In Silico Characterization of Atypical Kinase PFD0975w from Plasmodium Kinome: A Suitable Target For Drug Discovery

In Silico Characterization of Atypical KinasePFD0975w from Plasmodium Kinome: A SuitableTarget For Drug Discovery

Vishal Trivedi* and Swagata Nag

Department of Biotechnology, Malaria Research Group, IndianInstitute of Technology-Guwahati, Guwahati-781039, Assam, India*Corresponding author: Vishal Trivedi, [email protected];[email protected]

RIO-2 kinase is known to regulate ribosome bio-genesis and other cell cycle events. The 3D modelof ATP bound and an unbound form of PFD0975wwas generated using AfRIO-2 crystal structure1TQI, 1ZAO as template employing MODELLER9v7program. Structural characterization identified N-terminal winged helix domain (1–84), C-terminalkinase domain (148–275), and presence of othercritical residues known for ATP binding andkinase activity. Using Q-site and pocket finder, anumber of well-defined substrate (peptide) bindingregions were identified in the catalytic core of theprotein. The peptide binding regions were furthervalidated by molecular modeling a non-specificpolyalanine peptide and a sequence-specific pep-tide2 into these sites to generate a stablePFD0975w/peptide complexes. Peptide fits wellinto identified pocket on PFD0975w and makesextensive interaction with the protein residues.These newly identified peptide binding sitespotentially give opportunity to design a specificinhibitor against PFD0975w. There are subtle butsignificant differences between Plasmodium falci-parum and human RIO-2 to exploit PFD0975w fordrug development. In conclusion, our finding willlet us to design effective chemotherapy againstmalaria parasite exploiting PFD0975w as a drugtarget.

Key words: atypical kinase, kinome, malaria, ribosome biogenesis,RIO kinase

Received 31 March 2011, revised 5 December 2011 and accepted forpublication 28 December 2011

Malaria causes 250 million cases and more than one million deathsannually. The people living in unhygienic conditions, with inade-quate food intake and weak immunological systems, are more sus-ceptible to the disease (1). Parasite depends on hemoglobinmetabolism to provide nutrient, space for growth, and maintainingosmotic balance within an infected red blood cell (RBC) (2–4).

Besides these pro-supportive roles, hemoglobin degradation gener-ates heme, peroxide, and other free radicals (5). To detoxify toxiceffects of heme, the parasite is well equipped with the antioxidantsystem [glutathione (GSH) and thioredoxin dependent] and lowmolecular weight antioxidants such as GSH. Disturbance of theseredox balance causes parasite death because of apoptotic-like pro-cess, and it is a major mechanism of several drugs in a clinical set-ting (6–9). On the other hand, parasite up-regulate its ownantioxidant defense to acquire drug resistance against a number ofknown antimalarials (10–12). Thus, the disease situation is alarm-ing, and there is an urgent need to identify and characterize noveldrug target from parasites.

Transcriptome studies of intra-erythrocytic stages of parasite indi-cate that oxidative stress activates the transcriptional machinery toproduce antioxidant enzymes, proteins involved in merozoite inva-sion, and drive different phases of life-cycle in erythrocytes (13–16).In eukaryotic system, protein kinases (ePKs) are involved inresponses to oxidative stress and survival of the organism. Themalarial genome contains protein kinases, and bioinformatics analy-sis identified approximately 65 eukaryotic-like protein kinasesequences that belong to different established ePKs families.Besides known classes of protein kinases, parasite also containunique family of kinases known as atypical kinases (17).

Atypical kinases exhibit protein kinase activity with little sequencesimilarity to known typical kinase present in eukaryotic systems.These kinases are known to regulate a number of biological pro-cesses at the cellular level through their phosphorylation activityand proposed to be important drug targets (18,19). RIO kinase, oneof the members of the atypical kinase group, is known to regulateribosome biogenesis and cell cycle in yeast (20). These are essen-tial kinases for organism survival, and at least two members of thisgroup are present in each species from archaea to human, namedas RIO-1 and RIO-2. A third member of RIO family is identified andtermed as RIO-3. It is more prevalent in multicellular organisms andstructurally more similar to RIO-1, but has a distinct N-terminaldomain (21). RIO-1 or RIO-2 deletion in yeast is lethal, and theseobservations indicate a distinct physiological role of each subfamilymember (22,23). In general, RIO kinase contains N-terminal domainand a C-terminal kinase domain, which are connected with eachother by a linker region. RIO-2 has a characteristic N-terminalwinged helix (wHTH) domain, known to bind DNA and present intranscription factors & DNA binding proteins (24). The RIO kinasedomain is short because of truncation of subdomain VIII (activation

600

Chem Biol Drug Des 2012; 79: 600–609

Research Letter

ª 2012 John Wiley & Sons A/S

doi: 10.1111/j.1747-0285.2012.01321.x

loop), X, and XI, an important region known to provide substratespecificity and binding in typical protein kinase (22). ATP bindingpockets between RIO-1, RIO-2, and ePKs are quite distinct, espe-cially charge distribution and mode of nucleotide binding. Bothkinase subfamilies share a number of biochemical and structuralfeatures besides distinct features of their own respective group,and these differences are good enough to design drugs not onlyRIO-specific but also subfamily-specific as well (24,25).

