Role of mitochondrial processing peptidase and AAA proteases in processing of the yeast acetohydroxyacid synthase precursor

We studied presequence processing of the mitochondrial‐matrix targeted acetohydroxyacid synthase (Ilv2). C‐terminal 3HA‐tagging altered the cleavage pattern from a single step to sequential two‐step cleavage, giving rise to two Ilv2‐3HA forms (A and B). Both cleavage events were dependent on the mitochondrial processing peptidase (MPP). We present evidence for the involvement of three AAA ATPases, m‐ and i‐AAA proteases, and Mcx1, in Ilv2‐3HA processing. Both, precursor to A‐form and A‐form to B‐form cleavage were strongly affected in a ∆yme1 mutant. These defects could be suppressed by overexpression of MPP, suggesting that MPP activity is limiting in the ∆yme1 mutant. Our data suggest that for some substrates AAA ATPases could play an active role in the translocation of matrix‐targeted proteins.

We studied presequence processing of the mitochondrial-matrix targeted acetohydroxyacid synthase (Ilv2). C-terminal 3HA-tagging altered the cleavage pattern from a single step to sequential two-step cleavage, giving rise to two Ilv2-3HA forms (A and B). Both cleavage events were dependent on the mitochondrial processing peptidase (MPP). We present evidence for the involvement of three AAA ATPases, m-and i-AAA proteases, and Mcx1, in Ilv2-3HA processing. Both, precursor to A-form and A-form to B-form cleavage were strongly affected in a Δyme1 mutant. These defects could be suppressed by overexpression of MPP, suggesting that MPP activity is limiting in the Δyme1 mutant. Our data suggest that for some substrates AAA ATPases could play an active role in the translocation of matrix-targeted proteins.
Acetohydroxyacid synthase (Ilv2) catalyzes the first step in the biosynthesis of branched-chain amino acids [1], which mainly takes place in the mitochondrial matrix [2]. The nuclear encoded Ilv2 is synthesized in the cytoplasm with an N-terminal mitochondrial targeting sequence (MTS) directing import into the mitochondrial matrix [3]. Matrix-targeted proteins are recognized by receptors on the surface of mitochondria and are imported through multi-subunit translocases of the outer membrane (TOM) and inner membrane (TIM) into the mitochondrial matrix [4]. The presequences of matrix-targeted proteins are cleaved off by mitochondrial processing peptidase (MPP), a conserved hetero-dimeric metallopeptidase, which in yeast consists of the subunits Mas1 and Mas2 [5]. Preprotein cleavage does not appear to be essential for translocation into the matrix, since import continues in the absence of processing [6]. But, processing seems to be important for the activity of at least some proteins, since yeast cells are not viable in the absence of MPP [7]. Also, nonprocessed precursor proteins are less stable, which could explain why MPP activity is essential [8]. In some cases, additional sequences are removed after MPP cleavage. The mitochondrial intermediate peptidase (MIP) removes an octapeptide from the preproteins after MPP processing [9], the protease Icp55 removes one amino acid [10]. These additional cleavage events could play a role in regulating the stability of the proteins.
A certain fraction of Ilv2 seems to be localized to the cytoplasm [11]. Since the N-terminal extension of Ilv2 is unusually long compared with other matrix-targeted proteins (90 aa) [10], it could play a role in regulating the distribution between cytoplasm and mitochondrial matrix. Therefore, the N-terminal processing of the Ilv2 preprotein was examined more closely. N-terminal processing of Ilv2 was altered by C-terminal tagging. In contrast to native Ilv2, which was cleaved only once, two-step sequential cleavage was observed with C-terminally tagged Ilv2. The cleavage events were dependent on MPP and the m-and i-AAA proteases. Our results further suggest that MPP activity is limiting in the i-AAA protease mutant Δyme1.

Materials and methods
Yeast strains, plasmids, and media The yeast strains used are listed in Table 1. The yeast strains are derived from the CEN.PK and BY4741 backgrounds. Insertions and deletions were introduced into the yeast genome by one-step gene replacement with PCR-generated cassettes [12]. The insertions and deletions were verified by PCR. For overexpression of MAS1 and MAS2, the two genes were amplified by PCR from yeast chromosomal DNA (MAS1: pos. À630 to 1460; MAS2: pos. À630 to 1519) and cloned in tandem into the vector YEplac195 [13] to give plasmid pRK1109. All cloned genes were sequenced to ensure that no PCR mutations occurred. For standard experiments, yeast cells were grown in YPD medium (1% yeast extract, 2% bacto peptone, 2% glucose) or in SD/CAS medium (0.67% yeast nitrogen base w/o amino acids, 1% casein hydrolysate, 2% glucose).

