IntroductionThe phenomenon of dual localization of a single translation productin cells has been the subject of a significant number of reports inrecent years (reviewed in Karniely and Pines, 2005; Regev-Rudzkiet al., 2005). Yet, understanding the various mechanisms thatdetermine the dual localization (distribution) of these single-translation products is far from being achieved. Two knowndistributed proteins are the tricarboxylic acid (TCA) cycle enzymesaconitase and fumarase. Both proteins are single-translationproducts, and are distributed in yeast between the cytosol andmitochondria in extremely different patterns. Whereas the cytosolicpresence of fumarase is obvious – more than 50% of the protein’smolecules are localized in the cytosol (Sass et al., 2001), thecytosolic amount of aconitase is minute (less than 5%) – a situationthat has been termed ‘eclipsed distribution’ (Regev-Rudzki et al.,2005; Regev-Rudzki and Pines, 2007; Shlevin et al., 2007).Aconitase [citrate (isocitrate) hydroxylase] employs a [4Fe-4S]cluster to catalyze the stereospecific dehydration-rehydration ofcitrate to isocitrate via cis-aconitate in both: the TCA cycle of themitochondria and in the glyoxylate shunt in the cytosol (Beinert etal., 1996; Regev-Rudzki et al., 2005). The glyoxylate shunt enablesplant and fungal cells to make four-carbon organic compounds outof two-carbon compounds, thereby allowing them to grow onacetate, ethanol or oleate as their carbon source. The enzymefumarase (fumarate hydratase) catalyzes the conversion of L-malicacid to fumaric acid. The cytosolic fumarase was suggested toparticipate in the urea cycle and in metabolism of amino acids (aa)(Ratner et al., 1953).

Studies on the distribution mechanism, particularly of fumarase,reveal that the protein is partially imported into mitochondria so

that the N-terminal mitochondrial targeting sequence (MTS) iscleaved by the mitochondrial processing peptidase (MPP). Onlyafter cleavage of the MTS in the mitochondrial matrix, does asubpopulation of the molecules move back by retrogradetranslocation into the cytosol. This results in dual localization ofidentically processed proteins in two separate locations:mitochondria and cytosol. For fumarase, the driving force for thisretrograde movement, which enables cytosolic localization ofprocessed (mature) protein, appears to be the folding of the matureprotein. Such folding impedes fumarase anterograde movement andimport (Knox et al., 1998; Stein et al., 1994). Indeed, throughoutthe coding sequence of fumarase, mutations that alter itsconformation do not impair targeting to mitochondria but do causethe loss of the retrograde movement. Thus, the distribution of theprotein is impaired and fumarase loses its cytosolic localization (Sasset al., 2003). Two additional lines of evidences support the notionthat folding is the driving force of the retrograde movement: (1)cytosolic and mitochondrial isoenzymes of fumarase are, on thebasis of mass spectrometry analysis, identical and posses noposttranslational modifications that correspond to the molecularweight of the processed protein (Sass et al., 2001; Sass et al., 2003)and, (2) the level of Hsp70 chaperones in the yeast cytosol andmitochondria affect the subcellular distribution balance (Sass et al.,2003).

Given that folding of fumarase is the driving force for itsdistribution and that specific MTS cleavage occurs beforedistribution, it would seem that the MTS is not necessary foracquiring ‘distribution ability’. Indeed, the MTS itself does notappear to confer ‘distribution capability’ on a passenger protein(Eram Blachinski, Processing of the single translation product of

2423Research Article

Dual localization of proteins in the cell has appeared in recentyears to be a more abundant phenomenon than previouslyreported. One of the mechanisms by which a single translationproduct is distributed between two compartments, involvesretrograde movement of a subset of processed molecules backthrough the organelle-membrane. Here, we investigated thespecific contribution of the mitochondrial targeting sequence(MTS), as a cis element, in the distribution of two proteins,aconitase and fumarase. Whereas the cytosolic presence offumarase is obvious, the cytosolic amount of aconitase is minute.Therefore, we created (1) MTS-exchange mutants, exchangingthe MTS of aconitase and fumarase with each other as well aswith those of other proteins and, (2) a set of single mutations,

limited to the MTS of these proteins. Distribution of bothproteins is affected by mutations, a fact particularly evident foraconitase, which displays extraordinary amounts of processedprotein in the cytosol. Thus, we show for the first time, that theMTS has an additional role beyond targeting: it determines thelevel of retrograde movement of proteins back into the cytosol.Our results suggest that the translocation rate and folding ofproteins during import into mitochondria determines the extentto which molecules are withdrawn back into the cytosol.

Key words: Dual-localization, Distribution, Mitochondrial-targeting-sequence, Fumarase, Aconitase

Summary

The mitochondrial targeting sequence tilts thebalance between mitochondrial and cytosolic duallocalizationNeta Regev-Rudzki*, Ohad Yogev* and Ophry Pines‡

Department of Molecular Biology, Hebrew University Medical School, Jerusalem 91120, Israel*These authors contributed equally to this work‡Author for correspondence (e-mail: ophryp@ekmd.huji.ac.il)

Accepted 28 April 2008Journal of Cell Science 121, 2423-2431 Published by The Company of Biologists 2008doi:10.1242/jcs.029207

Jour

nal o

f Cel

l Sci

ence

2424

the FUM1 gene (fumarase) and its subcellular distribution in baker’syeast. PhD thesis, Hebrew University, 2001; Karniely et al., 2006;Ratner et al., 1953; Sass et al., 2003). Nevertheless, we askedwhether the MTS has a role in this retrograde driven dualdistribution, a question that has not been previously addressed.

The MTSs of most mitochondrial preproteins have a typical sizeof about 20-50 aa residues, which usually reside within the N-terminus (Rapaport, 2003). These targeting sequences have twogeneral features: first, enrichment in basic, hydrophobic andhydroxylated aa and, second, the ability to form an amphiphylicα-helix (Claros et al., 1997). A number of parameters are used toevaluate the characteristics of the MTS: (1) the hydrophobicmoment (μH), which is used as a measure of the helicalamphiphilicity and the asymmetry of the distribution ofhydrophobic side chains, (2) the maximal hydrophobicity of thehydrophobic face of the helical structure (Hmax) and, (3) the numberof positively charged residues within the N-terminus (von Heijne,1986).

