This journal is©The Royal Society of Chemistry 2015 Mol. BioSyst.

Cite this:DOI: 10.1039/c5mb00251f

L-Asparaginase as a new molecular target againstleishmaniasis: insights into the mechanism ofaction and structure-based inhibitor design†

Jasdeep Singh,‡a Ankit Srivastava,‡a Pravin Jha,b Kislay K. Sinhab andBishwajit Kundu*a

L-Asparaginases belong to a family of amidohydrolases that catalyze the conversion of L-asparagine into

L-aspartic acid and ammonia. Although bacterial L-asparaginases have been used extensively as anti-

leukemic agents, their possible role as potential drug targets for pathogenic organisms has not been

explored. The presence of genes coding for putative L-asparaginase enzymes in the Leishmania donovani

genome hinted towards the specific role of these enzymes in extending survival benefit to the organism. To

investigate whether this enzyme can serve as a potential drug target against the Leishmania pathogen, we

obtained structural models of one of the putative Leishmania L-asparaginase I (LdAI). Using an integrated

computational approach involving molecular modelling, docking and molecular dynamics simulations, we

found crucial differences between catalytic residues of LdAI as compared to bacterial L-asparaginases. The

deviation from the canonical acid–base pair at triad I, along with the structural reorganization of a b-hairpin

loop in the presence of a substrate, indicated an altogether new mechanism of action of the LdAI enzyme.

Moreover, the finding of compositional and functional differences between LdAI and human asparaginase

was used as a criterion to identify specific small molecule inhibitors. Through virtual screening of a library of

11 438 compounds, we report five compounds that showed favorable interactions with the active pocket of

LdAI, without adversely affecting human asparaginase. One of these compounds when tested on cultured

Leishmania promastigotes displayed a promising leishmanicidal effect. Overall, our work not only provides

first hand mechanistic insights of LdAI but also proposes five strongly active compounds which may prove

as effective anti-leishmaniasis molecules.

1. Introduction

Leishmaniasis is a prominent pathogenic disease in tropicalcountries caused by the protozoan parasites Leishmania spp.1–3

Although Leishmania can affect various tissues, the visceralinfection (known as kala-azar) is most lethal, as at the laterstages of infection the pathogen severely compromises the hostimmune response.4–6 During infection, this pathogen usuallybegins infiltrating the phagocytic cells, forms phagolysosomes,proliferates via complex host–parasite interactions and eventuallysurpasses host immune responses by strategically evadingdestruction by the host cells.7,8 Forward line chemotherapy of

leishmaniasis includes administration of pentavalent anti-monials, anti-fungal antibiotics and aromatic dimidines.9–13

Some of the anti-retro viral drugs and immunomodulators haverecently been introduced in conjunction with chemotherapyto potentiate anti-leishmanial effects.14–17 Besides being expensiveand highly toxic, acquisition of drug resistance by the protozoaagainst the chemotherapeutics is a major threat, necessitating theidentification of newer drugs.18–20 In the absence of effectivevaccines, inhibitors designed against pathogen-specific proteintargets may prove to be the most appropriate, next generation,anti-leishmaniasis drugs.

A good volume of literature supports the employment ofcomparative proteome and genome analysis for finding effectivedrug targets against protozoan pathogens.21,22 Recently, anL-asparagine synthase has been found to be important in render-ing survival benefit to Leishmania, and hence has been proposedas a potential drug target.23 Whole proteome BLAST showed thatamongst various protozoan pathogens, a distinct genomic regioncoding for putative L-asparaginase is present only in Leishmania,Giardia and Trichomonas spp. L-Asparaginases belong to the

a Kusuma School of Biological Sciences, Indian Institute of Technology Delhi,

New Delhi-110016, India. E-mail: bkundu@bioschool.iitd.ac.in;

Fax: +91 1126597523; Tel: +91 1126591037b National Institute of Pharmaceutical Education & Research, EPIP Complex,

Hajipur, Vaishali-844102, India

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5mb00251f‡ These authors have equal contribution.

Received 8th April 2015,Accepted 13th April 2015

DOI: 10.1039/c5mb00251f

www.rsc.org/molecularbiosystems

MolecularBioSystems

PAPER

Publ

ishe

d on

13

Apr

il 20

15. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Tec

hnol

ogy

New

Del

hi o

n 20

/04/

2015

15:

50:0

9.

View Article OnlineView Journal

Mol. BioSyst. This journal is©The Royal Society of Chemistry 2015

amidohydrolase class of enzymes and have been used as effectiveanti-leukemic agents.24 Their specific role in pathogenic proto-zoans has not been evaluated. An earlier report suggests that theamino acid L-asparagine (substrate for L-asparaginase) poses aninhibitory control over the leishmanial auto-phagosomal pathway.25

In a recent report it has been shown that the L-asparaginase ofMycobacterium tuberculosis (MyAnsA) renders survival benefit tothe pathogen by reducing the acidic environment inside thehost cell.26 Since a phagolysomal fusion mechanism is com-mon for the establishment of infection by both M. tuberculosisand Leishmania donovani, we hypothesized that LdAI mightbe serving identical neutralizing function for the survival ofLeishmania inside the host. While this requires experimentalverification, we present first hand computational analysis ofthe putative LdAI with a preliminary in vitro testing data tosupport our hypothesis.

Our data shows characteristic deviations of LdAI from theother known L-asparaginases in its catalytic site composition,hinting towards a novel mechanism of action. Based on this weidentified five potential inhibitors of LdAI from a large libraryof compounds. We confirmed our data about the active sitemechanism and the binding poses of the reported inhibitorsusing Molecular docking and MD simulations. Our resultsstrongly support the uniqueness of LdAI that can be exploitedas a potential and specific protein target against leishmaniasis.Hence, we believe that the information reported in our studywill be critical and helpful in designing new and efficaciousinhibitors against lieshmaniasis in future.

2. Materials and methods2.1 Material retrieval and sequence alignment

Protein sequences of L-asparaginase I (Uniprot: P0A962) and II(Uniprot: P00805) from E. coli (EcAI and EcAII) were retrievedfrom UniProtKB and a BLASTp analysis of these sequences wasdone to identify putative L-asparaginase genes associated withenteric pathogens. To identify residues in the catalytic triadsof different asparaginases, multiple sequence alignment ofvarious sequences of Lieshmania spp. (Leishmania donovani,Leishmania infantum, Leishmania major, Leishmania mexicana,Leishmania braziliensis) was done along with L-asparaginasesfrom E. coli using CLUSTALW2.27–29

2.2 Model building and evaluation

The Leishmania donovani L-asparaginase (LdAI) structure wasbuilt using comparative modelling from its primary sequence(Uniprot: E9BC85). Initially, the LdAI sequence was submittedto I-TASSER30,31 (Iterative Threading Assembly Refinement)server that generates atomic models based on multiple threadingalignments of sequences with continuous Z-score assignments.Following this, all the continuously aligned fragments wereassembled using the conformational constraints of the templatestructures. In the next step, structural matching of predictedmodels with known protein structures from the RCSB ProteinData Bank (PDB) was performed and C-scores were assigned

based on relative clustering structural density and consensussignificance. Finally, the best model based on C-scores amongthe top five modelled structures was chosen for further studies.The inspection of phi/psi angles using the Ramachandran plotand energy criterion comparison was done to validate the overallquality of the chosen model using PROCHECK and ProSA32,33

programs, respectively.

