Crystal structures of methionine adenosyltransferase complexed with substrates and products reveal the methionine-ATP recognition and give insights into the …

Crystal Structures of Methionine AdenosyltransferaseComplexed with Substrates and Products Reveal theMethionine-ATP Recognition and Give Insights intothe Catalytic Mechanism

Beatriz Gonzalez1, Marıa A. Pajares2, Juan A. Hermoso1

Danielle Guillerm3, Georges Guillerm3 and Julia Sanz-Aparicio1*

1Grupo de CristalografıaMacromolecular y BiologıaEstructural, Instituto deQuımica-Fısica “Rocasolano”CSIC, Serrano 119, 28006Madrid, Spain

2Instituto de InvestigacionesBiomedicas “Alberto Sols”CSIC-UAM, Arturo Duperier4, 28029 Madrid, Spain

3Universite de ReimsChampagne-Ardenne, UMR6519-UFR Sciences-BP 103951687 Reims Cedex, France

Methionine adenosyltransferases (MATs) are a family of enzymes incharge of synthesising S-adenosylmethionine (SAM), the most importantmethyl donor present in living organisms. These enzymes use methionineand ATP as reaction substrates, which react in a SN2 fashion where the sul-phur atom from methionine attacks C50 from ATP while triphosphatechain is cleaved. A MAT liver specific isoenzyme has been detected,which exists in two distinct oligomeric forms, a dimer (MAT III) and atetramer (MAT I). Our previously reported crystal structure of MAT Icomplexed with an inhibitor led to the identification of the methionine-binding site. We present here the results obtained from the complex ofMAT I with a competitive inhibitor of methionine, (2S,4S)-amino-4,5-epoxypentanoic acid (AEP), which presents the same features at the meth-ionine binding site reported before. We have also analysed several com-plexes of this enzyme with methionine and ATP and analogues of them,in order to characterise the interaction that is produced between both sub-strates. The crystal structures of the complexes reveal how the substratesrecognise each other at the active site of the enzyme, and suggest a puta-tive binding site for the product SAM. The residues involved in the inter-actions of substrates and products with MAT have been identified, andthe results agree with all the previous data concerning mutagenesis exper-iments and crystallographic work. Moreover, all the information providedfrom the analysis of the complexes has allowed us to postulate a catalyticmechanism for this family of enzymes. In particular, we propose a keyrole for Lys182 in the correct positioning of the substrates, and Asp135p,in stabilising the sulphonium group formed in the product (SAM).

q 2003 Elsevier Ltd. All rights reserved

Keywords: methionine adenosyltransferase complexes; methionine; ATPand SAM binding-site; X-ray structure*Corresponding author

Introduction

S-adenosylmethionine (SAM) is the most import-ant methyl donor found in living organisms.1 Thenumber of reactions that involve the use of SAMhas been calculated to be as large as those usingATP.2 These reactions are catalysed by methyltrans-ferases that participate in reactions as diverse asDNA methylation and synthesis of neurotransmit-ters or phospholipids. In addition, SAM is alsoused as a propylamine donor, after decarboxyl-ation, for the synthesis of polyamines (spermidineand spermine). Methionine adenosyltransferases

0022-2836/$ - see front matter q 2003 Elsevier Ltd. All rights reserved

E-mail address of the corresponding author:[email protected]

Abbreviations used: MAT I, methionineadenosyltransferase I from rat liver; cMAT, methionineadenosyltransferase from E. coli; SAM, S-adenosylmethionine; LcisAMB, L-2-amino-4-methoxy-cis-but-3-enoic acid; AEP, (2S,4S)-amino-4,5-epoxypentanoic acid; PPPi, triphosphate; Pi,orthophosphate; PPi, pyrophosphate; DTT, dithiothreitol.

doi:10.1016/S0022-2836(03)00728-9 J. Mol. Biol. (2003) 331, 407–416

(MAT, EC 2.5.1.6) are the family of cytosolicenzymes that synthesise SAM, using for thispurpose L-methionine and ATP. These highlyconserved proteins need divalent cations for cata-lysis, whereas monovalent cations are able to acti-vate them. In mammals two isoenzymes havebeen detected, one ubiquitous (MAT II) andanother mainly liver-specific. This last MAT existsas homodimers (MAT III) and homotetramers(MAT I), the normal oligomer rate being altered indiseases such as liver alcohol cirrhosis andcancer.2 – 4 The mechanisms that regulate the oligo-meric state are not known, although many studieshave shown a critical role for the cysteine residuesof the enzyme, some of which are specific of theliver MAT.5 – 7

