Miltefosine (MF) is the first oral compound used in the chemotherapy against leishmaniasis. Since the mechanism of action of this drug and the targets of MF in Leishmania are unclear, we generated in a step-by-step manner Leishmania major promastigote mutants highly resistant to MF. Two of the mutants were submitted to a short-read whole genome sequencing for identifying potential genes associated with MF resistance.
Analysis of the genome assemblies revealed several independent point mutations in a P-type ATPase involved in phospholipid translocation. Mutations in two other proteins—pyridoxal kinase and α-adaptin like protein—were also observed in independent mutants. The role of these proteins in the MF resistance was evaluated by gene transfection and gene disruption and both the P-type ATPase and pyridoxal kinase were implicated in MF susceptibility. The study also highlighted that resistance can be highly heterogeneous at the population level with individual clones derived from this population differing both in terms of genotypes but also susceptibility phenotypes.
Whole genome sequencing was used to pinpoint known and new resistance markers associated with MF resistance in the protozoan parasite Leishmania. The study also demonstrated the polyclonal nature of a resistant population with individual cells with varying susceptibilities and genotypes.
Leishmania spp. are parasitic protozoa responsible for a spectrum of diseases known as leishmaniasis. There are few drugs available for the treatment of these diseases, and miltefosine is the first oral drug used in treatment of visceral leishmaniasis, a form of the disease that can be lethal if not treated. In this study, we seek to understand the mechanism of action and identify targets of the drug by generating promastigote mutants highly resistant to miltefosine. Two independent mutants were submitted to short read whole genome sequencing. Genome analysis of these mutants has permitted us to identify point mutations in three genes (P-type ATPase, pyridoxal kinase and α-adaptin like protein) that were also present in other independent miltefosine resistant mutants. Some of the new genes identified here could be useful as potential markers for miltefosine resistance in Leishmania. Moreover, our approach has permitted us to highlight that resistance can be highly heterogeneous at the population level with individual clones derived from this population differing both in terms of genotypes but also susceptibility phenotypes. This may have practical applications while studying resistance.
Citation: Coelho AC, Boisvert S, Mukherjee A, Leprohon P, Corbeil J, et al. (2012) Multiple Mutations in Heterogeneous Miltefosine-Resistant Leishmania major Population as Determined by Whole Genome Sequencing. PLoS Negl Trop Dis 6(2): e1512. doi:10.1371/journal.pntd.0001512
Editor: Peter J. Myler, SBRI, United States of America
Received: August 26, 2011; Accepted: December 20, 2011; Published: February 14, 2012
Copyright: © 2012 Coelho et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: ACC and AM were Strategic Training Fellows of the Strategic Training Program in Microbial Resistance, a partnership of the Canadian Institutes of Health Research (CIHR) Institute of Infection and Immunity and the Fonds de Recherche en Santé du Québec. SB is recipient of a doctoral award from the CIHR (200910GSD-226209-172830). JC holds the Canada Research Chair in Medical Genomics and has a resource allocation project from Compute Canada (nne-790-ab, colosse/CLUMEQ/Calcul Québec). MO is supported by CIHR operating and group grants and is a Burroughs Wellcome Fund Scholar in Molecular Parasitology and holds the Canada Research Chair in Antimicrobial Resistance. The funders have no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Leishmania is a protozoan parasite responsible for a spectrum of diseases collectively known as leishmaniasis in tropical and subtropical areas of the world . There is no effective vaccine for the prevention of this parasitic disease and its control relies on chemotherapy. The arsenal of available drugs is limited with most compounds being compromised by toxicity, cost, or resistance . The alkyl-lysophospholipid analogue miltefosine (MF), a drug initially developed as an antitumoral compound, was the first effective oral drug against Leishmania. It has been successfully used in the last decade for the treatment of visceral leishmaniasis –. The exact mode of action of MF is not well understood, but it was shown to induce changes in the biosynthesis of phospholipids and the metabolism of alkyl-lipids in Leishmania , in addition to induce an apoptotic cell death , . The uptake of MF and other alkyl-glycerophospholipids in Leishmania requires a P-type ATPase, named miltefosine transporter (MT), which is responsible for the translocation of phospholipids from the exoplasmic to the cytoplasmic leaflet of the plasma membrane of the parasite . Based on findings in yeast , , the uptake of MF in L. donovani was further shown to require a protein named LdRos3, the β-subunit of the MT . Differences in susceptibility to alkyl-lysophospholipids between Leishmania species ,  has recently been associated with the low expression of this MF translocation machinery in species displaying a lack of intrinsic MF susceptibility .
Resistance to MF in Leishmania  and other organisms ,  involved a decreased accumulation of the drug due to reduced uptake or increased efflux. In L. donovani, a drastic reduction (40-fold decrease) in the ability to internalize the drug was shown to be accompanied by a decreased content of unsaturated phospholipids in the plasma membrane of resistant parasites , . The acquisition of inactivating mutations or the deletion of the MT gene in L. donovani ,  was shown to drastically increase MF resistance both in vitro and in vivo. In addition, the overexpression of ABCG4 and ABCG6, two ABC proteins localized to the flagellar pocket and plasma membrane of the parasite, was associated with an increased resistance to several alkyl-lysophospholipid analogues in Leishmania –. Parasites overexpressing ABCG4 or ABCG6 had a reduced accumulation of fluorescent phospholipids and [14C]-MF, which suggested an enhanced outward translocation of the drug across the plasma membrane. A L. tropica mutant resistant to daunomycin harboring an amplification of the ABC protein ABCB4 (MDR1) was also shown to display cross-resistance to MF , although the absence of similar amplicons in MF-resistant L. donovani suggested that the amplification of ABCB4 (MDR1) is not a frequent mechanism of resistance to MF . Finally, functional cloning experiments identified a hypothetical gene in L. infantum whose increased expression conferred resistance to MF and antimony .
The central role of MF in the control of leishmaniasis warranted further studies on its mode of action and the mechanisms involved in resistance. Indeed, the ease with which MF-resistant parasites are selected in vitro suggests that clinical failures resulting from MF-resistant isolates are likely. In this study, we sought to determine the mutational events involved in MF resistance on a whole genomic scale by sequencing the entire genome of two L. major mutants independently selected for high level MF resistance in vitro.
