The Plant Journal (2001) 27(5), 427±438
Mutagenesis and computer modelling approach to study determinants for recognition of signal peptides by the mitochondrial processing peptidase Xiao-Ping Zhang1,², Sara SjoÈling1,²,³, Marcel Tanudji2, LaÂszlo Somogyi3,§, David Andreu3, L. E. GoÈran Eriksson1, Astrid GraÈslund1, James Whelan2 and Elzbieta Glaser1,* 1 Department of Biochemistry and Biophysics, Arrhenius Laboratories for Natural Sciences, Stockholm University, 106 91 Stockholm, Sweden, 2Department of Biochemistry, University of Western Australia, Nedlands, Australia, and 3 Department of Organic Chemistry, University of Barcelona, Barcelona, Spain Received 18 May 2001; accepted 19 June 2001 *For correspondence (fax: +46 8 153679; email:
[email protected]) ² The contribution of the ®rst two authors to this work was equal. ³ Present address: University College London, Department of Biochemistry and Molecular Biology, Gower Street, London, WC1E 6BT, UK. § Present address: The Skaggs Institute for Chemical Biology, The Scripps Research Institute, 10550 N. Torrey Pines Road, La Jolla, CA 92037, USA.
Summary Determinants for the recognition of a mitochondrial presequence by the mitochondrial processing peptidase (MPP) have been investigated using mutagenesis and bioinformatics approaches. All plant mitochondrial presequences with a cleavage site that was con®rmed by experimental studies can be grouped into three classes. Two major classes contain an arginine residue at position ±2 or ±3, and the third class does not have any conserved arginines. Sequence logos revealed loosely conserved cleavage motifs for the ®rst two classes but no signi®cant amino acid conservation for the third class. Investigation of processing determinants for a class III precursor, Nicotiana plumbaginifolia F1b precursor of ATP synthase (pF1b), was performed using a series of pF1b presequence mutants and mutant presequence peptides derived from the C-terminal portion of the presequence. Replacement of ±2 Gln by Arg inhibited processing, whereas replacement of either the most proximally located ±5 Arg or ±15 Arg by Leu had only a low inhibitory effect. The C-terminal portion of the pF1b presequence forms a helix± turn±helix structure. Mutations disturbing or prolonging the helical element upstream of the cleavage site inhibited processing signi®cantly. Structural models of potato MPP and the C-terminal pF1b presequence peptide were built by homology modelling and empirical conformational energy search methods, respectively. Molecular docking of the pF1b presequence peptide to the MPP model suggested binding of the peptide to the negatively charged binding cleft formed by the a-MPP and b-MPP subunits in close proximity to the H111XXE114H115X(116±190)E191 proteolytic active site on b-MPP. Our results show for the ®rst time that the amino acid at the ±2 position, even if not an arginine, as well as structural properties of the C-terminal portion of the presequence are important determinants for the processing of a class III precursor by MPP. Keywords: mitochondrial processing peptidase, cytochrome bc1 presequence, mitochondrial biogenesis, computer modelling.
complex,
protein
processing,
Introduction Most mitochondrial proteins are nuclear-encoded and synthesized on cytosolic ribosomes with an N-terminal extension called the presequence. The presequence contains information required for targeting of the precursor protein to mitochondria and is cleaved off by the mitochondrial processing peptidase (MPP) after the preã 2001 Blackwell Science Ltd
cursor has been imported into the mitochondrion (for review see; Glaser et al., 1998; Neupert, 1997; Schatz and Dobberstein, 1996). The presequence also contains hsp70 binding motifs, which are important for translocation of the precursor across the mitochondrial membranes (Zhang et al., 1999). Studies of mitochondrial prese427
428
Xiao-Ping Zhang et al.
quences from different organisms showed a high variation in the presequence length, from 8 to 136 residues (Zhang and Glaser, unpublished data). Presequences do not share sequence similarity and no consensus for processing has been found, except that many mitochondrial precursors contain a loosely conserved motif with an arginine residue at the ±2 or ±3 position upstream of the cleavage site (Schneider et al., 1998; von Heijne et al., 1989). The importance of these arginines for presequence recognition by MPP has been shown by experimental studies (Arretz et al., 1994; Ogishima et al., 1995; Tanudji et al., 1999). Another characteristic feature of mitochondrial presequences is the high content of positively charged residues and the capacity to form an amphiphilic a-helix (Roise et al., 1986; von Heijne, 1986). The importance of an N-terminal amphiphilic a-helix and basic residues within the presequence for interaction with the mitochondrial import machinery is generally accepted (Voos et al., 1999). Recently, it has been shown that hydrophobic rather than ionic interactions mediate binding of a presequence to the cytoplasmic domain of the Tom20 receptor (Abe et al., 2000; Muto et al., 2001). Most mitochondrial presequences are cleaved off in a single step by MPP, resulting in mature proteins (Arretz et al., 1991). MPP is a heterodimeric enzyme composed of two related polypeptides, a-MPP and b-MPP. a-MPP is proposed to bind the precursor protein, whereas the bMPP contains the catalytic site (Kitada et al., 1995; Luciano et al., 1997). MPP belongs to a metalloendopeptidase protein family ± the pitrilysins (Rawlings and Barrett, 1995). The HXXEHX74±76E Zn2+-binding motif, the catalytic site in pitrilysins, is conserved in all b-MPPs but is degenerated in a-MPPs. MPP resides in various intra-mitochondrial locations in different species. In Saccharomyces cerevisiae and mammals, MPP is located in the matrix, whereas in Neurospora crassa b-MPP is identical to the core 1 subunit of the bc1 complex of the respiratory chain (Schulte et al., 1989). In plants, MPP is completely integrated into the cytochrome bc1 complex, where the core 1 and core 2 proteins are identical to the bMPP and a-MPP subunits, respectively (Braun et al., 1992; Eriksson et al., 1993; Eriksson et al., 1994; for review see Glaser and Dessi, 1999). Recent studies have shown that the plant MPP/bc1 complex is distant from the protein import channels (Dessi et al., 2000). MPP is a general peptidase, acting on several hundred mitochondrial precursor proteins, recognizing a distinct cleavage site and speci®cally removing the presequence. The mechanism of MPP recognition of the cleavage site on a precursor has yet to be clari®ed. The importance of arginine residues proximal and distal to the cleavage site has been shown experimentally by site-directed mutagenesis or with synthetic presequence peptides (Klaus et al., 1996; Niidome et al., 1994; Ogishima et al., 1995; Ou et al., 1994; Tanudji
