INTERMEDIATE CLEAVAGE PEPTIDASE55 Modifies Enzyme Amino Termini and Alters Protein Stability in Arabidopsis Mitochondria^,

A mitochondrial protease alters the stability of proteins in Arabidopsis by removal of a single amino acid from their sequence.

Abstract

Precursor proteins containing mitochondrial peptide signals are cleaved after import by a mitochondrial processing peptidase. In yeast (Saccharomyces cerevisiae) and human (Homo sapiens), INTERMEDIATE CLEAVAGE PEPTIDASE55 (ICP55) plays a role in stabilizing mitochondrial proteins by the removal of single amino acids from mitochondrial processing peptidase-processed proteins. We have investigated the role of a metallopeptidase (At1g09300) from Arabidopsis (Arabidopsis thaliana) that has sequence similarity to yeast ICP55. We identified this protein in mitochondria by mass spectrometry and have studied its function in a transfer DNA insertion line (icp55). Monitoring of amino-terminal peptides showed that Arabidopsis ICP55 was responsible for the removal of single amino acids, and its action explained the −3 arginine processing motif of a number of mitochondrial proteins. ICP55 also removed single amino acids from mitochondrial proteins known to be cleaved at nonconserved arginine sites, a subset of mitochondrial proteins specific to plants. Faster mitochondrial protein degradation rates not only for ICP55 cleaved protein but also for some non-ICP55 cleaved proteins were observed in Arabidopsis mitochondrial samples isolated from icp55 than from the wild type, indicating that a complicated protease degradation network has been affected. The lower protein stability of isolated mitochondria and the lack of processing of target proteins in icp55 were complemented by transformation with the full-length ICP55. Analysis of in vitro degradation rates and protein turnover rates in vivo of specific proteins indicated that serine hydroxymethyltransferase was affected in icp55. The maturation of serine hydroxymethyltransferase by ICP55 is unusual, as it involves breaking an amino-terminal diserine that is not known as an ICP55 substrate in other organisms and that is typically considered a sequence that stabilizes rather than destabilizes a protein.

Plant mitochondria provide energy production through respiration. Most mitochondrial proteins responsible for the machinery of respiration and metabolism are synthesized in the cytosol and imported into mitochondria. After import, N-terminal presequences containing targeting signals are cleaved from many proteins by the mitochondrial processing peptidase (MPP; Sjoling and Glaser, 1998; Zhang and Glaser, 2002), and the mitochondrial presequences themselves are then degraded by presequence peptidases (Ståhl et al., 2002; Moberg et al., 2003; Bhushan et al., 2005) and oligopeptidase (Kmiec et al., 2013). The specificity of the MPP cutting sites has been analyzed by comparison of the experimentally determined N-terminal sequences of mature proteins with the amino acid sequences of the precursor proteins. From this analysis, the most frequently observed MPP cleavage sites are referred to as −2 from an Arg (the −2R cleavage group) and −3 from an Arg (the −3R cleavage group) within the presequence (Zhang et al., 2001; Huang et al., 2009). However, despite the clear presence of both groups in experimental data, the −3R motif did not fit the high-resolution structure of MPP, which revealed −2R as the only probable cleavage site (Taylor et al., 2001). In yeast (Saccharomyces cerevisiae), there are also a group of mitochondrial proteins with a −10R motif that are now known to be first cleaved by MPP and then by OCTAPEPTIDYL AMINOPEPTIDASE1 (Oct1; Isaya et al., 1991; Vögtle et al., 2011). In plants, a group of mitochondrial proteins with nonconserved Arg cleavage sites has been reported (Huang et al., 2009). For many years, it has remained unclear why mitochondrial proteins from plants and yeast differed in the cleavage motif and what other proteases might be involved in these processes in plants.

The identification of INTERMEDIATE CLEAVAGE PEPTIDASE55 (P40005.1) in yeast mitochondria solved the long-standing problem of apparent −2R and −3R cleavage sites by MPP (Vögtle et al., 2009). In yeast, ICP55 removes one residue from the mature protein after cleavage by MPP, leading to the Arg residue in the presequence being −3 from the position of the final N terminus of the mature protein. Therefore, the −3R group of proteins undergo a two-step cleavage, first by MPP and then a single amino acid removal by ICP55. In yeast, ICP55 cleaves exclusively after a Tyr, Leu, or Phe, leaving the first residue of the mature protein to typically be Ser, Ala, or Thr (Vögtle et al., 2009). Our analysis of plant mitochondrial presequence cleavage motifs indicated that the major −3R group (55%–58%) in Arabidopsis (Arabidopsis thaliana) and rice (Oryza sativa; Huang et al., 2009) had a very similar motif to that cleaved by ICP55 in yeast (Vögtle et al., 2009). ICP55 belongs to the M24 metallopeptidase peptidase family and is critical for mitochondrial protein stability in yeast (Vögtle et al., 2009). The stability, abundance, and turnover of mitochondrial proteins in yeast are also determined by other mitochondrial proteases (Brandner et al., 2005). Plant mitochondria also contain many other proteases, such as filamentation temperature sensitive H protein, long undivided filaments protein, caseinolytic protease, and degradation of periplasmic protein classes (Gibala et al., 2009; Rigas et al., 2009; Kmiec et al., 2012; Kwasniak et al., 2012; Solheim et al., 2012). Metallopeptidase and other protease classes could be part of a complex network controlling protein stability and, thus, the rate of protein turnover in plant mitochondria (van Wijk, 2015).

We have compiled lists of Arabidopsis mitochondrial proteins using organelle isolation, fractionation, and proteomic analysis (Heazlewood et al., 2004; Taylor et al., 2011). Our in-depth analysis of mitochondrial matrix proteins has identified a protein with unknown function encoded by the gene At1g09300. This protein has some similarity to yeast ICP55 (P40051.1) and has been suggested to be an ICP55-like protein in plants based on sequence comparisons (Kwasniak et al., 2012). In this study, we have examined the role of this ICP55-like protein (At1g09300) in the cleavage of plant mitochondrial proteins using peptide mass spectrometry (MS) to compare the wild type with a transfer DNA (T-DNA) insertion line, icp55. The plant mitochondrial ICP55-like protein is not only responsible for −3R group protein cleavage, as observed in yeast, but also the cleavage of non-R group proteins that are only found in plant mitochondria, to our knowledge. The lack of ICP55 alters mitochondrial protein stability, as indicated by an analysis of protein degradation rates in isolated mitochondria. We also show that serine hydromethyltransferase (SHMT) is processed by ICP55, a degradation product of SHMT is stabilized in vitro in the absence of ICP55, and SHMT turns over more rapidly in vivo in icp55. The maturation of SHMT by cleavage of a diserine represents a new substrate class for ICP55 and appears to destabilize rather than stabilize this enzyme.

