doi:10.1016/j.jmb.2009.12.054 J. Mol. Biol. (2010) 396, 1197–1210

Available online at www.sciencedirect.com

Glycine-Rich Loop of Mitochondrial ProcessingPeptidase α-Subunit Is Responsible for SubstrateRecognition by a Mechanism Analogous toMitochondrial Receptor Tom20

Klára Dvořáková-Holá1, Anna Matušková1, Martin Kubala2,Michal Otyepka3, Tomáš Kučera1, Jaroslav Večeř4, Petr Heřman4,Natalya Parkhomenko1, Eva Kutejova1,5 and Jiří Janata1⁎

© 2010 Elsevier Ltd. All rights reserved.

1Institute of Microbiology,Academy of Sciences of theCzech Republic, Vídeňská 1083,142 20 Prague 4,Czech Republic2Department of Biophysics,Faculty of Science, PalackyUniversity, tř. Svobody 26,CZE-771 46 Olomouc,Czech Republic3Department of PhysicalChemistry, Faculty of Science,Palacky University, tř. 17.listopadu 12, 771 46 Olomouc,Czech Republic4Faculty of Mathematics andPhysics, Institute of Physics,Charles University, Ke Karlovu5, 12116 Prague,Czech Republic5Institute of Molecular Biology,Slovak Academy of Sciences,Dúbravská cesta 21, 845 51Bratislava, Slovak Republic

Received 19 June 2009;received in revised form23 December 2009;accepted 28 December 2009Available online4 January 2010

*Corresponding author. E-mail addjanata@biomed.cas.cz.Abbreviations used: GRL, glycine

mitochondrial processing peptidasedynamics; pMDH, malate dehydrogProtein Data Bank.

0022-2836/$ - see front matter © 2010 E

Tryptophan fluorescence measurements were used to characterize the localdynamics of the highly conserved glycine-rich loop (GRL) of themitochondrial processing peptidase (MPP) α-subunit in the presence ofthe substrate precursor. Reporter tryptophan residue was introduced intothe GRL of the yeast α-MPP (Y299W) or at a proximal site (Y303W). Time-resolved and steady-state fluorescence spectroscopy demonstrated that forTrp299, the primary contact with the yeast malate dehydrogenase precursorevokes a change of the local GRL mobility. Moreover, time-resolvedmeasurements showed that a functionless α-MPP with a single-residuedeletion in the loop (Y303W/ΔG292) is defective particularly in the primarycontact with substrate. Thus, the GRL was proved to be part of a contact siteof the enzyme specifically recognizing the substrate. Regarding the surfaceexposure and presence of the hydrophobic patches within the GRL, weproposed a functional analogy between the presequence recognition by thehydrophobic binding groove of the Tom20 mitochondrial import receptorand the GRL of the α-MPP. A molecular dynamics (MD) simulation of theMPP–substrate peptide complex model was employed to test thishypothesis. The initial positioning and conformation of the substratepeptide in the model fitting were chosen based on the analogy of itsinteraction with the Tom20 binding groove. MD simulation confirmed thestability of the proposed interaction and showed also a decrease in GRLflexibility in the presence of substrate, in agreement with fluorescencemeasurements. Moreover, conserved substrate hydrophobic residues inpositions +1 and −4 to the cleavage site remain in close contact with the sidechains of the GRL during the entire production part of MD simulation asstabilizing points of the hydrophobic interaction. We conclude that the GRLof the MPP α-subunit is the crucial evolutional outcome of the presequencerecognition by MPP and represents a functional parallel with Tom20 importreceptor.

Keywords: mitochondrial processing peptidase; presequence; substrate

recognition; hydrophobic patch; Tom20 Edited by M. Sternberg

ress:

-rich loop; MPP,; MD, molecularenase precursor; PDB,

lsevier Ltd. All rights reserve

Introduction

More than 98% of mitochondrial proteins aresynthesized as precursor proteins on cytosolicribosomes and posttranslationally transported intothe mitochondria. Specific targeting of the matrix

d.

1198 Primary Contact of α-MPP Subunit with Substrate

precursors is in most cases facilitated by their N-terminal signal presequences, which are recognizedby a series of protein translocases. First of them, theTOM complex, is the translocase in the outermitochondrial membrane and represents the entrygate of mitochondria. This complex includes theTom20 protein, an import receptor exposed tocytosol that is responsible for the recognition andbinding of the presequences and therefore enablesthe precursors to enter the import machinery.1–4

Once the precursor has been translocated into thematrix, the targeting signal is no longer necessaryand is proteolytically removed. Mitochondrial pro-cessing peptidase (MPP; EC 3.4.24.64) represents thefinal component of the import machinery. MPP is ametallopeptidase located in the matrix, specificallycleaving off most of the N-terminal presequencesfrom precursor proteins.5–7 The enzyme consists oftwo homologous subunits α and β, both essential forits enzymatic activity. The crystal structure of yeastMPP demonstrated nearly identical protein archi-tecture of the two subunits; each consists of twodomains of ∼210 residues with very similar foldingtopology, which are connected by a flexible linker of∼20 residues.8 The catalytic center containing a zincbinding motif is conserved only in the β-subunit andis localized in a polar internal cavity inside theenzyme dimer. Although the roles of each subunit insubstrate initial binding are still a matter ofcontention, most studies have emphasized therelevancy of the α-subunit in substrate bindingand recognition.9–11

Similarly to Tom20, MPP specifically recognizes alarge variety of diverse mitochondrial presequencesthat share a low sequence similarity and vary inlength, in most cases within the range of 20–50residues.6,12,13 Nevertheless, the recognized mito-chondrial presequences display common physico-chemical properties such as a positive charge andthe potential to form an amphiphilic α-helix.14,15

Abe et al.16 demonstrated that whereas the positivecharge of the presequences is required for somecomponents of the import machinery, binding to theTom20 receptor is mediated mainly by hydrophobicrather than ionic interactions. The NMR structure ofthe cytosolic domain of the rat Tom20 in complexwith the aldehyde dehydrogenase presequencepeptide demonstrated that the bound presequenceforms an amphiphilic helical structure with thehydrophobic side interacting with a hydrophobicgroove of the receptor.16 Based on fluorescenceresonance energy transfer measurement, a ratherstructure-less presequence form was predicted tobind the MPP enzyme.17 This prediction is consis-tent with the crystal structure of yeast MPP showinga presequence peptide bound in an extendedconformation within the dimer cavity,8 as well aswith the fluorescence resonance energy transfermeasurements investigating the spatial orientationof MPP and a presequence.18 Nevertheless, thepresequence fixation to the catalytic site of the β-subunit and to the β-subunit generally by strongelectrostatic interactions reflects the later steps of the

substrate–enzyme interaction. Regarding the factthat the catalytic center is localized deep inside theenzyme cavity, these binding sites can hardlyparticipate in the primary interaction with theprecursor presequence. However, a clear evidenceof the particular region(s) of the enzyme that attractsthe presequence and is responsible for the primarycontact with the substrate has not yet been reported.Similarly, the conformation of the substrate prese-quence at the moment of the initial interaction hasnot yet been clearly defined.We previously anticipated19 that a candidate for

the “primary contact site” is the most conserved partof the α-MPP—the glycine-rich loop (GRL). Thisflexible region is situated at the entrance into theMPP dimer cavity and is exposed to both theenzyme catalytic center and the substrate. Thehighly conserved glycine-rich motif (consensussequence HfbHfbGGGGSFSAGGPGKGMF/YSRLYT/LXVLN; Hfb indicates hydrophobic resi-dues; residues of the intrinsic unstructured GRL arein boldface) is present in all known α-MPPs. In thebest characterized yeast α-MPP, it occupies posi-tions 282–308. Besides the characteristic glycineresidues, there are also several conserved hydro-phobic and/or aromatic amino acid residues. Theintrinsic fully unstructured GRL is bordered on oneside by a short β-sheet structure with a cluster offour glycines and on the other side by a short α-helix. Nagao et al. reported that point mutations inthe GRL notably affected the MPP activity; inparticular, deletions completely destroyed the en-zyme activity as well as affinity for the presequenceof mouse malate dehydrogenase precursor(pMDH).20 It appears that the essential function ofthe α-subunit is localized just in this flexible region.The aim of the present study was to verify the

participation of the GRL region in the initial contactbetween MPP and a precursor presequence and tounderstand the mechanism of the recognition. Inorder to characterize the local dynamics of the GRLin the presence of the substrate precursor, wedescribed the interaction using steady-state andtime-resolved tryptophan fluorescence spectrosco-py. Moreover, we took advantage of moleculardynamics (MD) simulations to evaluate a newhypotheticalmodel ofMPPpresequence recognition.

