Protein Biosynthesis in Mitochondria: Past Simple, Present Perfect, Future Indefinite

Mitochondria are obligate organelles of most eukaryotic cells that perform many different functions important for cellular homeostasis. The main role of mitochondria is supplying cells with energy in a form of ATP, which is synthesized in a chain of oxidative phosphorylation reactions on the organelle inner membrane. It is commonly believed now that mitochondria have the endosymbiotic origin. In the course of evolution, they have lost most of their genetic material as a result of genome reduction and gene transfer to the nucleus. The majority of mitochondrial proteins are synthesized in the cytosol and then imported to the mitochondria. However, almost all known mitochondria still contain genomes that are maintained and expressed. The processes of protein biosynthesis in the mitochondria — mitochondrial translation — substantially differs from the analogous processes in bacteria and the cytosol of eukaryotic cells. Mitochondrial translation is characterized by a high degree of specialization and specific regulatory mechanisms. In this review, we analyze available information on the common principles of mitochondrial translation with emphasis on the molecular mechanisms of translation initiation in the mitochondria of yeast and mammalian cells.

Mitochondria are important organelles in almost all eukaryotic cells. Their main function is supplying cells with the energy, which is provided in the form of ATP synthesized in the reactions of oxidative phosphorylation. Mitochondria also participate in the biosynthesis of FeS clusters, metabolism of amino acids and nucleotides, syn thesis of lipids and steroids, and regulation of pro grammed cell death [1]. According to the widely accept ed endosymbiotic theory, mitochondria have originated exogenously from a prokaryotic precursor similar to mod ern alpha proteobacteria [2]. Mitochondria are semi autonomous organelles -during billions of years of evo lution, mitochondria of various groups of eukaryotes have lost a significant portion of genetic material of their pre cursor, preserving only a small fraction of their genome. Mitochondrial proteome is represented mostly by the proteins encoded in the nucleus, synthesized in the cyto plasm, and imported to the mitochondria [3].
Mitochondrial genomes of modern eukaryotes are usual ly significantly smaller than the bacterial ones and con tain a relatively small number of genes. Due to a signifi cant divergence, different groups of eukaryotes differ considerably in the mitochondrial genome composition and number of mitochondrial genes. The smallest mito chondrial genome (~6000 bp) has been found in the malaria infectious agent Plasmodium falciparum, while the total size of the multichromosomal mitochondrial DNA of the carnation family plant Silene conica exceeds 10,000,000 bp [4,5].
The mitochondrial genome of baker's yeast (Saccharomyces cerevisiae) contains ribosomal RNA genes, complete set of tRNA genes, and genes for eight proteins -ribosomal protein Var1p and seven hydropho bic core proteins of the electron transport chain com plexes [6]. Mammalian mitochondrial genomes contain genes coding for rRNAs, tRNAs, and 13 proteins, com ponents of the oxidative phosphorylation complexes [7]. At the same time, mitochondrial genomes of jakobids (Protozoa) have the largest number of mitochondrial genes among the known organisms and include genes of several ribosomal proteins, four subunit RNA poly BIOCHEMISTRY (Moscow) Vol. 85 No. 3 2020 merase, a reduced form of transfer messenger RNA, RNA component of RNase P, and proteins involved in the oxidative phosphorylation chain assembly, as well as rRNA and tRNA genes [8]. A significant degree of evolu tionary divergence is indicated by the fact that the num ber of mitochondrial genes in different eukaryotes does not correlate with the size of mitochondrial genome. In particular, human mitochondrial genome contains 37 genes and has a size of 16,569 bp, whereas S. cerevisiae mitochondria genome contains approximately the same number of genes but consists of ~80,000 bp, i.e., has the size similar to that of the mitochondrial genome of the abovementioned jakobids that includes ~70 genes [6 8].
Mitochondria of different eukaryotes also display signifi cant differences in the mechanisms of gene expression (transcription and translation).
In this review, we present the latest discoveries in the studies of protein biosynthesis in yeast and mammalian mitochondria with special emphasis on the mechanisms of the translation initiation regulation. Without trying to comprehensively review all the known facts on the mito chondrial translation, we discuss existing contradictions and recently introduced novel concepts in this field of the studies.

