Copyright 0 1993 by the Genetics Society of America

PETl 11 Acts in the 5”Leader of the Saccharomyces cereuisiae Mitochondrial COX2 mRNA to Promote Its Translation

Julio J. Mulero* and Thomas D. Fox?

*Section of Biochemistry, Molecular and Cell Biology and ?Section of Genetics and Development, Cornell University, Ithaca, New York 14853-2703

Manuscript received September 10, 1992 Accepted for publication November 13, 1992

ABSTRACT PETl 11 is a yeast nuclear gene specifically required for the expression of the mitochondrial gene

COX2, encoding cytochrome c oxidase subunit I1 (~0x11). Previous studies have shown that PETl 11 activates translation of the COX2 mRNA. To map the site of PETl 11 action we have constructed, in vitro, genes coding for chimeric mRNAs, introduced them into mitochondria by transformation and studied their expression. Translation of a chimeric mRNA with the 612-base 5”untranslated leader of the COX3 mRNA fused precisely to the structural gene for the coxII-precursor protein is independent of PETl 11, but does require a COX3 mRNA-specific translational activator known to work on the COX3 5”leader. This result demonstrates that PETl 11 is not required for any post- translational step. Translation of a chimeric mRNA with the 54-base 5”leader of the COX2 mRNA fused precisely to the structural gene for cytochrome c oxidase subunit I11 was dependent on PETl 11 activity. These results demonstrate that PETl 11 acts specifically at a site in the short COX2 5”leader to activate translation of downstream coding sequences.

M OST of the protein coding genes in mitochon- drial DNA (mtDNA) specify subunits of en-

zymes that carry out respiratory electron transport and oxidative phosphorylation. Expression of these mitochondrial genes depends on many proteins en- coded by nuclear genes [reviewed in ATTARDI and SCHATZ (1988), COSTANZO and Fox (1990), DUJON (1 98 l), PON and SCHATZ (1 99 1) and TZACOLOFF and DIECKMANN (1990)l. Several of these nuclear genes are specifically required for the expression of individ- ual mitochondrial gene products, and in all cases studied to date they exert their specific effects at post- transcriptional steps in mitochondrial gene expres- sion. A subset of these nuclear genes code for mito- chondrial mRNA-specific translational activators [re- viewed in CosTANzo and Fox (1 990) and HINNEBUSCH and LIEBMAN (1 99 l)].

Recessive mutations in the nuclear gene P E T l I I specifically block expression of the mitochondrial gene COX2, encoding cytochorome c oxidase subunit I1 (coxII), and thereby produce a respiration deficient phenotype (CABRAL and SCHATZ 1978). Phenotypic characterization suggests that the defect is at the level of translation since pet1 1 I mutants do not accumulate the cox11 protein but contain substantial amounts of its mRNA (POUTRE and FOX 1987). The PETl 1 1 protein is located in mitochondria (STRICK 1988; STRICK and FOX 1987), and thus acts directly within the organelle.

The requirement for P E T l 1 I function can be by- passed by genetically selected rearrangements in mtDNA that result in translational fusions of other

Genetics 133: 509-516 (March, 1993)

mitochondrial genes to the proximal portion of COX2 (POUTRE and Fox 1987; our unpublished results). All eight of the bypass suppressors characterized to date generate chimeric mRNAs predicted to code for coxII-precursor proteins with extended amino termini (POUTRE and FOX 1987; our unpublished results). However, these novel coxII-precursor proteins all re- tain the processing site between amino acids 15 and 16 (POUTRE and Fox 1987; PRATJE and GUIARD 1986; ourB unpublished results) and are apparently proc- essed, since mature-sized cox11 protein accumulates (POUTRE and Fox 1987). These studies strongly sug- gested that P E T l 1 I is normally required to act at a site encoded by the proximal portion of the COX2 gene. However, since the genetically selected bypass suppressors substituted both new 5”leaders on the COX2 mRNA and new amino termini on the coxII- precursor protein, they did not distinguish between two general models. In the first model, PETl I I might act on the 120x2-mRNA leader to activate translation initiation. Such a scheme applies to the translational activation of at least some other mitochondrial mRNAs (COSTANZO and FOX 1988; RODEL and FOX 1987). In the second model, analogous to the cotrans- lational model for protein translocation across the endoplasmic reticulum (WALTER and LINCAPPA 1986), P E T l 1 I might interact with the nascent amino terminus of the coxII-precursor to antagonize trans- lational arrest and allow continued synthesis of pre- cox11 (POUTRE and Fox 1987).

To distinguish these two models we constructed chimeric genes in vitro that code for chimeric mRNAs

510 J. J. Mulero and T. D. Fox

TABLE 1

Yeast strains used in this study

Strain Genotype Source

JJM 18rho" JJM18 JJM 100 JJM5lrho" JJM5 1 LSF6 JJM 190 JJM 195 JJM 102 DAUP JJM 194 JJM50 PTH44 PTY 7 DBY947 MCCl18

MATa ade2-101 ura3-52 p e t l 11-1 1 [rho"] MATa ade2-101 ura3-52 petl l l - l 1 [rho+] MATa ade2-101 [rho+ cox2-M461 I ] MATa ade2-101 ura3-52 petl 11-1 1 pet494-2 [rho"] MATa ade2-101 ura3-52petll l-11 pet494-2 [rho+] MATa adel op l [rho+ cox3-M7583] MATa ade2-101 ura3-52 pe t l l l -11 [rho- pJMSO] MATa ade2-101 ura3-52petll l-11 pet494-2 [rho- pJM411 MATaade2-I01 ura3-52petlll-I1 [rho+] MATa ade2-101 ura3 [rho+] MATa ade2-101 ura3-52 pe t l l l - I1 pet494-2 [rho- pJMJO] MATa ade2-101 ura3-52 pet1 11-1 1 pet4942 [rho'] MATa ade2-101 ura3-52 pet494-2 [rho+] MATa ade2-101 ura3-52 [rho+ cox2-171 MATa ade2-101 ura3-52 [rho+] MATa lys2 [rho+ cox3-A5]

