Portal de Periódicos da CAPES

Widespread use of unconventional targeting signals in mitochondrial ribosome proteins

2021; Springer Nature; Volume: 41; Issue: 1 Linguagem: Inglês

10.15252/embj.2021109519

1460-2075

Yury S. Bykov, Tamara Flohr, Felix Boos, Naama Zung, Johannes M. Herrmann, Maya Schuldiner,

RNA Research and Splicing

Article17 November 2021Open Access Source DataTransparent process Widespread use of unconventional targeting signals in mitochondrial ribosome proteins Yury S Bykov Yury S Bykov orcid.org/0000-0003-2959-4108 Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, Israel These authors contributed equally to this work Search for more papers by this author Tamara Flohr Tamara Flohr orcid.org/0000-0002-8775-7959 Division of Cell Biology, University of Kaiserslautern, Kaiserslautern, Germany These authors contributed equally to this work Search for more papers by this author Felix Boos Felix Boos Division of Cell Biology, University of Kaiserslautern, Kaiserslautern, Germany Search for more papers by this author Naama Zung Naama Zung orcid.org/0000-0002-1517-9774 Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, Israel Search for more papers by this author Johannes M Herrmann Corresponding Author Johannes M Herrmann [email protected] orcid.org/0000-0003-2081-4506 Division of Cell Biology, University of Kaiserslautern, Kaiserslautern, Germany Search for more papers by this author Maya Schuldiner Corresponding Author Maya Schuldiner [email protected] orcid.org/0000-0001-9947-115X Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, Israel Search for more papers by this author Yury S Bykov Yury S Bykov orcid.org/0000-0003-2959-4108 Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, Israel These authors contributed equally to this work Search for more papers by this author Tamara Flohr Tamara Flohr orcid.org/0000-0002-8775-7959 Division of Cell Biology, University of Kaiserslautern, Kaiserslautern, Germany These authors contributed equally to this work Search for more papers by this author Felix Boos Felix Boos Division of Cell Biology, University of Kaiserslautern, Kaiserslautern, Germany Search for more papers by this author Naama Zung Naama Zung orcid.org/0000-0002-1517-9774 Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, Israel Search for more papers by this author Johannes M Herrmann Corresponding Author Johannes M Herrmann [email protected] orcid.org/0000-0003-2081-4506 Division of Cell Biology, University of Kaiserslautern, Kaiserslautern, Germany Search for more papers by this author Maya Schuldiner Corresponding Author Maya Schuldiner [email protected] orcid.org/0000-0001-9947-115X Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, Israel Search for more papers by this author Author Information Yury S Bykov1, Tamara Flohr2, Felix Boos2,3, Naama Zung1, Johannes M Herrmann *,2 and Maya Schuldiner *,1 1Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, Israel 2Division of Cell Biology, University of Kaiserslautern, Kaiserslautern, Germany 3Present address: Department of Genetics, Stanford University, Stanford, CA, USA *Corresponding author. Tel: +49 631 2052406; E-mail: [email protected] *Corresponding author. Tel: +972 08 9346346; E-mail: [email protected] The EMBO Journal (2022)41:e109519https://doi.org/10.15252/embj.2021109519 PDFDownload PDF of article text and main figures. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions Figures & Info Abstract Mitochondrial ribosomes are complex molecular machines indispensable for respiration. Their assembly involves the import of several dozens of mitochondrial ribosomal proteins (MRPs), encoded in the nuclear genome, into the mitochondrial matrix. Proteomic and structural data as well as computational predictions indicate that up to 25% of yeast MRPs do not have a conventional N-terminal mitochondrial targeting signal (MTS). We experimentally characterized a set of 15 yeast MRPs in vivo and found that five use internal MTSs. Further analysis of a conserved model MRP, Mrp17/bS6m, revealed the identity of the internal targeting signal. Similar to conventional MTS-containing proteins, the internal sequence mediates binding to TOM complexes. The entire sequence of Mrp17 contains positive