PROTEIN BIOSYNTHESIS IN MITOCHONDRIA
Protein biosynthesis is carried out by ribosomesmolecular machines that translate genetic information encrypted in mRNA and catalyze formation of peptide bonds between amino acids [9]. A ribosome id formed by the small and large subunits. The small subunit decodes mRNA, and the large subunit catalyzes the synthesis of the polypeptide chain. Despite the common principles of mRNA decoding and protein synthesis, the structure of ribosomes and the mechanisms of ribosome functioning differ not only between prokaryotes and eukaryotes, but also between bacteria and archaea [10]. Similarly, signifi cant differences between the ribosomal structure and translation regulatory mechanisms have been observed for the mitochondrial ribosomes of different groups of organ isms. For a long time, mitochondrial translation had remained the terra incognita of molecular biology; from the moment of its discovery, it has been assumed that pro tein biosynthesis in the mitochondria is similar to the one in bacteria. However, functional and, to a greater degree, structural studies of the last decade have significantly pro moted our understanding of mitochondrial translation [11 13]. It has become obvious that this process is char acterized by certain features that distinguish it from simi lar processes in bacteria or cytosol of eukaryotic cells. Mitochondrial translation has been investigated most comprehensively in S. cerevisiae and mammalian cells.
At present, bacterial origin of mitochondrial ribo somes causes no doubts [14]. In the process of evolution, mitochondrial ribosomes have significantly diverged from their bacterial precursors in both spatial structure and composition of ribosomal RNAs and ribosomal proteins. Explosive methodological development of cryo electron microscopy in recent years has led to the resolving the structures of mitochondrial ribosomes from many organ isms (Trypanosoma brucei [15], S. cerevisiae [16], Sus scrofa [12], Homo sapiens [11], Arabidopsis thaliana [17], and Brassica oleracea var. botrytis [18]). Comparison of these structures has revealed that even if mitoribosomes from different groups of eukaryotes differ significantly, they exhibit some common features. First of all, in com parison to the bacterial or cytoplasmic ribosomes, mito chondrial ribosomes are characterized by a significantly higher protein content (relative to RNA), often carry insertions and deletions in rRNA, and possess a specific set of proteins. Most likely, this is related to the partial replacement of RNA functions by the proteins, as well as to the fact that mitochondria synthesize only a limited number of hydrophobic proteins that are incorporated into the inner mitochondrial membrane as components of the oxidative phosphorylation complexes [19]. Protein biosynthesis in the mitochondria is strictly coupled with the synthesis of mitochondrial proteins in the cytoplasm, which is essential for correct assembly of the oxidative phosphorylation complexes [20]. Below, we present the data on the translation regulation in the mitochondria in yeast and mammalian cells (as the two most investigated systems) at the stage that is most susceptible to regulation, i.e., translation initiation.

PAST SIMPLE
As mentioned above, mitochondrial translation has a bacterial origin, which leads to the logical suggestion that protein biosynthesis initiation in the mitochondria should be somewhat similar to that in eubacteria. Initiation of bacterial translation has been investigated in great detail. The process starts with the formation of the 30S transla tion initiation complex in which the start codon of mRNA interacts with the CAU anticodon of the initiator formylmethionine tRNA (fMet tRNA) at the P site of the small ribosomal subunit. Three initiation factors (IF1, IF2, and IF3) play the most important role in the initia tion complex formation; each of these proteins is essen tial for accurate and efficient translation initiation. All three factors interact with the corresponding sites on the 30S subunit and perform their specific functions. IF2 in a complex with GTP binds fMet tRNA, thus forming a ter nary complex that interacts with the 30S subunit. IF1 binds to the 30S subunit and prevents the access of aminoacyl tRNA to the A site. It also increases the affin ity of the IF2/GTP/fMet tRNA ternary complex to the 30S subunit. IF3 binds to the 30S subunit and prevents its association with the 50S subunit, as well as enables the accuracy of recognition of the fMet tRNA start codon in the P site. The complex of 30S with IF1, IF3, and IF2/GTP/fMet tRNA binds short polypurine sequence in the 5′ untranslated region (5′ UTR) of mRNA (Shine-Dalgarno sequence) due to the presence of par tially complementary sequence in 16S rRNA. This bind ing is followed by the recognition of the start codon by the fMet tRNA anticodon, GTP hydrolysis to GDP, dissoci ation of the initiation factors, and association of the small and large ribosomal subunits. The resulting 70S initiation complex represents an associated ribosome with the fMet tRNA positioned in the P site on the start AUG codon of mRNA and the A site containing the second codon that expects arrival of the corresponding amino acyl tRNA. The initiation factors IF1 and IF2 are con sidered universal and highly conserved, as their function al and structural homologs have been found in all investi gated bacteria and archaea [21].