As for the catalytic mechanism, it is well knownthat SAM synthesis takes place in two steps(Figure 1). First, L-methionine reacts with ATP toyield SAM and triphosphate (PPPi) through a SN2mechanism. The sulphur atom of methionineattacks the C50 from ATP, at the same time that thePPPi chain is cleaved. In a second step, PPPi ishydrolysed yielding pyrophosphate (PPi) andorthophosphate (Pi) as side products. This hydroly-sis is orientated since 98% of the Pi comes from thePg of the ATP. Finally, all the products are released.In order to carry out the mechanistic studies sev-eral analogues of methionine and ATP have beensynthesised, some of which are inhibitors of the

reaction. Among the methionine analogues, cyclo-leucine has been shown to reduce SAM levels, andserve for further inhibitor design, the most power-ful inhibitors being oxygenated analogues of theamino acid, LcisAMB and AEP (Figure 1(b)).8,9 Onthe basis of the results obtained by the use ofthese analogues and non-hydrolysable analoguesof ATP several hypotheses have been proposed toexplain the role of the hydrolytic step in the overallreaction. Thus, PPPi hydrolysis could be necessary,as an energy source for the conformational changesrequired to complete the reaction. It could also be away of producing more efficient leaving groups(PPi and Pi), or alternatively, a way of preventingthe reversal of the reaction.

The structural knowledge of this protein familyand their complexes with substrates or products isquite limited. The structures for MAT from E. coli(cMAT)10 and recombinant rat liver MAT I11 areknown, both being tetrameric and showing thesame overall folding. Tetramers are constituted oftwo tightly bound dimers rotated 908 one to theother (see Figure 2(a)). The subunits are organisedin three domains related by ternary pseudo sym-metry, with the face involved in monomer/mono-mer interaction being essentially hydrophobic. Theactive site is located in a broad cavity defined bythe dimer interface, residues of both subunits con-tributing to it. Thus, there are four active sites pertetramer, one on each side of the dimers. The main

Figure 1. Enzymatic mechanismof MAT. (a) Scheme of the reactioncatalysed by MAT in which bothsteps are represented: the sulphurof methionine attacks the C50 ofATP, producing enzyme-boundSAM and PPPi in a SN2 reactionand, subsequently, PPPi is hydro-lysed before all the products arereleased. (b) Competitive inhibitorsof MAT enzymes used in this work.

408 Catalytic Mechanism of MAT I

difference between cMAT and MAT I structures isrelated to the orientation of the C-terminal domain,leading to active sites of different sizes.11 Inaddition, a disordered loop is identified at theentrance to the active site in both proteins. Pre-vious studies suggested conformational changesin this loop related to effects in catalytic activity.12,13

Kinetic differences observed among this family ofenzymes have been related to this loop, due to itslow conservation at the sequence level, as well asthe changes observed in its length.14 Moreover,this loop could be responsible for the differencesin the orientation of the C-terminal domain foundbetween both structures, as it connects the centraland C-terminal domains.

The structure of several MAT-complexes is alsoknown (cMAT–ADP–Pi,

15 cMAT–PPi–Pi15 and

MAT I–LcisAMB11). ADP was found bound to theenzyme through interactions with both subunits ofthe dimer in crystals of cMAT–ATP, where Pg ofATP was hydrolysed. As for the methionine-bind-ing site, this was revealed in the MAT I–LcisAMBcomplex structure, involving the loop 251–260 inamino acid recognition. Further analysis of therole of the residues directly involved in methioninebinding (F251 and D180) was carried out by site-directed mutagenesis. However, with the infor-mation obtained from the binary complexes, theproblem arises from the distance observed betweenADP and LcisAMB sites that do not allow anexplanation for the reaction mechanism. Therefore,new complexes need to be studied in order toinvestigate how the substrates interact, by lookingat possible intermediate species of the enzymaticmechanism. Here we present results obtainedfrom the complex of MAT I with both substratesATP and methionine, and new ternary complexesin which methionine has been substituted by theinhibitors AEP or LcisAMB. The analysis of thesecomplexes allows a deeper characterisation of theactive site and suggests a catalytic mechanism forthis family of enzymes.

Figure 2. The MAT I–AEP complex: (a) MAT I tetra-mer with the ligands and metal ions represented in ball-and-stick. The subunit of each dimer is coloured in cyanand blue. (b) A detail of the active site showing thethree phosphate positions, and the inhibitor AEP. Theloop (251–260) involved in methionine binding is in yel-low, and the putative metal are represented as spheres,Mg2þ in orange, and Kþ in magenta. (c) Detail of the pro-posed atomic interactions: the methionine analoguestacks against Phe251 in a rather planar conformation. Acarboxylate oxygen atom is co-ordinating a Mg2þ

(Mgm), which in turn is linked to Asp180. A secondMg2þ cation (Mg3) is linked to Asp135p of the other sub-unit and is further co-ordinated to a phosphate anion(Pi3). The final 2Fo 2 Fc electron density map at theAEP and the ions is contoured at 1s. Co-ordination ofboth Mg2þ ions must be completed with water moleculesnot visible in the electron density.

Catalytic Mechanism of MAT I 409

Results and Discussion

The tetrameric form of MAT from rat liver waspurified and crystallised as described.11 Experi-mental details and structure determination pro-cedures are given in Materials and Methods andin Table 1.

MAT I–AEP complex

The AEP methionine analogue (Figure 1(b)) wasfound bound to the enzyme at the same sitereported previously for LcisAMB.11 Refinement ofthe model led to final Fourier electron densitymaps for the AEP molecule shown in Figure 2, thecorresponding refinement parameters being sum-marised in Table 1.