PFD0975w, RIO-like kinase is present in the Plasmodium falciparumkinome (17). Transcriptome studies of intra-erythrocytic stages indi-cate that PFD0975w expression does not change, and it exhibits aconstitutive expression throughout the life-cycle stages in RBC. Inthis study, we have described the first structural report of any atyp-ical kinase from P. falciparum. Moreover, in-depth structural detailsallow us to understand salient biochemical properties of theenzyme. In summary, our work clearly highlights several distinctstructural features of PFD0975w, which can be exploited to designspecific and better drug against P. falciparum.

Methods

Multiple sequence alignmentRIO-2 sequences from Archaeoglobus fulgidus (accession no.030245), Saccharomyces cerevisiae (accession no. NP_014192), andHomo sapiens (accession no. Q9BVS4.2) were retrieved from NCBIdatabase, and multiple sequence alignment was performed usingCLUSTALW 2.0.11 (26).

Molecular modeling and structure validationPFD0975w protein sequence was retrieved from the Plasmodiumgenome database, plasmodb (http://www.plasmodb.org). The suit-able templates for homology modeling were selected by blastingthe PFD0975w protein sequence into protein data bank of NCBIusing PSI-blast. Crystal structure of RIO-2 (native form) from A. ful-gidus (PDB code 1TQI) was found to be suitable with 83.39% simi-larity over 265 residues (24). Remaining residues of PFD0975w(approximately 316 residues) could not show any significant similar-ity, and hence, this part was not taken for homology modeling. Thecomplete protocol to generate 3D model was as follows: An initial3D model (1–300) was generated by the automodel module of MOD-

ELLER 9v7 and ranked based on DOPE scores. Ten high-score modelswere selected, and model quality was assessed by PROCHECK (27).All models were energy minimized by Swiss PDB VIEWER 4.0.1(Swiss Institute of Bioinformatics, Switzerland), and structural qual-ity of the refined models was again assessed by PROCHECK (28). Loopregions were modeled using loop refinement module of modeler.After each cycle of loop refinement, ERRAT plot was generated tocheck structure quality. This process was repeated in an iterativefashion until all the residues are not below 95% in ERRAT plot(29). The structural quality of final model was verified by Verify 3D,Procheck, and Ramachandran plot (30). The overall RMSD betweenthe template and modeled structure was 0.58 � over all Ca. A simi-lar homology modeling protocol was used to model ATP-bound formof PFD0975w using ATP-bound form of AfRIO-2 (PDB code 1ZAO) asa template (31). As the template and modeled structure were very

close to each other, ATP and Mg2+ were taken from template 1ZAOand added into modeled PfRIO-2 structure for studying ATPinteraction within PFD0975w structure.

Identification of hot spots on PFD0975wstructure for peptide bindingFinal model of PFD0975w was used to identify potential hot spotregions for binding peptide using Q-site and POCKET FINDER (32). Thecomplete protocol is as follows: PFD0975w protein surface wasscanned using a hydrophobic probe (CH3), and binding energy wascalculated for each probe cluster. These clusters were ranked basedon sum of total binding energy for each cluster. Clusters with highbinding energy were given as hot spots and were considered forpeptide modeling experiments.

Molecular modeling of ATP into PFD0975wstructureThe docking program AUTODOCK 4.1 (Molecular Graphics Laboratory,The Scripps Research Institute, La Jolla, CA, USA) was used to gen-erate ATP-bound PFD0975w and a HuRIO-2 model for interactionstudies (33). The complete docking protocol was as follows: first,the 3D coordinate of ATP was taken from PDB code 1ZAO andchecked for polar hydrogens, atomic charges, and flexible torsionsas defined. The resulting PDBQT file was generated for ATP. Formacromolecules, PFD0975w, polar hydrogen, kollman charges, andatomic solvation parameters were assigned, and a grid of 0.4 �resolution was centered using autogrid module of AUTODOCK 4.1. Anumber of docking parameters, number of generations, energy eval-uation, and the GA run were set as 27 000, 2 500 000, and 10,respectively. Finally, docked conformations were selected based oninteraction energy. Under the identical parameters, ATP was dockedinto the HuRIO-2 structure to generate HuRIO-2 ⁄ ATP complexes.The best complexes were used for comparative studies betweenPFD0975w and HuRIO-2 interaction with bound ATP.

Molecular modeling of the PFD0975w ⁄ peptidecomplexAs a number of potential peptide binding hot spots are identifiedon PFD0975w structure, we modeled peptide in these sites to fur-ther validate the results. As an optimal peptide sequence forPFD0975w ⁄ peptide complex was not available, we have used twodifferent types of peptides to generate RIO-2: peptide model. Ini-tially, a polyalanine peptide (from PDB code 2A79) was used toprobe the peptide binding site. A molecular modeling docking algo-rithm patchdock was used to generate a stable PFD0975w ⁄ Polyala-nine complex (34). The complete molecular modeling protocol wasas follows: patchdock scans the protein surface containing hotspots with highest shape complementarity to the given peptide. Thebest 10 solutions were energy minimized using Fire-dock to obtainstable PFD0975w ⁄ polyalanine complex (35). All the energy-mini-mized structures were analyzed and sorted based on their bindingenergy and local proximity to the predicted binding sites onPFD0975w. In another set of modeling experiments, we have useda sequence-specific peptide2 to represent specific proteinous sub-strates. 3D coordinates of peptide2 (TTYADFIASGRTGRRNAIHD) from

In Silico Analysis of PfRIO-2

Chem Biol Drug Des 2012; 79: 600–609 601

co-crystallized structure of the cAMP-dependent kinase (PDB code1APM) were taken to model PFD0975w ⁄ peptide2 molecular model.A similar molecular modeling protocol has been employed to gener-ate PFD0975w ⁄ peptide2 complex. The model will be very useful totest sequence variability (to mimic different protein substrate) andpeptide binding pocket.

