Dual localization of fumarase is dependent on the integrity of the glyoxylate shunt


*E-mail ophryp@ekmd.huji.ac.il; Tel. (+972) 2 6757203; Fax (+972) 2 6757260.


Fumarase and aconitase in yeast are dual localized to the cytosol and mitochondria by a similar targeting mechanism. These two tricarboxylic acid cycle enzymes are single translation products that are targeted to and processed by mitochondrial processing peptidase in mitochondria prior to distribution. The mechanism includes reverse translocation of a subset of processed molecules back into the cytosol. Here, we show that either depletion or overexpression of Cit2 (cytosolic citrate synthase) causes the vast majority of fumarase to be fully imported into mitochondria with a tiny amount or no fumarase in the cytosol. Normal dual distribution of fumarase (similar amounts in the cytosol and mitochondria) depends on an enzymatically active Cit2. Glyoxylate shunt deletion mutations (Δmls1, Δaco1 and Δicl1) exhibit an altered fumarase dual distribution (like in Δcit2). Finally, when succinic acid, a product of the glyoxylate shunt, is added to the growth medium, fumarase dual distribution is altered such that there are lower levels of fumarase in the cytosol. This study suggests that the cytosolic localization of a distributed mitochondrial protein is governed by intracellular metabolite cues. Specifically, we suggest that metabolites of the glyoxylate shunt act as ‘nanosensors’ for fumarase subcellular targeting and distribution. The possible mechanisms involved are discussed.


The localization of a protein to a specific subcellular compartment is one of the most fundamental processes of a living cell. In eukaryotic cells, each subcellular compartment is characterized by a unique composition of proteins enabling its biological functions. Nevertheless, some proteins are localized to more than one subcellular compartment and consequently participates in different biochemical pathways and might have completely different functions (Mackenzie, 2005). One of the mechanisms by which a single translation product is distributed between two compartments involves retrograde movement of a subset of processed molecules back through the organelle-membrane (Karniely and Pines, 2005; Regev-Rudzki and Pines, 2007). Two known distributed proteins are the tricarboxylic acid (TCA) cycle enzymes: aconitase and fumarase. These two proteins are single translation products that are distributed in yeast between the cytosol and mitochondria in very different patterns. While the cytosolic presence of fumarase is obvious (more than 50% of the protein's molecules are localized in the cytosol; Sass et al., 2001), the cytosolic amount of aconitase is very small (less than 5%; Regev-Rudzki et al., 2005; Regev-Rudzki et al., 2008). In this respect, during the past few years, intensive efforts have been invested in identifying intracellular elements that affect protein targeting and distribution between organized compartments.

Citrate synthase (CS) is a key enzyme in cellular metabolism and is present in virtually all cells capable of oxidative metabolism (Velot et al., 1999; Chen et al., 2002). In mammalian cells, the CS reaction seems to be exclusively located in mitochondria. In yeast and plant cells, additional CS enzymatic activities can be found outside mitochondria and are involved in the glyoxylate shunt, an anaplerotic pathway of the TCA cycle. In Saccharomyces cerev isiae, mitochondrial and extramitochondrial CS activities are encoded by three distinct nuclear genes: CIT1, CIT2 and CIT3. The products of the CIT1 and CIT3 genes have been shown to be mitochondrial proteins, whereas that of the CIT2 gene is extramitochondrial (Graybill et al., 2007).

The glyoxylate shunt enables synthesis of C4 dicarboxylic acids from acetyl-CoA units, bypassing the two decarboxylation reactions in the TCA cycle. As the C4 metabolites so produced can be used for gluconeogenesis and other biosynthetic pathways, the glyoxylate cycle allows cells to utilize C2 compounds (such as acetate and ethanol) and fatty acids as sole carbon sources (Velot et al., 1999). Of the glyoxylate cycle genes tested, CIT2 is the only one whose transcription in the nucleus is regulated by alterations in mitochondrial function, via an inter-organelle signalling pathway termed retrograde regulation (Parikh et al., 1987; Liao and Butow, 1993). Thus, retrograde regulation balances various cellular activities according to the changes in the mitochondrial function. The CIT2 gene is the best-studied target of the retrograde response (Chen et al., 2002), responding to various situations such as inhibition of respiration, loss of TCA cycle activity or loss of mitochondrial DNA (Liu and Butow, 2006; Liu et al., 2008). Hence, Cit2 may be an important control point for metabolic cross-feeding from the glyoxylate cycle in the cytosol to the mitochondria (Chelstowska et al., 1999). To date, studies on such metabolic cross-feeding effects indicate only modulation of expression of nuclear genes in response to the alterations in mitochondrial function.