Cycloheximide (CHX) chase
Yeast cultures were grown overnight to exponential phase (OD 600 ≤ 1.0) in YPD medium. CHX was added to a final concentration of 0.1 mgÁmL À1 to the culture. A constant amount of culture was harvested at 20 min intervals and cell extracts were prepared by glass-beading in lysis buffer (50 mM HEPES, pH 7.4; 0.3 M sorbitol; 10 mM NaN 3 ; with protease inhibitors) and analyzed by western blotting with anti-Ilv2 or anti-HA antibodies. Bound antibodies were detected with the Lumi-Light Kit (Roche, Mannheim, Germany) and luminescence signals were recorded with the LAS-3000 imaging system (FUJI, Kleve, Germany). Digital images of immunoblots were analyzed using ImageJ 1.47v (http://rsb.info.nih.gov/ij/). All experiments were performed at least twice, with the same result. A representative experiment is shown in each case.

Kinetic model
We hypothesized that the Ilv2 preprotein cleavages are sequential. Our system is therefore an example of three consecutive, irreversible, homogeneous first-order reactions in a closed system: (P is the precursor, A is the Ilv2 A-form and B is the Ilv2 B-form, X degradation products).
The rate equation connects the reaction rate with concentrations and constant parameters as follows: PðtÞ ¼ e Àkpat P 0 ; AðtÞ ¼ k pa e Àkabt k pa À k ab À k pa e Àkpat k pa À k ab P 0 þ e Àkabt A 0 ; However, we observed, that k pa ≫ k ab for most of the measured samples. Hence, the precursor can be neglected as it vanishes very quickly. The solution simplifies to: The matrix exponential was performed with GNU Maxima (http://maxima.sourceforge.net) and the exchange rate constants were estimated by utilizing this solution and applying a least square approach to the measured data as implemented in Octave-Forge (http://octave.sourceforge.net).

Results
Ilv2 presequence processing is altered by C-terminal tagging Potential presequence cleavage sites are predicted between aa 37 and 38 by the program TargetP [14], between aa 55 and 56 by MitoProt [15], and a study involving the whole yeast mitochondrial proteome places the main cleavage site between aa 22 and 23 [10] (Fig. 1A). To see which of these cleavage sites are used in our yeast strain, we compared the mobility of Ilv2 from yeast cell extracts on SDS/PAGE gels with the mobility of N-terminally truncated Ilv2 marker proteins expressed in Escherichia coli. One main Ilv2 band was detected that migrated slightly faster than the Ilv2Δ55 variant ( Fig. 1B), which places the mature N-terminus a few amino acids downstream from the predicted 55↓56 site. In addition, a faint band corresponding to the Ilv2 precursor was detected.
We also performed experiments with a C-terminally 3HA-tagged Ilv2 variant. Strains expressing tagged Ilv2 grew like wild-type on media lacking isoleucine, leucine, and valine proving that the protein is functional. Unexpectedly, a different banding pattern was observed with the tagged protein. In contrast to native Ilv2, two bands (designated A and B) and a faint precursor band were observed (Fig. 1C). The mobility of the A-band was close to the mobility of the Δ37 and Δ55 forms. The resolution of our gel system is not good enough to clearly assign the A-band to either one of these two forms. But, the A-form clearly runs faster than the Δ22 variant. Thus, the mobility of the A-form is compatible with cleavage between aa 37 and 38, and between aa 55 and 56. The B-form matches most closely the Δ90 deletion.
We then investigated the contribution of MPP and AAA proteins to llv2 cleavage. Two hexameric AAA proteases are present in the inner mitochondrial membrane, one with the catalytic domain pointing toward the matrix (m-AAA) and one with the catalytic domain pointing toward the intermembrane space (i-AAA) [16]. Another AAA-protein (Mcx1) is the yeast homologue of bacterial ClpX that assists the ClpP protease in substrate binding and degradation [17]. Since MPP activity is essential in yeast, a conditional temperature sensitive mas1 mutant [7] was utilized to assess the role of MPP in Ilv2 processing.
With native Ilv2, deletion of m-AAA (Δyta10) and i-AAA (Δyme1) subunits led to an accumulation of the Ilv2 precursor band ( Fig. 2A) pointing to a role of these proteins in Ilv2 translocation or processing. Deletion of MCX1 and the mas1 mutant at permissive temperature (25°C) had no discernible effect on the Ilv2 banding pattern. Deletion of YTA10 and YME1 also affected processing of Ilv2-3HA (Fig. 2B). Similar to native Ilv2, an accumulation of the precursor band was observed. An additional phenotype was observed for the YTA10 deletion and the mas1 mutant (grown at permissive temperature). In both cases, the B-band was barely detectable indicating that the A to B conversion is compromised in these mutants. The A to B ratio in Δyme1 and Δmcx1 resembled wild-type.