This study This study This study This study This study L. S. FOLLEY This study This study This study COSTANZO and FOX ( 1 988) This study This study P. T . HAFFTER P. E. THORSNESS NEFF et al. (1 983) FOLLEY and FOX (1 99 1)

with altered 5"leaders but wild-type polypeptide products. These chimeric genes were then returned by genetic transformation to mitochondria (FOX et al. 1991; FOX, SANFORD and MCMULLIN 1988) where their expression could be studied. We demonstrate first, that the 612-base cytochrome c oxidase subunit 111 (COX3) mRNA 5"untranslated leader, precisely attached to the structural gene for the coxII-precur- sor, bypasses PETl 11 function. This result demon- strates that PETl 11 is not required for any step sub- sequent to translation initiation. In addition, we show that expression of a chimeric mRNA with the 54-base COX2 mRNA 5"leader precisely attached to the COX3 structural gene is dependent on PETl 11 activity. Taken together, these results strongly indicate that the PETl 1 1 protein acts specifically at a site in the COX2 5"untranslated mRNA leader to activate trans- lation of downstream coding sequences.

MATERIALS AND METHODS

Yeast strains, media and genetic methods: The strains used in this study are listed in Table 1. All strains are isogenic or congenic to DBY947 (NEFF et al. 1983) except DAU2, MCC118 andJM100 which are isogenic to D273- 1 OB (ATCC 25627). The petl l l - l 1 allele is a double mu- tation isolated fortuitously during in vitro mutagenesis which lacks sequences encoding amino acids 637 through 640 and contains a small insertion at a second site (STRICK 1988). pet11 1-11 exhibits the null phenotype and does not revert easily.

Nonfermentable medium was YPEG (1% yeast extract, 2% peptone, 3% ethanol and 3% glycerol). Glucose-contain- ing medium (YPD) and minimal medium (SD) were pre- pared as described (SHERMAN, FINK and HICKS 1986). Stand- ard genetic procedures were as described (SHERMAN, FINK and HICKS 1986). Transfer of mitochondrial genomes to different nuclear backgrounds was accomplished by cyto- duction (CONDE and FINK 1976). Cytoductants were identi- fied genetically (Fox et al. 1991) following efficient mating of the parental strains (ROGERS and BUSSEY 1978).

Nucleic acid manipulation: Standard DNA manipula- tions were carried out as described (SAMBROOK, FRITSCH

and MANIATIS 1989). DNA sequence analysis was performed by the dideoxynucleotide chain termination method (SAN- GER, NICKLEN and COWLSON 1977). Total yeast DNA and RNA were prepared as described (SHERMAN, FINK and HICKS 1986). DNA-gel-blots were hybridized to 32P-labeled probes of either the pJM30 or the pJM4l construct. RNA- gel-blots were carried out using formaldehyde as a denatur- ant as described (SAMBROOK, FRITSCH and MANIATIS 1989). The probes used for the RNA hybridizations consisted of a COX2 0.45-kb RsaI fragment (CORUZZI and TZAGOLOFF 1979) and a COX3 1.9-kb XbaI fragment (THALENFELD and TZAGOLOFF 1980). Transcript sizes were estimated using the 0.85-kb wild-type COX2 mRNA (BORDONNE, DIRHEIMER and MARTIN 1988), the 1.68-kb 15s mitochondrial rRNA (LI et al. 1982) and the 3.6-kb wild-type COX3 mRNA (THALENFELD, HILL and TZAGOLOFF 1983) as standards.

Site-directed mutagenesis of the COX2 and COX3 genes: The mtDNA for all constructs was derived from strain D273-10B strain (ATCC 25627). T o generate pJM3O (Fig- ure l), we first inserted the COX2 gene into pTZ18U (Bio- Rad) as a 2.5-kb SalI-Hind111 fragment with the Sal1 site upstream of the gene. [This fragment corresponds to the 2.45-kb HpaII fragment of mtDNA (Fox 1979) flanked by polylinker sequences.] Then, the 1 .S-kb HaeIII-Sau3A frag- ment from the COX3 gene (DE ZAMAROCZY and BERNARDI 1986) (containing upstream sequences, including the pro- moter, 5'-untranslated leader and amino-terminal structural gene sequences) was inserted between the SmaI-BamHI sites of the plasmid. Oligonucleotide mutagenesis (Muta-gene kit, Bio-Rad) was performed to attach the 5"untranslated leader sequences of COX3 to the COX2 structural gene using an oligomer whose sequence was complementary to that shown in Figure 1 .

pJM41 (Figure 1) was generated by placing the 2.5-kb SalI-Hind111 fragment carrying COX2 into a Bluescript KS- plasmid (Stratagene) that already contained COX3 on a 3.8- kb XbaI-EcoRI fragment (DE ZAMAROCZY and BERNARDI 1986). T o attach the COX2 5'-untranslated leader to the COX3 structural gene, in vitro site directed mutagenesis was performed with the oligomer whose sequence is shown in Figure 1.

The leader-structural gene junctions in both pJM3O and pJM41 were checked by DNA sequence analysis.

