table S1 in Additional data file 1) and the expression levels of 75 genes were changed significantly (cut-off of 2, p < 0.05) in the L. major MTX60.4 mutant compared to the wild-type strain (Figure 2b; Table S1 in Additional data file 1). Secondly, a majority of genes whose expression was modulated by more than two-fold had increased expression levels in L. infantum MTX20.5 but the majority of another set of genes had decreased expression levels in L. major MTX60.4 (inserts of Figure 2; Table S1 in Additional data file 1). If the expression modulation cut-off was changed from 2 to 1.5 (p < 0.05), we found 251 and 372 genes that were differentially expressed in L. infantum MTX20.5 and L. major MTX60.4, respectively (Figure 2). Surprisingly, few differentially expressed genes were found to be modulated similarly in both mutants (Figure 3; Table S1 in Additional data file 1). One notable exception is a region of chromosome 6 that corresponds to a six gene locus including the DHFR-TS gene. DHFR-TS is the main target for MTX and its gene was frequently found amplified in L. major MTX resistant mutants as part of extrachromosomal circles (reviewed in 34).

The DNA microarray data were supported by selected quantitative real-time reverse transcription PCR (qRT-PCR) assays in both the L. major and L. infantum mutants (Figure 3). In only two cases we found a discrepancy between the two techniques. LmjF04.0160 and its orthologue LinJ04_V3.0160 were found down-regulated in both mutants using DNA microarrays, but this was confirmed only in the L. major mutant by qRT-PCR (Figure 3). The other discrepancy between microarray and qRT-PCR data was for FT1, but this is explained by a gene deletion event (see below). The only other gene that was modulated similarly in the two mutants was the ABC protein gene ABCA2 and this was confirmed by qRT-PCR (Figure 3). Other genes were modulated in both mutants but in different ways. While LmjF31.0720 was down-regulated in L. major MTX60.4, its orthologue LinJ31_V3.0750 in L. infantum MTX20.5 was overexpressed (Figure 3). Otherwise, genes differentially expressed were specific to individual mutants.

The differential gene expression of the MTX resistant mutants was also represented in a chromosome by chromosome fashion (Figure 2). This has permitted us to visualize regions that are differently expressed (red/orange, corresponding to overexpressed genes in the mutants). Two regions were clearly overexpressed in the L. infantum MTX20.5 mutant. One region was on chromosome 6 (DHFR-TS loci) and the second was in the left portion of chromosome 23 (Figure 2a). For the L. major MTX60.4 mutant, we also saw an increase in expression of selected genes present on chromosome 6 (DHFR-TS loci), but we also observed a number of whole chromosomes (for example, chromosome 22; colored predominantly red in Figure 2b).

DHFR-TS is present on chromosome 6 and by close examination of the expression data derived from the arrays we were able to precisely define the genes with increased expression in both the L. major and L. infantum mutants. In L. infantum, the genomic region overexpressed is delimited by genes LinJ06_V3.0860 and LinJ06_V3.0910 (Figure 4a). Most interestingly, the same region is overexpressed in L. major MTX60.4 (Figure 4a). As Leishmania is devoid of control for the initiation of transcription (no pol II promoter has yet been isolated in this parasite 58), it is possible that the amplification of a small genomic region containing the DHFR-TS gene is responsible for the increased gene expression as determined by DNA microarrays. This was tested by hybridization of a blotted pulsed-field gel electrophoresis (PFGE) gel with a DHFR probe. Wild-type cells gave rise to two hybridizing bands, suggesting that the two homologous chromosomes 6 have different sizes (Figure 4b, lanes 1 and 3), a well established phenomenon in Leishmania 59. The two mutants had an extra band hybridizing to the DHFR probe, which with its hybridizing smear is characteristic of extrachromosomal circles (Figure 4b, lanes 2 and 4). The genesis of circular DNA in Leishmania has been studied and is often due to homologous recombination between direct repeats bordering the regions amplified 222324. Close examination of the sequences flanking the regions amplified indeed pointed to the presence of repeated sequences (Figure 4a). The repeated sequences were highly similar between L. major (575 bp) and L. infantum (837 bp) (Figure S1 in Additional data file 2). To provide evidence that the DHFR-TS containing circles were generated through homologous recombination between these direct repeated sequences, we used two primers (6a and 6b in Figure 4a,c) that should give rise to a PCR amplification product only when an extrachromosomal circle is formed (Figure 4c). Indeed, when using this primer pair, PCR fragments of the expected size were observed in L. infantum MTX20.5 and L major MTX60.4 (Figure 4d, lanes 2 and 4) while no amplification was observed in the wild-type cells (Figure 4d, lanes 1 and 3). The difference in size of the PCR fragments between L. major and L. infantum is due to the difference in size of the repeats in the two species (Figure S1 in Additional data file 2). Sequencing of the PCR generated amplicon derived from L. major MTX60.4 [GenBank:EU346088] confirmed the scenario of homologous recombination between the repeated sequences (Figure S1d in Additional data file 2).