The information within these sequences enables the sequentialrecognition of the proteins by the general import machinery, theouter-membrane TOM complex and the inner-membrane TIM23complex which in concert with the translocation motor, leads totranslocation through the inner membrane into the mitochondrialmatrix (Neupert, 1997). During translocation, precursor proteins arefully unfolded and threaded through the import channel (Eilers andSchatz, 1986). Nevertheless, targeting signals appear to contain moreinformation and roles than simply targeting (Hegde and Bernstein,2006). Variations of signal sequences that target proteins to theendoplasmic reticulum (ER) interact differently either with thetranslocon (Kim et al., 2002; Plath et al., 1998) or with the cleavagemachinery (Kurys et al., 2000). It has been suggested that signalsequences regulate the timing of cleavage as a means of controllingprotein folding, protein modification and the translocation acrossthe ER membrane (Hegde and Bernstein, 2006).

Journal of Cell Science 121 (14)

Here, we examined the effect of the MTS on subcellulardistribution of aconitase and fumarase. We found that exchange ofsignals and substitution-mutations within the MTS cause changesin the pattern of distribution of both proteins; By introducing specificpoint mutations limited only to the MTS and that do not impair (1)the targeting of the protein to mitochondria and (2) the processingof the protein in the mitochondrial matrix, we can, nevertheless,significantly remodel the typical distribution patterns. This is thefirst time that the MTS has been shown to have a specific role indetermining the balance of (retrograde driven) dual localization, afunction separate from the targeting function of this sequence.

ResultsAconitase- and fumarase-exchange MTS mutants display analtered subcellular distribution patternThe parameters that allow mitochondrial single-translation productsto distribute into additional subcellular localizations remain largelyunexplored. Moreover, dual targeted mitochondrial proteins exhibita wide variety of distribution mechanisms and patterns. Asmentioned above, aconitase and fumarase demonstrate a significantdifference in their distribution pattern. Fumarase has a significantcytosolic presence, whereas the cytosolic amount of aconitase isminute (Regev-Rudzki et al., 2005; Sass et al., 2001; Shlevin et al.,2007).

To investigate the significance of the MTS in the retrograde-driven distribution of these proteins (both are single-translationproducts), we exchanged the native MTS of these proteins with theMTS of other mitochondrial proteins and then analyzed thedistribution patterns mutant proteins. We generated: (1) constructsin which the full MTS of aconitase was replaced with the full MTSof fumarase and visa versa (Table 1, mutants A and C) and, (2)chimeric constructs where the MTS of fumarase was replaced bythe MTS of other mitochondrial proteins, such as Hsp60 (Table 1,mutant B), Nif3, MDH1 or Cytb2 (not shown). Initially, we chose

Table 1. Fumarase- and aconitase-MTS-exchange-mutants

Plasmid Mutant protein MPII Score† Hmax‡ μH§ MTS sequence (3� to 5�) Distribution Mito vs Cyt¶

WT-FUM1 WT-Fum1 0.9265 4.56 8.38 MLRFTNCSCKTFVKSSYKLNIRRM*NSSFRT 30:70A AcoMTS-Fum1 0.9964 3.02 5.02 MLSARSAIKRPIVRGLATVSSFRT 40:60B Hsp60MTS-Fum1 0.9995 4.96 8.53 MLRSSVVRSRATLRPLLRRAYSSSFRT 58:42WT-ACO1 WT-Aco1 0.9950 5.01 6.23 MLSARSAIKRPIVRGL*ATV 94:6C FumMTS-Aco1 0.9653 4.56 8.3 MLRFTNCSCKTFVKSSYKLNIRRMNTV 50:50

†MitoProtII score (Claros et al., 1997). ‡Maximal hydrophobicity of the hydrophobic face of the helical structure (von Heijne, 1986). §Hydrophobic moment(von Heijne, 1986). ¶Percentage of mitochondrial and cytosolic fractions. Modifications of the amino acid sequences are in bold letters.

*, MPP cleavage site.

Table 2. Aconitase-MTS point mutants

Plasmid Mutant protein MPII Score† Hmax‡ μH§ MTS sequence (3� to 5�) Distribution Mito vs Cyt¶

WT-ACO1 WT-Aco1 0.9950 5.01 6.23 MLSARSAIKRPIVRGL*ATV 94:6E MTS4Aco1 0.9968 1.86 4.80 MLSRRSAIKRPIVRGLATV 45:55F MTS5Aco1 0.9933 5.82 7.93 MLSASSAIKRPIVRGLATV 92:8G MTS8Aco1 0.9879 1.1 2.46 MLSARSARKRPIVRGLATV 60:40H MTS9Aco1 0.9957 5.01 6.47 MLSARSAICRPIVRGLATV 63:37I MTS456Aco1 0.9900 5.95 8.04 MLSTAYAIKRPIVRGLATV 58:42J MTS910Aco1 0.8963 5.01 2.54 MLSARSAIPIPIVRGLATV 46:54K MTS48Aco1 0.9934 –2.05 3.04 MLSRRSARKRPIVRGLATV –L MTS91014Aco1 0.3291 5.01 1.37 MLSARSAIPIPIVCGLATV –

†MitoProtII score (Claros et al., 1997). ‡Maximal hydrophobicity of the hydrophobic face of the helical structure (von Heijne, 1986). §Hydrophobic moment(von Heijne, 1986). ¶Percentage of mitochondrial and cytosolic fractions. Modifications of the amino acid sequences are in bold letters.

*, MPP cleavage site.

Jour

nal o

f Cel

l Sci

ence

2425The MTS tilts the balance of dual localization

to exchange the MTS of fumarase, a protein with a substantialcytosolic presence, with the MTSs of proteins with eclipsed or nocytosolic presence. The notion was to see whether one can conveythe distribution pattern of one protein on the other by exchangingMTSs.

