Graphics & Design

A model for the interaction between plant GAPN and 14-3-3zeta using protein-protein docking calculations, electrostatic potentials and kinetics

Phosphorylated non-phosphorylating glyceraldehyde-3-phosphate dehydrogenase (EC; GAPN) found in heterotrophic cells of wheat is activated by MgCl(2). The divalent cation disrupts the interaction between GAPN and a 14-3-3 regulatory protein.
of 13
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
  A model for the interaction between plant GAPN and14-3-3 z  using protein–protein docking calculations,electrostatic potentials and kinetics Diego M. Bustos a , Alberto A. Iglesias b, * a  Instituto Tecnolo´ gico de Chascomu´ s (IIB-INTECH), Camino Circunv, Laguna km 6, CC 164, B7130IWA Chascomu´ s, Argentina b  Laboratorio de Enzimologı´ a Molecular, Bioquı´ mica Ba´ sica de Macromole´ culas, Facultad de Bioquı´ mica y Ciencias Biolo´ gicas,UNL, Paraje ‘‘El Pozo’’, CC 242, S3000ZAA Santa Fe, Argentina Received 17 June 2004; received in revised form 25 March 2005; accepted 29 March 2005Available online 17 May 2005 Abstract Phosphorylated non-phosphorylating glyceraldehyde-3-phosphate dehydrogenase (EC; GAPN) found in heterotrophic cells of wheat is activated by MgCl 2 . The divalent cation disrupts the interaction between GAPN and a 14-3-3 regulatory protein. This effect is quiteremarkable,sinceithaspreviouslybeenshownthat14-3-3bindingtoatargetproteinrequiresdivalentcationsasMg 2+ orCa 2+ .Bindingofthedivalent cation to 14-3-3 causes an increase in surface hydrophobicity. Crystal structure of a 14-3-3-target protein complex has been onlydetermined for serotinin  N  -acetyltransferase. We utilized a model of a subunit of plant GAPN and the crystallographic structure of human 14-3-3 z  to shape the complex between theses two proteins. Initial dockings were performed with the BiGGER program, which allows anexhaustive search of translational and rotational space. A filtering procedure was then applied to reduce the number of complexes to amanageable number. We predict the structural characteristics of GAPN–14-3-3 z  binding process, proposing that the main attractive force inthis complex derives from electrostatic interactions. The predicted model was corroborated by analysis of kinetic behavior of GAPN and itsrelationship with pH and ionic strength conditions. This study provides a variant on the interaction of 14-3-3 with target proteins, thusaffording a wider scenario to establish possible structural models for this remarkable family of regulatory proteins. # 2005 Elsevier Inc. All rights reserved. Keywords:  Glyceraldehyde-3-phosphate; Non-phosphorylating; GAPN; 14-3-3 Proteins; Protein–protein docking 1. Introduction Non-phosphorylating glyceraldehyde-3-phosphate dehy-drogenase (EC; GAPN) has been identified as amember of the aldehyde dehydrogenase superfamily [1].The enzyme catalyzes the irreversible, NADP + -dependent,oxidation of glyceraldehyde-3-phosphate (Ga3P) into 3-phosphoglycerate (3-PGA) [2,3]. GAPN has been found ingreen algae, higher plants [1,4], and some specializedbacteria [1,5].Inphotosynthetic cells theenzyme isinvolvedin a shuttle system exporting photogenerated NADPH fromthe chloroplast, and thus it plays a key role in the supply of reducing power to the cytosol [3]. In plants, GAPN is alsofound in non-photosynthetic tissues, such as endosperm,shoots, cotyledons, and roots [4,6,7]. The function played bythis enzyme in heterotrophic plant cells is less clear. It hasbeen proposed that in non-green cells GAPN could coupleNADPH production needed for anabolic reactions toglycolysis [1].GAPN activity is carried out by a catalytic domain in theenzymewhosestructureand specific aminoacidresidues arehighly or absolutely conserved in the aldehyde dehydro-genase superfamily [1]. This characteristic structure hasbeen found to involve three subdomains with highlyconserved sequences [7]. Crystallographic structures andkinetic studies of GAPN from the bacterium  Streptococcusmutans  (GAPN sm ) have