Cell lines, culture conditions and transfections
L. major Friedlin and L. infantum (MHOM/MA/67/ITMAP-263) wild-type promastigotes were grown at 25°C in SDM-79 medium supplemented with 10% heat inactivated fetal bovine serum and 10 µg/ml hemin. The L. major Friedlin MF80.1, MF80.2, MF80.3 and MF80.5 mutants were selected from a cloned parental population using a stepwise selection until they were resistant to 80–100 µM MF. MF was purchased from Cayman Chemical (Ann Harbor, USA). To obtain clones derivied from the mutants 5.3, 10.3, 15.3 40.3 and 80.3, parasites were spread on SDM-agar (1% Noble Agar, Nunc) plates in absence of MF.
Partial revertants were obtained by culturing the resistant lines in the absence of MF for 30 passages. The L. infantum MF200.3 and MF200.5 mutants selected for MF resistance had been generated previously . Growth curves were obtained by measuring absorbance at 600 nm as previously described . Gene transfection was performed by electroporation as reported previously . Statistical significance was determined by Student's t-test. Significance was considered as P<0.05.
Total DNA was isolated using DNAzol reagent (Invitrogen) as recommended by the manufacturer. For quantitative Southern blots, the genomic DNA was digested with appropriate restriction enzymes and migrated in 0.8% agarose gels. Southern blots, hybridizations and washes were performed following standard protocols . Probes were obtained by amplifying the upstream or downstream regions of the gene studied. The pSP72αHYGα-MT construct was generated by the amplification of the MF transporter gene from L. major Friedlin (LmjF13.1350) with primers containing the restriction enzymes XbaI and HindIII (Table S1) and cloned in the respective plasmid. All the constructs were checked by sequencing with internal primers (Table S1). Similarly, the genes pyridoxal kinase (LmjF30.1250) and α-adaptin like protein (LmjF07.0050) were cloned in pSP72αNEOα or pSP72αHYGα at XbaI and HindIII sites.
The pyridoxal kinase gene (LmjF30.1250) inactivation cassette was generated by overlap extension PCR ,  using the primer A-PK-KO and primer B-PK-NEO-KO to amplify a region of 550-bp upstream of the gene; primer E-PK-NEO-KO and primer F-PK-KO to amplify a DNA fragment of 600-bp downstream of the gene; and primers C-PK-NEO-KO and primer D-PK-NEO-KO, to amplify the NEO gene. The fragment generated was cloned in the pGEM-T-easy vector leading to the pGEM-T-KO-NEO-PK. The LmjF30.1250 inactivation cassette was isolated from pGEM-KO-NEO-PK by an EcoRI digestion and transfected in L. major Friedlin. A second inactivation cassette containing a HYG marker was generated using the A and F primers described above along with primers B-PK-HYG-KO, D-PK-HYG-KO and E-PK-HYG-KO (Table S1). The fragment generated was cloned in the pGEM-T-easy vector leading to the second construct pGEM-T-KO-HYG-PK to inactivate the PK gene. The integration of the inactivation cassette at the LmjF30.1250 locus was confirmed by PCR and Southern blots. A rescue vector containing the wild-type gene was generated by PCR amplification using the primers PKF and PKR, cloned in pGEM-T easy vector, sequenced and then subcloned in pSP72αZEOα generating the plasmid pSP72αZEOα-PK. All the primers used are listed in Table S1.
Genome resequencing and analysis
Genomic DNAs were prepared from mid-log phase cultures of L. major Friedlin mutants MF80.3 and MF80.5 and L. major Friedlin wild-type parasites and sequenced using Illumina Genome AnalyzerIIx short 76-nucleotide single-end reads. We had a genome coverage over 50-fold for the two independent mutants and the wild-type parasite. This strategy allowed us to identify point mutations and small indels (≤3 bp) when comparing with the known genome sequence of L. major Friedlin .
Sequence reads from each clone were aligned to the L. major Friedlin reference sequence available (TriTrypDB version 3.3)  using the software bwa (bwa aln, version 0.5.9) with default parameters . The maximum number of mismatches was 4, the seed length was 32 and 2 mismatches were allowed within the seed. The detection of single nucleotide polymorphisms (SNPs) was performed using samtools (version 0.1.18), bcftools (distributed with samtools) and vcfutils.pl (distributed with samtools) , with a minimum of three reads to call a potential variation prior to further analysis. The sequence data for L. major Friedlin wild-type, MF80.3 and MF80.5 are available at the EMBL European Nucleotide Archive (http://www.ebi.ac.uk/ena) (accession number ERP000917; samples ERS056015, ERS056016 and ERS056017 corresponding to L. major Friedlin wild-type; MF80.3 and MF80.5 respectively). Several python (version 2.4.3) scripts and bash (version 3.2) scripts were created to further analyse the data. The quality assessment software samstat (v1.08) was used to generate quality reports .
All the putative point mutations detected by whole genome sequencing were verified by PCR amplification and conventional DNA sequencing. The primers were designed using Primer3 program .
Characterization of L. major Friedlin MF-resistant mutants
Four independent L. major Friedlin MF-resistant mutants (MF80.1, 80.2, 80.3 and 80.5) were generated by a stepwise fashion until they reached a MF EC50 of at least 80 µM (Fig. 1). The MF80.3 and MF80.5 lines displayed the highest levels of resistance, with at least 15 fold increase in resistance compared to the wild-type (WT) parent. The phenotype of MF resistance appeared relatively stable, as culturing the parasites in absence of MF for 30 passages led to only a small albeit significant decrease in resistance in both MF80.3 and MF80.5 (Fig. 1, lanes 6, 7) but also in MF80.1 and MF80.2 (data not shown). Because of the small decrease of resistance, we tested whether MF resistance in our mutants was accompanied by gene amplification, a frequent mechanism of drug resistance in Leishmania , . However, neither alkaline lysis nor clamped homogeneous electrical field electrophoresis revealed the presence of circular or linear amplicons in any of MF resistant mutants (results not shown).
The EC50 values were determined by culturing promastigote parasites in the presence of increasing concentrations of MF. The average of three independent experiments is shown. EC50 values for MF in L. major wild-type parasites (1); MF80.1 (2); MF80.2 (3);MF80.3 (4); MF80.5 (5) MF-resistant mutants. The MF80.3 and MF80.5 mutants were grown in the absence of MF for 30 passages (6 and 7 respectively). * p<0.05; ** p<0.01.
