Genomewide Analysis of Mode of Action of the S-Adenosylmethionine Analogue Sinefungin in Leishmania infantum

The two main cellular metabolic one-carbon donors are reduced folates and S-adenosylmethionine, whose biosynthetic pathways have proven highly effective in chemotherapeutic interventions in various cell types. Sinefungin, a nucleoside analogue of S-adenosylmethionine, was shown to have potent activity against the protozoan parasite Leishmania. Here, we studied resistance to sinefungin using whole-genome approaches as a way to further our understanding of the role of S-adenosylmethionine in this parasite and to reveal novel potential drug targets. These approaches allowed the characterization of novel features related to S-adenosylmethionine function in Leishmania which could further help in the development of sinefungin-like compounds against this pathogenic parasite.

). One major CNV was detected in all four resistant mutants; this CNV corresponded to a 5-kb deletion in chromosome 10 ( Fig. 2A). LiSNFR50.2 differs, however, from the other three mutants. Indeed, while the number of sequence reads is also smaller in this mutant, it is not nil. This could be explained by the heterogeneity of the population with different copy numbers of this region of chromosome 10 that is emerging upon drug selection and passages. The deleted region comprises genes coding for FBT proteins, including the AdoMet transporter (LinJ. 10.0370) ( Fig. 2A). The deletion of AdoMetT1 was confirmed by Southern blotting using an AdoMetT1-specific probe ( Fig. S2A and S2B). Genomic DNAs of Leishmania parasites were digested with NheI, and a 2.1-kb band hybridized with the AdoMetT1 probe in WT cells (Fig. S2B, lane 1) but not in LiSNFR50.1, LiSNFR50.3, or in a resistant cell grown without SNF for several passages (Fig. S2B, lanes 2 to 4). The AdoMetT1 gene was cloned and transfected in LinSNFR50.1 and LiSNFR50.3 which resulted in a complete reversion of the resistance phenotype (Fig. 1C). Several folate-biopterin-transporter (FBT) genes, including AdoM-etT1, are located in tandem within a 37-kb region of chromosome 10. These genes possess high level of sequence identities (18,23). The deletion is characterized by the absence of sequencing reads between nucleotides ϳ151 kb and ϳ156 kb on chromosome 10 and is flanked by the gene LinJ. 10.0380 on its right side ( Fig. 2A). We found that recombination occurred within an identical 720-bp region shared between LinJ. 10.0370 (AdoMetT1) and LinJ. 10.0380 ( Fig. 2A). Primers were designed to map this rearrangement, and PCR indeed confirmed that the recombination site occurs within this 720-bp region ( Fig. S2C and S2D). Sequencing of the rearranged gene further supported the proposed rearrangement (Fig. S3). The AdoMetT1 rearrangement dynamics was further studied using SNF exposure. Using PCR, we monitored the recombination event between LinJ. 10.0370 (AdoMetT1) and LinJ. 10.0380 and the concomitant disappearance of the AdoMetT1-specific amplification. The recombined product started appearing at 2 M (32ϫ EC 50 ) and dominated at higher levels of SNF, with the parallel disappearance of the AdoMetT1 PCR band (Fig. 2B).
The sequence reads were further analyzed for the presence of single-nucleotide variants (SNVs) as previously described (24). Few coding SNVs were detected, and those passing our filtering criteria (see Materials and Methods) were mostly for genes coding for surface proteins, kinesin, or duplicated hypothetical proteins (Table S1), a phenomenon frequently observed in genomics studies on Leishmania (19). These were not analyzed further.
Leishmania parasites can both transport and synthesize AdoMet. Since the lack of AdoMetT1 might affect several important methylation events in the cell, we examined the expression of four SAM-metabolizing enzymes in our SNF-resistant parasites. The genes coding for AdoMet synthetase (MetK) (LinJ.30.3560) and for cobalamindependent methionine synthase (CoMS) (LinJ.07.0240) were upregulated in LiSNFR50.1 by 2.5-fold and 2-fold, respectively, but expression of the mitochondrial methionine synthase reductase (MMSR) (LinJ.36.4950) and AdoMet hydrolase (SAH) (LinJ.36.4100) were not significantly altered (Fig. 3A). The episomal expression of AdoMetT1 did not revert the upregulation of MetK. Quantitative reverse transcription-PCR (RT-PCR) revealed alteration of AdoMetT1 mRNA abundance between the growth phases, attaining maximum at stationary phase (Fig. 3B), and interestingly, in L. infantum promastigotes, SNF responsiveness was determined to be dependent on the growth phase of the inoculum as reflected by the two-to threefold variation in EC 50 values between early log and stationary phase (Fig. 3C). Expression of AdoMetT1 as measured by qRT-PCR during early (white bar), late logarithmic (gray bar), and stationary (black bar) phases of growth. Expression was quantified relative to GAPDH. (C) Response of Leishmania promastigotes to SNF pressure at early log (), late log (), and stationary (OE) phases of growth. All data are means plus SEM for three independent replicates. Statistical analyses were performed using paired (A and B) and unpaired (C) two-tailed t tests. Statistical significance is indicated as follows: **, P Ͻ 0.01; *, P Ͻ 0.05; ns, not significant.