enabled the determination of theenzyme catalyticmechanism [8].Itstartswiththe binding of of Molecular Graphics and Modelling 23 (2005) 490–502* Corresponding author. Tel.: +54 342457 5216x217;fax: +54 342457 5221. E-mail address: (A.A. Iglesias).1093-3263/$ – see front matter # 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.jmgm.2005.03.002  NADP + toapo-GAPN,whichinducesalocalconformationalchange of the active site [8], with at least a reorientation of side chains of amino acids Cys 302 and Glu 268 . Four otherresidues found conserved in the GAPN family (Arg 124 ,Tyr 170 , Arg 301 , and Arg 459 ) seem to be involved in therecognition of the other substrate,  D -Ga3P, as well as incatalysis [2,3].GAPN from wheat endosperm and shoots requires Mg 2+ to achieve maximal activity [9]. The latter could beassociated with recent  fi ndings from our laboratory on thephosphorylation of GAPN from plant heterotrophic tissues[7] and its interaction with 14-3-3 regulatory proteins [9]. Data derived from size exclusion chromatography and co-immunoprecipitation demonstrated that Mg 2+ exerts thedisruption of a protein – protein interaction between native,phosphorylated GAPN and a 14-3-3 protein [9]. These novel fi ndings showing that GAPN activity is under control bypost-translational modi fi cation, dark-to-light variation of Mg 2+ concentration [10], and interaction with regulatory proteins, open a new scenario on the way of howcarbohydrates can be utilized to produce ATP or NADPHin the cytosol of non-green cells from plants [7,9,11].One hallmark of signal transduction events is thephosphorylation-induced transition of a protein from oneactive state to another [12]. However, it is becoming clear that, in many cases, phosphorylation by itself is notsuf  fi cient to induce changes in activity. Rather, it is thephosphorylation-induced association with 14-3-3 proteinsthat results in the transition to changes in activity [12]. Inplants, 14-3-3 proteins were found to regulate the activity of enzymes involved in primary carbon metabolism, nitrogenassimilation and generation of protons gradient [13].Crystal structuresofboth t and z isoforms of14-3-3showthat they are highly helical, dimeric proteins [14,15]. Eachmonomer is composed of nine anti-parallel  a -helices,organized into an N-terminal and a C-terminal domain. Thedimer forms a large, negatively charged channel [16]. The14-3-3 proteins bind to targets containing the consensussequence RXXpS/pTXP (named mode 1) or RXXXpS/ pTXP (mode 2), with the coordination of pSer by threeabsolutely conserved amino acid residues, Lys 49 , Arg 56 andArg 127 , as srcinally described [17,18]. However, phosphor-ylation of the target protein is not the only factor controllingthe binding of 14-3-3. Although hydrophobic interactionscould be a main force determining 14-3-3-target interactions[19], electrostatic attraction has long been recognized as akey driving force for fast protein – protein associations [20].The interaction of 14-3-3s with their targets speci fi callyrequires millimolar concentrations of a divalent cation [21]suchasMg 2+ ,Ca 2+ orMn 2+ .Thestructuralmodelshowsthatbinding of the divalent cation to the 14-3-3 causes anincrease in surface hydrophobicity, as was monitored usingthe  fl uorescent probe bis-ANS [4,4 0 -bis(1-anilinonapthalene8-sulfonate) [19]. Interestingly, the interaction of plantphosphorylated GAPN with 14-3-3 was found to be releasedby Mg 2+ , thus constituting a distinctive example respect tothe way by which the regulatory protein binds to a targetenzyme [9]. The prediction of a structural model for the complex formed between GAPN and 14-3-3 is thusparticularly important to understand the odd behaviour of the divalent cation in this protein – protein interaction.The 14-3-3-af  fi nity puri fi cation of more than 200 humanphosphoproteins [22] reveals that 14-3-3 interacts withseveral cellular proteins to regulate metabolism, prolifera-tion and traf  fi cking. From this, it is clear that even whenautomated crystallography