Many proteins that contain a carboxyl-terminal CaaX sequence motif, including Ras and yeast a-factor, undergo a series of sequential posttranslational processing steps. Following the initial prenylation of the cysteine, the three C-terminal amino acids are proteolytically removed, and the newly formed prenylcysteine is carboxymethylated. The specific amino acids that comprise the CaaX sequence influence whether the protein can be prenylated and proteolyzed. In this study, we evaluated processing of a-factor variants with all possible single amino acid substitutions at either the a 1 , the a 2 , or the X position of the a-factor Ca 1 a 2 X sequence, CVIA. The substrate specificity of the two known yeast CaaX proteases, Afc1p and Rce1p, was investigated in vivo. Both Afc1p and Rce1p were able to proteolyze a-factor with A, V, L, I, C, or M at the a 1 position, V, L, I, C, or M at the a 2 position, or any amino acid at the X position that was acceptable for prenylation of the cysteine. Eight additional a-factor variants with a 1 substitutions were proteolyzed by Rce1p but not by Afc1p. In contrast, Afc1p was able to proteolyze additional a-factor variants that Rce1p may not be able to proteolyze. In vitro assays indicated that farnesylation was compromised or undetectable for 11 a-factor variants that produced no detectable halo in the wild-type AFC1 RCE1 strain. The isolation of mutations in RCE1 that improved proteolysis of a-factor-CAMQ, indicated that amino acid substitutions E139K, F189L, and Q201R in Rce1p affected its substrate specificity.

Two protein prenyltransferase enzymes, farnesyltransferase (FTase) and geranylgeranyltransferase-I (GGTase-I), catalyze the covalent attachment of a farnesyl or geranylgeranyl lipid group to the cysteine of a CaaX sequence (cysteine [C], two aliphatic amino acids [aa], and any amino acid [X]. In vitro studies reported here confirm previous reports that CaaX proteins with a C-terminal serine are farnesylated by FTase and those with a C-terminal leucine are geranylgeranylated by GGTase-I. In addition, we found that FTase can farnesylate CaaX proteins with a C-terminal leucine and can transfer a geranylgeranyl group to some CaaX proteins. Genetic data indicate that FTase and GGTase-I have the same substrate preferences in vivo as in vitro and also show that each enzyme can prenylate some of the preferred substrates of the other enzyme in vivo. Specifically, the viability of yeast cells lacking FTase is due to prenylation of Ras proteins by GGTase-I. Although this GGTase-I dependent prenylation of Ras is sufficient for growth, it is not sufficient for mutationally activated Ras proteins to exert deleterious effects on growth. The dependence of the activated Ras phenotype on FTase can be bypassed by replacing the C-terminal serine with leucine. This altered form of Ras appears to be prenylated by both GGTase-I and FTase, since it produces an activated phenotype in a strain lacking either FTase or GGTase-I. Yeast cells can grow in the absence of GGTase-I as long as two essential substrates are overexpressed, but their growth is slow. Such strains are dependent on FTase for viability and are able to grow faster when FTase is overproduced, suggesting that FTase can prenylate the essential substrates of GGTase-I when they are overproduced.

The post-translational processing of the yeast a-mating pheromone precursor, Ras proteins, nuclear lamins, and some subunits of trimeric G proteins requires a set of complex modifications at their carboxyl termini. This processing includes three steps: prenylation of a cysteine residue, proteolytic processing, and carboxymethylation. In the yeast Saccharomyces cerevisiae, the product of the DPR1-RAM1 gene participates in this type of processing. Through the use of an in vitro assay with peptide substrates modeled after a presumptive a-mating pheromone precursor, it was discovered that mutations in DPR1-RAM1 cause a defect in the prenylation reaction. It was further shown that DPR1-RAM1 encodes an essential and limiting component of a protein prenyltransferase. These studies also implied a fixed order of the three processing steps shared by prenylated proteins: prenylation, proteolysis, then carboxymethylation. Because the yeast protein prenyltransferase could also prenylate human H-ras p21 precursor, the human DPR1-RAM1 analogue may be a useful target for anticancer chemotherapy.