RESULTS

Subcellular Localization Sequence Similarity and Expression of At1g09300

Analysis of Arabidopsis cell culture mitochondria matrix samples identified nine peptides matched to the sequence of At1g09300 (Fig. 1). The nine peptides cover 31.2% of the amino acid sequence of At1g09300 and identified this protein with high confidence. At1g09300 is a member of the metallopeptidase M24 family and has two splice variants (At1g09300.1 and At1g09300.2; http://www.arabidopis.org). Full-length complementary DNAs (cDNAs) for At1g09300.1 (AK117456 and BX814500) but not Atg09300.2 have been reported, but only At1g09300.1 encodes a protein predicted to be located in mitochondria (http://suba.plantenergy.uwa.edu.au/; Tanz et al., 2013). Based on the AtGenExpress Visual database and Genevestigator (www.genevestigator.com), At1g09300.1 is relatively evenly expressed in all tissues, but its expression is induced by light treatment. However, the AtProteome database of plant tissues (fgcz-atproteome.unizh.ch) showed that At1g09300.1 was predominantly found in rapidly proliferating cell culture tissues like the ones used for our identification in mitochondrial samples, while only a single peptide was found in the flower, carpel, root, and seeds of whole plants (Fig. 1B). BLASTP of the At1g09300.1 protein sequence showed a 35% sequence identity to yeast ICP55 (P40051.1), with an E-value of 7E-88 (Fig. 1C). Both proteins contain the PF00557 domain for M24 metallopeptidases and conserved metal-binding Asp, His, and Glu residues defined for this family (Fig. 1D). The yeast ICP55 removes one residue from mature proteins after cleavage by MPP, leaving an apparent cutting motif of −3R from the start of the mature protein sequence (Vögtle et al., 2009). As Arabidopsis mitochondria contain a similar group of cleaved proteins with an apparent −3R cutting motif (Huang et al., 2009), it was hypothesized that At1g09300 may have a similar function to yeast ICP55 (Kwasniak et al., 2012), so we named this protein Arabidopsis ICP55.

Figure 1.

Identification, expression, and sequence similarity of At1g09300. A, At1g09300 peptides in isolated Arabidopsis mitochondria samples identified by peptide MS. Experimental and theoretical peptide mass and Mascot ion scores of identified peptides are shown. m/z, Mass-to-charge ratio. B, At1g09300 peptides in the published AtProteome study of different tissue types (http://fgcz-atproteome.unizh.ch/). C, Sequence similarity of At1g09300 to yeast ICP55. D, Sequence alignment and conservation of metal-binding sites in metallopeptidase M24 domains in At1g09300 and yeast P40051.

Characterization of icp55 T-DNA Insertion Lines with PCR, Quantitative Reverse Transcription-PCR, and Selected Reaction Monitoring Showed an Absence of ICP55 in One Insert Line

To experimentally investigate the role of ICP55 in Arabidopsis, we obtained four independent T-DNA insertion lines, Salk_123637, Salk_123635, Sail_672_D05, and Salk_001742, in At1g09300 (Fig. 2; primers used for genotyping are shown in Supplemental Figs. S1 and S2). We could not identify any T-DNA sequence in Salk_123637, Salk_123635, and Salk_001742 lines using PCR amplification (Supplemental Fig. S2). We could identify a homozygous T-DNA insertion in Sail_672_D05 using PCR amplification. The T-DNA insertion in Sail_672_D05 was located in the last intron (Fig. 2A), and we named this plant line icp55. To analyze the ICP55 expression level in the wild type and icp55, we conducted qRT-PCR based on specially designed primers (Fig. 2A; Supplemental Fig. S1). Primer pairs were located upstream of the T-DNA insertion and also across the T-DNA insertion (Fig. 2A). In icp55, the detected transcript levels upstream of the T-DNA insertion were approximately 50% of the wild type, suggesting the expression of ICP55. However, the transcript levels across the T-DNA were less than 1% of those in the wild type (Fig. 2B), indicating that T-DNA insertion disrupted expression of the full-length sequence. The transcript levels for three housekeeping genes (Protein Phosphatase 2A, At1g13320; Yellow Leaf-Specific Protein8, At5g08290; and Clathrin, At5g46630) in the wild type and icp55 were very similar (Fig. 2B). The detection of transcript from the region upstream of the insertion region, and the fact that the insert was in the C-terminal region of the protein, raised the possibility of a truncated ICP55 protein in icp55. To detect relative ICP55 abundance, we attempted polyclonal antibody production and MS-based selected reaction monitoring assays. Only the selected reaction monitoring assays for two unique peptides of ICP55 (LVELLPENSLAIISSAPVK [peptide 1] and YTNLDDFQNSASLGK [peptide 2]) that are located upstream of the T-DNA were successful (Fig. 2A). Peptides from mitochondrial aconitase were used as a control for mitochondrial protein (Taylor et al., 2014). Quantification of ICP55 peptides in trypsin-digested whole mitochondria showed that both peptides 1 and 2 were not detectable in icp55 when compared with the wild type (Fig. 2C), while similar abundances of aconitase were observed in icp55 and the wild type (Fig. 2C).

Figure 2.

Characterization of T-DNA insertion lines with analysis of gene expression and protein abundance. A, Locations of T-DNA insertions in At1g09300 and primers used for icp55 gene expression using quantitative reverse transcription (qRT)-PCR. The white boxes indicate no T-DNA amplification, and the gray box (Sail) indicates with T-DNA amplification. The primer pairs F12/R1 and F12/R2 are located upstream of the T-DNA insertion of the Sail line, and primers F3/R3 and F4/R4 cross the T-DNA insertion of the Sail line. The Sail T-DNA disruption at the protein sequence is indicated with a thick arrow. The positions of peptides used for ICP abundance are indicated with black boxes: peptide 1, LVELLPENSLAIISSAPVK (64–82); and peptide 2, YTNLDDFQNSASLGK (183–197). AA, Amino acids. B, Expression level of icp55 in the wild type (WT; black bars) and the T-DNA knockout line (white bars). The positions of primer pairs F12/R1, F12/R2, F3/R3, and F4/R4 are indicated in A. PPase, Yls8, and Clathrin are three housekeeping genes. C, Peptide levels detected by selected reaction monitoring of icp55 in the wild type (black bars) and the T-DNA knockout line (white bar). Peptides from the tricarboxylic acid cycle enzyme aconitase were used as a control (means ± se; n = 3).

ICP55 Has a Similar Whole-Plant Phenotype But Different Protein Cleavage Sites in Mitochondrial Proteins When Compared with the Wild Type

To test for any phenotypic differences between icp55 and the wild type, we grew plants in soil and on agar plates under long-day and short-day conditions. We did not find any gross difference between wild-type and icp55 seedlings (Fig. 3). In yeast, the deleted line icp55 was unable to grow in elevated temperature (37°C) using glycerol as a substrate (Vögtle et al., 2009). Microarray analysis indicated that Arabidopsis ICP55 expression was induced in response to light or high-light treatment in response to abiotic and biotic stresses (www.genevestigator.com; At1g09300; array 263707_et). To test whether heat stress or high light affected plant growth of icp55, we treated plants grown on agar plates or soil with high temperature (37°C and 40°C) or a combination of high temperature (37°C) and high light (700 µmol m⁻² s⁻¹). However, we did not find any gross difference in the appearance of wild-type and icp55 plants following the treatments (Supplemental Fig. S3).