et al., 1999). In addition, a proline residue between the distal and proximal arginines (Hammen et al., 1996; Niidome et al., 1994; Ogishima et al., 1995; Thornton et al., 1993) and individual residues at positions +1, +2 and +3 from the cleavage site were shown to be of importance (Klaus et al., 1996; Lain et al., 1998; Ogishima et al., 1995). It has also been suggested that MPP recognizes higher-order structural elements (Hammen et al., 1994; Luciano and GeÂli, 1996; SjoÈling and Glaser, 1998; SjoÈling et al., 1994; SjoÈling et al., 1996; Waltner and Weiner, 1995). In this study, we investigated the interaction of MPP with the presequence from the Nicotiana plumbaginifolia mitochondrial F1b precursor. In order to understand the recognition mechanism of MPP for a subset of mitochondrial precursors which do not have ±2 and ±3 conserved arginines, we have used mutagenesis and computer modelling techniques. This is the ®rst study concerning the substrate speci®city of MPP for class III precursors. We have introduced point mutations and changed the secondary structure of the presequence of pF1b. We have used mutant synthetic peptides derived from the C-terminal part of the pF1b presequence to study effects on processing. Our results show that recognition of the pF1b presequence by MPP is an event controlled by multiple factors. The ±2 position, secondary structural elements and presequence ¯exibility are important for the ef®cient processing of pF1b by MPP. We have built a structural model of MPP based on the three-dimensional structure of the bovine cytochrome bc1 complex and performed molecular docking studies of the wild-type presequence peptide to the MPP model. Determinants for processing are discussed on the basis of the information obtained from the mutagenesis and computer modelling studies. Results Analysis of mitochondrial presequences We analysed the position of conserved arginines in all available plant mitochondrial presequences with experimentally determined cleavage sites and found that 33.3% of the presequences have ±2R, 44.4% have ±3R and the remaining 22.3% have neither ±2R nor ±3R. We did not ®nd any over-representation of arginine at position ±10 nor a characteristic octapeptide found in mammalian and yeast precursors that are cleaved by mitochondrial intermediate peptidase (MIP) (Isaya et al., 1991). We classi®ed presequences into three groups, class I (±2R), class II (±3R) and class III (no conserved arginines) and analysed them using sequence logos (Schneider and Stephens, 1990) (Figure 1). The 15 N-terminal amino acid residues and 15 C-terminal residues of the presequences plus 15 residues from the N-terminus of the mature portion were investiã Blackwell Science Ltd, The Plant Journal, (2001), 27, 427±438
Speci®city of the mitochondrial processing peptidase gated, and the amino acid occurrence frequencies at each position were determined. In the ±2R and ±3R classes, we observed a loosely conserved motif around the cleavage site. In the ±2R class (one sequence with ±2K), there was a high abundance of serine at positions +1 and +2 and threonine at positions +2 and +3 (Figure 1a). The ±3R class also showed some interesting features: at position ±1, there was an over-representation of the aromatic amino acids phenylalanine and tyrosine (only one exception was observed), whereas at position +1, alanine and serine, and at position +2, threonine, serine and alanine were overrepresented (Figure 1b). A similar analysis for class III presequences showed no signi®cant conservation in amino acid content around the cleavage site (Figure 1c), indicating that other features are important for recognition of this subset of presequences by MPP. We used the precursor of the N. plumbaginifolia F1b subunit of ATP synthase as an example from class III to investigate determinants for processing. The 54 amino acid residue presequence of pF1b contains eight arginine residues, ®ve of which are located in the N-terminal half and three in the C-terminal half of the presequence. Sequence alignment of six higher plant pF1b presequences showed that these arginines are well conserved (Figure 2). Although the lengths of these presequences are different, the N- and C-terminal portions of the presequences are more conserved than the internal part. Both of the terminal regions have a high potential to form an a-helix according to secondary structure prediction. The internal segment between the N- and C-terminal domains is enriched in proline indicating that it is a more ¯exible part of the presequence. The C-terminal part of the pF1b presequence has been previously shown to interact with both the spinach MPP/bc1 complex and rat MPP (SjoÈling et al., 1994; SjoÈling et al., 1996). Therefore, this study focused on the Cterminal part of the presequence. Processing of wild-type pF1b in the presence of synthetic presequence peptides In order to investigate the role of proximal basic residues in the processing of pF1b, we replaced the three arginine residues in synthetic peptides derived from the 22 Cterminal amino acid residues of the pF1b presequence. We tested the ability of the peptides to inhibit processing of wild-type pF1b by the puri®ed spinach MPP/bc1 complex (Figure 3). Three peptides were synthesized with a leucine residue instead of an arginine at positions ±5 (R±5L), ±15 (R±15L) or ±19 (R±19L) relative to the cleavage site. In the fourth peptide, all three arginines were exchanged for leucines (R±5,15,19L) (Figure 3a). The R±5L peptide inhibited processing to the same extent as the wild-type peptide, whereas R±15L, R±19L and R±5,15,19L were less inhibitory than the wild-type peptide (Figure 3b). ã Blackwell Science Ltd, The Plant Journal, (2001), 27, 427±438
429
Secondary structure prediction of the pF1b showed that the C-terminal helix in the presequence goes through the cleavage site of the wild-type peptide (cf. Figure 2b). The region upstream of the cleavage site around positions ±18/±19 showed a propensity to form a short a-helix (see also SjoÈling et al., 1994). To test the role of the C-terminal helical structure in processing, we synthesized a peptide with proline at position ±17 replacing a serine residue (S±17P). This exchange destroyed the short C-terminal helical element according to secondary structure predictions. Circular dichroism analysis con®rmed a lower a-helix content (55% less compared to the wild-type peptide) (Figure 3a). The S-17P peptide inhibited processing less than the wild-type or the other peptides, indicating a lower af®nity for MPP. The R±5,15,19L peptide, which had a higher content of a-helix (74% more compared to the wild-type peptide according to circular dichroism analysis), showed also diminished inhibition of processing in comparison with the wild-type peptide (Figure 3b). Secondary structure prediction of this mutant peptide indicated a prolonged helical structure (not shown). Processing of the presequence mutants of the pF1b precursor by MPP To investigate the role of certain amino acid residues in the C-terminal part of the pF1b presequence for MPP cleavage, we performed site-directed mutagenesis directly on the pF1b precursor (Figure 4a) and measured processing activity (Figure 4b). A leucine residue at position ±8 is conserved in the presequences of nearly all pF1b from higher plants, bovine, human, rat, N. crassa and Chlamydomonas reinhardtii. Therefore, we exchanged this residue for an alanine (L±8A) or a glycine residue (L±8G). The mutations only slightly lowered the ef®ciency of processing when compared to the wild-type, showing that although the leucine is conserved in pF1b presequences, it is not essential for processing. Alanine and serine residues are enriched at the ®rst position, +1, of the mature part of plant precursor proteins (cf. Figure 1). We exchanged the +1 alanine residue to a leucine. This mutation had no effect on processing, indicating that this conserved residue is also not essential for processing. The closest arginine to the cleavage site of pF1b is at position ±5. Upon replacement of ±5 arginine by leucine (R±5L) or alanine (R±5A), processing was decreased to 87% and 77%, respectively, compared with the wild-type. To introduce an arginine at position ±2, a double mutant was created where glutamine at position ±2 was exchanged for an arginine and arginine at position ±5 was exchanged for a leucine (Q±2R, R±5L). Processing of the double mutant was substantially inhibited to 54% compared with the wildtype. Comparison with the processing data of the R±5L mutant reveals that the ±2 position is important for the
430
Xiao-Ping Zhang et al.