Translation initiation in the mitochondria and, espe cially, the role and functions of mitochondrial initiation factors, have been studied to a much lesser extent. In par ticular, mitochondria of all investigated organisms lack IF1, while IF2 is absolutely universal and IF3 is almost universal [22]. It is likely that the mitochondrial initiation factor 2 (mtIF2) plays a key role in the translation initia tion in mitochondria, and its functions will be discussed in detail further in this review.

PRESENT PERFECT
Bacterial IF2 is a GTPase consisting of six domains (I VI), where domain IV hydrolyzes GTP and domain VI interacts directly with fMet tRNA [23,24]. IF2 selects the initiator aminoacyl tRNA and facilitates association of ribosomal subunits during translation initiation (the latter function is evolutionary conserved in archaea and eukaryotes). Unlike bacteria, mammalian mitochondria do not have specialized methionine tRNA for the transla tion initiation but utilize the same tRNA used for methio nine incorporation during elongation [25]. A portion of Met tRNAs is formylated by a special enzyme; the result ing fMet tRNA has a high affinity to mtIF2 and low affinity to the mitochondrial elongation factor EF Tu. This dual function of Met tRNA has been also found for T. brucei mitochondria which do not encode tRNAs required for the mitochondrial translation but import them from the cytoplasm [26].
It has been shown that human mtIF2 can function ally replace not only IF2, but also both IF2 and IF1 in Escherichia coli. Moreover, it has been demonstrated that the function of IF1 is carried out by a relatively small unique 37 a.a. domain located between the domains II and III in mtIF2 [27,28]. The structure of the mtIF2 complex with porcine mitochondrial ribosome was resolved by Kummer et al. [29] by cryo electron microscopy. The authors confirmed that the abovemen tioned 37 a.a. domain was located in the A site of the small subunit. It prohibited the transfer of aminoacyl tRNA to this site and prevented ribosome sliding over mRNA, i.e., functioned similarly to the bacterial IF1. Furthermore, the authors have been able to show that mtIF2 does not directly interact with mRNA and does not participate in its recruitment to the ribosome. It was established for the COX3 mRNA that the pentatricopep tide repeat (PPR) containing mitoribosomal protein mS39 located in the region of the mRNA entry tunnel participated in the mtIF2 association with the mitoribo some [29]. Hence, it was suggested that mRNA interac tion with the mitochondrial ribosome is provided by either individual mitoribosomal proteins or specialized translation factors.
Mitochondrial IF2 from S. cerevisiae encoded by the IFM1 gene has been investigated to a much lesser extent. Despite the fact that mIF2 was identified about 30 years ago [30], almost no structure functional studies have been conducted on this protein. It was found that deletion of IFM1 results in the cessation of yeast growth on non fermentable carbon sources [30]. It was also demonstrat ed in vitro that recombinant mIF2 forms a ternary com plex with GTP and fMet tRNA (or met tRNA) which associates with the 30S subunit of bacterial ribosome in the presence of synthetic mRNAs. Recombinant yeast mtIF2 was capable of GTP hydrolysis in the content of the complex, as well as prevented non enzymatic hydro lysis of the indicated tRNAs [31]. Nevertheless, strict necessity of this factor for the translation initiation in the yeast mitochondria has not been demonstrated.
Bacterial translation initiation factor IF3 performs two important functions: it associates with the ribosomal 30S subunit following its dissociation after termination of translation, thus preventing association of the 30S and 50S subunits, and participates in the selection of tRNAs and mRNAs during initiation of a new round of protein synthesis by specifically destabilizing incorrectly formed codon-anticodon pairs [32,33]. IF3 is universal and conserved in bacteria but is absent from the mitochondria of some organisms [22]. The functions of IF3 in the cyto plasm of eukaryotes are performed by the multisubunit factor eIF3 that does not display noticeable homology with bacterial IF3 [34].
Bacterial IF3 is a globular protein consisting of two well defined N and C terminal domains connected by an α helical linker [35]. The C terminal domain interacts directly with 16S rRNA of the small ribosomal subunit, while the N terminal domain can assume several confor mations and provides interaction with fMet tRNA [23]. Experiments on the deletion of IF3 fragments have demonstrated that isolated C terminal domain could per form all functions characteristic for the full size IF3, while the N terminal domain provides additional strength to the IF3 binding to the ribosome and tRNA [36].