Cos-seq reveals targets for sinefungin along with the diversity of AdoMet function in Leishmania. L. infantum parasites transfected with a cosmid genomic library were selected with SNF at increasing concentrations in biological duplicates. The parasites adapted to each passage and reached stationary phase within 5 days of selection for all doses. Since the parasites showed rapid adaptation to drug selection, cosmids enriched at 4ϫ, 16ϫ, 32ϫ, and 64ϫ EC 50 selections were sequenced and analyzed. Using a cutoff 16-fold enrichment compared to a similarly cultured but untreated control population, cosmid enrichments of genomic loci derived from chromosomes 15, 28, 29, 30, 31, and 36 were identified, with the highest enrichment being for a locus on chromosome 30 ( Fig. 4A to C and Fig. S4). Visualization of cosmid abundance revealed gradual enrichment for each of these, with maximal enrichment occurring at drug selection equivalent to 64ϫ EC 50 (Fig. 4B). Each enriched cosmid is coding between 8 and 14 genes (Fig. 4C).
At least three gene products isolated from the Cos-seq screen, namely, MetK, CMT1, and LCMT, contain AdoMet binding sites as determined by molecular docking modeling of SNF and AdoMet with these targets. Docking of MetK revealed a binding affinity for SNF of Ϫ37.16 kcal/mol. It formed 11 hydrogen bonds (H-bonds) and hydrophobic interactions (Fig. 5). The binding site is similar to the docking site of AdoMet with at least two conserved H-bonds (Q184 and Y326) (Fig. S5). CMT1 showed similar binding affinity for SNF (Ϫ38.05 kcal/mol) with at least 13 potential H-bond and hydrophobic interactions. A binding affinity of Ϫ35.01 kcal/mol was showed by LCMT with SNF binding with at least eight H-bonds and hydrophobic interactions (Fig. 5). Binding of AdoMet by CMT1 and LCMT shares several conserved interactions as observed for SNF (Fig. S5).
Leucine carboxyl methyltransferase and sinefungin. The role of leucine carboxyl methyltransferase (LCMT) is to methylate the carboxyl group of C-terminal leucine residues of proteins. More than 10% of total proteins in L. infantum contain leucine at their C termini (Fig. S6A). Multiple alignment with bacterial and higher eukaryotic orthologs of LCMT demonstrated that LCMT from L. infantum and Trypanosoma brucei shares significant amount of conservation (Fig. S6B), although the kinetoplastid LCMTs cluster distantly from eukaryotic LCMTs (Fig. S6C). The LCMT gene was inactivated by integrating neomycin and puromycin resistance cassettes using a CRISPR-Cas9-based approach (21) in L. infantum (Fig. 6A). The knockout was confirmed by Southern blotting (Fig. 6B) and by PCR using open reading frame (ORF)-specific and untranslated Although LCMT Ϫ/Ϫ cells grew well in SDM medium, the growth in log phase was found to be impaired compared to WT (Cas9) cells (Fig. 6C). The phenotype was reverted by episomal expression of LCMT (Fig. 6C). LCMT Ϫ/Ϫ cells were unexpectedly four-to fivefold more resistant to SNF (Fig. 6D). The resistance was abolished when LCMT was expressed episomally in LCMT Ϫ/Ϫ cells (Fig. 6D). This is in contrast to the Cos-seq screen, where LCMT episomal overexpression led to resistance. This apparent dichotomy may be explained by the role of LCMT during growth phase. Indeed, while LCMT Ϫ/Ϫ cells are resistant to SNF both in log and stationary phase, cells expressing LCMT as part of an episome displayed maximum resistance in early log phase, and the resistance is gradually lost in late-log and stationary-phase promastigotes (Fig. S7B).