ã Blackwell Science Ltd, The Plant Journal, (2001), 27, 427±438
Speci®city of the mitochondrial processing peptidase recognition of the precursor by MPP, even if there is no arginine at this position. In mutant R±15L, the arginine distant to the cleavage site, at position ±15, was replaced by a non-charged leucine. The mutation of the distant basic residue had no effect on processing. The C-terminal helical structure may facilitate the interaction of the presequences with MPP (Figure 2 and SjoÈling et al., 1994). Two mutants were designed to affect this structure. An exchange of proline at position ±12 for a leucine (P±12L) was predicted to result in a prolonged helix at the C-terminal portion of the presequence (not shown). Substitution of arginine at position ±19 by a glycine (R± 19G) was predicted to abolish partially the helical region. Both these mutations resulted in signi®cant inhibitory effects, showing that the structural properties of the Cterminal part of the presequence are the most important factors for processing by MPP (Figure 4b). Presequence recognition by the MPP model In plants, the MPP subunits are integrated into the bc1 complex and are identical to core proteins of the complex (for review see Glaser and Dessi, 1999). Alignments of the potato b-MPP with bovine core 1 protein and a-MPP with core 2 protein reveal 43 and 28% sequence identities,
431
respectively. Therefore, although a crystal structure of MPP subunits is not available, the data from the threedimensional structures of the bovine bc1 complex (Iwata et al., 1998; Xia et al., 1997) can be used to derive the structural model of MPP. We generated a homology model of the potato MPP (Figure 5a±c) based on the X-ray structure of core 1 and core 2 of the bovine bc1 complex (Iwata et al., 1998). The plant MPP model shares a very similar folding topology with the core proteins of bovine bc1 complex. The model reveals that a-MPP and b-MPP subunits form a cleft, which extends deeply inside MPP and forms a large internal cavity (Figure 5b). There are seven negatively charged areas on the surface of the cavity. Inside the cavity, there are two negative patches (Glu200, Glu201, Glu361, Glu381 and Glu433, Asp434) on b-MPP, and another two (Glu97, Glu174 and Glu414, Asp415) on a-MPP (not visible in Figure 5b). On both sides of the crack there are negatively charged areas: Asp277, Glu279, Asp364 and Asp463 on a-MPP form a large negatively charged area (low, right) whereas Glu193± 194 and Asp301±303 on b-MPP form two strong negatively charged areas (left side of the crack). Furthermore, the Zn2+-binding site of b-MPP that consists of His111, Glu114, His115 and Glu191 faces the cavity of MPP (Figure 5c,d). The H111XXE114H115 motif and the distal E191 are spatially
Figure 2. Conserved domains within the plant presequence of the mitochondrial pF1b subunit of ATP synthase. (a) The alignment of all available statistically relevant plant pF1b presequences shows conserved regions in both the N-terminal and C-terminal regions of presequence peptides. GI525291 represents the wheat pF1b presequence. The alignment was performed in ClustalX 1.80 (Thompson et al., 1997) with the default parameters, the picture was generated in Gendoc (http://www.psc.edu/biomed/genedoc/). (b) Predicted secondary structures of the pF1b presequences presented in (a). The prediction was performed using nnPredict (Kneller et al., 1990).
Figure 1. Sequence logos of plant mitochondrial presequence peptides collected from the Swiss-Prot and NCBI databases. The cleavage sites of the presequence peptides by MPP were experimentally con®rmed (either from Swiss-Prot annotation or the literature). The left part of the logos show 15 amino residues from the N-terminal portion of the presequence peptide. The right part of the logos includes the 15 C-terminal amino acid residues plus 15 residues from the N-terminal portion of the mature protein from (a) 15 peptides with ±2R (class I), (b) 20 peptides with ±3R (class II), and (c) 10 peptides with no ±2R or ±3R (class III).
ã Blackwell Science Ltd, The Plant Journal, (2001), 27, 427±438
432
Xiao-Ping Zhang et al. et al., 1998), the peptide was successfully docked into theFigure 5c,d); the docked energy was ±8.12 kcal mol±1. The tyrosine residue at position ±1 of the C-terminal helix at the cleavage site of the pF1b presequence is very close to the active site of MPP (Figure 5d), with the turn region pointing into the interior of the cavity. The ±5, ±15 and ±19 arginines of the peptide interact with Glu201, Glu381 and Lys338 on b-MPP, respectively, by forming hydrogen bonds, indicating that these arginines might be involved in the interaction of the presequence peptide with MPP. Discussion
Figure 3. Effect of mutant presequence peptides on processing of wildtype N. plumbaginifolia pF1b by spinach MPP/bc1 complex. (a) Synthetic presequence peptides used. Values on the right side show the estimated percentage a-helix of the pF1b presequence peptides analysed using circular dicroism. (b) The pF1b processing was measured in the presence of the presequence peptides at 30°C after 40 min. MPP was incubated prior to the processing reaction with the synthetic peptides for 5 min at 30°C. The processing ef®ciency was calculated as the ratio of the mature form to the sum of the precursor and the mature form. The processing ef®ciency of wild-type (WT) precursor without addition of any peptide has been taken as 100%. The data set used in the diagram represents mean values from ®ve experiments.