Mitochondrial transformation: The mitochondrial transformation of JJMl8rho" with the plasmid pJM3O was performed by high velocity microprojectile bombardment

Translational Activation by PET1 I I 51 1

S COX3kader COXZcodlng sequence

I - Np"I

S 3' H - "L-

COXZkader COX3codlng sequence

pJM3O Sst I+ Hind 111 uncut "

1 2 3 4 5 6 7 8 m 2 1- Kpn I+ C/a I uncut

~ ~ ~ ~ M G A T T T ~ ~ ~ M A ~ z ~ ~ c ~ c ~ ~ ~ G ~ G ~ ~ G T A -

100 c H

FIGURE 1 ."Schematic diagram of plasmids bearing chimeric mi- tochondrial genes. Large rectangles correspond to structural genes. Smaller rectangles correspond to 5'-untranslated leader coding regions. Solid rectangles represent COX3 sequences. Open rectan- gles represent COX2 sequences. 5' and 3' ends of the predicted chimeric mRNA products are indicated. The written sequence shown for pJM3O is complementary to the oligonucleotide used to generate the fusion (see MATERIAL AND METHODS). The written sequence shown for pJM41 corresponds exactly to the oligonucle- otide used. The underlined ATG sequence indicates the initiation codon of each gene. The wavy lines indicate sequences derived from the vectors. The indicated restriction sitesare SstI ( S ) , Hind111 (H), Kpnl (K) and Clal (C).

(FOX et al. 1991; Fox , SANFORD and MCMULLIN 1988; JOHNSTON et al. 1988) in a Biolistics PDS-1000 (Du Pont Biolistic Particle Delivery System) driven by a gunpowder charge. The mitochondrial transformation of JJM5 1 rhoo with the plasmid pJM41 was achieved with the same appa- ratus using a pressured helium acceleration system. The bombardments were carried out as described (Fox et al. 1991) with an equimolar mixture of each plasmid carrying a chimeric mitochondrial gene and the URA3 shuttle vector YEP352 (HILL et al. 1986).

In vivo labeling of mitochondrial translational prod- ucts: T h e mass mating of the heteroplasmic diploids was performed as described (COSTANZO and FOX 1988; ROGERS and BUSSEY 1978). Yeast cells were labeled with Trans5S- label (ICN Radiochemicals) in the presence of cycloheximide as described (DOUGLAS and BUTOW 1976; FOX et al. 1991). Labeled mitochondrial proteins were separated through a 16% polyacrylamide gel (prepared from a stock solution containing 29.2% acrylamide, 0.8% bisacrylamide) contain- ing 10% glycerol in the presence of 0.1 % sodium dodecyl sulfate (SDS) and autoradiographed.

RESULTS

Construction and delivery of chimeric genes into mitochondria: T o determine whether the site of PET1 1 I action was in the 5"leader of the COX2 mRNA or in the amino-terminal domain of the coxII- precursor protein, we generated in vitro (see MATE- RIALS AND METHODS) two chimeric genes and analyzed their expression in vivo. The first chimeric gene (car- ried on plasmid pJM30; Figure 1) coded for an mRNA with the 5'4eader of the COX3 mRNA fused precisely to the structural gene coding the coxII-precursor protein. The COX3 mRNA 5"leader is known to contain the site (or sites) required for translational

FIGURE 2.-DNA-gel-blot analysis of DNA from mitochondrial transformants. The plasmids pJM30, pJM41 and total yeast DNA from the strains indicated below were subjected to agarose gel electrophoresis and then transferred to nitrocellulose filters. The filters were probed with radioactively labeled DNA of either pJM3O (lanes 1-8) or pJM41 (lanes 9-16). The samples analyzed in lanes 1-4 were cut with the enzymes SstI and HindII1, which have sites only in vector sequences of pJM30 (Figure 1). Cleavage of pJM3O with the enzymes Sstl and Hind111 produces two fragments of 2.9 kb (pTZ18u) and 3.4 kb (COX3-COX2 insert). Lanes 5-8 were not digested. DNA samples were as follows: lanes 1 and 5, plasmid pJM3O; lanes 2 and 6, the recipient strain JJM 1 8rho"; lanes 3 and 7, the synthetic rho- transformant.JJM190; lanes 4 and 8, the rho+ strain, JJM18. The samples analyzed in lanes 9-12 were cut with the enzymes KpnI and Clal, which have sites only in vectnr se- quences of pJM41. Cleavage of pJM41 with the enzymes Kpnl and ClaI produces two fragments of 2.9 kb (Bluescript KS- vector) and 3.8 kb (COX2-COX3 insert). Lanes 13-16 were undigested. The DNA samples were as follows: lanes 9 and 13, the plasmid pJM41; lanes 10 and 14, the recipient strain,JJM51rhoo; lanes 11 and 15, the synthetic rho- transformant,JJM195; lanes 12 and 16, the rho+ strain.JJM51.

activation of downstream coding sequences by the (20x3-specific activators PET54, PET122 and PET494 (COSTANZO and Fox 1988). The second chimeric gene (carried on plasmid pJM41; Figure 1) was reciprocal to the first, encoding an mRNA with the 5"leader of the COX2 mRNA fused precisely to the COX3 struc- tural gene. Both plasmids carried promoters corre- sponding to the leader sequences (BORDONNE, DIR- HEIMER and MARTIN 1988; THALENFELD, HILL and TZAGOLOFF 1983) and 3"processing sites (or tran- scriptional terminators) corresponding to the struc- tural genes (OSINGA et al. 1984; THALENFELD et al. 1983). As shown below, the chimeric genes directed the accumulation of the predicted chimeric mRNAs in mitochondria.