Since our objective was to examine the effect of the MTS onretrograde-driven dual targeting, two prerequisites were appliedbefore further analysis of the hybrid constructs (1) activity in vivoas verified by complementation of a yeast chromosomal deletionstrain and, (2) full import and processing by MPP in vivo. As shownin Fig. 1A, hybrid proteins A, B and C can significantly complementthe respective Aco1 or Fum1 chromosomal knockout strains (Δaco1and Δfum1). These strains express the respective MTS-hybridproteins and grow on galactose or on non-fermentable medium, suchas ethanol-acetate medium, whereas the control Δaco1 or Δfum1strains do not grow on these same medium. Not all the MTS-exchange hybrids exhibited activity. Nif3MTS-Fum1, Cytb2MTS-Fum1 and Mdh1MTS-Fum1 did not support growth on galactoseor on ethanol-acetate (data not shown). The active MTS-exchangemutants were imported and processed efficiently as detected bylabeling experiments in the presence (+) and absence (–) of carbonylcyanide m-chlorophenylhydrazone (CCCP), which dissipatesmembrane potential and blocks import. As shown in Fig. 1B, inthe presence of CCCP the active MTS-exchange mutants appear ashigher molecular weight bands corresponding to the precursors (p)of these proteins. In the absence of CCCP, the detected matureproteins (m), are the same size as the corresponding wild-type-processed proteins (Fig. 1B), indicating that these hybrid proteinsare fully processed. Worth mentioning is that the non-active mutantswe examined appeared unprocessed, as detected by such metaboliclabeling experiments (Nif3MTS-Fum1 is shown as an example inFig. 1B).

The three active MTS-exchange mutants were subjected tosubcellular fractionation experiments. The quality of eachfractionation was monitored with antibodies against mitochondrial(Hsp60) and cytosolic (hexokinase 1, HK) markers (Fig. 1C).Exchange of the aconitase-MTS for the fumarase-MTS, incurs ahighly significant increase in the cytosolic aconitase fraction (from6% to 50% of the molecules) (mutant C, Fig. 1C and Table 1).Replacing the fumarase MTS with the MTSs of aconitase or Hsp60caused a small (but reproducible) change in the fumarase-hybriddistribution pattern (mutants A and B, Fig. 1C); a lesser (comparedwith wild type) amount of fumarase was detected in the cytosol(60% and 42%, respectively, versus 70% in the wild type) (Table1 and Fig. 1C).

Single mutations in the aconitase MTS cause a major shift tothe cytosolic fractionTo further assess the impact of the MTS on protein distribution inthe cell, we introduced point mutations within the MTSs ofaconitase and fumarase, and analyzed their distribution patterns.As pointed out above, the MTS has two general features: enrichmentof specific aa and the ability to form amphiphilic α-helices (vonHeijne, 1986). Hence, we picked several candidate aa that, on onehand might have an effect on the MTS but on the other would notabolish the targeting capacity of the signal. In other words, we werein search of additional information within the MTS that can affectdistribution but not targeting. In this regard, all the mutant proteinsare still predicted with high probability by the MitoProtII programto be targeted to mitochondria [http://ihg.gsf.de/ihg/mitoprot.html(Claros and Vincens, 1996)] (Table 2). We mainly focused on

Fig. 1. MTS-exchange mutants. (A) Growth and complementation of aconitaseand fumarase-MTS-exchange mutants. Δaco1 or Δfum1 strains harboring theindicated plasmids that encode MTS-exchange mutants were diluted andgrown on galactose or ethanol-acetate medium plates as indicated (B)Aconitase- and fumarase-MTS-exchange mutants are processed. Wild-type,Δaco1 and Δfum1 strains induced for expression the indicated plasmids werelabeled with [35S]methionine for 30 minutes, either in the absence (–) orpresence (+) of 20 μM CCCP. Total cell extracts were prepared,immunoprecipitated with the indicated antiserum and analyzed using SDS-PAGE. Arrows show positions of precursor (p) and mature proteins (m).(C) MTS-exchange mutants exhibit alterations in the protein subcellulardistribution. Yeast cells expressing aconitase and fumarase variants weresubjected to subcellular fractionation. Equivalent portions from the total (Tot)cytosolic (Cyt) and mitochondrial (Mit) fractions were analyzed by westernblotting using antibodies against the indicated proteins. Representativeexamples of mitochondrial (Hsp60) and cytosolic (hexokinase1, HK) controlsare shown. Cytosolic and mitochondrial band intensities were quantifieddensitometrically using TINA Software.

Jour

nal o

f Cel

l Sci

ence

2426

exchanging the positively charged aa residues arginine or lysine;either removing or adding these within the aconitase and fumaraseMTS aa sequences.

We created a total of ten constructs of MTS point mutations inaconitase and three MTS point mutations in fumarase (see below).Tables 2 and 3 display aconitase-MTS mutants. MutantsMTS4Aco1, MTS5Aco1, MTS8Aco1 and MTS9Aco1 (Table 2;mutants E, F, G, H) contain substitution mutations of aa residues4, 5, 8, 9, respectively. Mutants I and J contain substitutionmutations of three or two aa (residues 4, 5, 6 and 9, 10, respectively;Table 2). Mutants MTS48Aco1 (K) and MTS91014Aco1 (L)contain substitution mutation of two and three aa (residues 4, 8 and9, 10, 14, respectively; Table 2). In essence, in mutants E, G andK we substituted hydrophobic alanine or isolucinea with basicarginine. In the aconitase-MTS mutants F, H, I, J and L weexchanged a basic for a non-charged aa.

We constructed two additional aconitase mutants in which theMTS was lengthened (Table 3) by repeating the last nine aa (9-17)of the aconitase-MTS (MTS9aaAco-Aco1, mutant O) or by addingthe last fumarase-MTS 9 aa (MTS9aaFum-Aco1, mutant N). Again,all the aconitase-MTS point mutants were subjected to the twoprerequisites: enzymatic activity in vivo and full processing in

Journal of Cell Science 121 (14)

mitochondria. As illustrated in Fig. 2A and Fig. 3A, the MTS-mutantproteins (except mutants K and L, see below) can complement aΔaco1 chromosomal knockout strain.