2.3 Virtual screening of compounds against LdAI

Virtual screening (VS) was carried out using a composite ligandbased and receptor based approach. Firstly, the filtering con-ditions for virtual screening of approximately 23 million com-pounds from the ZINC database were applied on the basisof the physicochemical properties of the enzyme substrate,L-asparagine. These screened ligands were further filtered usingthe AutoDock Vina34 docking engine, adopted for a receptor-based virtual screening via the iDOCK modification.35 Thedocking grid was defined to encompass the active site of LdAI(residues 11, 16, 21, 89, 91 in the modelled structure). Variousconformational orientations of ligands were analyzed andranked based on their binding scores. Only the top 100 com-pounds with highest binding scores (4L-asp; score �5.3, 100thcompound; score �5.7) were selected based on the Lipinski’srule of five36 and the respective iDock scores. The absorption,distribution, metabolism, excretion and toxicity (ADMET) pro-perties constitute a crucial part of the pharmacokinetic profileof drug candidates. The druggability of candidate compoundsand their clinical effectiveness is based on their ADMET profile.Thus, only the top 20 drug-like compounds were retained basedon predicted low toxicity values using admetSAR.37 The fivereported pharmacokinetic parameters and their evaluationmodels are described in the ESI.†

2.4 Re-docking and free energy calculation

For eliminating the possibility of the inhibitory effect ofthe selected ligands on the human asparaginase isoform(Type-III L-asparaginase, HLA), binding of the top 20 admet-SAR filtered compounds was re-assessed using MolecularOperating Environment (MOE) v2009.10.38 MOE utilizes aMonte Carlo simulated annealing process from a set of energygrids at the active site pocket of the protein. Compared toother docking algorithms such as AutoDock,39 Glide40 andGold,41 MOE can produce ligand poses close to originalconformers with minimum RMSD. Besides, it also sampleslarger conformational spaces which are close to the nativepose of the ligands upon interaction with receptors.42 Hence,MOE was employed for precise free energy calculations. Theligands were re-docked using MOE onto the active pockets ofLdAI and HLA using the above described active site gridparameters. Finally, a triangular matching placement methodwas employed to generate poses by aligning ligand triplets ofatoms with triplets of receptor site points. London dG scoringfunction was used to retain most favorable poses of ligandconformations.

Paper Molecular BioSystems

Publ

ishe

d on

13

Apr

il 20

15. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Tec

hnol

ogy

New

Del

hi o

n 20

/04/

2015

15:

50:0

9.

View Article Online

This journal is©The Royal Society of Chemistry 2015 Mol. BioSyst.

The free energy of ligand binding from a given pose isdefined as:

DG ¼ cþ Ef þX

h-bondsCbFbþ

Xm-ligand

CmFmþX

atoms i

DDi

where, c represents the average gain/loss of rotational andtranslational entropy, Ef is the energy due to the change inflexibility of the ligand upon binding, Fb measures geometricimperfections of hydrogen bonds, Cb is the energy of an idealhydrogen bond, Fm measures geometric imperfections of metalligations, Cm is the energy of an ideal metal ligation; and Di

represents desolvation energy of the ith atom.Difference in desolvation energies are calculated as:

DDi ¼ ciRi3

ðððu=2A[B

juj�6 dU

where, A and B are the protein and ligand volumes with atom ibelonging to volume B; Ri is the solvation radius of atom i and ci

is the desolvation coefficient of atom i. The best five conforma-tions were evaluated using the Generalized Born solvationmodel ranked by higher binding free energy to LdAI in com-parison to HLA.

Esolvation¼�wsolW d�1 � dx

�1� � e24pe

1

2

Xni�1

Xnj�1

qiqjffiffiffiffiffiffiffiffiffiffiGiGj

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiyijþexp �yijð Þ=4

p s rij� �

Tij

where, wsol and W are weights, d is the dielectric constant in theinterior of the solute, dx is the dielectric constant of the solventand s and T are as in the van der Waals energy. Gi is the selfenergy of atom i, defined as:

Gi ¼�12

ffiffiffiffiffiffiffi1

Ri3

3

r� Vi;

Vi �ððð

x2solute

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffix� xij j�6

pd x� xij j4Rið Þdx

where, Ri is the solvent model radius of atom i and Vi is anapproximation to the Born Integral. Subsequently, the resultingfive LdAI–inhibitor complexes were subjected to moleculardynamics simulations to ascertain stability of these complexes.The holo-enzyme complex of LdAI and its substrate L-asparaginewas similarly created by docking the substrate into the active siteof the enzyme.

2.5 Molecular dynamics simulations

MD simulations were carried out using GROMACS packagev4.5.5. The GROMOS 53A6 force field43 was employed foranalysing apo-LdAI (without the substrate) and holo-LdAI(with the substrate or inhibitors). The ligand topologies weregenerated using PRODRG.44 In a separate set of simulations,apo-LdAI, holo-LdAI and the five LdAI–ligand complexes wereplaced in cubical boxes, equidistantly at 15 Å distance from boxedges. Following this, hydrogen atoms were added using apdb2gmx module of GROMACS and were constrained usingthe LINCS algorithm. With periodic boundary conditionsapplied in all three dimensions, protein was explicitly solvatedusing a TIP3P water system and appropriate counter ions wereadded to neutralize the system. The solvated system was first

energy minimized by the steepest descent method (2000 steps)followed by the conjugant gradient (1000 steps) method. In twosteps of 400 ps each, the systems were equilibrated in an NVTensemble followed an NPT ensemble. The Particle mesh Ewald(PME) method was employed to treat long-range electrostaticinteractions with a cut off radius of 10 Å. The system tempera-ture of 300 K was kept constant by modified Berendsencoupling. The Leap frog integration was used for velocitygeneration with 2 fs time step. The structural coordinates wererecorded at 4 ps intervals in a total of 15 ns simulations forapo-LdAI and holo-LdAI along with 10 ns each for LdAI–ligandcomplexes. The resulting trajectories were later analyzed forRMSD, solvent accessible surface area, radius of gyration andEnergy. The structural coordinates were analysed using PyMOLand Discovery Studio visualizer v.3.1.

