Structural aspects of mitochondrial translational apparatus

Structures of Mitochondrial Ribosomes

The cryo-EM structures of two mitoribosomes (Fig. 1), studied so far, have revealed several of their unique features. Both mitoribosomes are found to be highly porous structures [14,15], as compared to their bacterial counterpart despite similar molecular masses (Table 1). The porosity in structure is primarily due to the absence of several bacterial rRNA segments in those mitoribosomes [see ref 16•, for comparative secondary structures of rRNA]. Even though the proportion of r-proteins is significantly increased as compared to that in the bacterial ribosome [20,21], many of new protein masses occupy new spatial positions in the mitoribosome structures. Thus, there is limited structural compensation for missing rRNA segments by new, or enlarged homologs of bacterial, r-proteins. For example, in the mammalian (Bos taurus) mitoribosome, only ~20% of missing rRNA segments is structurally replaced by mitochondrial r-proteins (MRPs) [14], while such a compensation is much higher (>50%) in the protistan (Leishmania tarantolae) mitoribosome [15].

Due to significant reduction in the size of their rRNAs, both subunits of the mitoribosome possess several tunnel-like features that appear to connect the inter-ribosomal subunit space with the bulk solvent. However, the functional core of both mitoribosomes, i.e., the regions involved in mRNA decoding on the SSU [22] and peptide-bond formation on the LSU [23], are predominantly made up of conserved rRNA segments, even though there are greater occurrences of protein masses on the interface side of both their subunits as compared to that in the subunits of the cytoplasmic ribosomes. Many of these protein masses participate in the formation of protein-protein inter-subunit bridges, and paths of mRNA and tRNA interactions [14,15]. The rRNA scaffolds within the mitoribosome structures are highly shielded by MRPs, such that the solvent-facing surfaces of the mitoribosomal subunits are predominantly made up of MRPs. This situation is in sharp contrast to the pattern of distribution of rRNA and r-protein masses within the structures of bacterial or chloroplast ribosomes, where proteins are present in relatively small patches on the solvent-side surfaces (Fig. 1) [16•]. Interestingly, most (11 out of 12) bacterial 23S rRNA segments that form the tRNA-exit site (E site) [24] are absent in the mito-LSU rRNAs [14,15,25,26], implying that the E site is either absent or very weak in both these mitoribosomes and that the deacylated mitochondrial tRNA (tRNA_mt) may not stay on either of these two mitoribosomes after its release from the ribosomal peptidyl site (P site) during the translation elongation cycle [27]. These unique features of the mitoribosome structures suggest that the modus operandi of these ribosomes has diverged significantly from their cytoplasmic or chloroplast counterparts.

Structure of the Mammalian Mitoribosome

In addition to the common features of the mitoribosome structures described in the previous section, here we highlight some of the unique features of the mammalian mitoribosome structure. Due to the absence of bacterial SSU rRNA helices 16 and 17 that are not compensated by MRPs, the mammalian mitoribosomal SSU has a narrower shoulder and body regions and a more exposed translation factor-binding region, as compared to its bacterial counterpart. However, the body of the mammalian mitoribosomal SSU is significantly elongated due to the addition of MRPs to than in its lower body region. In addition, a triangular gate-like structure, made entirely of MRPs, partially covers the mRNA entry site on the solvent side of the mitoribosomal SSU [14] (Fig. 1B). More recent 7.0 Å resolution cryo-EM maps of the mammalian mitoribosome (MR Sharma, TM Booth, ME Haque, LL Spremulli, RK Agrawal, in preparation) suggest that this gate-like feature is a highly dynamic structure, and may be involved in recruiting the generally leaderless mitochondrial mRNAs [28] to the mitoribosomal SSU.