The crystallisation buffer included 10 mM Na/Kphosphate and the complex presents three phos-phate molecules bound to the active site, thereinlabelled Pi1, Pi2 and Pi3. Pi1 and Pi2 positions aresimilar to those previously described in native andcomplexed cMAT structures (Figure 2(a) and (b)).Pi3 is next to the methionine-binding site and wasalso found in the MAT I–LcisAMB structure. Thisanion (Pi3) seems to be in a new position onlyobserved in MAT–methionine analogue com-

plexes, thus suggesting its relationship to the occu-pation of the methionine-binding site. Pi3 interactswith Lys182(NZ), Gln184(NE) and a magnesiumion which in turn is co-ordinated to Asp135p(OD)and Ser248(OG). Two Mg2þ ions (Mg2 and Mg3)are co-ordinated with phosphates (Pi2 and Pi3,respectively) and a third one, therein namedMgm, is involved in methionine-protein recog-nition (Figure 2(c)). The Mg sites are assigned onthe basis of the positions reported before for therat liver and E. coli complexes. However, althoughMg2þ is included in the protein liquor, the occu-pancy of these sites by a different cation such asNaþ cannot be excluded, due to the moderate res-olution of the data.

The AEP binding site is conserved with respectto the LcisAMB binding site, and the pattern ofinteractions with the protein is essentially thesame. First, polar interactions are producedbetween the inhibitor carboxylate group (COOH)and Asp180 (OD) mediated by a metal (Mgm).In addition, fixing hydrophobic interactions areproduced between the AEP carbonated side-chain and Phe251 aromatic ring. Therefore, theresults presented herein support the main fea-tures of the methionine binding-site previouslyreported.

Table 1. Data collection and refinement statistics

Complex 1 MAT I–AEP 2 MAT I–ATP–Met 3 MATI–ATP–AEP 4 MATI–ADP–LcisAMB

A. Data collectionReactants AEP ATP/Met ATP/AEP ATP/LcisAMBConcentration (mM) 10 30/10 10/10 10/10Temperature (K) 120 120 120 120Source Rotating-anode Rotating-anode Rotating-anode Rotating-anodeSpace group P4122 P4122 P4122 P4122Unit cell dimensions (A) 114.89, 161.08 114.99, 160.34 115.12, 160.89 115.14, 160.45Limiting resolution (A) (Outer shell) 3.02 (3.16–3.02) 2.70 (2.82–2.70) 3.35 (3.48–3.35) 3.0 (3.19–3.0)Unique reflections 18,937 27,790 14,731 20,749Rsym 10.1 (49.8) 7.8 (50.2) 14.5 (49.5) 12.3 (48.2)Completeness (%) 99.3 (100.0) 97.5 (97.5) 100.0 (98.6) 96.6 (98.3)Mean multiplicity 4.2 (4.2) 4.1 (3.8) 4.8 (4.8) 8.4 (8.4)Mean I/sI 3.0 (1.5) 7.3 (1.6) 4.5 (1.5) 4.9 (1.6)

B. Final refinement parametersResolution range (A) 25.0–3.02 25.0–2.70 25.0–3.35 25.0–3.00Protein atoms (non-H) 5692 5692 5692 5692Heterogen atoms (non-H) 29 55 55 56Solvent molecules 180 181 190 183R-factor 0.23 0.24 0.22 0.23Rfree 0.28 0.29 0.27 0.26

R.m.s. deviationsBond lengths (A) 0.008 0.009 0.010 0.009Bond angles (deg.) 1.4 1.6 1.5 1.4Dihedral angles (deg.) 24.7 24.6 24.7 24.7Improper angles (deg.) 1.3 1.2 1.2 1.2

Averaged B-factors (A2)Main-chain 25.2 25.7 22.7 22.6Side-chain 27.8 27.2 21.9 25.7Methionine /analogue 78.1 75.1 45.6 66.2ATP/ADP 81.5 79.7 52.5

Rsym ¼P

lIðhÞi 2 kIðhÞll=P

kIðhÞl; Rfactor ¼P

ðlFobs 2 FcalclÞ=P

lFobsl (Rfree is equivalent to R factor for a randomly selected 7% subsetof reflections not used in structure refinement).