2.6 Anti-leishmanial activity assay

The effect of compound L1 on the viability of L. donovaniAG83 promastigote cells was determined by a standard 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenylterazolium bromide (MTT)assay. The assay was performed as described previously withminor modifications.54,55 To mimic the leishmanial auto-phagosomal pathway, the assay was performed on low pHadapted promastigotes. For adaptation to lower pH, L.donovaniAg83 promastigotes were seeded gradually to M199 mediumwith Hanks salt at pH 7.0, 6.5, 6.0 and 5.5 with 15% FBS andincubated at 25 1C in a BOD incubator. After the secondpassage at pH 5.5, cells were used for MTT assay. At theexponential phase (B1 � 107 cells per ml), cells were collectedby centrifugation at 1000 � g and were resuspended in freshM199 medium with 15% FBS to make 2.22 � 106 cells per ml.90 ml of cells (2 � 105 cells) were seeded in each well of three96 well plates and incubated at 25 1C in a BOD incubator. After1 hour, cells were treated with equal volumes of L1 (10 ml) atdifferent concentrations and a known concentration (135 nM)of Amphotericin B as control. Plates were incubated at 25 1C ina BOD incubator. After 48 and 72 hours, plates were treatedwith MTT solution (5 mg ml�1 in PBS) with 10 ml added in eachwell. After incubating at 25 1C for 3 hours, 100 ml of stopsolution (50% isopropanol and 10% SDS) were added to eachwell and left for incubation for another 15 minutes. Followingincubation, the optical density at 570 nm was recorded usinga microplate reader and the percentage cytotoxicity was calcu-lated with respect to untreated control cells.

3. Results and discussion3.1 Sequence alignment and model building

Comparative genome analysis highlighted the presence ofthe putative L-asparaginase gene only in few enteric protozoa(Table S1, ESI†). In the case of Leishmania donovani, interest-ingly two gene fragments coding for L-asparaginase were found.One of these present at chromosome number 15, has beenannotated as the cytoplasmic L-asparaginase in the UniProt (E9BC85)database, possibly corresponding to bacterial Type I enzyme.

Molecular BioSystems Paper

Publ

ishe

d on

13

Apr

il 20

15. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Tec

hnol

ogy

New

Del

hi o

n 20

/04/

2015

15:

50:0

9.

View Article Online

Mol. BioSyst. This journal is©The Royal Society of Chemistry 2015

The other (putative) may therefore correspond to the Type IIbacterial L-asparaginase (Fig. S1a, ESI†). Thus, the presenceof the L-asparaginase gene may signify its important role inthe metabolism and survival of organisms possessing it. Uponcomparing the protein sequences of L-asparaginases from variousLeishmania genomes with that of bacterial L-asparaginases, ahigh degree of conservation was noted (Fig. 1). However, fewcrucial differences were also identified including Tyr to Alareplacement in catalytic triad I. The other variations and theirinterpretations are discussed in detail in the following sections.

The human L-asparaginase (HLA) belongs to Type III family,representing a substantially different class and catalytic mecha-nism as compared to bacterial and protozoan L-asparaginases.45

Moreover, there is negligible sequence similarity between HLAand LdAI (BLASTp, E value 1.8). Assuming that LdAI is an essentialenzyme for Leishmania, inhibiting its function will selectivelyprevent the survival of the pathogen, without adversely affectingthe human host. This projects LdAI as a promising candidatefor rational design of selective inhibitors. To study similarity/dissimilarity among various L-asparaginases, a structural modelof LdAI was generated, using EcAI as the template (sequenceidentity 37%). The Ramachandran plot of the modeled LdAIstructure showed 92.5% residues in the most favourable region,6.2% in the allowed region, and 0.4% in the disallowed region(Fig. S1b, ESI†). This was very similar to the EcAI template(Pdb id: 2p2d) that showed the presence of 92.2% residues inthe most favourable region, 7.4% in the allowed region, none inthe disallowed region. For LdAI, it was suspected that theresidues present in the disallowed region (0.4%) could repre-sent the active site amino acids. A PROCHECK analysis howeversuggested that they are not. Further, PROCHECK analysisshowed that the main-chain and side-chain parameters werealso satisfactory, thus confirming the reliability of the LdAIstructure. Finally, the ProSA analysis of energy criteria showednegative interaction energies for most of the residues in LdAI,confirming a stable structure (Fig. S1c, ESI†).

3.2 MD simulations of apo- and holo-enzyme

In the absence of a crystal structure of LdAI, MD simulations of itsapo- and holo-form provided preliminary insights on its 3D con-formation. The simulations of the apo- and holo-forms were carriedout for 10 ns each till the backbone RMSD stabilized (Fig. S2a, ESI†).

Furthermore, comparative stability of the enzyme in the absenceand presence of its substrate was confirmed by consistency intotal energy (potential, kinetic and columbic energies) of boththe apo- and holo-forms during the entire simulation time(Fig. S2b, ESI†). Upon analyzing the overall structural scaffold,it was evident that the LdAI monomer was very similar to thoseof the related asparaginases. Each LdAI monomer is composedof two distinct a/b domains connected by an un-structured linker(Fig. 2). The larger N-terminal domain (green) comprises of 10b-pleated sheets and 3 a-helices along with a rigid hairpincomposition at the active site loop (residues 11–27 in themodelled structure). This is followed by a 21 residue (residues187–207 in the modelled structure) linker region (grey) contain-ing a b-sheet that extends to the C-terminal domain consistingof 2 b-pleated sheets and 4 a-helices (magenta). As a whole, thesecondary structure content is similar to the other reported type Ienzymes i.e. E. coli and Pyrococcus horikosshi (PhA).46,47 The simi-larity in the secondary structural folds is evident from the highdegree of superimposition of EcAI and LdAI models (Fig. S3a, ESI†).

Fig. 1 Multiple sequence alignment of L-asparaginase Type I sequences from available Leishmania genomes with bacterial Type I and Type II (EcAI andEcAII, respectively). The fully conserved residues across the species are shown in red, residues conserved only in Leishmania are shown in pink, Glu 3010

and Val 3020 found conserved in all Leishmania sequences and EcAII only are highlighted in yellow. The green star represents crucial Tyr to Alareplacement in Leishmania L-asparaginases.

Fig. 2 Homology modelled LdAI dimer after 15 ns of MD simulation. Thetwo monomers are shown in violet and green respectively. The dotted redbox highlights the dimer interface, with arrows indicating the two activesites formed at the dimer interface. Enlarged LdAI monomer (blue dottedline) showing the domain architecture comprising of a separate C-terminaldomain (green), an N-terminal domain (violet) and an unstructured linker(orange).

Paper Molecular BioSystems

Publ

ishe

d on

13

Apr

il 20

15. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Tec

hnol

ogy

New

Del

hi o

n 20

/04/

2015

15:

50:0

9.