On the mammalian mitoribosomal LSU, the central protuberance is much larger in size as compared to that in a bacterial ribosome, even though density for the 5S rRNA has not been detected within the cryo-EM map of the mitoribosomal LSU. [It should be noted that only two rRNAs, 12S and 16S, corresponding to the mitoribosomal SSU and LSU, respectively, are encoded by the mammalian mitochondrial genome.] However, a recent biochemical study suggests that the 5S rRNA is transported into the mitochondria along with MRP L18 [29]. Thus, the possibility that the 5S rRNA is present only in a very small subset of isolated mitoribosomal population, and therefore goes undetected in the averaged cryo-EM maps, cannot be ruled out. In all of the cryo-EM maps, a mito-specific MRP emerges from the inter-subunit face of the central protuberance of the LSU and contacts the P-site tRNA_mt at its T-loop side [14]. This extended structure, referred to as the P-site finger, is a dynamic feature that may stabilize and regulate the binding and movements of generally short mammalian tRNA_mts [30]. Another important structural aspect of the mammalian mitoribosomal LSU is its nascent polypeptide exit tunnel, which has been separately described later in this article.

Structure of the Mammalian Mitochondrial Translation Initiation Factor 2

Steps in the mechanism of mammalian mitochondrial translation initiation are somewhat similar to those in a bacterial cell. However, of the three bacterial initiation factors (IF1, IF2, and IF3), homologs of only two, IF2_mt and IF3_mt, have been identified in the mammalian mitochondria [31]. Both these factors show significant alterations in their structural organization as compared to their bacterial counterparts [32,33•]. Interestingly, IF2_mt, which stimulates the binding of fMet-tRNA to the SSU of the mitoribosome, lacks domains I and II of E. coli IF2 [34] but contains a functionally important 37 amino-acid insertion domain (Fig. 2A).

Cryo-EM maps of the IF2_mt•GMPPNP•fMet-tRNA in complex with both the 70S bacterial ribosome [35••] and the 55S mitoribosome (MR Sharma, AS Yassin, ME Haque, LL Spremulli, RK Agrawal, in preparation) have been determined. The overall binding positions of conserved domains of IF2_mt to the ribosome are similar to those of the bacterial IF2 [36]. However, the density corresponding to the insertion domain protrudes out from the rest of the IF2_mt density (Fig. 2B) to partially occupy the ribosomal aminoacyl-tRNA binding site (A-site) on the SSU, overlapping with the binding site of bacterial IF1 [37]. Thus, this structural study [35••] shows that IF2_mt possesses dual functions corresponding to bacterial IF1 and IF2, corroborating a previous genetic study [38]. Despite the fact that there are significant size mismatch and no sequence homology or structural similarity between the insertion domain (37 amino acid) of IF2_mt and the bacterial IF1 (71 amino acids) (Fig. 2C,D), both structures bear similar surface charge distributions, which might play an important role in recognition of the same ribosomal binding pocket [35••]. The presence of long linkers between the insertion domain and the rest of the IF2_mt molecule [35••,39] suggests that the relative position of the insertion domain might vary between its native and ribosome-bound states, as also observed for the bacterial release factors [40]. An integration of a relatively small IF1-mimicking feature onto the IF2_mt might have occurred during the course of evolution to serve a dual purpose: (i) one less protein had to be transported into the mitochondrial matrix, and (ii) to ensure an efficient transport of the IF1 feature to the mitochondrial protein synthesis site within highly dense mitochondrial matrix [41].

The Polypeptide-Exit Tunnel and Its Interaction with mtIM

In contrast to the variety of possible destinations for the nascent polypeptide chain synthesized by the cytoplasmic and chloroplast ribosomes, all of the nascent chains synthesized by the mammalian mitoribosome are inserted into the mtIM, and it is most likely that these polypeptides are inserted co-translationally into mtIM [18,19]. The nascent chain exit tunnel in the mitoribosomes appears to be tailor made for this purpose [16•]. The exit tunnel in the mitoribosome LSU has two openings on the solvent side, one corresponding to the conventional polypeptide exit site (PES) [42], and the second site at a distance of ~25 Å from the PES. This site of premature exposure of the nascent chain is located closer to the peptidyl-transferase center and allows free access to the solvent, and is designated as the polypeptide-accessible site [PAS, 14,15]. PAS is formed because the majority of the rRNA domain I and a significant portion of domain III of LSU rRNA, which together form the inner lining of the lower portion of the exit tunnel in bacterial and archaeal ribosomes [42], are absent in the mitoribosome and are not compensated by MRPs. The solvent-side openings of both PES and PAS are predominantly encircled by mito-specific MRPs [16•], which may be involved in the adherence of the mitoribosome to the mtIM [43,44] through either PES (Fig. 3A) or PAS, such that the lower portion of the mitoribosome is embedded into the mtIM with PAS and PES exposed at opposite sides of mtIM (Fig. 3B).