charges mediating translocation. The fact that these sequence properties could not be reliably predicted by standard methods shows that mitochondrial protein targeting is more versatile than expected. We hypothesize that structural constraints imposed by ribosome assembly interfaces may have disfavored N-terminal presequences and driven the evolution of internal targeting signals in MRPs. Synopsis Unlike most mitochondrial matrix proteins synthesized in the cytosol, many subunits of mitochondrial ribosomes lack N-terminal targeting sequences. Dissection of Mrp17 reveals TOM-binding sites and multiple positive charges as directive elements for mitochondrial targeting and translocation. 25% of mitochondrial ribosomal proteins (MRPs) lack predicted N-terminal matrix-targeting presequences. 5 out of 15 studied MRPs utilize internal mitochondrial targeting determinants. Dissection of Mrp17/bS6m reveals an internal sequence carrying its targeting information. The Mrp17 internal signal contains TOM-complex binding sites for targeting and scattered positive charges for translocation. Introduction Mitochondria are descendants of ancient bacteria that formed eukaryotic cells together with their archaeal host (Sagan, 1967; Zaremba-Niedzwiedzka et al, 2017; Martijn et al, 2018). Since then, mitochondria have lost their autonomy and their reproduction depends entirely on the nuclear genome, which encodes the majority of mitochondrial proteins. However, all mitochondria capable of respiration have retained small vestigial genomes of their own and fully functional gene expression machineries of bacterial origin (Roger et al, 2017). Mitochondrial ribosomes (mitoribosomes) are the most complex components of the mitochondrial gene expression system and consist of several RNA molecules and 60 to 80 different proteins (Greber & Ban, 2016; Ott et al, 2016). Mitoribosome dysfunction has adverse consequences leading to a broad spectrum of diseases (Boczonadi & Horvath, 2014). While it took many years to solve the first ribosome structures (Ban et al, 2000; Carter et al, 2000; Schluenzen et al, 2000), the progress in cryo-electron microscopy is now rapidly revealing the structural details of mitoribosomes of many different taxonomic groups (Amunts et al, 2015; Desai et al, 2017; Kummer et al, 2018; Ramrath et al, 2018; Itoh et al, 2020; Tobiasson & Amunts, 2020; Waltz et al, 2020). The availability of so many structures highlighted an interesting feature of mitoribosomes—their incredible evolutionary diversity (Waltz & Giegé, 2019; Kummer & Ban, 2021). The composition of mitochondrial ribosomes in different eukaryotic lineages underwent dramatic changes caused by multiple losses of RNA segments and mitoribosomal proteins (MRPs) as well as acquisition of new, lineage-specific RNA segments and MRPs (Smits et al, 2007; Desmond et al, 2011; Sluis et al, 2015; Petrov et al, 2019). As a result, mitoribosomes contain a core set of MRPs homologous to the bacterial ribosomal proteins (BRPs) and a variable set of MRPs that can be common for all mitochondrial ribosomes or specific only to certain eukaryotic lineages. In addition, during their evolution, many MRPs acquired significant expansions of their C- and N-termini while retaining structurally conserved domains of their BRP ancestors (Vishwanath et al, 2004; Sluis et al, 2015; Melnikov et al, 2018). Mitochondrial genomes in many eukaryotic organisms still contain genes for a number of ribosomal proteins, indicating that their successful transfer to the nuclear genome might be less easily feasible than that of many other matrix proteins (Bertgen et al, 2020). However, most eukaryotic, and in Metazoa even all, MRPs are nuclear encoded. Thus, similar to the majority of mitochondrial proteins (numbering from around 800 in yeast to around 1500 in mammals), they must be imported from the cytosol (Pagliarini et al, 2008; Morgenstern et al, 2017; Vögtle et al, 2017). The import of mitochondrial proteins can be conceptually subdivided in two steps: (i) targeting of the newly synthesized mitochondrial protein precursors to the mitochondrial membrane. This can occur either post-translationally or co-translationally involving ribosome-nascent chain complexes. (ii) Translocation