Here we show, for the first time, that the cytosolic localization of a dual targeted mitochondrial TCA cycle enzyme, fumarase, is extremely sensitive to glyoxylate shunt deficiency and in particular deficiency of the Cit2 activity. We hypothesize that metabolic stress in the cytosol results in the accumulation of specific metabolites which in turn lead to changes in dual distribution of mitochondrial proteins (which have a second function in the second location). We suggest that one (or more) of the glyoxylate metabolites such as succinate may determine the expression and extent to which fumarase molecules are withdrawn back into the cytosol.


Analysis of the yeast R strain and its altered fumarase dual localization

In an earlier study (Regev-Rudzki et al., 2005), we isolated a 5-fluoroorotic acid resistant yeast strain, expressing an Aco1-URA3 hybrid (designated the ‘R’ strain). This strain which was selected for misdistribution of aconitase also exhibited an altered distribution of chromosomal encoded fumarase (Regev-Rudzki et al., 2005). Aconitase and fumarase demonstrate highly different distribution patterns; fumarase has a significant cytosolic presence (more than 50% of the protein), whereas aconitase is eclipsed (less than 5% in the cytosol) (Regev-Rudzki et al., 2008). In the R strain these two distributed proteins lose their cytosolic localization (Regev-Rudzki et al., 2005). These results were used by us to further investigate the R strain in order to detect trans elements which may influence subcellular localization of these proteins.

DNA microarray analysis representing the full yeast genome was applied to characterize the expression profile for the R strain and compare it with the corresponding wild-type (WT) strain. Analysis of microarray images revealed 205 upregulated genes which were categorized by their functional categories of GOstat annotations (http://gostat.wehi.edu.au). We found up to 40 times enrichment value of genes involved in metabolism of organic acids, glutamate, TCA intermediates and acetyl-CoA metabolism (not shown). Among the upregulated genes (see below) (Fig. 2) which stood out and captured our interest were FUM1 encoding cytosolic and mitochondrial fumarase, CIT2 encoding cytosolic CS and ICL1 encoding cytosolic isocitrate lyase. The latter two, like ACO1 (aconitase gene), encode glyoxylate shunt enzymes.

Figure 2.

Gene expression profiles of ‘Cit2-modified’ strains by microarray analysis. The heat map shows fold changes in gene expression which are graded by different intensities of red (upregulated) and green (downregulated). Consistent results were obtained in at least three separate experiments (for each sample) on multiple technical replicates as well as after reverse labelling of the hybridized samples.

In a second approach, we applied suppressor analysis to the R strain. The R strain was transformed with a yeast high copy number 2-micron library of yeast genes under the PGK (phosphoglycerate kinase) promoter, and cells were selected for their ability to grow on ethanol-acetate plates (requiring a functional glyoxylate shunt). One of the few genes identified by this approach was again CIT2 (data not shown). As both our microarray and suppressor screen pointed to organic acid metabolism and in particular to Cit2 as probable factors in the phenomenon of altered subcellular distribution, we decided to focus on the relationship between Cit2 and the dual targeting of fumarase.

Expression levels of Cit2 affect dual targeting of fumarase

The WT and Cit2 deleted (Δcit2) strains or these yeast strains (WT and Δcit2) expressing Cit2 from a plasmid (pCit2) were examined. Yeast cells were subjected to a subcellular fractionation procedure, and equivalent portions from fractions of the total cell extract (T), cytosol (C) and mitochondria (M) were analysed by Western blotting using antibodies against fumarase. Hsp60 was used as a mitochondrial marker while hexokinase1 (HK1) and Bmh1 were used as cytosolic markers, to assess cross-contamination of cellular fractions. Remarkably, the subcellular distribution of fumarase in a CIT2 chromosomal knockout strain (Δcit2) is completely altered; the amount of cytosolic fumarase is drastically reduced in the Δcit2 strain, when compared with the WT strain (Fig. 1A; compare WT with Δcit2). To test for complementation, Δcit2 strain harbouring pCit2 (Cit2 under the GAL promoter on a 2 μ plasmid) was also examined. A nearly WT pattern of fumarase distribution was observed with significant amounts of fumarase in the cytosol and in mitochondria (Fig. 1A, compare Δcit2 with Δcit2+ pCit2). Interestingly, in a WT strain harbouring pCit2, a larger fraction of the fumarase molecules is found in the mitochondria and a smaller amount of the fumarase is retained in the cytosol (Fig. 1A; compare WT with WT + pCit2). Worth pointing out here is that in these strains the level of fumarase is not reduced compared with the WT suggesting that cytosolic fumarase is not degraded (Fig. 1B and the next sections). One way to interpret these results is that there are three levels of Cit2 expression: (i) depletion of Cit2, in a Δcit2 strain which alters fumarase distribution, (ii) low to medium levels of Cit2, in either a WT strain expressing the chromosomal CIT2 or in a Δcit2 strain harbouring the pCit2 plasmid (both strains display WT fumarase distribution) and (iii) high levels of Cit2, in a WT strain harbouring the pCit2 plasmid which alters fumarase distribution. Indeed, from microarray analysis presented in the next section, one sees that the Δcit2 strain is devoid of CIT2 transcripts while the Δcit2+ pCit2 and WT + pCit2 strains have, respectively, 2.7 (± 0.2) and 8.8 (± 0.5) fold higher levels of CIT2 transcripts than the WT yeast strain. Furthermore, we measured the cit2 mRNA levels in all of these strains (Δcit2, WT, Δcit2+ pcit2 and WT + pcit2) by real time-PCR using total yeast RNA. CIT2 expression was normalized to actin (Act1) expression (Table 1). While the CIT2 mRNA levels in Δcit2+ pcit2 strains are only 3.7-fold higher than in WT CIT2 strain, the levels of this CIT2 mRNA in the WT + pcit2 strain are profoundly higher, by 22.8-fold, than in WT CIT2 (Table 1).