Whole-genome sequencing of L. major Friedlin MF-resistant mutants
The genomes of one clone derived from the wild-type L. major, MF80.3 and MF80.5 mutants were resequenced using the single-end reads Illumina platform. We obtained a total of 30,083,681, 28,369,175 and 28,259,044 reads leading to an average genome coverage of 52-fold for the wild-type L. major Friedlin strain and MF80.3 and MF80.5 respectively. In all cases, close to 80% of the reads were aligned to the L. major Friedlin genome. The unmapped reads (25.1%, 20.2% and 18.5% respectively for L. major wild-type, MF80.3 and MF80.5) had low mean base quality according to samstat (version 0.08)  with slightly less than half mapping to mitochondrial DNA. Read depth coverage can be used to predict copy number variations being either amplifications or deletions. We investigated copy number of chromosomal regions by read depth coverage for each chromosome of the resistant mutants which were compared to the wild-type sensitive L. major Friedlin. This analysis did not reveal changes in read depth between sensitive and resistant isolates that would constitute signatures of copy variation, hence confirming the absence of gene amplification in these mutants but also indicating no obvious gene/locus deletion or changes in ploidy (data not show). While no change in copy number was related to resistance, we found that some chromosomes were polyploid in both L. major Friedlin wild-type cells and in the two MF-resistant clones. Chromosome 1, 5, 11, 20 and 23 were in three copies while chromosome 31 was at least in six copies in the wild-type cells but also in the mutants (Fig. S1A–C).
While no changes in copy number were observed, the resequencing of the L. major Friedlin wild-type strain and of the mutants differed from the published sequence . Indeed, 20 homozygous (Table S2) and 42 heterozygous SNPs (Table S3) were observed. The polymorphic sites identified here in the coding sequences were highlighted by resequencing the wild-type Friedlin strain, the two MF-resistant strains and also other L. major Friedlin resistant mutants that we are currently analysing in the lab. A majority of polymorphisms (68%) changed the amino acids of the encoded proteins (Table S2 and S3). Thus, the majority of putative mutations found by this next-generation sequencing approach corresponded to natural polymorphisms present in the parental clones and that were not selected for during the drug pressure.
However, whole genome sequencing of the two miltefosine resistant mutant clones also revealed genuine mutations (Table 1). Indeed, a three nucleotide deletion was seen in in the MT gene of MF80.3 (LmjF13.1530) and a transition mutation leading to a mutated G852D protein version was observed in mutant MF80.5 (Table 1). These mutations highlighted by next generation sequencing were confirmed by PCR amplification and standard DNA sequencing and the same validation approach was used for all other mutations that will be discussed afterwards. Interestingly, another mutation was in a common gene for both mutants. It corresponded to the LmjF30.1250 gene that code for a putative pyridoxal kinase (PK). The MF80.3 mutant acquired a mutation in both alleles that led to a T216M amino acid substitution while one wild-type allele and one allelic version leading to a G269D protein were observed in the MF80.5 line (Table 1). Other mutations were specific to either MF80.3 or MF80.5. Three heterozygous mutations in LmjF15.0800 and LmjF.27.1250, both coding for hypothetical proteins, and in LmjF36.5880 coding for a small GTPase were found in MF80.3 (Table 1). The sequences of the MF80.5 clone allowed the detection of a homozygous mutation in LmjF07.0050 coding for an α-adaptin like protein and an heterozygous mutation in the hypothetical protein coding gene LmjF21.0835 (Table 1).
Since few genuine point mutations were highlighted by whole genome sequencing, we next amplified and sequenced these specific mutated genes in additional L. major and L. infantum MF resistant mutants that were available. This targeted approach revealed that only the MT and PK genes were mutated in the four independent L. major resistant mutants (Table 2), and that MT was also mutated in the two L. infantum mutants investigated (Table 2). The mutations were at different sites in the L. infantum or L. major mutants (Table 2), but these mutations occurred at conserved sites within the MT protein (Fig. S2). The targeted sequencing of the PK gene in the other L. major mutants revealed that they were also heterozygous for mutations leading to G269D or T216M amino acid substitutions, respectively (Table 2). However, the PK gene was not mutated in the two L. infantum MF resistant mutants analysed (Table 2). The LmjF07.0050 α-adaptin gene was the only other mutation that was not strain-specific since the gene was also mutated in the two independent L. infantum MF resistant mutants (Table 2 and Fig. S3). The four remaining mutations highlighted by next-generation sequencing were unique to either MF80.3 or MF80.5 (Table 2). Surprisingly, we did not observe point mutations in the Ros3 β-subunit of MT, a known MF-resistance gene , in any of the miltefosine resistant mutants analyzed (Table 2).
We were intrigued by the diversity of mutations in the MT gene between the different mutants (Table 2) suggesting that mutations in MT could be polyclonal. We investigated the MT status in clones (and population) isolated at intermediate passages leading to the generation of MF80.3. Most interestingly, no mutations in MT were found in any of the resistant populations leading to MF80.3 (Fig. 2), suggesting indeed that resistance can be polyclonal, and since no clones are dominant, mutations can be missed when sequencing the population. In none of the 10 clones investigated derived from MF5.3, MF10.3, or MF15.3 could we detect any mutations in MT. However in the 10 clones derived from MF40.3 where the MT gene was sequenced, all clones had a mutated MT and 5 different genotypes were observed. Three clones had a three nucleotide deletion leading to the deletion of M547, two had a deletion of M546, one clone had a D567V mutation and 4 clones had a L235R mutation (3 homozygous and one heterozygous) (Fig. 2). The three clones derived from MF80.3 had the deletion of M547 (Fig. 2). The situation was different in the lineage leading to MF80.5. Indeed, the homozygous G852D mutation found in the MT of the sequenced clone (Table 1) was also detected in the MF80.5 population but also in the MF60.5 population and the mutation was heterozygous at the earlier passages of MF20.5 and MF40.5 (Table 3). Similarly to the lineage leading to MF80.3, mutations in the MT gene were observed only at the highest drug concentration. In contrast to MT, mutations in the PK gene LmjF30.1250 appeared during the first selection of MF selection in both MF80.3 and MF80.5 mutants (Fig. 2 and Table 3). While the mutation remained heterozygous in the MF80.5 lineage (Table 3), the number of mutated alleles in the PK gene in the MF80.3 lineage, with the exception of MF40.3 (see below), correlated with resistance levels (Fig. 2).
