Identification and initial characterization of translational initiation factor 2 from bovine mitochondria.

The bovine liver mitochondrial factor that promotes the binding of fMet-tRNA to mitochondrial ribosomes, initiation factor 2 (IF-2mt), has been identified in the postribosomal supernatant fraction of isolated liver mitochondria. This factor has been purified approximately 5,000-fold and present preparations are estimated to be about 10% pure. IF-2mt has an apparent molecular weight of about 140,000 as determined by gel filtration chromatography. IF-2mt is active in stimulating fMet-tRNA binding to Escherichia coli ribosomes but E. coli IF-2 is not active in promoting initiator tRNA binding to animal mitochondrial ribosomes. The IF-2mt-mediated binding of fMet-tRNAi(Met) to mitochondrial ribosomes is dependent on the presence of a message such as poly(A,U,G) and on GTP. Nonhydrolyzable analogs of GTP are 2-3-fold less effective in promoting initiation complex formation on mitochondrial ribosomes than is GTP suggesting that IF-2mt is capable of recycling to some extent under the current assay conditions.


Identification
and Initial Characterization of Translational Initiation Factor 2 from Bovine Mitochondria" ( The bovine liver mitochondrial factor that promotes the binding of fMet-tRNA to mitochondrial ribosomes, initiation factor 2 (IF-2,,J, has been identified in the postribosomal supernatant fraction of isolated liver mitochondria. This factor has been purified approximately 5,000-fold and present preparations are estimated to be about 10% pure. IF-2,t has an apparent molecular weight of about 140,000 as determined by gel filtration chromatography. IF-2,t is active in stimulating fMet-tRNA binding to Escherichia coli ribosomes but E. coli IF-2 is not active in promoting initiator tRNA binding to animal mitochondrial ribosomes. The IF-2,,-mediated binding of fMet-tRNAFt to mitochondrial ribosomes is dependent on the presence of a message such as poly(A,U,G) and on GTP. Nonhydrolyzable analogs of GTP are 2-3-fold less effective in promoting initiation complex formation on mitochondrial ribosomes than is GTP suggesting that 1F-2,t is capable of recycling to some extent under the current assay conditions. Although it has been known for many years that mitochondria contain their own translational system, a great deal remains to be learned about the process of chain initiation in this organelle. The mechanism of initiation complex formation in prokaryotes is quite different from the process observed in the eukaryotic cell cytoplasm (1,2). In E. co& IF-2l directs the binding of fMet-tRNA to 30 S ribosomal subunits in a message-dependent reaction. In the eukaryotic cell cytoplasm, eIF-2 forms a ternary complex with Met-tRNA and GTP and this complex is then transferred to the 40 S ribosomal subunit in a message-independent pathway. Mitochondrial protein synthesis resembles the bacterial system in the use of the formylated initiator tRNA (3); however, this system has a number of unique features that distinguish it from other translational systems. Animal mitochondrial DNA has a single gene coding for tRNAMet and this gene is thought to give rise, in an unknown manner, to the synthesis of both initiator and elongator tRNAMet species (4). This unusual feature has not been found in any other known mRNA binding to small subunits requires the presence of fMet-tRNA while in the eukaryotic cytoplasmic system, the initial interaction of the mRNA with the 40 S subunit is mediated by initiation factors which bind to the cap structure of the message (2,5). In contrast, the pathway for the assembly of the initiation complex in animal mitochondria appears to be quite different from the pathways found in either the prokaryotic or eukaryotic cytoplasmic systems. The small (28 S) subunit of bovine mitochondrial ribosomes has the ability to bind mRNAs tightly in a sequence and initiation factorindependent manner (6). Efficient binding appears to require mRNAs that are several hundred nucleotides in length.' In addition, it has recently been shown (7)  ion-exchange chromatography as described (11). E. coli W ribosomes were prepared as described (12). Assays on E. coli ribosomes were carried out as described (8).

Miscellnneous
Procedures-Protein concentrations were determined by the method of Lowry et al. (13) or by the method of Sedmak and Grossberg (14) with modifications by Bearden (15) using bovine serum albumin as a standard.

AND DISCUSSION
One of the difficulties facing the development of an animal mitochondrial system that will promote the binding of the initiator tRNA to mitochondrial ribosomes arises because the initiator tRNA from this organelle has not been isolated and no information is available on its aminoacylation or formylation. To circumvent this problem, we have used yeast tRNAp which has been charged by the E. coli methionyl-tRNA synthetase and subsequently formylated by the bacterial transformylase.
We have chosen to work with this tRNA because an examination of the single tRNAMet gene present in bovine mitochondrial DNA (4) indicates that it has features that are slightly closer to the eukaryotic cytoplasmic initiator tRNA than to E. coli tRNAp.