techniques were utilized, theelucidationofthe fi nestructureofallthecomplexesbetween14-3-3 and target proteins is far from being available. Underthis scenario, the developing of new in silico structuralapproaches become practically relevant. Docking algo-rithms are currently being developed for the correct analysisof protein – ligand complexes. However, to predict protein – protein interactions is more dif  fi cult, requiring of other datato select for the best model. In the present work, we utilizedan in silico structural approach to study the plantphosphorylated GAPN – 14-3-3 z  complex using proteindocking calculations, biochemical data  fi ltering and kineticand hysteretic behaviour experiments. The derived model iscompared with those previously developed, to understandthe distinctive effect of the divalent cation in the interactionbetween GAPN and 14-3-3. 2. Material and methods 2.1. Multiple sequence alignment  The sequences of GAPN monomers were aligned usingthe DIALIGN multiple sequence alignment algorithm [23],that includes the following steps: (1) pairwise alignmentusing Needleman and Wunsch algorithm modi fi ed to usezero end gap penalties; (2) evolutionary tree constructionwith a  ‘‘ neighbour-joining algorithm ’’ , to determine theorder of alignment and to calculate relative weight of sequences and pro fi les from the branch lengths; and (3)traverse the tree from top to bottom, aligning the closestsequences or pro fi les. A multi-protein sequence alignmentof GAPN monomers was generated automatically by theDIALIGN [23] method and then it was manually adjustedaround residues 240 – 255 (in plant GAPN), in order tocorrectly align the motif GxxxG/A between prokaryotic andplant GAPNs. 2.2. Plant GAPN and its interaction with 14-3-3:sequences and 3D model building Three-dimensional models (GAPNpls) of   fi ve knownplant GAPNs were constructed using crystal structurecoordinates of GAPN sm  (PDBcode: 1EUH) as template andModeller v6.0 program [24]. Sequences of plant GAPNswere obtained from the Genebank. Modeller program iscompletely automated and capable of generating energy  D.M. Bustos, A.A. Iglesias/Journal of Molecular Graphics and Modelling 23 (2005) 490  –  502  491  minimized protein models by satisfying restraints on bonddistances and dihedral angles. The identity percentagebetween models and template sequences is 52%, which is ingood agreement with Modeller requirements. GAPNplswere based in the protein at the unphosphorylated state, andits evaluation was conducted using the WHATIF program[25].Protein coordinates for human 14-3-3 z  were obtainedfrom the Protein Databank (PDB code 1A4O, subunit A andB). In this structure, monomeric 14-3-3 contains nine anti-parallel  a -helices arranged into an N-terminal and a C-terminal domain. Although it has been pointed out theexistence of a tenth  a -helix in plant 14-3-3 (unsolved in thehuman isoform, see reference [26]), it would localize faraway from the target binding site, and consequently playingno relevant role in the proposed model.All GAPN residues within 8 A ˚  of the putative 14-3-3binding site ( 401 RINSVEE) differing from the GAPNtemplate residue were energy-minimized by AMBER force fi eld [27], implemented on HyperChem 5.0, according to thefollowing procedure. Coordinates of the backbone atoms inplant GAPN were set to the values of the correspondingatoms in GAPN sm . Similarly, coordinates of side chainatoms of the conserved residues were set to the values of thecorresponding coordinates of GAPN sm  atoms as well. Incontrast, side chains of no conserved residues were set in anextended conformation. The resulting structure of plantGAPN was energy minimized in vacuo using a distance-dependent dielectric constant, and equal electrostatic andvan der Waals forces. Minimization was carried out in onestep, where the residues were completely relaxed. Theresulting structure exhibited a root mean square deviation(rmsd)minorto1.6 A ˚  fromthestartingX-ray structurewhenall the backbone atoms were compared, which represents areasonable value. 