Results

Site-directed mutagenesis of the GRL in theyeast α-MPP

Reporter tryptophan residues were individuallyintroduced into two positions within the GRL or inits vicinity to prepare a system for monitoring thedynamics of the GRL using tryptophan fluorescenceassays. The mutations were introduced into the α-MPP*—the previously prepared α-MPP form with areduced number of native tryptophan residues19

(described in Expression vectors and protein

Table 1. Activities of MPPs mutated in GRL of α-subunit

α-MPP mutantform

Relativeenzymatic activity(% of wt activity)

Tryptophanresiduesa

α-MPP*b 100 Trp223α-MPP*-Y303W 100 Trp303, Trp223α-MPP*-Y299W 60 Trp299, Trp223α-MPP*-Y303W/

ΔG292b0.5 Trp303, Trp223

a Introduced reporter Trp residues are in boldface.b α-MPP*=α-subunit with minimized number of native

tryptophan residues (see Expression vectors and proteinproducts).

1199Primary Contact of α-MPP Subunit with Substrate

products). The positions for the mutations werechosen in order to place the fluorescence markers inoptimal locations for the interaction measurements,with minimal steric modifications of the side chain.Individual substitutions of aromatic amino acidresidues within the GRL region (Y299W andY303W) were performed (Fig. 1). The introductionof Trp299 residues directly to the GRL led to adecrease in MPP activity; nevertheless, a significantportion of the enzyme activity (∼60%) remainedpreserved (Table 1). On the other hand, theintroduction of the reporter residue within a shortconserved α-helix immediately adjacent to the GRL(Y303W; Fig. 1) did not influence enzymatic activity

Fig. 1. Structure of the yeast MPP dimer. PDB with ID1HR6.6 The figure was generated by PyMOL software. (a)α-Subunit (blue) and β-subunit (gray) form the internalenzyme cavity. The glycine-rich site of the α-MPP (red;residues between positions 284–299) formed by anunstructured loop and a short β-sheet structure with acluster of four glycines is exposed to both the enzymesurface and to the catalytic center of β-MPP (turquoise)and connected with the short α-helix (green; residuesbetween positions 300 and 308). (b) Details of the glycine-rich site of the α-MPP. Tyr299 and Tyr303 wereindividually replaced by reporter Trp residues. Gly292(yellow) was deleted in the α-MPP*-Y303W/ΔG292mutant form.

at all (Table 1). Consequently, we combined thismutation (Y303W) that does not affect proteinfunction with the deletion of a single glycine residuein the center of the GRL (ΔG292) that was expectedto affect the protein function. Indeed, the MPP dimercomprising the α-MPP*-Y303W/ΔG292 did notshow any detectable activity (Table 1). Thus, twoforms of α-MPP carrying the reporter residueTrp303 were prepared to observe the GRL dynam-ics: the fully functional and the functionless one withthe conformation of the loop modified by itsshortening.

Time-resolved fluorescence spectroscopy

A partial system employing the α-MPP subunitalone was chosen for monitoring the initial steps ofenzyme–substrate interaction that eliminates theinterfering influence of β-MPP. The presence of theβ-subunit in the system would not allow us todistinguish the weak primary interaction of sub-strate recognition by α-MPP from the strong finalfixation of the presequences to the catalytic β-subunit by electrostatic interactions.Measurements of the time-resolved fluorescence

anisotropy revealed that the α-MPP* subunit con-taining a single Trp223 residue exhibits a highlyheterogeneous anisotropy decay with three rota-tional correlation times (Table 2, line 1). The shortcorrelation times ϕ̄1 and ϕ̄2 can be associated withrapid Trp223movements that include a combinationof subnanosecond “in-cone” movement and itssegmental motions that typically occurs on thenanosecond time scale. For rigid globular proteins,the longest correlation time ϕ̄ 3 should reflectrotational diffusion of the protein as a whole.

Table 2. Fluorescence anisotropy decays of α-MPP* in TNbuffer, pH 8.6, with 0%, 10%, or 15% (v/v) glycerol

Glycerolτ̄

(ns)a β1

ϕ̄ 1(ns) β̄ 2

ϕ̄ 2(ns) β̄ 3

ϕ̄ 3(ns)

0% 6.21 0.090 b0.3 0.018 1.2±0.3 0.112 7±210% 3.28 0.051 b0.3 — — 0.179 23±515% 3.26 0.011 b0.3 — — 0.219 34±5

Uncertainties are estimated frommultiplemeasurements (n=3–5).a Mean fluorescence lifetime (±0.03 ns).

1200 Primary Contact of α-MPP Subunit with Substrate

This correlation time should scale with the molec-ular mass of the protein. Interestingly, the mea-sured value of ϕ̄3=7 ns does not correlate with thesize of α-MPP* (54.3 kDa). It rather corresponds tothe rotation of a smaller body with a molecularmass not larger than about 15–25 kDa,21 hence notthe entire subunit. From this observation, weconclude that α-MPP* exhibits a high degree ofinternal flexibility and the Trp-carrying domainmoves almost independently.In order to detect the dynamics of the GRL in the

double Trp mutants, it was essential to reduce theinterfering depolarization caused by Trp223 motion.For this reason, a fraction of glycerol was added tothe sample. Figure 2 shows anisotropy decays of α-MPP* in the presence of a varying amount ofglycerol. It is seen that an increasing amount ofglycerol causes rigidization of α-MPP* as judgedfrom a lower fluorescence depolarization. Table 2shows that in the presence of glycerol, the anisotro-py decay of α-MPP* became essentially bimodal. Inparticular, the amplitude β̄1 of the fastest sub-nanosecond motion decreases and the amplitude β̄2attributed to the nanosecond segmental flexibilityfalls below a detectable limit. Moreover, ϕ̄3 isincreased from 7 to 23 ns at 10% glycerol. Theseresults indicate that the presence of 10% glycerolsubstantially restricts motional freedom of Trp223and slows down the motion of the Trp-containingdomain of α-MPP*. The disappearance of thenanosecond correlation component opens a timewindow for the observation of a GRL movement inthe Y303W and Y299W mutants. At the same time,the activity of α-MPP* and both mutants remainsunchanged in the presence of 10% glycerol (data notshown). All further experiments were thereforedone in the presence of this glycerol concentration.The addition of 10% (v/v) glycerol to the buffer

causes about 30% increase of the solvent viscosityand about 3% decrease of the dielectric constant.These changes cannot fully account for the observed

increase of ϕ̄3 and for the decrease of the meanfluorescence lifetime. We propose that some specificinteraction of glycerol with the protein should takeplace. Fortunately, this interaction does not affectthe protein functionality and we can therefore drawproper conclusions about the interactions of thesubstrate with α-MPP* and its mutants. Impor-tantly, glycerol also suppressed the tendency foraggregation of the double Trp mutants.Fluorescence intensity and anisotropy decays

were used for examination of the effect of substratebinding on the mobility of the GRL. We used Trp-free yeast pMDH as a substrate. The results obtainedfor α-MPP* and both Y303W and Y299W mutantsare listed in Table 3 and are summarized below:

• Native Trp223 of the α-MPP* subunit—Addition of the substrate causes both a signif-icant decrease in the fluorescence lifetime and arestriction in the subnanosecond Trp motion asseen by the decrease in the β̄1 amplitude. Theseobservations indicate that the protein binds thesubstrate (lines 1 and 2). From our experiments,we are unable to decide whether the decreasedrotational freedom of Trp223 is caused by abinding-induced conformational change of theprotein or by direct steric hindrance caused bythe substrate binding.