View Article Online

This journal is©The Royal Society of Chemistry 2015 Mol. BioSyst.

Similar to bacterial and archeal L-asparaginases, LdAI also exhibitedtwo oppositely oriented active sites along the vertical axis.48

Moreover, the C-terminal domain has six extra amino acidsprotruding out of the globular structure when compared toEcAI (Fig. S3b, ESI†). It can be hypothesized that the C-terminaltail is essential for inter-domain stabilization and consequentlyformation of higher order oligomeric assemblies. Furthermore,the linker region was found structured and comparativelylonger in LdAI. Since, Leishmania is an early eukaryote;the linker domain might confer flexibility to C-terminal andN-terminal domains, thus modulating macromolecular associa-tions and subsequently biological functions. This is in accordwith the hypothesis by Wang et al., which sheds light on therole of terminal tails and the varied length of internal linkerson eukaryotic, bacterial and archaeal proteomes.49 These nondomain regions play a crucial role in stabilization of macro-molecular complexes and higher order tertiary structuresthrough various non bonding interactions. Also, a rigidb-hairpin loop at the active site was suggestive of limitedflexibility and restricted solvent accessibility to the catalyticsite (Fig. 2). This might confer LdAI with activity similar to theone seen in the case of EcAI and PhA.48 We confirmed this fromsolvent accessible surfaces of the loop region in LdAI. Thecomparably similar solvent accessibility hints toward similaractivity (EcAI 208 Å2, EcA II 264 Å2, LdAI 240 Å2). Moreover, thecomparative analysis of solvent accessibility of the entire activesite pocket also showed similar results (Fig. S3c, ESI†).

3.3 Active site

Previously reported L-asparaginase crystal structures haveshown the presence of two catalytic triads (I and II) at eachactive site. The triad I is composed of a nucleophile (Thr), abase (Tyr) and a acidic moiety (Glu) which functions on theflexibility of the active site loop. Similarly, triad II is alsocomposed of a second nucleophile (Thr) that activates a watermolecule along with a basic (Lys) and an acidic residue (Asp).These two triads function in a concerted manner and convertthe substrate asparagine into aspartate in the presence of awater molecule (acting as a second nucleophile).50

The multiple sequence alignment of all available LeishmaniaL-asparaginase sequences along with the bacterial L-asparaginase(EcAI and EcAII) yielded interesting insights into the active siteconformation of LdAI (Fig. 1). There were significant overlaps inthe conserved residues of Leishmania asparaginases and thebacterial asparaginases (red). The first nucleophile in triad I(Thr 38) along with the entire triad II (Asp 116, Thr 117 andLys 187) were found conserved (red). However, few evolutionaryreplacements in the LdAI active site make its active site con-formationally different from the known asparaginases. Thecrucial tyrosine moiety (residue 24 on the active site loop) ofEcAI is found replaced with an alanine (residue 48 on the activesite loop) of LdAI (Green star, Fig. 1). Previous studies onbacterial asparaginases have shown that flipping of this tyrosineresidue inside and outside the plain is responsible for activatingthe nucleophilic threonine residue that attacks the substrate.51,52

Hence, an alanine replacement in LdAI (Tyr 48 - Ala) createsa certain degree of muddiness on the mechanism of action ofthis enzyme.

For deciphering this, we did a rigorous sequence and structureanalysis of the simulated structure. We found that there were twolysine residues on the active site loop that were conserved in theLeishmania asparaginases (highlighted in magenta, Fig. 1). How-ever, upon closer look at their structural orientations, only the Lys43 was found in the plane of the triad I. Further, we delved intoidentifying a possible acidic residue that must interact with thebase lysine so as to activate it. From the sequence alignment, it wasfound that glutamic acid (3010 chain B, highlighted in orange,Fig. 1) was conserved in all type I Leishmania asparaginases.Interestingly, this residue is also found conserved for most othertype II asparaginases50 (green star). This led us to hypothesise thatLeishmania being an early eukarya shares the structural andsequence motifs of both type I and type II asparaginases. Hence,we propose the presence of a different version of active siteresidues working in concert to bring upon substrate conversionin LdAI (Fig. 3a). The major difference being at the triad I where thebase Lys 43 after getting deprotonated by acid Glu 3010 activatesthe first nucleophile Thr 38 that attacks the substrate L-asparagine.

We confirmed our results by performing simulations of LdAIin the presence of its substrate asparagine (-holo enzyme).

Fig. 3 Proposed active site conformation of LdAI. (a) Residue compositions of triad I and triad II are shown by dotted red and blue lines, respectively.(b) Secondary structure reorganization of the LdAI active site loop shown in dotted circle in its apo-(beta hairpin, cyan) and holo-(unstructured, green)forms. (c) RMSF of last 4 ns simulations of apo- and holo-LdAI is shown. The higher fluctuation in the active site loop region of holo-enzyme (dottedrectangle) corresponds to the highly flexible unstructured loop region.

Molecular BioSystems Paper

Publ

ishe

d on

13

Apr

il 20

15. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Tec

hnol

ogy

New

Del

hi o

n 20

/04/

2015

15:

50:0

9.

View Article Online

Mol. BioSyst. This journal is©The Royal Society of Chemistry 2015

Interestingly, it was found that the b-hairpin loop at the activesite underwent reorganization and became flexible in thepresence of asparagine (Fig. 3b). The flexibility of the activesite loop gave support to our hypothesis that glutamic acid(residue 3010) de-protonates the base lysine (residue 43) whichin turn activates the nucleophile (Thr 38). The comparison ofRMSF of the apo- and holo-enzymes further confirmed thestabilization of the entire protein upon substrate binding(Fig. S2a and b, ESI†) and reorganization of the active site loopleading to increase of flexibility at the loop region (Fig. 3c). Thisfinding is in contrast to the previous reports that suggested amobile gate mechanism of the active site loop where openingand closing of the loop was found to be ligand-dependent.53

Interestingly, we have recently reported a similar reorganization ofthe structured active site loop in another type I asparaginases(P. furiosus) that assists substrate entry and its conversionto product.56 Therefore, LdAI appears to employ a distinctmechanism of action and warrants further investigation. Inter-estingly, a recently reported L-asparaginase from Mycobacteriumtuberculosis (MyAnsA) shows a similar Val substitution for Tyr.26

Such a finding is pertinent in the context of Leishmania survivalinside the phagolysosome. Leishmania probably involves ananalogous mechanism of action to neutralize the acidicenvironment as is done by MyAnsA of Mycobacterium.