Complex of MAT I with both substrates: theMAT I–ATP–Met complex

From the crystal structures of the binary com-plexes of cMAT and MAT I, it was difficult to envi-sage how both substrates approach for theenzymatic reaction to take place. This fact made itnecessary to study ternary complexes with bothsubstrate binding-sites occupied. The crystals withboth substrates were prepared by co-crystallisationof MAT I and L-methionine, and by soaking the

resulting crystal in an ATP solution, since co-crys-tallisation with both substrates did not yield suit-able crystals. The electron density maps allowedthe clear identification of ATP and methionine atthe active site groove, both substrates being mod-elled as shown in Figure 3(a). Interestingly, thedifference Fourier electron map of this structurecomprised additional electron density at the ATPsite compatible with the presence of a SAM mol-ecule in a different orientation to ATP. This densitywas defined and contoured only very weakly and,

Figure 3. Stereo views of the MAT I active groove showing substrates, inhibitors and products and final 2Fo 2 Fc

density maps at the ATP and methionine sites of the complexes. (a) MAT I–ATP–Met, contoured at 1s level. (b)MAT I–ATP–AEP, at 0.9s level. (c) MAT I–ADP–LcisAMB complex contoured at 0.8s level. The same residues ofMAT I are represented as sticks in all the plots. Residues labelled with an asterisk correspond to subunitB. Methionine (a) and the analogues (b) and (c) are at the same site shown in Figure 2. The ADP in (c) is in a differentorientation to that of ATP (a) and (b). As it can be observed, AEP (b) is stacking against Phe251 in a different orien-tation to that of methionine (a) but, nevertheless, the same interactions with ATP–Pg and Pb are produced.


due to the moderate resolution of the analysis(2.7 A), model building of an alternate occupancywas not attempted. However, this fact suggeststhat a partial SAM-synthesis has taken place, andthe putative position of the product SAM couldhave some implications in the mechanism, thatwill be discussed below.

Complexes of MAT I with ATP and Metanalogues: the MAT I–ATP–AEP andMAT I–ADP–LcisAMB complexes

L-Methionine was replaced by the competitiveinhibitors, AEP or LcisAMB, and crystals wereobtained by the co-crystallisation method. Bothinhibitors occupy the methionine-binding site pre-viously characterised. On the other hand, electrondensity maps show density in the ATP site of bothcomplexes. The ATP molecule in the complex withAEP adopts a position similar to that of ATP inMAT I–ATP–Met complex (Figure 3(b)). However,in the complex with LcisAMB, the appearance ofthe typical L-shaped density can only be fittedunambiguously with an ADP molecule disposedin a different orientation to that of ATP (Figure3(c)). Remarkably, this position is similar to thelocation suggested for the product SAM in theactive site groove of the previously described com-plex of MAT I with both substrates. Therefore, wehave observed in the crystal the previouslyreported ATPase function of this enzyme,15 the pro-duct ADP adopting a position which resemblesthat envisioned for the main product of reaction.

The ATP binding site

We have observed the ATP binding site in MATI, in the presence of methionine (MAT I–ATP–Met

complex, Figure 3(a)) and an analogue of it (MATI–ATP–AEP complex, Figure 3(b)). The ATP posi-tion is essentially the same in both complexes. Fur-thermore, this site is similar to that occupied byADP in cMAT–ADP crystal complex,15 although itpresents some interesting differences. Structuralanalysis of the highest resolution ATP-complexpresented here (MAT I–ATP–Met) shows themain interactions stabilising the ATP moiety(Figure 4). The NH2 group from adenine (N6)interacts with Glu71p (OE), while the adenine N3is hydrogen bonded to Lys290p (NZ). However,the strongest interaction between MAT I and ATPseems to be that produced through Pg, which inter-acts with several protein residues, His30 (NE2),Glu24 (OE) through Mg2, Lys182 (NE), Lys266(NE) and Asp180 through Mgm. Remarkably, thePi2 position is occupied in our case by the Pg,whereas the position is occupied by the Pb ofADP in the cMAT–ADP–Pi complex (Figure 5).Therefore, the Pb position described by Takusagawaet al. (1996) is coincident with the Pg positiondescribed herein. The ribose portion and the phos-phates Pa and Pb are, on the contrary, more weaklybound to the enzyme and their positions are not con-served in both the complexes, as can be seen inFigure 5.

The role of the above mentioned residues in sta-bilising ATP is further confirmed by site-directedmutagenesis studies. Thus, replacement of His14in cMAT (His30 in MAT I) has been shown to abol-ish the SAM synthesis whereas Lys245 mutation(Lys266 in MAT I) seems to impair the triphospha-tase activity.16,17 In rat liver MAT I, we found thatthe mutant Lys182Gly shows an absolute loss ofSAM synthesis activity, while preserving triphos-phatase activity and the ability to bind ATP. There-fore, Lys182 and Lys266 seem to play different

Figure 4. Atomic interactions at the active site of theMAT I–ATP–Met complex. Pg of ATP is interactingwith several residues of the enzyme and the methion-ine-COOH, while Pb is linked to the methionine-NH2.Lys182 is hydrogen bonded to both, Pa and Pg of ATP,and also to the phosphate Pi3 (not shown in the Figurefor clarity). On the other hand, the sulphur of methionineis co-ordinating a putative Mg which can be involved inthe subsequent nucleophylic attack.

Figure 5. MAT’s active site. Superposition of the Ca

MAT I (yellow) and cMAT (grey) structures, in whichthe following molecules are represented: ATP found inMAT I–ATP–Met complex (green), ATP found in MATI–ATP–AEP complex (red) and ADP reported in cMATcomplex15 (brown). As can be seen, Pg of both ATP andPb of ADP are in a similar position, which in turn isequivalent to Pi2.


roles in the enzymatic mechanism that will be dis-cussed below.