The top and bottom half circle represents the MT and PK genes respectively. White represents wild-type sequences; semi-colored an heterozygous mutation and fully colored an homozygous mutation. The type of mutation is also indicated.
Functional analysis of genes involved in MF resistance
The role in resistance of the MT mutations highlighted in one clone of MF80.3 and MF80.5 was further studied by gene transfection experiments. Episomal transfection of the wild-type copy of MT sensitized wild-type cells to MF by 3-fold whereas transfections of the G852D or M547- versions cloned in the same vector did not change the phenotype (Fig. 3A, bars 2–4). Transfection of the wild-type gene copy of MT in MF80.3 and MF80.5 sensitized the mutants to MF although not to wild-type levels (Fig. 3B and C, bar 2). Significantly, transfection of the mutated versions of MT neither sensitized the mutant cells nor the wild-type cells, showing that these mutations alter the function of MT (Fig. 3, bars 3, 4).
The EC50 values were determined by culturing promastigote parasites in the presence of increasing concentrations of MF. The average of three independent experiments is shown. LmjF WT (A), MF80.3 (B) and MF80.5 (C) parasites were transfected with pSP72αHYGα (1); pSP72αHYGα-MT (2); pSP72αHYGα-MT* (G852D) (3); pSP72αHYGα-MT* (M547del) (4); pSP72αNEOα-PK (5); and pSP72αHYGα-adaptin like protein (6). * p<0.01; ** p<0.001.
The role of the PK and α-adaptin-like proteins that were mutated in more than one mutant (Table 1), in MF resistance was also assessed by gene transfection experiments. In contrast to the MT gene, however, the overexpression of WT versions of the PK or α-adaptin-like genes failed to alter the MF resistance levels of L. major WT (Fig. 3A, bars 5 and 6), MF80.3 (Fig. 3B, bars 5 and 6) and MF80.5 mutants (Fig. 3C, bars 5 and 6). Transfection of mutated versions of the PK and α-adaptin-like genes also failed to alter the MF susceptibility status of the wild-type L. major Friedlin strain (data not shown). The PK mutations were acquired during the first step of selection in both lines, prior to the selection of mutations in the MT (Fig. 2 and Table 3). To further investigate the role of PK in MF resistance, we investigated clones at different level of selection giving rise to MF80.3. The genotype of PK varied with clones being either heterozygous (WT/T216M) or T216M homozygous mutants (Fig. 2). The level of resistance to MF and the number of mutated PK alleles was not straightforward but a wild-type PK allele has never been found in MF resistant cells (Fig. 2, and Table 3). Only heterozygous mutations were found in clones derived from MF5.3 but in MF10.3 we found a mixture of clones with either homozygous or heterozygous mutations in PK, while all clones derived from MF15.3 had homozygous T216M mutation in PK (Fig. 2). In the MF40.3 clone, we observed cells that were surprisingly heterozygous for PK but these cells have a mutation in MT, a mutation not present in the MF15.3 derived clones (Fig. 2) and possibly the establishment of mutations in MT is facilitated in cells still having some wild-type PK activity. When the MF40.3 population was selected for higher resistance to yield MF80.3, all clones exhibited the same PK homozygous mutation (Fig. 2).
We also investigated the susceptibility levels of selected clones for the earlier passages. While, as expected, the population of MF5.3, MF10.3 and MF15.3 became more and more resistant to MF (Fig. 4) we found variations in the EC50 between the individual clones derived from the population. Indeed, some clones were as resistant as the population and others had susceptibilities close to wild-type levels (Fig. 4). For example, clones 4 and 10 derived from MF5.3 had the same genotype (for MT and PK) but clone 4 had an EC50 close to wild-type while clone 10 was resistant to MF (Fig. 4). Similarly, the cloning of MF10.3 led to parasite with either homozygous (clone 3) or heterozygous (clone 7) mutations in PK and clone 3 was found to be more resistant (Fig. 4). However, clones 1 and 2 derived from MF15.3 had the same genotype (for MT and PK) but their resistance to MF varied (Fig. 4).
Drug susceptibility of wild-type cells (white bars) and MF-resistant mutant population (P) or clones (gray shaded bars) were measured. The clones were derived from MF5.3, MF10.3 and MF15.3 population and clone numbers are relate to Fig. 2. The average of three independent measures is shown. * p<0.05; ** p<0.01; *** p<0.001.
While no susceptibility phenotype was observed in transfecting PK in MF80.3 (Fig. 3B), the MF15.3 mutant is homozygous for the T216M PK substitution and still has a wild-type version of MT (Fig. 2). The transfection of the wild-type PK-containing plasmid in MF15.3 resulted in the reversion of MF resistance (Fig. 5). In contrast, the transfection of theT216M version had no effect on MF susceptibility (Fig. 5). We next investigated whether deletion of PK was possible and tested how this could modulate MF susceptibility. Two replacement cassettes containing the neomycin phosphotransferase (NEO) or hygromycin phosphotransferase (HYG) genes were generated for replacing PK (Fig. 6). Since Leishmania is diploid, the sequential integration of the NEO or HYG inactivation cassette should lead to a 1.5 and 1.8 kb DNA fragment, respectively, in addition to the remaining WT PK fragment of 1.2 kb when the DNA of transformants is digested with PstI and hybridized to a 3′UTR PK probe corresponding to a 600 bp fragment upstream the PK gene stop codon (Fig. 6A). The generation of NEO single disruptant led to the expected genotype (Fig. 6B, lane 2). Introduction of the HYG construct into a PK/NEO parasite indicated that while both NEO and HYG alleles integrated properly, there was always a remaining wild-type allele (Fig. 6B, lane 3), a phenomenon often observed when attempting to inactivate essential genes –. Parasites with a single allele of PK inactivated with NEO were more sensitive to MF than the control-transfected parasites (Fig. 6C) and this phenotype could be partially rescued by the addition of an episomal construct encoding a functional copy of PK (Fig. 6C).
The EC50 values were determined by culturing promastigotes parasites in the presence of increasing concentrations of MF. The average of three independent experiments is shown. L. major MF15.3 transfected with the empty vector pSP72αNEOα (1); pSP72αNEOα-PK (2); or pSP72αNEOα-PK* (T216M) (3). * p<0.05.