Results

PFD0975w belongs to RIO-2 kinase familyIn plasmodium genome database (http://www.plasmodb.org),PFD0975w is annotated as RIO like kinases. RIO kinases are presentin archaea to human, and at least two different types of RIO kinas-es (RIO-1 and RIO-2) are found in every organism (22). These kinas-es are known to share a high level of sequence similarity, but atlarge, every class has their own distinct features both at sequenceand structure level. The first question we ask is to which classPFD0975w belongs. To address this question, we blasted thePFD0975w protein sequence into a protein sequence database (Ref-seq_protein) at NCBI, and it picked up more than 100 blast hits.The majority of sequences found in blast searches belong to RIO-1and RIO-2 kinases, serine ⁄ threonine kinase, etc. To further identifydifferent domains in PFD0975w, we performed a multiple sequencealignment with the RIO-2 kinase sequence from A. fulgidus, S. cere-visiae, and H. sapiens. Multiple sequence alignment clearly showsthe presence of winged helix (wHWH) domain at the N-terminal ofthe protein (1–84), a long linker region (85–147), and a C-terminalcatalytic kinase domain (148–275). N-terminal wHWH domain is thecharacteristic of RIO-2 kinase and one of the distinguishing featurebetween RIO-1 and RIO-2 kinases, and this domain is completelyabsent in RIO-1 kinase (36). Sequence alignment also allows us toidentify different kinase subdomains present in PFD0975w(Figure 1). As a member of atypical kinase, PFD0975w lacks VIII, X,XI kinase subdomains, and a well-defined activation loop.

Quality of modeled PFD0975w modelTo obtain the structural details and arrangement of these domainswithin PFD0975w, we have modeled the structure of the PFD0975wusing crystal structure of A. fulgidus (AfRIO-2) as a template (PDBaccession number 1TQI). Most of the residues were present withinthe allowed region of Ramachandran plot (Figure 2A). ERRAT PLOTanalysis of the final model shows that most of the residues below95% with model structural quality factor 91.667. Overall, no shortcontact was observed in the final model (Figure 2B). In Verify 3Dplot, 91.32% residues of the model are above 0.22 except unstruc-tured loop region 123–149 (Figure 2C). PFD0975w-ATP, HuRIO-2, andHuRIO-2-ATP models were of high quality, and structure validationresults were given in Table S1.

Overall structure of PFD0975wStructural features present in PFD0975w are comparable to thestructure of AfRIO-2 with similar arrangements of structural ele-ments (Figure 3A). N-terminal wHWH domain comprised of three a-helices followed by b-sheet and a fourth a helix, followed by along linker region and at C-terminal domain with three a-helix and

two b-sheets. The C-terminal kinase domain is conserved in bothRIO-1 and RIO-2 proteins and are present in other protein kinases.A topology diagram of the PFD0975w was given in Figure 3B todepict the organization of structural units. In structural comparisonwith AfRIO-1, AfRIO-2, and protein kinase A (PKA) structures, anumber of conserved residues have been identified. Phe-229 (Tyr222 in AfRIO-2) is present very close to c-phosphate of ATP, and itsbulky side chain probably regulates the substrate size or entry intothe active site. The loop residues (123–149) are unstructured inPFD0975w, and this region contains most of the conserved residuesknown for this class of kinase (Figure 3A, green colored). The com-parable region in the crystal structure of AfRIO-2 (template used inthis study) has a very weak electron density, and this clearly indi-cates that a high degree of mobility exists in this region of the pro-tein (24). This region in PFD0975w might play a role in substraterecognition, or it regulates the opening of active site or brings resi-due in the active site during catalysis. Although more detailedstructure–function analysis is required to clearly understand the roleof unstructured loop region in PFD0975w catalysis.

ATP binding regionTo understand the nucleotide binding pocket and interaction of ATPwithin the PFD0975w, we superimposed the modeled structure withATP-bound structure of AfRIO-2 (PDB accession code 1ZAO). ATP isin extensive interaction with protein residues present in the ATPbinding pocket of PFD0975w (Table 1). The binding pocket has a

Figure 1: Characterization of different domains in PFD0975w.Multiple sequence alignments of PFD0975w (1-300) with a RIO-2kinase from Archaeoglobus Fulgidus (AfRio2), Saccharomyces cerevi-siae (ScRio2), and Homo sapiens (HsRio2). A shaded block is usedto mark N-terminal winged helix domain and C-terminal catalyticdomain. Residues highlighted in gray and black represent similarand identical residues, whereas * denotes residues involved incatalysis and interaction with substrate.