2.3. Protein docking Molecular interaction simulations were performed usingthe docking program BiGGER [28]. Although this programonly succeeded in modelling one of the  fi ve criticalassessment of prediction of interactions (CAPRI) targets,these results seem to be more due to the dif  fi culty of the task than to shortcomings in the method, and BiGGER couldsuggest useful models in a setting where some experimentaldata were available, which is our case. This algorithmperforms a complete and systematic search in the bindingspace of both molecules. The surface of each partner isrepresented as a binary (0, 1) grid at 1 A ˚  resolution, andsurface contact is estimated by counting the superposition of surface cells (those marked with 1 in each grid). The coreregion of each partner is represented as another binarymatrix, and superposition of these core regions indicatesexcessive overlap and is forbidden. A population of 1000candidate protein – protein-docked geometries is generatedand selected, based on the geometric complementarities andamino acid pairwise af  fi nities between the two molecularsurfaces. In this process, the algorithm enables implicittreatment of molecular  fl exibility.A candidate model is rejected if it does not have a highersurface contact score than the lowest ranking of the modelsretained at that point. If it has a higher score, it replaces thelowest ranking model if it passes the second  fi lter, which is aneuralnetworkthatevaluatestheaminoacidcontactsbetweenthepartners.Thisnetworkwastrainedwithalargenumberof examples to distinguish between correct and incorrectcomplexes. In a subsequent step, putative docked structuresare ranked using an interaction scoring function, whichcombines several interaction terms thought to be relevant forthe stabilization of protein complexes. The latter includes:geometric packing of the surfaces, explicit electrostaticinteractions, desolvation energy, and pairwise propensities of the amino acid side chains to contact across the molecularinterface. In the ab initio simulations, the entire molecularsurface was searched using absolutely no additionalinformation regarding the binding sites. After this  fi rst step,abiochemical fi lterwasutilizedinordertoreducethequantityof complexes to a manageable number. Speci fi cally, theamino acid residues Ser 404 from GAPN and Lys 49 , Arg 56 andArg 127 from 14-3-3 protein were selected and distancesbetween these moieties sorted all complexes.The peptide ARAApSAPA (designed by Carol MacK-intosh, University of Dundee, UK) was docked usingGRAMM v1.03 program [29] that may be used for a proteinand for a smaller compound as well. We performed thisdocking because the peptide was previously shown tointeractwith14-3-3competitivelywithGAPN[9]andthusitafford useful information on the region in 14-3-3 where theinteraction with plant GAPN takes place. The followingparameters were utilized: atching mode (generic/helix) = -generic; grid step = 1.7; repulsion (attraction is always  1) = 5.; attraction double range (fraction of singlerange) = 0.; potential range type (atom_radius, grid_ste-p) = atom_radius; projection (blackwhite, grey) = black-white; representation (all, hydrophobic) = all; number of matches to output = 1000; angle for rotations, (10 8 ,12 8 , 15 8 ,18 8 , 20 8 , 30 8 , 0 8 — no rot.) = 10. 2.4. Electrostatic potentials Electrostatic potentials were obtained from Swiss pdbviewer v3.7 program. The solvent was described in terms of a bulk dielectric constants (=80), whereas GAPN and 14-3-3were described in terms of the coordinates of individualatoms as well as atomic partial charges and dielectricconstant (protein) = 4.0. Computational methods werePoisson – Boltzmann equations; and solvent ionicstrength = 0.0 mol/l. Coordinates for the experimentallydetermined 14-3-3 z  without any divalent cation andmodelled GAPN structure were used in the calculations.No difference was observed between the structures of thepeptide bound and peptide free states