In mutant L. infantum MTX20.5 we observed a region of chromosome 23 that was overexpressed (increased RNA levels; Figure 2a). This region contains the gene for pteridine reductase 1 (PTR1), a well established MTX resistance gene whose product can reduce folic acid, hence by-passing the need for DHFR-TS 3031. Similarly to the DHFR-TS loci, the microarray expression data have allowed the precise determination of the region that was overexpressed, which started at the telomeric end and extended 120 kb up to gene LinJ23_V3.0380 (Figure 5a). The putative presence of telomeric sequences would suggest a linear amplification instead of a circular amplification. Hybridization of a chromosome PFGE blot has shown that PTR1 hybridized to the approximately 800 kb chromosome in both wild-type and resistant cells but also to a smaller linear amplicon of approximately 230 kb in L. infantum MTX20.5 (Figure 5b). This amplicon also hybridized to a telomere probe (Figure 5b). The size of the amplicon suggests that the amplified region was duplicated. The LinJ23_V3.0390 gene is clearly not overexpressed and thus not part of the amplicon (Figure 5a). Three genes, LinJ23_V3.0360, LinJ23_V3-0370 and Lin23_V3.0380, were less overexpressed than the other genes that are part of the amplicon (Figure 5a). Examination of the sequences where expression changed enabled the detection of inverted homologous repeats of 578 bp (Figure S2 in Additional data file 2) between LinJ23_V3.0350 and Lin23_V3.0360, and between LinJ23_V3.0380 and Lin23_V3.0390 (Figure 5a). Interestingly, similar repeats of 574 bp with 91% identity were found at the same position in the L. major genome 60. The presence of these inverted repeats and the microarray expression data would suggest the formation of a linear amplicon with large inverted duplications that was formed by annealing of the identical 578 bp inverted repeats (Figure 5c). To obtain support for this scenario, we used PCR primer pairs (23a and 23b, or 23c and 23d) that would lead to a PCR product only if the rearrangement had occurred at the level of the inverted repeats (as, for example, during a block in DNA replication). Indeed, we obtained a product of the expected size with these pairs of primers in L. infantum MTX20.5 but no product was obtained from DNA derived from wild-type cells (Figure 5d). The nucleotide sequence of the PCR amplicon obtained with primer pair 23a/23b [GenBank:EU346089] is entirely consistent with the model shown in Figure 5c (Figure S2 in Additional data file 2).

Leishmania spp. have a large gene family of conserved folate transporters with 14 FBT members (AA Ouameur et al., unpublished data). Part of this family located on chromosome 10 is shown in Figure 6a. Microarray expression data indicated that FT1, coding for the main Leishmania folate transporter 1819, is down-regulated in L. major MTX60.4 but not in L. infantum MTX20.5 (Figure 3). The level of conservation of the various FBTs precluded that the 70-mer oligonucleotides spotted on the arrays would discriminate several of these closely related genes. The use of qRT-PCR to confirm the microarray data indicated that FT1 may be absent (Figure 3). This was suggestive of a gene deletion event and indeed a Southern blot of L. major MTX60.4 DNA hybridized with a probe recognizing the majority of FBT genes confirmed this extensive gene rearrangement (Figure 6b) and bands corresponding to LmjF10.0380, LmjF10.0385 (FT1) and LmjF10.0390 were either lacking or rearranged. Using PCR primers (labeled F and R in Figure 6a,c), we were able to demonstrate that FT1 (LmjF10.0385) was deleted following an event of homologous recombination between conserved sequences between LmjF10.0380 and LmjF10.0390 (Figure 6c). Indeed, primers F and R gave rise to a PCR fragment of 2.2 kb in L. major MTX60.4 (Figure 6d, lane 2) while under the conditions tested no fragments were found with L. major wild-type cells. Sequencing of the amplicon [GenBank:EU346090] validated the scenario of homologous recombination between two FBT genes leading to the diploid deletion of FT1 (Figure 6c; Figure S3 in Additional data file 2).

