Gravitational and mechanical forces shape mitochondrial translation

Abstract

Life on Earth has evolved in a form suitable for the gravitational force. Although the pivotal role of gravity in gene expression has been suggested, the molecular details remain unclear. Here, we show that mitochondria utilize gravity to activate protein synthesis within the organelle. Genome-wide ribosome profiling reveals reduced mitochondrial translation in mammalian cells and Caenorhabditis elegans under microgravity. We found that attenuation of cell adhesion through laminin–integrin interactions caused the phenotype. Mitochondrial translation is activated by a signal relayed by FAK, RAC1, PAK1, BAD, and Bcl-2 family proteins in the cytosol, and the mitochondrial fatty acid synthesis (mtFAS) pathway in the matrix. Consumption of mitochondrial malonyl-CoA by mtFAS reduces the malonylation of the translational machinery and accelerates the rates of translational initiation and elongation. Physiologically, this system operates in mechano-response of skeletal muscles. Our work provides mechanistic insights into how cells convert gravitational and mechanical forces into translation in mitochondria.

In this study, microgravity was found to disrupt mitochondrial translation through inhibition of laminin–integrin signaling and the downstream pathway, which is usually activated by mechanical stress.

Introduction

Since the 1950s, when the first dog was sent to space, biological and medical research has revealed unique responses to life in the space environment¹. In addition to radiation, microgravity is one of the main hazards to cells in space and potentially causes oxidative stress^2,3, mitochondrial dysfunction^3–5, and epigenetic changes^5,6. However, due to the limitations of spaceflight experiments, the global view of the cell response to microgravity is still incomplete.

To fill this gap, multi-omics approaches have been applied to a cohort of flight samples, including those from astronauts^7,8. These studies explored the epigenome, transcriptome, proteome, and metabolome^3,9–18 and revealed microgravity-mediated gene expression profile changes associated with the hazardous phenotypes mentioned above. However, these conventional spaceflight studies have lacked investigations of translational regulation, an acute and inducible means of reprogramming protein expression^19–22. Given that translational control is typically followed by transcriptome changes, dynamic shifts in the protein synthesis rate under microgravity could be envisioned, but the “translatome” has not been assessed in previous spaceflight studies.

Here, we comprehensively surveyed the translational response of human tissue cultures and Caenorhabditis elegans to microgravity and found that mitochondrial translation is dramatically reduced by microgravity. By elucidating the mechanism of this response, we found that cell adhesion mediated by laminin–integrin and downstream signaling transmission through FAK, RAC1, PAK1, BAD, and Bcl-2 family proteins relays gravitational forces for mitochondrial translation. This leads to the activation of the mitochondrial fatty acid synthesis (mtFAS) pathway in the matrix and reduces malonyl-coenzyme A (malonyl-CoA), maintaining a low level of malonylated lysines in the translation machinery. We found that this regulation increases the kinetics of translation initiation and elongation in mitochondria. Considering that laminin–integrin functions physiologically as a mechanosensor, the microgravity response may be the manifestation of an impairment of mechanical stress. Consistent with this scenario, we recapitulated the attenuation of mitochondrial translation in unloaded and immobilized mouse hindlimbs. This study revealed the previously uncharacterized unique connection between the extracellular environment and mitochondrial translation.

Results

Mitochondrial translation repression under microgravity in space

To investigate the translational landscape under microgravity, we conducted genome-wide ribosome profiling^23,24 on HEK293 cells cultured at the International Space Station (ISS) (Fig. 1A)²⁵. For this purpose, cells were first cultured at 1 × g, artificially generated by centrifugation in the ISS, and then under microgravity conditions for 24 and 48 h (Fig. 1A). In contrast, control cells were maintained at 1 × g until harvesting. The cells were treated with translation inhibitors (cycloheximide and chloramphenicol), frozen, and returned to the laboratory on Earth for library preparation.

Fig. 1. Genome-wide ribosome profiling reveals mitochondrial translation repression by microgravity in humans and nematodes.

Schematic representation of the experiments in the International Space Station (ISS) for HEK293 cells (A) and C. elegans (D). µg: microgravity. B MA (M, log ratio; A, mean average) plots for ribosome footprint change (left), RNA abundance change (middle), and translation efficiency change (right) after 24-h microgravity (μg) culture in HEK293 cells. Significantly altered transcripts (false discovery rate [FDR] < 0.05) and mitochondrial genome-encoded mRNAs are highlighted. C Box plots for ribosome footprint change (left), RNA abundance change (middle), and translation efficiency change (right) in mitochondrial genome-encoded mRNAs (n = 13) over the course of microgravity culture. D Cumulative distributions of ribosome footprint change (left), RNA abundance change (middle), and translation efficiency change (right) in C. elegans with respect to nuclear genome-encoded mRNAs and mitochondrial genome-encoded mRNAs (n = 12) after 4-d microgravity culture. E The shaded area represents the range from 2.5 percentile to 97.5 percentile of a 1000-times random sampling of 12 genes encoded in the nuclear genome from the dataset. For (C, E), the pvalue was calculated by the Mann‒Whitney U test (two-tailed). In box plots (C), the median (centerline), upper/lower quartiles (box limits), and 1.5× interquartile range (whiskers) are shown. See also Supplementary Fig. 1 and Supplementary Data 2.

Due to spaceflight restrictions, the amount of material available was limited for standard ribosome profiling experiments. To overcome this issue, we harnessed Thor-Ribo-Seq, a ribosome profiling derivative tailored for low input based on RNA-dependent RNA amplification by in vitro transcription²⁶. The application allowed us to obtain high-quality data from samples returned from the ISS, which presented hallmarks of ribosome footprints: a sharp peak in the read length at 28-29 nt (Supplementary Fig. 1A), 3-nt periodicity (Supplementary Fig. 1B), and high reproducibility (Supplementary Fig. 1C left).