located close to each other to form the putative active site (Figure 5c). A structural model of the peptide that corresponds to the 22 C-terminal amino acids of the pF1b presequence (identical to the synthetic wild-type presequence peptide used in the competition experiments, cf. Figure 3) was obtained through an extensive iteration with the Empirical Conformational Energy Program for Peptides (ECEPPAK, http:// www.tc.cornell.edu/Research/Biomed/CompBiologyTools/ eceppak/Manual.html) (Ripoll et al., 1998) (Figure 5d). The overall structure of this peptide consists of a helix±turn± helix. The C-terminal helix is formed by residues ±1 to ±11, the turn structure by residues ±12 to ±14 (Pro, Ser and Ala) and the second helix by residues ±15 to ±21. By using the Lamarckian Genetic Algorithm in Autodock 3.0.5 (Morris
This study showed that the majority of plant mitochondrial presequences belong to the ±2R and ±3R classes, as in yeast and mammals (Gavel and von Heijne, 1990). These classes share loosely de®ned cleavage motifs, RX¯S(S/T)T and RX(F/Y)¯(A/S)(T/S/A) (arrow represents cleavage site), respectively (cf. Figure 1). The class II precursors have an aromatic amino acid at position ±1. Class III presequences that do not contain any conserved motif account for 22.3% of all presequences. There was only one presequence with ±10 R and no ±2R or ±3R. However, we found two entries for soybean in NCBI (AW567635, AW759833) and four entries for Arabidopsis in TIGR (K17N15±9, K21L13±14, F12B17± 110, F12A21.28) with a substantial (> 30%) sequence identity and 41±64% sequence similarity to human MIP. The presence of MIP in plants has to be studied further. The existence of class III presequences suggests that elements other than the ±2 or ±3 arginines and loosely de®ned sequence motifs play an important role in mitochondrial presequence recognition by MPP. Mutagenesis experiments of a class III presequence, pF1b, showed that replacing either of the two arginines ±5R or ±19R by leucine reduced the processing ef®ciency, whereas substitution of ±15R by leucine did not affect processing. The replacement of ±5R with a small hydrophobic residue, alanine, disturbed processing ef®ciency more than replacement with a bulky hydrophobic residue, leucine, indicating an interaction between this residue and MPP. However, these results imply that the role of ±5R is not crucial for pF1b processing by MPP. Replacement of ±19R with glycine not only removed a positive charge from the presequence but also changed the secondary structure of the C-terminal helical element of the presequence, and consequently had a strong inhibitory effect on processing. It has been suggested that distal arginines interact with negative charges on a-MPP (Shimokata et al., 1998). The interaction of proximal arginine(s) with b-MPP and of the distal arginine(s) with a-MPP may place the presequence in the cavity in the proper position and expose the cleavage site to the MPP active site. The electrostatic interaction between arginines of the presequence with MPP might have an accumulative effect (Ito, 1999). There ã Blackwell Science Ltd, The Plant Journal, (2001), 27, 427±438
Speci®city of the mitochondrial processing peptidase
433
Figure 4. Effect of point mutations in the C-terminal portion of the pF1b presequence on processing by spinach MPP/bc1 complex. (a) Mutants tested. (b) Cleavage of mutant pF1b precursors by puri®ed spinach MPP/bc1 complex after 5, 10 and 40 min incubation. The ef®ciency of processing was calculated as the ratio of the mature form to the sum of the precursor and the mature form. The processing ef®ciency of the wild-type (WT) precursor after 40 min has been taken as 100%. The data set in the diagram represents average values from six experiments. Standard errors for different mutants after 40 min were as follows (%): WT, 5.50; A+1L, 10.5; Q±2R, 14.3; R±5A, 12.6; R±5L, 7.4; L±8A, 9.9; L±8G, 11.5; P±12L, 11.7; R±15L, 9.3; R±19G, 13.6.
are three arginines on the C-terminal peptide of the pF1b presequence and therefore changing one arginine may not be suf®cient to abolish processing. In contrast to the mutagenesis experiments, R±5L and R±19L mutant peptides inhibited processing to the same extent as the wildtype peptide. The R±15L and R±5,15,19L peptides inhibited processing less ef®ciently. These results show that mutation of the arginines in presequence peptides did not affect binding af®nity to MPP or affected it only partially. We interpret this as indicating that mutations of arginines do not cause dramatic structural changes of the peptides and therefore the mutant peptides still bind to MPP. The docking orientation of the mutant peptides on MPP may be changed in comparison to the wild-type peptide. The processing event includes both presequence binding and cleavage site recognition. The competition experiments in the presence of presequence peptides re¯ect the effect of ã Blackwell Science Ltd, The Plant Journal, (2001), 27, 427±438
mutations on binding af®nity, whereas direct pF1b precursor mutagenesis experiments supply information on both binding and cleavage site recognition. Therefore, the results may be apparently not consistent. Within the pF1b presequence, a glutamine residue is located at the ±2 position. To our surprise, we could not replace this glutamine with a basic arginine residue without substantial loss of the processing activity, showing that the residue at position ±2 affects processing even though it is not an arginine. In contrast to pF1b, the ±2R or ±3R motifs in class I and II precursors seem to be the most important factors for MPP recognition. If the ±2 or ±3 arginine is changed, the processing is substantially inhibited (Tanudji et al., 1999). Certain distal basic residues are also important for processing of class I signal peptides (Ogishima et al., 1995; Tanudji et al., 1999), implying an accumulative effect similar to that seen in this study.
434
Xiao-Ping Zhang et al.
Figure 5. Structural model of MPP. (a) A model of potato MPP (see Experimental procedures for details on model building). (b) GRASP surface of the MPP. The negatively charged cleft is anticipated to be the substrate-binding site. (c) A transparency view of MPP and the docked C-terminal 22 amino residue peptide of pF1b presequence showing the global position of the active site and the docked peptide in the MPP. Molecular docking was performed in Autodock 3.0.5 using the Lamarckian Genetic Algorithm. The docked energy is ±8.12 kcal mol±1. (d) Interaction of docked peptide with b-MPP. The putative active site consists of His111, Glu114, His115 and Glu191 on the b-MPP. The pF1b presequence peptide was found to be in a helix±turn±helix structure. The three arginines ±5R, ±15R and ±19R on the peptide potentially interact with the residues Glu201, Glu381 and Lys338 on b-MPP, respectively, by forming hydrogen bonds. The cleavage site of the presequence was in the vicinity of the active site. (a) was produced using Molscript and Raster3D, (b) with GRASP, (c) with Molscript, GRASP and Raster3D and (d) with WebLab ViewerLiteä 3.2. The labelling was done in GIMP and ImageMagick.