(A) A schematic drawing of the PK locus with PstI sites. (B) Southern blot analysis of L. major genomic DNA digested with PstI and hybridized with a 3′ UTR probe of PK gene (a 600 bp fragment just upstream the stop codon of PK gene). Lane 1, L. major wild-type; lane 2, L. major with one PK allele disrupted with the NEO marker; lane 3, L. major with two alleles disrupted (HYG and NEO). Molecular weight is indicated on the right and the alleles are indicated on the left. (C) Miltefosine susceptibilities of Leishmania major wild-type (1), SKO-PK (2) and SKO-PK complemented with a wild-type PK gene (3) parasites. The EC50 values were determined by culturing promastigotes parasites in the presence of increasing concentrations of MF. The average of three independent experiments is shown. * p<0.01.
We sequenced two L. major MF-resistant mutants in order to identify mutations putatively associated with MF resistance. Our strategy was based on sequencing two independent resistant strains and to assess the role in resistance of recurrent mutations. Leishmania often amplifies regions of its genome in response to drug selection , , but this does not appear to be the case for the MF resistant strains as determined by read depth sequence analysis (Fig. S1A–C) and also further confirmed by comparative genomic hybridization experiments (unpublished observations). The absence of amplicons in our MF-resistant L. major is consistent with the previous characterization of MF resistant strains of L. donovani . However, the analysis revealed the presence of point mutations in protein coding genes for both L. major MF-resistant mutants (Table 1). Three mutations were found in more than one mutant: the miltefosine transporter MT LmjF13.1530, the pyridoxal kinase PK LmjF30.1250, and the α-adaptin like protein LmjF07.0050 (Table 2).
The MT protein is a P-type ATPase that mediates the inward translocation of MF and glycerophospholipid analogues in Leishmania , . The acquisition of inactivating mutations in the MT gene was shown to be a major determinant of resistance to MF in L. donovani and it appears that several distinct mutations in MT can confer resistance in both promastigote and amastigotes parasites , . The mutant Li MF200.5 with the MT gene mutated, was indeed shown to be MF resistant using an intramacrophagic essay (results not shown). Although several distinct MT point mutations have also been identified in our panel of L. major MF-resistant mutants (Table 2), they all differed from those previously reported in L. donovani. These mutations were located at conserved residues within the P-type ATPase protein (Fig. S2). The number of mutated alleles in MT correlated with the level of resistance to MF with the most resistant strains being homozygous mutants. Mutations in MT only occur at later step of drug selection (Fig. 2, Table 3).
Prior to MT mutations, the only other recurring mutation in low-level resistant cells was the acquisition of the T216M or G269D amino acid substitutions in the PK gene. Similarly as for MT, the level of MF resistance correlated with the number of mutated PK alleles in the MF 80.3 lineage (Fig. 2). As explained in the result section, it is possible that mutations in MT happens in a minority population of cells with heterozygous PK alleles and this could explain how cells appear to be mostly homozygous for a mutant allele in PK (15.3), and then becomes heterozygous (40.3).. Pyridoxal-5′-phosphate (PLP) is the active form of vitamin B6 and is an essential cofactor of several enzymes, predominantly involved in the transformation of amino acids , . Leishmania cannot synthesize PLP de novo and rely on the scavenging of vitamin B6 precursors like pyridoxal, pyridoxine and pyridoxamine . PK was shown to catalyze the ATP-dependent phosphorylation of these precursors and to play an essential role in the formation of the biologically active PLP . The role of PK in MF susceptibility warrants further studies and may be due to the many roles of PLP in cell physiology. Point mutations in PK are found in the four L. major MF-resistant mutants (Table 2) and transfection of a wild-type gene in MF15.3 restored susceptibility (Fig. 5). Mutations in PK were not observed in MF resistant L. infantum (Table 2), possibly suggesting additional species-specific differences. For example, the expression of the MF translocation machinery was found to vary between species and this altered their MF susceptibilities . Mutations in PK were in conserved regions between Leishmania and other species but not in active sites (Fig. S4), suggesting that the mutated PK possibly retains activity. These mutations may modulate activities necessary to mitigate the effect of MF. Reintroduction of wild-type alleles eliminates this selective advantage. Our inability to generate a PK null mutant may suggest that this gene is essential and cells lacking one allele were shown to be more susceptible showing that PK is not likely to be a target of MF and that other activities of PK or some enzymatic reactions requiring PLP are essential to deal with MF stresses. MF was shown to induce cellular events similar to those resulting from alterations in the metabolism of PLP. Indeed, MF was shown to induce altered content of phosphatidylcholine (PC) and phosphatidylethanolamine (PE) in MF-sensitive but not MF-resistant L. donovani parasites , . Interestingly, vitamin B6 deficiency in rat liver cells was shown to lead to a drastic reduction in the methylation of PE to PC . MF was also shown to induce the accumulation of reactive oxygen species (ROS) in a dose-dependent manner in Leishmania and this could be reverted by reduced thiols . The major reduced thiol in Leishmania is the glutathione-spermidine conjugate trypanothione. The rate limiting step of spermidine biosynthesis is catalyzed by the PLP-dependent enzyme ornithine decarboxylase . Moreover, vitamin B6 has recently been associated with resistance to oxidative stresses and to be itself an efficient scavenger of ROS . Mutations in PK may thus help in sustaining MF pressure by maintaining the PLP levels, the homeostasis of phospholipids, or the redox state of the parasite. Obviously, further experiments will be required to confirm these hypotheses.
Other mutations were identified in the mutants MF80.3 and MF80.5 but these were specific to a mutant with the exception of the α-adaptin protein where mutations were also observed in two L. infantum mutants (Table 2). However, these mutations occurred only in the latest step of selection and transfection of the α-adaptin gene did not change MF susceptibility (Fig. 3). While most of the other mutated genes code for hypothetical proteins and are found in single mutants they may be worth further investigating in additional studies.
This study highlighted a number of concepts of general interest in the genesis of drug resistance with also possible practical implications. Resistance appears to be highly heterogeneous at the level of the population with individual clones within the population differing both in terms of genotype (Fig. 2) but also phenotype (Fig. 4). This may be similar to the recent description of bacterial charity where in a resistant bacterial population there are clones with variable susceptibility, various clones helping others depending on the selective pressure . This heterogeneity both in genotype and phenotype in resistant populations may explain the difficulties when studying field isolates. Analysis of drug resistance in several clones may be more revealing than population analysis. Indeed, the sequencing of the MF40.3 (or MF80.3) population failed to highlight mutations in MT but the sequencing of several clones showed mutations associated with resistance (Fig. 2). This study also showed the sequential order of appearance of mutations, with mutations during the selection of resistance in PK arising always before mutations in MT and heterozygous mutation being more frequent in the lower resistant cells and homozygous mutations prevailing in more resistant isolates.
