Resource
Global Analysis of the Mitochondrial
N-Proteome Identifies a Processing
Peptidase Critical for Protein Stability
F.-Nora Vo¨gtle,1,2,10Stefanie Wortelkamp,3,10Rene´ P. Zahedi,3Dorothea Becker,1,4Claudia Leidhold,1Kris Gevaert,5,6 Josef Kellermann,7Wolfgang Voos,1,4Albert Sickmann,3,8,*Nikolaus Pfanner,1,9,*and Chris Meisinger1,9
1Institut fu¨r Biochemie und Molekularbiologie, ZBMZ, Universita¨t Freiburg, 79104 Freiburg, Germany 2Fakulta¨t fu¨r Biologie, Universita¨t Freiburg, 79104 Freiburg, Germany
3ISAS, Institute for Analytical Sciences, 44139 Dortmund, Germany
4Institut fu¨r Biochemie und Molekularbiologie, Universita¨t Bonn, 53115 Bonn, Germany 5Department of Medical Protein Research, VIB, 9000 Ghent, Belgium
6Department of Biochemistry, Ghent University, 9000 Ghent, Belgium 7Max Planck Institute of Biochemistry, 82152 Martinsried, Germany
8Medizinisches Proteom-Center (MPC), Ruhr-Universita¨t Bochum, 44801 Bochum, Germany 9Centre for Biological Signalling Studies (bioss), Universita¨t Freiburg, 79104 Freiburg, Germany 10These authors contributed equally to this work
*Correspondence:[email protected](A.S.),[email protected](N.P.) DOI 10.1016/j.cell.2009.07.045
SUMMARY
Many mitochondrial proteins are synthesized with
N-terminal presequences that are removed by
specific peptidases. The N-termini of the mature
pro-teins and thus peptidase cleavage sites have only
been determined for a small fraction of mitochondrial
proteins and yielded a controversial situation for the
cleavage site specificity of the major mitochondrial
processing peptidase (MPP). We report a global
analysis of the N-proteome of yeast mitochondria,
revealing the N-termini of 615 different proteins.
Significantly more proteins than predicted contained
cleavable presequences. We identified the
interme-diate cleaving peptidase Icp55, which removes an
amino acid from a characteristic set of
MPP-gener-ated N-termini, solving the controversial situation of
MPP specificity and suggesting that Icp55 converts
instable intermediates into stable proteins. Our
results suggest that Icp55 is critical for stabilization
of the mitochondrial proteome and illustrate how
the N-proteome can serve as rich source for a
systematic analysis of mitochondrial protein
target-ing, cleavage and turnover.
INTRODUCTION
Many proteins of cell organelles such as mitochondria, chloro-plasts and endoplasmic reticulum are synthesized with signal sequences that are proteolytically removed by specific enzymes. For the majority of organellar proteins, however, the mature N-termini have not been determined experimentally. In case of
mitochondria, the most comprehensive studies on their protein composition have been performed in the yeast Saccharomyces
cerevisiae. Proteomic analyses indicated that yeast
mitochon-dria contain1000 and human mitochondria 1200–1500 dif-ferent proteins that function in numerous processes, including bioenergetics, protein translocation, folding and degradation, metabolism of amino acids, lipids, heme and iron, genome main-tenance, membrane dynamics, signaling and apoptosis ( Sick-mann et al., 2003; Prokisch et al., 2004; Reinders et al., 2006; Pagliarini et al., 2008).
Ninety-nine percent of mitochondrial proteins are synthesized in the cytosol and translocated into the organelle. Many proteins carry positively charged N-terminal targeting signals, termed presequences, which are removed upon import (Dolezal et al.,
2006; Neupert and Herrmann, 2007; Bolender et al., 2008).
Pre-sequences direct the import of most matrix proteins, a significant fraction of inner membrane proteins and several intermembrane space proteins. The major presequence protease is the mito-chondrial processing peptidase (MPP) located in the matrix
(Taylor et al., 2001; Neupert and Herrmann, 2007). The
mitochon-drial intermediate peptidase (Oct1) functions after MPP and typically removes an octapeptide from several preproteins
(Gakh et al., 2002). The inner membrane protease (IMP) also
functions after MPP and removes a hydrophobic sorting signal that is located behind the matrix targeting signal (Gakh et al.,
2002; Neupert and Herrmann, 2007). In addition, the m-AAA
protease and the rhomboid protease Pcp1 of the inner membrane were shown to cleave a few preproteins (Esser et al., 2002;
Nolden et al., 2005). Mitochondrial processing defects are
involved in human diseases such as hereditary spastic paraplegia (Nolden et al., 2005).
However, these observations are based on the analysis of a small subset of preproteins. Since the exact N-termini of the mature proteins and thus the processing sites have only been
determined for a minor fraction of the mitochondrial proteome, we have only limited information on preprotein processing and its functional relevance. As a consequence, the widely used prediction programs (Emanuelsson et al., 2007) could only be trained with a small number of proteins and the prediction of cleavage sites is often not successful. A major unsolved question is the cleavage site specificity of MPP. While an arginine in the C-terminal region of the presequence seems to be a critical determinant for cleavage, it is not understood why in the two most frequent MPP motifs, the arginine residue is either located at position 2 (R-2 motif) or at position 3 (R-3 motif) from the cleavage site (Schneider et al., 1998; Emanuelsson et al., 2001,
2007; Gakh et al., 2002; Habib et al., 2007). Numerous mutational
studies indicated that R-2 as well as R-3 were important for cleavage by MPP (summarized inGakh et al., 2002) and thus an experimental solution has not been achieved so far.
Here we report a global analysis of mitochondrial processing sites by determining the mature N-termini of the majority of known mitochondrial proteins. The resulting N-proteome re-vealed that cleavable presequences are significantly more frequent than predicted. We identified a new processing pepti-dase, Icp55, which removes a single amino acid residue after the MPP cleavage site and thus converts the R-3 motif to an R-2 motif, solving the controversial situation regarding MPP cleavage. A systematic analysis suggests that Icp55 is critical for the stability of the mitochondrial proteome. Icp55 cleaves preprotein intermediates that carry a destabilizing N-terminal amino acid residue according to the N-end rule for bacterial proteins (Varshavsky, 1996, 2008; Mogk et al., 2007). Our study suggests that the intermediate peptidase stabilizes mitochon-drial proteins by converting destabilizing N-termini, which were generated by the action of MPP, into stable N-termini. The N-proteome represents the first systematic resource for the characterization of authentic mitochondrial presequences and their processing and thus provides an important foundation for the analysis of mitochondrial protein biogenesis and turnover.
RESULTS
Identification and Validation of the Mitochondrial N-Proteome
The experimental strategy for a systematic identification of the N-termini of mature yeast mitochondrial proteins is outlined in
Figure 1. A crude mitochondrial preparation was isolated from
S. cerevisiae by differential centrifugation. A highly purified
mito-chondrial fraction was then prepared by sucrose gradient centri-fugation (Sickmann et al., 2003) and subjected to combined fractional diagonal chromatography (COFRADIC) according to
Gevaert et al. (2003). Free amino groups were blocked by
acet-ylation. After digestion with trypsin, chymotrypsin or V8 pro-tease, a first reverse-phase (RP)-HPLC run was performed (
Fig-ure 1A). The free N-termini of internal peptides (generated by the
proteases) were modified by 2,4,6-trinitrobenzenesulfonic acid (TNBS). The strong hydrophobicity of the TNB-group led to a retention time shift of internal peptides in a secondary, identical RP-HPLC run. The N-terminal peptides were eluted in the same fraction in both HPLC runs and analyzed by LC-MS/MS. The N-peptides were compared with the 851 proteins of the most
comprehensive proteomic analysis of yeast mitochondria, the PROMITO dataset (peptides starting from residues 1–100; see
Supplemental Experimental Procedures available with this
article online;Reinders et al., 2006). This led to the identification of the N-termini of 615 different proteins (Figure 1B and Table S1).