Figure 1.

A. Fumarase dual distribution is affected by Cit2 expression. Yeast cells expressing different CIT2 genes were subjected to subcellular fractionation. Equivalent portions from the total (T) cytosolic (C) and mitochondrial (M) fractions were analysed by Western blotting using antibodies against the indicated proteins.
B. Western blot analysis of different Cit2 strains. Cultures from the indicated strains grown in galactose medium and were broken with glass beads in TE buffer. Equivalent portions from the total cell extracts were analysed by Western blotting using antibodies against the indicated proteins.
C. Fumarase activity in fractions. Yeast cells harbouring the indicated plasmids were grow in galactose medium and subjected to a subcellular fractionation. Equivalent extracts from the cytosolic (C) and mitochondrial (M) fractions were analysed for fumarase-specific activity. inline image Fumarase activity; inline image G6PD activity.
D. Cytosolic fumarase (lacking the MTS) is stable in the cytosol. Yeast cells expressing pFumΔMTS were subjected to subcellular fractionation. Equivalent portions from the total cell extract (T) cytosolic (C) and mitochondrial (M) fractions were analysed by Western blotting using antibodies against the indicated proteins.

Table 1.  Cit2 transcript level in ‘Cit2-modified’ strains.
  1. Total RNA was isolated from yeast cultures of the indicated strains. The amounts of Cit2 transcripts were determined by RT-PCR. The quantity of cDNA was normalized to that of ACT1 in each RNA sample. The values shown are the mean ± standard deviation of three independent isolated RNA preparation analysed in duplicates.

WT1.0 ± 0.19
Δcit20 ± 0
WT + pCit222.8 ± 1.14
Δcit2+ pCit23.7 ± 0.0

Another way to determine dual targeting of fumarase is to measure enzymatic activities in the subcellular fractions. The specific activities of cytosolic and mitochondrial fumarase and a control cytosolic enzyme were determined in the subcellular fractions of the yeast strains Δcit2, Δcit2+ pCit2, WT + pCit2 and WT. In full agreement with the Western blot analysis (Fig. 1A), the proportion of fumarase activity is very low in the cytosolic versus mitochondrial fractions obtained from Δcit2 and WT + pCit2 strains when compared with the WT or Δcit2+ pCit2 strains, in which the fumarase activity is divided equally between the corresponding subcellular fractions (Fig. 1C). It should be noted that in the first two ‘Cit2-modified’ strains there is approximately 10-fold more-fumarase activity in the mitochondrial fraction than in the cytosolic fraction. Taken together (Western blot and enzymatic activities), we suggest that a depletion or an overexpression of Cit2 causes misdistribution of fumarase.

One explanation for the reduction (or the total loss of cytosolic fumarase) in yeast depleted of Cit2 can be in the specific degradation of fumarase in the cytosol. We have previously shown that fumarase is a very stable enzyme both in the cytosol and in the mitochondria (Regev-Rudzki et al., 2005; Regev-Rudzki et al., 2008). In Fig. 1D, we examined the levels of a cytosolically targeted fumarase [lacking a mitochondrial targeting sequence (MTS)] in Δcit2 and R strains. A plasmid encoding fumarase lacking an MTS was transformed into the WT, Δcit2 and R strains. Subcellular fractions of these strains which were analysed by Western blot revealed highly significant amounts of mature fumarase in the cytosolic fraction (Fig. 1D, middle lanes). These results support the conclusion that cytosolic fumarase is stable in these strains and that fumarase dual localization is not determined by the stability of this protein in a specific cell compartment.