• Compared to the α-MPP* subunit, a newnanosecond correlation time ϕ̄2 appeared inthe anisotropy decay of the α-MPP*-Y299Wmutant. The same pattern was observed bothin the presence and in the absence of thesubstrate (lines 3 and 4). ϕ̄2 should be linked tothe motion of Trp299 located on the GRL. ThepMDH addition caused a drop of the β̄2amplitude associated with this motion from0.020 to 0.012. The amplitude β̄1 linked to thefastest motion decreases as well. Binding-induced changes of both the raw anisotropydata and correlation time distributions are

Fig. 2. Fluorescence anisotropydecays of α-MPP* in TN buffer,pH 8.6, containing 0% (△), 10% (●),or 15% (○) glycerol. Continuousline represents the best fit. Excita-tion, 298 nm; emission, 360 nm. Thetime scale is 36.5 ps/channel.

Table 3. Anisotropy decay parameters of α-MPP* with reporter tryptophan residues in GRL

Sample τ̄ (ns)a β̄ 1 ϕ̄ 1 (ns) β̄ 2 ϕ̄ 2 (ns) β̄ 3 ϕ̄ 3 (ns)

α-MPP*b 3.28 0.051 b0.3 — — 0.179 23±5α-MPP*+S 2.89 0.017 b0.3 — — 0.213 28±5α-MPP*-Y299W 3.74 0.019 b0.3 0.020 1.6±0.3 0.191 32±5α-MPP*-Y299W+S 3.52 0.005 b0.3 0.012 1.0±0.3 0.213 38±5α-MPP*-Y303W 4.04 0.008 b0.3 — — 0.222 44±5α-MPP*-Y303W+S 3.92 — — — — 0.230 60 (−10, +∞)

+S, in the presence of pMDH substrate.Uncertainties are estimated from multiple measurements (n=3–5).Results were obtained in the presence of 10% glycerol.

a Mean fluorescence lifetime (±0.03 ns).b α-MPP*=α-subunit with the single Trp223 residue (see Expression vectors and protein products).

1201Primary Contact of α-MPP Subunit with Substrate

presented in Fig. 3. We conclude that theprotein binds pMDH and the complex forma-tion causes restriction of the GRL flexibility.

Difference in overall rotation of α-MPP* and α-MPP*-299W reflected by ϕ̄3 is marginal withinuncertainties reported in Table 3.

Fig. 3. Fluorescence anisotropydecays of the α-MPP*-Y299W mu-tant form in TN buffer, pH 8.6,containing 10% glycerol. (a) Anisot-ropy decays in the absence (●) andin the presence (○) of substrate.Continuous lines are the best fits.The time scale is 36.5 ps/channel.(b) Analysis of the fluorescenceanisotropy decay data. Distribu-tions of correlation times corres-ponding to the best fits from Fig. 3ain the absence (broken line) and inthe presence (continuous line) ofsubstrate.

1202 Primary Contact of α-MPP Subunit with Substrate

• Table 3 shows that the incorporation of Trp303into the α-MPP* subunit does not qualitativelychange the character of the anisotropy decay,which remains bimodal. Lower β̄1 and longerϕ̄3 could indicate that unliganded α-MPP*-Y303W is slightly more rigid than free α-MPP*.Substrate addition causes further rigidizationof the α-MPP*-Y303W mutant as judged fromthe complete disappearance of the fast anisot-ropy component ϕ̄1. The mean fluorescencelifetime of α-MPP*-Y303W slightly decreasesupon ligand binding as well.

Different behavior was observed for the function-less α-MPP*-Y303W/ΔG292 mutant where rawanisotropy decays in the presence and in the absenceof the substrate fully overlapped. The anisotropydecay was monoexponential, and after the substrateaddition, the longest correlation time ϕ̄3 remainedessentially the same (46±5 ns versus 48±5 ns). Thefluorescence lifetime remained unaffected as well(τ̄ =3.95±0.005). This result provides strong evi-dence that the functionless mutant with the deletionwithin the GRL does not bind the pMDH substrate.These results clearly demonstrate that even in the

absence of the catalytic β-subunit, the α-subunititself is capable of substrate binding and this bindingis connected with changes in the local GRL mobility.

Steady-state fluorescence spectroscopy

Results of steady-state fluorescence measurementof the reporter tryptophans in the α-MPP subunitare summarized in Table 4. The apparent emissionspectrum of Trp303 exhibited a maximum at339 nm, suggesting a moderate exposure of theresidue to the solvent. Notably, emission of Trp303was not substantially changed in theΔG292 mutant,indicating that the deletion of the glycine residuedid not radically affect GRL folding. The position ofTrp299 is much more exposed to the solvent, asrevealed by its maximum at 347 nm. The interactionof the α-MPP mutants with 3-fold molar excess ofthe yeast pMDH resulted in a 9-nm blue shift of theTrp299 spectrum (maximum, 338 nm). The shiftindicates burying of the tryptophan within a proteinor its interaction with some of the hydrophobicresidues of the substrate presequence. Only insig-nificant changes were observed in the fluorescenceof Trp303 following substrate binding. Magnitudes

Table 4. Maximum of the emission spectrum of thereporter tryptophan residue in GRL

α-MPP mutantform

Reportertryptophanresidue

Maximum of theemission

spectrum (nm)

Withoutsubstrate

Withsubstrate

α-MPP*-Y299W Trp299 347 338α-MPP*-Y303W Trp303 339 340α-MPP*-Y303W/ΔG292 Trp303 338 340

of the spectral shifts and steady-state intensitychanges correlate well with the extent of thebinding-induced lifetime changes.The intensity of fluorescence decreased after the

substrate binding in all mutants. This finding is inagreement with the time-resolved measurements,where substrate binding caused a decrease in themean fluorescence lifetime (Table 3).