Two approaches were used to assess the validity of the deter-mined N-termini (Figure 1B). (1) We performed a systematic search of the original literature and found that the N-termini of 50 N-proteome proteins had been experimentally determined in single protein studies (control set,Table S2). The N-proteome agreed with 47 of the 50 reported N-termini (94% accuracy). (2) We imported radiolabeled preproteins into mitochondria and compared the in organello processing product with in vitro synthesized truncated versions of the proteins according to the determined N-termini of the N-proteome. In each case, the mobility of the processed preprotein on SDS-PAGE was indistin-guishable from the truncated version derived from the N-pro-teome (Figure S1A). We also tested proteins that were found not to contain presequences according to the N-proteome (N-terminal peptide starting with amino acid residue 1 or 2). These proteins, which are termed noncleavable precursors in the mitochondrial import field, were imported into mitochondria but did not undergo further cleavage (Figure S1B). The in organ-ello import experiments thus fully agreed with the N-proteome data for both cleavable and noncleavable precursors.
For a number of proteins such as Gif1 and Pdb1, more than one N-terminus was found (Tables S1 and S3). We thus syn-thesized two truncated versions of the proteins and compared them to the in organello processing product (Figure S1A). The N-terminus identified with the higher frequency typically corre-sponded to the in organello processed mature protein, whereas the low frequency peptides were likely derived from processing intermediates (Tables S1–S3; see also below the detection of intermediates of Icp55 and Oct1). We generated the N-proteome master set of 585 N-termini that were reproducibly identified (see
Supplemental Experimental Procedures) and deduced the
pre-sequences (Table S3).
The distribution of proteins over the four mitochondrial sub-compartments (14% outer membrane, 8% intermembrane space, 39% inner membrane and 39% matrix;Table S3) agreed well with the typical distribution of mitochondrial proteins in the
Saccharomyces genome database (SGD), indicating that the
analysis was not biased against individual mitochondrial sub-compartments.
A possible limitation of the N-proteome is a cleavage of mito-chondrial proteins by unspecific or nonmitomito-chondrial proteases that are copurified during isolation. The protocol for the isolation of highly pure mitochondria used here minimized this problem since the cells were homogenized in the presence of a protease inhibitor cocktail, other organelles such as lysosomes were rapidly removed and the mitochondria were only lysed after removal of the contaminants (Sickmann et al., 2003). Surface-exposed mitochondrial outer membrane proteins like Tom proteins are the most sensitive markers for unspecific cleavage. Western blot analysis indicated that the isolated mitochondria contained intact Tom proteins (Sickmann et al., 2003). Tom22 and the small Tom proteins expose their N-termini to the cytosol
centrifugation Sucrose gradient
S. cerevisiae Crude mitochondria
Trypsin V8
chymotrypsin Acetylation of free
amino groups Digestion Internal peptides N-terminal peptides Lysis AU t 16 12 Internal peptides shift due to TNBS 1st RP-HPLC with TNBS 2nd RP-HPLC 615 Edman sequencing (615 proteins) ~27,000 spectra of icp55Δmitochondria Oct1 substrates 7 publ. 38 Icp55 sub- strates Ac-N K K COOH Ac-N K K COOH (~665,000 spectra) 12 Protease screen Comparison to prediction programs (TargetP, MitoProt II) Validation by in vitro import Validation by published N-termini (Edman degradation) 47 3 PROMITO 100 Dataset In vitro import A B 2D-DIGE analysis In vitro protein import MRM COOH K N COFRADIC Ac Ac Ac Ac Ac AU t 16 12 N-terminal peptides C K Ac K Ac Total y east
MitochondriaHighly pure mi to. - Tim23 (IM) - Aco1 (Matrix) - Sss1 (ER) - Vam3 (Vac.) - Pex14 (Perox.) - Cox4 (IM)
Figure 1. Analysis of the Mitochondrial N-Proteome
(A) Scheme of COFRADIC analysis. Highly pure yeast mitochondria were obtained by sucrose gradient centrifugation and the purity was confirmed by immuno-blotting. N-terminal and internal peptides were separated by COFRADIC. Abbreviations are as follows: Ac, acetyl; IM, mitochondrial inner membrane; Perox., peroxisomes; Vac., vacuole.
(B) Overview of data processing and analysis. Validation of determined N-termini by in vitro import and processing by isolated mitochondria is illustrated in
and thus a cleavage could be monitored by the COFRADIC anal-ysis. In each case, peptides starting with residue 1 or 2 were the most abundant ones and peptides starting with internal residues were of lower abundance (Tables S1 and S3). A further potential limitation would be an incomplete labeling of internal peptides with TNBS. We thus performed four consecutive cycles of TNBS labeling for each sample, leading a labeling yield of more than 98% (Gevaert et al., 2003). In addition, to distinguish N-terminal peptides from internal peptides that escaped TNBS-labeling, the difference in acetylation with a trideutero-acetyl group was taken into account. We conclude that our experi-mental approach combined with the analysis of the most frequently identified N-peptides strongly minimized the prob-lems of unspecific proteolysis and incomplete TNBS-labeling and yielded the true N-terminus with high confidence also for surface exposed proteins.
Presequences and Predictions
So far, the characteristics of most mitochondrial presequences were derived from prediction programs that were trained on a small set of proteins with experimentally determined N-termini. Using the N-proteome presequences, the majority of authentic mitochondrial presequences could now be systematically analyzed. We observed a good agreement for the basic charac-teristics of presequences (length, charge), but noted strong differences between the N-proteome and prediction programs
with regard to cleavage sites for the majority of precursor proteins.
Presequences were predicted to typically possess a positive net charge and a length of20–60 amino acid residues (Gakh
et al., 2002; Habib et al., 2007; Emanuelsson et al., 2007).
Simi-larly, the most frequent net charges of the N-proteome prese-quences were +3 to +6 and the predominant length distribution was 15–55 amino acid residues (Figure 2A). In addition to a signif-icant number of long presequences (>65 residues), very short presequences were also observed. The 6-residue extension of Atp17 was the shortest presequence of the N-proteome, for which import and processing by isolated mitochondria in a membrane potential (Dc)- and MPP-dependent manner was observed (Figure S1C andTable S3).
Seven substrates of the mitochondrial intermediate peptidase Oct1 have been known in yeast mitochondria (Branda and Isaya, 1995; Schneider et al., 1998; Gakh et al., 2002). We identified five more substrates by inspection of the N-proteome and importing radiolabeled preproteins into wild-type and oct1D mitochondria (accumulation of the intermediate forms in oct1D mitochondria;
Figure S1D and Supplemental Experimental Procedures).
