INTRODUCTION

c-myc is a major regulator of cell growth and differentiation, and it is a key player in human oncogenesis (for reviews see Refs. 1, 2, 3). The myc family of proto-oncogenes has been implicated to regulate gene transcription. All Myc proteins harbor several motifs characteristic to known transcription factors and together with another nuclear protein, Max, they bind a specific DNA sequence (4, 5). The pathways by which c-Myc induces cellular processes is still obscure, and targets for Myc activity are at large unknown. Yet, in the past years, several genes were suggested to be directly regulated by c-Myc. The gene ECA39 has been isolated by a subtraction/co-expression strategy using c-Myc-induced tumors of transgenic mice (6, 7). ECA39 gene was shown to harbor a functional c-Myc binding motif and was proven to be a direct target for c-Myc activity in both mice and man (6, 8). Other genes suggested as targets for c-Myc regulation are -prothymosin (9, 10), p53 (11, 12). and ornithine decarboxylase (13, 14).

c-Myc oncoprotein has been shown to induce both proliferation and apoptosis (3). In addition, its targets have also been implicated to be involved in one or both of these processes. p53 was suggested to directly mediate c-Myc-induced apoptosis, since activation of c-myc failed to induce apoptosis in p53-null fibroblasts (15). Ornithine decarboxylase, a rate-limiting enzyme in the biosynthesis of polyamines, was suggested to mediate both proliferation and apoptosis of cells (16, 17, 18, 19).

To study the function of ECA39, we have recently characterized several homologs of ECA39 from vertebrates and invertebrates (8). In studying a Saccharomyces cerevisiae homolog of ECA39, we disrupted the gene and showed that it is involved in the regulation of cell growth and in the transition from G1 to S phase in the cell cycle (8). In the present study, we characterize another yeast homolog of ECA39 and two human homologs. Moreover, we demonstrate that the two yeast homologs code for mitochondrial and cytoplasmic branched-chain amino acid aminotransferases (BCAT).¹ We thus suggest that BCAT is a target for c-Myc regulation.

MATERIALS AND METHODS

YPH857 (MAT, ura3-52, lys2-801, ade2-101, trp163, his3200, leu21, cyh2R) (20). YPH858 is the isogenic MATa strain. AE16 is isogenic to YPH857 with eca39::URA3, and AE1632 is the eca39 homozygous diploid strain. AE25 is isogenic to YPH857 with eca40::HIS3. YEPD, SC, sporulation, and dropout media were prepared as described (21).

Yeast genomic DNA was extracted as described (21). Total RNA was extracted as described previously (22). Southern and Northern blot analyses were performed as described (23, 24). Radiolabeling of DNA probes was performed by random priming (25) using [-³²P]dCTP (3000 Ci/mM, Rotem Industries, Israel). DNA sequencing was performed automatically using the dideoxynucleotide chain termination method (26). Computer analysis of protein and DNA sequences was performed using the GCG Wisconsin package (27) and Psort server (28).

The ECA40 gene was amplified from S. cerevisiae DNA with specific oligonucleotide primers (5 oligomer: CTGAGGGTCATTCACGACCTAG and 3 oligomer: CTTCTAAGGTATGTATGGGCC), using standard PCR methodology (22). The 1.8-kilobase PCR product was cloned into a pBluescript SK() phagemid. A vector for ECA40 gene disruption was constructed by deleting a 1145-base pair StyI-SnaBI fragment and inserting a 2257-base pair ScaI-NaeI fragment from pRS303 (29) containing the HIS3 gene. A ClaI-XbaI fragment from this plasmid was used for homologous recombination and introduced by LiAc transformation (30) into the YPH857 yeast and into an eca39 diploid homozygous strain (8).

The procedure is a modification of the method described by Yaffe (31). Cells were grown on YEPD to A₆₀₀ of about 1.6, collected, and resuspended to 0.2 g/ml in 0.1 M Tris-SO₄, pH 9.3, with 10 mM dithiothreitol, incubated for 10 min in 30 °C, collected again, washed in spheroplasting buffer (1.2 M sorbitol, 20 mM potassium phosphate, pH 7.4), and resuspended to 0.1 g/ml in spheroplasting buffer containing 7,000 units/ml of yeast lytic enzyme (ICN Immunobiologicals). Cells were incubated for 45 min in 30 °C, and spheroplasts were collected and washed twice in spheroplasting buffer. Spheroplasts were resuspended in MIB buffer (0.6 M mannitol, 20 mM HEPES, pH 7.4) to a final concentration of 0.5 g of cells/ml and broken in a Dounce homogenizer. The homogenate was centrifuged at 3,000 × g for 5 min at 4 °C to avoid intact cells. The supernatant was centrifuged at 9,500 × g for 10 min at 4 °C. The pellet contained the crude mitochondrial fraction. The supernatant from this centrifugation was further centrifuged at 200,000 × g for 20 min at 4 °C, and the supernatant from this spin was frozen and used as the cytosolic extract. The crude mitochondrial fraction was washed three times in MIB buffer and sonicated, and the supernatant was frozen as the mitochondrial extract.