Figure 3.

Plant growth and mitochondrial protein cleavage sites in icp55. A, Plant growth of icp55 (Sail_672_D05) in soil under long-day (left) and short-day (middle) conditions and on plates under long-day (right) conditions. B, Coomassie Blue-stained gel of wild-type (WT) and icp55 mitochondrial proteins and cleavage sites of mitochondrial proteins in wild-type and icp55 protein bands based on MS analysis (Supplemental Data Set S1). Cleavage sites are shown with black arrowheads in the wild type and with red arrowheads in icp55. C, Comparison of observed cutting sites with predicted ICP55 cutting sites using MitoFates (mitf.cbrc.jp/MitoFates). n.d, Not detected; n.p, not predicted.

To determine if ICP55 influenced the cleavage site of mitochondrial proteins, we isolated mitochondria from 3-week-old shoots of hydroponically cultured wild-type and icp55 plants. We separated mitochondrial proteins using one-dimensional (1D) SDS-PAGE, stained with Coomassie Blue, and the wild type and icp55 showed similar patterns (Fig. 3B). We then cut protein bands from stained gel bands in regions where we knew from previous work that non-R and −3R group proteins were located. These protein bands were digested with trypsin and peptides identified using quadrupole time of flight (Q-TOF)-MS. In the four bands indicated (Fig. 3B), we detected N-terminal semitryptic peptides of mitochondrial proteins using our in-house Arabidopsis mitochondrial protein database (Huang et al., 2009).

In band 1 of the wild type, we detected the semitryptic peptide (SLPSEAVDEK) from the SHMT (At4g37930) sequence (Fig. 3B). The tandem mass spectrometry (MS/MS) spectra for semitryptic peptides are presented in Supplemental Data Set S1. Based on this sequence of the N terminus of the mature protein, the predicted presequence of SHMT (At4g37930) has a length of 29 amino acids and pI at 11.2, which are consistent with our previous characteristics of Arabidopsis mitochondrial presequences (Huang et al., 2009). The cleavage site according to the mature protein was eight amino acids away from the last Arg in the presequence (Fig. 3B). This showed that SHMT has a nonconserved Arg cleavage motif. In samples from the icp55 line in the same position on the gel, we could not detect this peptide, but we detected a new semitryptic peptide for SHMT that is one more amino acid in length (SSLPSEAVDEK; Fig. 3B), indicating that ICP55 is needed for cutting between the Ser residues (S↓S) in this protein sequence. It should be noted that cleavage of a Ser is very unusual, because Ser is normally a protein-stabilizing N-terminal amino acid (Varshavsky, 2008). In band 2, we found the semitryptic N-terminal peptide of ATP synthase β-chain1 (At5g08670 and At5g08690), which is another protein with a non-R conserved cutting site (Fig. 3B). In icp55, we could not detect the wild-type peptide but instead detected a semitryptic N-terminal peptide with the Tyr retained (Fig. 3B). The detection of differential cleavage of non-R group proteins, such as SHMT (At4g37930) and ATP synthase β-chain1 (At5g08690 and At5g08670), highlight that AtICP55 has a role in the cleavage of non-R group proteins.

In bands 3 and 4, we detected an ATP synthase Δ-chain protein (At5g13450) and two glycine decarboxylase (GDC) H proteins (At2g35370 and At1g32470), which are −3R group proteins in the wild type. Each had one more amino acid in icp55 (RT↓Y↓A in At5g13450; RC↓F↓S in At2g35370 and At1g32470). These cleavage sites and the amino acids cleaved are entirely consistent with the reported properties of ICP55 in yeast (Vögtle et al., 2009). We did observe some retained wild-type peptide for At5g13450 in icp55 (Fig. 3B), indicating that some cleavage of this Phe is still possible in icp55, but it is incomplete. In the same gel bands, we observed the semitryptic N-terminal peptide of dihydrolipoamide dehydrogenase (At1g48030) and ubiquinone cytochrome C reductase iron-sulfur (At5g13440), both of which have classic −3R cutting sites (RGF↓A) in the wild type. The same peptide defining the −3R site was observed in icp55 with no evidence of longer peptides (Fig. 3B).

We compared the observed cutting site with the prediction of an ICP55 cutting site using MitoFates (Fukasawa et al., 2015; http://mitf.cbrc.jp/MitoFates) for mitochondrial presequence cleavage. The observed ICP55 cutting sites of −3R proteins At5g13450, At1g32470, and At2g35370 could also be predicted, but predicted cutting sites for At1g58030 and At5g13440 were not detected in icp55 (Fig. 3C). For the non-R group proteins, At4g37930, At5g08670, and At5g8690, the observed sites of cutting in icp55 were not predicted by MitoFates (Fig. 3C). Notably, the training sequences used for MitoFates development were restricted to −3R group proteins (Fukasawa et al., 2015).

To investigate the potential impact of AtICP55 knockout on mitochondrial protein abundance, we isolated mitochondria and compared the proteomes of icp55 and the wild type using differential in-gel electrophoresis (DIGE). There were very few differences in protein pattern and abundance of mitochondrial protein between icp55 and the wild type (Supplemental Fig. S4), indicating that mitochondrial protein abundance is likely to be controlled by a sophisticated network that cannot be disrupted by the loss of a single peptidase such as ICP55. We could only identify one protein spot that changed in abundance, which appeared to be a protein that differed in pI by approximately 0.3 units in icp55 and the wild type (Supplemental Fig. S4). Both these protein spots were identified as ATP synthase Δ-subunit (At5g13450), which belongs to the −3R cutting site group (RTY↓A) in the wild type and the RT↓YA cutting site group in icp55 (Fig. 3B). We calculated the pI of each mature At5g13450 protein using simple calculators of pI, which suggested no change in pI. However, more sophisticated algorithms such as ProMoST (http://proteomics.mcw.edu/), which take into account the location of changed residues and changes in the acid dissociation constant of N- and C-terminal amino acids, showed that the removal of the N-terminal Tyr and the generation of an N-terminal Ala alter the predicted pI of the whole mature protein by 0.27 pI units. This prediction indicates that the protein in icp55 would be more acidic than the protein from the wild type, which is entirely consistent with the experimental data in Supplemental Figure S4.