of 14-3-3 [30],  D.M. Bustos, A.A. Iglesias/Journal of Molecular Graphics and Modelling 23 (2005) 490  –  502 492  suggesting that no conformational change is induced bybindingoftheligandto14-3-3.Indeed,thesecrystallographicstructures were obtained at different concentrations of Mg 2+ ,so it is possible that no conformational change occurs in thepresence of this cation as well. Although an increase of surface hydrophobicity was monitored in 14-3-3 [19] in thepresence of divalent cations, no experimental evidence of conformational change has been reported at the present time. 2.5. Kinetic studies Initial velocity studies were performed by followingNADPH formation at 340 nm in a Spectronic 20 spectro-photometer, essentially as previously described [7,9].Concentrations of NADP + and  D -Ga3P were  fi xed at 0.11and 1.2 mM, respectively. Studies on the hystereticbehaviour were performed as before [31]. Samples partially puri fi ed from wheat endosperm and containing the GAPN – 14-3-3 complex [7,9] were incubated at 30  8 C, thetemperatureusedfor activity assays, in a medium containing50 mM Tricine – NaOH (pH 8.5), 1 mM 2-mercaptoethanol,and NaCl at the speci fi ed concentrations. After 5 min pre-incubation with the different concentrations of NaCl, analiquot was withdrawn and immediately assayed for activityin media containing 10 mM MgCl 2 . The time course of thereaction was continuously registered and the non-linear (lagexhibiting) transient analyzed. The kinetic transient fol-lowed a  fi rst order exponential increase, with data  fi tting theequation described by Neet and Ainslie [32]: ½ P  ¼  V  ss t    ð V  ss    V  i Þ k  obs   ½ 1    exp ð k  obs t  Þ where  V  i  and  V  ss  are the initial and the steady state (linear)velocities, respectively;  t   is time of reaction and  k  obs  is theapparent rate constant for the transition between  V  i  and  V  ss .For a more compressive analysis  k  obs  was transformed into k   1obs  ¼  t  . The experimental data obtained under linear reac-tion conditions ( V  ss ) were  fi tted to the generalized Hillequation by a non-linear least square regression kineticscomputerprogram[33].Thisprogramwasusedtodetermine sigmoidicity in the respective substrate saturation curve, bycalculating Hill coef  fi cients ( n H ). Since  n H  was not signi fi -cantly different from 1 in all cases, the same computerprogram was utilized to  fi t data to the Michaelis Mentenequation. One unit (U) is de fi ned as the amount of enzymethat catalyzes the formation of 1  m mol NADPH/min underthe speci fi ed assay conditions. 3. Results and discussion 3.1. Homology modelling of GAPN  A crystal structure (1.8 A ˚  resolution) corresponding to amonomer of apo-GAPN from  S. mutans  (GAPN sm ; PDBcode 1EUH) was utilized as template to build a homologymodel of   fi ve known plant GAPN homotetrameric enzymes.Prokaryotic and eukaryotic protein sequences have 52%identities, being thus in good agreement to performhomology modelling procedure. The multi-protein sequencealignment of GAPN, automatically generated by theDIALIGN [23] method and manually adjusted betweenprokaryotic and plant GAPNs, exhibited sequence blocks of interest as shown in Fig. 1. In addition, even when goodhomology exists between plant and GAPN sm  subunits theoverall model was further re fi ned (see Section 2 for details).The overall topology of the GAPN models was similar toX-ray structure of GAPN sm  [8]. Each monomer can bedivided into three domains (Fig. 2): (i) a cofactor bindingdomain composed of a core which resembles the Rossmannfold (residues 145 – 252)  fl anked by  fi ve helices and fourstrands (residues 2 – 118, 450 – 464), (ii) a catalytic  a  /  b domain (residues 253 – 449), and (iii) a small protrudingdomain (residues 119 – 144 and 465 – 475) enabling oligo-merization.The most remarkable feature in our models is that twoCys residues (Cys 271 and Cys 422 in  Triticum aestivum )strictly conserved in the sequence of plant GAPNs, butabsent in the enzyme from prokaryotes (Fig. 