In Saccharomyces cerevisiae, subunit V of the inner mitochondrial membrane protein complex cytochrome c oxidase is encoded by two Cytochrome c oxidase is a key enzyme in the regulation of cellular energy production in eucaryotes (18). As a member of the electron transport chain of the inner mitochondrial membrane, this enzyme catalyzes the reduction of oxygen in a reaction that is essentially irreversible (57, 59). Since this is the only irreversible reaction in the mitochondrial electron transport chain, it is equivalent to the committed step in a metabolic pathway. As such it is an important control point in the regulation of electron flow through the chain (18). At present, it is unclear how eucaryotic cells alter their cytochrome c oxidase activity levels in response to changes in energy demand. Is it by changing (increasing or decreasing) the number of assembled holoenzyme molecules in the inner membrane? Is it by modulating the activity of individual enzyme molecules? Or is it by the assembly of isologous forms of subunits, which have different effects on catalysis, into holoenzyme molecules (4,12,23,24,26)?To study these and related questions, we are examining various aspects of the biogenesis of cytochrome c oxidase in Saccharomyces cerevisiae. This enzyme consists of nine different polypeptide subunits (46); subunits I, II, and III are encoded by mitochondrial DNA, whereas subunits IV, V, VI, VII, Vlla, and VIII are encoded by nuclear DNA. The structural genes encoding all but one of these subunits (subunit VII) have been identified and characterized (6, 11-13, 20, 25, 30, 33, 42, 51, 53, 60, 61). Subunits I, II, and III as well as subunits IV, VI, and VIIa are each encoded by a unique gene. In contrast, subunit V is encoded by a small gene family composed of two nonidentical genes, COXSa and COXSb (12).Previously, we isolated and partially characterized both COXS genes (12, 13) and showed that the protein product of either can assemble into holocytochrome c oxidase. Subsequently, two other laboratories also isolated and sequenced * Corresponding author. (25,51). However, because only partial nucleotide sequences for COXSa and COXSb have been available until now the relation between COXSa and COX5b and this COXS gene has been unclear. In studies described here we demonstrate that COXSa and this recently described COXS gene are identical. We also demonstrate that COXSa and COXSb exist in single copy and are markedly different. They encode proteins that differ in 55 of 154 amino acids, their 5'-and 3'-flanking regions lack homology, and one of them, COXSb, possesses an intron that interrupts its initiation codon. In the accompanying paper (54) we describe the effects of null mutations in COX5a and COXSb on in vivo cytochrome c oxidase activity, cell respiration, and growth on nonfermentable carbon sources. We also demonstrate that these two genes encode interchangeable mature protein subunits that are expressed at markedly different levels. MATERIALS AND METHODSPlasmids, strains, and growth media. The...

In Saccharomyces cerevisiae, the COXSa and COXSb genes encode two forms of cytochrome c oxidase subunit V, Va and Vb. We report here that heme increases COX5a expression and decreases COX5b expression and that the HAP2 and REOI genes are involved in positive regulation of COX5a and negative regulation of COX5b, respectively. Heme regulation of COX5a and COX5b may dictate which subunit V isoform is available for assembly into cytochrome c oxidase under conditions of high-and low-oxygen tension.The recent discovery of different forms (isologs) for some of the nuclear-encoded subunits of cytochrome c oxidase in mammals and yeasts has led to the hypothesis that isologs play a role in altering the functional properties of the holoenzyme (2,9,10,21 The effect of heme deprivation on expression of COXSa and COX5b was tested with strains carrying hem] mutations. hemi mutant strains are heme deficient because of an inability to synthesize b-amino levulinic acid (8-ALA) (7, 23); they can be made heme proficient by the addition of b-ALA to the growth medium. RNA blot analysis (Fig. 1) demonstrated that a hemi mutant strain grown in the presence of b-ALA (lanes 1 and 3) had increased steady-state levels of COXSa transcripts and decreased levels of COX5b transcripts compared with transcript levels of the same strain grown in the absence of b-ALA (lanes 2 and 4).Additional evidence for inverse regulation of COXSa and COX5b by heme was provided by ,-galactosidase activities measured in hemi mutant strains transformed with COXSalacZ and COX5b-lacZ fusion genes, carried on the 2,umbased plasmids pCT5aL and pMC5bL, which were constructed as described previously (21). Cultures carrying these plasmids were grown to exponential phase in synthetic glucose medium (2) lacking uracil (to select for the plasmids) but containing 40 ,ug of histidine, 40 CYCI (8,15,16). The hapl::LEU2 mutation resulted in a small decrease (about 1.5-fold) in COXSa-lacZ expression and had little or no effect on COXSb-lacZ expression in aerobically grown heme-proficient cells. The significance of these small changes is unclear. In contrast, the hap2-1 mutation in strain LGW1 had a dramatic effect on expression of COXSa-lacZ, reducing it about eightfold relative to expression in the isogenic HAP2+ strain, BWG1-7A, but had no effect on the expression of COXSb-lacZ (