Disruption of ICP55 Alters Mitochondrial Protein Stability in Vitro

In yeast, the loss of ICP55 causes an unusual phenotype in mitochondria isolated from mutants. Mitochondrial ICP55 target proteins in icp55 were shown to be degraded more rapidly than the wild type when isolated mitochondria were incubated in buffer at 37°C or 22°C and the protein abundances were followed over time after SDS-PAGE separation (Vögtle et al., 2009). To detect if the loss of ICP55 impacts protein degradation rate in Arabidopsis mitochondria in a similar manner, we conducted in vitro time-course experiments using isolated mitochondria. We incubated freshly isolated mitochondria in washing buffer (0.3 m Suc and 10 mm TES, pH 7.4) at 22°C for 12, 16, and 20 h and then analyzed their composition with SDS-PAGE (Fig. 4). After 12 h of incubation, the total protein abundance pattern was slightly lower in icp55 than in the wild type. But after 16 and 20 h of incubation, the abundance of protein in icp55 was much less than that in the wild type (Fig. 4A), suggesting a greater instability of isolated mitochondrial proteins in icp55 than in the wild type. To ensure that the observed degradation of protein was via proteases, the effect of proteinase inhibitors (Roche) on this assay was studied. The enhanced mitochondrial protein degradation in icp55 was totally prevented by the addition of proteinase inhibitors (Supplemental Fig. S5), indicating the contribution of mitochondrial proteases in the differential protein degradation observed.

Figure 4.

In vitro protein degradation in mitochondria isolated from the wild type (WT) and icp55 in 3-week-old shoots. A, Coomassie Blue-stained gel of wild-type and icp55 mitochondrial proteins after 12, 16, and 20 h of incubation of isolated mitochondria at 22°C. B, Two-dimensional DIGE analysis of protein abundance between wild-type and icp55 mitochondrial extracts after in vitro degradation at 22°C for 20 h. The identities of the differentially abundant protein spots 1 to 18 are given in Table I.

To further identify if there was specificity of the protein degradation observed in vitro, we conducted a DIGE experiment comparing the wild type and icp55 after 20 h of treatment at 22°C (Fig. 4B; Table I). Again, the most notable difference was spots 16 and 17, the ATP synthase Δ-subunit (At5g13450) that appeared at a different pI in icp55 and the wild type. The only other protein spot (spot 15) that was more abundant in icp55 was a partial SHMT (At4g37930) product (Supplemental Fig. S6). This protein has a nonconserved Arg cleavage site and is altered in icp55, as one more Ser is retained in the mature protein sequence (Fig. 3B).

Table I. Identification of proteins with different abundances in wild-type and icp55 mitochondria after 20 h of in vitro degradation in Figure 4B.

Proteins were identified by MS/MS, and the predicted M_r (MW) and pI of the match are shown along with the Mascot scores (P < 0.05 when the score is greater than 37). The number of peptides matched to tandem mass spectra and the percentage coverage of the matched sequence are also shown. The classification of ICP55 cleavage site (Y, Yes) was based on data in this report (Fig. 3; Supplemental Data Set S1) and our previous report (Huang et al., 2009; O, observed), and the predicted ICP55 cleavage was based on MitoFates (Fukasawa et al., 2015; P, predicted). FC, Fold change (icp55/wild type).

Spot No.	Arabidopsis Genome Initiative Accession No.	FC	Description	Precursor Cleavage Site Group	ICP55 Cleavage	Mascot Score	MW	Peptide No.	Percentage Coverage	pI
Spot 1	At4g26970	−1.38	Aconitate hydratase	Unknown	Y (P)	235	108,413	4	7.4	6.71
Spot 2	At5g37510	−1.73	NADH dehydrogenase 75-kD subunit	Unknown		491	81,130	7	13.4	6.24
Spot 3	At2g26080	−2.20	GDC P subunit	−2R		106	113,703	4	4.3	6.18
	At4g33010
Spot 4	At3g02090	−3.37	MPP β-subunit	Unknown		608	59,123	8	20	6.3
Spot 5	At5g08670	−1.85	ATP synthase β-subunit	Non-R	Y (O)	348	59,594	8	19.4	6.13
	At5g08680				Y (P, O)
	At5g08690				Y (O)
Spot 6	At1g51980	−2.14	MPP α-subunit	Unknown		476	54,368	7	14.3	5.94
Spot 7	AtMg01190	−1.78	ATP synthase α-subunit	Unknown		478	54,937	8	19.5	6.23
Spot 8	At2g44350	−1.52	Citrate synthase	−2R		446	52,621	7	17.1	6.41
Spot 9	At3g20000	−1.52	Translocase of the outer membrane40	Unknown		146	34,228	4	16.8	6.32
Spot 10	At3g09260	−1.56	β-Glucosidase; PYK10	Unknown		106	59,683	2	4.8	6.45
Spot 11	At5g08300	−3.84	Succinyl-CoA ligase	−3R	Y (P)	213	36,129	5	19.9	8.55
Spot 12	At5g63400	−1.72	Adenylate kinase1	Noncutting		178	26,915	5	25.6	6.91
Spot 13	At4g35260	−1.77	Isocitrate dehydrogenase	Unknown		138	39,602	4	11.4	8.36
Spot 14	At5g40770	−1.56	Prohibitin3	Noncutting		274	30,381	7	29.6	6.99
Spot 15	At4g37930	3.55	SHMT	Non-R	Y (O)	222	57,364	7	13.2	8.13
Spot 16	At5g13450	−8.62	ATP synthase Δ-subunit	−3R	Y (P, O)	94	26,305	3	10.9	9.25
Spot 17	At5g13450	4.98	ATP synthase Δ-subunit	−3R	Y (P, O)	135	26,305	5	16	9.25

Other proteins identified had significantly decreased abundance in icp55 (Fig. 4B; Table I). Those with known cleavage sites (Huang et al., 2009) were succinyl-CoA ligase (At5g08300; −3R group), citrate synthase (At2g44350; −2R), and Gly decarboxylase P-protein2 (At2g26080 and At4g33010; −2R group). A nonconserved Arg motif protein, ATP synthase β-chain1 (At5g08670, At5g08680, and At5g08690; non-R group), was also detected to be 1.9-fold lower in abundance in icp55 than in the wild type. Proteins with unknown cleavage sites that decreased in abundance included aconitate hydratase (At4g26970), NADH dehydrogenase (At5g37510), mitochondrial processing peptidase α- and β-subunits (At1g51980 and At3g02090), translocase of the outer membrane40 (At3g20000), and isocitrate dehydrogenase1 (At4g35260). Proteins known not to undergo cleavage but decreased in icp55 included adenylate kinase1 (At5g63400), prohibitin3 (At5g40770), and ATPase subunit1 (AtMg01190). Aconitase (At4g26970) and succinyl-CoA ligase (At5g08300) are predicted to have ICP55 cleavage sites based on MitoFates (Fukasawa et al., 2015; Table I). These data suggested that elevated in vitro mitochondrial protein degradation in icp55 was not simply a direct effect of ICP55 function; rather, it was likely to be due to a changed network of proteolytic events. This claim is supported by the inhibition of in vitro protein degradation in ICP55 by the addition of a protease inhibitor (Supplemental Fig. S5). These observations are distinct from the report in yeast Δicp55, where only ICP55 targets were found to be more unstable in in vitro assays (Vögtle et al., 2009).