Trivedi and Nag

602 Chem Biol Drug Des 2012; 79: 600–609

mixed environment to support different types of chemical moietiespresent in the ATP molecule (Figure 4A). A hydrophobic platformmade up of A171, V177, and F236 supports ribose ring, and otheramino acids such as A157, I161, and M186 make several inter-actions with the adenine ring, whereas negatively charged PO4group remains in a positive-charged tunnel (surface volume approxi-

mately 14.509 �3) lined by I148, S150, R152, P237 and other nearbyresidues such as E104, S105, K121, H123, R127, D243, P245.Besides the protein residues that make up the binding pocket forATP, several other residues are also in interaction with differentparts of the nucleotide (Figure 4B).The adenine ring forms a hydro-gen bond with the E104, whereas ATP a-phosphate is stabilized by

A

C

B

Figure 2: Quality of modeled PFD0975w structure. The PFD0975w model was validated by (A) Ramachandran plot (B) Errat plot and (C)Verify 3D. Analysis indicates that modeled structure was of high quality.

A B

Figure 3: The overall fold of native PFD0975w from Plasmodium falciparum. (A) The modeled structure of PFD0975w shows differentstructural elements, N-terminal helix winged domain (cyan blue), linker region (yellow), unstructured region (green), and C-terminal domain(red). N-terminal helix-winged-helix domain (wHWH) and C-terminal kinase are marked by shaded block. (B) Topology diagram of PFD0975 todepict the organization of structural elements.



nearby K121. The b-phosphate is in close contact to S105. The c-phosphate, an important leaving group from ATP, makes hydrogenbonding with the R127 and D243.

Catalytic core of the proteinCatalytic core of the protein formed by the several residues contrib-uted by both domains and inter-domain interaction seems to benecessary for enzyme activity. Catalytic core of the majority of kin-ases is made up of 11 conserved subdomains (37). Similar to Af-RIO-2, kinase domain in PFD0975w is also shorter than typicalprotein kinases present in protein kinase C (PKC) and protein kinaseA (PKA). It lacks critical activation loop (subdomain VIII) and helicespresent in subdomain X XI, and this region is responsible for sub-strate specificity and binding (22).

PfRIO-2 is a suitable candidate for drugdevelopmentProtein kinases present within parasite kinome are suitable targetfor drug development (38,39), and in comparison to other kinase(s)present within plasmodium kinome, PFD0975w offers a number ofadvantages as drug target. (i) The PFD0975w kinase domain is struc-turally very different from the classical eukaryotic kinase(s). (ii) ATPbinding pocket lacks critical residues present in eukaryotic kinases,(iii) There is no defined substrate binding region. Apart from thesedifferences from the classical eukaryotic kinase(s), we have also

explored how much it is different at sequence and structure levelfrom human RIO-2. Both kinases are very much similar at sequence(similarity 71.73%) or structural level (RMSD 0.64 �) except subtledifferences in winged helix domain and catalytic core of the protein.In PFD0975w, ATP binding pocket is more well defined to harbor themolecule, especially c-phosphate, and it is more compact. In humanRIO-2, it is wide open and not well defined (Figure 4A,C). These dif-ferences also reflect on their interaction within residues of bindingpockets, and PFD0975w-ATP makes several strong interactions withthe ATP atoms, whereas interactions present in the huRIO-2-ATPstructure is much weaker (Figure 4B,D and Table 1). As ATP-boundcomplexes for this analysis has been generated by taking ATP fromtemplate (1ZAO), in many instances, structure biasness might mis-guide interpretations. To rule out such a possibility, we have mod-eled ATP-bound PFD0975w and HuRIO-2 structures by docking ATPinto their respective native model using AUTODOCK4.1. In the ATP-bound model of PFD0975w (PFD0975w-ATP) or HuRIO-2 (HuRIO-2-ATP), ATP is present in the similar conformation, and there are nosignificant differences from earlier observations.

Substrate binding sitesAs compared to typical protein kinase that has well-defined peptidebinding site, RIO kinases lack these structural features, and it ishypothesized that protein surface might serve as supporting plat-form for the substrate protein docking, but there is no idea aboutthe region of the protein involved in substrate binding ⁄ interactions.The substrate binding ⁄ interaction region is an important factor todesign inhibitors against this class of enzymes. To address thisquestion, we started looking for available free space (hot spot)within the protein structure by performing a generic search for cavi-ties using Q-site and POCKET FINDER (32). Results were encouraging,and we could be able to locate a number of potential binding siteswithin PFD0975w (Figure 5A). Detailed properties of different sitesare given in Table 2. A number of sites are present near to theunstructured loop region; these sites can easily be able to accom-modate substrate, especially when this part of PFD0975w will haveso much flexibility and mobility in the structure. These sites or com-bination of sites could be potentially be used by PFD0975w to bindthe protein substrate. As an example of how a peptide substratecould accommodate within these sites, we have chosen two sites(site 1 & 2, Figure 5A) and analyzed their probability to bind sub-strate and drive phosphate transfer from ATP present in the activesite. These two sites are in close proximity to each other and areconnected with each other by a tunnel. Both sites are lined by anumber of residues that provide suitable environment to enablebinding of a variety of peptide substrates (Figure 5B,C). A peptidebinding in either site can be able to reach close to c-phosphate ofATP through the tunnel. It is difficult to say whether only these twosites are only involved or other unknown regions are also requiredfor catalysis. Moreover, previous structural studies in AfRIO-2 indi-cate a specificity of enzyme toward serine residues, and a largesurface area of the enzyme is responsible for substrate binding andspecificity. To address this aspect, we put efforts to model apeptide within these predicted sites. RIO-2 kinase exhibits kinaseactivity against a number of protein substrates under in vitro and invivo conditions (31). Considering this fact, we have used two differ-ent kinds of peptides, a polyalanine peptide and sequence-specific