In eukaryotes, LCMT is known to methylate and regulate PP2A catalytic subunit (PP2AC) and related protein phosphatases (25). Intriguingly, cosmids derived from chromosomes 15 and 29 isolated from the Cos-seq screen carry genes that encode regulators of phosphatase. In order to determine whether PP2AC and LCMT do interact, a C-terminal hemagglutinin (HA)-tagged version of LCMT was cotransfected with a N-terminal Ty1-tagged PP2AC in LCMT Ϫ/Ϫ parasites. Immunoprecipitation with anti-HA antibody followed by Western blotting with anti-Ty1 antibody confirmed that LCMT interacts with PP2AC (Fig. 7A). To further investigate the LCMT-PP2AC interactions, we carried out high-ambiguity-driven protein-protein docking (26) with quality models of the Leishmania LCMT and PP2AC. The model for the Leishmania predicted complex structure exhibited similarity in orientation and organization with crystal structure of the human LCMT1 and PP2AC complex with the C-terminal leucine of PP2AC buried inside LCMT for both structures (Fig. S8). Analysis of the interaction interface and stabilization energy revealed that both the proteins have considerable continuous patches of interaction with stabilization energy of Ϫ661.6545 kJ/mol, indicating strong  interaction between the two proteins (Fig. S8). To examine whether methylation of the C-terminal leucine of PP2AC (L308) is indeed linked to the SNF response, we independently transfected WT L. infantum parasites with episomes coding for WT PP2AC or for a PP2AC L308G variant in which the C-terminal leucine was mutated to a glycine residue. The PP2AC L308G -expressing parasites elicited 1.75-fold-higher resistance compared to cells expressing the WT version of PP2AC (Fig. 7B).

DISCUSSION
SNF is a nucleoside antibiotic structurally related to AdoMet that competitively inhibits AdoMet-synthesizing and -dependent enzymes (27). The antileishmanial activity of SNF is well established (15), but the drug was not further developed due to nephrotoxicity and toxicity to bone marrow cells (28,29). Nonetheless, work is still ongoing in developing SNF analogues with higher therapeutic indexes (29)(30)(31)(32). This is certainly justified in light of the recent discovery that lead antitrypanosomal boroncontaining compounds perturb AdoMet metabolism and seem to act similarly to SNF (33).
Whole-genome sequencing of independent resistant mutants could detect a deletion of AdoMetT1, the AdoMet transporter. This is reasonable, as SNF was previously shown to use the AdoMet transporter in Leishmania (16, 17). This paralleled observations made in SNF-resistant yeast or Toxoplasma gondii where mutations in their . Puromycin (PURO) and neomycin (NEO) repair cassettes were generated by PCR using primers with overhangs corresponding to the first 30 nucleotides (nt) of the LCMT 5= and 3= untranslated regions (UTRs) for integration by homologous recombination at the cut site. The repair templates were transfected in LiCas9 along with a gRNA-crRNA hybrid before the selection of transfectants with puromycin or G418. The puromycin-or G418-selected populations were then plated, and five individual clones were isolated for each transfection. (B) Confirmation of LCMT knockout was obtained by Southern blotting analysis where genomic DNAs from WT (LiCas9) (lanes 1) and LCMT Ϫ/Ϫ cells (lanes 2) were digested with Afe1 and hybridized to LCMT-specific (left) or PTR1-specific (right) DNA probes. (C) The growth of WT L. infantum, LCMT Ϫ/Ϫ , and LCMT Ϫ/Ϫ add-back was monitored in SDM medium for 7 days by OD 600 measurements. Data are means Ϯ SEM from at least three independent experiments. (D) L. infantum LCMT Ϫ/Ϫ parasites are resistant to SNF as determined by EC 50 measurements for WT (LiCas9) and LCMT Ϫ/Ϫ or LCMT Ϫ/Ϫ add-back cells. Data are means plus SEM for at least three biological replicates. Statistical analyses were performed using unpaired two-tailed t tests. ***, P Ͻ 0.001. AdoMet transporters were found to be the main driver of resistance (34,35). However, in those cases, the AdoMet transporters are part of the amino acid permease superfamily, and resistance was mediated by point mutations rather than gene deletion as in our L. infantum SNF-resistant mutants. Our next-generation sequencing (NGS) work allowed us to precisely define the molecular mechanism of gene deletion that was mediated by homologous recombination between conserved regions of FBT genes, a frequent mechanism of gene rearrangement in Leishmania (36). The deletion took place in the same region between AdoMetT1 and LinJ. 10.0380 in all four mutants, despite the presence of other homologous repeats within the FBT paralogues of chromosome 10. One possible reason could be the fitness cost that the loss of folate transport would have brought from the deletion of the nearby FT1 or FT5 (23,37). The AdoMetT1 gene is preferentially expressed in stationary phase that is correlated with SNF susceptibility (Fig. 3). Possibly Leishmania uses primarily the AdoMet biosynthetic route during logarithmic phase, but because of metabolic reprogramming during stationary phase, the parasite may rely to a greater extent on transport of AdoMet to meet its AdoMet requirements. No phenotypic SNVs could be associated with SNF resistance in our NGS effort.