The plasmids bearing these chimeric genes were delivered to mitochondria of rho' strains (lacking en- dogenous mtDNA) by high velocity microprojectile bombardment (see MATERIALS AND METHODS). The COX3-leader, COX2-structural gene chimera (pJM3O) was introduced into the mitochondria of a p e t l l l , [rho'] strain (LTM 18rho"). The synthetic rho-mitochondrial transformants were identified (FOX et al. 1991) by their ability to produce respiring dip- loids, by recombination, when mated to the cox2 mu- tant tester strain, JJM100. The COX24eader, COX3-

512 J. J. Mulero and T. D. Fox

structural gene chimera (pJM4 1) was introduced into the mitochondria of a p e t l l l , p e t 4 9 4 , [rho"] strain UJM5 lrho"). These mitochondrial transformants were identified by mating to the cox3 tester strain, LSF6, and scoring for respiration in the resulting diploids.

The structure of plasmid DNA sequences in the mitochondrial transformant strains obtained, JJM 190 and JJM 195 (Table l), was analyzed by DNA-gel-blot hybridization analysis, using as probes the plasmids pJM3O and pJM41, respectively (Figure 2). As ob- served previously for other plasmids introduced into mitochondria (FOLLEY and Fox 199 1; Fox, SANFORD and MCMULLIN 1988), pJM30 and pJM41 appeared to have become concatemerized in mitochondria, re- sembling standard rho- mtDNAs (Figure 2, lanes 7 and 15). The concatemers appeared to be simple tandem repeats, since cleavage of the DNA with re- striction enzymes that cut in the vector sequences produced fragments with the same mobilities as the those of the original plasmids (Figure 2, lanes 3 and

Translation of the COX3-leader, COX2-structural gene chimeric mRNA is independent of PETl 11 function: If P E T l 11 normally promotes translation by acting through a site in the COX2 mRNA, then the requirement for P E T l I 1 function should be bypassed in strains containing a chimeric mRNA composed of the COX3-leader and COX2-structural gene. On the other hand, if P E T 1 1 1 normally functions to allow continued COX2 translation by acting through a site in the amino-terminal domain of the pre-cox11 pre- cursor protein, then translation of such a chimeric nlRNA, which encodes the wild-type coxII-precursor protein, should still require P E T l 11 activity.

To determine whether the presence of the COX3- leader, COX2-structural gene chimeric mRNA would bypass the requirement for PETI I 1 activity, the p e t l 11, [rho-, pJM301 transformant JJM19O was mated to a pe t l 11 , [rho+] strain UJM 102), generating pet l 1 I l p e t l l I homozygous zygotes that contained both pJM3O and wild-type mtDNA in a heteroplasmic state. The resulting diploid was respiratory-compe- tent, indicating that the presence of the COX3-leader on the chimeric mRNA bypassed the requirement for PETI 11. [The respiratory-competency of the diploid was mitotically unstable, as has been previously ob- served for other similar heteroplasmic strains (COS- TANZO and Fox 1986; MULLER et al. 1984; POUTRE and Fox 1987).]

RNA-gel-blot hybridization analysis was carried out to confirm that the heteroplasmic diploid contained the predicted chimeric mRNA bearing the COX3- leader (Figure 3; MATERIALS AND METHODS). The wild- type COX2 transcript is 0.85 kb (BORDONNE, DIRHEI- MER and MARTIN 1988) while the predicted size of the chimeric COX3-leader, COX2-structural gene

11).

wt c o n , mRNA

1 2 3 4

wtc0x3+ mRNA - &

FIGURE 3.-Analysis of chimeric mRNA transcripts by RNA-gel- blot hybridization. Total RNA was isolated from the following strains and subjected to electrophoresis (see MATERIALS AND METH- ODS) as follows: lanes 1 and 3, the wild-type strain, DAU2; lane 2, the petl 1 llpetlll, PET4941pet494 [rho', rho- pJM301 diploid, JJM194 XJMlOZ;lane4, thepet494/pet494,PETlll/petlll [rho+, rho- pJM411 diploid.JM195 X PTH44. Lanes 1 and 2 were probed with a labeled 0.45-kb Rsal fragment internal to the COX2 structural gene. Lanes 3 and 4 were probed with a 1.9-kb Xbal fragment carrying the entire COX3 structural gene and flanking sequences.

mRNA is approximately 1.46 kb. Hybridization of a labeled COX2 probe to total RNA from wild-type (strain DAU2) and from a p e t l 1 l / p e t l l I diploid het- eroplasmic for wild-type mtDNA and pJM30 revealed the 0.85-kb wild-type transcript in both strains. As expected, the heteroplasmic diploid (Figure 3, lane 2) also contained a prominent RNA species with the predicted size of the chimeric mRNA, supporting the idea that it was the presence of the COX3-leader that allowed cox11 synthesis in the absence of PET111 function.

The level of the wild-type COX2 mRNA in the pet1 1 l / p e t l l l diploid was low relative to that of the chimeric mRNA. We have not directly investigated this phenomenon since it does not affect the interpre- tation of our results. However, it is probably due to the fact that populations of cells heteroplasmic for rho+ and rho- genomes generally contain many more copies of the rho- mtDNA than the rho+ (POUTRE and FOX 1987). In addition, the stability of the wild-type COX2 mRNA is apparently decreased in the absence of translation (POUTRE and Fox 1987).