MD simulations

To interpret the data obtained from the fluores-cence measurements and to gain insight into theinitial interaction of the GRL with the substratepresequence at an atomic level, we performed MDsimulations. The Tom20 receptor represents theinitial contact site of the presequence on the outermitochondrial membrane. Analogously, MPP recog-nizes an identical set of presequences after theimport into the mitochondria. One could expect thatthese two components of import machinery couldemploy a similar mechanism in recognition of thepresequence. We took advantage of the fact that thestructure of the rat ALDH presequence peptide(residues 12–22) in a complex with the Tom20receptor was previously elucidated by NMR.16 Forthe initial setting of the MPP–substrate complexmodel, we used the conformation of the ALDHpresequence peptide identified in the complex withTom20 receptor. Initial fitting of the GRL–peptideinteraction, that is, positioning and orientation of thepeptide relative to the GRL surface, was based onthe analogy with the binding of the Tom20 bindinggroove: the hydrophobic residues (Leu15, Leu18,and Leu19) of the peptide amphiphilic α-helix wereoriented to the hydrophobic spots on the GRLsurface. Slight adjustments of the peptide structureand position with respect to the GRL surfaceshowed a good complementarity of the GRL surfaceand the peptide (Fig. 4).The all-atomic classical molecular simulation with

explicit solvent was chosen as our tool. Twosimulations were set up, (i) MPP with the interactingsubstrate peptide and (ii) MPP without substrate forcomparison. All discussed trajectories were stableduring the production parts of the MD simulations.Trajectory stability was monitored and confirmedby the analysis of the secondary-structure elements,the radius of gyration, the RMSD of backbone atomsfrom the crystal structure, and the total energy (datanot shown). The curve of the backbone atoms'RMSD from the X-ray structure against time reachedits plateau at ∼2 ns (MPP without peptide) and∼1 ns (MPP with peptide) fluctuating around amean value of 2.2 Å.Snapshots from MD simulation of the MPP–

pALDH interaction are shown in Fig. 5. Initially,the presequence peptide structure relaxed in ∼1 nsand its initial α-helical fold melted (Fig. 5a).However, the peptide has remained in contactwith MPP with no sign of dissociation during theentire production part of MD simulation (23 ns).Although the interaction is dynamic, the side chains

Fig. 4. Representation of the initial fitting of the α-MPPGRL with the presequence peptide. The all-atomic MDsimulation of the yeast MPP dimer interaction with the ratALDH presequence peptide (residues 12–22) was per-formed using the elucidated structures of both interactingpartners.8,16 The molecular surface of the GRL is drawn ingray (residues 284–308 of the α-MPP). The bound ratALDH presequence peptide (residues 12–22) is drawn as ayellow tube with side chains for two highly conservedhydrophobic residues Leu15 (red) and Leu19 (purple),respectively, and in blue for non-conserved Leu18.

1203Primary Contact of α-MPP Subunit with Substrate

of the two leucines oriented toward the GRL, Leu15and Leu19 (positions −4 and +1 to the cleavage site),remained in close contact with the side chains of theGRL during the entire simulation. Leu18 (position−1 to the cleavage site) was fluctuating and not incontact with the GRL in the initial 10 ns of thesimulation (Fig. 5a and b). Nevertheless, in the lastthird of the production phase, Leu18 came to theproximity of the GRL. Finally, the α-helical confor-mation of the precursor peptide is partially refolded(see Figs. 5c and 4).The flexibility of the system was monitored using

temperature B-factors calculated from the last 1 ns ofthe simulations. In general, the B-factors calculatedfrom the MD simulation agree well with thoseextracted from crystallographic data. The highestpeaks in the B-factor plot (Fig. 6) correspond to themost flexible regions, namely, (i) residue G251, (ii)residue A291 at the tip of the GRL, and (iii) residuesat the termini of both subunits. The presence of thepeptide substrate leads to a significant decrease inthe B-factors of the GRL, for example, from BA291=52in the free MPP to BA291=23 in the MPP–substratepeptide complex (note: BY299=17 free MPP; 13 withpeptide). The presence of substrate peptide in theproximity of the GRL decreases the flexibility of theloop, restricting its conformational space.To verify the data obtained from the MD

simulation of the MPP–ALDH presequence peptideinteraction for the homologous yeast MPP–pMDHsystem employed in the time-resolved and steady-state fluorescence spectroscopy experiments, weperformed the following simulation: an MD simu-

lation of yeast MPP–ALDH presequence peptidesystem was run for an initial 2 ns. After peptidestructure relaxation, amino acid side chains weresubstituted for appropriate residues of the yeastMDH presequence. The MD simulation was contin-ued for an additional 24 ns.MD simulation of the MPP–pMDH interaction

confirmed the general feature observed previouslyfor the MPP–pALDH system. The presequencepeptide remained in contact with MPP, due to thehydrophobic interaction between the side chain ofthe residue Ala6 (position −4 to the cleavage site)and the GRL, during the whole period of simulation.Similarly, with the MPP–pALDH system, the sidechain of Phe10 (position +1 to the cleavage site)remained oriented toward theGRL for the first∼6 nsof MD simulation, confirming primary contact of theconservative residue in position +1 with the GRL.Subsequently, the side chain of Phe10 was slowlymoving away from the GRL in two steps.

Discussion

Our previous results of MPP tryptophan fluores-cence measurements showed that the presence ofthe substrate evokes a conformational change of theα-MPP, but not β-MPP.11 Subsequently, we dem-onstrated a translocation of both the N- and C-terminal domains of α-MPP evoked by thesubstrate.19 In this study, measurements werefocused on the local dynamics of the conservedGRL of the yeast α-subunit. The fluorescenceexperiments have resulted in two main findings.First, a change of the local GRL mobility is clearlydetected upon the substrate addition. Second, even aminimal shortening of the GRL makes the proteinunable to bind the substrate.Glycine-rich elements are described in enzymes

that recognize a variety of protein substrates(protein kinases) as universal binding sites, inwhich a high number of the glycine residues isnecessary to avoid steric restrictions for any poten-tial substrate.22 The crystallographic evidence of aweak electron density for the GRL in the wild-typeMPP indicates that the loop is highly flexible.8 Asexpected, an internal dynamics of the reporterresidue Trp299 located directly in the loop wasdetected both in the absence and in the presence ofsubstrate. A change in the loop dynamics aftersubstrate addition was clearly proved. The decreasein the protein internal motion induced by thesubstrate binding is significant and reflects aprimary contact of the GRL with the presequence.The results of the time-resolved fluorescence experi-ments agree with the results of the MD simulation.Moreover, a change in the local microenvironmentof Trp299 that is located on the loop interacting withthe substrate was shown by steady-state fluores-cence measurements. The significant blue shift of theTrp299 emission could result from a binding-induced conformational change of the loop or froma direct shielding of Trp299 by the substrate.

Fig. 5. MD simulation of the α-MPP GRL interaction with the ALDH presequence peptide. A detailed structure of theα-MPP GRL interaction (residues 284–308 gray tube) with the substrate peptide is shown (yellow tube). The boundpeptide is drawn as a tube with side chains for two highly conserved hydrophobic residues Leu15 (red) and Leu19(purple), respectively, and for non-conserved Leu18 (blue). The MD simulation is shown for three different snapshots(23 ns; 1 frame=2 ps): (a) T=1346th frame (∼2.7 ns), (b) T=1927th frame (∼3.8 ns), and (c) T=9751th frame (∼19.5 ns).Tyr303 residue of the GRL is in green and Tyr299 is in orange. The distances between atoms of the three marked Leuresidues of the peptide and of Tyr303 during the interval of the MD simulation are summarized in (d)-(f): (d) distancebetween γ-carbon of Leu18 andΔ-carbon of Tyr303, (e) γ-carbon of Leu19 and the γ-carbon of Tyr303, and (f) γ-carbon ofLeu15 and the ɛ-carbon of Tyr303. Three vertical broken lines in (d)–(f) mark the snapshots at frames 1346, 1927, and 9751corresponding to the structures shown in (a)–(c).

1204 Primary Contact of α-MPP Subunit with Substrate

Both the steady-state and the time-resolvedfluorescence measurements of Trp303 indicatedthat a short α-helix immediately adjacent to theGRL, where this reporter is situated (Fig. 1b), is onlyslightly affected by ligand binding. In other words,the conformational change of the GRL induced bythe substrate binding is probably localized to the

GRL and does not propagate into its vicinity.However, the Trp303 served as an excellent obser-vation point for substrate binding using the time-resolved measurements. In the case of the fullyactive α-MPP*-Y303W, an increase in the longestcorrelation time (ϕ̄3) after substrate addition clearlyindicates binding of the 30-kDa-sized yeast pMDH.