Assay procedure is based on the method described by Ichihara and Koyama (32). Cytosolic or mitochondrial extracts were incubated with 37 mM sodium pyrophosphate buffer, pH 9.2, 6.7 mM -ketoglutarate, 67 µM pyridoxal phosphate, and 6.7 mM branched-chain amino acid (either leucine, isoleucine, or valine). The reaction was terminated after 10 min by addition of trichloroacetic acid. Samples were centrifuged and clear solution was transferred to a new tube and incubated at 25 °C with 2,4-dinitrophenylhydrazine. The hydrazones of the branched-chain keto acids were first extracted with toluene and then with sodium carbonate. The amount of the hydrazone was determined by measuring the optical absorbance at 440 nm. An extinction coefficient of 1.4 A/µmol was found for the hydrazones of all three keto acids (33). Specific activity was calculated in units defined as micromoles of branched-chain keto acids produced × min¹ × mg of protein¹.

RESULTS

We have previously identified homologs to the murine ECA39 gene from human, Caenorhabditis elegans, and S. cerevisiae genomes (8). Recently, in the S. cerevisiae genome project, another gene homologous to ECA39 was identified on chromosome X (ORF YJR148w, accession number Z49648[GenBank]), referred by us here as yeast ECA40 (Fig. 1). The amino acid sequence of the yeast ECA40 shows 77% identity to the yeast ECA39 protein. In parallel, we have identified a human expressed sequence tag (cloned by the Washington University-Merck & Co. EST Project) homologous to the human ECA39. A cDNA clone was obtained from the American Type Culture Collection (ATCC), and we determined its full nucleotide sequences. Our analysis revealed a new gene homologous to the human ECA39, and we address it as human ECA40 (Fig. 1). The amino acid sequence of human ECA40 shows a 54% identity to the human ECA39 protein.

The mouse ECA39 gene is highly expressed in dividing cells and during embryogenesis (6). Similarly, the yeast ECA39 gene is expressed at high levels in dividing cells (logarithmic phase of growth) and at low levels in nondividing cells (stationary phase of growth) (8). We thus examined the pattern of expression of the yeast ECA40 gene, in the same growth conditions. Contrary to ECA39, ECA40 is expressed at higher levels during stationary phase of growth than during the logarithmic phase of growth (Fig. 2). As controls we refer to levels of ACT1 mRNA and of 25 S and 18 S rRNA (Fig. 2).

So far, five eukaryotic genes homologous to the mouse ECA39 were identified. A tree representation of alignment of the amino acid sequences homologous to mouse ECA39 is given in Fig. 3. The alignment revealed that the eukaryotic ECA39/ECA40 genes show significant similarity to prokaryotic genes. The highest similarity is to Bacillus subtilis (accession number P39576[GenBank]) and to Hemophilus influenzae (34) genes. More informative is the remote similarity to the known prokaryotic enzymes, branched-chain amino acid aminotransferase (BCAT) (ilvE gene (35, 36)), D-alanine aminotransferase (D-AAT (37)), and 4-amino-4-deoxychorismate lyase (pabC (38)), which were suggested to form a separate subgroup in the aminotransferase family (39).

To study the function of ECA39, we have previously disrupted the gene in S. cerevisiae (8). To extend our analysis, we have now disrupted the gene ECA40. Deletion of most of the ECA40 gene was carried out by one-step replacement (40), using a HIS3 disruption vector. Deletions (eca40::HIS3) were introduced into YPH857 haploid strains and into AE1632, eca39::URA3 homozygous diploid strains. The gene disruptions were verified at the DNA level (results not shown). Haploid eca40 strains were viable, and no growth phenotypes were observed. When sporulation was induced in the eca39/eca39, ECA40/eca40 heterozygous diploid cells, asci appeared normal. When dissected on YEPD medium, two spores of Ura⁺ His phenotype (eca39,ECA40) were obtained from each tetrad (12 tetrads). The other two spores germinated and gave rise to tiny clusters of cells which developed very slowly into colonies of double mutant genotype.

Based on the homology of ECA39 and ECA40 to the prokaryotic ilvE genes (Fig. 3) encoding BCAT (EC), we tested this activity in yeast cells mutated in either ECA39 or ECA40 genes. In mammalian cells, activities of both cytosolic and mitochondrial isoenzymes have been measured (32, 41, 42). Therefore, we tested BCAT activity in either cytosolic or mitochondrial fractions. The results show loss of cytosolic BCAT activity in the strain lacking ECA40, suggesting that ECA40 codes for the cytosolic BCAT (Fig. 4A). In contrast, mitochondrial BCAT activity was lost in the strains lacking ECA39, suggesting that ECA39 codes for mitochondrial BCAT (Fig. 4B). The same results were found when using as substrate each of the three different branched-chain amino acids (isoleucine, leucine, and valine) (Fig. 4).

eca39

eca40

DISCUSSION

BCATs are found in all organisms, catalyzing transamination of branched-chain amino acids to branched-chain -keto acids. In mammals they execute the first step in catabolism of branched-chain amino acids, while in lower organisms, they are involved in biosynthesis as well (43). It seems that in prokaryotes only one form of the enzyme exists as a single gene exists in H. influenzae, where the entire genome was sequenced (34). In mammals, both cytosolic and mitochondrial activities were measured (32, 41, 42). The two eukaryotic isoenzymes are encoded by two different genes, as we describe a pair of genes in the unicellular yeast S. cerevisiae and in man.