The amino acid sequence of mammalian mitochon drial translation initiation factor 3 (mtIF3) exhibits only limited homology to the sequence of the bacterial factor; however, these proteins are structurally similar, differing only in the presence of short N and C terminal exten sions in the mitochondrial factor [37]. In vitro studies have demonstrated that mammalian mtIF3 performs functions similar to those of the bacterial protein. In par ticular, mtIF3 is capable of dissociating mitochondrial ribosomes and facilitates formation of the initiation complex together with mtIF2 [37]. Similarly to the bac terial IF3, mammalian mtIF3 interacts with the mitori bosome predominantly via the C terminal domain with some contribution of the linker sequence [38]. Moreover, mammalian mtIF3 also destabilizes the initiation com plexes that do not contain mRNA or contain incorrect tRNA, although the latter activity is pronounced signifi cantly less than in the bacterial IF3 [39,40]. The N and C terminal extensions play an important role in the mtIF3 functioning. Thus, deletion of the C terminal extension results in the inability of mtIF3 to dissociate incorrect initiation complexes, while deletion of the N terminal extension significantly increases the mtIF3 affinity to the 30S subunit [39,40]. In 2019, Koripella et al. reported [41] the structure of the mammalian mtIF3 complex with the mitochondrial ribosome resolved by cryo electron microscopy. It was found that in contrast to bacterial IF3, virtually all structural sites in mtIF3 participate in the interactions with the mitoribosome except the C terminal extension. According to the mod eling data, this structural element occupies the P site during interaction of the mitoribosomal small subunit with the ternary mtIF2/GTP/fMet tRNA complex in the absence of mRNA, thus preventing formation of the initiation complex without a template [41]. In the same study, the authors suggested the mtIF3 might be involved in the mRNA recruitment to the forming initiation com plex.
For a long time, all attempts to identify the third translation initiation factor in S. cerevisiae mitochondria had been unsuccessful. However, recent studies have convincingly identified Aim23p protein as the mtIF3 in the yeast. The similarity between the ternary structures of Aim23p and mammalian mtIF3 was demonstrated using bioinformatics analysis; it was also found that the effect of the Aim23p gene deletion could be suppressed via expression of human and Schizosaccharomyces pombe mtIF3 proteins [22]. Aim23p can associate with the small subunits of the yeast mitochondrial ribosome [42], as well as with the bacterial ribosome, causing their dissoci ation. Interestingly, in the case of bacterial ribosomes, this dissociation occurs via an unusual mechanismafter binding to the associated 70S ribosome, Aim23p interacts with both 30S and 50S subunits, facilitating for mation of the intermediate dissociation state with the sedimentation coefficient of ~60S [43]. Both N and C terminal extensions of Aim23p have been found to be functionally important: deletion of these fragments results in the factor inactivation, while a hybrid protein obtained by addition of these sequences to IF3 from E. coli is fully active in the yeast mitochondria [44]. However, the most intriguing fact is that Aim23p is not essential for the mitochondrial translation. Deletion of the AIM23 gene results only in the reduced efficiency of COX1 and COX2 biosynthesis in the mitochondria, which phenotypically is manifested as slower adaptation of yeasts to the growth on non fermentable carbon sources, but does not affect translation of other mito chondrial mRNAs [45]. This fact is quite surprising con sidering that the presence of IF3 is absolutely necessary for bacterial translation.
As mentioned above, most mitochondrial genes encode components of the oxidative phosphorylation chain. Therefore, mitochondrial translation should be coordinated with the cytosolic biosynthesis of the nuclear DNA encoded subunits of these complexes. Regulation of mitotranslation remains poorly understood despite a significant progress achieved in the recent years. The mechanisms of translation regulation in S. cerevisiae mitochondria are the ones most extensively studied. A distinctive feature of this unique system is the presence of translational activators -groups of proteins that establish the efficiency of translation of one or another mRNA by interacting with the extended 5′ UTRs capable of form ing the secondary structure. Translational activators have been discussed in detail in the recent reviews [46,47].