3.4 Virtual screening of compounds against LdAI

The initial library of compounds was created by screeningsubstances from ZINC database that had 90% similarities instructural and physicochemical parameters with L-asparagine.The resulting 11 438 drug-like ligands were identified andretained for the receptor based screening. The compoundsshowing higher binding scores compared to L-asparagine werethen filtered based on the Lipinski’s rule of five and relativeiDock scores. Only the top 100 screened compounds possessinghigher binding affinities compared to the original substrateL-asparagine (4�5.3 kcal mol�1, score for L-asparagine) wereselected. Following this, second screening was done to identify20 potential candidate molecules on the basis of their lowertoxicity profiles (LD50 o 2.5 mol kg�1). The final screening stepinvolved free energy calculations and comparative bindingstudies with HLA using MOE. This resulted in five compoundsshowing higher binding energies with LdAI compared to HLA(Table 1, Fig. S4, ESI†). The compounds are as follows: com-pound L1 (ZINC 565127), compound L2 (ZINC 9013886),

compound L5 (ZINC 86862222), compound L7 (ZINC 83260818)and compound L12 (ZINC 58950950) (Fig. 4). All these potentialinhibitors were successfully filtered by Lipinski’s rule of five(Table 1) and various ADMET parameters (Table 2) usingadmetSAR (ESI†). These ligands screened on the basis ofsimilarity with the substrate showed favorable interactions atthe enzyme’s active site (Fig. 5, Fig. S5, ESI†). This is becauseof the extra number of non-covalent bonding interactions atthe active site pocket, resulting in more stable enzyme–inhibitor complexes.

All five compounds showed excellent probabilities for intestinalabsorption and minimal permeability for blood brain barrier(40.7). The caco-2 permeability scores indicating a moderateparacellular movement of compounds across the intestinalmonolayers also aid in drug-likeliness of the five compounds.Moreover, owing to their moderate P-glycoprotein substratepropensities (B0.6) and high CYP2D6 inducing probabilities(40.9), all five compounds are predicted to have moderatechances of drug efflux with higher probability of forming activemetabolites. Thus, the ADMET profiles of the screened com-pounds suffice appropriate pharmacokinetic parameters andhave reasonable druggability for using as potential drug candi-dates. In the absence of any previous reports on the inhibitionof leishmanial L-asparaginase, these identified compoundsawait in vitro and in vivo testing for their effectiveness againstleishmaniasis.

3.5 Ligand L1

The ligand L1 corresponding to ZINC ID 565127 is recognizedby IUPAC name 2-{3-[2-(3-aminophenyl)-2-oxoethyl]-3-hydroxy-2-oxo-2,3-dihydro-1H-indol-1-yl}acetamide. With net zero chargeunder physiological conditions, it binds to the primary nucleo-phile Thr 38 of catalytic triad I through a hydrogen bondbetween its HG1 and O1. The bulky indole ring gets accommo-dated inside the active site groove and is further stabilized by fivehydrogen bonds with four active site residues at the peripheryi.e. Leu 49, Ser86, Met41 and Asp85 (Fig. 5a).

Table 1 Docking, binding free energy and Lipinski’s parameters for thesubstrate and identified compounds

Compound

Molecularweight(g mol�1) x log P

iDockscore HBA HBD

PSA(Å2)

MOE DGbind(kJ mol�1)

L-Asparagine 132.11 �2.81 �5.37 5 5 110 �11.34L1 339.35 �0.23 �6.13 7 5 126 �22.14L2 302.29 �0.25 �6.17 8 5 130 �20.13L5 266.32 �0.95 �5.92 5 6 110 �18.61L7 336.34 �1.49 �6.27 8 6 135 �14.88L12 274.28 �0.13 �5.98 7 5 120 �13.99

Fig. 4 Molecular structures of five identified small molecule inhibitors(L1, L2, L5, L7, and L12) against LdAI.

Paper Molecular BioSystems

Publ

ishe

d on

13

Apr

il 20

15. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Tec

hnol

ogy

New

Del

hi o

n 20

/04/

2015

15:

50:0

9.

View Article Online

This journal is©The Royal Society of Chemistry 2015 Mol. BioSyst.

3.6 Ligand L2

The ligand L2 corresponding to ZINC ID 9013886 is recognizedby IUPAC name 2-amino-4-hydroxy-5-(4-hydroxy-3-methoxyphenyl)-5,8-dihydropyrido[2,3-d]pyrimidin-7(6H)-one. With net zero chargeunder physiological conditions, it binds to the primary nucleophileThr 38 of catalytic triad I through hydrogen bonding between itsHG1 and O1. Additional hydrogen bonding with six other residuesof the active pocket that include Leu 49, His 192, Asp 85 and Ser 87along with Asp 118 and Met 144, belonging to the triad II groove(Fig. 5b).

3.7 Ligand L5

The ligand L5 corresponding to ZINC ID 86862222 is recognizedby IUPAC name 3,30-diamino-4,40-dihydroxydiphenyl sulfone.With net zero charge under physiological conditions, it bindsto the primary nucleophile Thr 38 of catalytic triad I throughdual-hydrogen bonding between OG1 and H13 and H14 moietiesof L5. The bulky phenyl ring helps in stabilization of L5 insidethe active site groove. The protein–ligand complex is furtherstabilized by nine additional hydrogen bonds between L5 andAsp 85, Ser 86 and Asp 118 residues belonging to the active sitepocket of the protein (Fig. 5c).

3.8 Ligand L7

The ligand L7 corresponding to ZINC ID 83260818 is recognizedby IUPAC name N-[(2R,3S,4R,5S,6R)-2,4,5-trihydroxy-6-(hydroxy-methyl)-4-(1H-indol-3-yl)tetrahydropyran-3yl] acetamide. Withnet zero charge under physiological conditions, L7 also bindsto the primary nucleophile Thr 38 of catalytic triad I throughdual-hydrogen bonding between HG1 and O2 and O5 of L7.Additionally, another hydrogen bond between OG1 of Thr 38 andH2 of the ligand L7 exists. Besides these essential hydrogenbonds, the protein–ligand complex is further stabilized by five

additional hydrogen bonds with Leu 49, Glu 50, Asp 85 and Asp118 residues of the LdAI active site (Fig. 5d).

3.9 Ligand L12

The ligand L12 corresponding to ZINC ID 58950950 is recog-nized by IUPAC name 6-amino-3-methyl-4-(5-methyl-2-furyl)-1,4-dihydropyrano[2,3-c]pyrazole-5-carboxamide. With net zerocharge under physiological conditions, it strongly binds to theprimary nucleophile Thr 38 of catalytic triad I via hydrogenbonding between its OG1 and H8, H9 and H10. The stability ofthis complex is further stabilized by seven additional hydrogenbonds with five residues at the active site pocket that includesLeu49, Glu50, Asp85 along with Asp118 and Met114 belongingto triad 2 pocket (Fig. 5e).