of the unfolded precursors through the mitochondrial membrane(s) to deliver them to their final destination within mitochondria (Bykov et al, 2020). Effective targeting and translocation are mediated by specialized protein complexes that recognize targeting and translocation signals within precursor protein sequences. Transport through the outer membrane is mediated by the TOM (translocon of the outer membrane) complex and through the inner membrane by TIM23 or TIM22 (translocon of the inner membrane) complexes (reviewed in Neupert & Herrmann, 2007). Most matrix and inner membrane proteins are synthesized with N-terminal matrix targeting sequences (MTSs), also called presequences, which are both necessary and sufficient for mitochondrial targeting. MTSs have a characteristic structure that can be predicted computationally (Claros & Vincens, 1996; Emanuelsson et al, 2000; Fukasawa et al, 2015; Armenteros et al, 2019). MTSs are typically between 10 and 60 residues in length and can form an amphipathic α-helix with one positively charged surface and one hydrophobic surface. On the outer membrane, MTSs are recognized by the receptor subunits of the TOM complex, Tom20 and Tom22, and then threaded through the β-barrel pore of Tom40. MTS-containing proteins destined to the matrix are transported through the TIM23 complex that has two pore-forming subunits Tim23 and Tim17 while some inner membrane proteins without MTS can get inserted via the TIM22 complex (reviewed in Neupert & Herrmann, 2007). In most cases, MTSs are proteolytically removed during protein import, giving rise to mature forms of mitochondrial matrix or inner-membrane proteins (von Heijne, 1986; Bedwell et al, 1989; Vögtle et al, 2009). In contrast to all other proteins of the mitochondrial matrix, many MRPs lack N-terminal MTSs (Woellhaf et al, 2014). In some cases, MRPs use N-terminal regions that mimic the properties of MTSs but are not cleaved (un-cleaved MTSs). Such un-cleaved MTSs are also found in some matrix proteins that are not associated with the ribosome, such as Hsp10 (Poveda-Huertes et al, 2020). Surprisingly, a number of MRPs do not contain any regions that show MTS-like features and it is unknown how mitochondria recognize and import these proteins. For now, there are only two well characterized examples of MRPs with unconventional MTSs—Mrpl32 (bL32m, by new nomenclature (Ban et al, 2014)) and Mrp10 (mS37) whose import path deviates from the canonical matrix-targeting route (Nolden et al, 2005; Bonn et al, 2011; Longen et al, 2014). In this work, we studied the mechanisms by which MRPs are targeted and translocated into mitochondria. We systematically examined N-termini of unconventional MRPs and analyzed them in silico and experimentally. We further focused on the biogenesis of Mrp17 (bS6m) as a representative of the unconventional group of MTS-less MRPs. We discovered a novel mitochondrial matrix targeting region that is displayed in the internal sequence of the protein. This stretch shares properties with mitochondrial targeting sequences such as positive charges for receptor binding and membrane potential-dependent translocation, but differs in its structural features and position in the protein. The efficient import of Mrp17 shows that the mitochondrial import machinery is much more versatile in its substrate spectrum than expected. More generally, our work shows how structural restrictions favored the generation of unconventional targeting motifs. Results Mapping unconventional MRP targeting signals To systematically investigate MRP targeting signals in detail, we compiled all existing data on the maturation of their N-termini in yeast (Dataset EV1). We used direct N-terminal sequencing data (Graack et al, 1988, 1991; Grohmann et al, 1989, 1991; Matsushita et al, 1989; Dang & Ellis, 1990; Kitakawa et al, 1990, 1997; Boguta et al, 1992; Davis et al, 1992; Matsushita & Isono, 1993), N-terminal proteomics (Vögtle et al, 2009) and predictions performed by UniProt annotators, as well as by ourselves using MitoFates for cleavage site prediction (Fukasawa et al, 2015). Importantly, we also used available structural information (Desai et al, 2017). In particular, mitoribosome structures were helpful to identify