To examine whether the mutant proteins are processed, yeast cellsexpressing them were labeled in the presence or absence of CCCP.As shown in Fig. 2B the mutants (except mutants K and L) areprocessed efficiently as deduced from the apparent size shift(compare p with m) of the proteins in the presence (+) or absence(–) of CCCP. These experiments indicate that, although the nativeMTS sequences were modified, the N-termini of these proteins werestill imported into mitochondria and processed by the MPP. Twomutants MTS48Aco1 (K) and MTS91014Aco1 (L), in which theamphiphilic helix was most probably excessively damaged, werenot imported and processed (Fig. 2B). Accordingly, these mutantsdid not fully complement the aconitase cytosolic activity in a Δaco1strain and do not support growth on EtOH-acetate plates (Fig. 2A)(Regev-Rudzki et al., 2005).

To investigate the distribution patterns of aconitase-MTS pointmutants, subcellular fractionation experiments were employed. Fiveout of the six active and processed aconitase-MTS mutantsexpressed in a Δaco1 background displayed significantly differentdistribution patterns. Mutants E, G, H, I and J exhibited 37-55

Fig. 2. Aconitase-MTS point mutations. (A) Growth and complementation of aconitase-MTS point mutants. Δaco1 strain harboring the indicated plasmidsencoding MTS point mutants were diluted and grown on galactose or ethanol-acetate medium plates as indicated. (B) Aconitase-MTS point mutants are processed.MTS point mutants of aconitase were labeled and examined for processing as described in the legend to Fig. 1B. (C) MTS point mutants exhibit alterations in theprotein subcellular distribution. Yeast cells expressing aconitase variants were subjected to subcellular fractionation. Equivalent portions from the total (Tot)cytosolic (Cyt) and mitochondrial (Mit) fractions were analyzed by western blotting using antibodies against the indicated proteins.

Jour

nal o

f Cel

l Sci

ence

2427The MTS tilts the balance of dual localization

percent of their aconitase isoenzyme in the cytosol compared toabout 6% for the wild type (Fig. 2C and Table 2). Thus, thesemutations abolish the natural eclipsed distribution of aconitasewithout impairing processing by MPP, and cause a shift of asignificant portion of the molecules from mitochondria to thecytosol. Mutant MTS5Aco1 (F) exhibited a distribution patternsimilar to that of the wild type protein with more than 90% of theprotein localized to mitochondria (Table 2, Fig. 2C).

Of interest are the constructs in which the aconitase MTS waselongated with wild-type sequences of the aconitase or the fumaraseMTS. These constructs (mutants N and O, Table 3) were active andprocessed (Fig. 3A,B) and behaved similarly to most of the pointmutations in that more protein is diverted to the cytosol (Fig. 3C).Thus, it is not simply the sequence of the MTS but also, maybe,its compatibility with the mature portion of the protein and/orpossibly interaction with trans-acting elements.

Single mutations in the fumarase MTS cause a minor shift tothe mitochondrial fractionA limited but similar analysis to that described in the previoussection was carried out for fumarase. Table 4 displays fumarase-MTS mutants; MTS3Fum1, MTS5Fum1 and MTS9Fum1 whichare substitution mutants of aa 3, 5 and 9, respectively, in the fumaraseMTS (Table 4, mutants R, S and T, respectively). In MTS5Fum1(S) and MTS9Fum1 (T) a basic arginine replaces a non-chargedaa, whereas in mutant MTS3Fum1 (R) a non-charged aa was addedinstead of a positively charged one.

These fumarase-MTS point mutants were active similarly to thewild-type protein and can fully complement a Δfum1 chromosomalknockout strain (Fig. 4A). As shown in Fig. 4B fumarase-MTSmutants are also processed efficiently, similar to the wild-typeprotein; as deduced from the apparent size shift of the proteins inthe presence (+) or absence (–) of CCCP. Fractionation of Δfum1cells expressing the mutants MTS3Fum1, MTS5Fum1 andMTS9Fum1 (Table 4; mutants R, S and T, respectively), revealeda slight but reproducible enrichment of the protein in mitochondriawhen compared with wild-type fumarase (50-58% versus 30%respectively, Fig. 4C, Table 4).

Slowed down translocation affects the subcellular distributionof aconitaseWe have previously shown that protein folding and translocationefficiency affect dual targeting of fumarase (Sass et al., 2003;Karniely et al., 2006). Prior to addressing the possibility that theMTS mutations might exert their effects through the above, wewished to rule out a more trivial possibility, which is that cytosolicaconitase and fumarase might be less stable, thereby affecting theapparent distribution of the proteins. Both fumarase and aconitaseappear to be stable, as previous studies have suggested (Karnielyet al., 2006; Regev-Rudzki et al., 2005; Sass et al., 2003; Shlevinet al., 2007; Yogev et al., 2007). Here we examined, using pulse-chase labeling (Fig. 5), the stability of aconitase and fumarase, andthe stability of their cytosolic derivatives that lack their respectiveMTS (ΔMTS-Aco1 and ΔMTS-Fum1). Both fumarase and aconitaseappear to be stable – in contrast to a non-stable aconitase fused toa degron (ΔMTS-Aco1-SL17), which is recognized and degradedby the ubiquitin-proteasome system (Shlevin et al., 2007). ThisΔMTS-Aco1-SL17 is rapidly degraded and, within 30 minutes, isreduced to less than a half of its initial amount (Fig. 5, third panel).

Table 3. Aconitase-MTS elongation mutants

Plasmid Mutant protein MPII Score† Hmax‡ μH§ MTS sequence (3� to 5�) Distribution Mito vs Cyt

WT-ACO1 WT-Aco1 0.9950 5.01 6.23 MLSARSAIKRPIVRGL*ATV 94:6N MTS9aa Fum-Aco1 0.9989 4.26 5.78 MLSARSAIKRPIVRGYKLNIRRMNTV 50:500 MTS9aa Aco-Aco1 0.9996 5.48 8.54 MLSARSAIKRPIVRGKRPIVRGLATV 40:60

†MitoProtII score (Claros et al., 1997). ‡Maximal hydrophobicity of the hydrophobic face of the helical structure (von Heijne, 1986). §Hydrophobic moment(von Heijne, 1986). ¶Percentage of mitochondrial and cytosolic fractions. Modifications of the amino acid sequences are in bold letters.