The principles behind the regulation of biosynthesis of 13 proteins encoded in the mammalian mitochondrial genome, all of which are components of the oxidative phosphorylation chain complexes, are less obvious. First of all, this is due to the differences in the structure of mitochondrial mRNAs in mammals. These mRNAs lack the 5′ UTRs [48], which precludes an existence of trans lational activators similar to those found in yeast cells. So far, only one translational activator (TACO1) was identi fied in mammalian mitochondria, that determines the efficiency of cytochrome c oxidase (COX1) biosynthesis; however, the mechanism of its action remains unknown [49].
Recent structural studies have shed the light on the mechanisms of recognition of mRNAs lacking the leader sequence by the mitochondrial ribosomes [29]. The structure of the initiation complex of mammalian mito chondrial ribosome with the COX3 mRNA assembled in vitro was resolved by cryo electron microscopy. It has been established that translation initiation of this mRNA is mediated mostly by interaction of the PPR protein mS39 with the U rich region after the seventh mRNA codon, which is conserved in all 11 mRNAs from the mammalian mitochondria [29]. These data, however, do not explain how the efficiency of translation initiation of any mRNA is determined.

FUTURE INDEFINITE
To conclude our review, we will first summarize the data presented above. It had been assumed before that due to an indisputable origin of mitochondria from a bac terial precursor, mitochondrial translation is organized in a way similar to that in prokaryotes. However, the studies of recent 10 15 years have demonstrated that despite their common origin, the systems of protein biosynthesis in modern prokaryotes and mitochondria are very different, as well as the systems of mitochondrial translation in dif ferent groups of eukaryotes. First of all, these differences are manifested in the structure of mitochondrial ribo somes, which are specialized for the biosynthesis of most ly hydrophobic proteins encoded in the mitochondrial genome. The latter could explain the evolutionary origin of systems for the regulation of mitochondrial translation, the most studied of which involves translational activators in S. cerevisiae. The lack of similar system in mammalian mitochondria is due to significant differences in the mRNA structure (i.e., the absence of 5′ UTRs). Nevertheless, the existence of translation regulation in mammalian mitochondria is beyond doubts, because the biosynthesis of proteins encoded in the mitochondrial genome must be strictly coordinated with the biosynthe sis of other components of the electron transport chain in the cytosol. It is likely that one of the mechanisms of mitotranslation regulation involves coupling with the assembly of respiratory complexes in the inner mitochon drial membrane, although, this explanation is more rele vant for the regulation of the elongation process, rather than translation initiation.
In this connection, of special interest are the data on the mitochondrial translation in yeast in the absence of the third initiation factor, which has little effect on the biosynthesis of the majority of mitochondrially encoded proteins [45]. In essence, the profile of mitotranslation in the case of AIM23 deletion is similar to the one observed when a translational activator of a corresponding mRNA (e.g., Pet111p, an activator of the COX2 mRNA) is absent [50]. Based on these data, we suggest that mitochondrial evolution resulting in the loss of a majority of ancestral genes and mitotranslation specialization for the synthesis of a very limited number of hydrophobic proteins, as well as the need for a strict regulation of translation, have caused initiation factors to lose their versatility and become regulators of biosynthesis of individual proteins. Apparently, this hypothesis requires verification, for example, by investigating mitochondrial translation in the cells lacking genes of mitochondrial translation initiation factors. At the same time, it does not contradict present ly available data, because all structural and functional studies of initiation factors have been performed in vitro, and the only in vivo model -absence of Aim23pdemonstrated that the third translation initiation factor is not essential for the protein biosynthesis in the mito chondria. Apparently, this observation could be an excep tion characteristic only for the yeast system and this par ticular factor. However, it is our opinion that the question whether translation initiation factors are necessary for the biosynthesis of all mitochondrial proteins should be elu cidated experimentally. Currently, these experiments are underway in our laboratory. Our hypothesis is corroborat ed in part by the recently reported data by Rudler et al. [51] who have demonstrated that deletion of mouse mtIF3 gene (Aim23p ortholog) does not abolish protein synthesis in the mitochondria but results only in its quan titative imbalance.
Funding. This work was supported by the Russian Science Foundation (project 17 14 01005; testing hypotheses on the evolution of mitochondrial translation apparatus) and Russian Foundation for Basic Research (project 19 14 50206).
Conflict of interest. The authors declare no conflict of interest in financial or any other sphere.
Ethical approval. This article does not contain description of studies with human participants or animals performed by any of the authors.
Open access. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits unrestricted use, distri bution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.