Translation of mitochondrial mRNAs bearing the COX3-leader depends on at least three specific activa- tors (COSTANZO and Fox 1988). To further confirm that the bypass of P E T l 1 I function was due to trans- lational activation from within the COX3-leader of the chimeric mRNA, we asked whether translation of this chimeric mRNA was dependent upon one of the COX3-specific translational activators, PET494. T o ac- complish this, a p e t l I I , pe t494 , [rho-, pJM3OI strain, JJM194, was mated with the p e t l l l , p e t 4 9 4 , [rho+] strain, JJM50, generating a population of diploid zyg- otes homozygous for the nuclear mutations and het-

Translational Activation by PETl 1 I 513

cox II c

cox 111 - FIGURE 4.-Analysis of mitochondrial translation products. Mi-

tochondrial translation products were labeled in vivo and separated by SDS-polyacrylamide gel electrophoresis as described in MATE- RIA= AND METHODS. The positions of the cox11 and cox^^^ proteins are indicated. The cells labeled were either defined haploid strains, or zygote populations formed by mass mating (MATERIALS AND METHODS), as indicated below. Lane 1, the rho+, ~0x2-17 strain, PTY7; lane 2, p e t l l l / p e t l l l , P E T 4 9 4 l p e t 4 9 4 [rho+, rho- pJMSO] zygotes (JJM194 XJM102); lane 3, p e t l l l / p e t l l I , p e t 4 9 4 1 p e t 4 9 4 [rho+, rho-pJMSO] zygotes (JJM194 xJJM50); lane 4, the wild-type strain DBY947; lane 5 , the rho+, cox3-A5 strain, MCCl18; lane 6, pet4941pet494, PET1 1 I lpe t l I I [rho+, rho- pJM4 I] zygotes (JJM 195 X PTH44);lane7,pet494/pet494,petlll/petlll [rho+,rho-pJM41] zygotes (JJM195 XJJMSO).

eroplasmic for wild-type mtDNA and pJM30. The unstable mitochondrial heteroplasmic state can only be maintained during growth by selection for respi- ration. Since the homozygouspet494 mutation in these diploids prevents respiration by blocking synthesis of cytochrome c oxidase subunit 111, these diploids could not be grown selectively. We therefore monitored transient expression of COX2 in zygotes immediately after mass mating by labeling mitochondrial transla- tion products in the presence of cycloheximide (AN- ZIANO and BUTOW 1991 ; COSTANZO and Fox 1988). (Cycloheximide inhibits cytoplasmic, but not mito- chondrial, protein synthesis.) As a positive control we performed a parallel mass mating experiment in which JJM 194 was mated to the petl I I , PET494, [rho+] strain, JJM102. The labeled proteins were analyzed by gel electrophoresis and autoradiography (Figure 4; MATERIALS AND METHODS). The control zygotes that were homozygous for the petl 1 I mutation but heter- ozygous for wild-type PET494 were able to synthesize the cox11 protein as expected (Figure 4, lane 2). However, zygotes homozygous for both the petl 11 and pet494 mutations were unable to translate the cox11 protein (Figure 4, lane 3). Thus, PET494 was required to activate translation of the cox11 protein from the chimeric mRNA bearing the COX3-leader.

Translation of the COX2-leader, COX3-structural gene chimeric mRNA is dependent on PETl 11 func-

tion: If PET111 normally promotes translation of downstream coding sequences by acting through a site in the COX2 mRNA 5’-leader, then translation of a chimeric mRNA with the COX2-leader fused to the COX3 structural gene should be independent of PET494 but dependent on PETl I I activity.

T o test first whether the presence of a COXZ-leader, COX3-structural gene chimeric mRNA would allow cox111 translation in the absence of PET494 activity, the pe t1 11, pet494, [rho-, pJM411 mitochondrial trans- formant strain JJM 195 was mated to the pet494, [rho+] strain, PTH44. The resulting diploid had a functional PETl 11 gene, lacked PET494 activity and contained both pMJ41 and wild-type mtDNA. This diploid was respiratory-competent, indicating that the presence of the chimeric mRNA bypassed the requirement for PET494. As expected, this heteroplasmic diploid con- tained both the wild-type 3.6-kb COX3 mRNA (THAL- ENFELD, HILL and TZACOLOFF 1983), and a novel transcript of the size, 3.0 kb, predicted for the chi- meric mRNA specified by pJM41 (Figure 3, lane 4).

T o determine next whether translation of the COXZ-leader, COX3-structural gene chimeric mRNA was dependent on PET1 11 activity, we asked whether cox111 synthesis was blocked in zygotes homozygous for petl I 1 and pet494 mutations, and heteroplasmic for wild-type mtDNA and pJM41. To perform this experiment we employed the same scheme of mass mating followed by in vivo labeling described above. The petl I I, pet494, [rho-, pJM411 mitochondrial transformant strain JJM 195 was mated to the petl 11, pet494 [rho+] strain, JJM50. Analysis of mitochondrial translation products from the resulting zygotes re- vealed that they were unable to synthesize cox111 protein (Figure 4, lane 7). Control zygotes, containing a wild-type PETl 11 gene (formed by mating JJM 195 with PTH44) did synthesize cox111 (Figure 4, lane 6). Therefore, PETl 11 is required to activate the trans- lation of this chimeric mRNA bearing the COX2- leader sequence.

DISCUSSION

Translation of at least five of the eight major yeast mitochondrial mRNAs requires, or appears to re- quire, mRNA-specific activation by the products of nuclear genes (ACKERMAN et al. 199 1 ; COSTANZO and Fox 1986; COSTANZO and Fox 1988; COSTANZO, SEAVER and Fox 1986; DECOSTER et al. 1990; Fox et al. 1988; KLOECKENER-GRUISSEM, MCEWEN and POY- TON 1988; MULLER et a/ . 1984; PAYNE, SCHWEIZER and LUKINS 199 1; POUTRE and FOX 1987; RODEL 1986; RODEL and FOX 1987). In the two cases studied most intensively to date, the nuclearly coded transla- tional activators have been shown genetically to work through sites in the long 5’-leader-s of the COX3 (COSTANZO and FOX 1988) and COB (RODEL 1986; RODEL and FOX 1987) mitochondrial mRNAs, which

514 J. J. Mulero and T. D. Fox

\ “+y

PET111 +

FIGURE 5.-Model for PET1 I I activation of COX2 nlRNA trans- lation. The wavy line represents the COX2 mRNA, with the coding sequence in bold. Experiments described in this paper map the site of PET1 I I action to the 54-base 5‘-untranslated leader of the mitochondrially coded COX2 mRNA.

are 6 12 bases (KLOECKENER-GRUISSEM, MCEWEN and POYTON 1988; M. C. COSTANZO, unpublished results) and 954 bases (DIECKMANN and MITTELMEIER 1987) in length, respectively. Such long untranslated 5’- leaders are typical of the major yeast mitochondrial mRNAs (GRIVELL 1989).