Microarray analysis of strains with altered Cit2 expression

A microarray analysis was used to compare the expression profiles of strains with different levels of the Cit2 expression (Δcit2, WT + pCit2 and the R strain) to the WT strain. Expression profiles were analysed and presented in a heat map format, which is a graphical representation of microarray in grid form, with columns representing samples and rows representing differentially (up and down) expressed genes. For the heat map presented, we chose the R strain as the primary sample represented by continuous colours (red and green), and its differentially expressed genes were accordingly tested in every other sample (Fig. 2).

The expression profile of the R strain resembles that of Δcit2 and less that of WT + pCit2 profiles. Thus, when comparing each of the three strains above to the WT strain, these three strains exhibit similar differentially expressed gene profiles (Fig. 2). The functional analysis programmes (e.g. GO annotation http://gostat.wehi.edu.au) were used in order to characterize the groups of impaired functions within these strains. Similar to the results obtained for the R strain, a significant enrichment was obtained in the upregulated genes that are involved in metabolism of organic acids (data not shown). In particular, the level of fumarase mRNA is upregulated both in the Δcit2 and WT + pCit2 strains (P-values of 2.47E-03, 4.22E-05 and 3.81–06 respectively). In addition, an RT-PCR analysis showed an approximately twofold higher Fum1 mRNA level in a Δcit2 strain (2.18 ± 0.208) when compared with the WT strain (1.217 ± 0.2370); the amounts of Fum1 transcripts were determined by RT-PCR, and the quantity of cDNA was normalized to that of Act1 in each RNA sample. The values are the mean ± standard deviation of three independent isolated RNA preparations analysed in duplicates.

Consequently, we asked whether these differences in fumarase mRNA levels also result in higher cellular fumarase activity. We detected a more than twofold increase in total fumarase enzyme activity in the Δcit2 and WT + pCit2 strains when compared with the WT or Δcit2+ pCit2 strains (not shown).

Dual distribution of fumarase depends on an enzymatically active Cit2

We were intrigued in determining whether Cit2 expression affecting fumarase dual targeting requires the active enzyme or whether the presence of the Cit2 protein has a second regulatory function unrelated to its enzymatic activity. To answer this question, a mutant of Cit2 (designated Mut-Cit2) was constructed by changing three conserved amino acids in the active site of the protein (amino acids 339–341: Histidine, Alanine, Valine to Alanine, Aspartic acid, Alanine). The plasmid pMut-Cit2 encoding this mutant enzyme was transformed into a yeast Δcit2 strain and as expected (in contrast to the WT gene) supported weaker growth on oleic acid as the carbon source in the medium. This growth phenotype indicated a glyoxylate shunt defect (Fig. 3A, compare Δcit2 and Δcit2+ pMut-Cit2 with the WT and Δcit2+ pCit2). The reason for this weak growth can be attributed to the fact that yeast encodes additional CS genes (Cit1 and Cit3) which may partially complement the lack of Cit2 in the glyoxylate shunt. Accordingly, a complete block of the glyoxylate shunt, for example in a Δmls1 strain, fully impairs growth on oleic acid (Fig. 3A, lower panels). As presented in Fig. 3B, the cytosolic amount of fumarase is diminished in subcellular fractions of the Δcit2+ pMut-Cit2 strain. In fact, the Δcit2+ pMut-Cit2 strain is essentially identical to the Δcit2 strain. It is important to point out that the levels of Cit2p and Mut-Cit2p protein in these cells (according to Western blots) are identical (data not shown). It appears that the native cytosolic localization of fumarase depends on an enzymatically active Cit2 protein.

Figure 3.

A. A Cit2 active site mutant gene does not complement the Δcit2 strain. WT or Δcit2 strains harbouring the indicated plasmids were diluted and grown on galactose (Gal) or oleate medium plates. Mls1, malate synthase 1.
B. A Cit2 active site mutant gene does not restore fumarase distribution. Yeast cells expressing pCit2 or pMut-Cit2 were subjected to subcellular fractionation. Equivalent portions from the total cell extract (T) cytosolic (C) and mitochondrial (M) fractions were analysed by Western blotting using antibodies against the indicated proteins.