Disruption of ICP55 Changes the Protein Turnover Rate of Selected Proteins in Vivo

The significant differences in mitochondrial proteome proteolysis in vitro (Fig. 4B) did not appear to be consistent with the lack of differences in the abundance of proteins in the isolated mitochondrial proteome (Supplemental Fig. S4) or the absence of whole-plant visual phenotypes (Fig. 3A; Supplemental Fig. S3). To measure protein turnover rates in vivo, we applied our recently developed technology by labeling hydroponically grown plants with ¹⁵N inorganic salts and extracting mitochondrial proteins at a defined time point (Nelson et al., 2013). Using 1D gels for separation, we focused on the proteins studied in Figure 3B. Most of the mitochondrial proteins had very similar turnover rates between the wild type and icp55 (Fig. 5), which is consistent with the lack of obvious phenotypic differences between the plants. However, we found that SHMT (At4g37930) had a higher in vivo turnover rate in the wild type than in icp55 (Fig. 5). Therefore, SHMT had an unusual cutting site (S↓S; Fig. 3B), the N-terminal portion of the protein was retained in in vitro degradation (spot 15 in Fig. 4B), and SHMT had a slower in vivo protein turnover rate in icp55 (Fig. 5).

Figure 5.

Protein turnover rates of selected proteins. In vivo protein turnover rates of selected proteins (Fig. 3B) are shown from shoots of icp55 and the wild type (WT). *P < 0.05 (n = 3).

Complementation of icp55 Restores Wild-Type Protein Stability and Mitochondrial Protein N Termini

To confirm that the cutting events recorded in icp55 were caused by the ICP55 protein, we cloned an At1g09300 full-length cDNA (for primer information, see Supplemental Fig. S1) from wild-type Columbia-0. To our surprise, there was one nucleotide close to the stop codon that was not present in at least four independent clones when compared with the full-length At1g09300 cDNA sequence from The Arabidopsis Information Resource (TAIR). We checked genomic DNA sequences of At1g09300 in other ecotypes and observed the same sequences we cloned from Columbia-0. We have deposited this full-length cDNA sequence for ICP55 (At1g09300) into the National Center for Biotechnology Information database with accession number KF921304. Due to the absence of this one nucleotide, the predicted amino acid sequence encoded in the cloned full-length cDNA was noticeably shorter than that proposed in the TAIR database (Supplemental Fig. S7). This places the disruption in Sail_672_D05 at position 57 instead of position 70 from the C terminus of the protein (Supplemental Fig. S7), which is within the PF00557 metallopeptidase domain of the protein and upstream of one conserved residue involved in metal binding (Fig. 1D). We transformed the cloned At1g09300 full-length cDNA under the control of the 35S cucumber mosaic virus promoter in pGreen pMDC43 vector into icp55. We obtained 23 independent lines following selection on hygromycin and PCR confirmation. To analyze the ICP55 expression level in wild-type, icp55, and complemented lines, we conducted qRT-PCR based on specially designed primers (Fig. 2A; Supplemental Fig. S1). In the complemented line, the transcript levels both in the upstream region of insertion (F12R1 and F12R2) and in the insertion region (F3R3 and F4R4) were 20- to 30-fold those in the wild type (Fig. 6), indicating the overexpression of the ICP55 cDNA. The transcript levels of three housekeeping genes (PPase2A, At1g13320; Yls8, At5g08290; and Clathrin, At5g46630) in wild-type, icp55, and complemented lines were very similar (Fig. 6A). To check the protein abundance of ICP55 in the complemented line, we analyzed the abundance of peptides 1 and 2 (Fig. 2C) using selected reaction monitoring. Compared with the wild type, the complemented line had approximately 20-fold higher abundance of both peptides (Fig. 6B), indicating strong accumulation of ICP55.

Figure 6.

Characterization of the complementation of icp55. A, Expression levels of the ICP55 gene in the wild type (WT; black bars), icp55 (white bars), and the complemented lines (gray bars). The locations of primers used for qRT-PCR are given in Figure 2A. PPase, Yls8, and Clathrin are three housekeeping transcripts. B, Relative peptide levels detected by selected reaction monitoring in the wild type (black bars) and the complemented line (gray bars). Peptides 1 and 2 are not detectable in icp55, as shown in Figure 3. Aconitase (At4g26970) is a tricarboxylic acid cycle enzyme used as a control. C, Coomassie Blue-stained 1D SDS-PAGE gel of wild-type, icp55, and complemented line proteins after 0 and 20 h of incubation of isolated mitochondria from 2-week-old plants. EDTA at 4 mm was added in the incubation medium. The quantification of selected bands (A–C) using ImageJ is given at the bottom. D, Cleavage sites of mitochondrial proteins in the wild type, icp55, and the complemented line based on MS analysis (for details, see Supplemental Data Set S1).

As there was no clear phenotypic difference between icp55 and the wild type, we used in vitro protein degradation from freshly isolated mitochondria from 2-week-old (Fig. 6C) and 3-week-old (Supplemental Fig. S8) plants as a complementation test. The 2-week-old plants are consistent with the developmental stage for the protein turnover study (Fig. 5), and the 3-week-old plants are consistent with the protein degradation analysis (Fig. 4). Again, icp55 showed faster in vitro protein degradation of a number of protein bands when compared with the wild type and the complemented line (Fig. 6C; Supplemental Fig. S8). These results confirmed that the absence of ICP55 was responsible for the more rapid in vitro degradation of a selection of mitochondrial proteins. The addition of EDTA in the medium blocked protein degradation in icp55, indicating the contribution of metals to protein degradation in icp55 (Fig. 6C). We also confirmed the complementation of the cutting sites of proteins At4g37930 (SHMT1), At5g08670 (ATP synthase β-chain1), At5g13450 (ATP synthase Δ-chain1), At2g35370 (GDC H protein1), and At1g32470 (GDC H protein2; Fig. 6D). The MS/MS spectra for the complementation of the N-terminal peptides for At4g37930, At5g08670, At5g13450, At2g35370, and At1g32470 are presented in Supplemental Data Set S1 and show the recovery of the wild-type cutting site in each case. These results show, through the overexpression of ICP55 in the icp55 background, that At1g09300 is responsible for the differential cutting sites of mitochondrial proteins and in vitro mitochondrial protein stability.