*, MPP cleavage site.

Fig. 3. Aconitase-MTS-elongated mutants. (A) Growth and complementationof aconitase-MTS elongated mutants. Δaco1 strains harboring the indicatedplasmids encoding MTS elongated mutants were diluted and grown ongalactose or ethanol-acetate medium plates as indicated. (B) Aconitaseelongated mutants are processed. MTS-elongated mutants of aconitase werelabeled and examined for processing as described in the legend to Fig. 1B.(C) Aconitase elongated mutants exhibit alterations in the protein subcellulardistribution. Subcellular fractionations of yeast cells expressing aconitaseelongated mutants. Equivalent portions from the total (Tot) cytosolic (Cyt) andmitochondrial (Mit) fractions were analyzed by western blotting usingantibodies against the indicated proteins.

Jour

nal o

f Cel

l Sci

ence

2428

Additionally, it is important to note that the MTS-mutant precursors,and mature forms of aconitase and fumarase analyzed in this studyalso appear to be stable in vivo, as can be deduced from pulse-

Journal of Cell Science 121 (14)

labeling experiments in the presence and absence of CCCP (Fig.2B, Fig. 3B and Fig. 4B). Thus, protein turnover is not likely to bethe reason for MTS-mutation effects on distribution.

The possibility that folding of the passenger protein is involvedin dual localization is referred to in the discussion section. Here,we address the possibility that the effect of the MTS-mutations ondistribution in vivo occurs through slowing down of import. Wehave recently shown that slowed down translocation can affectfumarase distribution (Yogev et al., 2007). Here, we asked whetherthe relationship between the distribution and translocation holdstrue for aconitase. We performed subcellular fractionations on strainsharboring mutations that affect the inner and outer membranetranslocases (Δmim1, tom40ts and tim23ts). These mutations havebeen recently shown to slow down translocation of fumarase underspecific growth conditions (Waizenegger et al., 2005; Yogev et al.,2007). As shown in Fig. 6A, temperature-sensitive mutations inTim23 and Tom40, and a deletion of Mim1, lead – under the semipermissive growth conditions (30°C) – to an apparent largerfraction of aconitase (up to 50% of the molecules) in the cytosol,when compared with wild-type cells (or corresponding wild-typestrains of the tom40ts, tim23ts and Δmim1 mutants; not shown).These results indicate that mutations that affect the translocationrate can in turn change the aconitase distribution pattern.

Further evidence showing that translocation rate affects theprotein distribution pattern was obtained by growth-temperatureexperiments. We have previously shown that, when yeast cells aregrown at low temperatures, the rate of fumarase translocation into

Table 4. Fumarase-MTS point mutants

Plasmids Mutant protein MPII Score† Hmax‡ μH§ MTS sequence (3� to 5�) Distribution Mito vs Cyt¶

WT-FUM1 WT-Fum1 0.9265 4.56 8.38 MLRFTNCSCKTFVKSSYKLNIRRM*N 30:70R MTS3Fum1 0.9291 4.56 6.08 MLSFTNCSCKTFVKSSYKLNIRRMN 58:42S MTS5Fum1 0.9546 2.11 5.94 MLRFRNCSCKTFVKSSYKLNIRRMN 58:42T MTS9Fum1 0.8856 3.04 7.33 MLRFTNCSKKTFVKSSYKLNIRRMN 50:50

†MitoProtII score (Claros et al., 1997). ‡Maximal hydrophobicity of the hydrophobic face of the helical structure (von Heijne, 1986). §Hydrophobic moment(von Heijne, 1986). ¶Percentage of mitochondrial and cytosolic fractions. Modifications of the amino acid sequences are in bold letters.

*, MPP cleavage site.

Fig. 4. Fumarase-MTS point mutations. (A) Growth and complementation offumarase. MTS point mutants–Δfum1 strain harboring the indicated plasmidsencoding MTS point mutants ware diluted and grown on galactose or ethanol-acetate medium plates as indicated. (B) Fumarase-MTS point mutants areprocessed. MTS point mutants of fumarase were labeled and examined forprocessing as described in the legend to Fig. 1B. (C) Aconitase-MTS pointmutants exhibit alterations in the protein subcellular distribution. Yeast cellsexpressing fumarase variants were subjected to subcellular fractionation.Equivalent portions from the total (Tot) cytosolic (Cyt) and mitochondrial(Mit) fractions were analyzed by western blotting using antibodies against theindicated proteins.

Fig. 5. Aconitase and fumarase and their cytosolic derivates are stable. Yeaststrains harboring plasmids encoding the indicated proteins were induced ingalactose medium and pulse-labeled with [35S]methionine-cysteine for 15minutes, followed by a chase (addition of unlabeled methionine-cysteine andcycloheximide) for the times indicated. Total cell extracts were prepared,immunoprecipitated with aconitase or fumarase antiserum and analyzed bySDS-PAGE and autoradiography.

Jour

nal o

f Cel

l Sci

ence

2429The MTS tilts the balance of dual localization

mitochondria is reduced (Yogev et al., 2007). If this is also true foraconitase, a reduction of the translocation rate at low temperaturesshould then affect the distribution pattern of aconitase. Wild-typeS. cerevisiae was grown at 20°C or 30°C on galactose medium andexamined by subcellular fractionation. As presented in Fig. 6B, cellsgrown at 20°C exhibit a significant presence of aconitase in thecytosol, in contrast to cells grown at 30°C.

To gain direct evidence that the aconitase-MTS mutations slowsdown the translocation process in vivo, we performed a metabolic-labeling experiment. The approach we took was first to block themitochondrial import in vivo (using CCCP) and then reverse theimport block (using DTT). Under these conditions, the fullytranslated precursor is accumulated in the cytosol and only thenis it translocated posttranslationally. This allowed us to examinethe rate of the appearance of aconitase in processed form as anindication of the translocation rate. We find that, in contrast tothe wild-type protein – for which we find that 50% of themolecules are processed within 10 minutes – the MTS mutantsMTS4Aco1 (E), MTS9Aco1 (H) and MTS910Aco1 (J) werehardly processed (Fig. 7). Taken together, these results suggestthat slowed down translocation leads to a larger cytosolic aconitasefraction.