proteins that do not have a cleavable MTS—such proteins had their N-termini contained within the structure and hence could not have been cleaved after import into the mitochondrial matrix. We reanalyzed ribosome profiling data on translation initiation in yeast to ascertain that none of these proteins has mis-annotated translation start sites that might produce an N-terminal extension accounting for a missing cleavable MTS (Appendix Fig S1). In the yeast mitochondrial ribosome structure (PDB:5MRC), the detectable sequence of six proteins started with amino acid number 1 (Met), that of 12 started with amino acid number 2, five—with amino acids 3–9, and the rest, 50, with amino acid number 10 and more. The number of the first amino acid present in the structure was moderately conserved among the determined mitoribosome structures (Appendix Fig S2) and was not restricted to any particular group of MRPs classified by origin (bacterial, mitochondria-specific, or yeast-specific) or position in the structure (Fig EV1, Appendix Fig S2). Click here to expand this figure. Figure EV1. MRPs can be classified according to the presence of their most N-terminus inside the mitoribosome structure Left—MTS prediction scores for yeast MRPs first sorted in groups by the first amino acid with reported atomic coordinates in the structure PDB:5MRC, then by subunit and then by length with protein origin and subunit noted for each MRP. Right—primary and secondary sequence properties for 15 MRPs selected for further characterization showing a variety of N-terminal and internal targeting signal predictions, overall positive charge, presence of documented cleavable MTS, and a variety of N-terminal secondary structures. Universal ribosomal protein nomenclature is used (Ban et al, 2014), except for Mhr1 which is a yeast-specific MRP. Download figure Download PowerPoint Interestingly, a simple distinction by the first amino acid appearing in the structure separates MRPs into two classes. In the first group are those MRPs that are derived from cleaved precursors (which consistently have high MTS prediction scores). In addition, this group may contain proteins with an uncleavable N-terminus of a flexible nature which would then be unresolved in the available structures. Some of the latter may have poor mitochondrial targeting scores in prediction algorithms. In the second group are those whose structure starts with amino acid number less than 10. Most of these proteins score very poorly with different software predicting N-terminal MTS (Figs 1A and EV1). Many MRPs of this group lack conventional, N-terminal import signals, and their targeting signals are not predicted by available software. Thus, the available structures of mitochondrial ribosomes confirm the previous conclusion that many MRPs are made without N-terminal MTSs (Woellhaf et al, 2014). Figure 1. Mitochondrial ribosomal proteins (MRPs) have various types of targeting signals A. Yeast MRPs having uncleaved N-termini that can be tracked in the mitoribosome structure (PDB:5MRC) score much lower with MTS prediction algorithms (average of TargetP2 and MitoFates) compared with other MRPs that have their N-termini cleaved off or are not present in the structure (so might be flexible and outside the mitoribosome body). Mean ± SD is indicated by the bars, no replicates were performed as the values are predictions. See Fig EV1 for more detailed data. B. Schematic of MRP truncations used to characterize targeting properties of MRP N-termini: MRPFull as control, MRP∆30 to check if the N-terminus is necessary, MRP1–30 to check if it is sufficient (top) and the schematics of expected GFP localization in case the N-terminus is MTS-like (necessary and sufficient) or not (bottom). C–E. Micrographs collected in the GFP channel for each truncation (columns) of each studied MRP (rows) grouped by the N-terminus targeting properties based on theoretical expectation summarized in (B) with the MRPs possessing MTS-like N-termini in panel (C), MRPs with intermediate phenotype in panel (D) and MRPs without N-terminal signal in panel (E), for each MRP a yeast gene name and new nomenclature protein name is shown on the left. Scale bar for all micrographs is 5 µm. Download figure