ATP-methionine recognition

Despite previous results indicating the long dis-tance observed between the binding site of the sub-strates, ATP and methionine seem to interactclosely in the crystal (Figure 4). First, the methioninecarboxylate and the ATP Pg interact through a Mgion (Mgm) that is also involved in Met recognitionby MAT I, as stated before.11 Moreover, a secondinteraction is produced between methionine aminegroup (NH2) and the ATP Pb phosphate. This lastinteraction seems to be essential since methionineanalogues without amine group are not reactionsubstrates, as has been reported.18 Furthermore,the interaction is observed in both ternary com-plexes (Figure 3(a) and (b)) despite the constrainedgeometry of the AEP epoxy group, which conse-quently is occupying the methionine binding sitein a different orientation from LcisAMB and meth-ionine. Therefore, we conclude that methionine orits analogues orient the ATP Pa and Pb, theposition of the amine group of methionine beingcrucial in determining this Pb position (Figure 5).

In conclusion, ATP and methionine closely inter-act at the active site groove and the presence ofboth substrates is needed to have a representativepicture of the enzymatic reaction. Nevertheless,the distance between the sulphur of methionineand C50 from ATP is still too long (around 9 A) forthe nucleophylic attack to occur, which suggeststhat in subsequent states an approximation shouldbe produced.

ADP site: putative product binding site?

ADP molecule has been clearly identified fromelectron density maps of the crystal obtained fromMAT I with ATP and LcisAMB, in which the SAMsynthesis reaction has been blocked by using themethionine analogue. Consequently, the ATPasefunction of MAT has taken place, as observed inthe E. coli15 enzyme. But, in contrast to the complexfrom the bacterial enzyme, our complex presentsthe methionine binding-site occupied. This factsuggests that we may be looking at different stepsof the mechanism, and the ADP found in our com-plex might be mimicking the product bound state.This hypothesis is also in agreement with theobservation of the residual density suggesting thepresence of SAM in the MAT I–ATP–Met complexmentioned above, which seems to be located in asimilar way. On the other hand, the ADP in ourcomplex weakly interacts with MAT I through H-bonds of the amine group with Lys286p andAsp292p, whereas the adenine ring is packed toLys290p side-chain (Figure 3(c)). This weak inter-action would be consistent with a step of the mech-anism before the product release. Furthermore, thehypothesis of this ADP moiety mimicking theSAM binding-site would imply the sulphonium

interaction with Asp135p (Asp118p in cMAT). Thesulphonium positive charge stabilisation by a car-boxylate is frequent in proteins, as seen in struc-tures of SAM-dependent methyltransferases.19 – 22

This interaction, could explain the effect of themutation at this aspartate in cMAT, since themutant presents a reduced SAM synthesis activityand lacks the activation of PPPi hydrolysisobserved in the presence of SAM,16 typical of thisfamily of enzymes.

Proposed enzymatic mechanism

The structure of MAT I ternary complexes withboth reaction substrates has been determined forthe first time, revealing the interactions producedbetween them. At the mean time, reaction hasbeen produced to certain extension in some of thecomplexes, suggesting a possible location of theproduct binding-site. Furthermore, we know thatthere are three positions with phosphate affinity.Pi1 and Pi2 might be related to ATP binding andPPPi hydrolysis, on the basis of the results obtainedhere and on the cMAT–ADP–Pi and in cMAT–PPi–Pi complexes. Pi3 is next to the methioninebinding site and co-ordinates to Lys182, whosemutation abolishes SAM synthesis activity,suggesting a crucial role for Pi3 site. Finally, thesubstrates interact in MAT I active site, but thereaction centres are not disposed in a suitable pos-ition for the nucleophylic attack to proceed.

On the basis of the available experimental infor-mation and the structures here reported, we pro-pose a mechanism to explain how the enzymaticreaction can take place. Starting from the ATP con-formation observed in the MAT I–ATP–Met crys-tal (Figure 6(a)), the substrates could approach bysimple torsion of the triphosphate chain whilekeeping the Pg position, which seems fixedthrough many links to the enzyme (Figure 4).Besides, the substrates recognition through themain interactions (Pg–Mgm–COOH and Pb–NH2) is also maintained. Such a rotation throughthe polyphosphate chain can situate Pa in Pi3position, which is not far away, while conservingthe interaction with Lys182. Thus, Lys182 could beinvolved in the ATP conformational changethrough its long side-chain, as we proposedbefore.11 The resulting “intermediate step” (seeFigure 6(c)) would place the reaction centres in asuitable position, converting the Pi3 position in a“reactive position”. Moreover, Mg3 would remainbetween the methionine sulphur atom and O50

from ATP, therefore assisting in the nucleophylicattack by weakening the C5–O50 bond. Phe251, inaddition to orientating the methionine side-chain,could have a role in stabilising the positive chargethat is being produced in the transition statethrough a cation–p interaction, as described inmany SAM-dependent enzymes.23 Moreover, thisproposed ATP state situates the ribose and adeninerings in an orientation intermediate between thosefound for ATP in the crystals, and the ADP


mimicking the putative SAM binding site(Figure 6(b)). This makes the proposed mechanismcompatible with the different orientations observedbetween substrate and the product ADP. Probably,the active-site loop, which is disordered in the crystalstructures of MAT I and cMAT, must be involved insuch a conformational change.