DISCUSSION

Plant ICP55 Cleaves −3R and Nonconserved R Proteins in Mitochondria at Recognized Residues

We previously characterized Arabidopsis and rice mitochondrial presequence cleavage motifs by searching for semitryptic N-terminal peptides using MS (Huang et al., 2009). The similarity in mitochondrial presequence cleavage motifs between plants and yeast (Huang et al., 2009; Vögtle et al., 2009) and the existence of orthologs of ICP55 in plants led us to investigate if At1g09300 was an ICP55-like protein in Arabidopsis that might be responsible for generating −3R mature proteins in plant mitochondria. Our data show that ICP55-dependent cleavage of −3R group proteins in Arabidopsis removes Phe or Try to reveal Ala or Ser N termini (Fig. 3B), in a similar manner to that shown in yeast (Vögtle et al., 2009). These observed cleavages were all complemented when the Arabidopsis icp55 insert line was complemented with a cDNA encoding ICP55. This result is generally consistent with a recent independent study using the charge-based fractional diagonal chromatography (ChaFRADIC) technique for the enrichment of N-terminal peptides (Carrie et al., 2015). We found the cleavage of F↓S and Y↓A but not F↓A −3R substrates in icp55 (Fig. 3B). In the yeast icp55 null, there are cases where both cleaved and uncleaved N termini are found in the wild type and/or mutant (Vögtle et al., 2009), indicating that complete cleavage is not always attained and that ICP55-independent pathways for the cleavage of destabilizing N termini are also possible. ChaFRADIC analysis (Carrie et al., 2015) also detected both cleaved and uncleaved peptides, including for dihydrolipoamide dehydrogenase (At1g48030), for which we only found uncleaved peptides in this study. The differences may reflect the different methods of N-terminal identification by ChaFRADIC analysis (Carrie et al., 2015), where Arg residues in peptides are not cleaved, leading to longer peptides aiding identification. The cleavage of ATP synthase β-subunit (At5g08670), while not a −3R group protein, is a Y↓A cleavage and therefore is consistent with the other cleavages catalyzed for ICP55 in plants and yeast. The nonconserved Arg cutting sites in plant proteins could not be predicted by MitoFates (Fig. 3C), further indicating a specific cleavage system. It should also be pointed out that Arabidopsis At4g29490 has some homology to ICP55, with 26% identity and 45% similarity at the amino acid level (Supplemental Fig. S9). At4g29490 is predicted to be located in plastid or mitochondria by various prediction software programs (http://suba3.plantenergy.uwa.edu.au/), but experimental evidence such as GFP localization will be required to confirm its organelle localization. If At4g29490 is dual targeted to mitochondria and chloroplast, it may explain the coexistence of cleaved and uncleaved N termini in icp55 both in this study (Fig. 3B) and as described by Carrie et al. (2015). The functional characterization of At4g29490 is under way to test this hypothesis.

ICP55-Dependent Cleavage of SHMT

Our data show that plant ICP55 is also responsible for the maturation of SHMT (At4g37930). In contrast to the other cleavages, SHMT is not only in the non-R cutting site class of cleavages, but ICP55 appears to cut this protein between Ser residues, which is distinct from any of the other ICP55 target motifs identified in plants (Fig. 3B) or yeast (Vögtle et al., 2009), which remove Tyr, Leu, or Phe. In addition to Tyr, Leu, or Phe residues, ChaFRADIC analysis indicated that AtICP55 can cleave Ile, Met, and Gln but, interestingly, did not find examples of cleavage of Ser (Carrie et al., 2015). The unusual cleaved PheCys, PheThr, TyrAsn, and PheAsn residues were also recently observed by Carrie et al. (2015). Herein, the unusual Ser cutting site of SHMT (At4g37930; Fig. 3B), the in vitro accumulation of the N-terminal portion of SHMT in icp55 mitochondria during proteolysis (Fig. 4B), and the slower protein turnover of SHMT in icp55 in vivo (Fig. 5) all point to a specific biological effect of ICP55 on the stability of the SHMT protein. SHMT is also noted to change in abundance in response to light and dark diurnal cycles (Lee et al., 2010), suggesting that its turnover under natural conditions could be significant. However, we have not yet found any gross phenotypic difference between icp55 and the wild type in long days, short days, or after treatment with different temperatures or elevated light intensity.

The Role of ICP55 with Other Proteases in Plant Mitochondria

Nonconserved Arg cleavage has also been reported to be catalyzed by Oct1, which has been found in mammals and yeast to cleave after MPP and yield −10R products (Isaya et al., 1991; Vögtle et al., 2011). Our data indicated that ICP55 is responsible for amino acid cleavage after presequence cleavage from several nonconserved R proteins (Fig. 3B). In the recent ChaFRADIC analysis, 28 (including At5g08670 found in this study) of the 88 apparent substrates belong to the non-R cleavage group (Carrie et al., 2015). However, as these 88 cleavages were only shown in a single mutant line and none of these reported cleavages were shown to be complemented by the expression of ICP55 in the mutant background, they still await final confirmation. In Arabidopsis, it has been suggested that At5g65620 and At5g51540 could be potential candidates for Oct1-like proteins in mitochondria (Kwasniak et al., 2012). However, there are no clear −10R group proteins in our survey of Arabidopsis and rice mitochondrial protein cleavage motifs (Huang et al., 2009). Indeed, analysis of seven substrates cleaved by the Oct1-like protein encoded by At5g51540 indicated that Arabidopsis does not have a classical −10R motif (Carrie et al., 2015). Coexpression analysis of Arabidopsis genes (http://atted.jp) reveals the expression of ICP55 clusters with the expression of three other mitochondrial components, At2g31060, At3g17910, and Lon1 (At5g26860). At2g31060 is an elongation factor family protein and has been found in Arabidopsis mitochondrial membranes by MS (Nikolovski et al., 2012). At3g17910 is similar in sequence to human SURFEIT1, which is known to be involved in cytochrome c oxidase assembly, and is also reported to be found in Arabidopsis mitochondria (Heazlewood et al., 2004; Klodmann et al., 2011; Nikolovski et al., 2012). Lon1 is a mitochondrial protease located in the matrix (Rigas et al., 2009). Point mutations and knockout lines of Arabidopsis Lon1 have a clear growth deficit compared with the wild type. Lon1 mutants also show changes in the mitochondrial proteome, including increases in Heat shock protein70, prohibitins, peroxiredoxin, SHMT, aconitase, and lipoamide dehydrogenase and lower abundance of complexes I, II, and IV (Rigas et al., 2009; Solheim et al., 2012). Interestingly, the ICP55 homolog At4g29490 is also coexpressed with another Lon protein, At2g25740 (http://atted.jp). The complicated protein degradation of the mitochondrial network related to ICP55 and other proteases is also indicated by the in vitro nonselected protein degradation in mitochondria isolated from icp55 (Fig. 4). The inhibition of protein degradation by the metal chelator EDTA (Fig. 6C) suggested that metal-dependent proteinase(s) is involved in this process. The potential that Lon1, ICP55, and other metalloproteinases work closely together in the stabilization and/or degradation of mitochondrial proteins, including SHMT, adds to the process of better defining the plant mitochondria protease network.