Fig. 6. Comparison of backboneB-factors as calculated from the lastnanosecond of the MD simulationsof the free MPP (blue) and MPPwith ALDH precursor peptidebound (green) with that extractedfrom the X-ray structure (1HR6,chains A and B); numbering fromthe MD simulation is used. Foreasier comparison, the B-factorswere scaled yi′=(yi−min(y))/s(y),where s represents the RMSD.

1205Primary Contact of α-MPP Subunit with Substrate

On the contrary, only insignificant change of ϕ̄3 wasobserved for the α-MPP*-Y303W/ΔG292 mutantwhere the α-subunit with the shortened GRL isdefective in substrate binding. Thus, comparison ofa pair of the α-MPP mutant forms carrying the samereporter tryptophan (Y303W) but differing in theglycine-loop length provides evidence of the essen-tial role of the GRL in the initial contact with thesubstrate. Accordingly, activity assays using yeastpMDH as a substrate showed that deletion of even asingle glycine residue of the GRL (Y303W/ΔG292)resulted in complete loss of enzyme activity.Interestingly, Kitada et al. recently reported thatthe GRL is required only for the processing of longpresequences: MPP enzyme containing α-MPP withthe whole GRL deleted (positions 284–296) was stillable to cleave yeast pMDH, albeit with a lowerprocessing activity than the wild type.23 Thus, thesignificant difference between the published resultsand our data is unexpected and indeed interesting.As our experimental results clearly proved that

the GRL is the “primary contact site” of the MPPenzyme with the yeast pMDH precursor, weformulated a new hypothesis concerning substraterecognition by MPP: Regarding the surface-exposedlocalization of GRL and the presence of hydrophobicpatches on the GRL surface, we proposed ananalogy between presequence recognition by thehydrophobic binding groove of the Tom20 importreceptor and the GRL of α-MPP. As both systemswere designed to recognize the same set of signalpresequences, they might share a common “ID codefor entrance to a system”. Tom20 recognizes thepotential of the presequence to form an amphiphilicα-helix and the presequence binding is mediatedprimarily by hydrophobic interactions.16,24 Wepropose that these features could be the recognition“ID code” required also by MPP enzyme and thatthe GRL (similarly to Tom20 receptor) induces theformation of a transient helical structural motif inthe approaching presequence, whereas the struc-ture-less form is assumed to be present in the cytosolor mitochondrial matrix.17

To support the results of the fluorescence experi-ments indicating interaction of the GRL with thetransient α-helix of the presequence, we employedMD simulation. For the rat pALDH peptide, the MDsimulation displayed stability of interaction be-tween the substrate peptide and the GRL surfaceduring the entire production phase (23 ns). Initial α-helical conformation of the presequence peptidetemporarily melted, but in the last third of theproduction phase, its α-helix was partially refolded.Final conformation of the pALDH in GRL hassimilar features (e.g., helical character of thepresequence peptide, orientation of Leu residues,and contact of both interacting surfaces) to theanalogous and inspiring system pALDH in Tom20receptor given by NMR structure.16

Similarly, in the case ofMDHpresequence peptide,the Ala6 in position −4 to the cleavage site remains incontact with the GRL during the entire productionphase (24 ns). However, the other interacting sidechain of Phe10 (position +1 to the cleavage site) wasoriented toward the GRL only for the first ∼6 ns,later moving slowly away. It is possible that thisconformation change may reflect the subsequentsteps in the MPP–pMDH interaction. Nevertheless,we are reluctant to interpret the differences observedin the latter part of bothMD simulations just becauseof the modeling of the MDH presequence structure.Simulations could reflect real differences in thesubsequent mechanism of the MPP–presequenceinteraction and may even explain different require-ment of the GRL for recognition of various substratesdescribed previously23 but they may have arisen outof the modeling of the MDH presequence structure.Without any doubt, both simulated systems empha-sized an impact of the hydrophobic interaction oftwo conservative hydrophobic residues in positions−4 and +1 to the cleavage site of the substrate withthe surface of the GRL.Interestingly, MD simulations show that these

hydrophobic residues in position +1 are in theproximity of the GRL residues Met298 and Tyr299during the simulation. This direct shielding of

1206 Primary Contact of α-MPP Subunit with Substrate

Tyr299 by the conserved bulky hydrophobic resi-dues can explain the significant blue shift of theTrp299 spectrum observed by the steady-statefluorescence measurements in the presence of thepMDH substrate. Similarly to position +1, hydro-phobic residues are frequent at positions −4 and/or−5 to the substrate cleavage site (Leu15 in pALDH;Val5 and Ala6, in pMDH). Thus, the conservedhydrophobic residues in positions +1 and −4oriented toward the GRL surface during thetransient interaction appear to be the stabilizingpoints of the initial hydrophobic interaction betweenthe signal presequence and the GRL, the substrate-recognizing element of the α-MPP. In agreementwith the time-resolved fluorescence measurements(decrease of ϕ̄2 of the Trp299), the MD simulationconfirmed a decrease in the loop flexibility in thepresence of substrate (decreased value of B-factor).Similarly as suggested for Tom20 substraterecognition,25 the orientation of the presequenceconserved hydrophobic residues towards the GRLsurface facilitates to display the conserved positivelycharged arginine residues as docking surface forother, negatively charged components of the cata-lytic β-MPP subunit participating in the prese-quence fixation by strong electrostatic interaction.We hypothesize that the GRL of the MPP α-

subunit is the crucial evolutional outcome ofpresequence recognition by MPP and it representsa functional parallel with the Tom20 importreceptor. Sequence homology cannot be expectedas the two recognition systems have developedfrom different evolutionary ancestors. Recently, anexample of the evolution of mitochondrial prese-quence recognition was presented, where differentorganisms adapted distinct proteins to fulfill thesame function:26 Although Tom20 from plants andanimals presumably evolved from two distinctancestral genes, the need to bind equivalentmitochondrial presequences has driven the conver-gent evolution of two distinct proteins to acommon structure and function. In case of the GRLof α-MPP, there are no structural features equivalentto the binding site of Tom20s, but we suggest ananalogy in the participation of hydrophobic patchesof both sites and thus in the principle of substratebinding.The proposed analogy relates just to the moment

of the presequence recognition, that is, the initialstep of the protein–protein interaction. Obviously,the immediately succeeding steps are different inthe two systems. The detailed mechanism ofpresequence translocation from the GRL of α-MPPto the distant active site at the β-MPP subunitaccompanied by unfolding of the unstable α-helixof the presequence remains to be elucidated. Verypreliminary data of the MD simulation of a longerperiod of enzyme–substrate interaction indicatemelting of the secondary structure of the substrateand shifting of the presequence deeper inside theenzyme cavity, but other modeling tools andexperimental approaches should be used to verifythis movement.

Materials and Methods

Expression vectors and protein products

The pETYB construct27 was used for the preparation ofthe mature form of yeast β-MPP and the pETYAHconstruct19 for the production of the N-terminally His-tagged yeast α-MPP. The pETYA-W147N/W481Yconstruct19 was used to produce the α-MPP-W147N/W481Y, His-tagged yeast α-subunit with the number ofnative tryptophan residues minimized to a single remain-ing Trp223, which was shown as essential for proteinsolubility.19 The α-MPP-W147N/W481Y protein isdenoted in the following text as α-MPP*. The pETyPMDHplasmid carrying the yeast full-length mdh sequence wasused for the preparation of the MPP substrate—yeastpMDH.19 The plasmid encoding the C-terminally His-tagged yeast pMDH (pETyPMDH-H) was prepared byPCR from pETyPMDH. The primers generating appro-priate restriction sites were the following: forward, 5′aactgcagcatatgttgtcaagagtagctaaa3′ (NdeI site underlined)and reverse, 5′ggaattctcgagtttactagcaacaaagttgac3′ (XhoIsite underlined). The PCR product was ligated into theNdeI/XhoI doubly digested pET42b(+) vector (Novagen)and sequenced.