In yeast, Oxa1p, an integral membrane protein [45] and a mitochondrial homolog of bacterial YidC [46], interacts with the nascent chain while it is still on the ribosome. The human homolog of Oxa1p is known as Oxa1L [47]. The C-terminal tail of Oxa1L (Oxa1L-CTT) is exposed to the mammalian mitochondrial matrix and has been cross-linked to the mitoribosomal LSU [48]. Interestingly, Oxa1L-CTT does not interact with the conserved components (homologs of bacterial r-proteins L22, L23, L24, and L29) of conventional PES. Instead, it crosslinks to other homologs of bacterial r-proteins (L13, L20, and L28) and several mito-specific MRPs (MRPL48, MRPL49, and MRPL51), implying that these MRPs are situated close to the PAS, which would then be close to the matrix side of the mtIM (as depicted in Fig. 3B). A recent study showed that a mitochondrial protein ATAD3 can interact with both the mitochondrial DNA nucleoid and mitoribosome, as well as with mtIM [49•]. Another study showed an interaction between MRPL7 and the mitochondrial RNA polymerase [50•]. Both these studies suggest a coupling between mitochondrial transcription and translation machineries, and strengthen the view that mitoribosomes are either situated very close to, or partially embedded into, the mtIM. Based on these findings it is tempting to suggest that each mitoribosome could be associated with its own mitochondrial DNA (as there are more than 1000 copies of DNA in a given mitochondrion) to facilitate the synthesis and mtIM insertion of a defined set of polypeptide(s) and to ensure the high level of accuracy and efficiency needed to assemble the complexes of oxidative phosphorylation.

Concluding Remarks

From the cryo-EM structures of mitoribosomes and their comparison with the atomic structures of the cytoplasmic ribosomes, some novel aspects of mitochondrial protein synthesis have begun to emerge, suggesting that mitoribosomes have diverged significantly to adopt different host cell types during the course of evolution. However, higher resolution structures are necessary to unravel details of the mechanisms of three very fundamental processes associated with the mammalian mitoribosome: (i) recruitment of leaderless mRNAs, (ii) decoding of atypical codons by unusual tRNA_mt, and (iii) the release of nascent chains through the dramatically remodeled exit tunnels. Structural characterization of the protein components present near key functional sites, such as at the mRNA entrance, and PAS and PES of the nascent chain exit tunnel, will be critical in establishing detailed molecular mechanism of these processes. While the tracing of the rRNA scaffolds can be achieved with high confidence in current 7 Å resolution maps, identification and modeling of mito-specific MRPs, especially those with undefined secondary structure motifs, into the cryo-EM map is still challenging. In the absence of any atomic structures of MRPs or mitochondrial translational factors, molecular interpretation of structures of all mitoribosomes and their functional complexes has to be based currently on docking of homology models into the cryo-EM maps. Furthermore, mitoribosomes seem to be inherently heterogeneous in composition due to (i) dramatic reduction in size of their rRNAs and significant increase in the number of MRPs; consequently, some of the mito-specific MRPs may be less firmly attached to the rest of the mitoribosome in the absence of their direct interaction with the main rRNA scaffolds, and (ii) presence of multiple isoforms of MRPs (e.g., MRPS18) in a given mitoribosome population. Thus, the progress in determining their higher resolution structure is hindered mainly due to challenges in obtaining highly homogeneous preparations. Nevertheless, because of improved algorithms to classify particle images, cryo-EM has been very effective in providing the overall structures of the mitoribosome and associated translation factors at reasonable resolutions.