The COX2 mRNA is unique among major mito- chondrial messages in having a short 5’4eader of only 54 bases (BORDONNE, DIRHEIMER and MARTIN 1988), suggesting that mRNA-specific activation of COX2 translation by the nuclear gene P E T l 1 I might have been mechanistically different from that of COX3 and COB. Consistent with this possibility was the fact that analysis of genetically selected mitochondrial muta- tions bypassing PETI 1 I function had failed to rule out a model for PETI I I action in which it recognized the nascent amino terminus of the coxII-precursor to allow continued elongation (POUTRE and Fox 1987). Precedent for such a model is provided by the mech- anism of cotranslational protein translocation across the endoplasmic reticulum (WALTER and LINGAPPA 1986).

The experiments reported here show that the re- quirement for P E T l 11 in translation of the COX2 mRNA coding sequence can be circumvented by plac- ing the COX3 mRNA 5’-untranslated leader in front of the COX2 structural gene. Translation of this chi- meric mRNA is dependent on the nuclear gene PET494, a translational activator known to work spe- cifically on the COX3 5”leader. Furthermore, mito- chondrial translation of the COX3 mRNA coding se- quences becomes dependent on the activity of the PET1 11 gene when the 5‘-untranslated leader of the COX2 mRNA is attached to the COX3 structural gene. As expected, translation of this chimeric mRNA does not require the COX3-mRNA-specific activator, PET494. Taken together, these results place the site of action of PET11 I within the 54-base 5’-untrans- lated leader of the COX2 mRNA (Figure 5), ruling out any important role for interaction between

P E T l I I and the nascent amino terminus of the coxII- precursor.

The PETl 1 1 protein appears to be located in mi- tochondria since a fusion protein consisting of the amino-terminal 154 amino acids of PETl 1 1 fused to ,%galactosidase was specifically associated with mito- chondria (STRICK and Fox 1987). Furthermore, when overproduced, the PETl 1 1 protein was immunologi- cally detectable only in mitochondria and was tightly associated with the inner membrane (STRICK 1988). Therefore we assume that PETl 11 works within the organelle to activate translation of the COX2 mRNA. An attractive hypothesis is that the PETl 11 protein may bind specifically with the COX2 mRNA 5’-leader, but this has not been demonstrated. An abundant nuclearly coded yeast mitochondrial protein has been shown to bind the 5’-untranslated leaders of mito- chondrial mRNAs, including the COX2 mRNA, but its role in mitochondrial gene expression remains unclear ( DEKKER, PAPADOPOULOU and GRIVELL 199 1 ; DEKKER et al. 1992; PAPADOPOULOU et al. 1990).

One of the three COX3-mRNA-specific translational activators that function on the COX3 mRNA 5’- leader, PET122, has been shown to interact function- ally with the small subunit of mitochondrial ribo- somes: mutations that truncate the PET122 carboxy terminus are suppressed allele specifically by muta- tions in at least three nuclear genes encoding mito- chondrial small subunit ribosomal proteins (HAFFTER and Fox 1992; HAFFTER, MCMULLIN and Fox 1990, 1991 ; MCMULLIN, HAFFTER and Fox 1990). These findings suggest that at least some yeast mitochondrial translational activators may function by mediating the interaction between their respective mRNAs and the ribosomal small subunit (HAFFTER, MCMULLIN and Fox 199 1). The fact that the P E T l 11 site of action has now been mapped to the COX2 mRNA 5’4eader indicates that it too could function in this way.

We thank MARIA COSTANZO, LINDA FOLLEY, PASCAL HAFFTER and PETER THORSNESS for their gifts of plasmids and strains. J.J.M. was a recipient of a National Science Foundation predoctoral fel- lowship and partially supported by a fellowship from the Cornell Biotechnology Program, which is sponsored by the New York State Science and Technology Foundation, a consortium of industries, the U.S. Army Research Office and the National Science Founda- tion. This work was supported by a grant (GM 29362) to T.D.F. from the National Institutes of Health.

LITERATURE CITED

ACKERMAN, S . H., D. L. GATTI, P. GELLEFORS, M. G. DOUGLAS and A. TZAGOLOFF, 1991 A T P 1 3 a nuclear gene of Sacchnromyccs cerevisiae essential for the expression of subunit 9 of the mito- chondrial ATPase. FEBS Lett. 2 7 8 234-238.

ANZIANO, P. Q., and R. A. BUTOW, 1991 Splicingdefective mu- tants of the yeast mitochondrial COX1 gene can be corrected by transformation with a hybrid maturase gene. Proc. Natl. Acad. Sci. USA 8 8 5592-5596.

AITARDI, G., and G. SCHATZ, 1988 Biogenesis of mitochondria. Annu. Rev. Cell Biol. 4: 289-333.

BORDONNE, R., G . DIRHEIMER and R. P. MARTIN,

Translational Activation by PET1 11 515

1988 Expression of the axil and maturase-related RFI genes in yeast mitochondria. Curr. Genet. 13: 227-233.

CARRAL, F., and G. SCHATZ, 1978 Identification of cytochrome c oxidase subunits in nuclear yeast mutants lacking the functional enzyme. J. Biol. Chem. 253: 4396-4401.