The importance of structural elements in processing has been previously suggested (Hammen et al., 1994; Ou et al., 1994; SjoÈling and Glaser, 1998; SjoÈling et al., 1994; SjoÈling et al., 1996; Waltner and Weiner, 1995). Our studies of structural mutants of a class III precursor show that the structure of the pF1b presequence C-terminal domain is
very important for processing. The ECEPPAK model (Ripoll et al., 1998) of the C-terminal part of pF1b (cf. Figure 5d) reveals a helix±turn±helix structure, consistent with the typical structures of cleavable mitochondrial presequences elucidated by NMR studies (Lancelin et al., 1996; Thornton et al., 1993). The replacement of the ±17 serine of the ã Blackwell Science Ltd, The Plant Journal, (2001), 27, 427±438
Speci®city of the mitochondrial processing peptidase synthetic pF1b presequence peptide with a proline residue, diminishing the a-helical content of the peptide, strongly decreased the af®nity of the peptide for MPP. Also, the R± 5,15,19L mutant peptide with a higher content of the helical structure showed a decreased af®nity for MPP. In addition, direct mutations in pF1b affecting the presequence secondary structure inhibited processing ef®ciency. The P±12L mutation caused a prolonged Cterminal helix and dramatically affected processing. These results are consistent with data on mammalian aldehyde dehydrogenase presequence showing that deletion of a ¯exible linker Arg-Gly-Pro in the presequence inhibited the processing activity (Thornton et al., 1993). A few non-cleavable mitochondrial presequences form a continuous a-helix (Hammen et al., 1994; Jarvis et al., 1995), supporting our processing data on the P±12L mutant and indicating that the linker between the two helixes is important for processing. This linker might give enough ¯exibility to the signal sequence to adapt to the MPP substrate binding cleft. The R±19G mutation that removed a short helical structure upstream of the C-terminal a-helix also greatly decreased the processing ef®ciency. According to docking studies (cf. Figure 5c,d), this helix may be necessary to position the presequence and expose the cleavage site to the active site of b-MPP. Molecular modelling has proved to be a valuable approach to study the structure and function of biomolecules when the X-ray structure is not yet available. The interaction of a class I presequence peptide with the mammalian matrix-located soluble MPP has been discussed based on a MPP model (Ito, 1999). Our homology model of the plant membrane-bound MPP (cf. Figure 5a) suggests a bowl-like structure of a-MPP and b-MPP, forming a global complex with the putative peptide binding cleft between the subunits (Figure 5b). This cleft opens to the matrix (Iwata et al., 1998; Xia et al., 1997), providing the active site with the maximal accessibility for mitochondrial presequences without spatial hindrance. Negatively charged patches on the presequence binding cleft of the MPP model and the abundance of positively charged residues on mitochondrial presequences imply that electrostatic interactions occur during the recognition event. For the class III precursor, pF1b, it seems that both charge and hydrophobic interactions play a role in the recognition of the presequence by MPP. Docking studies showed that the cleft may form a large cavity, in which a peptide containing 22 amino acid residues derived from the C-terminal of pF1b could be easily docked. These data are consistent with the ¯uorescence quenching analysis, showing that a substrate peptide bound to MPP was buried inside the MPP (Kojima et al., 1998). However, the model reveals that the depth of the cleft is limited and a long presequence must be bent or folded in the right con®rmation to expose the cleavage site to the active site ã Blackwell Science Ltd, The Plant Journal, (2001), 27, 427±438
435
on b-MPP. The C-terminus helix±turn±helix structure is probably a satisfactory solution for this requirement. Experimental procedures Analysis of mitochondrial presequences Our data set includes 45 plant mitochondrial presequences collected from Swiss-Prot (release 39.7, 2 October 2000) and NCBI databases. The signal peptides were classi®ed according to the presence of a conserved arginine at the ±2 and ±3 positions from the cleavage site or the absence of any conserved amino acid. We found only one presequence which contains ±10R but neither ±2R nor ±3R, HS7M_PHAVU (temporarily put in class III as the presence of MIP in plants is not yet con®rmed). The 15 Nterminal amino acids residues and 15 C-terminal residues plus 15 amino acid residues from the mature portion were analysed by sequence logos generated using alpro and makelogo (Schneider and Stephens, 1990).
Isolation of the MPP/bc1 complex Spinach (Spinacia oleracea L.) leaf mitochondria were isolated according to the method described by Hamasur et al. (1990). Fractionation of mitochondria was performed by sonication in the presence of 30 mM MgCl2, and the spinach MPP/bc1 complex was isolated by dodecyl-b-D-maltoside extraction of the mitochondrial membranes followed by separation on an anionic MonoQ FPLC column and size exclusion chromatography as described previously (Eriksson et al., 1994). Protein content was determined using the method described by Peterson (1977).
In vitro transcription and translation of the mitochondrial precursor protein Transcription and translation of N. plumbaginifolia F1b in pTZ18U vector (a gift from Professor M Boutry) (Boutry and Chua, 1985) was carried out using the TNTâ-coupled reticulocyte lysate system (Promega, http://www.sdsscan.se) with T7 RNA polymerase in the presence of [35S]-methionine.
Processing of precursor proteins Reactions for processing contained 1 ml radiolabelled translation product (about 10 000±15 000 counts min±1) and 1 mg spinach MPP/bc1 complex in 20 mM HEPES, pH 8.0, 2.5 mM MnCl2, in a ®nal volume of 17 ml. MPP was incubated prior to the processing reaction with the synthetic peptides derived from the presequence, for 5 min at 30°C. Processing was carried out for 5, 10, 20 and 40 min at 30°C, and stopped by the addition of double strength sample buffer (1:1, by volume) and solubilized at 90°C for 2 min. The solubilized samples were applied to 12% SDS±PAGE, according to Laemmli (1970), and analysed using a phosphoimager. The ef®ciency of processing was calculated as the ratio of the mature form to the sum of the precursor and the mature form.