Glyoxylate shunt deletion mutants exhibit an altered fumarase dual distribution

In yeast, the enzyme Cit2 participates in the glyoxylate shunt together with aconitase (Aco1), malate synthase (Mls1), isocitrate lyase (Icl1) and malate dehydrogenase (Mdh3). This shunt enables the cell to synthesize four-carbon organic compounds such as oxaloacetate from two-carbon compounds such as ethanol or acetate. As the dual distribution phenotype of fumarase depends on an active Cit2, it is possible that substrates or products of the enzymatic reaction may be involved. To ask whether fumarase cytosolic localization is exclusively linked to Cit2 activity or to the entire glyoxylate shunt and its intermediates, fumarase subcellular localization was analysed in three of the chromosomal deletion strains of the glyoxylate shunt enzymes: Δmls1, Δaco1 and Δicl1. The subcellular fractionations of these three strains exhibited a pattern of distribution that is reminiscent of the results with the Δcit2 strain. The cytosolic amount of fumarase in these three strains was significantly reduced when compared with the WT strain (Fig. 4A, upper panel). Furthermore, subcellular fractionation experiments for these deletion strains, each expressing the respective missing protein from a plasmid (Δicl1+ pIcl1, Δmls1+ pMls1, Δaco1+ pAco1) revealed that the cytosolic presence of fumarase is restored (Fig. 4A, second panel). The fascinating finding that depletion of any one of the enzymes of the glyoxylate shunt affects fumarase dual localization strongly suggests that intracellular metabolic cues which are intermediates of the glyoxylate shunt, may be responsible for the dual localization phenotype described above.

Figure 4.

A. Fumarase dual localization in glyoxylate shunt defective strains. Yeast glyoxylate shunt knockout strains harbouring the indicated plasmids were subject to subcellular fractionation. Equivalent portions from the total cell extract (T) cytosolic (C) and mitochondrial (M) fractions were analysed by Western blotting using antibodies against the indicated proteins.
B. Higher-fumarase activity in glyoxylate shunt defective strains. WT or knockout strains harbouring the indicated plasmids were grown in galactose medium, and cell extracts were assayed for fumarase and G6PD activity. G6PD, a control cytosolic enzymatic; inline image fumarase activity; inline image G6PD activity.
C. Western blot analysis of glyoxylate knockout strains. Cultures of the indicated strains grown in galactose medium were broken with glass beads in buffer. Equivalent portions from the total cell extract were analysed by Western blotting using antibodies against the indicated proteins. The relative fumarase band intensities divided by the respective HK1 intensities: WT = 1, Δcit2 = 1.4, Δcit2 = 1.6, Δmls1 = 1.7.

In addition to the change in fumarase distribution, we detected a significant increase in fumarase expression and cellular enzymatic activity in the glyoxylate deletion strains: Δmls1, Δicl1, Δaco1and Δcit2. The total (cytosolic plus mitochondrial) specific activity of fumarase was more than 100% higher in these cells bearing glyoxylate shunt gene deletions than in WT cells (Fig. 4B). Accordingly, the fumarase protein is expressed to higher levels in these strains (Fig. 4C compare deletion strains with the WT). Moreover when the respective missing glyoxylate enzyme is expressed from a plasmid in such deletion strains, the fumarase activity levels reverted back to that of the WT level (Δcit2 + pCit2, Δicl1+ pIcl1, Δmls1+ pMls1; Fig. 4B). Thus, glyoxylate shunt abrogation affects both fumarase distribution and expression.

Succinic acid in the growth medium affects fumarase dual distribution

Our assumption is that the presence of fumarase in the cytosol is related to a functional glyoxylate shunt and possibly to intermediates of this pathway. This was examined by studying whether organic acid intermediates of the glyoxylate pathway may alter fumarase distribution, when present at high levels in the growth media. For this, yeast WT strains were grown in either rich or minimal media, in the presence of l-malic, citric, succinic (Fig. 5), fumaric and oxaloacetic acid (not shown). The yeast cultures were then subjected to a standard subcellular fractionation. A normal strong presence of cytosolic fumarase was detected in cells grown with the above organic acids; relative band intensities of 47% and 51% of the total were detected by densitometric of Western blots of cells grown with citric acid and malic acid, respectively. However, for cells grown with succinic acid we detected a weaker band of cytosolic fumarase, comprising only 20% of the total (Fig. 5). The results with fumaric and oxaloacetic acids were similar to those obtained with l-malic and citric acids (not shown). It is interesting to point out that total fumarase activity increased approximately by 20% in cells grown with succinic acid as compared with the control cells to which no acid was added to the growth medium (data not shown). Thus, succinic acid may play a role in controlling the dual targeting of fumarase as a way for cross-feeding information between pathways in the mitochondria and the cytosol.

Figure 5.