Whole genome sequencing is becoming a useful technique to study resistance in bacteria – and presently applied to the larger parasite genomes , . This study pinpointed known and new resistance markers and also allowed the demonstration of the polyclonal nature of a resistant population with individual cells with varying susceptibility and genotypes. In further studies we hope to study more deeply the clonal variety within a resistant population and how these contribute to either resistance or fitness, as recently highlighted for a population of resistant bacteria .
Copy number sequencing analysis for all Leishmania chromosomes for L. major Friedlin wild-type parasites (A) and the mutants MF80.3 (B) and MF80.5 (C). The 36 chromosomes have been mapped as circles representing the normalized read counts for each chromosome (blue line). For the MF80.3 and MF80.5 mutants, the log2 ratios of normalized read counts compared to the L. major Friedlin wild-type reference are shown as black lines. Log2 ratios of at least 1 are shown in green and log2 ratios of at least −1 are shown in red. Methods - Sequence reads were aligned with BWA (Burrow-Wheeler Alignment)  and alignments were outputted in the SAM (Sequence Alignment/Map) format . Read alignment positions were grouped in 4 kb windows  and the number of reads in each window was normalized by multiplying it with the ratio of the total number of reads for the L. major Friedlin wild-type compared to the total number of reads obtained for the corresponding resistant mutant. This allowed us to remove the bias associated to the different number of reads per sequencing run. For each sample, a ratio was computed for each window in respect to the corresponding window in the reference sample (to remove some biases such as GC bias). The logarithm in base 2 was then computed for each ratio to reduce the span of observed values and to obtain a continuous spectrum . Data was plotted in a circular fashion using Circos, a versatile information-aesthetic framework for data visualisation . Log-2 ratios of at least 1 are shown in green whereas of at least −1 are shown in red.
Protein sequence alignment of L. major MT (LmjF13.1530) with its respective orthologs of L. infantum (LinJ13_V3.1590) and L. braziliensis (LbrM13_V2.1380). Alignment was performed using the ClustalW algorithm implemented in the Lasergene software (DNASTAR, Inc.). Identical residues are shaded in black. The amino acids mutated in the MT of L. major MF mutants are marked by the lower arrowheads.
Protein sequence alignment of L. majorα-adaptin like protein (LmjF07.0050) with its respective orthologs of L. infantum (LinJ07_V3.0060) and L. braziliensis (LbrM7_V2.0050). Alignment was performed using the ClustalW algorithm implemented in the Lasergene software (DNASTAR, Inc.). Identical residues are shaded in black. The amino acids mutated in the α-adaptin like protein of L. major and L. infantum MF mutants are marked by the lower arrowheads.
Protein sequence alignment of L. major PK (LmjF31.1250) with its respective orthologs of L. infantum (LinJ13_V3.1590), L. braziliensis (LbrM13_V2.1380), T. brucei (Tb06.5F5.240), E. coli (NP_416153) and Homo sapiens (NP_003672). Alignment was performed using the ClustalW algorithm implemented in the Lasergene software (DNASTAR, Inc.). Identical residues are shaded in black. The amino acids mutated in the PK of L. major MF mutants are marked by the lower arrowheads.
Primers used in this study for amplifying the genes MT (LmjF13.1530), PK (LmjF30.1250) and the α-adaptin like protein (LmjF07.0050) of L. major. Restriction sites are underlined. Primers are also listed to sequence the genes MT, the α-adaptin like protein and Ros3 (LmjF32.0510) of L. major and to generate PK-KO constructs.
Homozygous SNPs found after resequencing the genome of L. major Friedlin. All the SNPs were identified in L. major Friedlin wild-type and in the mutants MF80.3 and MF80.5. The symbol * corresponds to a potential stop codon.
Heterozygous allelic polymorphisms in L. major Friedlin. All the same SNPs were identified in L. major Friedlin wild-type and in the mutants MF80.3 and MF80.5.
Conceived and designed the experiments: ACC JC MO. Performed the experiments: ACC SB AM. Analyzed the data: ACC SB AM PL JC MO. Contributed reagents/materials/analysis tools: SB JC. Wrote the paper: ACC PL MO.
- 1. Herwaldt BL (1999) Leishmaniasis. Lancet 354: 1191–1199.
- 2. Croft SL, Sundar S, Fairlamb AH (2006) Drug resistance in leishmaniasis. Clin Microbiol Rev 19: 111–126.
- 3. Jha TK, Sundar S, Thakur CP, Bachmann P, Karbwang J, et al. (1999) Miltefosine, an oral agent, for the treatment of Indian visceral leishmaniasis. N Engl J Med 341: 1795–1800.
- 4. Soto J, Toledo J, Gutierrez P, Nicholls RS, Padilla J, et al. (2001) Treatment of American cutaneous leishmaniasis with miltefosine, an oral agent. Clin Infect Dis 33: E57–61.
- 5. Sundar S, Jha TK, Thakur CP, Bhattacharya SK, Rai M (2006) Oral miltefosine for the treatment of Indian visceral leishmaniasis. Trans R Soc Trop Med Hyg 100: Suppl 1S26–33.
- 6. Lux H, Heise N, Klenner T, Hart D, Opperdoes FR (2000) Ether–lipid (alkyl-phospholipid) metabolism and the mechanism of action of ether–lipid analogues in Leishmania. Mol Biochem Parasitol 111: 1–14.
- 7. Paris C, Loiseau PM, Bories C, Breard J (2004) Miltefosine induces apoptosis-like death in Leishmania donovani promastigotes. Antimicrob Agents Chemother 48: 852–859.
- 8. Tewari R, Sharma V, Koul N, Sen E (2008) Involvement of miltefosine-mediated ERK activation in glioma cell apoptosis through Fas regulation. J Neurochem 107: 616–627.
- 9. Perez-Victoria FJ, Gamarro F, Ouellette M, Castanys S (2003) Functional cloning of the miltefosine transporter. A novel P-type phospholipid translocase from Leishmania involved in drug resistance. J Biol Chem 278: 49965–49971.