Site-directed mutagenesis of the presequence Mutants of the presequence of the F1b subunit of the ATPase in N. plumbaginifolia were created using a Quick-Change Site Directed Mutagenesis kit (Stratagene, http://www.stratagene.com/).
436
Xiao-Ping Zhang et al.
Peptide synthesis
Molecular docking of presequence peptides to MPP
The 22-residue peptide sequences shown in Figure 3(a) were synthesized by solid-phase methods as C-terminal carboxamides on p-methylbenzhydrylamine resin using standard Boc chemistry protocols (Merri®eld, 1986). Treatment of the peptide resins with HF/p-cresol (9:1, 0°C, 1 h) and reverse-phase HPLC puri®cation of the resulting crude products (C18-silica, MeCN±water gradients (+0.05% TFA)) led to the target peptides, which had the expected amino acid compositions and matrix assisted laser desorption/ ionization time-of-¯ight (MALDI-TOF) mass spectra.
Polar hydrogen atoms were added and partial charges were assigned for the peptide in Quanta (MSI, San Diego, California, USA; http://www.msi.com/). The pdbq format of a peptide was produced manually by merging charge information from a bgf ®le into a pdb ®le in a text editor. The interaction of the signal peptides with MPP was analysed in Autodock 3.0.5 (Morris et al., 1998) by docking a fully ¯exible presequence peptide to MPP peptide binding cleft using the Lamarckian Genetic Algorithm (LGA). The parameters used were: grid spacing 0.375 AÊ, grid point 120,120,120, 50 runs, a population size of 50, and a runtermination criterion of a maximum of 27 000 generations or a maximum of 150 000 energy evaluations, whichever came ®rst. The rmsd conformational clustering tolerance was 0.5 AÊ and other parameters were by default.
Circular dichroism analysis Circular dichroism spectra of the peptides were recorded on a JASCO J-720 spectropolarimeter. The samples were scanned from 300 to 190 nm at 20°C with a path length of 0.1 cm and accumulated 15±30 times. Peptide concentrations were typically 25 mM. The buffer was 50 mM sodium phosphate, pH 7.5. The ellipticity was measured in buffer as well as in 50% HFP (1,1,1,3,3,3-hexa¯uoro-2-propanol) (1:1, by volume). A mixed aqueous solvent with 50% HFP was used to stabilize the a-helix secondary structures to facilitate comparison between the helixforming propensities of the different peptides (Backlund et al., 1994). Spectra were corrected with respect to baseline and for dilution. From the measured circular dichroism spectra, the mean residue molar ellipticity value at wavelengths 208 nm (Green®eld and Fasman, 1969) and 222 nm (Morrisett et al., 1977) was used to estimate the helical content.
Acknowledgements This work was supported by grants from the Swedish Natural Science Research Council to E.G. and A.G. Work in Barcelona was supported by the Centre de RefereÁncia en Biotecnologia (Generalitat de Catalunya, Spain). X.P.Z. is grateful to Dr Arne Elofsson for helpful discussions on protein modelling and data processing, Professor PaÈr Nordlund and Professor Peter Brzezinski for making their Silicon Graphic Inc. computers accessible (Quanta and GRASP, respectively) and the Helge Ax:son Johnsons foundation for the stipend for purchasing a computer. We thank Professor Gunnar von Heijne for stimulating discussions and Dr Patrick Dessi for critical reading of the manuscript.
Computer modelling of MPP and pF1b presequence peptide
References
Sequences of Salanum tuberosum a-MPP and b-MPP were from SwissProt-P29677 and NCBI-X80237, respectively (Emmermann et al., 1994). The sequences were aligned with core 2 and core 1 subunits (SwissProt-P23004 and NCBI-X59692, respectively) of bovine bc1 complex [protein data bank (PDB) code, 1BGY] (Iwata et al., 1998) in Phi-BLAST. The three-dimensional models of the subunits of MPP were built in MODELLER 4.0 (SÏali et al., 1993) with default model-building procedures. The quality of the model was checked by WHATIF 97 (Vriend, 1990) and PROCHECK 3.0 (Laskowski et al., 1993). No serious error was found, except that ®ve residues, Glu262, Ser384 and Ile402 on a-MPP and Ile61 and Leu423 on b-MPP, were in disallowed regions on the Ramachandran plot. These residues are all in loop regions on the outside of the molecule. The Empirical Conformational Energy Program for Peptides (ECEPPAK) (Ripoll et al., 1998) was used to search the native-like conformation of the pF1b presequence peptide (22 amino acid residues from the C-terminal part of the presequence) on an Intel/ Pentium III PC-Linux workstation using the electrostatically driven Monte Carlo (EDMC) module. The parameters were maxit = 1000, seed = ±513 482, thermal_shock t_low = 200, t_up = 50 000, OMEGA_180. The conformation with the lowest energy (± 302.84 kcal mol±1) was found in iteration 634. The molecular images were created with MOLSCRIPT 2.1.2 (Kraulis, 1991), GRASP 1.3.6 (Nicholls et al., 1991), Raster3D 2.5 (Merritt and Bacon, 1997) and WebLab ViewerLiteä 3.2 (MSI, San Diego, California, USA; http://www.msi.com/). Labelling was done in GIMP 1.1.2 (http://www.gimp.org/) and ImageMagick 5.2.0 (ImageMagick Studio, http://www.imagemagick.org/).