Assays of initiation complex formation in prokaryotic and eukaryotic cytoplasmic systems are often carried out using the AUG triplet alone since this trinucleotide is readily bound by the ribosomes from these sources. It has been reported (16) that AUG, by itself, will not bind to bovine mitochondrial ribosomes. However, the polynucleotide poly(A,U,G) will bind to 28 S subunits at high concentrations' and we chose to use it as the mRNA in our initial attempts to identify an initiation factor that would promote the binding of fMet-tRNAy to mitochondrial ribosomes (IF-2,J. We have sought to detect IF-2,t using a postribosomal supernatant fraction (SlOO) from purified mitochondria as the starting material. The rationale for seeking to detect IF-2,t in the SlOO fraction was as follows. First, mitochondrial ribosomes are generally prepared in buffers containing 300 mM KC1 since ribosomes prepared under these ionic conditions are the most active in poly(U)-directed polymerization and in the ability to bind natural mRNAs. Under these ionic conditions, the initiation factors are not likely to remain associated with the ribosomal fraction. Mitochondrial ribosomes have a much lower RNA content than other ribosomes (17), hence, electrostatic interactions between translation factors and these ribosomes are probably not as strong as those observed in prokaryotic or eukaryotic cytoplasmic systems. Second, when prepared under our standard conditions, mitochondrial ribosomes have an extremely low background activity in fMet-tRNA binding suggesting that factors essential for this process have been removed during their preparation.
Assays designed to test the ability of a postribosomal supernatant to promote fMet-tRNA binding to mitochondrial ribosomes failed to detect any activity. This observation is not surprising since one would expect that the concentration of any putative initiation factor would be quite low in these extracts.
In addition, these extracts are toxic and appear to degrade the input tRNA even in the presence of RNasin. In order to enrich for possible initiation factors, a portion of the SlOO was subjected to chromatography on DEAE-cellulose. Less than 20% of the protein in these extracts was retained by this resin and the protein could be eluted by buffers containing 0.5 M KC1 (Table I). When this material was tested for the ability to promote yeast fMet-tRNAp binding to mitochondrial ribosomes in the presence of poly(A,U,G), a significant enhancement of binding was detected using a nitrocellulose filter binding assay described under "Experimental Procedures" (data not shown). The binding observed was dependent on the presence of mitochondrial ribosomes indicating that the assay was not simply detecting a complex between fMet-tRNA and some protein in this crude preparation. About 100 units of IF-2,, could be obtained from 25 g of washed bovine liver mitochondria.
The amount of IF-2,t observed correlates well with the amount expected based on the ribosome content of this organelle. It is estimated that there are only about 100 ribosomes in each mitochondrion (18). In comparison, about 15,000 units of E. coli IF-2 would be obtained from a comparable weight of bacterial cells and there are about 25,000 ribosomes/cell in a culture of E. coli growing exponentially in rich media. To confirm that the fMet-tRNA binding observed was actually occurring on mitochondrial ribosomes, reaction mixtures were analyzed on sucrose density gradients. As indicated in Fig. 1, fMet-tRNA can be seen associated with both the 28 S subunit and the 55 S monosome fraction after density gradient analysis. Essentially no binding of fMet-tRNA was observed in the absence of IF-2,, (data not shown). These observations indicate that mitochondrial ribosomes will bind the yeast initiator tRNA when it has been formylated and that both 28 and 55 S complexes can be formed in the presence of this mitochondrial initiation factor. There is no change in the absorbance profile of the mitochondrial ribosome and its subunits following incubation with this mitochondrial fraction indicating that it does not contain a ribosome dissociation factor comparable to E. coli IF-3. The binding of fMet-tRNA observed is not a fortuitous reaction promoted by mitochon- Reaction mixtures were prepared as described under "Experimental Procedures" and then analyzed on sucrose density gradients. The amount of fMet-tRNA present in each fraction was quantitated and the absorbance at 254 nm was used to determine the position of elution of the 28, 39, and 55 S particles. drial EF-Tu/Ts since these preparations are free of this elongation factor (data not shown). In addition, purified mitochondrial EF-Tu/Ts will not interact with yeast fMet-tRNA.3 E. coli IF-Z occurs in two forms with molecular weights of about 80,000 and 100,000 (19) while the cytoplasmic factor eIF-2 exists primarily as a M, 145,000 trimeric protein (2). In contrast, the only other characterized organellar IF-2, chloroplast IF-2 from Euglena grucilis, is a complex protein with multiple high molecular weight forms4 In order to assess the approximate size and complexity of IF-2,t, we subjected a portion of the DEAE-cellulose preparation of this factor to gel filtration chromatography.
As indicated in Fig. 2, IF-2,t elutes as a symmetrical peak with an apparent molecular weight of about 140,000. This observation indicates that the 3 C. Schwartzbach, unpublished observations. a Ma, L., and Spremulli, L. (1990) J. Biol. Chem., in press. mitochondrial factor has a size close to that observed with the corresponding bacterial and cytoplasmic factors. Further work will be required in order to determine whether it is composed of a single polypeptide chain or whether it is an oligomeric protein.