COFRADIC analysis led to the identification of both intermediate and mature forms for the new substrate Cpr3 in wild-type mito-chondria, with the typical 8-residue difference (Tables S1 and S3). The relative frequency of amino acid residues of the twelve Oct1 substrates is shown in Figure S1E, including arginine A
Length of presequence (# amino acids)
10 20 30 5-9 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89 90-94 95-99 0 Number of proteins N-proteome Control set
Net charge of presequence
10 20 30 40 50 Number of proteins 0 - 13 - 11 - 9 - 7 - 5 - 3 - 1 + 1 + 3 + 5 + 7 + 9 + 11 + 13 + 15 Control set N-proteome C 0 100 50 1 2 MitoProt II TargetP Cleavable proteins 25 75 MitoProt II TargetP Non-cleavable proteins 0 100 50 3 4 25 75 5 6 MitoProt IITargetP Total Clea v age sit e pr edic ted (% of N-pr oteome) 0 100 50 25 75 C orr ec t clea v age site (% of control set)
0
100
50
1
2
3
Predicted 25 75 B MitoProt II TargetP N-proteome +_1 +_1Figure 2. Mitochondrial Presequences Deduced from the N-Proteome
(A) Analysis of presequence length and net charge for the proteins identified in the N-proteome master set (Table S3andSupplemental Experimental Procedures)
and control set (N-termini previously determined in single protein studies;Table S2andSupplemental References).
(B) Comparison of the cleavage sites of proteins from the control set (Table S2) with the cleavage sites determined in the N-proteome master set (Table S3) and
cleavage sites predicted by MitoProt II and TargetP (for the prediction programs, the cleavage site ± 1 residue was counted).
(C) Comparison of the cleavage sites determined in the N-proteome master set (Table S3) with the cleavage sites predicted by MitoProt II and TargetP
(the cleavage site ± 1 residue was counted). Columns 3 and 4, and 5 and 6 separate between cleavable (279) and noncleavable precursor proteins (122), respectively.
at 10 (for cleavage by MPP) and a large hydrophobic residue at position 8 (beginning of the octapeptide) (Branda and Isaya, 1995; Schneider et al., 1998; Gakh et al., 2002; Emanuelsson et al., 2007).
Cleavable preproteins (69.6%) were significantly more fre-quent than noncleavable proteins (30.4%) in the N-proteome master set. We used the N-proteome to test two widely used programs for the prediction of mitochondrial cleavage sites, MitoProt II (Claros and Vincens, 1996) and TargetP (
Emanuels-son et al., 2001, 2007). With regard to the control set of 50
pro-teins, where the N-termini were determined in previous studies
(Table S2), TargetP correctly predicted 62% of the mature
N-termini and MitoProt II 36%, including the prediction of non-cleavable proteins—compared to the accuracy of 94% of the N-proteome (Figure 2B). We then tested the prediction programs for the large number of proteins of the N-proteome. TargetP pre-dicted 41% of the mature N-termini and MitoProt II 34%, when both cleavable and noncleavable proteins were counted (
Fig-ure 2C, left panel). The accuracy of prediction differed
signifi-cantly between cleavable and noncleavable proteins. For pre-proteins with cleavable presequences, the processing site was only predicted in 21% (MitoProt II) to 33% (TargetP) of the cases, while the lack of cleavage was predicted in 61%–64% (Figure 2C, columns 3–6). For the prediction programs, we also counted the sites that differed by one residue from the correct site (as the
1-residue processing by Icp55 described below was not known when the programs were designed), whereas for the determina-tion of the accuracy of the N-proteome (Figure 2B, column 1) only the exact sites were used.
In summary, cleavable preproteins form the most abundant class of mitochondrial proteins, yet the mature N-termini of the vast majority of cleaved proteins are not correctly predicted. Tar-getP showed a better performance than MitoProt II, yet only cor-rectly predicted a third of the cleavage sites of presequence-carrying proteins within a range of ± 1 residue. Both programs were significantly better for predicting a lack of cleavage (up to two third of noncleavable proteins). TargetP predicted the pub-lished N-termini (control set) with significantly higher accuracy than the N-termini of the N-proteome, likely because the pro-gram was trained on published N-termini.
Identification of the Intermediate Cleaving Peptidase Icp55
To define the cleavage motif in mitochondrial preproteins, we compiled a reference set of very high confidence presequences derived from N-termini that were identified at least ten times (see
Supplemental Experimental Procedures). The reference set
showed a predominance of arginine residues at positions 2
and 3 (Figure 3, upper panel). For several proteins with R-3
we found two N-termini by COFRADIC, typically the mature
bits
0
1
Position (relative to mature N-termini)
+1 -2 -3 -5 -4 -7 -8 -9 -10 -11 -6 -12 -13 -14 -16 -15 -18 -19 -20 -17 bits
Position (relative to mature N-termini)
0
1
-1 -2 -3 -5 -4 -7 -8 -9 -10 -11 -6 -12 -13 -14 -16 -15 -18 -19 -20 -172
+2 +3 +4 +5 -1 +1 +2 +3 +4 +52
e r u t a M e c n e u q e s e r P Presequence (without residue removed by Icp55)Mature (+1 residue for Icp55 substrates)
Figure 3. Amino Acid Frequency at Cleavage Sites of Mitochondrial Proteins
Relative frequency of amino acids in presequences (up to 20 residues of the C-terminal segment) and the first five residues of mature proteins identified in the
N-proteome reference set (seeSupplemental Experimental Procedures). Only presequences with 10 or more residues were analyzed. Upper panel, alignment
according to N-proteome reference set; lower panel, Icp55 substrates were aligned according to the MPP cleavage site, i.e., one residue before the mature N-terminus.
N-terminus with high frequency and a one residue larger protein with low frequency (Tables S3 and S4). This suggested the hypothesis of a two-step cleavage, first by MPP (cleavage motif R-2, leading to an intermediate) and then by a peptidase that re-moves a single amino acid residue.
To screen for such a putative peptidase, we isolated mito-chondria from mutant yeast strains of the known and predicted proteases that were found in the mitochondrial proteome (
Sick-mann et al., 2003; Reinders et al., 2006). By western blotting, we
compared the SDS-PAGE mobility of 29 mitochondrial proteins that are synthesized with presequence (and for which
high-affinity antisera were available). We observed a slight migration difference for Mge1 and Nfs1 in one of the mutant mitochondria
(Figure 4A) (mtHsp70 is shown as control). The corresponding
yeast cells lacked the uncharacterized open reading frame YER078c, now termed ICP55 for intermediate cleaving pepti-dase of 55 kDa. Sequence homology indicates that Icp55 is a member of the aminopeptidase P (APP) family of metallopro-teases (Ersahin et al., 2005; Rawlings et al., 2008). icp55D cells were defective in growth on nonfermentable medium at elevated temperature (37C) (Figure 4B, left panel) (in wild-type, the
levels of Icp55 were similar at 25 and 37C [Figure S2A]).
pI SDS-P A GE kDa 97 6 1 1 29 24 36 45 66 205 10 8 3 4 5 6 7 9 20 Atp16 Mge1 B A I H mtHsp70 yta12 Δ qri7 Δ oma1 Δ mgr1 Δ yme1 Δ pcp1 Δ cym 1Δ icp55 Δ afg3 Δ oct1 Δ prd1 Δ ste24 Δ Mge1 WT + WT -WT -WT -WT + WT + 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Nfs1 + -Δψ -+ - + - + + - + -WT icp55Δ [35S]Mge1 Prot. K i m 1 2 3 4 5 6 7 8 9 P recursor [35S]Nfs1 i m Atp16 Mge1 Nfs1 icp55 Δ WT icp55 Δ 1 2 3 m i m i m i F WT WT + Apst. icp55 Δ [35S]Atp16 m i 1 2 3 G D 5 6 Mge1 Tim23 Icp55 pH 11.5 C icp55Δ Glucose Glycerol 30 °C 37 °C 25 °C WT icp55Δ WT WT icp55Δ mas1 icp55Δ mas1 mas1 icp55Δ mas1 mas1 icp55Δ mas1 S100 P100 Total Mi to. Pgk1 Sss1 Aco1 Icp55 1 2 3 4 Glucose Glycerol E Mcr1OM p p 30°C 18°C 23°C P SN Glucose 30°C 37°C 39°C oct1Δ icp55Δ oct1Δ icp55Δ oct1Δ icp55Δ oct1Δ oct1Δ oct1Δ 6 7 Swelling Mge1 Tim23 Mcr1OM + + - - -- + + + + Icp55 5 1 2 3 4 TX-100 Zim17 Tim23 Mcr1OM Mito. Prot. K - + + + Icp55 Mito. Prot. K Mge1 Icp55 P SN 100 250 100 250 Tim44 NaCl [mM] 3 4 1 2 [35S]Icp55
Figure 4. Identification of the Intermediate Cleaving Peptidase Icp55
(A) Mitochondria were isolated from wild-type (WT) and mutant yeast lacking (predicted) proteases and subjected to SDS-PAGE and immunoblotting. WT+
, rho+
strain; WT-, rho-strain.