Introduction of reporter tryptophan residues

A PCR-based strategy was used to prepare a cassetteallowing the replacement of 74 bp in the sequencecoding for the whole glycine-rich region and its closevicinity. A pair of silence mutations generating restric-tion sites (PstI and AatII) that border the region wasintroduced. First PCR was performed using primersBsaI-PstIfor (5′atcggatttgaaggtctccctatagatcatccagatatc-tatgctttagcaacactgcagactttacttggaggtgg3′) and AatIIrev(5′gtaatactggttcaagacgtcggtatacaaacgagaata3′); a secondPCR was performed using the PCRI product and primerBseRIfor (5′caaccgcttgaggagctgcctgcggtatacaagataatga3′).The PCRII product was digested by BsaI and BseRI andligated into a doubly digested construct pETA-HB [pET28b(+) containing the HindIII-BamHI fragment of theyeast α-MPP DNA sequence]. Subsequently, the HindIII-BamHI fragment was transferred into pBSII SK vector(Stratagene) to yield a plasmid construct with thecassette in which both PstI and AatII were unique sites.The fragments with cohesive ends (PstI-AatII), formed

by hybridization of two complementary oligonucleo-tides carrying the respective mutation (Table 5), wereligated to the cassette (replacing part of the cassette).Finally, the BsaI-BamHI fragment excised from thecassette was ligated into the doubly digested (BsaI/BamHI) construct pETYA-W147N/W481Y. Using thisstrategy, we separately prepared substitutions Y299Wand Y303W as well as the double mutation Y303W/ΔG292. TheDNA sequences of allmutantswere confirmedby sequencing.

Preparation of yeast MPP subunits and pMDH

The yeast pMDH precursor and β-MPP were producedin Escherichia coli BL21(DE3) strain transformed withpETyPMDH-H or with pETYB, respectively. His-taggedα-MPP was produced in E. coli BL21(DE3) co-transformedby vectors pETYAH and pGroESL (DuPont; vectorbearing sequences coding for GroES and GroEL chaper-

Table 5. Mutation oligonucleotides used for introduction of reporter Trp residues

Substitution Oligonucleotides

Y299W 5′gactttacttggaggtggtggttctttcagtgctggtgggccaggaaaggggatgtggtcccgtttgtatacccacgt3′5′gggtatacaaacgggaccacatcccctttcctggcccaccagcactgaaagaaccaccacctccaagtaaagtctgca3′

Y303W 5′gactttacttggaggtggtggttctttcagtgctggtgggccaggaaaggggatgtattctcgtttgtggacccacgt3′5′gggtccacaaacgagaatacatcccctttcctggcccaccagcactgaaagaaccaccacctccaagtaaagtctgca3′

Y303W/ΔG292 5′gactttacttggaggtggtggttctttcagtgctgggccaggaaaggggatgtattctcgtttgtggacccacgt3′5′gggtccacaaacgagaatacatcccctttcctggcccagcactgaaagaaccaccacctccaagtaaagtctgca3′

1207Primary Contact of α-MPP Subunit with Substrate

onins). Mutant forms of α-MPP were produced accord-ingly using appropriate constructs. Transformed E. colicells were grown on LB medium (1000 ml) at 37 °C toOD600=0.6, induced with 0.4 mM IPTG, and cultivated foradditional 2 h at 37 °C or for 36 h at 12 °C in the case of theα-MPP and derived mutant forms to assist proper proteinfolding.28

Protein purification

The N-terminally His-tagged α-MPP forms were isolat-ed using metal affinity chromatography. The cell pelletwas resuspended in TN buffer, pH 7.4 (20 mM Tris–HCland 100 mMNaCl), containing 20% glycerol, disrupted bysonication, and centrifuged at 15,000g for 15 min. Thesupernatant was loaded on a 1-ml Hi-Trap Chelatingcolumn (Amersham) equilibrated with TN buffer contain-ing 20% glycerol. The α-MPP was eluted with TN buffer,pH 7.4, containing 20% glycerol and 250 mM imidazole,and desalted on a Sephadex G-25 column (Sigma-Aldrich)using TN buffer, pH 8.6. The insoluble β-MPP and yeastpMDH precursor were isolated from inclusion bodies. Inthe case of the β-MPP, the solubilization and renaturationprocedure was used as described by Géli.29 C-terminallyHis-tagged yeast pMDH was purified using a Hi-TrapChelating column (Amersham) equilibrated with dena-turation buffer (20 mM Tris, pH 7.4, 20 mMNaCl, and 8 Murea), eluted by 250 mM imidazole in denaturation buffer,and renatured by dialysis for 18 h at 4 °C against 5×1 L ofTN buffer, pH 8.6.

Activity assay and protein quantification

Each α-MPP mutant subunit was mixed with the wild-type MPP β-subunit to reconstitute the enzyme complexand purified as described previously.19 MPP dimers weretested for activity using previously described protocols.19

Yeast pMDH (naturally with no tryptophan residues) wasused as a substrate and the reaction products wereanalyzed on 10% SDS-PAGE. Protein concentration wasquantified using the BCA Protein Assay Reagent Kit(Pierce).

Time-resolved fluorescence spectroscopy

An apparatus with time-correlated single photoncounting detection equipped with a cooled MCP PMT(Hamamatsu, R3809U-50) was used for fluorescencedecay experiments.30 Fluorescence was excited at 298 nmby a 4-MHz train of picosecond pulses generated by thefrequency-doubled output of a cavity-dumped Rhoda-mine 6G dye laser (Spectra Physics, model 375). The 298-nm excitation served both for a selective excitation of Trpresidues only and for obtaining the high initial emissionanisotropy r0.

31 Tryptophan emission was collected

through a monochromator at 360 nm. In order to enhancethe suppression of both excitation light and Ramanscattering, we placed a color glass filter (Zeiss, UG1) infront of the input slit. Data were typically accumulateduntil at least 107 counts per decay were reached. We used1024 channels per decay curve with a time scale of 36.5 ps/channel. Fluorescence decays were measured under the“magic angle” conditions when the measured intensitydecays I(t) were not biased by rotational diffusion of thechromophore. In order to minimize drifts, we sequentiallyacquired the polarized decays I||(t) and I⊥(t) with analternating frequency of 30 s. The G-factor, correctinguneven transmittances of the detection system for I||(t)and I⊥(t), was determined in a separate experiment.In order to detect the dynamics of the GRL by time-

resolved fluorescence anisotropy measurements, it wasrequired to reduce the interfering movement of Trp223.Reduction of Trp223 mobility was achieved by theaddition of glycerol, which was found to have no effecton MPP activity (not shown). The experiments wereconducted using a protein concentration of 0.2 mg/ml in abuffer (20 mM Tris–HCl, pH 8.6, and 100 mM NaCl)containing 0%, 10%, and 15% of glycerol, and 10% glycerol(v/v) was used for theMPP–substrate experiments (unlessotherwise stated). Samples were measured at roomtemperature.Both the fluorescence decays I(t) and fluorescence

anisotropies r(t) were analyzed by the maximum entropymethod for oversampled data (SVD-MEM) coded in ourlaboratory according to an algorithm described else-where.32 For a complex multiexponential fluorescencedecay I(t), the program returns a set of amplitudes αi thatrepresent a distribution of the corresponding lifetimes τi:

I tð Þ =X

i

aie− t= si ð1Þ

Typically, 150 lifetimes τi ranging from 20 ps to 20 nswere chosen. The lifetimes were equidistantly spaced atthe logarithmic time scale and the data analysis startedfrom a flat initial distribution when all amplitudes αi hadthe same value. This gave the same prior probability to alllifetimes in the decay.Since the intensity fraction fi of the ith lifetime

component is given by the formula

fi = aisi =X

i

aisi; ð2Þ

the mean lifetime of the decay was calculated from theequation:

s =X

i

fisi ð3Þ

The dynamic anisotropies r(t) were analyzed fordistributions of rotational correlation times by a methodsimilar to the one described elsewhere.33 We have used a

1208 Primary Contact of α-MPP Subunit with Substrate

model-independent SVD-MEM approach that did not setprior limits on the shape of the distribution. Fluorescenceanisotropy decays were analyzed as a series of exponen-tials:

r tð Þ =X

i

bie− t=/i ð4Þ

where the set of the amplitudes βi represents a distributionof the correlation times ϕi. The βi are related to the initialanisotropy r0 by the formula:

X

i

bi = r0 ð5Þ

Typically, 150 correlation times ϕi ranging from 50 ps to200 ns and equidistantly spaced at the logarithmic scalewere used. For multimodal distributions, the meancorrelation time associated with the mth peak of thedistribution was calculated from the formula:

/m =X

k

bm;k/m;k =X

i

bm;k ð6Þ

where index k runs over the nonzero amplitudes of themth peak of the distribution only. The area of the peakrepresents the associated mean amplitude β̄m.In this work, we did not consider any specific

associations between τ and Φ values. Strong associationwould be reflected in a specific shape of decay-associatedanisotropy curves that could not be simply fitted by anexponential series. The anisotropy decays could exhibiteven non-monotonic dependence. Our data did notexhibit such features. Moreover, in our rather complexsystem with multiple Trps, each of them having multi-exponetial emission decay and a complex emission-depolarizing movement, we were not convinced thatsuch association could be reliably extracted from evenhigh-quality data.

Steady-state fluorescence spectroscopy

Steady-state spectra of the tryptophan fluorescencewere recorded on a Fluoromax-2 fluorometer (Jobin-Yvon, Inc.) at room temperature (22 °C). The excitationwavelength was set to 300 nm to eliminate tyrosinefluorescence. Both excitation and emission bandpasseswere set to 3 nm. The α-MPPmutant forms were dissolvedin TN buffer, pH 8.6, to a final 0.5-μM concentration. Thesubstrate was added in 3-fold molar excess. The signals ofthe buffer and the substrate (if present) were subtracted asbackground.Tryptophan emission is sensitive to the general solvent

effect. The position of the maximum in the emissionspectrum depends on tryptophan exposure to theaqueous solvent.34 The emission maximum ranges forvarious proteins from 308 to 360 nm. A higher exposureto the solvent results in a longer wavelength of theemission maximum. Trp223 was present in all examinedmutants. The subtraction of Trp223 emission from theemission spectrum of the double-tryptophan mutantyielded an apparent spectrum of the tryptophan residuein the GRL.

MD simulation

MD simulations of MPP systems were carried out usingthe AMBER suite of programs35 with the parm99SB forcefield.36 The starting structure of the free yeast MPP was

taken from the Protein Data Bank (PDB) (ID: 1H6R). Themodel structure of MPP with a bound rat ALDHpresequence peptide GPRLSRL↓LSYA (positions 12–22;cleavage site is marked by the arrow and residues inpositions −4 and +1 are in boldface) was generated bysuperposition with the analogous Tom20 receptor protein(PDB ID: 1OM2). The MD simulation protocol was usedas follows.37,38 First, the protonation states of allhistidines were checked by visual inspection to createan optimal H-bond network. Then, all hydrogens wereadded using the Leap program from the AMBERpackage. The structures containing one Zn2+ ion in theactive site were neutralized by adding six or four Na+

counterions for MPP and MPP with peptide-boundsystems, respectively. Each system was inserted in arectangular water box where the layer of the watermolecules was at least 9 Å thick. Then, each system wasminimized prior to the production part of the MD run inthe following way. The protein was frozen and thesolvent molecules with counterions were allowed tomove during a 1000-step minimization followed by a10-ps-long MD run under NpT conditions (p=1 atm,T=298.15 K). Then, the side chains were relaxed byseveral sequential minimizations with decreasing forceconstants applied to the backbone atoms. After relaxa-tion, the system was heated to 50 K for 20 ps and then upto 298.15 K for 90 ps. The particle-mesh Ewald method fortreating electrostatic interaction was used. All simulationswere run under periodic boundary conditions in the NpTensemble at 298.16 K and at a constant pressure of 1 atmusing a 2-fs time integration step. The SHAKE algorithmwith a tolerance of 10−5 Å was applied to fix all bondscontaining hydrogen atoms. The 9.0-Ǻ cutoff was used totreat non-bonding interactions. Coordinates were storedevery 2 ps. The total duration of the production phasesfor the studied systems is equal to 19 ns for the free MPPand 23 ns for MPP with bound pALDH and 24 ns withbound pMDH peptides. The free MPP contained 93.651atoms in 896 residues (457 in the α-subunit and 439 in theβ-subunit), 1 Zn2+ ion, 6 Na+ counterions, and 26.595water molecules. The MPP with bound peptide contained93.696 atoms in 907 residues (457 in the α-subunit, 439 inthe β-subunit, and 11 in the peptide substrate), 1 Zn2+

ion, 6 Na+ counterions, and 26.550 water molecules.For the simulation of the yeast MPP–MDH presequence

interaction, MD simulation of the yeast MPP–ALDHpresequence peptide system was run for an initial 2 nsusing an identical protocol as described above. Afterpeptide structure relaxation, ALDH peptide side chainswere substituted for the yeast MDH presequenceLSRVAKRA↓FSST (positions 2–13; cleavage site ismarked by the arrow and residues in positions −4 and+1 are in boldface). The simulation was continued for anadditional 23 ns using the same protocol.

Acknowledgements

The authors thank Frantisek Kalousek, JacobBauer, and Jan Tachezy for helpful discussion andfor review of the manuscript.This work was supported by grant no.

A501110631 of Grant Agency of the ASCR; bygrants MSM 0021620835, MSM 6198959215, andMSM 6198959216 of the Ministry of Education,Youth and Sports of the Czech Republic; by grant

1209Primary Contact of α-MPP Subunit with Substrate

APVV-0024-07 of the Slovak Research and Devel-opment Agency; and by Institutional ResearchConcept AV 0Z 50200510.

References

1. Model, K., Meisinger, C. & Kuhlbrandt, W. (2008).Cryo-electron microscopy structure of a yeast mito-chondrial preprotein translocase. J. Mol. Biol. 383,1049–1057.

2. Model, K., Prinz, T., Ruiz, T., Radermacher, M.,Krimmer, T., Kuhlbrandt, W. et al. (2002). Proteintranslocase of the outer mitochondrial membrane: roleof import receptors in the structural organization ofthe TOM complex. J. Mol. Biol. 316, 657–666.

3. Pfanner, N. & Geissler, A. (2001). Versatility of themitochondrial protein import machinery. Nat. Rev.,Mol. Cell Biol. 2, 339–349.

4. Schatz, G. & Dobberstein, B. (1996). Common princi-ples of protein translocation across membranes.Science, 271, 1519–1526.

5. Pfanner, N. & Neupert, W. (1990). The mitochondrialprotein import apparatus. Annu. Rev. Biochem. 59,331–353.

6. Gakh, O., Cavadini, P. & Isaya, G. (2002). Mitochon-drial processing peptidases. Biochim. Biophys. Acta,1592, 63–77.

7. Ito, A. (1999). Mitochondrial processing peptidase:multiple-site recognition of precursor proteins. Bio-chem. Biophys. Res. Commun. 265, 611–616.