Download PowerPoint Next, we experimentally analyzed the targeting information in the sequences of different MRPs by GFP fusion proteins. To this end, we selected 15 MRPs with different properties (Fig EV1, Dataset EV2). Then we tested whether the N-terminal 30 residues of these proteins were necessary and/or sufficient for mitochondrial targeting. The length of 30 residues was chosen as it corresponds to the most common size of a cleavable yeast MTS (Vögtle et al, 2009). To test this, we expressed each MRP in diploid yeast fused to GFP. To assay if the N-terminus is necessary, we expressed a truncated version with the first 30 amino acids deleted (MRPtype="InMathematical_Operators">∆30-GFP). To test if the N-terminus is sufficient, we expressed a version with only the first 30 amino acids (MRP1–30-GFP). As a control, we used the full-length version (MRPFull-GFP; Fig 1B). The distribution of GFP signals was imaged in cells in which mitochondria were stained with MitoTracker Orange (Fig 1C–E, Appendix Fig S3, Dataset EV2). Six proteins (Mrpl15, Rsm26, Rsm18, Rsm25, Mrpl40, and Rsm27) contained targeting information within their N-termini (Fig 1C); of them, only Mrpl15 (mL57) had high MTS prediction scores consistent with highly confident annotation of a cleavable 29-amino acid long MTS (Dataset EV1). Other proteins whose N-termini were able to target GFP to mitochondria had low MTS prediction scores (Dataset EV2) indicating that their N-terminal signals have distinct properties, not similar to conventional MTSs. For four proteins (Mrps16, Mrp51, Mrpl38, and Mrpl28) neither the N-terminal 30 residues nor the internal segment on its own were sufficient for targeting, indicating that the necessary targeting information is contained in an N-terminal segment longer than 30 amino acids or distributed over the whole length of these proteins (Fig 1D). Finally, five proteins (Mrp17, Pet123, Mrpl23, Mrp35, and Mrp20) were targeted to mitochondria independently of their N-terminal regions indicating that the targeting signals in these proteins are internal (Fig 1E). Interestingly, many of the N-terminally truncated MRP versions accumulated outside mitochondria in the cytosol or, in many cases, in the nucleus and nucleolus (Figs 1C–E and EV2). These observations agree with the recent discovery that mistargeted mitochondrial proteins can accumulate in the nucleus and get degraded in perinuclear puncta (Shakya et al, 2021). Despite the mislocalization of several of these forms, none of them resulted in obvious growth defects (Appendix Fig S4). Click here to expand this figure. Figure EV2. MRPs have N-termini with various targeting properties MRP-GFP truncations mistargeted to the nucleus were transformed with NLS (nuclear localization signal)-tdTomato plasmid, stained with MitoView 405 dye and visualized by fluorescent microscopy. Scale bar is 10 µm. MRP-GFP truncations mistargeted to the nucleus and enriched in the nucleolus with nucleolar protein Nop2 genomically tagged using mCherry, stained with MitoView 405 dye and visualized by fluorescent microscopy. Observed MRP-GFP aggregates are highlighted with white arrowheads. Scale bar is 10 µm. Summary of mistargeting locations for one or more truncations of each MRP, if the location is observed for any of the MRP truncations, it is marked with a black circle. Download figure Download PowerPoint To summarize, we selected a subset of MRPs with diverse structural and sequence features and characterized the mitochondrial targeting capacity of their N termini. We observed that many of these MRPs contain unconventional targeting signals, often outside of the 30 N-terminal residues, and apparently scattered over their sequence. One particularly intriguing MRP was Mrp17 (bS6m), a protein of the small subunit of the yeast mitoribosome. Mrp17 lacks any identifiable targeting signal and is present in all mitoribosome structures studied to date with its N-terminus visualized in all of these structures (Appendix Fig S2). Hence, we chose Mrp17 for further investigation. Defining Mrp17 targeting and translocation signals To investigate the unconventional targeting signals of Mrp17 in more detail, we created a systematic set of Mrp17 truncations fused to GFP and