IF-2,t has been further purified by chromatography on a TSKgel SP-5PW column. The majority of the protein in the starting material is not retained by this resin and IF-2,t can be purified nearly lOOO-fold by this procedure (Table I). We estimate that we have purified IF-2,t about 5000-fold over the original SlOO fraction and that it is about 10% pure. The more highly purified preparations of IF-2,t are quite labile. Samples are fast-frozen in small aliquots and each aliquot is generally thawed and used only once.
We have examined a number of properties of the partially purified IF-2,,. As indicated in Fig. 3, the binding observed is completely dependent on the presence of poly(A,U,G). It should be noted that fMet-tRNA binding catalyzed by E. coli IF-2 is also mRNA-dependent (5) while the eIF-2-dependent binding of the initiator tRNA to 40 S subunits can occur in the absence of a message (2). Hence, in this respect, IF-2,* appears to resemble its prokaryotic counterpart. Attempts to replace poly(A,U,G) with natural mRNA such as that for cytochrome oxidase subunit II in the fMet-tRNA binding assay have not yet been successful. Presumably, other factors are required to promote initiation complex formation with natural mRNAs in mitochondria as in other systems. It has been reported (16) that the trinucleotide AUG does not bind to mitochondrial ribosomes. This idea is in agreement with our own work which indicates that there is a minimum length required for the efficient binding of RNAs to these ribosomes.' To our surprise, however, AUG is as efficient as poly(A,U,G) in promoting the IF-Smt-dependent binding of fMet-tRNA to mitochondrial ribosomes (data not shown). This observation suggests that AUG can interact with these ribosomes but that its stable association requires the presence of codon:anticodon interactions.
Both monovalent and divalent cations effect the IF-2,t dependent binding of fMet-tRNA to mitochondrial ribosomes. As indicated in Fig. 4A, the Mg2' optimum for this reaction is about 7.5 mM. This value is comparable to that observed for the binding of mRNAs to these ribosomes (6). Monovalent cations are generally inhibitory to the binding assay (Fig. 4B) and the lowest concentrations that can be tested (35 mM) give the best binding results. This observation suggests that there may be important electrostatic interactions that stabilize the mitochondrial initiation complex. The binding of fMet-tRNA to mitochondrial ribosomes is sensitive to the temperature of incubation.
The reaction proceeds somewhat better at 27 "C than at 37 "C. Substantially less binding is observed at 0 "C than at elevated temperatures (data not shown). The binding of the initiator tRNA in both prokaryotic and eukaryotic cytoplasmic systems is dependent on the presence of GTP. We have tested whether this nucleotide will promote initiation complex formation in the mitochondrial system. AS indicated in Fig. 5, essentially no fMet-tRNA binding can be observed in the absence of GTP and this nucleotide promotes the binding reaction over lo-fold. GDP can not replace GTP in this reaction (Fig. 5). A similar observation has been made with other translational systems. In general, GTP hydrolysis appears to be important for the recycling of translation factors rather than for the binding reaction itself. One way to determine whether IF-2,t is acting catalytically or stoichiometrically under our assay conditions is to investigate the effect of nonhydrolyzable nucleotides on the observed binding. The use of the nonhydrolyzable analog GMP-PNP reduces the Reaction mixtures were prepared as described under prepared as described under "Experimental Procedures" and con-"Experimental Procedures" except that they contained 0.4 ASO 6). This observation suggests that IF-2,t may be functioning catalytically and is used several times during the course of the assay. In E. coli, subunit joining is required for GTP hydrolysis and the subsequent release of IF-2 (5). The observation that a portion of the fMet-tRNA bound is present on 55 S particles ( Fig. 1) would agree with this idea. It has recently been observed (7) that mitochondrial 28 S subunits have a binding site for guanine nucleotides and it will be interesting to investigate the possible role of subunit bound nucleotides in IF-2,,-promoted binding of fMet-tRNA to mitochondrial ribosomes. Finally, we have examined whether IF-2,t will promote initiation complex formation on prokaryotic ribosomes as well as on its homologous ribosomes. As indicated in Fig. 7A, this factor is actually more active when tested on bacterial ribosomes than when tested on mitochondrial ribosomes. This difference probably simply reflects the quality of the ribosome preparations currently available or a difference in the ability of the factor to recycle on the E. coli ribosomes. In contrast, E. coli IF-2 is not active in fMet-tRNA binding on mitochondrial ribosomes (Fig. 7B)  EF-G is active on E. coli ribosomes5 while the corresponding prokaryotic factor is not active on mitochondrial ribosomes (10). As indicated above, we have detected and partially characterized the first mitochondrial initiation factor reported to date. This factor has several properties that are analogous to its prokaryotic counterpart.
The binding of fMet-tRNA to ribosomes promoted by IF-2,t is dependent on the presence of a mRNA such as poly(A,U,G) and requires the presence of guanine nucleotides.
The observation that efficient binding cannot be carried out in the presence of a natural mRNA in the present system, suggests that additional factors may be required to position the initiation codon located at the 5' end of these mRNAs into the correct site of the ribosome. We are currently seeking to detect and characterize additional initiation factors from this organelle.