This mechanism may also explain how inhibitionis produced when using methionine analogues.The inhibitors, having oxygen instead of sulphur,could form stable co-ordination complexes withMg3, precluding the reaction to proceed. Thereforethe mechanism can be initiated in a similar way,but ATPase reaction takes place instead of SAMsynthesis. The enzymatic process is then mimickedand the product ADP reorients in a similar way asSAM. That would explain why ADP in cMAT isplaced in a similar orientation to that of ATP, inthe absence of methionine. Moreover, this obser-vation is consistent with previous reported kineticanalyses that revealed conformational changesonly when all ligand binding sites were occupied.24

On the other hand, a model for the triphosphatechain after SAM formation could be traced fromPi1 to Pi2 positions. Following this model, Mg2and Lys266 (Figure 4) could have an importantrole in PPPi hydrolysis, since they are located nextto the bridge oxygen of the PPPi model. In agree-ment with such a role, the Lys266 mutantsobtained in cMAT (Lys245 in the bacteria) havereduced PPPi hydrolysis.17

In summary, we propose here a singular mech-anism for MAT I, based on results from severalcomplexes. This mechanism is coherent with theavailable biochemical data, as well as with thenecessity of divalent ions for SAM synthesis25 andPPPase activity and also with the mutagenesisresults on cMAT16,17 and MAT I. Moreover, most ofthe residues proposed as involved in substrate/product binding and catalysis are in sequence seg-ments comprising characteristic motifs. It isremarkable that Lys286 is in a P-loop typical ofphosphate binding proteins26 and Lys266 is in apyrophosphate binding-like motif of inorganicpryrophosphatases.27 Furthermore, Asp135p, whichwould stabilise the sulphonium charge createdupon SAM formation, and the crucial Phe251 arelocated in motifs I and II, respectively, identified inSAM binding proteins.28

Nevertheless, many details of the singular mech-anism remain unknown. In particular, it would beclarifying to have a clear picture of the crucial roleof the active site loop in the progress of the reac-tion, and more work is necessary to fully under-stand the functionality of these very interestingenzymes.

Figure 6. Molecular surface at the active site of MAT Ishowing (a) the substrates ATP and methionine in greyand (b) the ADP molecule in cyan. The phosphate ionsare shown in yellow and two Mg2þ are represented asmagenta spheres. As it can be seen, the ADP is in adifferent orientation from that of ATP, suggesting thatthe product reorients in the active site after reaction isproduced and before it is released. (c) Proposed enzy-matic mechanism and superposition of the differentstates: the binding of ATP and Met, as observed in thecrystal (a) is represented in dark grey and the putativeSAM, superimposed to ADP found in the crystal (b), is

represented in cyan. A model for the proposed “reac-tive” conformation of ATP is shown in white. In thismodel, ATP–Pg remains at the position coincident withPi2, while ATP–Pa is situated at Pi3 (both Pi sites arecircled in yellow).


Materials and Methods

Recombinant rat liver MAT expression, purificationand crystallisation

Competent E. coli BL21 (DE3) cells were transformedwith the plasmid pSSRL-T7N that contains the sequenceof wild-type MAT.5 The inclusion bodies were isolatedafter disruption of the cells by sonication, and purified asdescribed by Lopez-Vara et al.29 Refolding was carried outusing 10 mM DTT, and the DTT-refolded protein was puri-fied as previously described, but using a Q-Sepharose col-umn. The purity of the samples was tested by SDS-PAGEelectrophoresis, and the final protein samples character-ised by fluorescence and circular dichroism spectroscopies.Crystallisation conditions were as described by Gonzalezet al.11 The protein samples (30 mg/ml protein in 50 mMTris–HCl, pH 8, 10 mM DTT, 50 mM KCl, 10 mM MgSO4

and 1 mM EDTA) were mixed with equal amounts of pre-cipitant solution (16% (w/v) PEG 10000 in 100 mMHepes, pH 7.5) and equilibrated by vapour-diffusion inhanging drops. Only very fresh purified sample was suit-able for crystallisation experiments, as reported before.The complexes were obtained by the co-crystallisationmethod, by adding substrate or analogues to a concen-tration of 10 mM over the protein solution. Apparition ofcrystals was very scarce and difficult to reproduce. Only afew crystals reached a size suitable for diffraction trialsshowing patterns of poor quality. All the complexes belongto the P4122 space group, with two molecules in the asym-metric unit and a 60% solvent content.

Crystallographic data collection

Full data sets were obtained at 120 K in-house on a 345MAR Research Imaging Plate detector using an ENRAF-NONIUS rotating anode generator (l ¼ 1.54 A) which isfitted with a nitrogen Oxford-Cryosystem. Attempts toimprove the resolution of the data by using synchrotronsources were unsuccessful. Data were processed usingthe MOSFLM program package30 and merged with theCCP4 suite.31 A summary of data collection andreduction statistics is given in Table 1.