DiscussionAt present, fumarase and aconitase serve as models for retrograde-driven dual localization (Knox et al., 1998; Regev-Rudzki et al., 2005;Regev-Rudzki and Pines, 2007). These two proteins share two

common features in their distribution mechanism: (1) a single nucleargene encodes both the mitochondrial and cytosolic forms and, (2) allthe molecules are processed in mitochondria before distribution (Knoxet al., 1998; Regev-Rudzki et al., 2005; Sass et al., 2001). In thisstudy we show, for the first time (through the effect of MTS mutations)that the MTS has a role in the dual-localization of fumarase andaconitase. For aconitase it appears to be a major cis element thatdetermines the balance between cytosolic and mitochondrial locationswhereas for fumarase it is minor. The reason for this might be therelative contribution of the MTS and the mature protein sequencesto distribution. As pointed out in the results section, we have shownthat both folding and translocation might be important factors indetermining the relative distribution of fumarase between the cytosoland mitochondria (Karniely et al., 2006; Sass et al., 2003; Yogev etal., 2007). Consistent with fumarase folding as the major cisdeterminant for distribution is the finding that mutations throughoutthe coding sequence of fumarase, that do not impair targeting tomitochondria, cause a loss in retrograde movement and distribution(Sass et al., 2003). For aconitase, however, the situation is verydifferent because mutations and deletions within the mature sequencedo not necessarily abrogate the distribution of the protein (N.R.-R.,Reut Hazan and O.P., unpublished data). Thus, the folding constituentin aconitase dual targeting is probably much lower than that forfumarase.

Traditionally, the MTS is associated with import that includesfunctions of targeting to the membrane and translocation throughmitochondrial translocases. The results presented in this study,are consistent with a post-targeting function (i.e. translocation)of the MTS in determining the protein subcellular distribution:(1) MTS mutants of aconitase and fumarase displaying an altereddistribution are active and fully processed (suggesting no problemin targeting). (2) Regardless of whether mutations were designedto strengthen or weaken the targeting ability of the MTS(according to μH, Hmax and the MitoprotII score), they turnedout to have similar effects on distribution. (3) Mutations incomponents of the mitochondrial translocation machinery(Tim23, Mim1, Tom40) or growth at low temperatures that slowsdown translocation, cause an increase in the cytosolic fractionof aconitase (this study) and fumarase (Karniely et al., 2006;Yogev et al., 2007). (4) Aconitase-MTS mutants exhibited aslower translocation rate into the mitochondria according to in

Fig. 6. (A) Translocation machinery mutant strains exhibit alterations inprotein subcellular distribution. Indicated mutant strains were grown at thepermissive temperature. The strains were subjected to subcellularfractionation. Equivalent portions from the total (Tot) cytosolic (Cyt) andmitochondrial (Mit) fractions were analyzed by western blotting usingantibodies against the indicated proteins. (B) Growth at low temperatureincreases the aconitase cytosolic presence. Wild-type yeast cells were grown at20°C or at 30°C and subjected to subcellular fractionation as described above.

Fig. 7. Translocation of aconitase-MTS mutants is slower. Yeast cellsexpressing the indicated plasmids were incubated for 1 minute with CCCPfollowed by labeling for 15 minutes with [35S]methionine (in the presence ofCCCP). DTT, an excess unlabeled methionine-cysteine and cycloheximidewere added (start of the chase). Aliquots were taken at the indicated times,immunoprecipitated with aconitase antiserum and analyzed by SDS-PAGE.Arrows show positions of precursor (p) and mature proteins (m).

Jour

nal o

f Cel

l Sci

ence

2430 Journal of Cell Science 121 (14)

vivo post-translational pulse-chase experiments. These results areconsistent with the notion that MTS affects distribution throughits effect on the translocation rate.

Our working hypothesis is that the folding and rate oftranslocation of the proteins determine the dual distribution pattern.Yet, folding and translocation rate are dependent on one another;modifying the translocation rate allows the proteins to fold and/orinteract with cytosolic factors prior to translocation, thereby causinga change in their distribution.

Finally, one can not exclude the possibility that there arealternative explanations for our observations besides affects ontranslocation rate and folding. First, the efficiency of targetingmight be subtly impaired, which in turn might affect the balanceof dual targeting (even though all the relevant mutants in this studyare fully processed). Second, the MTS mutations might affect aninteraction with a trans-acting element, such as mitochondrialHsp70, which has been shown to affect fumarase distribution(Karniely et al., 2006). Third, the cleaved MTS peptide mightaffect retrograde translocation. In this regard, an example of apost-cleavage function of a signal are cleaved ER signal peptides,which have been proposed to have a role in antigen presentationfor the MHC-class I proteins (Borrego et al., 1998; Braud et al.,1998).

We suggest that during evolution the combination of signal andmature portions of fumarase have been optimized for substantialretrograde-driven cytosolic presence. Any change in either thefumarase MTS or the mature protein sequence causes less cytosolicretention. Aconitase, however, has – during evolution – beenoptimized for eclipsed distribution and any change in its MTS causesmore cytosolic retention. How this evolutionary goal is specificallyachieved and how cis- and trans-acting elements might be involved,remains to be determined in future studies.

Materials and MethodsStrains, plasmid constructs and growth conditionsStrainsS. cerevisiae strains used were BY4741 (Mat a; his3Δ1; leu2Δ0; met15Δ0; ura3Δ0)and YPH499 (Mat a ura3;lys2;ade;trp1;his3;leu2), BY4742 (Mat α; his3Δ1; leu2Δ0;lys2Δ0; ura3Δ0; YLR304c::kanMX4), and BY4743 (Mat a/α; his3Δ1/his3Δ1;leu2Δ0/leu2Δ0; lys2Δ0/LYS2; MET15/met15Δ0; ura3Δ0/ura3Δ0;YLR304c::kanMX4/YLR304c). Δaco1 was obtained as previously described (Regev-Rudzki et al., 2005). Δfum1 was obtained as previously described (Sass et al., 2003).tim23ts strain was kindly provided by Abdussalam Azem (Department ofBiochemistry, Tel-Aviv University, Israel), Δmim1 and tom40ts strains were kindlyprovided by Doron Rapaport (Department of Biochemistry, University of Tübingen,Germany).