3.10 MD simulations and the overall stability of structure

To validate the stability of enzyme–inhibitor complexesobtained from docking studies, MD simulations were carriedup to 10 ns as described previously in the methods section.In the presence of ligands, the backbone RMSD stabilized

Table 2 ADMET parameters of the identified inhibitors

Compound Blood–brain barrier (�) Intestinal absorption (+) Caco-2 permeability (�) P-glycoprotein substrate CYP2D6 inhibitor

L1 0.97 0.99 0.70 0.71 0.91L2 0.98 0.99 0.50 0.68 0.91L5 0.87 0.98 0.66 0.64 0.84L7 0.98 0.89 0.63 0.66 0.93L12 0.92 0.91 0.67 0.53 0.96

Fig. 5 Molecular representation of active site–inhibitor complexes. Active site molecular interactions for each of the protein–inhibitor complexes isshown. (a) L1; (b) L2; (c) L5; (d) L7; and (e) L12.

Fig. 6 All atom RMSD of LdAI–inhibitor complexes. Different ligands areshown in different color codes.

Molecular BioSystems Paper

Publ

ishe

d on

13

Apr

il 20

15. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Tec

hnol

ogy

New

Del

hi o

n 20

/04/

2015

15:

50:0

9.

View Article Online

Mol. BioSyst. This journal is©The Royal Society of Chemistry 2015

after 6 ns, indicating stabilization of docked complex in allcases (Fig. 6). Changes in radius of gyration over the simula-tion indicate folding behavior of the protein. In the case ofboth docked and native LdAI, the radius of gyration variedbetween 2.6 nm and 2.7 nm (Fig. 7). This suggests that thebinding of the ligands to LdAI enzyme does not affect theoriginal protein architecture. However, under physiologicalconditions, enzymes are dynamic in nature. While someregions are static in nature, others confer flexibility to theenzymes to carry specific functions. RMSF analysis of Caatoms of the protein during the last 4 ns of simulation ofdocked complexes revealed residues contributing to regionsof higher flexibility in the enzyme (Fig. 8). The analysis ofthe active site loop region in the complexes showed that inthe case of L1, L7 and L12 ligands, comparatively lowerresidue-fluctuation was present (dotted rectangle, Fig. 3).Finally, the total energy profile of the enzyme–ligand com-plexes during the entire simulation was also analyzed toascertain the overall stability of the systems in terms ofpotential, kinetic and columbic energies. It was observed thatall five complexes showed s constant energy profile during thetrajectory of simulation (Fig. S6, ESI†). So, based on stronghydrogen bonding interactions at the active site groove thatremained throughout the MD simulation trajectory, these fiveidentified compounds show excellent stability and reliability.

3.11 Assessment of anti-leishmanial activity

Following the stringent screening criteria, we examined theanti-leishmanial activity of one of the five identified com-pounds. Upon testing, L1 showed appreciable reduction in cellviability of L. donovani AG83 promastigote cells in a dosedependent manner. At 72 hours, the inhibitory effect wascomparable to that of the standard drug (Amphotericin B)(Fig. 9). Although the dose requirement was on the higher sidebut given the ADMET criteria of selecting relatively low toxicitycompounds, L1 appears promising. Overall, these results representa starting point of rigorous in vivo experimentation for develop-ing newer and potent drugs from the identified anti-leishmanialtarget.

4. Conclusions

In the present study, we have explored the possibility oftargeting L-asparaginases in pathogenic Leishmania donovani(LdAI) as a new therapeutic strategy against lesihmaniasis.Based on our sequence and structural analysis, we report anovel mechanism of active site remodelling involving differentresidues at the active site compared to previously reportedasparaginase. Following this, we screened a large library ofcompounds for identifying inhibitors which could favourablyinteract with the active site of LdAI. Based on an integratedapproach involving molecular modelling, docking and MDsimulations, we selected five compounds which could stronglybind at the LdAI active site. The binding mode analysis of allfive compounds showed the presence of a strong hydrogenbonding network involving the active site residues. To the bestof our knowledge this is a firsthand report on targetingL-asparaginase in a pathogenic protozoan.

The major conclusions of this study are as follows: (1) acomparison with previously reported asparaginases revealed aflexible domain architecture of LdAI, owing to the longer inter-domain linker length. (2) The catalytic triad I of LdAI constitutesdifferent residues (Lys 43 as base, Glu 3010 as acid and Thr 38 asnucleophile), not reported in other L-asparaginases. (3) The rigid

Fig. 7 Radius of gyration of backbone atoms of LdAI–inhibitor complexes.Different ligands are shown in different color codes.

Fig. 8 RMSF of LdAI–inhibitor complexes. Different ligands are shown indifferent color codes. Black stars represent active site residues.

Fig. 9 Cytotoxicity profile of compound L1 against L. donovani AG83promastigote cells. Cell toxicity of different concentrations of L1 is repre-sented as a percentage of the untreated control after 72 h treatment.Amphotericin B (135 nM) was used as a positive control.

Paper Molecular BioSystems

Publ

ishe

d on

13

Apr

il 20

15. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Tec

hnol

ogy

New

Del

hi o

n 20

/04/

2015

15:

50:0

9.

View Article Online

This journal is©The Royal Society of Chemistry 2015 Mol. BioSyst.

active site loop of LdAI gets un-structured in the presence of asubstrate, possibly enhancing substrate accessibility. (4) Theselected inhibitors showed high affinity to LdAI and are devoidof any significant affinity to the human asaparaginase. (5) Theheterocyclic rings in the five identified inhibitors help in stabili-zation at the active site groove along with the hydrogen bondingwith the surrounding residues. (6) Compound L1 works at a higherdose but produces comparable anti-leishmanial activity to that ofstandard compounds. In summary, these mechanistic insightsinto the functioning of pathogenic protozoan L-asparaginase canhelp in deciphering the survival strategies of these organisms andhence can also aid in developing novel therapeutic strategies. Thepreliminary results suggest that the identified lead molecules mayserve as models for developing new and efficient inhibitors againstleishmaniasis.

Acknowledgements

This work was supported financially and infrastructurally by IITDelhi. JS and AS acknowledge the financial support receivedfrom IIT Delhi and ICMR India, respectively. PJ and KKSacknowledge the financial and infrastructural support of NIPERHajipur. The authors also acknowledge Banpreet Kaur andVimal Kant for technical discussions and suggestions.

References

1 P. Kaye and P. Scott, Leishmaniasis: complexity at the host-pathogen interface, Nat. Rev. Microbiol., 2011, 9, 604–615.

2 P. J. Guerin, P. Olliaro, S. Sundar, M. Boelaert, S. L. Croft andP. Desjeux, et al., Visceral Leishmaniasis: current status ofcontrol, diagnosis, and treatment, and a proposed researchand development agenda, Lancet Infect. Dis., 2002, 2, 494–501.