Analysis of gene expression on a chromosome by chromosome basis (Figure 2) suggested that the expression of whole chromosomes is modulated in L. major MTX60.4. Indeed, the majority of genes present on chromosomes 11 and 12 appeared down-regulated while the expression of genes located on chromosomes 7, 22, 28 and 32 seemed up-regulated (Figure 2). Chromosome 6 of L. infantum MTX20.5 also appears to be in more than two copies. This chromosome-wide uniform modulation of expression was represented more thoroughly for selected chromosomes by plotting the fold modulation in gene expression along the chromosome (Figure 7). The normalized microarray data indicated that genes of chromosomes 22 and 28 were overexpressed 1.7- and 1.5-fold, respectively, in the resistant strain L. major MTX60.4 compared to the wild-type strain. The expression of genes on chromosomes 11 and 12 seemed, in general, to be 50% underexpressed in the mutant strain compared to wild-type cells (Figure 7).

A number of hypotheses can explain this whole chromosome-specific gene regulation and we tested whether the copy number of specific chromosomes changed upon MTX selection in L. major MTX60.4. Quantitative Southern blot analyses with two distinct probes derived from chromosome 22 revealed that if the wild-type cells contain two homologous copies of chromosome 22 (Leishmania is a diploid organism), L. major MTX60.4 had four copies (Figure 7a, lanes 1 and 2). Similarly, L. major MTX60.4 had three copies of chromosome 28 compared to wild-type cells (Figure 7b, lanes 1 and 2). The probes used are physically far apart, indicating a change in ploidy of the whole chromosome. However, this change in chromosome copy number was not observed for chromosomes 11 and 12 (Figure 7c,d). Aneuploidy of specific chromosomes and drug resistance has been described in cancer cells (reviewed in 61) and fungi 6263. To test this possibility, we generated a revertant line of L. major MTX60.4 by successive passages in the absence of MTX; under these conditions, resistance to the drug decreased (Figure 1b). Revertant cells were not as sensitive as wild-type cells to MTX but this is expected as a deletion of FT1 (Figure 6) will lead to resistant parasites 19. The aneuploidy of chromosomes 22 and 28 regressed to diploidy (similar to wild-type diploidy) after 30 passages, thus circumstantially linking resistance levels (Figure 1b) and copy number of these chromosomes (Figure 7a,b, lanes 2-6). With the cells now diploid, additional passages (for example, passage 42) did not decrease resistance further.

Since several of the changes in RNA levels were correlated with gene amplification or gene deletion, we undertook a comparative genomic hybridization (CGH) study using the full genome array. The DNA of mutant L. major MTX60.4 was labeled and changes in copy number in comparison to sensitive wild-type cells were measured using CGH. The CGH data are represented in a chromosome by chromosome fashion in Figure S4 in Additional data file 3. A qualitative correlation was observed between CGH and RNA-based hybridization (Figure 8). Indeed, amplification of the DHFR-TS locus, derived from chromosome 6, was easily detected by both techniques and quantification of the DNA amplification was compared to RNA levels (Figure 4). The deletion of FT1 was also detected by CGH and the latter technique was found to be quantitative. Indeed, the 70-mers recognizing FT1 recognized three conserved FT genes. In the MTX60.4 mutant two of these genes are deleted, hence explaining the ratio of 0.33 obtained by CGH (Figure 6). Polyploidy was also easily detected by CGH (Figure 8). Indeed, a similar qualitative pattern of hybridization intensities was obtained for both RNA expression profiling and CGH (Figure 8). Interestingly, while RNA expression profiling showed that chromosome 11 was down-regulated, quantitative Southern blots indicated that the copy number of the chromosome remained unchanged (Figure 7). This was also confirmed by CGH (Figure 8). There are some differences, however, between RNA expression profiling and CGH. For example, the latter technique showed that chromosome 2 is polyploid (Figure S4 in Additional data file 3) but this is likely due to the dynamic process of cell culture and parasite evolution, as DNA and RNA were prepared 1.5 years apart, rather than a difference in the techniques.