Succinic acid reduces fumarase cytosolic presence. Yeast WT cells were grown in galactose medium containing 50 mM of the indicated organic acids. Cells were subjected to subcellular fractionation, and equivalent aliquots of the total (T) cytosolic (C) and mitochondrial (M) fractions were analysed by Western blotting using antibodies against the indicated proteins. The percentage of fumarase in mitochondrial and cytosolic fractions was quantified from densitometric analysis of band intensities; %Cytosolic–%Mitochondrial, untreated 51–49, citrate 47–53, succinate 20–80, malate 51–49.


In the present study, we have uncovered a link between primary metabolism and protein distribution in different cellular compartments. Our results show that the fumarase distribution pattern between the cytosol and the mitochondrial fractions in yeast is subject to remodelling which is linked to metabolic cues originating from the glyoxylate shunt. Five observations support this notion: (i) deletion mutant strains of the glyoxylate shunt exhibit a diminished cytosolic presence of fumarase, (ii) when the respective missing glyoxylate genes (ACO1, CIT2, ICL1 and MLS1) are restored on a plasmid in these cells, fumarase distribution reverts back to the WT pattern, (iii) depletion or overexpression of CIT2 reduces the level of cytosolic fumarase, (iv) when succinate, a metabolite of the glyoxylate shunt, was added to the growth medium, a significant reduction in the level of cytosolic fumarase was observed and (v) a yeast strain (the ‘R’ strain) selected for misdistribution of aconitase also misdistributes fumarase (Regev-Rudzki et al., 2005) and exhibits upregulation of genes involved in the metabolism of organic acids.

Fumarase mislocalization is correlated with increased mRNA and protein expression when compared with the WT state; deletion or overexpression of CIT2, deletion of glyoxylate shunt genes, growth in the presence of succinic acid and expression in the R strain, all exhibit approximately a twofold overexpression of fumarase and an alteration in fumarase distribution. While our finding that metabolic cues affect protein distribution and localization is novel, the fact that intracellular metabolic signals can affect gene expression is well documented. This latter phenomenon resembles the mitochondrial retrograde signalling (RTG) pathway in which mitochondrial dysfunction in combination with the lack of glutamate and glutamine leads to the induction of specific nuclear encoded genes (Chen et al., 2002; Butow and Avadhani, 2004). The precise pathway involved in fumarase induction remains to be deciphered.

Our hypothesis is that glyoxylate shunt intermediates and/or products control the dual distribution balance of fumarase and probably other proteins in the cell. Thus, we suggest that in yeast intermediates of primary metabolism can act as ‘nanosensors’ for remodelling protein localization. Possible scenarios as to how these metabolic cues affect protein subcellular distribution, based on our knowledge about the mechanism of fumarase targeting and dual distribution, are briefly described below.

Previous studies revealed that the single initial translation products of the FUM1 gene are first partially translocated; while a subset of these molecules continues to be fully translocated into the organelle, the rest are released by the retrograde movement of the molecules back into the cytosol (Knox et al., 1998; Sass et al., 2003). The rapid folding of fumarase blocks anterograde movement and appears to provide the driving force for retrograde movement (of the processed protein) back to the cytosol through the translocation pore. (Sass et al., 2003; Yogev et al., 2007). Another unique feature of fumarase is that it cannot be imported into mitochondria after the termination of translation. The results indicate that import of fumarase into mitochondria occurs while the ribosome is still attached to the nascent chain (Karniely et al., 2006; Yogev et al., 2007). Nevertheless, the nascent chain is exposed to the cytosol and this exposure of fumarase depends on both translocation and translation rates. Accordingly, mitochondrially attached polysomes are enriched for the FUM1 message in comparison with free polysomes (Marc et al., 2002; Reut Hazan and Ophry Pines, unpublished results).

We suggest three possible chains of events to explain how metabolic cues affect protein subcellular distribution:

  • i. Organic acids such as succinic acid (a possible effector metabolite) directly interact with fumarase and slow down its folding thereby causing more fumarase to be fully imported into mitochondria. In preliminary experiments, we detect an increase in fumarase stability to proteinase K in the presence of organic acids which appears to be non-specific; any one of the organic acids, even acetate, affects fumarase stability similarly (not shown). In addition, the presence of succinate does not affect the enzymatic activity of fumarase (not shown). Thus, at present we do not have support for a direct interaction of succinate with fumarase, and its affect could rather occur indirectly via a trans factor that binds the effector metabolite and in turn affects fumarase folding.
  • ii. The effector metabolite interacts directly or indirectly with the mitochondrial translocon(s), thereby enhancing translocation and providing less opportunity for fumarase to fold prior to full import. In preliminary in vitro experiments, we have not found significant effects of succinate or in fact any other organic acid such as malate or fumarate on the kinetics or final yield of Su9-DHFR translocation into isolated mitochondria (not shown). It is possible that succinate either affects translocation indirectly or that we simply have not found the specific in vitro conditions under which succinate can exert its effect.
  • iii. The effector metabolite directly (or indirectly) affects the coupling of translocation to translation, either by inducing the transcription of a different FUM1 mRNA or by affecting the interaction of the ribosome with the translocon. The earlier is consistent with the upregulation of FUM1 under the variety conditions and strains reported in the present work. The notion is that metabolic cues alter fumarase import so that the ribosome intimately feeds the translating polypeptide into the translocon, thereby limiting the opportunity of fumarase folding prior to full import.