Table 1: Interaction of the ATP within the molecular model ofPFD0975w and HuRIO-2

S. NoATPatom

PFD0975w HuRIO-2

Proteinresidue Atom Distance

Proteinresidue Atom Distance

1 N6 M186 SD 3.23 E189 O 2.812 S187 O 2.95 M188 SD 3.363 N1 I189 O 3.31 I191 N 2.934 C2 I189 O 2.68 L190 CD1 3.835 C1 I99 CD1 3.84 Absent6 O4¢ I99 CD1 2.82 Absent7 O1G E104 OE1 2.75 Absent8 R127 NH1 2.31 E106 OE1 3.779 O2G E104 OE1 2.24 D246 OD1 2.37

10 D243 O2D 2.44 E106 OE1 3.3611 O1B E104 OE1 3.18 D246 OD2 3.1212 PG E104 OE1 2.81 Absent13 O1 E104 OE2 2.85 Absent14 O1A K121 NZ 3.07 K123 NZ 2.7215 O3G K121 NZ 2.36 D246 OD1 3.1916 O3A K121 NZ 3.18 D246 OD1 3.4617 O3B S105 OG 3.21 Absent18 O1B D243 OD2 2.97 D246 OD2 3.1219 Absent E106 OE1 3.6220 OE2 3.5821 O2A Absent N233 OD1 3.28

PFD0975w and Human RIO-2 structure were build as described in Methodsection and then compared by taking ATP from ATP-bound form of AfRIO-2(PDB code 1ZAO). Both structures were identical except local changes withinATP binding pocket.

Trivedi and Nag


peptide2 (TTYADFIASGRTGRRNAIHD), to probe the non-specific andspecific binding pockets. PFD0975w ⁄ polyalanine model binds poly-alanine peptide within the space cover by site no 6, 8, 10 (Fig-ure 5A) and forms a number of interactions with protein residues(Figure 6A). N-terminal of the peptide is stabilized by ionic interac-tions with Y75, K103, R124, and R137, whereas middle segment ismaking extensive hydrophobic interactions with L125, N135, N147,W148, and polar contacts with Y31 and K142, and the other end ofpolyalanine peptide is in contact with Y58, K66 (Figure 6C).

Interestingly, sequence-specific peptide2 fit well to two pockets. Thegeometric shape complementarity score of peptide for first pocketand second pocket was 10204 and 10122, respectively. First pocketis present near the unstructured region and exhibits a similar kindof interactions with the protein residues as polyalanine peptide.Second binding site is closer to ATP binding region (site 1 and 2,Figure 5A), and a number of interactions with active site residuesstabilize its conformation (Figure 6B). N-terminal of peptide (T5-A12)is stabilized by K103, R127, N136, and R261, whereas middle seg-ment (S13-R19) is in contact with H196, F229, N227, and D243, andC-terminal end (N20-D24) of peptide is in contact with G191 andY192. RIO kinases are serine–threonine class of kinases, and inter-estingly we found that Ser 13 of the peptide2 is in close proximity

(2.7 �) to c-phosphate of bound ATP, and geometry is very favorablefor phosphate transfer (Figure 6D). Control docking of peptide2 in1APM using patchdock gives a geometric shape complementaryscore of 9736, and peptide2 binds in the extended conformation asgiven in co-crystal structure. Detailed biochemical experimentation isrequired to resolve how substrate binds to PDF0975w, but explora-tion of these sites might help in narrowing down the searches.

Discussion

Atypical kinases present within kinome of an organism are knownto regulate various biologically important processes through phos-phorylation and are potential targets for drug development (18). Inplasmodium, at least two atypical kinases are present whose func-tion is currently unknown. In this study, we attempted to performan in-depth structural characterization of PFD0975w through build-ing a modeled structure based on the crystal structure of RIO-2from A. fulgidus (24). In multiple sequence alignment, we identifiedthe presence of HWH domain in PFD0975w, a characteristic domainpresent in RIO-2 kinase family (Figure 1). The partial modeledstructure (1–300) in this study is enough to characterize structuralfeatures required for phosphorylation mediated by PFD0975w.

A B

C D

Figure 4: Surface structure of ATP binding pocket and interaction of ATP within modeled structure of PFD0975w-ATP and HuRIO-2-ATP. (Aand C) Surface structure of the ATP binding pocket of PFD0975w-ATP and HuRIO-2-ATP, Blue is positive, red is negative, and ATP is displayedas in ball-stick model. (B and D) Interaction of ATP with key residues present within the binding pocket of PFD0975w-ATP and HuRIO-2-ATP. Ahydrogen-bonding interaction exists between ATP and Asn 233 of HuRIO-2-ATP, absent in PFD0975w-ATP structure.



A B

C

Figure 5: Potential ligand binding site within PFD0975w. (A) Structure of PFD0975w in wire frame with the potential cavities as space-filled model. All the cavities were numbered (1–10) for better understanding. (B) Zoomed view of site 1 and 2 on surface of PFD0975w. (C)Communication mode of site 2 with ATP present at the active site through the tunnel. In B and C, suitable size objects in site 1 (Red color)and site 2 (salmon) are placed to show site 1 and 2 communication to allow bound substrate (peptide fragment) to access c-phosphate ofATP present within the active site.