Kinetics of Ilv2 processing
We wanted to know whether the two Ilv2-3HA forms result from independent or sequential cleavage events. To this end, the turnover of the various Ilv2 forms was examined by cycloheximide (CHX)-chase. The mature form of native Ilv2 proved to be fairly stable. No significant turnover was observed within 100 min (Fig. 3A). Ilv2-3HA, however, behaved differently. At t 0 , the precursor band was barely visible and disappeared quickly during the chase period, the A-form decreased steadily with time, while the amount of the B-form stayed more or less constant (Fig. 3B). To determine the half-lives of these two Ilv2 forms, we set The simulations showed a perfect match to the experimental data (Fig. 3C). We therefore conclude that the Ilv2-3HA cleavages occur sequentially. The A-form    (Table 2).
Next, the turnover of the Ilv2-3HA forms was examined in the mas1 and in the AAA-protein mutants by CHX-chase (Fig. 4A). In the mas1 mutant at permissive temperature (25°C) at t 0 , a clear precursor band (PP), a strong A-band signal, and only a very faint B-band signal were observed as described above. The deviation from the wild-type pattern suggests that MPP activity is already compromised at permissive temperature in the mas1 mutant. After shifting to nonpermissive temperature (37°C), the precursor band was completely stable (Fig. 4B). Since PP to A conversion is completely blocked and since the amount of B-form is very low in the mutant, MPP seems to be involved in both PP?A and A?B cleavage events.
The YTA10 deletion mutant showed a similar pattern as the mas1 mutant at permissive temperature (Fig. 4C). The A-form was strongly stabilized (s = 121 min), while the turnover of the B-form was   not affected (s = 16 min) ( Table 2). An interesting pattern was observed for the Δmcx1 mutant, here both forms were stabilized (s = 138 min and s = 102 min) (Fig. 4D). Since both forms are equally stabilized, the ratio of the two forms is similar to the wild-type ratio, which explains why no clear difference to wild-type in the banding pattern could be observed under steady state conditions (Fig. 2B).
MPP activity is limiting in the Δyme1 mutant All our kinetic data obtained so far were compatible with our sequential cleavage model. However, when we examined the kinetics of Ilv2 processing in the Δyme1 mutant, a deviation from the sequential cleavage model was observed (Fig. 5A). At earlier time points during the CHX-chase, there was 'too much B-form'. The experimental data could not be fitted to our sequential cleavage model (Fig. 5A, middle panel). The simplest interpretation for the occurrence of this extra amount of B-form in the Δyme1 mutant is that part of the Ilv2 precursor is directly converted to the B-form. If we incorporate this assumption into our kinetic model, the data can be perfectly fitted (Fig. 5A, lower panel). From this model, the halflives of the Ilv2 forms were calculated. It turned out that both A to B conversion (s = 116 min) and B degradation (s = 71 min) are slowed down in Δyme1. This again explains why the ratio of A-and B-forms is not so much different from the wild-type. (Fig. 2B). One possible reason for the direct PP to B cleavage could be that MPP activity is limiting in the Δyme1 mutant. Thus, PP to A conversion may occur less efficiently increasing the likelihood of direct B cleavage. In line with this interpretation, we also observed accumulation of some precursor protein in Δyme1. To test this assumption, the MPP subunits, Mas1 and Mas2 were overexpressed from a multicopy plasmid in the Δyme1 mutant. By this means, the altered kinetics of Ilv2 processing could be completely reverted to a wild-type pattern (Fig. 5C, Table 2). This finding indeed suggests that limiting MPP activity is responsible for the altered kinetics of Ilv2 processing in Δyme1.