3 S. L. Croft and G. H. Coombs, Leishmaniasis-current chemo-therapy and recent advances in the search for novel drugs,Trends Parasitol., 2003, 19, 502–508.

4 M. L. Fernandez-Guerrero, J. M. Aguado, L. Buzon, C. Barros,C. Montalban and T. Martin, et al., Visceral leishmaniasis inimmunocompromised hosts, Am. J. Med., 1987, 83, 1098–1102.

5 S. L. Reiner, S. Zheng, Z. E. Wang, L. Stowring andR. M. Locksley, Leishmania promastigotes evade interleukin12 (IL-12) induction by macrophages and stimulate a broadrange of cytokines from CD4 + T cells during initiation ofinfection, J. Exp. Med., 1994, 179, 447–456.

6 T. S. Tiuman, A. O. Santos, T. Ueda-Nakamura, B. P. Filhoand C. V. Nakamura, Recent advances in leishmaniasistreatment, Int. J. Infect. Dis., 2011, 15, e525–e532.

7 A. C. Cunningham, Parasitic adaptive mechanisms in infectionby leishmania, Exp. Mol. Pathol., 2002, 72, 132–141.

8 M. A. Vannier-Santos, A. Martiny and W. de Souza,Cell biology of Leishmania spp.: invading and evading,Curr. Pharm. Des., 2002, 8, 297–318.

9 F. Frezard, C. Demicheli and R. R. Ribeiro, Pentavalentantimonials: new perspectives for old drugs, Molecules,2009, 14, 2317–2336.

10 R. R. New, M. L. Chance and S. Heath, Antileishmanial activityof amphotericin and other antifungal agents entrapped inliposomes, J. Antimicrob. Chemother., 1981, 8, 371–381.

11 P. L. Olliaro and A. D. Bryceson, Practical progress and newdrugs for changing patterns of leishmaniasis, Parasitol.Today, 1993, 9, 323–328.

12 C. A. Bell, M. Cory, T. A. Fairley, J. E. Hall and R. R. Tidwell,Structure-activity relationships of pentamidine analogsagainst Giardia lamblia and correlation of antigiardialactivity with DNA-binding affinity, Antimicrob. AgentsChemother., 1991, 35, 1099–1107.

13 C. A. Bell, J. E. Hall, D. E. Kyle, M. Grogl, K. A. Ohemeng andM. A. Allen, et al., Structure-activity relationships of analogsof pentamidine against Plasmodium falciparum andLeishmania mexicana amazonensis, Antimicrob. AgentsChemother., 1990, 34, 1381–1386.

14 A. C. Smith, V. Yardley, J. Rhodes and S. L. Croft, Activityof the novel immunomodulatory compound tucaresolagainst experimental visceral leishmaniasis, Antimicrob.Agents Chemother., 2000, 44, 1494–1498.

15 L. C. Afonso, T. M. Scharton, L. Q. Vieira, M. Wysocka,G. Trinchieri and P. Scott, The adjuvant effect ofinterleukin-12 in a vaccine against Leishmania major,Science, 1994, 263, 235–237.

16 S. St George, J. V. Bishop, R. G. Titus and C. P. Selitrennikoff,Novel compounds active against Leishmania major, Antimicrob.Agents Chemother., 2006, 50, 474–479.

17 J. van Griensven, E. Diro, R. Lopez-Velez and M. Boelaert, HIV-1protease inhibitors for treatment of visceral leishmaniasis inHIV-co-infected individuals, Lancet Infect. Dis., 2013, 13,251–259.

18 T. K. Jha, S. Sundar, C. P. Thakur, P. Bachmann,J. Karbwang and C. Fischer, et al., Miltefosine, an oralagent, for the treatment of Indian visceral leishmaniasis,N. Engl. J. Med., 1999, 341, 1795–1800.

19 J. Soto-Mancipe, M. Grogl and J. D. Berman, Evaluation ofpentamidine for the treatment of cutaneous leishmaniasisin Colombia, Clin. Infect. Dis., 1993, 16, 417–425.

20 S. Sundar, T. K. Jha, H. Sindermann, K. Junge, P. Bachmannand J. Berman, Oral miltefosine treatment in children withmild to moderate Indian visceral leishmaniasis, Pediatr.Infect. Dis. J., 2003, 22, 434–438.

21 L. Florens, M. P. Washburn, J. D. Raine and R. M. Anthony,et al., A proteomic view of the Plasmodium falciparum lifecycle, Nature, 2002, 419, 520–526.

22 D. Damte, J. W. Suh, S. J. Lee, S. B. Yohannes, M. A. Hossainand S. C. Park, Putative drug and vaccine target proteinidentification using comparative genomic analysis of KEGGannotated metabolic pathways of Mycoplasma hyopneumoniae,Genomics, 2013, 102, 47–56.

23 R. Manhas, P. Tripathi, S. Khan, B. Sethu Lakshmi, S. K.Lal, V. S. Gowri, A. Sharma and R. Madhubala, Identifi-cation and functional characterization of a novel bacterialtype asparagine synthetase A: a tRNA synthetase paralogfrom Leishmania donovani, J. Biol. Chem., 2014, 289,12096–12108.

Molecular BioSystems Paper

Publ

ishe

d on

13

Apr

il 20

15. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Tec

hnol

ogy

New

Del

hi o

n 20

/04/

2015

15:

50:0

9.

View Article Online

Mol. BioSyst. This journal is©The Royal Society of Chemistry 2015

24 S. Bansal, D. Gnaneswari, P. Mishra and B. Kundu, Struc-tural stability and functional analysis of L-asparaginasefrom Pyrococcus furiosus, Biochemistry, 2010, 75, 375–381.

25 S. St George, J. V. Bishop, R. G. Titus and C. P. Selitrennikoff,Novel compounds active against Leishmania major,Antimicrob. Agents Chemother., 2006, 50, 474–479.

26 A. Gouzy, G. Larrouy-Maumus, D. Bottai, F. Levillain, A. Dumasand J. B. Wallach, et al., Mycobacterium tuberculosis exploitsasparagine to assimilate nitrogen and resist acid stress duringinfection, PLoS Pathog., 2014, 10, e1003928.

27 M. A. Larkin, G. Blackshields, N. P. Brown, R. Chenna,P. A. McGettigan and H. McWilliam, et al., Clustal W andClustal X version 2.0, Bioinformatics, 2007, 23, 2947–2948.

28 H. McWilliam, W. Li, M. Uludag, S. Squizzato, Y. M. Parkand N. Buso, et al., Analysis Tool Web Services from theEMBL-EBI, Nucleic Acids Res., 2013, 41, W597–W600.

29 M. Goujon, H. McWilliam, W. Li, F. Valentin, S. Squizzato andJ. Paern, et al., A new bioinformatics analysis tools framework atEMBL-EBI, Nucleic Acids Res., 2010, 38, W695–W699.