Table 2: Properties of potential peptide binding pockets present within PFD0975w

BindingSite Protein residues surrounding site Vol (�3)

�Capacity toaccommodateNo. Of Ala residues

1 I99, I107, C119, K121, E159, L163, P174, M186, S187, Y188, I189, G191, P193, L232, I241, I242, D243, F244 791 92 E104, H123, R124, L125, G126, R127, I128, L151, I154, A155, K158, P245, Q246, I247, V248, Y258 603 73 K158, E159, A161, Y162, V165, A219, D220, I221, Q246, I247, V248, S249, H252 351 44 Y203, I206, D207, F210, Q279, L283, Y284, E285, K292, D293, N296 324 45 A217, K218, D220, L250, R251, A255, K256, F259, L289, K294, I297, D298, N300 191 2.06 N264, I267, N268, F271, I276, K277, E280, E280, Y284, E285, D286, V287, N291 92 17 M26, R27, N28, H29, E30, F145, R146, N147, Y150 93 18 M1, K2, L3, D4, G96, N97, Q98, Y108 72 �19 E30, K134, N134, N136, D138, Y139, K142, K143, C144, F145, R146, N147 66 �110 L3, D4, F8, L83, L87, L92, K93, S94, I95 56 �1

Binding sites were identified using qsite ⁄ POCKET FINDER. The important conserved residues known in catalysis are underlined for each binding site.

Trivedi and Nag


RIO kinases show a very low level of kinase activity against mostof the known protein substrate in in vitro kinase assay (36). YeastRIO-1 mutant where C-terminal 80 amino acids were deletedshows an enhanced kinase activity, but this mutation is lethal inyeast (22). The RIO kinases present in eukaryotes (and PFD0975w)are longer than archaea, and it is interesting to explore a compar-ative study from different organisms to understand how this regionregulates protein kinase activity. The extra longer portion ofPFD0975w (288 amino acid beyond 300) for which we were failedto build a model because of low structural similarity to otherstructure present in the database might be important for regulat-ing kinase activity. The comparative study also might give anopportunity for drug development exploiting this futile region ofthe protein.

The N-terminal wHWH domain is present in transcription factor andother DNA binding proteins (40,41). In vitro assay with AfRIO-2

exhibits non-specific oligonucleotide binding activity, but it is notclear whether this domain plays any role in kinase activity of aRIO-2 kinase. In PFD0975w structure, this domain is in constantinteraction with the kinase domain and contributes a number ofimportant residues for catalysis; moreover, subtle changes in wHWHdomain affect catalytic kinase domain as well. It is interesting toexplore what type of conformational changes a DNA or RNA bind-ing induces to the PFD0975w structure and how these changesaffects its protein kinase activity. RIO-2 kinases are known to playan important role in ribosome maturation pathway, and it is quitepossible that these interactions (DNA ⁄ RNA-wHWH binding) willenhance RIO-2 kinase activity toward protein involved in ribosomematuration. Exactly what mechanism is involved to activate RIO-2kinase is currently unknown, but presence of autophosphorylationactivity in the presence of ATP and Mg2+ only gives indications thatthe substrate binding may be responsible for the activation ofkinase activity.

A B

C D

Figure 6: Molecular model of PFD0975w ⁄ peptide complex. (A) PFD0975w ⁄ polyalanine complex mimics non-specific proteinous substrates,whereas (B) PFD0975w ⁄ peptide2 represent specific substrates. Peptide2 binds to two sites, a unstructured loop site and another site close toATP binding pocket. Both complexes exhibit binding of peptide in an extended conformation to the large surface area of binding pockets andshow extensive interactions with the PFD0975w residues. (C) Interaction of the polyalanine peptide with the residues present within bindingsite, (D) Interaction of peptide2 with the residues present within the binding pocket close to the ATP binding site.



In typical kinase, a well-defined region is present to assist peptidebinding, and a number of residues make hydrogen bonding with thepeptide substrate; these regions are unknown in atypical RIO-2 kinas-es. Substrate binding to RIO-2 is still at the stage of speculation, anda number of theories put forward to explain how peptide substratecould bind to RIO-2 kinase (42). Presence of conserved residues onthe protein surface in the crystal structure of AfRIO-2 indicates thatthese surfaces could serve as a platform to bind proteins, but it wasfailed to explain how these peptides will access ATP c-phosphatethat is buried deep into the active site. In this study, we have identi-fied many cavities present in the PFD0975w, as it is quite possiblethat one or more cavities might give way to bound protein to accessthe active site for phosphate transfer (Figure 5). In a proposed modelfor substrate binding to PFD0975w, conserved charged surface resi-dues will bring the target protein and provide a large surface forinteraction, and then peptide fragment of target protein will utilizecavities and could be able to reach ATP for phosphate transfer (Fig-ure 5C). A PFD0975w ⁄ polyalanine and PFD0975w ⁄ peptide2 modelswere generated to test whether these identified pockets have dock-ing sites for peptide chains. Two different kinds of peptides weretaken to represent both specific and non-specific substrates. In boththe modeled structures, peptide docks very nicely into the pocket andmakes extensive interaction with the protein residues (Figure 6).Many of the interacting residues are reported in a homologous struc-ture important for autophosphorylation and do not rule out the possi-bility of contributing for substrate binding (31). Although a number ofevidences support the output of current study, rigorous validationthrough biochemical experimentation is essential to confirm the cur-rent findings, which is beyond the scope of the current work.