(B) Growth of WT, single and double mutant yeast on YPD and YPG plates.
(C) Subcellular fractionation of yeast cells by differential centrifugation. Samples were analyzed by SDS-PAGE and immunoblotting. P100, S100, pellet and
super-natant of 100,000 3 g centrifugation.
(D) Mitochondria were subjected to treatment in hypotonic buffer or Triton X-100 (TX-100), followed by proteinase K treatment, SDS-PAGE and immunoblotting.
(E) Mitochondria were incubated with [35S]Icp55 and sonicated in the presence of 100-250 mM salt (samples 1-4). Mitochondria were treated with sodium
carbonate (lanes 5 and 6). Abbreviations are as follows: P, pellet; SN, supernatant.
(F) Mitochondria were incubated with [35
S]precursors, treated with proteinase K and analyzed by SDS-PAGE and autoradiography. Abbreviations are as follows: p, precursor; i, intermediate; m, mature.
(G) 2D-DIGE analysis of WT (blue) and icp55D (green) mitochondria. (H) Isolated mitochondria were subjected to SDS-PAGE and immunoblotting.
(I) Mitochondria were incubated with 10 mM apstatin (Apst.) as indicated. [35
Re-expression of ICP55 from a plasmid restored the growth of
icp55D cells and the processing of Mge1 and Nfs1 (Figure S2B),
confirming that the lack of Icp55 was responsible for the observed defects. Deletion of ICP55 led to synthetic growth defects with mutants of MPP and Oct1, suggesting a functional relation of Icp55 to mitochondrial processing enzymes. For MPP we used the temperature-sensitive mutant mas1 (Witte et al., 1988). mas1 icp55D cells were impaired in growth on ferment-able medium at 30C and inviable on nonfermentable medium
at any temperature (Figure 4B, middle panel); oct1D icp55D cells showed a strong growth defect at elevated temperature on fermentable medium (Figure 4B, right panel) (analysis on a non-fermentable medium was not possible due to the rho-genotype of oct1D yeast).
In a fractionation of whole yeast cells, Icp55 was present in the mitochondrial fraction (Figure 4C). Icp55 remained protected against externally added protease after swelling of mitochondria and was digested only after lysis of the mitochondrial mem-branes with detergent (Figure 4D). Upon sonication of mitochon-dria in the presence of salt, Icp55 remained in the membrane fraction (Figure 4E, lanes 1 and 2). After treatment at alkaline pH, Icp55 was extracted (Figure 4E, lane 6), indicating that it was not embedded into the lipid phase of the membrane, consis-tent with the lack of prediction of hydrophobic transmembrane segments. Thus Icp55 behaves like a protein that is peripherally attached to the mitochondrial inner membrane from the matrix side.
icp55D mitochondria imported the precursors of Mge1 and
Nfs1 in a Dc-dependent manner to a protease-protected loca-tion like wild-type mitochondria with the exceploca-tion that a slight mobility difference of the processed forms was observed
(Figure 4F) in agreement with the western blot analysis of
Fig-ure 4A. In mas1 mutant mitochondria, the processing of Mge1
and Nfs1 was inhibited such that mainly the precursor form was observed (Figure S2C), indicating that MPP is the first protease cleaving the precursors.
We compared icp55D and wild-type mitochondria by the 2D-DIGE technique upon labeling with different fluorescent dyes. The proteins were separated by isoelectric focusing and SDS-PAGE. Proteins that migrated slightly more slowly on SDS-PAGE in icp55D mitochondria were subjected to mass spectrometry, leading to the identification of Mge1 and Atp16 (Figure 4G). To better visualize the small difference in gel mobility by standard western blotting, we loaded a wild-type sample between two identical icp55D samples and thus confirmed the mobility difference for all three proteins, Mge1, Nfs1 and Atp16 (Figure 4H).
The APP metalloproteases are conserved from bacteria to humans (Wilce et al., 1998; Ersahin et al., 2005; Rawlings et al., 2008). Humans contain three genes for APP homologs. APP1 and APP2 are located in the cytosol and at the plasma membrane, respectively, whereas APP3 has not been analyzed at the protein level. Two APP members are predicted from the yeast genome: ORF Yll029w (Fra1) that is closely related to APP1 and APP2; and Icp55, which shows greater similarity to APP3 and E. coli APP. Two splice variants of human APP3 mRNA are known: APP3c and APP3m, which includes a pre-dicted mitochondrial presequence (Ersahin et al., 2005). We
synthesized the APP3m protein in vitro. Upon incubation with yeast mitochondria, APP3m was imported in a Dc-dependent manner and processed (Fig. S2D), whereas APP1 and APP2 were not imported into mitochondria. APP1 and APP2 are in-hibited by the specific inhibitor apstatin (Wolfrum et al., 2001). The addition of apstatin to wild-type yeast mitochondria inhibited the processing of Atp16 to the mature form like the lack of Icp55 did (Figure 4I, lane 3 versus lane 1). Taken together, these find-ings suggest that Icp55 and APP3m are mitochondrial members of the APP family.
Substrate Specificity of Icp55
To characterize the Icp55 cleavage site and identify further substrate proteins, we analyzed the N-termini of mitochondrial proteins from icp55D versus wild-type yeast by three ap-proaches: (1) COFRADIC analysis followed by LC-MS/MS, (2) Edman sequencing, and (3) multiple-reaction monitoring (MRM). Taken together we identified 38 substrates of Icp55 and the cleavage site in each case (Table S4).
(1) The N-proteome of icp55D mitochondria determined by COFRADIC analysis yielded proteins whose N-terminus differed by one residue from the mature N-terminus of wild-type mito-chondria. In each case, this residue was tyrosine, phenylalanine or leucine (Figure 5A;Tables S4 and S5). Whereas in the icp55D mitochondria virtually only N-termini starting with one of the three amino acids were identified, wild-type mitochondria typi-cally yielded both forms, the intermediate as well as the mature form. In the majority of cases in wild-type mitochondria, the pep-tides corresponding to the smaller protein (mature N-terminus) were more frequently found than the intermediate peptides (Table S4).
(2) The cochaperone Mge1 can be specifically copurified via its ATP-sensitive interaction with matrix Hsp70. We purified Mge1 from wild-type and icp55D mitochondria and sequenced its N-terminus by Edman degradation, demonstrating that the N-terminus of icp55D mitochondria contained an additional tyro-sine (Y44) (Figures 5B andS3).