expressed them in diploid yeast (Fig EV3, Appendix Fig S5). We observed that the internal fragment of Mrp17 between amino acids 20 and 100 was the minimal fragment able to target GFP to mitochondria similarly to full-length Mrp17 (131 amino acids) without producing cytosolic background signal (Fig 2A). This indicates that similarly to the N-terminus, the C-terminus is dispensable for targeting. Splitting this fragment in two halves showed that the N-terminal part (Mrp1721–60) was still able to target GFP to mitochondria although with significant cytosolic background while the C-terminal part (Mrp1761–100) was cytosolic (Fig 2A). We conclude that, in vivo, Mrp17 region 21–60 is necessary for mitochondrial targeting but is not sufficient for efficient targeting, which is promoted by additional signals distributed over the whole length of the protein (Fig EV3A and B). Click here to expand this figure. Figure EV3. The noncanonical targeting and translocation signal of Mrp17 is located between amino acids 30 and 60 A, B. In vivo characterization of mitochondrial targeting capacity of different Mrp17 truncations fused to GFP: (A) GFP localization examples for truncations, and demonstrating effective targeting with only mitochondrial GFP signal (Mrp1721–131-GFP, same data as in Appendix Fig S5); ineffective targeting with mitochondrial GFP signal accompanied by strong cytosolic signal (Mrp1721–60-GFP, same data as in Fig 2A and Appendix Fig S5); ineffective targeting with additional nuclear signal (Mrp1781–131-GFP, same data as in Appendix Fig S5); and no detectable mitochondrial targeting with exclusively cytosolic GFP (Mrp17100–131-GFP, same data as in Appendix Fig S5). Scale bar is 10 µm. (B) localization summary of different Mrp17 truncations fused to GFP and colored according to the color-code for effective, ineffective, and no targeting introduced in panel (A), truncations additionally targeted to the nucleus are marked with asterisks. C. Mrp17-DHFRmut is translocated into isolated mitochondria at the same rate as WT Mrp17 but gives better signal in the autoradiograph. D. In vitro import assays for additional truncations of Mrp17 fused to DHFRmut not shown in Fig 2B, import was performed as described in the legend for Fig 2. Download figure Download PowerPoint Figure 2. The noncanonical targeting and translocation signal of Mrp17 is located between amino acids 30 and 60 In vivo characterization of mitochondrial targeting capacity of different Mrp17 truncations fused to GFP visualized by fluorescent microscopy with MitoTracker Orange straining. Scale bar for all micrographs is 10 µm. Characterization of Mrp17 translocation signal using an in vitro import assay: shown are autoradiographs of full-length Mrp17 or its truncations fused to DHFRmut, translated in vitro with radiolabeled amino-acids, incubated with isolated yeast mitochondria for 2, 5, or 10 min, treated with proteinase K (PK) to remove nonimported proteins and visualized by 16% SDS–PAGE/autoradiography. As a negative control, mitochondria were treated with valinomycin, antimycin, and oligomycin (VAO) that eliminate membrane potential. For comparison, 20% of the protein used per import reaction was loaded on the first lane. Download figure Download PowerPoint The microscopic analysis does not allow us to discriminate between targeting to the mitochondrial surface from complete translocation into the matrix and is affected by truncated MRP stability in vivo. To elucidate the translocation efficiency of different Mrp17 regions, we used in vitro import assays into isolated yeast mitochondria. Since Mrp17 is very small and many fragments lacked methionine residues that are necessary for radiolabeling, we fused Mrp17 to an unfolded mutant of the mouse dihydrofolate reductase —DHFRmut (Vestweber & Schatz, 1988). The full-length Mrp17-DHFRmut fusion was effectively imported into isolated yeast mitochondria at the same rate as untagged Mrp17 but gave much stronger signals in autoradiography (Figs 2B and EV3C). In agreement with the targeting experiments performed in vivo, the short N-terminal region of Mrp17 was neither necessary nor sufficient for efficient translocation (Mrp1721–131, Mrp171–20 in Fig 2B). The first 60 amino acids of