CONDE, J., and G. R. FINK, 1976 A mutant of S. cerevisiae defec- tive for nuclear fusion. Proc. Natl. Acad. Sci. USA 73: 3651.

CORUZZI, G., and A. TZAGOLOFF, 1979 Assembly of the mito- chondrial membrane system: DNA sequence of of subunit I1 of yeast cytochrome c oxidase. J. Biol. Chem. 254: 9324-9330.

COSTANZO, M. C., and T. D. Fox, 1986 Product of Saccharomyces cerevisiae nuclear gene PET494 activates translation of a specific mitochondrial mRNA. Mol. Cell. Biol. 6 3694-3703.

COSTANZO, M. C., and T. D. Fox, 1988 Specific translational activation by nuclear gene products occurs in the 5’ untrans- lated leader of a yeast mitochondrial mRNA. Proc. Natl. Acad. Sci. USA 8 5 2677-2681.

COSTANZO, M. C., and T . D. Fox, 1990 Control of mitochondrial gene expression in Saccharomyces cerevisiae. Annu. Rev. Genet. 24: 91-113.

COSTANZO, M. C., E. C. SEAVER and T. D. FOX, 1986 At least two nuclear gene products are specifically required for translation of a single yeast mitochondrial mRNA. EMBO J. 5: 3637- 3641.

DE ZAMAROCZY, M., and G. BERNARDI, 1986 The primary struc- ture of the mitochondrial genome of Saccharomyces cerevisiae- a review. Gene 47: 155-177.

DECOSTER, E., M. SIMON, D. HATAT and G. FAYE, 1990 T h e MSS51 gene product is required for the translation of the COX1 mRNA in yeast mitochondria. Mol. Gen. Genet. 224: 1 1 1- 118.

DEKKER, P. J. T., B. PAPADOPOWLOW and L. A. GRIVELL, 1991 Properties of an abundant RNA-binding protein in yeast mitochondria. Biochimie 73: 1487-1492.

DEKKER, P. J. T., J. STUURMAN, K. VAN OOSTERUM and L. A. GRIVELL, 1992 Determinants for binding of a 40 kDa protein to the leaders of yeast mitochondrial mRNAs. Nucleic Acids Res. 2 0 2647-2655.

DIECKMANN, C. L., and T. M. MITTELMEIER, 1987 Nuclearly- encoded CBPl interacts with the 5’ end of mitochondrial cytochrome b pre-mRNA. Curr. Genet. 12: 391-397.

DOUGLAS, M., and R. A. BUTOW, 1976 Variant forms of mito- chondrial translation products in yeast: evidence for location of determinants on mitochondrial DNA. Proc. Natl. Acad. Sci. USA 73: 1083-1096.

DUJON, B., 1981 Mitochondrial genetics and functions pp. 505- 635 In: The Molecular Biology of the Yeast Saccharomyces, L fe Cycle and Inheritance, edited by J. N. STRATHERN, E. W. JONES

and J. R. BROACH, Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.

FOLLEY, L. S., a n d T . D. Fox, 1991 Site-directed mutagenesis of a Saccharomyces cerevisiae mitochondrial translation initiation codon. Genetics 1 2 9 659-668.

FOX, T. D., 1979 Genetic and physical analysis of the mitochon- drial gene for subunit I1 of yeast cytochrome c oxidase. J. Mol. Biol. 1 3 0 63-82.

Fox, T. D., J. C. SANFORD a n d T . W. MCMULLIN, 1988 Plasmids can stably transform yeast mitochondria lacking endogenous mtDNA. Proc. Natl. Acad. Sci. USA 85: 7288-7292.

Fox, T. D., M. C. COSTANZO, C. A. STRICK, D. L. MARYKWAS, E. C. SEAVER and J. K. ROSENTHAL, 1988 Translational regula- tion of mitochondrial gene expression by nuclear genes of Saccharomyces cerevisiae. Philos. Trans. R. SOC. Lond. Ser. B 319: 97-105.

Fox, T. D., L. S. FOLLEY, J. J. MULERO, T . W. MCMULLIN, p. E. THORSNESS, L. 0. HEDIN and M. C. COSTANZO, 1991 Analysis and manipulation of yeast mitochondrial genes. Methods En- zymol. 194 149-165.

GRIVELL, L. A., 1989 Nucleo-mitochondrial interactions in yeast

mitochondrial biogenesis. Eur. J. Biochem. 182: 477-493. HAFFTER, P., a n d T . D. FOX, 1992 Suppression of carboxy-ter-

minal truncations of the yeast mitochondrial mRNA-specific translational activator PET122 by mutations in two new genes, MRP17 and PET127. Mol. Gen. Genet. 235: 64-73.

HAFFTER, P., T . W. MCMULLIN and T. D. Fox, 1990 A genetic link between an mRNA-specific translational activator and the translation system in yeast mitochondria. Genetics 125: 495- 503.

HAFFTER, P., T. W. MCMULLIN a n d T . D. FOX, 1991 Functional interactions among two yeast mitochondrial ribosomal proteins and an mRNA-specific translational activator. Genetics 127: 319-326.

HILL, J. E., A. M. MYERS, T . J. KOERNER and A. TZAGOLOFF, 1986 YeastlE. coli shuttle vectors with multiple unique re- striction sites. Yeast 2: 163-167.

HINNEBUSCH, A. G., and S. W. LIEBMAN, 1991 Protein synthesis and translational control in Saccharomyces cerevisiae, pp. 627- 735 in The Molecular and Cellular Biology of the Yeast Saccha- romyces: Genome Dynamics, Protein Synthesis and Energetics, Vol. 1, edited by J. R. BROACH, J. R. PRINGLE and E. W. JONES. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.