(3) Intermediate and mature N-terminal peptides of icp55D and wild-type mitochondria were also analyzed by MRM (Table S6), confirming the findings of the COFRADIC analysis that
icp55D mitochondria accumulated intermediates that are one
residue longer than the mature protein. A typical result is shown
in Figure 5C. The N-terminal peptide corresponding to the
mature form of Atp3 was found in high abundance in wild-type mitochondria and only in tiny amounts in icp55D mitochondria, whereas the intermediate peptide containing an N-terminal tyro-sine was dominant in icp55D mitochondria.
To define the cleavage site of Icp55, the 38 substrate proteins are shown as amino acid blot of relative frequency (Figure 5D). Icp55 cleaves between the last amino acid of the presequence (tyrosine, leucine or phenylalanine) and the first residue of the mature protein, preferentially serine, alanine and threonine. When the frequency of COFRADIC identification of intermediate peptides versus mature N-termini was considered (Table S4, wild-type mitochondria), the most frequent residue tyrosine was efficiently removed by Icp55 and thus was the best substrate of the new peptidase. The efficiency of cleavage was lower for leucine and phenylalanine (Table S4).
The characteristic arginine for MPP cleavage was highly abun-dant in position 3 of the presequences of Icp55 substrates and thus in position 2 of the MPP cleavage site (Figure 5D). Most importantly, when the N-proteome reference set of presequen-ces, which contain arginine residues in 2 and 3 (Figure 3, upper panel), was modified by removing the last amino acid of the presequences in case of Icp55 substrates, a high abundance of arginine at position 2 was found (Figure 3, lower panel). This explains the puzzling situation with the MPP cleavage site: R-3 is not a MPP recognition motif but indicates a second processing to remove one further amino acid residue.
A Role of Icp55 for the Stability of Mitochondrial Proteins
To assess the functional relevance of Icp55, we systematically compared the amino acid residues at the positions 1 and +1 of the Icp55 cleavage sites and noticed a striking correlation
with the N-end rule for prokaryotic proteins (Figure 6A). The N-end rule connects the N-terminal amino acid with the half-life of proteins in bacteria and the eukaryotic cytosol by differen-tiating between stabilizing and destabilizing amino acids (
Var-shavsky, 1996, 2008; Mogk et al., 2007). We found that of the
four primary destabilizing amino acid residues of prokaryotes (tyrosine, leucine, phenylalanine and tryptophane), three were recognized and cleaved by Icp55 (no other amino acid was found at position 1 in the 38 substrates). Moreover, serine, alanine and threonine, the three amino acids that form the N-terminus after Icp55 processing in nearly all substrates, are categorized as stabilizing residues (Figure 6A).
We applied the N-end rule to the large list of mature N-termini identified in the N-proteome.Figure 6B shows the frequency of amino acids at the N-terminal position of cleaved proteins (upper panel). For comparison, we determined the amino acid compo-sition of the entire mitochondrial proteome derived from 851
bits 0 1 -1 -2 -3 -5 -4 -7 -8 -9 -10 -11 -6 -12 -13 -14 -16 -15 -18 -19 -20 -17
Position (relative to mature N-termini)
+1 +2 +3 +4 +5 +6 +7 +8 C
Mge1
Y
mtHsp70 Mge1 icp55 Δ WT icp55 Δ WT icp55 Δ WT icp55 Δ WT+ATP EDTA +ATP EDTA
Mito Eluate 1 2 3 4 5 6 7 8 B A
icp55
Δ
WT
Y
icp55
Δ
WT
Atp16
F
icp55
Δ
WT
Ald4
L
icp55
Δ
WT
Aco1
D -12 000 10 000 8000 6000 4000 2000 0 -0 - - - - -20 40 60 80 100 - - - - -20 40 60 80 100 - - - - -20 40 60 80 100 - - - - -20 40 60 80 100 -8000 6000 4000 2000 0 600 400 200 -400 300 200 100 0 2 3 4 Presequence MatureFigure 5. Cleavage Sites of Icp55
(A) Examples of Icp55 substrate proteins and their N-termini (arrows) identified by COFRADIC analysis in wild-type (WT) and icp55D mitochondria (further Icp55
substrates are listed inTables S4 and S5). The amino acid labeled in red is removed by Icp55.
(B) Coimmunoprecipitation of Mge1 with antibodies directed against mtHsp70 (Ssc1) analyzed by SDS-PAGE and immunoblotting. Scheme of presequence pro-cessing of Mge1 in WT and icp55D mitochondria.
(C) Multiple Reaction Monitoring of N-terminal peptides of Atp3 in wild-type and icp55D mitochondria.
(D) Amino acid blot of relative frequency of the presequences (up to 20 residues of the C-terminal segment) and first eight mature amino acids of 38 Icp55 substrate proteins.
proteins (Reinders et al., 2006) (Figure 6B, lower panel). The stabilizing residues serine and alanine were strongly enriched at the N-terminus (38.0% compared to 14.2% in the mitochon-drial proteome), whereas the six destabilizing residues accord-ing to the prokaryotic N-end rule (Leu, Lys, Tyr, Phe, Arg,
Trp; Varshavsky, 1996, 2008) were underrepresented at the
N-terminus (19.7% versus 31.1% in the proteome). Thus, stabi-lizing amino acid residues dominate at the mature N-terminus of cleaved mitochondrial proteins.
The noncleaved mitochondrial proteins start either with the first residue (methionine) or the second residue of the predicted amino acid sequence. Methionine aminopeptidases remove the first methionine from the majority of nascent polypeptides of a cell, yet only when the second residue is a stabilizing (small) residue
(Meinnel et al., 2006; Mogk et al., 2007). A systematic analysis
of the N-proteome demonstrated that the noncleavable mito-chondrial proteins agreed well with these observations (
Fig-ure 6C). 77% of the noncleavable mitochondrial proteins lacked
the first methionine. When the methionine was maintained, the second residue was destabilizing in most cases. In contrast, when the start methionine was removed, the N-terminal residue, i.e., the second residue of the predicted sequence, was stabilizing in the vast majority of proteins. Since methionine aminopepti-dases function cotranslationally, the N-end rule of the yeast cytosol was used (Varshavsky, 1996, 2008; Meinnel et al., 2006). Taken together, stabilizing amino acid residues according to the N-end rule are enriched at the mature N-termini of mito-chondrial proteins, suggesting that the N-end rule may be valid B C A Number of prot eins 0 N-terminal residue 5 10 15 20 25 N-T yr N-L eu N-Phe icp55Δ Number of proteins 0 5 10 15 20 25 N-S er N-A la N-T hr N-P ro N-terminal residue WT
Primary destabilizing amino acid Stabilizing amino acid
Secondary destabilizing amino acid Tertiary destabilizing amino acid
Ser Ala Leu Asn Thr Gln Lys Ile Asp Phe Pro Glu Gly Tyr Val Me t Arg His Cy s Trp 0 10 5
Abundance of amino acids in mitochondrial proteome
Number of amino acids
(% of t otal) N-L eu N-Ph e N-Ala N-Gln N-Th r N-I le N-Ser N-Gly Number of proteins 0 20 40 60 N-M et N-A sn N-Tyr N-Glu N-Cy s N-His N-Tr p N-Asp N-Va l N-A rg N-L ys
Cleaved proteins of N-proteome
N-terminal residue N-P ro Aco1 Tim21 WT icp55Δ F1β Rip1 Time Substrate of Icp55 + + D Icp55 substrates
N-end rule of E.coli
N-end rule
Met-1 present
Non-cleaved proteins of N-proteome
N-end rule of yeast cytosol
Met-1 removed Number of proteins Second residue 0 2 4 6 Gly -2 Pro-2 As p-2 Glu-2 His-2 Ile-2 Leu-2 Phe-2 Tyr-2 Asn-2 Ly s-2 8 Mdh1 Pgk1 Tim21 WT icp55Δ Time Substrate of Icp55 + + Pgk1 F1β Atp3 F1β Atp3 + E Number of proteins Second residue 0 10 20 30 40 50 Ser-2 Th r-2 Ala-2 Gly-2 Leu-2 Phe-2 Val-2 Asp -2 Pro-2
Figure 6. N-Termini of Mitochondrial Proteins and Relation to N-End Rule
(A) N-terminal amino acids of 38 Icp55 substrate proteins in WT and icp55D mitochondria. Classification according to the N-end rule for E. coli (Mogk et al., 2007;
Varshavsky, 2008).