MATERIALS AND METHODS

Plant Lines and Growth Conditions

Seeds of Arabidopsis (Arabidopsis thaliana) ecotype Columbia-0 (wild type) and T-DNA insertion lines (Salk_123637, Salk_123635, Sail_672_D05, and Salk_001742) were sown on a 1:3:1 perlite:shamrock compost:vermiculite soil mix and covered with transparent acrylic hoods. After 3 d at 4°C in the dark, plants were grown under controlled conditions as follows: a short-day photoperiod (8 h of light/16 h of dark) with a light intensity of 250 µmol m⁻² s⁻¹, a relative humidity of 70%, and a temperature cycle of 22°C day/17°C night; and a long-day photoperiod (16 h of light/8 h of dark) with a light intensity of 200 µmol m⁻² s⁻¹, a relative humidity of 70%, and a temperature cycle of 22°C day/17°C night. For plate growth, seeds were sown on one-half-strength Murashige and Skoog Gamborg B5 plates containing 0.8% (w/v) agar, 1% (w/v) Suc, and 1.8 mm MES, with pH adjusted by KOH to 5.8. After 3 d at 4°C in the dark, plates were transferred to long-day controlled conditions as described above and set in a vertical position to facilitate the analysis of root length. For heat and high-light stress, plants on soil or plates were moved into a chamber set at 37°C or 40°C, or high light was applied as described in “Results.”

Analysis of the Arabidopsis Mitochondria Proteome, and Identification of At1g09300

Previously, we had conducted in-depth analysis of the mitochondrial proteome to define mitochondrial protein presequences using mitochondrial matrix proteins isolated from Arabidopsis cell culture. We did not detect a presequence of At1g09300, but in these analyses of separated matrix proteins, we found high-quality MS/MS spectra matching At1g09300 as listed in Figure 1. Data collection was according to our previous methods (Huang et al., 2009; Taylor et al., 2011).

Isolation of Mitochondria from Hydroponic Plants Using Gradient Centrifugation

Mitochondria from shoots or seedlings were isolated from 2- and 3-week-old hydroponically grown Arabidopsis seedlings using a method described previously (Lee et al., 2008). The final mitochondrial samples was collected and washed three times by dilution and centrifugation. Aliquots were used directly for assays such as in vitro protein degradation analysis or were frozen for other analyses.

Quantitative Real-Time PCR for Transcripts

RNA was extracted from the leaves of 2-week old seedlings using the RNeasy Plant Mini Kit from Qiagen (www.qiagen.com) according to the manufacturer’s instructions. Reverse transcription analysis and the quantification of mRNAs were performed by qRT-PCR as described previously (de Longevialle et al., 2007). qRT-PCR was performed using LightCycler 480 SYBR Green I Master Mix (Roche Diagnostics; http://www.roche.com) and a Roche LightCycler 480 real-time PCR system with the following thermal cycling program: 95°C for 10 min, followed by 45 cycles of 95°C for 10 s, 60°C for 10 s, and 72°C for 10 s. The data were analyzed using the LightCycler 480 software version 1.5 (Roche Diagnostics). Three housekeeping genes (PPase 2A, At1g13320; Yls8, At5g08290; and Clathrin, At5g46630) were selected as internal controls. All primers used for qRT-PCR are listed in Supplemental Figure S1.

Quantification of Peptide Abundance Using Selected Reaction Monitoring

Protein extracts from nongel liquid chromatography-MS/MS of whole mitochondrial trypsin digests were analyzed on an Agilent 6430 QqQ mass spectrometer with an HPLC Chip Cube source (Agilent Technologies). The chip consisted of a 160-nL enrichment column (Zorbax 300SB-C18; 5-µm pore size) and a 150-mm separation column (Zorbax 300SB-C18; 5-µm pore size) driven by the Agilent Technologies 1200 series nano/capillary liquid chromatography system. Both systems were controlled by MassHunter Workstation Data Acquisition for QqQ (version B.03.01, build 2600; Agilent Technologies). Peptides were loaded onto the trapping column at 3 µL min⁻¹ in 5% (v/v) acetonitrile and 0.1% (v/v) formic acid with the chip switched to enrichment and using the capillary pump. The chip was then switched to separation, and peptides were eluted during a 30-min gradient (5% [v/v] acetonitrile to 100% [v/v] acetonitrile) directly into the mass spectrometer. Parent ions and transitions (three transitions per parent ion, one quantifier and two qualifiers) were selected based on previous Q-TOF experiments and were 997.1/1,425.8 (997.1/713.4 and 997.1/1,199.7) for LVELLPENSLAIISSAPVK and 836.9/1,181.5 (836.9/951.5 and 836.9/1,066.5) for YTNLDDFQNSASLGK. The mass spectrometer was run in positive ion mode, and for each parent/transition, the fragmentor was set to 130, dwell time was 5 ms, and collision energy was 22.9 V for LVELLPENSLAIISSAPVK and 23.99 V for YTNLDDFQNSASLGK. Detailed information for parent ions, transition, and collision energy for peptides (FSYNGQPAEIK, ILDWENTSTK, and GVISEDFNSYGSR) from tricarboxylic acid cycle enzyme aconitase2 (At4g26970) that was used as a control has already been reported (Taylor et al., 2014). Selected reaction monitoring chromatograms were analyzed in MassHunter Workstation Quantitative Analysis (version B.04.00, build 4.0.225.0; Agilent Technologies), and quantitative results were obtained by integrating the area under the peak of each quantifier transition. Qualifiers and retention time information were used to confirm that the right peak was integrated.

In Vitro Mitochondrial Protein Degradation Assay

The freshly isolated and ice-cold mitochondria were divided into aliquots in Eppendorf tubes (50 or 200 µg) with a protein concentration of approximately 1 µg µL⁻¹ in washing buffer (0.3 m Suc and 10 mm TES, pH 7.4). The tubes were set on the thermomixer at 22°C to promote in vitro protein degradation. After 0, 12, 16, and 20 h, the Eppendorf tubes containing mitochondria were stored at −80°C for further gel separation as described below. For proteinase inhibitor assays, proteinase inhibitor cocktail (Roche Diagnostics) was used with one tablet dissolved in 50 mL of washing buffer. EDTA at the concentration of 5 mm in washing buffer was used to inhibit metal-dependent proteolysis.