JOHNSTON, S. A,, P. Q. ANZIANO, K. SHARK, J. C. SANFORD and R. A. BUTOW, 1988 Mitochondrial transformation in yeast by bombardment with microprojectiles. Science 240: 1538-1541.

KLOECKENER-GRUISSEM, B., J. E. MCEWEN and R. 0. POYTON, 1988 Identification of a third nuclear protein-coding gene required specifically for posttranscriptional expression of the mitochondrial COX3 gene in Saccharomyces cerevisiae. J. Bacte- riol. 1 7 0 1399-1402.

LI, M., A. TZAGOLOFF, K. UNDERBRINK-LYON and N. C. MARTIN, 1982 Identification of the paromomycin-resistance mutation in the 15s rRNA gene of yeast mitochondria. J. Biol. Chem. 257: 5921-5928.

MCMULLIN, T . W., P. HAFFTER a n d T . D. FOX, 1990 A novel small subunit ribosomal protein of yeast mitochondria that interacts functionally with an mRNA-specific translational ac- tivator. Mol. Cell. Biol. 10: 4590-4595.

MULLER, P. P., M. K. REIF, S. ZONGHOU, C. SENGSTAG, T . L. MASON a n d T . D. FOX, 1984 A nuclear mutation that post- transcriptionally blocks accumulation of a yeast mitochondrial gene product can be suppressed by a mitochondrial gene re- arrangement. J. Mol. Biol. 175: 431-452.

NEFF, N. F., J. H. THOMAS, P. GRISAFI and D. BOTSTEIN, 1983 lsolation of the P-tubulin gene from yeast and demon- stration of its essential function in vivo. Cell 33: 21 1-219.

OSINGA, K. A., E. DE VRIES, G. VAN DER HORST and H. F. TABAK, 1984 Processing of yeast mitochondrial messenger RNAs at a conserved dodecamer sequence. EMBO J. 3: 829-834.

PAPADOPOULOU, B., P. DEKKER, J. BLOM and L. A. GRIVELL, 1990 A 40 kd protein binds specifically to the 5’-untranslated regions of yeast mitochondrial mRNAs. EMBO J. 9: 4135- 4143.

PAYNE, M. J., E. SCHWEIZER and H. B. LUKINS, 1991 Properties of two nuclear pet mutants affecting expression of the mito- chondrial olil gene of Saccharomyces cerevisiae. Curr. Genet. 1 9 343-351.

PON, L., and G. SCHATZ, 1991 Biogenesis of yeast mitochondria, pp. 333-406 in The Molecular and Cellular Biology of the Yeast Saccharomyces: Genome Dynamics, Protein Synthesis and Energetics, Vol. 1 , edited by J. R. BROACH, J. R. PRINGLE and E. W. JONES. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.

POUTRE, C. G., and T. D. FOX, 1987 P E T I I I , a Saccharomyces cerevisiae nuclear gene required for translation of the mito- chondrial mRNA encoding cytochrome c oxidase subunit 11. Genetics 115: 637-647.

PRATJE, E., and B. GUIARD, 1986 One nuclear gene controls the removal of transient pre-sequences from two yeast proteins:

516 J. J. Mulero and T. D. Fox

one encoded by the nuclear the other by the mitochondrial genome. EMBO J. 5: 1313-1317.

RODEL, G., 1986 Two yeast nuclear genes, CBSZ and CBS2, are required for translation of mitochondrial transcripts bearing the 5'-untranslated COB leader. Curr. Genet. 11: 41-45.

RODEL, G., and T. D. Fox, 1987 The yeast nuclear gene CBSZ is required for translation of mitochondrial mRNAs bearing the cob 5'-untranslated leader. Mol. Gen. Genet. 206 45-50.

ROGERS, D., and H. BUSSEY, 1978 Fidelity of conjugation in Saccharomyces cereuisiae. Mol. Gen. Genet. 162 173-182.

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Cloning. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.

SANCER, F., S. NICKLEN and A. R. COULSON, 1977 DNA sequenc- ing with chain-terminating inhibitors. Proc. Natl. Acad. Sci.

SHERMAN, F., G. R. FINK and J. B. HICKS, 1986 Methods in Yeast Genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.

STRICK, C. A., and T. D. FOX, 1987 Saccharomyces cereuisiae posi- tive regulatory gene PET2 2 2 encodes a mitochondrial protein

USA 74: 5463-5467.

that is translated from an mRNA with a long 5' leader. Mol. Cell. Biol. 7: 2728-2734.

STRICK, C.A., 1988 A study of the expression and protein product of the yeast positive regulatory gene PETZZZ. Ph.D. Thesis, Cornell University, Ithaca.

THALENFELD, B. E., J. HILL and A. TZAGOLOFF, 1983 Assembly of the mitochondrial membrane system: characterization of the oxi2 transcript and localization of its promoter in Saccharomyces cereuisiae D273-10B. J. Biol. Chem. 258: 610-615.

THALENFELD, B. E., and A. TZAGOLOFF, 1980 Assembly of the mitochondrial membrane system; sequence of the Oxi2 gene of yeast mitochondrial DNA. J. Biol. Chem. 255: 6173-6180.

THALENFELD, B. E., S. G. BONITZ, F. G. NOBRECA, G. MACINO and A. TZACOLOFF, 1983 olil transcripts in wild type and in cytoplasmic "petite" mutant of yeast. J. Biol. Chem. 258: 14065-14068.

TZACOLOFF, A., and C. L. DIECKMANN, 1990 PET genes of Sac- charomyces cereuisiue. Microbiol. Rev. 5 4 21 1-225.

WALTER, P., and V. R. LINCAPPA, 1986 Mechanism of protein translocation across the endoplasmic reticulum membrane. Annu. Rev. Cell Biol. 2 499-516.

Communicating editor: M. JOHNSTON