Analysis of the resistant mutants led to an understanding of the main strategy to resist SNF, but we had to rely on a functional genomic screen to isolate genes that could help in our understanding of the physiological role of AdoMet in Leishmania. A Cos-seq screen (20) highlighted the enrichment of at least six cosmids by SNF selection and identified MetK as a target for SNF in Leishmania. This gene produced six-to sevenfold-higher resistance to SNF. This is lower than the last drug concentration (64ϫ) used during the Cos-seq selection, but our experience indicates that we can seldom reach the level of resistance used for selection while transfecting individual genes. It is salient to point out, however, that even if cells are selected at 64ϫ EC 50 , they may not be resistant to that level of drug. Indeed, during continuous drug selection, a population may arise where there is transient physiological adaptation or tolerance that may facilitate growth in the presence of the drug. Of the other cosmids/genes that were revealed, one was coding for the mRNA cap-(guanine-N7)-methyltransferase CMT1. SNF is known to inhibit several viral N7 cap methyltransferases (38) and fungal enzymes (39), The level of expression of 2ϫ-Ty1-PP2AC (second panel) (marked with an arrowhead) and LCMT-HA (third panel) was tested by immunoblotting using anti-Ty1 and anti-HA antibodies, respectively. Our anti-Ty1 antibody reacts with an unknown 30-kDa protein, which is marked with an asterisk. ␣-Tubulin (␣-Tub) was detected as loading control (bottom panel). (B) Impact of PP2AC on SNF responsiveness was studied by expressing PP2AC WT and PP2AC L308G in L. infantum WT cells. EC 50 values were determined by dose-response curves against SNF. Data are means plus SEM for at least three biological replicates. Statistical analyses were performed using unpaired two-tailed t tests. ***, P Ͻ 0.001; ns, not significant. and it was suggested to be the target of SNF antifungal activity (40). Cotransfection of the yeast MetK (SAM1) and N7 cap methyltransferase (ABD1) was shown to produce resistance to SNF in yeast (34). Work on CMT1 has been carried out in the related parasite Trypanosoma brucei (41), and this gene appears to be nonessential (42). Here, we suggest, in line with viral work, that along with MetK the Leishmania CMT1 may be a secondary target for SNF. Modeling studies (Fig. 5) support this suggestion.
Our Cos-seq screen also led to the characterization of the leucine carboxyl methyltransferase LCMT. Molecular docking studies identified crucial residues for the interaction between AdoMet or SNF with LCMT as well ( Fig. 5; see also Fig. S5 in the supplemental material). In mammals, LCMT-1 methylates the C-terminal leucine of the C-subunits of protein phosphatases of the PP2A subfamily. Its methylation facilitates the formation of PP2A heterodimers that are involved in a plethora of physiological processes related to cell growth and proliferation (43). PP2AC-LCMT interactions were verified in Leishmania by immunoprecipitation of the two coexpressing tagged versions of the proteins. In contrast to Leishmania, LCMT is essential in mice (44). Overexpression of LCMT produces SNF resistance (Table 1), but its inactivation produced even more resistance (Fig. 6C). We found that overexpression of LCMT produces significant resistance only in early log phase of growth, while the LCMT Ϫ/Ϫ cells are resistant to SNF at every growth phase of the parasite. Thus, one possible explanation is that the Leishmania protein has several targets and their state of methylation (possibly linked to growth phase) is implicated in SNF resistance. For example, replacement of the terminal leucine in PP2AC contributed to SNF resistance (Fig. 7B). Interestingly, about 10% of the proteins in L. infantum possess leucine residues at their C termini. The Leishmania LCMT is 28% identical to the mammalian enzyme and phylogenetically distinct among LCMT orthologues, and it remains to be seen whether it also has PP2A as a substrate. Protein-protein docking would suggest that this is quite possible (Fig. S8). It is intriguing that the Cos-seq screen led to a PP2A regulatory subunit (LinJ.15.0980) which was found to elicit (low) resistance against SNF ( Table 1). The LCMT Ϫ/Ϫ mutant is resistant to SNF (Fig. 6C), and this is consistent with our observation that WT cells with episomal expression of a PP2A L308G version were also slightly resistant to SNF. The protein LinJ.29.0180, also involved in the response to SNF, has a Pfam motif for protein phosphatase inhibitor 2, and possibly the modulation in the activity of a number of phosphatases influences AdoMet metabolism and thus, the SNF response.