(B) N-terminal amino acids of the cleaved proteins of the N-proteome master set (upper panel). Lower panel, frequency of all amino acids in the yeast
mitochon-drial proteome (Reinders et al., 2006). Dashed horizontal line, average amino acid frequency.
(C) N-terminal amino acids of noncleavable proteins of the N-proteome master set.
(D) Mitochondria from WT and icp55D yeast were incubated at 37C (Aco1/Rip1, Atp3/F
1b) or 22C (Tim21/F1b) up to 141 hr as described inSupplemental
Exper-imental Proceduresand analyzed by SDS-PAGE and immunoblotting.
(E) In vivo degradation of proteins in WT and icp55D cells. After addition of cycloheximide, cells were incubated at 37C (Atp3/F
1b/Pgk1) or 22C (Tim21/Mdh1/
for mitochondrial proteins and thus Icp55 would be required to remove destabilizing residues. According to this hypothesis, the activity of Icp55 should lead to a stabilization of mitochon-drial proteins, i.e., should increase the half-life of its substrate proteins. For an experimental analysis, we isolated mitochondria from wild-type and icp55D cells and after incubation for different times spans, analyzed the protein levels (Figure 6D). The levels of the Icp55 substrates Aco1, Tim21 and Atp3 were decreased in a time-dependent manner in icp55D mitochondria, whereas the protein levels were stable in wild-type mitochondria. In contrast, the Rieske Fe/S protein (Rip1) and F1-ATPase subunit
b, (F1b), which are not substrates of Icp55, were of comparable
stability in wild-type and icp55D mitochondria. In addition, we analyzed protein stability in vivo after blocking cytosolic protein synthesis with cycloheximide. The levels of Icp55 substrate proteins decreased in a time-dependent manner in icp55D cells, while nonsubstrate proteins like Mdh1 and F1bremained stable
(Figure 6E). As a loading control we used the cytosolic protein
Pgk1. These results suggest that the cleavage by Icp55 stabi-lizes the substrate proteins by removing destabilizing amino acids.
DISCUSSION
We report a comprehensive approach for the identification of the protein N-termini of an entire cell organelle. The proteomic anal-ysis of yeast mitochondria yielded up to 851 different proteins (PROMITO;Reinders et al., 2006). Here we have identified the N-termini of 615 proteins, covering 72% of the PROMITO pro-teins. The mitochondrial N-proteome represents a rich source for the characterization of the N-terminal targeting signals and their cleavage sites. Mitochondria import precursor proteins by at least four different pathways, the presequence pathway and three pathways for noncleavable proteins (carrier, MIA and b-barrel proteins) (Dolezal et al., 2006; Neupert and Herrmann,
2007; Bolender et al., 2008). Application of presequence
predic-tion programs on the mitochondrial proteome suggested that only43%–50% of the known proteins contain a presequence, suggesting that noncleavable precursor proteins are at least as abundant as cleavable precursors (Sickmann et al., 2003;
Pro-kisch et al., 2006; Habib et al., 2007). The experimental
determi-nation of cleavable N-termini in the N-proteome, however, indi-cated that up to 70% of mitochondrial proteins are synthesized with a presequence. We conclude that presequences are signif-icantly more abundant than previously assessed and thus the presequence pathway is by far the major pathway of mitochon-drial protein import.
Based on the observation that a number of proteins of internal mitochondrial compartments yielded two N-termini that differed by one amino acid residue, we identified the intermediate cleaving peptidase Icp55 that removes one residue from the pre-sequence after the cleavage by the major processing peptidase MPP. The identification of Icp55 solves the long-standing problem that the two most frequently observed MPP cleavage site motifs (R-2 and R-3) were not readily compatible with a specific recognition and cleavage mechanism of MPP (Branda and Isaya, 1995; Gakh et al., 2002; Emanuelsson et al., 2007;
Habib et al., 2007). In particular, the R-3 motif does not fit to
the high-resolution structure of MPP that revealed an arginine at 2 of the cleavage site (Taylor et al., 2001). However,Gakh
et al. (2002)pointed out that the two presequences
cocrystal-lized with MPP were Oct1 substrates and may thus represent a special situation. Different explanations were discussed, from a high degree of amino acid sequence degeneracy at the MPP cleavage site (Gakh et al., 2002; Habib et al., 2007; Emanuelsson
et al., 2007) to the existence of an unknown proteolytic activity
that removes one residue in the case of R-3 (Schneider et al., 1998), yet no experimental answer was found. The N-proteome revealed that the vast majority of Icp55 substrates contain an arginine residue at R-3. The action of Icp55 converts them into R-2 substrates that fit perfectly into the mechanism of prese-quence binding and cleavage derived from the high resolution structures of MPP-presequence complexes (Taylor et al., 2001). We conclude that R-3 does not represent a MPP-cleavage site motif but is part of a motif for two-step MPP-cleavage, first by MPP and then by Icp55.
The identification of Icp55 revealed a new function of the aminopeptidase P family (Ersahin et al., 2005; Rawlings et al., 2008). Of the three human APP members, only the function of APP2 has been defined so far. APP2 is located at the plasma membrane of vascular endothelial cells and involved in inactiva-tion of bioactive peptides like bradykinin (Wolfrum et al., 2001;
Ersahin et al., 2005). Icp55 is related to human APP3 and
bacte-rial APP with so far unknown biological function. We observed that the removal of a single amino acid residue by Icp55 led to a stabilization of the substrates—in contrast to the inactivating function of APP2. The characterization of Icp55 substrates suggests the hypothesis that the N-end rule for protein degrada-tion is also applicable to mitochondria. The N-end rule connects the N-terminal amino acid residue with the half-life of proteins and was shown to exist in bacteria and the cytosol of eukaryotes
(Varshavsky, 1996, 2008; Mogk et al., 2007). Though it has been
speculated that a similar rule may exist in organelles, no exper-imental evidence has been reported so far (Meinnel et al., 2006). We identified 38 substrates of Icp55, each containing a ‘primary destabilizing residue’ (prokaryotic N-end rule) after the initial cleavage by MPP. Icp55, which is associated with the mitochondrial inner membrane at the matrix side, removes the ‘destabilizing residue’ and generates an N-terminus with a ‘stabilizing residue’. Icp55 substrate proteins were indeed much faster degraded in icp55D mitochondria than in wild-type mitochondria.
Noncleavable proteins of the cytosol and mitochondria typi-cally contain N-termini with stabilizing amino acids according to the N-end rule since methionine aminopeptidases remove the first methionine from nascent polypeptides only when the second residue is a stabilizing one (Meinnel et al., 2006; Mogk
et al., 2007). The removal of targeting sequences by organellar
processing enzymes like mitochondrial MPP, however, raises the problem that various new N-termini are generated, including also destabilizing residues. We propose the hypothesis that Icp55 and possibly further processing enzymes of mitochondria have the function to convert MPP-generated destabilizing N-termini into stable ones.