The wild-type strain L. major LV39 and the mutants L. major MTX60.4 have been described previously 65. The L. infantum strain (MHOM/MA/67/ITMAP-263) was selected in vitro in a stepwise fashion starting with its EC₅₀ (0.5 μM) with doubling concentrations of MTX when cells were adapted to yield L. infantum MTX20.5 growing at 20 μM of MTX. All cells were grown in M199 medium supplemented with 10% heat-inactivated fetal bovine serum and 5 μg/ml hemin at 25°C.

Chromosomes in agarose blocks were prepared and separated by PFGE as described previously 38. For Southern blot and PCR, genomic DNA was isolated using the DNAzol technique (Invitrogen, Carlsbad, CA, USA) as recommended by the manufacturer. Southern blots, hybridization, and washing conditions were done following standard protocols 80. For chromosome copy number investigation, Southern spots were quantified using ImageQuant 5.2 (GE Healthcare, Upsala, Sweden) and the reference gene α-tubulin was used for normalization.

The recent completion of the sequence of the L. major 81 and L. infantum 82 genomes, allowed the generation of multispecies high-density oligonucleotide microarrays. Our analysis of open reading frame sequence conservation between L. major and L. infantum revealed that these two species share 91-96% nucleotide identity, suggesting that interspecies microarray probes can be designed. Therefore, 70-mer oligonucleotides were designed for each open reading frame of L. infantum and L. major using automated bioinformatic procedures. The genomes of both species were first compared using BLAST and homologous genes were grouped together. Probes were designed with consistent thermodynamic properties. Probes were initially designed for L. infantum with the added requirement that the region targeted by the probes had perfect homology between both species. For common probes, up to 2 mismatches (out of 70 nucleotides) were tolerated. In the case that more than two mismatches were present in a given gene between L. infantum and L. major, a new probe was designed specifically for L. major (956 probes). The microarray included a total of 8,978 70-mer probes that recognized with no mismatches all L. infantum genes (8,184, GeneDB version 3) and also all L. major genes (8,370, GeneDB version 5.1) with a small percentage of the probes having at most 2 mismatches. Also, 372 control probes were included in the microarray for assessing synthesis variability, and location of the probe within a given open reading frame and of mismatches on hybridization. The probes were synthesized in 384-well plates by Invitrogen. The microarrays were printed on SuperChip (Erie Scientific, Portsmouth, NH, USA) using a BioRobotics MicroGrid (Genomic solutions Inc, Ann Arbor, MI, USA). Each probe was printed in duplicate. Our microarray platform is described in the Gene Expression Omnibus (GEO) with accession number GPL6855.

Total RNA was isolated from 10⁸ Leishmania cells during the mid-log phase using RNeasy Plus Mini Kit (QIAGEN, Hilden, Germany). The RNA preparation was treated with TURBO DNase (Ambion, Austin, TX, USA) to avoid any genomic contamination. The purity, integrity and quantity of the RNA were assessed on the Agilent 2100 bioanalyzer with the RNA 6000 Nano LabChip reagent set (Agilent Technologies, Santa Clara, CA, USA). For each probe, 10 μg of RNA were converted to aminoallyl-dUTP incorporated cDNA using random hexamers (Roche, Basel, Switzerland) and the SuperScript III RNase H Reverse Transcriptase (Invitrogen). Probes were thereafter coupled to the fluorescent dye Alexa Fluor555 or Alexa Fluor647 (Invitrogen) following the manufacturer's recommendations. Fluorescent probes were then purified with MinElute Spin Columns (QIAGEN) and quantified spectrophotometrically.

Genomic DNA from 10⁸ cells was isolated using the DNAzol technique (Invitrogen) as recommended by the manufacturer. Total DNA was then fragmented by successive passages through 22G1" and 27G 1/2" needles (Becton Dickinson Franklin Lakes, NJ, USA). Fragmented DNA was then double digested with PvuII and MseI restriction enzymes. Digested DNA was purified by phenol-chloroform, followed by an ethanol precipitation. For each probe, 4 μg of purified fragmented and digested genomic DNA were converted to fluorescently labeled DNA using Cy5- or Cy3-dCTP (Amersham, Piscataway, NJ, USA), random hexamers (Roche) and the exo^- Klenow DNA polymerase (NEB, Ipswich, MA, USA). Fluorescent probes were then purified with ArrayIt columns (TeleChem International, Sunnyvale, CA, USA) and quantified spectrophotometrically.