Two genes encoding tryparedoxin (TXN3 and TXN4) were also shown to contribute to SNF resistance ( Table 1). Overexpression of tryparedoxin possibly helped the parasite to circumvent the redox imbalance imposed by SNF.
The use of independent genomic approaches for studying the mode of action of SNF in Leishmania allowed the characterization of novel features related to AdoMet function in Leishmania. SNF has interesting activity against Leishmania, and reducing its toxicity may bring it further along the development pipeline. Potential targets have now been found which could further help in the development of SNF-like compounds. Ideally, these analogues would maintain specific activity against multiple Leishmania targets while being more lipophilic, hence escaping the need to enter the cells through AdoMetT1, a locus frequently deleted when cells are in contact with SNF.
Whole-genome sequencing and analysis. Paired-end sequencing libraries were prepared from L. infantum genomic DNA with the Nextera DNA sample prep kit and sequenced on an Illumina HiSeq platform with 101-nucleotide paired-end reads. An average genome coverage of more than 50-fold was achieved (see Fig. S1A in the supplemental material). Sequence reads were aligned to the L. infantum JPCM5 genome using bwa-mem (45). Read duplicates were marked using Picard, and GATK was applied for discovering single-nucleotide variants (SNVs) and small insertions or deletions (indels) (46). SNVs and indels from the vcf files were filtered using the following hard filtering criteria: mappingQual (MQ) of Ն40, FisherStrand (FS) of Յ60, QualByDepth (QD) of Ն5, MappingQualityRankSumTest (MQRankSum) of ՆϪ2.5, and ReadPosRankSumTest (ReadPosRankSum) of ՆϪ4. Single-nucleotide variants (SNVs) re-vealed by next-generation sequencing (NGS) were confirmed by PCR amplification and conventional DNA sequencing. Copy number variations (CNVs) were derived from read depth coverage as described earlier (47).
Cosmid extraction, purification, and paired-end sequencing library preparation. Cosmid extraction was conducted as previously described (20). Purified total DNA was treated with RiboShredder RNase blend (Epicentre) to remove potential RNA contaminations. Genomic DNA was removed with plasmidsafe ATP-dependent DNase (Epicentre) following the manufacturer's instructions. In addition, kinetoplastid DNA was removed by electrophoresis of DNase-treated cosmid extracts on 1% low-melting-point agarose (Invitrogen) followed by excision and purification of the bands corresponding to high-molecularweight cosmid DNA (ϳ50 kb). Purified cosmid DNA was quantified with the QuantiFluor dsDNA system staining kit (Promega). Fifty nanograms of purified cosmid DNA was used for paired-end library preparation using Nextera DNA sample preparation kit (Illumina). Sequencing libraries were quantified with the QuantiFluor dsDNA system and sequenced using an Illumina HiSeq system at a final concentration of 8 pM.
Cosmid enrichment analysis. Sequencing reads from each sample were independently aligned to the L. infantum JPCM5 reference genome (version 8.0) obtained from TritrypDB (http://tritrypdb.org/ tritrypdb/) using the bwa-mem software (48). BAM files were converted to BED files by using BEDTools (49), and the read depth and genome coverage were visualized using the SignalMap software (Roche NimbleGen). The detection of enriched genes relied on the Trinity software version 2.1.1 (50), which includes all third-party tools required for the analysis. Gene abundance within samples was quantified using the kallisto software (51). Clusters of genes significantly enriched by drug selection were retrieved with edgeR (52) using the default parameters (false-discovery rate of Յ0.001). Gene clusters were then plotted according to the median-centered log 2 -transformed fragment per kilobase per million mapped reads (FPKM) values using R scripts included in the Trinity package. Only genes with a log 2 fold change of Ն4 were retained. The cosmid fold enrichment was computed by extracting the mean FPKM ratio for the genes on enriched cosmids in the drug-selected samples normalized to the mean FPKM ratio for these genes in the control sample passaged in the absence of drug.