Acknowledgments

This work was supported by the Department of Biotechnology; Govtof India Grants (BT ⁄ PR13436 ⁄ MED ⁄ 12 ⁄ 450 ⁄ 2009 &BT ⁄ 41 ⁄ NE ⁄ TBP ⁄ 2010) to V.T. Both authors acknowledge the infra-structure facility provided by the Department of Biotechnology,Indian Institute of Technology-Guwahati, Assam, India.

References

1. Snow R.W., Guerra C.A., Noor A.M., Myint H.Y., Hay S.I. (2005)The global distribution of clinical episodes of Plasmodium falci-parum malaria. Nature;434:214–217.

2. Francis S.E., Gluzman I.Y., Oksman A., Knickerbocker A., MuellerR., Bryant M.L., Sherman D.R., Russell D.G., Goldberg D.E (1994)Molecular characterization and inhibition of a Plasmodium falci-parum aspartic hemoglobinase. EMBO J;13:306–317.

3. Gluzman I.Y., Francis S.E., Oksman A., Smith C.E., Duffin K.L., Gold-berg D.E. (1994) Order and specificity of the Plasmodium falcipa-rum hemoglobin degradation pathway. J Clin Invest;93:1602–1608.

4. Keaney J.F. Jr, Puyana J.C., Francis S., Loscalzo J.F., StamlerJ.S., Loscalzo J. (1994) Methylene blue reverses endotoxin-induced hypotension. Circ Res;74:1121–1125.

5. Guha M., Maity P., Choubey V., Mitra K., Reiter R.J., Bandyopad-hyay U. (2007) Melatonin inhibits free radical-mediated

mitochondrial-dependent hepatocyte apoptosis and liver damageinduced during malarial infection. J Pineal Res;43:372–381.

6. Becker K., Tilley L., Vennerstrom J.L., Roberts D., Rogerson S., Gins-burg H. (2004) Oxidative stress in malaria parasite-infected erythro-cytes: host–parasite interactions. Int J Parasitol;34:163–189.

7. Gromer S., Urig S., Becker K. (2004) The thioredoxin system –from science to clinic. Med Res Rev;24:40–89.

8. Kumar S., Bandyopadhyay U. (2005) Free heme toxicity and itsdetoxification systems in human. Toxicol Lett;157:175–188.

9. Trivedi V., Chand P., Srivastava K., Puri S.K., Maulik P.R., Bandyopad-hyay U. (2005) Clotrimazole inhibits hemoperoxidase of Plasmodiumfalciparum and induces oxidative stress. Proposed antimalarialmechanism of clotrimazole. J Biol Chem;280:41129–41136.

10. Dubois V., Couissi D., Schonne E., Remacle C., Trouet A. (1995)Intracellular levels and secretion of insulin-like-growth-factor-binding proteins in MCF-7 ⁄ 6, MCF-7 ⁄ AZ and MDA-MB-231breast cancer cells. Differential modulation by estrogens inserum-free medium. Eur J Biochem;232:47–53.

11. Meierjohann S., Walter R.D., Muller S. (2002) Regulation ofintracellular glutathione levels in erythrocytes infected with chlo-roquine-sensitive and chloroquine-resistant Plasmodium falcipa-rum. Biochem J;368:761–768.

12. Trivedi V., Srivastava K., Puri S.K., Maulik P.R., Bandyopadhyay U.(2005) Purification and biochemical characterization of a hemecontaining peroxidase from the human parasite P. falciparum.Protein Expr Purif;41:154–161.

13. Bozdech Z., Llinas M., Pulliam B.L., Wong E.D., Zhu J., DeRisiJ.L. (2003) The transcriptome of the intraerythrocytic develop-mental cycle of Plasmodium falciparum. PLoS Biol;1:E5.

14. Bozdech Z., Zhu J., Joachimiak M.P., Cohen F.E., Pulliam B., De-Risi J.L. (2003) Expression profiling of the schizont and trophozo-ite stages of Plasmodium falciparum with a long-oligonucleotidemicroarray. Genome Biol;4:R9.

15. Ginsburg H., Stein W.D. (2004) The new permeability pathwaysinduced by the malaria parasite in the membrane of the infectederythrocyte: comparison of results using different experimentaltechniques. J Membr Biol;197:113–134.

16. Perez-Rosado J., Gervais G.W., Ferrer-Rodriguez I., Peters W.,Serrano A.E. (2002) Plasmodium berghei: analysis of the gamma-glutamylcysteine synthetase gene in drug-resistant lines. ExpParasitol;101:175–182.

17. Ward P., Equinet L., Packer J., Doerig C. (2004) Protein kinasesof the human malaria parasite Plasmodium falciparum: the ki-nome of a divergent eukaryote. BMC Genomics;5:79.

18. Manning G., Plowman G.D., Hunter T., Sudarsanam S. (2002)Evolution of protein kinase signaling from yeast to man. TrendsBiochem Sci;27:514–520.