Structure solution and refinement

Initial phases of the MAT I-complexes were obtainedby rigid body refinement with the co-ordinates from theMAT I–LcisAMB crystal (pdb code: 1QM4). The appro-priate substrates or inhibitors were manually built onthe SA-omit maps calculated with a 3 A exclusion sphereat the methionine and ATP binding sites. Several ions(Kþ, Mg2þ, PO4

22) have also been modelled into positiveelectronic density peaks, taking into account the compo-sition of the protein liquor and chemical criteria. Theinitial model was improved by automated and manualbuilding, positional refinement and simulated annealingusing X-PLOR.32 The final model presents two areas ofdisorder previously reported, the N-terminal residues1–16 and the loop at the active site (117–129). The stereo-chemistry of the final models for the complexes waschecked by using the program PROCHECK.33 A sum-mary of the refinement statistics is shown in Table 1.Figures were prepared with MOLSCRIPT,34 renderedwith Raster3D,35 and WEBLAB†.

Accession number

The atomic co-ordinates have been deposited in theRCSB Protein Data Bank (accession codes: 1O90, 1O92,1O93 and 1O9T).

Acknowledgements

Work has been partially supported by the grants:FIS 01/1077, RCMN CO3/O8 (Ministerio de Sani-dad y Consumo) and BMC2002-00243, BIO2000-1279-C02-02 (Ministerio de Ciencia y Tecnologıa).Authors thank F. Garrido for his technical assist-ance, Dr J. R. Sufrin for supplying the LcisAMBinhibitor and Brenda Ashley Morris for style andgrammatical corrections.

References

1. Cantoni, G. L. (1975). Biochemical methylations:selected aspects. Annu. Rev. Biochem. 44, 435–451.

2. Mato, J. M., Alvarez, L., Ortiz, P. & Pajares, M. A.(1997). S-adenosylmethionine synthesis: molecularmechanisms and clinical implications. Pharmacol.Ther. 73, 265–280.

3. Cabrero, C., Martın-Duce, A., Ortiz, P., Alemany, S. &Mato, J. M. (1988). Specific loss of the high-molecular-weight form of S-adenosylmethionine synthetase inhuman liver cirrhosis. Hepatology, 8, 1530–1534.

4. Tsukada, K. & Okada, G. (1980). S-Adenosylmethio-nine synthetase isozyme patterns from rat hepatomainduced by N-2-fluorenylacetamide. Biochem. Bio-phys. Res. Commun. 94, 1078–1082.

5. Mingorance, J., Alvarez, L., Sanchez-Gongora, E.,Mato, J. M. & Pajares, M. A. (1996). Site-directedmutagenesis of rat liver S-adenosylmethioninesynthetase. Identification of a cysteine residue criticalfor the oligomeric state. Biochem. J. 315, 761–766.

6. Avila, M. A., Mingorance, J., Martınez-Chantar, M. L.,Casado, M., Martın-Sanz, P., Bosca, L. & Mato, J. M.(1997). Regulation of rat liver S-adenosylmethioninesynthetase during septic shock: role of nitric oxide.Hepatology, 25, 391–396.

7. Martınez-Chantar, M. L. & Pajares, M. A. (2000).Assignment of a single disulphide bridge in rat livermethionine adenosyltransferase. Eur. J. Biochem. 267,132–137.

8. Sufrin, J. R., Lombardini, J. B. & Alks, V. (1993).Differential kinetic properties of L-2-amino-4-methylthio-cis-but-3-enoic acid, a methionine analoginhibitor of S-adenosylmethionine synthetase. Bio-chim. Biophys. Acta, 1202, 87–91. and referencestherein.

9. Lavrador, K., Allart, B., Guillerm, D. & Guillerm, G.(1998). A new series of S-adenosylmethioninesynthetase inhibitors. J. Enzyme Inhibit. 13, 361–367.

10. Takusagawa, F., Kamitori, S., Misaki, S. & Markham,G. D. (1996). Crystal structure of S-adenosylmethio-nine synthetase. J. Biol. Chem. 271, 136–147.

11. Gonzalez, B., Pajares, M. A., Hermoso, J., Alvarez, L.,Garrido, F., Sufrin, J. R. & Sanz-Aparicio, J. (2000).The crystal structure of tetrameric methionine adeno-syltransferase from rat liver reveals the methionine-binding site. J. Mol. Biol. 300, 363–375.† http://www.msi.com


http://www.msi.com

12. Ruiz, F., Corrales, F. J., Miqueo, C. & Mato, J. M.(1998). Nitric oxide inactivates rat hepatic methionineadenosyltransferase in vivo by S-nitrosylation. Hepa-tology, 28, 1051–1057.

13. Perez-Mato, I., Castro, C., Ruiz, F. A., Corrales, F. J. &Mato, J. M. (1999). Methionine adenosyltransferaseS-nitrosylation is regulated by the basic and acidicamino acids surrounding the target thiol. J. Biol.Chem. 274, 17075–17079.