DNA constructs, cosmid isolation, and transfection. The genes of L. infantum were amplified from genomic DNA using compatible primer pairs and cloned in the Leishmania expression vectors pSP72␣Zeo␣, pSP72␣Puro␣, or pSP72␣HYG␣ unless mentioned otherwise. A total of 20 g of plasmid DNA for episomal expression was transfected into Leishmania promastigotes by electroporation.
The enriched cosmids used for paired-end sequencing library preparation were transformed in Escherichia coli DH5␣ and were either recovered by random picking of transformed colonies or by colony hybridization as described earlier (20). Candidate cosmids were transfected in wild-type (WT) L. infantum parasites.
Knockout cell lines were generated in WT L. infantum expressing Cas9 (LiCas9) (21). Puromycin and neomycin resistance genes were amplified with primers containing 30-to 40-bp sequences upstream and downstream of the target gene. A CRISPR RNA (crRNA) targeting the open reading frame (ORF) of LCMT (LinJ.36.0090) was designed targeting the following sequence: gRNALCMT, GCGACCTGTATGACG CCAGG. The guide RNA (gRNA) was generated by hybridizing 5 l of 0.1 nmol/l crRNA with 5 l of equimolar trans-activating crRNA (tracrRNA) (IDT) as described earlier (21). Eight micrograms of each repair template and 5 l of each crRNA-tracrRNA hybrid were transfected simultaneously using Amaxa Nucleofector transfection kit (Lonza). The selection was done with puromycin at 100 g/ml or with neomycin at 400 g/ml. Allelic substitutions were confirmed by PCR amplification of target genes followed by standard sequencing.
Immunoprecipitation and Western blot analysis. Immunoprecipitation was done using Pierce HA-tag magnetic IP/co-IP kit according to the manufacturer's protocol. Lysis of pellets derived from mid-log-phase cells was performed using lysis buffer supplemented with Halt protease inhibitor cocktail and Halt phosphatase inhibitor cocktail (Thermo Scientific) with 20 to 30 strokes of a Dounce homogenizer with the cells on ice. Clear supernatants obtained after centrifugation (10,000 ϫ g; 30 min) were incubated with antihemagglutinin (anti-HA) magnetic beads at 4°C for 4 h on a gentle rotator. The beads were separated and washed, and SDS-PAGE was performed on 10% or 12% acrylamide gels by standard procedures. Protein expression Immobilon western chemiluminescence kit (Millipore, Billerica, MA, USA) was used to detect proteins. Antibodies and dilutions used are as follows: mouse anti-HA IgG (Santa Cruz), (1:5,000), mouse antitubulin IgG (Millipore) (1:5,000), horseradish peroxidase (HRP)-conjugated anti-mouse IgG (Cell Signaling) (1:10,000), rabbit anti-Ty-1 IgG (Genscript) (1:500), and HRP-conjugated goat anti-rabbit IgG (GE Life Sciences) (1:5,000).
Homology model and molecular docking. The homology model structures of METK (LinJ.30.3560), CMT1 (LinJ.36.0130), andLCMT (LinJ.36.0090) are built using the protein template structures from the Protein Data Bank (PDB) entries or accession nos. 4ODJ, 4FYU, 3IEI, and 5E8J. The model structures were built using a fully automated protein structure homology modeling server SWISS-MODEL (http:// swissmodel.expasy.org/) (53). The model quality was estimated based on the QMEAN scoring functions of 0.80, 0.55, 0.68, and 0.56 which are within the acceptable range (54). PyMOL v1.3 was used to visualize the structural models (55). In silico docking of structural models of METK, CMT1, and LCMT with SNF and AdoMet was conducted using the PATCHDOCK server (56) and FireDock, an efficient method for the refinement and rescoring of rigid-body docking solutions (57). The binding site residues are identified from the LigPlotϩ (58) for representation of hydrophobic and hydrogen-bond interactions.
Protein-protein docking. Homology models for LiLCMT and PP2A individually were built using SWISS-MODEL (59) with complex structure of human LCMT-1 and PP2AC␣ as the template with QMEAN scores of Ϫ1.84 and Ϫ1.59, which are within the allowable limit. Each of the structures was validated by