1), areseparated 2.2 A ˚ and in the right orientation to form adisul fi de bond [25] (Fig. 2). To the best of our knowledge, the existence of this disul fi de bond has never been describedand its relevance for the GAPN stability during oxidativestress in plants is under experimentation (unpublished resultfrom our laboratory).Quality of the model was assessed by using differentvalidation tools. We performed proline puckering, packingquality, and stereochemistry of main-chain and side-chainresidues using program WHATIF [25]. The rmsd between template and models using all main-chain atoms was foundto be minor to 1.6 A ˚ . Fig. 2 shows a superposition of C-Abackbones between plant and  S. mutants  GAPNs .  Rama-chandran plot statistics indicated that 92    96.5% of themain-chain dihedral angles ( F  and  C  ) are found in the mostfavourable region. The resulting Ramachandran  Z  -score =   2.511, expresses how well the backbone con-formations of all residues are corresponding to the knownallowed areas in the Ramachandran plot, being the valuewithin expected ranges for well-re fi ned structures [25]. 3.2. GAPN domains interacting with 14-3-3 proteins We have previously analyzed [7,9] possible phosphor-ylation sites in plant GAPN, which could be relevant in theinteraction of the enzyme with 14-3-3 proteins. In such ananalysis, Ser 404 in the enzyme from  T. aestivum  exhibits thehighest phosphorylation probability, being the only siteexposed to solvent (and accessible to a putative kinase) intoGAPN homotetramer. Indeed, the sequence around Ser 404 (see Fig. 1) is more potentially able to bind a 14-3-3 protein.Analysis of this sequence shows that it lacks a Pro residue(+2 from pSer) present in other proteins interacting with 14-  D.M. Bustos, A.A. Iglesias/Journal of Molecular Graphics and Modelling 23 (2005) 490  –  502  493  3-3 [34]. However, there are many evidences that proteinslacking this residue bind 14-3-3 proteins as well [35,36].Indeed, sequence analysis reveals that sites lacking the Proresidue have a Val in position +1, similar to what is found inthis site from plant GAPNs (Fig. 1). Several proteins bindto 14-3-3 via unconventional phosphopeptide bindingsequences, like cardiac phosphofructokinase-2 (PFK-2),whosebindingto14-3-3sisdependentonphosphorylationatSer 483 (RNYpSVGS) by protein kinase B, which does notprecisely conform to a canonical 14-3-3-binding motif  [35].Also, three different K  + -channels and the nicotinic receptorhave a 14-3-3 binding motif without Pro residue and with aValresidue in+1 position [36].Thesite inGAPNpotentiallyinteracting with 14-3-3 proteins is exposed to the solventeither in the monomeric or tetrameric structures of theenzyme (data not shown). Taking this into account, buildingof a structural model that includes the 14-3-3 binding pocketwas performed from the alignment onto template monomer.In our previous report [9], we showed that the peptideARAApSAPA (designed by Carol MacKintosh, University  D.M. Bustos, A.A. Iglesias/Journal of Molecular Graphics and Modelling 23 (2005) 490  –  502 494Fig. 1. C-terminal alignment of primary structure corresponding to plant and prokaryotic GAPNs. Highly conserved residues: Cys, canonical 14-3-3 bindingsites, and His are in black background. Gaps in the sequences were replaced by dashes.  A. thaliana :  Arabidopsis thaliana .  A. graveolens :  Apium graveolens .  O.sativa :  Oryza sativa .  P. sativum :  Pisum sativum .  N. plumbagini :  Nicotiana plumbaginifolia .  T. aestivum :  Triticum aestivum .  Z. mays :  Zea mays .  S. mutans : Streptococcus mutans .  S. solfataric :  Sulfolobus solfataricus isozyme 1 .  S. solfataric2 :  Sulfolobus solfataricus isozyme 2 .  T. tenax :  Thermoproteus tenax .  M. jannaschii :  Methanococcus jannaschii .  S. coelicolor  :  Streptomyces coelicolor  .  S. pyogenes :  Streptococcus pyogenes isozyme 1 .  S. pyogenes2 :  Streptococcus pyogenes isozyme 2 .  S. pneumoniae :  Streptococcus pneumoniae .
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!