Discussion
Here, we examined the processing of the Ilv2 presequence and uncovered a role for AAA-proteins in translocation/processing of Ilv2.
Mature Ilv2 migrates slightly faster on SDS gels than a truncated variant missing the first 55 amino acids. This suggests that either cleavage occurs downstream from the 55↓56 site predicted by MitoProt or that initial cleavage by MPP is followed by an additional proteolytic event. Direct cleavage by MPP between aa 55 and 90 seems unlikely because an arginine residue is usually present at the À2 position from the cleavage site ('R-2 rule') [5], and there is no arginine residue present in this region.
C-terminal tagging alters the Ilv2 presequence cleavage pattern. In contrast to native Ilv2, two cleavage products (A and B) were observed for C-terminally 3HA-tagged Ilv2. The A-band could not be unequivocally assigned to either one of the two predicted MPP cleavage sites (37↓38 and 55↓56), since the corresponding truncated Ilv2 forms migrate very close together on SDS gels. The mobility of the A-band is therefore compatible with cleavage at both predicted MPP cleavage sites. The mobility of the B-band is compatible with cleavage at another potential MPP cleavage site between aa 90 and 91 (arginine at position À2). The homology to E. coli acetohydroxyacid synthase starts at position 91. Cleavage occurs in a sequential manner. MPP seems to be involved in both cleavage events, since PP to A conversion is completely blocked in mas1 at nonpermissive temperature and the B-band is barely detectable already at permissive temperature. Why tagging interferes with Ilv2 processing is not clear. We could offer the following interpretation: Tagging slows down Ilv2 translocation. B cleavage occurs only upon prolonged association of the precursor protein with the translocation channel. If translocation is fast, cleavage occurs at the 55↓56 site followed by an additional proteolytic event giving rise to the native Ilv2 form. The cleavage pattern would thus reflect the speed of translocation.
In contrast to native Ilv2, Ilv2-3HA is an unstable protein. The higher turnover could be the result of the altered presequence cleavage. For the subunit IV of cytochrome c oxidase (Cox4), it has been shown that differential processing of the preprotein affects its halflife [9]. Cox4 is subjected to two-step processing by MPP and MIP. If the second cleavage by MIP is prevented, the protein becomes unstable. Furthermore, evidence has been presented that the N-terminal amino acid generated by cleavage of the preprotein may determine the half-life of the cleaved protein [10]. Differential cleavage of the presequence could therefore be a mechanism to regulate the turnover of key metabolic enzymes. The Ilv2-3HA B-form is strongly stabilized in the Δmcx1 mutant. This suggests that the ClpX homologue Mcx1 assists in the degradation of Ilv2-3HA.
Ilv2 presequence processing is affected by loss of m-AAA protease function. For native Ilv2 as well as Ilv2-3HA, we observed an accumulation of the precursor band. Also we found that A to B conversion is slowed down in case of Ilv2-3HA, whereas degradation of the B-band seemed to be unaffected. The simplest interpretation of these findings is that the Ilv2-3HA presequence is a substrate for m-AAA cleavage. Alternatively, m-AAA could be involved in Ilv2-3HA translocation, e.g. by keeping the Ilv2-3HA precursor in an unfolded, translocation competent conformation.
Although m-AAA is mainly known as a protease that degrades a large number on non-native membrane proteins, it also appears to have specific regulatory functions. It is required for the proteolytic activation of the mitochondrial ribosomal protein MrpL32, a crucial step in ribosome assembly [18]. It also plays a role in the two-step processing of cytochrome c peroxidase (Ccp1), a protein that is targeted to the intermembrane space (IMS) [19].
We found that the i-AAA protease is also involved in A to B conversion (and in B degradation), but the mechanism appears to be different. In the i-AAA mutant Δyme1, the Ilv2-3HA preprotein processing is no longer purely sequential. Instead, we also observed direct precursor to B cleavage. This altered kinetics and the stabilization of A-and B-forms in the mutant could be completely suppressed by the overexpression of MPP. From this, we conclude that MPP activity is limiting in the Δyme1 mutant.
An active role of Yme1 in translocation of mitochondrial proteins has already been suggested by a previous study [20]. There, it has been shown that Yme1 is required for the targeting of mammalian polynucleotide phosphorylase (PNPase) to the IMS in yeast. In line with our observations, it was also noted that in Δyme1, PNPase was no longer processed by MPP (which occurred normally in a wild-type strain). Basically, there are two possibilities, why Δyme1 is suppressed by the overexpression of MPP. Either Yme1 is required for full activity of MPP, or MPP as well plays an active role in translocation, and overexpression of MPP compensates for a reduction in translocation efficiency due to loss of Yme1.
Our data are summarized in the model shown in Fig. 6. It is not yet clear why the C-terminal tag influences the N-terminal processing, nor whether the effect is specific for a particular tag and/or for a single matrix protein. It remains to be seen whether the involvement in N-terminal processing by the enzymes identified in the present work is widespread.