The APP cleavage motif reported for bacteria and humans is X-Pro with X being the N-terminal residue of peptides/polypeptides
(Wilce et al., 1998; Ersahin et al., 2005). APP thus removes the terminal residue before a proline. We observed that Icp55 has a much broader substrate specificity. Icp55 can remove the terminal residue in front of a proline, however, the majority of substrates contained serine, alanine or threonine instead of proline (Figure 5A). We speculate that Icp55 needs a broader substrate specificity in order to perform its stabilizing function for the mitochondrial proteome by processing different unstable intermediates generated by MPP. Interestingly,Graham et al.
(2004) solved the high resolution structure of E. coli APP in
complex with apstatin and reported that the structure could not explain the proline-specificity (amino acids other than proline would also be compatible in the active center). The analysis of Icp55 revealed that at least this APP family member recognizes a broader range of substrates.
Possible limitations of the N-proteome would be a cleavage by unspecific proteases during mitochondrial isolation or an incom-plete labeling of internal peptides with TNBS. We thus performed a number of analyses to assess the validity of the N-proteome. (1) The noncleavable Tom proteins of the outer membrane were used as sensitive markers for unspecific proteolysis or incom-plete TNBS-labeling. In each case, the most frequent N-peptide corresponded to the noncleaved, authentic N-terminus. (2) A comparison of the entire N-proteome with the 50 proteins, whose N-termini had been previously determined by single protein studies, revealed an agreement of 94%. For several proteins we tested the N-proteome findings by (3) in vitro pro-cessing, (4) Edman degradation or (5) multiple reaction moni-toring and observed a complete agreement with the COFRADIC results. (6) For several Icp55 substrates, more than one N-terminus was reproducibly found and identified to belong to intermediate and mature protein, respectively. Thus the multiple N-termini were not caused by an unspecific cleavage but indi-cated a high sensitivity of the approach in identifying processing intermediates. (7) The complete separation between ‘destabiliz-ing’ and ‘stabilizing residues’ according to the prokaryotic N-end rule before and after cleavage by Icp55 for all 38 substrates iden-tified also supports a high specificity of the N-terminal determi-nation and argues against unspecific cleavage. In summary, we conclude that the vast majority of deduced presequences and N-termini of the mitochondrial N-proteome are authentic.
The identification of Icp55 and the clarification of the contro-versial situation with the MPP cleavage site illustrate that the systematic analysis of the mitochondrial N-proteome opens many possibilities for future experimental and bioinformatics studies. The N-proteome strongly broadens the basis for the development and training of presequence and cleavage site prediction programs. A number of mitochondrial presequences contain an arginine residue at position 18 or 11 (Figure 3). This raises the possibility of a three-step cleavage MPP-Oct1-Oct1 or MPP-Icp55/MPP-Oct1-Oct1. The N-proteome provides the basis to systematically test the hypothesis of a mitochondrial N-end rule, not only with artificial substrates but also with a large number of physiological substrates (by comparing proteins before and after cleavage by Icp55). It will be interesting to address if the two splice variants of human APP3, APP3m (mito-chondrial) and APP3c (likely cytosolic), function in stabilization of proteins in mitochondria and the cytosol, respectively. It is also
conceivable that mitochondria contain further processing enzymes. A systematic N-proteome analysis of mutant mito-chondria lacking the various proteases found in the mitochon-drial proteome, as performed here for icp55D mitochondria, may be a promising approach. Whereas mitochondria lack a ubiquitin-proteasome system, homologs of bacterial prote-ases are promising candidates for identification of the protease system that functions in a mitochondrial N-end rule pathway: ClpP found in mitochondria of higher eukaryotes and Pim1 (Lon) in yeast mitochondria (that lack a ClpP homolog) (Meinnel
et al., 2006; Mogk et al., 2007). Finally, our findings raise the
possibility that stabilizing intermediate peptidases may also be relevant for other organelles with cleavable preproteins such as chloroplasts and endoplasmic reticulum. The generation of a comprehensive N-proteome of these cell organelles will be an important task.
EXPERIMENTAL PROCEDURES
Purification of Yeast Mitochondria and Protein Import
Mitochondria were isolated from S. cerevisiae. Crude mitochondrial fractions were obtained by differential centrifugation of yeast homogenate and purified
by sucrose gradient centrifugation (Sickmann et al., 2003). Highly pure
mito-chondria were stored in SEM buffer (250 mM sucrose, 1 mM EDTA, 10 mM
MOPS/KOH, [pH 7.2]). 35
S-labeled precursor proteins were generated by in vitro transcription/translation and incubated with mitochondria in import
buffer (3% [w/v] bovine serum albumin, 250 mM sucrose, 5 mM MgCl2,
80 mM KCl, 5 mM KPi, 10 mM MOPS/KOH, [pH 7.2]). After treatment with
proteinase K (50 mg/mL), mitochondria were reisolated and analyzed by SDS-PAGE and digital autoradiography. As control, Dc was dissipated prior to the import reaction by addition of valinomycin, antimycin A and oligomycin.
Fractionation of Mitochondria and Analysis of Protein Turnover Isolated mitochondria were sonicated in SEM buffer containing 100–250 mM NaCl. For alkaline treatment, mitochondria were incubated in 100 mM sodium carbonate, followed by separation of pellet and supernatant by ultracentrifuga-tion. Swelling of mitochondria was performed in 31 mM sucrose, 1 mM EDTA, 10 mM MOPS/KOH, (pH 7.2).
For coimmunoprecipitation of Mge1, mitochondria were depleted of ATP by incubation with apyrase and oligomycin, followed by lysis with Triton X-100. The lysates were incubated with mtHsp70-Protein A Sepharose and complexes were eluted with glycine, (pH 2.5).
The turnover of proteins in isolated mitochondria was analyzed by
incuba-tion of mitochondria in SEM buffer at 37or 22C, followed by SDS-PAGE
and immunoblotting. Protein turnover in vivo was analyzed by adding
cyclo-heximide to yeast cells growing at 37or 22C, followed by incubation for
different time spans, preparation of yeast cell extracts, SDS-PAGE and immu-noblotting.
Combined Fractional Diagonal Chromatography and Mass Spectrometry
For separation of N-terminal peptides by COFRADIC, highly pure mitochondria were lysed, disulfide bonds were reduced and alkylated prior to acetylation of N-termini with trideutero-acetyl N-Hydroxy-Succinimide. Samples were
di-gested with trypsin, V8 protease or chymotrypsin overnight at 25C or 37C,
respectively. Purified dried peptides were reconstituted in 0.1% trifluoroacetic acid (TFA) and separated on an Ultimate 3000 LC system. During the primary run, 16 fractions of 4 min each were collected and dried under vacuum.
N-termini of internal peptides were derivatized with TNBS (Gevaert et al.,
2003). Afterwards, fractions were applied to a secondary RP-HPLC run with
identical chromatographic conditions. Fractions were collected in the same time intervals as before, dried under vacuum and prepared for LC-MS/MS analysis.
The LC-MS/MS analysis yielded 665,571 spectra. For interpretation of the MS/MS data, a peptide database of the mitochondrial proteome including N-terminally truncated versions was generated, including reverse sequences
for estimation of false discovery rates. SeeSupplemental Experimental
Proce-duresfor more details. SUPPLEMENTAL DATA
Supplemental Data include Supplemental Experimental Procedures, Supple-mental References, three figures, and six tables and can be found with this
article online athttp://www.cell.com/supplemental/S0092-8674(09)01032-0.