Abe, Y., Shodai, T., Muto, T., Mihara, K., Torii, H., Nishikawa, S., Endo, T. and Kohda, D. (2000) Structural basis of presequence recognition by the mitochondrial protein import receptor Tom20. Cell, 100, 551±560. Arretz, M., Schneider, H., Wienhues, U. and Neupert, W. (1991) Processing of mitochondrial precursor proteins. Biomed. Biochim. Acta, 50, 403±412. Arretz, M., Schneider, H., Guiard, B., Brunner, M. and Neupert, W. (1994) Characterization of the mitochondrial processing peptidase of Neurospora crassa. J. Biol. Chem. 269, 4959±4967. Backlund, B.M., Wikander, G., Peeters, T.L. and GraÈslund, A. (1994) Induction of secondary structure in the peptide hormone motilin by interaction with phospholipid vesicles. Biochim. Biophys. Acta, 1190, 337±344. Boutry, M. and Chua, N.H. (1985) A nuclear gene encoding the beta subunit of the mitochondrial ATP synthase in Nicotiana plumbaginifolia. EMBO J. 4, 2159±2165. Braun, H.P., Emmermann, M., Kruft, V. and Schmitz, U.K. (1992) The general mitochondrial processing peptidase from potato is an integral part of cytochrome c reductase of the respiratory chain. EMBO J. 11, 3219±3227. Dessi, P., Rudhe, C. and Glaser, E. (2000) Studies on the topology of the protein import channel in relation to the plant mitochondrial processing peptidase integrated into the cytochrome bc1 complex. Plant J. 24, 637±644. Emmermann, M., Braun, H.P. and Schmitz, U.K. (1994) The mitochondrial processing peptidase from potato: a selfprocessing enzyme encoded by two differentially expressed genes. Mol. Gen. Genet. 245, 237±245. ã Blackwell Science Ltd, The Plant Journal, (2001), 27, 427±438
Speci®city of the mitochondrial processing peptidase Eriksson, A.C., SjoÈling, S. and Glaser, E. (1993) A general processing proteinase of spinach leaf mitochondria is associated with the bc1 complex of the respiratory chain. In: Plant Mitochondria (Brennicke, A. and Kuch, U., eds). Weinheim: VCH Verlagsgesellschaft, pp. 233±241. Eriksson, A.C., SjoÈling, S. and Glaser, E. (1994) The ubiquinol cytochrome c oxidoreductase complex of spinach leaf mitochondria is involved in both respiration and protein processing. Biochim. Biophys. Acta, 1186, 221±231. Gavel, Y. and von Heijne, G. (1990) Cleavage-site motifs in mitochondrial targeting peptides. Protein Eng. 4, 33±37. Glaser, E. and Dessi, P. (1999) Integration of the mitochondrialprocessing peptidase into the cytochrome bc1 complex in plants. J. Bioenerg. Biomembr. 31, 259±274. Glaser, E., SjoÈling, S., Tanudji, M. and Whelan, J. (1998) Mitochondrial protein import in plants. Signals, sorting, targeting, processing and regulation. Plant Mol. Biol. 38, 311±338. Green®eld, N. and Fasman, G.D. (1969) Computed circular dichroism spectra for the evaluation of protein conformation. Biochemistry, 8, 4108±4116. Hamasur, B., Birgersson, U., Eriksson, A.C. and Glaser, E. (1990) Large-scale puri®cation procedure of spinach leaf mitochondria ± isolation and immunological studies of the F1-ATPase. Physiol. Plant. 78, 367±373. Hammen, P.K., Gorenstein, D.G. and Weiner, H. (1994) Structure of the signal sequences for two mitochondrial matrix proteins that are not proteolytically processed upon import. Biochemistry, 33, 8610±8617. Hammen, P.K., Waltner, M., Hahnemann, B., Heard, T.S. and Weiner, H. (1996) The role of positive charges and structural segments in the presequence of rat liver aldehyde dehydrogenase in import into mitochondria. J. Biol. Chem. 271, 21041±21048. von Heijne, G. (1986) Mitochondrial targeting sequences may form amphiphilic helices. EMBO J. 5, 1335±1342. von Heijne, G., Steppuhn, J. and Herrmann, R.G. (1989) Domain structure of mitochondrial and chloroplast targeting peptides. Eur. J. Biochem. 180, 535±545. Isaya, G., Kalousek, F., Fenton, W.A. and Rosenberg, L.E. (1991) Cleavage of precursors by the mitochondrial processing peptidase requires a compatible mature protein or an intermediate octapeptide. J. Cell Biol. 113, 65±76. Ito, A. (1999) Mitochondrial processing peptidase: multiple-site recognition of precursor proteins. Biochem. Biophys. Res. Commun. 265, 611±616. Iwata, S., Lee, J.W., Okada, K., Lee, J.K., Iwata, M., Rasmussen, B., Link, T.A., Ramaswamy, S. and Jap, B.K. (1998) Complete structure of the 11-subunit bovine mitochondrial cytochrome bc1 complex. Science, 281, 64±71. Jarvis, J.A., Ryan, M.T., Hoogenraad, N.J., Craik, D.J. and Hoj, P.B. (1995) Solution structure of the acetylated and noncleavable mitochondrial targeting signal of rat chaperonin 10. J. Biol. Chem. 270, 1323±1331. Kitada, S., Shimokata, K., Niidome, T., Ogishima, T. and Ito, A. (1995) A putative metal-binding site in the b subunit of rat mitochondrial processing peptidase is essential for its catalytic activity. J. Biochem. 117, 1148±1150. Klaus, C., Guiard, B., Neupert, W. and Brunner, M. (1996) Determinants in the presequence of cytochrome b2 for import into mitochondria and for proteolytic processing. Eur. J. Biochem. 236, 856±861. Kneller, D.G., Cohen, F.E. and Langridge, R. (1990) Improvements in protein secondary structure prediction by an enhanced neural network. J. Mol. Biol. 214, 171±182. ã Blackwell Science Ltd, The Plant Journal, (2001), 27, 427±438
437