Prehybridization and hybridization were performed at 42°C under immersion (Corning chambers, Corning, NY, USA). Slides were prehybridized for 90 minutes in PreHYB Solution (5× Denhardt, 30% formamide, 6× SSPE, 0.5% SDS, 100 μg/ml salmon sperm DNA). Then, slides were first washed 2 times at 42°C for 5 minutes in 2× SSC, 0.1% SDS with gentle agitation. Subsequent washes were at room temperature, 3 minutes each, in 1× SSC, 0.2× SSC and 0.05× SSC. Slides were then dipped in 100% isopropanol and dried by centrifugation. For hybridization, Alexa Fluor555 and 647 cDNA probes were dried and resuspended in the HYB solution (2.5× Denhardt, 30% formamide, 6× SSPE, 0.5% SDS, 100 μg/ml salmon sperm DNA, 750 μg/ml yeast tRNA), then mixed, denatured 5 minutes at 95°C and cooled slowly to 42°C. Mixed probes were applied on the array under a lifterslip. Hybridization was performed for 16 h. Washes after hybridization were the same as those described for the prehybridization.

The Perkin Elmer ScanArray 4000XL Scanner was used for image acquisition (Perkin Elmer, Waltham, MA, USA). GenePix Pro 6.0 image analysis software (Axon Instruments, Union City, CA, USA) was used to quantify the fluorescence signal intensities of the array features. Four different RNA preparations of each mutant and their respective wild-type strain were analyzed, including dye-swaps. Raw data from GenePix were imported in R 2.2.1 for normalization and statistical analyses were performed using the LIMMA (version 2.7.3) package 838485. Before processing, probes were flagged according to the hybridization signal quality 86. Weights were assigned to each array in order to give less weight to arrays of lesser quality 87. Data were corrected using background subtraction based on convolution of normal and exponential distributions 88. Intra-array normalization was carried out using the 'print-tip loess' statistical method and inter-array normalization was done by using the 'quantiles of A' method for each array 89. Statistical analysis was done using linear model fitting and standard errors were moderated using a simple empirical Bayes 83. Multiple testing corrections were done using the FDR method with a threshold p-value of 0.05. Only genes statistically significant with an absolute log ratio greater than 0.58 (log₂ 1.5) were considered as differentially expressed. Species comparison was performed only on probes that had less than two mismatches when hybridized to either Leishmania species. GeneSpring GX 3.1 was used for the generation of scatter plots and for chromosome by chromosome analysis. The entire data set has been deposited in GEO under the accession number series GSE9949. The comparative genomic hybridization data are deposited under reference number GSE11623.

Three independent RNA preparations were conducted for each condition. First-strand cDNA was synthesized from 2 μg of total RNA using the Superscript III RNase H Reverse Transcriptase enzyme and random hexamers (Roche) according to the manufacturer's instructions. The resulting cDNA samples were stored at -20°C until use. Control PCR amplification was carried out using primers from different internal controls (GAPDH and actin) to evaluate the uniformity of cDNA synthesis in different samples. Primers, TaqMan probes, experimental procedures and quantification for qRT-PCR of the folate transporter genes was as described (AA Ouameur et al., unpublished data) using the glyceraldehyde-3-phosphate dehydrogenase gene (GAPDH) for normalization. For all other genes, equal amounts of cDNA were run in triplicate and amplified in a 15 μl reaction containing 7.5 μl of 2× Universal PCR Master Mix (Applied Biosystems, Foster City, CA, USA), 10 nM of Z-tailed forward primer, 100 nM of reverse primer, 250 nM of Amplifluor Uniprimer probe (Chemicon Int., Temecula, CA, USA), and 1 μl of cDNA target. Reactions were performed at the Gene Quantification core laboratory of the Centre de Génomique de Québec using the Applied Biosystems Prism 7900 Sequence Detector 90. Amplification was normalized to two genes showing a highly stable expression in wild-type and resistant strains: LinJ18_V3.0630/LmjF18.0620 encoding a putative 60S ribosomal protein L10a, and LinJ36_V3.0850/LmjF36.2500 encoding a chromatin assembly factor 1 subunit b-like protein.