- 10. Grant AM, Hanson PK, Malone L, Nichols JW (2001) NBD-labeled phosphatidylcholine and phosphatidylethanolamine are internalized by transbilayer transport across the yeast plasma membrane. Traffic 2: 37–50.
- 11. Hanson PK, Malone L, Birchmore JL, Nichols JW (2003) Lem3p is essential for the uptake and potency of alkylphosphocholine drugs, edelfosine and miltefosine. J Biol Chem 278: 36041–36050.
- 12. Perez-Victoria FJ, Sanchez-Canete MP, Castanys S, Gamarro F (2006) Phospholipid translocation and miltefosine potency require both L. donovani miltefosine transporter and the new protein LdRos3 in Leishmania parasites. J Biol Chem 281: 23766–23775.
- 13. Escobar P, Matu S, Marques C, Croft SL (2002) Sensitivities of Leishmania species to hexadecylphosphocholine (miltefosine), ET-18-OCH(3) (edelfosine) and amphotericin B. Acta Trop 81: 151–157.
- 14. Yardley V, Croft SL, De Doncker S, Dujardin JC, Koirala S, et al. (2005) The sensitivity of clinical isolates of Leishmania from Peru and Nepal to miltefosine. Am J Trop Med Hyg 73: 272–275.
- 15. Sanchez-Canete MP, Carvalho L, Perez-Victoria FJ, Gamarro F, Castanys S (2009) Low plasma membrane expression of the miltefosine transport complex renders Leishmania braziliensis refractory to the drug. Antimicrob Agents Chemother 53: 1305–1313.
- 16. Mollinedo F, Fernandez-Luna JL, Gajate C, Martin-Martin B, Benito A, et al. (1997) Selective induction of apoptosis in cancer cells by the ether lipid ET-18-OCH3 (Edelfosine): molecular structure requirements, cellular uptake, and protection by Bcl-2 and Bcl-X(L). Cancer Res 57: 1320–1328.
- 17. Zoeller RA, Layne MD, Modest EJ (1995) Animal cell mutants unable to take up biologically active glycerophospholipids. J Lipid Res 36: 1866–1875.
- 18. Perez-Victoria FJ, Castanys S, Gamarro F (2003) Leishmania donovani Resistance to Miltefosine Involves a Defective Inward Translocation of the Drug. Antimicrob Agents Chemother 47: 2397–2403.
- 19. Rakotomanga M, Saint-Pierre-Chazalet M, Loiseau PM (2005) Alteration of fatty acid and sterol metabolism in miltefosine-resistant Leishmania donovani promastigotes and consequences for drug-membrane interactions. Antimicrob Agents Chemother 49: 2677–2686.
- 20. Seifert K, Perez-Victoria FJ, Stettler M, Sanchez-Canete MP, Castanys S, et al. (2007) Inactivation of the miltefosine transporter, LdMT, causes miltefosine resistance that is conferred to the amastigote stage of Leishmania donovani and persists in vivo. Int J Antimicrob Agents 30: 229–235.
- 21. Castanys-Munoz E, Alder-Baerens N, Pomorski T, Gamarro F, Castanys S (2007) A novel ATP-binding cassette transporter from Leishmania is involved in transport of phosphatidylcholine analogues and resistance to alkyl-phospholipids. Mol Microbiol 64: 1141–1153.
- 22. BoseDasgupta S, Ganguly A, Roy A, Mukherjee T, Majumder HK (2008) A novel ATP-binding cassette transporter, ABCG6 is involved in chemoresistance of Leishmania. Mol Biochem Parasitol 158: 176–188.
- 23. Castanys-Munoz E, Perez-Victoria JM, Gamarro F, Castanys S (2008) Characterization of an ABCG-like transporter from the protozoan parasite Leishmania with a role in drug resistance and transbilayer lipid movement. Antimicrob Agents Chemother 52: 3573–3579.
- 24. Perez-Victoria JM, Perez-Victoria FJ, Parodi-Talice A, Jimenez IA, Ravelo AG, et al. (2001) Alkyl-lysophospholipid resistance in multidrug-resistant Leishmania tropica and chemosensitization by a novel P-glycoprotein-like transporter modulator. Antimicrob Agents Chemother 45: 2468–2474.
- 25. Seifert K, Matu S, Javier Perez-Victoria F, Castanys S, Gamarro F, et al. (2003) Characterisation of Leishmania donovani promastigotes resistant to hexadecylphosphocholine (miltefosine). Int J Antimicrob Agents 22: 380–387.
- 26. Choudhury K, Zander D, Kube M, Reinhardt R, Clos J (2008) Identification of a Leishmania infantum gene mediating resistance to miltefosine and SbIII. Int J Parasitol 38: 1411–1423.
- 27. Moreira W, Leprohon P, Ouellette M (2011) Tolerance to drug-induced cell death favours the acquisition of multidrug resistance in Leishmania. Cell Death Dis 2: e201.
- 28. Ouellette M, Fase-Fowler F, Borst P (1990) The amplified H circle of methotrexate-resistant Leishmania tarentolae contains a novel P-glycoprotein gene. Embo J 9: 1027–1033.
- 29. Papadopoulou B, Roy G, Ouellette M (1992) A novel antifolate resistance gene on the amplified H circle of Leishmania. Embo J 11: 3601–3608.
- 30. Sambrook J, Fritsch EF, Maniatis T (1989) Molecular Cloning: a Laboratory Manual. New York: Cold Spring Harbor: Cold Spring Harbor Laboratory Press.
- 31. El Fadili A, Kundig C, Ouellette M (2002) Characterization of the folylpolyglutamate synthetase gene and polyglutamylation of folates in the protozoan parasite Leishmania. Mol Biochem Parasitol 124: 63–71.
- 32. Horton RM, Hunt HD, Ho SN, Pullen JK, Pease LR (1989) Engineering hybrid genes without the use of restriction enzymes: gene splicing by overlap extension. Gene 77: 61–68.
- 33. Moreira W, Leblanc E, Ouellette M (2009) The role of reduced pterins in resistance to reactive oxygen and nitrogen intermediates in the protozoan parasite Leishmania. Free Radic Biol Med 46: 367–375.
- 34. Ivens AC, Peacock CS, Worthey EA, Murphy L, Aggarwal G, et al. (2005) The genome of the kinetoplastid parasite, Leishmania major. Science 309: 436–442.