Kojima, K., Kitada, S., Shimokata, K., Ogishima, T. and Ito, A. (1998) Cooperative formation of a substrate binding pocket by a- and b-subunits of mitochondrial processing peptidase. J. Biol. Chem. 273, 32542±32546. Kraulis, P.J. (1991) MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallogr. 24, 994±950. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature, 227, 680±685. Lain, B., Yanez, A., Iriarte, A. and Martinez-Carrion, M. (1998) Aminotransferase variants as probes for the role of the Nterminal region of a mature protein in mitochondrial precursor import and processing. J. Biol. Chem. 273, 4406±4415. Lancelin, J.M., Gans, P., Bouchayer, E., Bally, I., Arlaud, G.J. and Jacquot, J.P. (1996) NMR structures of a mitochondrial transit peptide from the green alga Chlamydomonas reinhardtii. FEBS Lett. 391, 203±208. Laskowski, R.A., MacArthur, M.W., Moss, D.S. and Thornton, J.M. (1993) PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallogr. 26, 283±291. Luciano, P. and GeÂli, V. (1996) The mitochondrial processing peptidase: function and speci®city. Experientia, 52, 1077±1082. Luciano, P., Geoffroy, S., Brandt, A., Hernandez, J.F. and GeÂli, V. (1997) Functional cooperation of the mitochondrial processing peptidase subunits. J. Mol. Biol. 272, 213±225. Merri®eld, R. (1986) Solid phase synthesis. Science, 232, 341±347. Merritt, E.A. and Bacon, D.J. (1997) Raster3D: photorealistic molecular graphics. Methods Enzymol. 277, 505±524. Morris, G.M., Goodsell, D.S., Halliday, R.S., Huey, R., Hart, W.E., Belew, R.K. and Olson, A.J. (1998) Automated docking using a Lamarckian genetic algorithm and an empirical binding free energy function. J. Comput. Chem. 19, 1639±1662. Morrisett, J.D., Jackson, R.L. and Gotto, A.M., Jr (1977) Lipid±protein interactions in the plasma lipoproteins. Biochim. Biophys. Acta, 472, 93±133. Muto, T., Obita, T., Abe, Y., Shodai, T., Endo, T. and Kohda, D. (2001) NMR and identi®cation of the Tom20 binding segment in mitochondrial presequences. J. Mol. Biol. 306, 137±143. Neupert, W. (1997) Protein import into mitochondria. Annu. Rev. Biochem. 66, 863±917. Nicholls, A., Sharp, K.A. and Honig, B. (1991) Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins, 11, 281±296. Niidome, T., Kitada, S., Shimokata, K., Ogishima, T. and Ito, A. (1994) Arginine residues in the extension peptide are required for cleavage of a precursor by mitochondrial processing peptidase. Demonstration using synthetic peptide as a substrate. J. Biol. Chem. 269, 24719±24722. Ogishima, T., Niidome, T., Shimokata, K., Kitada, S. and Ito, A. (1995) Analysis of elements in the substrate required for processing by mitochondrial processing peptidase. J. Biol. Chem. 270, 30322±30326. Ou, W.J., Kumamoto, T., Mihara, K., Kitada, S., Niidome, T., Ito, A. and Omura, T. (1994) Structural requirement for recognition of the precursor proteins by the mitochondrial processing peptidase. J. Biol. Chem. 269, 24673±24678. Peterson, G.L. (1977) A simpli®cation of the protein assay method of Lowry et al. which is more generally applicable. Anal. Biochem. 83, 346±356. Rawlings, N.D. and Barrett, A.J. (1995) Evolutionary families of metallopeptidases. Methods Enzymol. 248, 183±228. Ripoll, D.R., Liwo, A. and Scheraga, H.A. (1998) New developments of the electrostatically driven Monte Carlo
438
Xiao-Ping Zhang et al.
method: test on the membrane-bound portion of melittin. Biopolymers, 46, 117±126. Roise, D., Horvath, S.J., Tomich, J.M., Richards, J.H. and Schatz, G. (1986) A chemically synthesized pre-sequence of an imported mitochondrial protein can form an amphiphilic helix and perturb natural and arti®cial phospholipid bilayers. EMBO J. 5, 1327±1334. SÏali, A., Matsumoto, R., McNeil, H.P., Karplus, M. and Stevens, R.L. (1993) Three-dimensional models of four mouse mast cell chymases. Identi®cation of proteoglycan binding regions and protease-speci®c antigenic epitopes. J. Biol. Chem. 268, 9023±9034. Schatz, G. and Dobberstein, B. (1996) Common principles of protein translocation across membranes. Science, 271, 1519±1526. Schneider, G., SjoÈling, S., Wallin, E., Wrede, P., Glaser, E. and von Heijne, G. (1998) Feature-extraction from endopeptidase cleavage sites in mitochondrial targeting peptides. Proteins, 30, 49±60. Schneider, T.D. and Stephens, R.M. (1990) Sequence logos: a new way to display consensus sequences. Nucl. Acids Res. 18, 6097±6100. Schulte, U., Arretz, M., Schneider, H., Tropschug, M., Wachter, E., Neupert, W. and Weiss, H. (1989) A family of mitochondrial proteins involved in bioenergetics and biogenesis. Nature, 339, 147±149. Shimokata, K., Kitada, S., Ogishima, T. and Ito, A. (1998) Role of a-subunit of mitochondrial processing peptidase in substrate recognition. J. Biol. Chem. 273, 25158±25163. SjoÈling, S. and Glaser, E. (1998) Mitochondrial targeting peptides in plants. Trends Plant Sci. 3, 136±140. SjoÈling, S., Eriksson, A.C. and Glaser, E. (1994) A helical element in the C-terminal domain of the N. plumbaginifolia F1b presequence is important for recognition by the
mitochondrial processing peptidase. J. Biol. Chem. 269, 32059±32062. SjoÈling, S., Waltner, M., Glaser, E. and Weiner, H. (1996) Studies on protein processing for membrane-bound spinach leaf mitochondrial processing peptidase (MPP) integrated into the cytochrome bc1 complex and the soluble rat liver matrix mitochondrial processing peptidase. Eur. J. Biochem. 242, 114±121. Tanudji, M., SjoÈling, S., Glaser, E. and Whelan, J. (1999) Signals required for the import and processing of the alternative oxidase into mitochondria. J. Biol. Chem. 274, 1286±1293. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F. and Higgins, D.G. (1997) The CLUSTAL_X windows interface: ¯exible strategies for multiple sequence alignment aided by quality analysis tools. Nucl. Acids Res. 25, 4876±4882. Thornton, K., Wang, Y., Weiner, H. and Gorenstein, D.G. (1993) Import, processing, and two-dimensional NMR structure of a linker-deleted signal peptide of rat liver mitochondrial aldehyde dehydrogenase. J. Biol. Chem. 268, 19906±19914. Voos, W., Martin, H., Krimmer, T. and Pfanner, N. (1999) Mechanisms of protein translocation into mitochondria. Biochim. Biophys. Acta, 1422, 235±254. Vriend, G. (1990) WHAT IF: a molecular modeling and drug design program. J. Mol. Graph. 8, 52±56. Waltner, M. and Weiner, H. (1995) Conversion of a nonprocessed mitochondrial precursor protein into one that is processed by the mitochondrial processing peptidase. J. Biol. Chem. 270, 26311±26317. Xia, D., Yu, C.A., Kim, H., Xia, J.Z., Kachurin, A.M., Zhang, L., Yu, L. and Deisenhofer, J. (1997) Crystal structure of the cytochrome bc1 complex from bovine heart mitochondria. Science, 277, 60±66. Zhang, X.P., Elofsson, A., Andreu, D. and Glaser, E. (1999) Interaction of mitochondrial presequences with DnaK and mitochondrial hsp70. J. Mol. Biol. 288, 177±190.
ã Blackwell Science Ltd, The Plant Journal, (2001), 27, 427±438