19. Campbell B.E., Boag P.R., Hofmann A., Cantacessi C., Wang C.K.,Taylor P. et al. (2011) Atypical (RIO) protein kinases from Hae-monchus contortus – promise as new targets for nematocidaldrugs. Biotechnol Adv;29:338–350.

20. Campbell M.G., Karbstein K. (2011) Protein–protein interactionswithin late pre-40S ribosomes. PLoS One;6:e16194.

21. Manning G., Whyte D.B., Martinez R., Hunter T., Sudarsanam S.(2002) The protein kinase complement of the human genome.Science;298:1912–1934.

22. Angermayr M., Roidl A., Bandlow W. (2002) Yeast Rio1p is thefounding member of a novel subfamily of protein serine kinases

Trivedi and Nag


involved in the control of cell cycle progression. Mol Micro-biol;44:309–324.

23. Vanrobays E., Gleizes P.E., Bousquet-Antonelli C., Noaillac-Dep-eyre J., Caizergues-Ferrer M., Gelugne J.P. (2001) Processing of20S pre-rRNA to 18S ribosomal RNA in yeast requires Rrp10p,an essential non-ribosomal cytoplasmic protein. EMBOJ;20:4204–4213.

24. LaRonde-LeBlanc N., Wlodawer A. (2004) Crystal structure ofA. fulgidus Rio2 defines a new family of serine protein kinases.Structure;12:1585–1594.

25. LaRonde-LeBlanc N., Wlodawer A. (2005) The RIO kinases: anatypical protein kinase family required for ribosome biogenesisand cell cycle progression. Biochim Biophys Acta;1754:14–24.

26. Larkin M.A., Blackshields G., Brown N.P., Chenna R., McGettiganP.A., McWilliam H. et al. (2007) Clustal W and Clustal X version2.0. Bioinformatics;23:2947–2948.

27. Laskowski R.A., Moss D.S., Thornton J.M. (1993) Main-chainbond lengths and bond angles in protein structures. J MolBiol;231:1049–1067.

28. Guex N., Peitsch M.C. (1997) SWISS-MODEL and the Swiss-Pdb-Viewer: an environment for comparative protein modeling. Elec-trophoresis;18:2714–2723.

29. Colovos C., Yeates T.O. (1993) Verification of protein structures:patterns of nonbonded atomic interactions. Protein Sci;2:1511–1519.

30. Ramachandran G.N., Ramakrishnan C., Sasisekharan V. (1963)Stereochemistry of polypeptide chain configurations. J MolBiol;7:95–99.

31. LaRonde-LeBlanc N., Guszczynski T., Copeland T., Wlodawer A.(2005) Autophosphorylation of Archaeoglobus fulgidus Rio2 andcrystal structures of its nucleotide-metal ion complexes. FEBSJ;272:2800–2810.

32. Laurie A.T., Jackson R.M. (2005) Q-SiteFinder: an energy-basedmethod for the prediction of protein-ligand binding sites. Bioin-formatics;21:1908–1916.

33. Schames J.R., Henchman R.H., Siegel J.S., Sotriffer C.A., Ni H.,McCammon J.A. (2004) Discovery of a novel binding trench inHIV integrase. J Med Chem;47:1879–1881.

34. Schneidman-Duhovny D., Inbar Y., Nussinov R., Wolfson H.J.(2005) PatchDock and SymmDock: servers for rigid and symmet-ric docking. Nucleic Acids Res;33:W363–W367.

35. Andrusier N., Nussinov R., Wolfson H.J. (2007) FireDock: fastinteraction refinement in molecular docking. Proteins;69:139–159.

36. Laronde-Leblanc N., Guszczynski T., Copeland T., Wlodawer A.(2005) Structure and activity of the atypical serine kinase Rio1.FEBS J;272:3698–3713.

37. Knighton D.R., Zheng J.H., Ten Eyck L.F., Xuong N.H., Taylor S.S.,Sowadski J.M. (1991) Structure of a peptide inhibitor bound tothe catalytic subunit of cyclic adenosine monophosphate-depen-dent protein kinase. Science;253:414–420.

38. Doerig C. (2004) Protein kinases as targets for anti-parasitic che-motherapy. Biochim Biophys Acta;1697:155–168.

39. Doerig C., Meijer L. (2007) Antimalarial drug discovery: targetingprotein kinases. Expert Opin Ther Targets;11:279–290.

40. Dong G., Chakshusmathi G., Wolin S.L., Reinisch K.M. (2004)Structure of the La motif: a winged helix domain mediatesRNA binding via a conserved aromatic patch. EMBO J;23:1000–1007.

41. Wolberger C., Campbell R. (2000) New perch for the wingedhelix. Nat Struct Biol;7:261–262.

42. LaRonde-LeBlanc N., Wlodawer A. (2005) A family portrait ofthe RIO kinases. J Biol Chem;280:37297–37300.

Supporting Information

Additional Supporting Information may be found in the onlineversion of this article:

Table S1. Quality of modeled PFD0975w and human RIO-2structures.

Please note: Wiley-Blackwell is not responsible for the content orfunctionality of any supporting materials supplied by the authors.Any queries (other than missing material) should be directed to thecorresponding author for the article.



Documents

In Silico Characterization of Atypical Kinase PFD0975w from Plasmodium Kinome: A Suitable Target For Drug Discovery