- 35. Aslett M, Aurrecoechea C, Berriman M, Brestelli J, Brunk BP, et al. (2010) TriTrypDB: a functional genomic resource for the Trypanosomatidae. Nucleic Acids Res 38: D457–462.
- 36. Li H, Durbin R (2009) Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25: 1754–1760.
- 37. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, et al. (2009) The Sequence Alignment/Map format and SAMtools. Bioinformatics 25: 2078–2079.
- 38. Lassmann T, Hayashizaki Y, Daub CO (2011) SAMStat: monitoring biases in next generation sequencing data. Bioinformatics 27: 130–131.
- 39. Koressaar T, Remm M (2007) Enhancements and modifications of primer design program Primer3. Bioinformatics 23: 1289–1291.
- 40. Beverley SM (1991) Gene amplification in Leishmania. Annu Rev Microbiol 45: 417–444.
- 41. Borst P, Ouellette M (1995) New mechanisms of drug resistance in parasitic protozoa. Annu Rev Microbiol 49: 427–460.
- 42. Cruz AK, Titus R, Beverley SM (1993) Plasticity in chromosome number and testing of essential genes in Leishmania by targeting. Proc Natl Acad Sci U S A 90: 1599–1603.
- 43. Dumas C, Ouellette M, Tovar J, Cunningham ML, Fairlamb AH, et al. (1997) Disruption of the trypanothione reductase gene of Leishmania decreases its ability to survive oxidative stress in macrophages. Embo J 16: 2590–2598.
- 44. Mottram JC, McCready BP, Brown KG, Grant KM (1996) Gene disruptions indicate an essential function for the LmmCRK1 cdc2-related kinase of Leishmania mexicana. Mol Microbiol 22: 573–583.
- 45. Mukherjee A, Roy G, Guimond C, Ouellette M (2009) The gamma-glutamylcysteine synthetase gene of Leishmania is essential and involved in response to oxidants. Mol Microbiol 74: 914–927.
- 46. Neary JT, Diven WF (1970) Purification, properties, and a possible mechanism for pyridoxal kinase from bovine brain. J Biol Chem 245: 5585–5593.
- 47. Newman JA, Das SK, Sedelnikova SE, Rice DW (2006) Cloning, purification and preliminary crystallographic analysis of a putative pyridoxal kinase from Bacillus subtilis. Acta Crystallogr Sect F Struct Biol Cryst Commun 62: 1006–1009.
- 48. Scott TC, Phillips MA (1997) Characterization of Trypanosoma brucei pyridoxal kinase: purification, gene isolation and expression in Escherichia coli. Mol Biochem Parasitol 88: 1–11.
- 49. Rakotomanga M, Blanc S, Gaudin K, Chaminade P, Loiseau PM (2007) Miltefosine affects lipid metabolism in Leishmania donovani promastigotes. Antimicrob Agents Chemother 51: 1425–1430.
- 50. She QB, Hayakawa T, Tsuge H (1995) Alteration in the phosphatidylcholine biosynthesis of rat liver microsomes caused by vitamin B6 deficiency. Biosci Biotechnol Biochem 59: 163–167.
- 51. Sanchez CP, Sidrauski C, Freire SM, Gonzalez NS, Algranati ID (1995) Ornithine decarboxylase from Leishmania mexicana promastigotes: interaction with pyridoxal 5′-phosphate and alpha-difluoromethylornithine. Biochem Biophys Res Commun 212: 396–403.
- 52. Thotala D, Chetyrkin S, Hudson B, Hallahan D, Voziyan P, et al. (2009) Pyridoxamine protects intestinal epithelium from ionizing radiation-induced apoptosis. Free Radic Biol Med 47: 779–785.
- 53. Lee HH, Molla MN, Cantor CR, Collins JJ (2010) Bacterial charity work leads to population-wide resistance. Nature 467: 82–85.
- 54. Cui L, Isii T, Fukuda M, Ochiai T, Neoh HM, et al. (2010) An RpoB mutation confers dual heteroresistance to daptomycin and vancomycin in Staphylococcus aureus. Antimicrob Agents Chemother 54: 5222–5233.
- 55. Feng J, Lupien A, Gingras H, Wasserscheid J, Dewar K, et al. (2009) Genome sequencing of linezolid-resistant Streptococcus pneumoniae mutants reveals novel mechanisms of resistance. Genome Res 19: 1214–1223.
- 56. Howden BP, Stinear TP, Allen DL, Johnson PD, Ward PB, et al. (2008) Genomic analysis reveals a point mutation in the two-component sensor gene graS that leads to intermediate vancomycin resistance in clinical Staphylococcus aureus. Antimicrob Agents Chemother 52: 3755–3762.
- 57. Lescat M, Calteau A, Hoede C, Barbe V, Touchon M, et al. (2009) A module located at a chromosomal integration hot spot is responsible for the multidrug resistance of a reference strain from Escherichia coli clonal group A. Antimicrob Agents Chemother 53: 2283–2288.
- 58. Mwangi MM, Wu SW, Zhou Y, Sieradzki K, de Lencastre H, et al. (2007) Tracking the in vivo evolution of multidrug resistance in Staphylococcus aureus by whole-genome sequencing. Proc Natl Acad Sci U S A 104: 9451–9456.
- 59. Hunt P, Martinelli A, Modrzynska K, Borges S, Creasey A, et al. (2010) Experimental evolution, genetic analysis and genome re-sequencing reveal the mutation conferring artemisinin resistance in an isogenic lineage of malaria parasites. BMC Genomics 11: 499.
- 60. Martinelli A, Henriques G, Cravo P, Hunt P (2011) Whole genome re-sequencing identifies a mutation in an ABC transporter (mdr2) in a Plasmodium chabaudi clone with altered susceptibility to antifolate drugs. Int J Parasitol 41: 165–171.
- 61. Chiang DY, Getz G, Jaffe DB, O'Kelly MJ, Zhao X, et al. (2009) High-resolution mapping of copy-number alterations with massively parallel sequencing. Nat Methods 6: 99–103.
- 62. Quackenbush J (2002) Microarray data normalization and transformation. Nat Genet 32: Suppl496–501.
- 63. Krzywinski M, Schein J, Birol I, Connors J, Gascoyne R, et al. (2009) Circos: an information aesthetic for comparative genomics. Genome Res 19: 1639–1645.