Plasmid constructsMutation were created using the QuickChange®II kit (Stratagene) or PCR reactionsusing the indicated oligonucleotides (Table 5).

Growth conditionsStrains harboring the appropriate plasmids were grown overnight at 30°C or 20°C(when indicated) in synthetic depleted (SD) medium containing 0.67% (w/v) yeastnitrogen base plus 2% ethanol or 3% galactose, 2% glucose, 2% acetate and (w/v),supplemented with the appropriate amino acids (50 μg/ml). For agar plates, 2% agarwas added.

Metabolic labelingCultures or induced cultures (in galactose) were harvested and labeled with 10 μCi/ml[35S]methionine and further incubated for 30 minutes at 30°C. When required, 20μM carbonyl cyanide m-chlorophenylhydrazone (CCCP) was added before labeling.Labeling was stopped by addition of 10 mM sodium azide. Labeled cells werecollected by centrifugation, resuspended in Tris/EDTA buffer pH 8.0, containing 1mM phenylmethylsulfonyl fluoride, broken with glass beads for 5 minutes, andcentrifuged to obtain the supernatant fraction. Supernatants were denatured by boilingin 1% SDS, immunoprecipitated with anti-aconitase or anti-fumarase rabbit antiserumand protein A-Sepharose (Amersham Biosciences, Piscataway, NJ), and then analyzedby SDS-PAGE.

For pulse-chase experiments, yeast cultures were incubated for 1 minute with 20μM CCCP and labeled for 15 minutes with 10 μCi/ml [35S]methionine, and then, inthe presence of 40 mM DTT, chased using unlabeled 0.03% methionine, 0.04%cysteine and 0.01% cycloheximide.

Reverse

GGGTCTCTTGATGGCAGAACGTGCAGACAGCATGATTGATACTCTTATCCG

GCCTTGGAGATCTTGGGGGCCTTAAAGTAGCGCGACTACGAACAACGGATGATCTC

AACATGATTGTATCTCTTATCCGGCCTTGGAGATCTTGGGGGGTCTCTTGATGGCAGAACGTCTAGACAGCATGGATCCGGGTCTCTTGATGGCAGAAGATGCAGACAGCATGGATCCCGAACAATGGGTCTCTTACGGGCAGAACGTGCAGACAGCCCACGAACAATGGGTCTACAGATGGCAGAACGTGCCAATGGGTCTCTTGATGGCGGTAAATCTAGACAGCATGGA

TCCACTAGGCAAGACCACGAACAATGGGTCTCTTGATGGCAGAACGT

GCAGACGGTCTGCAGGTGGAGC

GGTCTGCAGGTGGAGC

GGTCTGCAGGTGGAGC

GGTCTGCAGGTGGAGC

GTCTTGCAACTACAATTGGTAAAACTCAACATGATTGTATCTC

CGAAAGTCTTGCAACTACAATTTCTAAATCTCAACATGATTGTATCTC

GCTTATATGACGATTTTACGAAAGTCTTCTTACTACAATTGGTAAATCTCAACATG

TCACCCGGGGGTACCACCTTGTGCCA

Forward

GGGTCGACCTGCAGCGCCATCAAGAGACCCATTGTTCGTGGTCTTGCGTCCTCGT

TCAGAACTGAAACCGGGTCGACCTGCAGCCGTAGTCGCGCTACTTTAAGGCCTTATTGCGTCGTGCTTA

CTCC TCCTCGTTCAGAACTGGGATCCATGCTGTCTAGACGTTCTGCCATCAAGAGACCCGGATCCATGCTGTCTGCATCTTCTGCCATCAAGAGACCCGCTGTCTGCACGTTCTGCCCGTAAGAGACCCATTGTTCGGCACGTTCTGCCATCTGTAGACCCATTGTTCGTGGCTAGTGGATCCATGCTGTCTAGATTTACCGCCATCAAGAG

ACCCATTGGTCTGCACGTTCTGCCATCAAGAGACCCATTGTTCGTGGT

CTTGCATGCTGAAGAGACGTTCTGCCAGAAAGAGACCCATTGTT

CGTGATGCTGTCTGCACGTTCTGCCATCCCCATCCCCATTGTTTG

TGGTCTTGCGACAGTCTCGGATCCATGCTGTCTGCACGTTCTGCCATCAAGAGACCCA

TTGTTCGTGGTCTTGCGAAGAGACCCATTGTTCGGGATCCATGCTGTCTGCACGTTCTGCCATCAAGAGACCCA

TTGTTCGTGGTGCCATCAAGAGACCCAGAGATACAATCATGTTGAGTTTTACCAATTGTAGTTGCAA

GACGAGATACAATCATGTTGAGATTTAGAAATTGTAGTTGCAA

GACTTTCGCATGTTGAGATTTACCAATTGTAGTAAGAAGACTTTCGTA

AAATCGTCATATAAGCGGACAAACTTGTGCGTAGCATTACCAAGTTCTACCCTCAA

AAGTACTCCTCGTTCAGAACTGAAAC

Mutant name

AcoMTS-Fum1

Hsp60MTS-Fum1

MTS4Aco1MTS5Aco1MTS8Aco1MTS9Aco1MTS456Aco1

MTS910Aco1

MTS48Aco1

MTS91014Aco1

MTS9aaFum-Aco1

MTS9aaAco-Aco1

MTS3Fum1

MTS5Fum1

MTS9Fum1

Nif3MTSFum1

Table 5. Oligonucleotides (5�� to 3��) used in this study

Jour

nal o

f Cel

l Sci

ence

2431The MTS tilts the balance of dual localization

Subcellular fractionationInduced yeast cultures were grown to an OD600 of 1.5. Mitochondria were isolatedas described previously (Knox et al., 1998). Spheroplasts were prepared in the presenceof Zymolyase-20T (MP Biomedicals, Irvine, CA). Each of our subcellular fractionationexperiments was assayed for cross-contaminations by using anti-Hsp60 or anti-mtHsp70 antibodies as mitochondrial markers and anti-hexokinase 1 (anti-HK)antibody as a cytosolic marker. Cytosolic and mitochondrial band intensities werequantified densitometrically using TINA Software.