ACKNOWLEDGMENTS
We thank Drs. M. Yaffe, T. Lithgow, J. Rassow, and P. Rehling for strains and antisera; Dr. N. Wiedemann for discussion; and C. Prinz and B. Scho¨nfisch for expert technical assistance. This work was supported by the Deutsche For-schungsgemeinschaft, Sonderforschungsbereich 746, Excellence Initiative of the German Federal and State Governments (EXC 294), Gottfried Wilhelm Leibniz Program, Trinational Research Training Group (GRK 1478, fellowship to F.N.V.), the Ministerium fu¨r Innovation, Wissenschaft, Forschung und Tech-nologie des Landes Nordrhein-Westfalen, the Bundesministerium fu¨r Bildung und Forschung, the Fonds der Chemischen Industrie, the Fund for Scientific Research–Flanders (Belgium), the Concerted Research Actions from Ghent University, the Inter University Attraction Poles (IUAP06) and the European
Union Interaction Proteome (6thFramework Program).
Received: March 31, 2009 Revised: June 14, 2009 Accepted: July 24, 2009 Published: October 15, 2009 REFERENCES
Bolender, N., Sickmann, A., Wagner, R., Meisinger, C., and Pfanner, N. (2008). Multiple pathways for sorting mitochondrial precursor proteins. EMBO Rep. 9, 42–49.
Branda, S.S., and Isaya, G. (1995). Prediction and identification of new natural substrates of the yeast mitochondrial intermediate peptidase. J. Biol. Chem. 270, 27366–27373.
Claros, M.G., and Vincens, P. (1996). Computational method to predict mito-chondrially imported proteins and their targeting sequences. Eur. J. Biochem. 241, 779–786.
Dolezal, P., Likic, V., Tachezy, J., and Lithgow, T. (2006). Evolution of the molecular machines for protein import into mitochondria. Science 313, 314–318.
Emanuelsson, O., von Heijne, G., and Schneider, G. (2001). Analysis and prediction of mitochondrial targeting peptides. Methods Cell Biol. 65, 175–187.
Emanuelsson, O., Brunak, S., von Heijne, G., and Nielsen, H. (2007). Locating proteins in the cell using TargetP, SignalP and related tools. Nat. Protocols 2, 953–971.
Ersahin, C., Szpaderska, A.M., Orawski, A.T., and Simmons, W.H. (2005). Aminopeptidase P isozyme expression in human tissues and peripheral blood mononuclear cell fractions. Arch. Biochem. Biophys. 435, 303–310. Esser, K., Tursun, B., Ingenhoven, M., Michaelis, G., and Pratje, E. (2002). A novel two-step mechanism for removal of a mitochondrial signal sequence involves the mAAA complex and the putative rhomboid protease Pcp1. J. Mol. Biol. 323, 835–843.
Gakh, O., Cavadini, P., and Isaya, G. (2002). Mitochondrial processing pepti-dases. Biochim. Biophys. Acta 1592, 63–77.
Gevaert, K., Goethals, M., Martens, L., Van Damme, J., Staes, A., Thomas, G.R., and Vandekerckhove, J. (2003). Exploring proteomes and analyzing protein processing by mass spectrometric identification of sorted N-terminal peptides. Nat. Biotechnol. 21, 566–569.
Graham, S.C., Maher, M.J., Simmons, W.H., Freeman, H.C., and Guss, J.M. (2004). Structure of Escherichia coli aminopeptidase P in complex with the inhibitor apstatin. Acta Crystallogr. D Biol. Crystallogr. 60, 1770–1779. Habib, S.J., Neupert, W., and Rapaport, D. (2007). Analysis and prediction of mitochondrial targeting signals. Methods Cell Biol. 80, 761–781.
Meinnel, T., Serero, A., and Giglione, C. (2006). Impact of the N-terminal amino acid on targeted protein degradation. Biol. Chem. 387, 839–851.
Mogk, A., Schmidt, R., and Bukau, B. (2007). The N-end rule pathway for regu-lated proteolysis: prokaryotic and eukaryotic strategies. Trends Cell Biol. 17, 165–172.
Neupert, W., and Herrmann, J.M. (2007). Translocation of proteins into mito-chondria. Annu. Rev. Biochem. 76, 723–749.
Nolden, M., Ehses, S., Koppen, M., Bernacchia, A., Rugarli, E.I., and Langer, T. (2005). The m-AAA protease defective in hereditary spastic paraplegia controls ribosome assembly in mitochondria. Cell 123, 277–289.
Pagliarini, D.J., Calvo, S.E., Chang, B., Sheth, S.A., Vafai, S.B., Ong, S.E., Wal-ford, G.A., Sugiana, C., Boneh, A., Chen, W.K., et al. (2008). A mitochondrial protein compendium elucidates complex I disease biology. Cell 134, 112–123. Prokisch, H., Scharfe, C., Camp, D.G., 2nd, Xiao, W., David, L., Andreoli, C., Monroe, M.E., Moore, R.J., Gritsenko, M.A., Kozany, C., et al. (2004). Integra-tive analysis of the mitochondrial proteome in yeast. PLoS Biol. 2, 160. Prokisch, H., Andreoli, C., Ahting, U., Heiss, K., Ruepp, A., Scharfe, C., and Meitinger, T. (2006). MitoP2: the mitochondrial proteome database–now including mouse data. Nucleic Acids Res. 34, 705–711.
Rawlings, N.D., Morton, F.R., Kok, C.Y., Kong, J., and Barrett, A.J. (2008). MEROPS: the peptidase database. Nucleic Acids Res. 36, 320–325. Reinders, J., Zahedi, R.P., Pfanner, N., Meisinger, C., and Sickmann, A. (2006). Toward the complete yeast mitochondrial proteome: multidimensional separation techniques for mitochondrial proteomics. J. Proteome Res. 5, 1543–1554.
Schneider, G., Sjo¨ling, 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.
Sickmann, A., Reinders, J., Wagner, Y., Joppich, C., Zahedi, R., Meyer, H.E., Scho¨nfisch, B., Perschil, I., Chacinska, A., Guiard, B., et al. (2003). The pro-teome of Saccharomyces cerevisiae mitochondria. Proc. Natl. Acad. Sci. USA 100, 13207–13212.
Taylor, A.B., Smith, B.S., Kitada, S., Kojima, K., Miyaura, H., Otwinowski, Z., Ito, A., and Deisenhofer, J. (2001). Crystal structures of mitochondrial process-ing peptidase reveal the mode for specific cleavage of import signal sequences. Structure 9, 615–625.
Varshavsky, A. (1996). The N-end rule: functions, mysteries, uses. Proc. Natl. Acad. Sci. USA 93, 12142–12149.
Varshavsky, A. (2008). The N-end rule at atomic resolution. Nat. Struct. Mol. Biol. 15, 1238–1240.
Wilce, M.C., Bond, C.S., Dixon, N.E., Freeman, H.C., Guss, J.M., Lilley, P.E., and Wilce, J.A. (1998). Structure and mechanism of a proline-specific amino-peptidase from Escherichia coli. Proc. Natl. Acad. Sci. USA 95, 3472–3477. Witte, C., Jensen, R.E., Yaffe, M.P., and Schatz, G. (1988). MAS1, a gene essential for yeast mitochondrial assembly, encodes a subunit of the mito-chondrial processing protease. EMBO J. 7, 1439–1447.
Wolfrum, S., Richardt, G., Dominiak, P., Katus, H.A., and Dendorfer, A. (2001). Apstatin, a selective inhibitor of aminopeptidase P, reduces myocardial infarct size by a kinin-dependent pathway. Br. J. Pharmacol. 134, 370–374.