14. Taylor, J. C., Takusagawa, F. & Markham, G. D.(2002). The active site loop of S-adenosylmethioninesynthetase modulates catalytic efficiency. Biochemistry,41, 9358–9369.

15. Takusagawa, F., Kamitori, S. & Markham, G. D.(1996). Structure and function of S-adenosylmethio-nine synthetase with ADP, BrADP, and PPi at 2.8 Aresolution. Biochemistry, 35, 2586–2596.

16. Taylor, J. C. & Markham, G. D. (1999). The bifunc-tional active site of S-adenosylmethionine synthe-tase. Roles of the active site aspartates. J. Biol. Chem.274, 32909–32914.

17. Taylor, J. C. & Markham, G. D. (2000). The bifunc-tional active site of S-adenosylmethionine synthe-tase. Roles of the basic residues. J. Biol. Chem. 275,4060–4065.

18. Markham, G. D. (1981). Spatial proximity of twodivalent metal ions at the active site of S-adenosyl-methionine synthetase. J. Biol. Chem. 256, 1903–1909.

19. Cheng, X., Kumar, S., Posfai, J., Pflugrath, J. V. &Roberts, R. J. (1993). Crystal structure of the HhalDNA methyltransferase complexed with S-adeno-syl-L-methionine. Cell, 74, 299–307.

20. Dixon, M. M., Huang, S., Matthews, R. G. & Ludwig,L. (1996). The structure of the terminal domain ofmethionine synthase: S-adenosylmethionine forreductive methylation. Structure, 4, 1263–1275.

21. Fu, Z., Hu, Y., Konishi, K., Takata, Y., Gawa, H. O.,Gomi, T. et al. (1996). Crystal structure of glycine N-methyltransferase from rat liver. Biochemistry, 35,11985–11993.

22. Rafferty, J. B., Somers, W. S., Saint-Girons, I. & Phillips,S. E. (1989). Three dimensional crystal structures ofEscherichia coli met repressor with and without core-pressor. Nature, 341, 705–710.

23. Ma, J. C. & Dougherty, D. A. (1997). The cation–pinteraction. Chem. Rev. 97, 1303–1324.

24. Mcqueney, M. S., Anderson, K. S. & Markham, G. D.(2000). Energetics of S-adenosylmethionine synthe-tase. Catal. Biochem. 39, 4443–4454.

25. Markham, G. D., Hafner, E. W., Tabor, C. W. & Tabor,H. (1980). S-Adenosylmethionine synthetase fromEscherichia coli. J. Biol. Chem. 255, 9082–9092.

26. Saraste, M., Sibbald, P. R. & Wittinghoffer, A. (1990).The P-loop a common motif in ATP and GTP-bind-ing proteins. Trends Biochem. Sci. 15, 431–434.

27. Lahti, R., Kolakowski, L. F., Jr, Heinonen, J., Vihinen,M., Pohjanoksa, K. & Cooperman, B. S. (1990). Con-servation of functional residues between yeast andE. coli inorganic pyrophosphatases. Biochim. Biophys.Acta, 1038, 338–345.

28. Kagan, M. R. & Clarke, S. (1994). Widespread occur-ence of three sequence motifs in diverse S-adenosyl-methionine-dependent methyltransferases suggestsa common structure for these enzymes. Arch. Bio-chem. Biophys. 310, 417–427.

29. Lopez-Vara, M. C., Gasset, M. & Pajares, M. A.(2000). Refolding and characterization of rat livermethionine adenosyltransferase from Escherichia coliinclusion bodies. Protein Expt. Purif. 19, 219–226.

30. Leslie, A. G. W. (1992). Molecular data processing. InCrystallographic Computing 5: From Chemistry toBiology (Moras, D., Podjarny, A. D. & Thierri, J. C.,eds), pp. 39–50, Oxford University Press, Oxford.

31. Collaborative Computational Project Number 4(1994). The CCP4 suite: programs for protein crystal-lography. Acta Crystallog. Sect. D, 50, 760–763.

32. Brunger, A. T. (1993). X-PLOR: a System for X-rayCrystallography and NMR, Yale University Press,New Haven.

33. Laskowsky, R. A., MacArthur, M. W., Moss, D. S. &Thornton, J. M. (1993). PROCHECK: a program tocheck the stereochemical quality of protein struc-tures. J. Appl. Crystallog. 26, 283–291.

34. Kraulis, P. J. (1991). MOLSCRIPT: a program to pro-duce both detailed and schematic plots of proteinstructures. J. Appl. Crystallog. 24, 946–950.

35. Merrit, E. A. & Bacon, D. J. (1997). Raster3D: photo-realistic molecular graphics. Methods Enzymol. 277,505–524.

Edited by R. Huber

(Received 24 January 2003; received in revised form 27 May 2003; accepted 30 May 2003)


Documents

Crystal structures of methionine adenosyltransferase complexed with substrates and products reveal the methionine-ATP recognition and give insights into the …