We thank Abdussalam Azem, Eitan Bibi, Sharon Karniely and LeeShlevin for critical reading of this manuscript. Special thanks to MeravTal and Yudit Karp for their dedicated assistance. This research wassupported by the German Israeli Foundation (GIF), the Israel ScienceFoundation (ISF) and the German Israeli Project Cooperation (DIP).

ReferencesBeinert, H., Kennedy, M. C. and Stout, C. D. (1996). Aconitase as ironminus signSulfur

protein, enzyme, and iron-regulatory protein. Chem. Rev. 96, 2335-2374.Borrego, F., Ulbrecht, M., Weiss, E. H., Coligan, J. E. and Brooks, A. G. (1998).

Recognition of human histocompatibility leukocyte antigen (HLA)-E complexed withHLA class I signal sequence-derived peptides by CD94/NKG2 confers protection fromnatural killer cell-mediated lysis. J. Exp. Med. 187, 813-818.

Braud, V. M., Allan, D. S., O’Callaghan, C. A., Soderstrom, K., D’Andrea, A., Ogg,G. S., Lazetic, S., Young, N. T., Bell, J. I., Phillips, J. H. et al. (1998). HLA-E bindsto natural killer cell receptors CD94/NKG2A, B and C. Nature 391, 795-799.

Claros, M. G. and Vincens, P. (1996). Computational method to predict mitochondriallyimported proteins and their targeting sequences. Eur. J. Biochem. 241, 779-786.

Claros, M. G., Brunak, S. and von Heijne, G. (1997). Prediction of N-terminal proteinsorting signals. Curr. Opin. Struct. Biol. 7, 394-398.

Eilers, M. and Schatz, G. (1986). Binding of a specific ligand inhibits import of a purifiedprecursor protein into mitochondria. Nature 322, 228-232.

Hegde, R. S. and Bernstein, H. D. (2006). The surprising complexity of signal sequences.Trends Biochem. Sci. 31, 563-571.

Karniely, S. and Pines, O. (2005). Single translation-dual destination: mechanisms of dualprotein targeting in eukaryotes. EMBO Rep. 6, 420-425.

Karniely, S., Regev-Rudzki, N. and Pines, O. (2006). The presequence of fumarase isexposed to the cytosol during import into mitochondria. J. Mol. Biol. 358, 396-405.

Kim, S. J., Mitra, D., Salerno, J. R. and Hegde, R. S. (2002). Signal sequences controlgating of the protein translocation channel in a substrate-specific manner. Dev. Cell 2,207-217.

Knox, C., Sass, E., Neupert, W. and Pines, O. (1998). Import into mitochondria, foldingand retrograde movement of fumarase in yeast. J. Biol. Chem. 273, 25587-25593.

Kurys, G., Tagaya, Y., Bamford, R., Hanover, J. A. and Waldmann, T. A. (2000). Thelong signal peptide isoform and its alternative processing direct the intracellulartrafficking of interleukin-15. J. Biol. Chem. 275, 30653-30659.

Neupert, W. (1997). Protein import into mitochondria. Annu. Rev. Biochem. 66, 863-917.Plath, K., Mothes, W., Wilkinson, B. M., Stirling, C. J. and Rapoport, T. A. (1998).

Signal sequence recognition in posttranslational protein transport across the yeast ERmembrane. Cell 94, 795-807.

Rapaport, D. (2003). Finding the right organelle. Targeting signals in mitochondrial outer-membrane proteins. EMBO Rep. 4, 948-952.

Ratner, S., Anslow, W. P., Jr and Petrack, B. (1953). Biosynthesis of urea. VI. Enzymaticcleavage of argininosuccinic acid to arginine and fumaric acid. J. Biol. Chem. 204, 115-125.

Regev-Rudzki, N. and Pines, O. (2007). Eclipsed distribution: a phenomenon of dualtargeting of protein and its significance. BioEssays 29, 772-782.

Regev-Rudzki, N., Karniely, S., Ben-Haim, N. N. and Pines, O. (2005). Yeast aconitasein two locations and two metabolic pathways: seeing small amounts is believing. Mol.Biol. Cell 16, 4163-4171.

Sass, E., Blachinsky, E., Karniely, S. and Pines, O. (2001). Mitochondrial and cytosolicisoforms of yeast fumarase are derivatives of a single translation product and haveidentical amino termini. J. Biol. Chem. 276, 46111-46117.

Sass, E., Karniely, S. and Pines, O. (2003). Folding of fumarase during mitochondrialimport determines its dual targeting in yeast. J. Biol. Chem. 278, 45109-45116.

Shlevin, L., Regev-Rudzki, N., Karniely, S. and Pines, O. (2007). Location-specificdepletion of a dual-localized protein. Traffic 8, 169-176.

Stein, I., Peleg, Y., Even-Ram, S. and Pines, O. (1994). The single translation productof the FUM1 gene (fumarase) is processed in mitochondria before being distributedbetween the cytosol and mitochondria in Saccharomyces cerevisiae. Mol. Cell. Biol. 14,4770-4778.

von Heijne, G. (1986). Mitochondrial targeting sequences may form amphiphilic helices.EMBO J. 5, 1335-1342.

Waizenegger, T., Schmitt, S., Zivkovic, J., Neupert, W. and Rapaport, D. (2005). Mim1,a protein required for the assembly of the TOM complex of mitochondria. EMBO Rep.6, 57-62.

Yogev, O., Karniely, S. and Pines, O. (2007). Translation coupled translocation of yeastfumarase into mitochondria in vivo. J. Biol. Chem. 282, 29222-29229.

Jour

nal o

f Cel

l Sci

ence