Mrp17 were sufficient for translocation narrowing down the import signal to the N-terminal half of the protein (Mrp171–60 in Fig 2B). Leaving only the first 40 amino acids or removing them from the N-terminus reduced the translocation speed indicating that regions 20–40 and 40–60 are equally important parts of the signal (Mrp171–40 and Mrp1741–131 in Fig 2B). Finally, we narrowed down the Mrp17 region containing the translocation signal to amino acids 30–60 (Fig 2B, bottom; Fig EV3D). However, similar to the results of in vivo experiments, even short fragments of Mrp17 outside this region retained some residual translocation capacity (Fig 2B). To summarize, we determined that the main mitochondrial targeting and translocation signal of Mrp17 is positioned between amino acids 30 and 60. However, there exist additional signals that improve mitochondrial targeting efficiency or stability in vivo. These additional signals reside in the C-terminal half of Mrp17. This again indicates, that the mitochondrial targeting regions are scattered over the Mrp17 sequence, and for this protein, the most N-terminal region is irrelevant for efficient mitochondrial import. Characterizing features of the Mrp17 targeting and translocation signal Next, we analyzed the unconventional internal targeting region of Mrp17 located between residues 30 and 60 in more detail. Standard prediction algorithms do not find an MTS-like sequence in this region (Fig EV1). The Mrp17 structure mostly contains β-strands in this region and only a part of a helical stretch (Fig 3A). Mrp17 is generally rich in positive charges (its pI is 10.5). While a high content of positive charges is a general feature of ribosomal proteins that interact with negatively charged mRNA, during evolution, the content of positive charges in MRPs (and particularly their lysine content) was further increased. This suggests that positive charges might play a role beyond their relevance for neutralizing the negative charges of ribosomal RNA (Fig EV4A and B). Figure 3. Positive charge and Tom20-binding motifs are important features of the Mrp17 targeting signal Primary sequence of Mrp17 highlighting charged amino-acids, Tom20-binding motifs (TBM1, TBM2) and secondary structure (from PDB:5MRC). In vitro mitochondria translocation capacity of WT Mrp17, Mrp17K-R with all lysines (K) substituted with arginines (R), and Mrp17K-A with all lysines (K) substituted with alanines (A) fused to DHFRmut. Import was performed as described in Fig 2 legend. In vivo mitochondrial targeting capability of WT Mrp17, Mrp17K-R, and Mrp17K-A fused to GFP. Scale bar is 10 µm. Substitution of lysines with alanines in the regions 1–60 or 30–60 of Mrp17 abolishes mitochondrial import capacity: (top) schematics of substitution positions, TBMs are in green, targeting signal outlined in grey, substitutions are denoted by arrows; (bottom) in vitro translocation of the mutants. Mrp17 variants rescue ∆mrp17 strain growth defect on respiratory media: the indicated variants or empty vector (EV) were introduced in WT yeast and then the genomic MRP17 was disrupted by knock out, the resulting mutants were serially diluted 10× and spotted on media containing glucose or glycerol as a sole carbon source. Download figure Download PowerPoint Click here to expand this figure. Figure EV4. Mrp17 sequence features important for targeting and translocation to mitochondria Ribosomal proteins are positively charged and mitochondrial proteins acquired even more positive net charge compared with their bacterial homologs. Total amino acid gain of MRPs (calculated as the difference between total count of each amino acid in all yeast MRPs, including mitochondria-specific, and all bacterial RPs) compared with bacterial RPs shows over-representation of lysines (K). Lysines in Mrp17 are not important for mitochondrial targeting and can be substituted with arginines, same micrographs for constructs MRP17WT-GFP, MRP17K-R-GFP, and MRP17K-A-GFP as in Fig 3C shown in all channels beside micrographs of yeast not expressing any GFP (bottom row) as control for autofluorescence relative to cytosolic signal. All micrographs in the GFP channel are shown at the same contrast and brightness for comparison; Scale bar is 10