30 Y. Zhang, I-TASSER server for protein 3D structure prediction,BMC Bioinf., 2008, 9, 40–48.

31 A. Roy, A. Kucukural and Y. Zhang, I-TASSER: a unifiedplatform for automated protein structure and functionprediction, Nat. Protoc., 2010, 5, 725–738.

32 R. A. Laskowski, J. A. Rullmannn, M. W. MacArthur, R. Kapteinand J. M. Thornton, AQUA and PROCHECK-NMR: programs forchecking the quality of protein structures solved by NMR,J. Biomol. NMR, 1996, 8, 477–486.

33 M. Wiederstein and M. J. Sippl, ProSA-web: interactive webservice for the recognition of errors in three-dimensional struc-tures of proteins, Nucleic Acids Res., 2007, 35, W407–W410.

34 O. Trott and A. J. Olson, AutoDock Vina: improving the speed andaccuracy of docking with a new scoring function, efficient optimi-zation, and multithreading, J. Comput. Chem., 2010, 31, 455–461.

35 L. Hongjian, K. Leung, P. J. Ballester and M. Wong, istar: AWeb Platform for Large-Scale Protein-Ligand Docking, PLoSOne, 2014, 9, e85678.

36 C. A. Lipinski, Chris Lipinski discusses life and chemistryafter the Rule of Five, Drug Discovery Today, 2003, 8, 12–16.

37 F. Cheng, W. Li, Y. Zhou, J. Shen, Z. Wu and G. Liu, et al.,admetSAR: a comprehensive source and free tool for assess-ment of chemical ADMET properties, J. Chem. Inf. Model.,2012, 52, 3099–3105.

38 S. Vilar, G. Cozza and S. Moro, Medicinal chemistry and themolecular operating environment (MOE): application ofQSAR and molecular docking to drug discovery, Curr. Top.Med. Chem., 2008, 8, 1555–1572.

39 G. M. Morris, R. Huey, W. Lindstrom, M. F. Sanner,R. K. Belew, D. S. Goodsell and A. J. Olson, Autodock4 andAutoDockTools4: automated docking with selective receptorflexiblity, J. Comput. Chem., 2009, 16, 2785–2791.

40 T. A. Halgren, R. B. Murphy, R. A. Friesner, H. S. Beard,L. L. Frye, W. T. Pollard and J. L. Banks, Glide: A NewApproach for Rapid, Accurate Docking and Scoring. 2.Enrichment Factors in Database Screening, J. Med. Chem.,2004, 47, 1750–1759.

41 M. L. Verdonk, J. C. Cole, M. J. Hartshorn, C. W. Murray andR. D. Taylor, Improved Protein-Ligand Docking Using GOLD,Proteins, 2003, 52, 609–623.

42 M. Feher and C. I. Williams, Effect of input differences onthe results of docking calculations, J. Chem. Inf. Model.,2009, 49, 1704–1714.

43 S. Pronk, S. Pall, R. Schulz, P. Larsson, P. Bjelkmar andR. Apostolov, et al., GROMACS 4.5: a high-throughput andhighly parallel open source molecular simulation toolkit,Bioinformatics, 2013, 29, 845–854.

44 A. W. Schuttelkopf and D. M. F. van Aalten, ‘‘PRODRG: a tool forhigh-throughput crystallography of protein-ligand complexes’’,Acta Crystallogr., Sect. D: Biol. Crystallogr., 2004, 60, 1355–1363.

45 J. Nomme, Y. Su, M. Konrad and A. Lavie, Structures of apoand product-bound human L-asparaginase: insights into themechanism of autoproteolysis and substrate hydrolysis,Biochemistry, 2012, 51, 6816–6826.

46 M. Yao, Y. Yasutake, H. Morita and I. Tanaka, Structure of thetype I L-asparaginase from the hyperthermophilic archaeonPyrococcus horikoshii at 2.16 angstroms resolution, ActaCrystallogr., Sect. D: Biol. Crystallogr., 2005, 61, 294–301.

47 A. L. Swain, M. Jaskolski, D. Housset, J. K. Rao and A. Wlodawer,Crystal structure of Escherichia coli L-asparaginase, an enzyme usedin cancer therapy, Proc. Natl. Acad. Sci. U. S. A., 1993, 90, 1474–1478.

48 S. Bansal, A. Srivastava, G. Mukherjee, R. Pandey, A. K. Verma,P. Mishra and B. Kundu, Hyperthermophilic asparaginasemutants with enhanced substrate affinity and antineoplasticactivity: structural insights on their mechanism of action,FASEB J., 2012, 26, 1161–1171.

49 M. Wang, C. G. Kurland and G. Caetano-Anolles, Reductiveevolution of proteomes and protein structures, Proc. Natl.Acad. Sci. U. S. A., 2011, 108, 11954–11958.

50 M. Sanches, S. Krauchenco and I. Polikarpov, Structure,Substrate Complexation and Reaction Mechanism of Bac-terial Asparaginases, Curr. Chem. Biol., 2007, 12, 75–86.

51 U. Bagert and K. H. Rohm, On the role of histidine andtyrosine residues in E. coli asparaginase, Chemical modifi-cation and 1H-nuclear magnetic resonance studies, Biochim.Biophys. Acta, 1989, 9991, 36–41.

52 C. Derst, A. Wehner, V. Specht and K. H. Rohm, States andfunctions of tyrosine residues in Escherichia-coli asparaginase-II,Eur. J. Biochem., 1994, 224, 533–540.

53 H. P. Aung, M. Bocola, S. Schleper and K. H. Rohm, Dynamics ofa mobile loop at the active site of Escherichia coli asparaginase,Biochim. Biophys. Acta, 2000, 14812, 349–359.

54 M. Mirzaie, S. J. Nosratabadi, A. Derakhshanfar andI. Sharifi, Antileishmanial activity of Peganum harmalaextract on the in vitro growth of Leishmania major promas-tigotes in comparison to a trivalent antimony drug, Vet. Arh.,2007, 77, 365–375.

55 M. Niks and M. Otto, Towards an optimized MTT assay,J. Immunol. Methods, 1990, 130, 149–151.

56 R. Tomar, P. Sharma, A. Srivastava, S. Bansal, Ashish and B. Kundu,Structural and functional insights into an archael L-asparaginaseobtained through linker-less assembly of constituent domains, ActaCrystallogr., Sect. D: Biol. Crystallogr., 2014, 70, 3187–3197.

Paper Molecular BioSystems

Publ

ishe

d on

13

Apr

il 20

15. D

ownl

oade

d by

Ind

ian

Inst

itute

of

Tec

hnol

ogy

New

Del

hi o

n 20

/04/

2015

15:

50:0

9.

View Article Online