1D SDS-PAGE and Two-Dimensional DIGE Analysis

Mitochondrial proteins (100 µg) from wild-type, icp55, and complemented lines were dissolved in sampling buffer (62.5 mm Tris-HCl, pH 6.8, 2% [w/v] SDS, 10% [v/v] glycerol, 5% [v/v] β-mercaptoethanol, and 0.2% [w/v] bromophenol blue) and heated at 90°C for 5 min before loading onto a 4% stack and 12% (v/v) separation gel for 1D SDS-PAGE separation. Two-dimensional DIGE was conducted as follows. Mitochondrial proteins (50 µg) of the wild type and icp55, as well as 50 µg of a 1:1 mixture of both samples, were acetone precipitated and resolubilized in lysis buffer (7 m urea, 2 m thiourea, 4% [w/v] CHAPS, and 40 mm Tris base, pH 8.5). The individual samples were labeled with 400 pmol of the weight- and pI-matched fluorescent dyes Cy2, Cy3, and Cy5 (GE Healthcare) and then combined and separated on IEF strips (pH 3–10NL, 24 cm; GE Healthcare) according to the manufacturer’s instructions. After a brief wash in 1× gel buffer, the strips were transferred on top of a 12% (v/v) polyacrylamide gel, where they were covered with 1.2% (w/v) agarose in 1× gel buffer. Two-dimensional gels were run at 50 mA per gel for 6 h. The fluorescence-labeled proteins were visualized on a Typhoon Trio laser scanner (GE Healthcare), and image comparison was conducted using the DECYDER software package (version 6.5; GE Healthcare) with statistical analysis of three independent experiments. Gel images were overlaid using the ImageQuant TL software (GE Healthcare). MS identification of protein spots using matrix-assisted laser-desorption ionization time of flight/time of flight was conducted as outlined previously (Huang et al., 2013).

Semipeptide Assay from a 1D SDS-PAGE Gel for Cutting Site Analysis

After mitochondrial proteins were separated by 1D SDS-PAGE, gel bands to be analyzed were cut from the gels and were in-gel digested with trypsin according to previous methods (Taylor et al., 2005). The digested proteins were then dried in a vacuum centrifuge, resuspended in 5% (v/v) acetonitrile and 0.1% (v/v) formic acid, and analyzed on a 6510 series Q-TOF mass spectrometer (Agilent Technologies) in a manner similar to previous methods (Nelson et al., 2013). Resulting MS/MS-derived spectra were analyzed against our in-house Arabidopsis mitochondrial protein database extracted from SUBA (http://suba.plantenergy.uwa.edu.au/). The database was searched using the Mascot search engine version 2.3.02 with selection of none enzyme digestion and utilizing error tolerances of ±1.2 D for MS and ±0.6 D for MS/MS, maximum missed cleavages set to 1, with variable modification of oxidation (M), carbamidomethyl (C), and acetylation (N-term), instrument set to ESI-QUAD-TOF, and peptide charge set at 2⁺ and 3⁺. Results were filtered using standard scoring, maximum number of hits set to AUTO, and ion score cutoff at 20. The significance threshold P ≤ 0.05 and require bold red were also set.

In Vivo Protein Turnover Rate Assay by ¹⁵N Labeling

Wild-type or icp55 seeds (approximately 30 mg) were sterilized and washed in sterile water five times before being dispensed into a 250-mL plastic vessel that contained 80 mL of growth medium (one-half-strength Murashige and Skoog medium without vitamins, one-half-strength Gamborg B5 vitamin solution, 5 mm MES, and 2.5% [w/v] Suc, pH 5.8). Plants were grown under long-day conditions, as described above, on a shaker (rotation speed, 90 rpm) for 10 d. Three biological replicates were collected for control mitochondrial preparations (¹⁴N labeled). After washing three times with the same growth medium without nitrogen, the plants were transferred to ¹⁵N growth medium containing ammonium-¹⁵N nitrate-¹⁵N (0.96 g L⁻¹) and potassium nitrate-¹⁵N (0.83 g L⁻¹). After 4 d of growth, three biological replicates were collected for treated mitochondrial preparations (¹⁵N labeled). Ten to 15 g of seedlings was homogenized with a mortar and pestle, and mitochondria were isolated as reported previously (Lee et al., 2008). Fresh isolated mitochondria were quantified with the Bradford assay protocol, and proteins were separated by SDS-PAGE (100 μg of protein per lane). Bands corresponding to the target proteins (Fig. 3B) were cut and in-gel digested by trypsin. An Agilent 6550 Q-TOF device was used for protein identification and quantification. MS data were processed as described recently (Nelson et al., 2014), and Mathematica (Wolfram Research) scripts written in-house were used for measurements of labeled protein fractions. Growth (i.e. fold change in protein abundance [FCP]) of the wild type (FCP = 2.65 ± 0.76) and icp55 (FCP = 2.01 ± 0.25) was determined by the fresh weight change of the three biological replicates during the labeling process. The protein degradation rates were determined by the labeled protein fractions and FCP as described previously (Li et al., 2012). The protein degradation rates from each biological replicate in the wild type or icp55 were averaged and compared by Student’s t test.

Complementation of icp55

The complementation of the icp55 knockout line (Sail_672_D05) was conducted using the full-length At1g09300 cDNA (accession no. KF921304). The resulting At1g09300 cDNA was recombined into the Gateway pDONR vector (Invitrogen), verified by DNA sequencing, and recombined into the modified binary vector pMDC43 with cauliflower mosaic virus 35S promoter (Curtis and Grossniklaus, 2003). This construct was introduced into Agrobacterium tumefaciens, which was then used to transform homozygous icp55 by a modified floral dip method (Pracharoenwattana et al., 2005). Transformed plants were selected by germinating seedlings on agar plates containing 25 mg L⁻¹ hygromycin.

The full-length cDNA sequence was deposited into the National Center for Biotechnology Information database with accession number KF921304.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Primers used for genotyping icp55 insertion lines, full-length cDNA amplification, and qRT-PCR.

• Supplemental Figure S2. PCR genotype of four T-DNA insertion lines for ICP55.

• Supplemental Figure S3. Response of the wild type and icp55 to high temperature and high light.

• Supplemental Figure S4. DIGE analysis of the proteome of mitochondria isolated from leaves of 3-week-old hydroponically grown plants.

• Supplemental Figure S5. In vitro protein degradation of freshly isolated mitochondrial proteins from the wild type and icp55 when incubated with (+) or without (−) proteinase inhibitors.

• Supplemental Figure S6. Peptides (red) detected in spot 15 in Figure 4B and their positions in the SHMT (At4g37930) protein sequence.

• Supplemental Figure S7. Predicted amino acid sequence alignment of amplified full-length cDNA (KF921304) and the TAIR At1g09300 coding sequence.

• Supplemental Figure S8. Quantification of protein degradation after 20 h in mitochondria from 3-week-old shoot from wild-type, icp55, and complemented lines.

• Supplemental Figure S9. Sequence similarity of ICP55 and At4g29490.

• Supplemental Data Set S1. MS spectra and data on N-terminal peptides of proteins in icp55, wild-type, and complemented lines.

Glossary

MPP

mitochondrial processing peptidase

T-DNA

transfer DNA

SHMT

serine hydromethyltransferase

cDNA

complementary DNA

qRT

quantitative reverse transcription

1D

one-dimensional

Q-TOF

quadrupole time of flight

MS

mass spectrometry

MS/MS

tandem mass spectrometry

DIGE

differential in-gel electrophoresis

TAIR

The Arabidopsis Information Resource

FCP

fold change in protein abundance