Aminotransferases require pyridoxal phosphate (PLP) as a co-factor for their activity. Sugio et al. (44) identified specific residues involved in the stabilization of the PLP molecule in D-AAT, an aminotransferase related to BCAT (39). These residues are highly conserved in the eukaryotic BCAT enzymes (results not shown), supporting the notion that binding of the co-factor in BCATs is similar to that described for D-AAT.

The existence of two BCAT homologous genes in yeast directed us to the finding that one (ECA39) codes for the mitochondrial enzyme and the other (ECA40) for the cytosolic enzyme (Fig. 4). The mitochondrial enzyme harbors a potential amphipathic helix in the most N-terminal region, typical to mitochondrial signal peptide, with a putative PRPNEE cleavage site (28, 45). Moreover, as suggested by Hartmann et al. (46), the pI value for mitochondrial isoenzymes is frequently more basic than the pI of cytosolic isoenzymes, and, thus, as noted by Hutson et al. (47), may serve to distinguish between cytosolic and mitochondrial BCATs. Indeed, the mitochondrial yeast enzyme has a basic pI of 9.0, while the cytosolic enzyme has a more acidic pI of 6.9. Based on this analysis, we suggest that the human ECA40, which has a calculated basic pI of 8.7, codes for the mitochondrial isoenzyme. We could not recognize an apparent mitochondrial signal peptide in human ECA40, however. The N-terminal domain may form an amphipathic helix; yet, in contrast to classical signal sequences, several negatively charged residues are found in this region. The mouse and human ECA39 genes seem to code for the cytosolic enzyme, as the proteins have an acidic calculated pI (5.1 and 5.2, respectively). The recent identification of the cytosolic BCAT from rat, showing a better homology to human ECA39 than to ECA40 and an acidic pI, supports this conclusion (47).

What is the significance of branched-chain amino acid aminotransferases being regulated by the c-myc oncogene? One of the few known targets for c-Myc regulation is the gene encoding the metabolic enzyme ornithine decarboxylase, a rate-limiting enzyme in polyamine biosynthesis. This enzyme was suggested to play a major role in proliferation, regulating entrance to S phase in the cell cycle (16, 17). The same enzyme was suggested to mediate c-Myc-induced apoptosis (18). We now identify another c-Myc target as a metabolic enzyme, the branched-chain amino acid aminotransferase. Biochemical studies have shown that cytosolic BCAT activity in mammals is limited to very few adult tissues (mainly brain) but is found to be abundant in fetal tissues and in transformed cell lines (48, 49). Interestingly, the expression pattern of the mouse ECA39 gene, isolated as a target for c-Myc activity, is in agreement with these observations (6), supporting the conclusion that BCAT is involved in c-Myc oncogenesis. We have recently suggested that the yeast ECA39 (mitochondrial BCAT) is involved in the regulation of the transition from G1 to S phase in the cell cycle (8). eca39 cells grow faster than wild type cells, G1 phase is shorter, and cells in G1 are smaller than wild type (8). Cells lacking ECA40 (cytosolic BCAT) do not show a growth phenotype. We have shown that ECA39 is expressed at high levels in proliferating cells while ECA40 expression is higher during stationary phase (Fig. 2). We thus suggest that the mitochondrial BCAT is the prominent isoenzyme during the logarithmic phase of growth, and, therefore, yeast cells lacking ECA39 experience an alteration in regulation of the cell cycle. One way to explain the involvement of BCAT in cell cycle regulation is through the level of the metabolites in the transamination reaction. It has been shown previously that mammalian cells exposed to high levels of branched-chain keto acids grow more slowly and that their G1 phase is extended (50). Our results fit such an observation, as in cells lacking ECA39 (mitochondrial BCAT), the production of the branched-chain keto acids is reduced and G1 phase is shorter (8). It is still possible that BCATs use other substrates, besides the three branched-chain amino acids, or that they carry out another function. In this regard, the mitochondrial BCAT was already suggested to function as a branched chain -keto acid transporter (51). This and/or another role for the BCATs may explain the difficulty in growing yeast cells mutated in both cytosolic and mitochondrial BCAT even on rich medium.

Studying the function of genetic targets for c-Myc regulation should direct us to pathways involved in normal development and in tumor formation. The realization that BCAT is a target for c-Myc activity suggests that regulation of catabolism of essential amino acids (branched-chain amino acids) is part of oncogenesis and embryogenesis.