The idea that intermediates of primary metabolism may also serve as signalling molecules changes the way we think about protein function and intracellular signalling. For example, human fumarase has been identified as a tumour suppressor gene whose bi-allelic deletion causes fumaric acid accumulation which leads to stabilization of the transcription factor HIF and tumour proliferation. Here, we suggest that metabolic intermediates can affect protein localization which should influence our perceptive of dynamic eukaryotic cell biology.

Experimental procedures

Strains, plasmid constructs and growth conditions

Strains.  The Scerevisiae strains used in the present study were BY4741 (Mat a, his3Δ1, leu2Δ0, met15Δ0, ura3Δ0) and BY4743 (Mat a/α, his3Δ1/his3Δ1, leu2Δ0/leu2Δ0, lys2Δ0/LYS2, MET15/met15Δ0, ura3Δ0/ura3Δ0, YLR304c::kanMX4/YLR304c). Strain Δaco1 (Regev-Rudzki et al., 2005) and strain Δfum1 (Sass et al., 2003) were described previously. Strains Δmls1, Δcit2 and Δicl1 were provided by the EUROSCARF Company, Johann Wolfgang Goethe-University Frankfurt.

Plasmid constructs.  Plasmids pAco1 and pFumΔMTS were created as previously described (Regev-Rudzki et al., 2005). Plasmids pCit2, pIcl1 and pMls1 were provided by the Open Biosystems Company. pMut-cit2 mutations were created by side-directed PCR using the QuickChange®II kit (Stratagene) and pCit2 as a template.

Growth conditions.  Strains harbouring the appropriate plasmids were grown overnight at 30°C on a synthetic depleted medium containing yeast nitrogen base (0.67% wt/vol) and supplemented with the appropriate amino acids (50 μg ml−1). The following carbon sources were added separately to the medium: galactose (2%), glucose (2%) and oleate (0.125%). For growth on plates, agar (2% wt/vol) was added.

Subcellular fractionation

Induced yeast cultures were grown to a concentration of 1.5 OD600nm. Mitochondria were isolated as described previously (Knox et al., 1998). Spheroplasts were prepared in the presence of Zymolyase-20T (MP Biomedicals, Irvine, CA). Each of subcellular fractionation experiments was assayed for cross-contaminations between the subcellular fractions using αHsp60 as mitochondrial marker and αHexoKinase1 (HK1) or αBmh1 as a cytosolic marker. Cytosolic and mitochondrial band intensities were quantified densitometrically using TINA software.

Enzyme activities

Induced cultures (in galactose) were harvested and resuspended in TE buffer (pH 8.0) containing 1 mM phenylmethylsulfonyl fluoride. Cell suspensions were broken with glass beads or by a French pressure apparatus. The resulting suspensions were centrifuged to obtain the supernatant fractions. Fumarase was assayed by the method of Kanarek and Hill (Kanarek et al., 1964) at 250 nm with l-malic acid as substrate. Glucose 6-phosphate dehydrogenase (G6PD) was assayed by following the formation of NADH in the presence of glucose 6-phosphate at 340 nm (Worthington, 1988). Protein concentration was determined by the method of Bradford (Bradford, 1976).


Total RNA was prepared from yeast cells, using the MasterPure™ yeast RNA purification kit (MPY03100, Epicentre Biotechnologies) according to the manufacturer instructions. First strand cDNA used as template was synthesized by reverse transcription using oligo(dT) as a primer and the Reverse-iTMAX First Strand Kit (ABgene). Fumarase and citrate synthase2 expression were normalized by actin1 expression and were quantified by real time-PCR (Rotor Gene RG-3000A) using SYBER green Master Mix (Absolute Syber Green ROX Mix, ABgene) using the following primers:













Microarray analysis

RNA extraction and labelling.  Total RNA was extracted using the MasterPure™ Yeast RNA Purification Kit (MPY03100, Epicentre Biotechnologies) and reverse transcribed using the superscript II reverse transcriptase (Invitrogen). Details of the microarrays and probe labelling, hybridization and washing conditions can be found at http://www.microarrays.ca/support/support.html. Shortly, the reverse transcription reaction was carried out at 42°C for 2 h with aminoallyl (aa)-dUTP. Removal of unincorporated (aa)-dUTP and free amines was carried out using Microcon YM-30 (Millipore) filters according to the manufacturer's recommendations. Coupling of aminoallyl-labelled cDNA to Cy-dye esters was performed in 0.1 M sodium carbonate buffer (pH = 8.6) for 1 h at room temperature. Removal of free dyes was accomplished with Qiagen QIAquick PCR purification columns. The labelling level was quantified using a spectrophotometer (ND-100, Nanodrop Technologies). Each Cy3-labelled sample was mixed with an equal amount of Cy5-labelled sample, and the mixtures were allowed to dry in a speed vacuum for about 1 h.

Microarray hybridization.  The S. cerevisiae expression double-spotted arrays containing 6240 yeast ORFs (plus control spots totalling 6.4K) were obtained from the Toronto Microarray Center (based at the University Health Network, Canada). Genes are represented by 70-bases oligonucleotides. Hybridization was performed basically according to TIGR protocol (http://pfgrc.jcvi.org/index.php/microarray.html). The samples were resuspended in hybridization buffer (25% formamide, 5X SSC, 0.1% SDS and 20 mg yeast tRNA) and heated at 95°C for 5 min. Following a quick centrifugation, the Cy-dyes coupled probes were applied to a microarray slide and incubated overnight at 42°C. Hybridization was carried out in corning hybridization chambers submerged in a water bath. The hybridized slides were then washed and scanned by using an Axon GenePix 4000B Scanner (Axon Instruments) with settings adjusted to obtain a similar green and red overall intensity.

Array data: normalization and statistical analysis

The results represent the findings from three independent biological replicate arrays performed with three different RNA samples for each specimen. In two of the arrays, the WT RNA sample was labelled with Cy5 and the target sample with Cy3 while the third array was reversibly labelled. The array images were quantitatively analysed using the GenePix Pro 4.1 software (Axon). This software localizes every spot within a programmed grid to obtain a mean signal intensity of each spot and background (offset) of the area surrounding the spot. The resulting files (gpr) produced by GenePix software were further analysed utilizing the LIMMA (Smyth, 2004) software package, available from the CRAN site (http://www.r-project.org). Spots flagged as not found or absent in GenePix were removed by filtering. Another filter was applied for saturated spots. After filtering, the data within the same slide were normalized using global loess normalization with the default smoothing span of 0.3 (Smyth and Speed, 2003). In order to identify differentially expressed genes, a parametric empirical Bayesian approach implemented in LIMMA was used (Lonnstedt and Speed, 2002). A moderated t-test was performed in parallel, with the use of a false discovery rate (Reiner et al., 2003) correction for multiple testing. A P-value < 0.05 confidence level was used to pinpoint those significantly differentiated genes. For heat map graphs, the Limma method was used for the selection of differentially expressed genes in each sample compared with the WT. The overall differentially expressed gene is represented in a heat map. Similarities of gene expression patterns between primary (Pr) strain (sequential colours) and other strains are represented by the vertical dendrogram. The relative gene expression in each sample is represented with respect to the WT strain. Expression levels are colour coded: red, expression above; green, expression below the mean of normalized gene expression across all samples.

cDNA library screen

R strain cells following eradicating of the Aco1-URA3 (Regev-Rudzki et al., 2005) and harbouring plasmid pAco1-α were transformed with a yeast cDNA library kindly provided by Nataly Bonfoy. This library contains yeast cDNA cloned under the PGK3 constitutive promoter in the centromeric pFL61 shuttle vector (URA3 marker) (Minet et al., 1992). Following transformation, cells were plated on galactose plates and ETOH-acetate plates and screened for colonies, exhibiting growth on both kinds of plates. A total of 32 000 cells were examined; plasmids from positive clones were isolated from collected cells and were sequenced using the FL61FW primer.


We thank Jerry Kaplan for discussions, help and critical reading of the manuscript. We thank Miriam Kott-Gutkowski from The Core Research Facility of the Hebrew University Faculty of Medicine for the microarray analysis. We thank Narmen Azazmeh for help with the real time-PCR analysis and Gideon D. Matthews for help with the in vitro import. We thank Merav Tal, Yudit Karp and Adi Naamati for their dedicated assistance. This research was supported by the Israel Science Foundation, the German Israeli Foundation and the German Israeli Project Cooperation.