8. Taylor, A. B., Smith, B. S., Kitada, S., Kojima, K.,Miyaura, H., Otwinowski, Z. et al. (2001). Crystalstructures of mitochondrial processing peptidasereveal the mode for specific cleavage of import signalsequences. Structure, 9, 615–625.

9. Yang, M. J., Geli, V., Oppliger, W., Suda, K., James, P.& Schatz, G. (1991). The MAS-encoded processingprotease of yeast mitochondria. Interaction of thepurified enzyme with signal peptides and a purifiedprecursor protein. J. Biol. Chem. 266, 6416–6423.

10. Luciano, P., Geoffroy, S., Brandt, A., Hernandez, J. F.& Geli, V. (1997). Functional cooperation of themitochondrial processing peptidase subunits. J. Mol.Biol. 272, 213–225.

11. Gakh, O., Obsil, T., Adamec, J., Spizek, J., Amler, E.,Janata, J. & Kalousek, F. (2001). Substrate bindingchanges conformation of the alpha-, but not the beta-subunit of mitochondrial processing peptidase. Arch.Biochem. Biophys. 385, 392–396.

12. Daley, D. O., Adams, K. L., Clifton, R., Qualmann, S.,Millar, A. H., Palmer, J. D. et al. (2002). Gene transferfrommitochondrion to nucleus: novel mechanisms forgene activation from Cox2. Plant J. 30, 11–21.

13. Branda, S. S. & Isaya, G. (1995). Prediction andidentification of new natural substrates of the yeastmitochondrial intermediate peptidase. J. Biol. Chem.270, 27366–27373.

14. von Heijne, G. (1986). Mitochondrial targetingsequences may form amphiphilic helices. EMBO J. 5,1335–1342.

15. Roise, D., Theiler, F., Horvath, S. J., Tomich, J. M.,Richards, J. H., Allison, D. S. & Schatz, G. (1988).Amphiphilicity is essential for mitochondrial prese-quence function. EMBO J. 7, 649–653.

16. Abe, Y., Shodai, T., Muto, T., Mihara, K., Torii, H.,Nishikawa, S. et al. (2000). Structural basis ofpresequence recognition by the mitochondrial proteinimport receptor Tom20. Cell, 100, 551–560.

17. Kojima, K., Kitada, S., Ogishima, T. & Ito, A. (2001). Aproposed common structure of substrates bound tomitochondrial processing peptidase. J. Biol. Chem. 276,2115–2121.

18. Nishino, T. G., Kitano, K., Kojima, K., Ogishima, T.,Ito, A. & Kitada, S. (2007). Spatial orientation ofmitochondrial processing peptidase and a preproteinrevealed by fluorescence resonance energy transfer. J.Biochem. 141, 889–895.

19. Janata, J., Hola, K., Kubala, M., Gakh, O., Parkho-menko, N., Matuskova, A. et al. (2004). Substrateevokes translocation of both domains in the mito-chondrial processing peptidase alpha-subunit duringwhich the C-terminus acts as a stabilizing element.Biochem. Biophys. Res. Commun. 316, 211–217.

20. Nagao, Y., Kitada, S., Kojima, K., Toh, H., Kuhara, S.,Ogishima, T. & Ito, A. (2000). Glycine-rich region ofmitochondrial processing peptidase alpha-subunit isessential for binding and cleavage of the precursorproteins. J. Biol. Chem. 275, 34552–34556.

21. Eftink, M. R. (2000). Intrinsic fluorescence of proteins. InTopics in Fluorescence Spectroscopy (Lakowicz, J. R., ed.),vol. 6, pp. 1–15, Kluwer Academic/Plenum Publishers,New York, NY.

22. Bossemeyer, D. (1994). The glycine-rich sequence ofprotein kinases: a multifunctional element. TrendsBiochem. Sci. 19, 201–205.

23. Kitada, S., Uchiyama, T., Funatsu, T., Kitada, Y.,Ogishima, T. & Ito, A. (2007). A protein from aparasitic microorganism, Rickettsia prowazekii, cancleave the signal sequences of proteins targetingmitochondria. J. Bacteriol. 189, 844–850.

24. Saitoh, T., Igura, M., Obita, T., Ose, T., Kojima, R.,Maenaka, K. et al. (2007). Tom20 recognizes mito-chondrial presequences through dynamic equilibriumamong multiple bound states. EMBO J. 26, 4777–4787.

25. Likic, V. A., Perry, A., Hulett, J., Derby, M., Traven, A.,Waller, R. F. et al. (2005). Patterns that define the fourdomains conserved in known and novel isoforms ofthe protein import receptor Tom20. J. Mol. Biol. 347,81–93.

26. Perry, A. J., Hulett, J. M., Likic, V. A., Lithgow, T. &Gooley, P. R. (2006). Convergent evolution of recep-tors for protein import into mitochondria. Curr. Biol.16, 221–229.

27. Adamec, J., Gakh, O., Spizek, J. & Kalousek, F. (1999).Complementation between mitochondrial processingpeptidase (MPP) subunits from different species.Arch.Biochem. Biophys. 370, 77–85.

28. Janata, J., Kogekar, N. & Fenton, W. A. (1997).Expression and kinetic characterization of methyl-malonyl-CoA mutase from patients with the mut-phenotype: evidence for naturally occurring inter-allelic complementation. Hum. Mol. Genet. 6,1457–1464.

29. Geli, V. (1993). Functional reconstitution in Escherichiacoli of the yeast mitochondrial matrix peptidase fromits two inactive subunits. Proc. Natl Acad. Sci. USA, 90,6247–6251.

30. Obsilova, V., Herman, P., Vecer, J., Sulc, M., Teisinger,J. & Obsil, T. (2004). 14-3-3zeta C-terminal stretchchanges its conformation upon ligand binding andphosphorylation at Thr232. J. Biol. Chem. 279,4531–4540.

31. Eftink, M. R., Selvidge, L. A., Callis, P. R. &Rehms, A. A. (1990). Photophysics of indolederivatives—experimental resolution of La and Lbtransitions and comparison with theory. J. Phys.Chem. 94, 3469–3479.

1210 Primary Contact of α-MPP Subunit with Substrate

32. Bryan, R. K. (1990). Maximum-entropy analysis ofoversampled data problems. Eur. Biophys. J. 18,165–174.

33. Brochon, J. C. (1994). Maximum-entropy method ofdata-analysis in time-resolved spectroscopy. MethodsEnzymol. 240, 262–311.

34. Eftink, M. R. & Ghiron, C. A. (1976). Exposure oftryptophanyl residues in proteins—quantitative de-termination by fluorescence quenching studies.Biochemistry, 15, 672–680.

35. Case, D. A., Cheatham, T. E., Darden, T., Gohlke, H.,Luo, R., Merz, K. M. et al. (2005). The Amberbiomolecular simulation programs. J. Comput. Chem.26, 1668–1688.

36. Hornak, V., Abel, R., Okur, A., Strockbine, B., Roitberg,A. & Simmerling, C. (2006). Comparison of multipleamber force fields and development of improvedprotein backbone parameters. Proteins, 65, 712–725.

37. Bartova, I., Otyepka, M., Kriz, Z. & Koca, J. (2004).Activation and inhibition of cyclin-dependent kinase-2 by phosphorylation; a molecular dynamics studyreveals the functional importance of the glycine-richloop. Protein Sci. 13, 1449–1457.

38. Bartova, I., Otyepka, M., Kriz, Z. & Koca, J. (2005). Themechanism of inhibition of the cyclin-dependentkinase-2 as revealed by the molecular dynamicsstudy on the complex CDK2 with the peptidesubstrate HHASPRK. Protein Sci. 14, 445–451.