Our analysis revealed the impacts of microgravity on translation, especially a reduction in a subset of mRNAs, at 24 h of microgravity culture (Fig. 1B and Supplementary Fig. 1E), rather than at 48 h (Supplementary Fig. 1D). Among those changed at 24 h, we found that protein synthesis from mitochondrial mRNAs, which are encoded in the organelle genome, was remarkably sensitive to microgravity (Fig. 1B left). Since our method captured footprints from both cytosolic ribosomes (cytoribosomes) and mitochondrial ribosomes (mitoribosomes)^27,28, we could interrogate both translation systems simultaneously. The reduction in mitoribosome footprints could not be explained by a reduction in the corresponding mRNAs, as measured by RNA sequencing (RNA-Seq) of the same materials (Fig. 1B middle and Supplementary Fig. 1C right and 1E). Due to this difference between Ribo-Seq and RNA-Seq, the translation efficiency, calculated by over- or underrepresentation of ribosome footprints over RNA-Seq fragments, still showed a decrease in mitochondrial mRNAs (Fig. 1B right). In contrast, the translation of mitochondrial proteins encoded in the nuclear genome was insensitive to microgravity (Supplementary Fig. 1F, G).

Mitochondrial translation regulation may be an early response to microgravity. We observed that the change in mitochondrial translation was more dramatic after 24 h of incubation compared to 48 h under microgravity conditions (Fig. 1C right and Supplementary Fig. 1D right), suggesting adaptation to the microgravity environment. This phenomenon may be due to the constant reduction in the abundance of these mRNAs (Fig. 1C middle and Supplementary Fig. 1D middle), which allows long-term suppression of mitochondrial protein production.

To further extend our exploration from cell culture to the whole-body level, we used nematode samples flown on the ISS^16,29,30. The L1 larvae were cultured under microgravity conditions for 4 d, while the control worms were incubated in a 1 × g centrifuge (Fig. 1D). Similar to human cells, ribosome profiling and RNA-Seq in nematodes (Supplementary Fig. 1I–K) revealed translational alterations in a subset of mRNAs (Supplementary Fig. 1L, M). Although individual mitochondrial mRNAs did not reach statistical threshold (Supplementary Fig. 1L), a significant reduction in translation efficiency of mitochondrial mRNAs as a group was observed (Fig. 1E and Supplementary Fig. 1N). Again, the translation of mitochondria-targeted proteins encoded in the nuclear genome was not reduced (Supplementary Fig. 1O).

Through these spaceflight experiments, we concluded that attenuation of mitochondrial translation is a pervasive response to microgravity in higher eukaryotes.

The mitochondrial translation response is recapitulated by simulated microgravity on the ground

To survey microgravity-mediated mitochondrial translation repression in the laboratory on the ground, we harnessed a clinostat that rotates the cell culture flask in three dimensions to reduce gravitational forces on the cells and simulate microgravity conditions (Fig. 2A). This cultural condition did not inhibit cell growth (Supplementary Fig. 2A). Ribosome profiling and RNA-Seq of HEK293 cells revealed that simulated microgravity phenocopied the attenuated mitochondrial translation observed in the ISS (Fig. 2B, C and Supplementary Fig. 2B)²⁵; we detected the reduction in mitoribosome footprints even after 1 h of simulated microgravity treatment (Fig. 2C). Again, we did not find a significant impact on nuclear genome-encoded mitochondrial proteins (Supplementary Fig. 2C, D). Gravity-mediated mitochondrial translation regulation was further ensured by mitochondria-specific fluorescent noncanonical amino acid tagging on gels (on-gel mito-FUNCAT)^31–35 by 24-h incubation of cells under a clinostat (Fig. 2D and Supplementary Fig. 2E–G). On-gel separation of labeled proteins assigned the 13 proteins synthesized within mitochondria as in the earlier studies^31–34 and detected their global reduction by simulated microgravity (Supplementary Fig. 2H). Since the same molecular phenotype was observed in human cells (HEK293) and mouse cells (C2C12 and 3T3) (Fig. 2D and Supplementary Fig. 2E–H), the mitochondrial translation response to microgravity is likely driven by conserved mechanisms. The longer incubation under microgravity for 48 h and 72 h led to the further reduction of mitochondrial translation (Fig. 2E and Supplementary Fig. 2I), suggesting steady-state adaptation occurred over time.

Fig. 2. Mitochondrial translation is a function of gravity.

A Schematic representation of experiments with a 3D clinostat to induce simulated microgravity (s-μg) in HEK293 cells. B MA plots for ribosome footprint change (left), RNA abundance change (middle), and translation efficiency change (right) after 24-h simulated microgravity culture in HEK293 cells. Significantly altered transcripts (false discovery rate [FDR] <0.05) and mitochondrial genome-encoded mRNAs are highlighted. C Box plots for ribosome footprint change (left), RNA abundance change (middle), and translation efficiency change (right) in mitochondrial genome-encoded mRNAs (n = 13) over the course of simulated microgravity culture. D–G Quantification of on-gel FUNCAT experiments to monitor total and mitochondrial translation in the indicated cell lines and conditions. For D, the cells were cultured under s-μg for 24 h before cell harvesting. For the recovery experiments in F, the cells were incubated under s-μg for 24 h and then at 1 × g for 24 h. For G, the cells were cultured at 10 × g centrifugation for 24 h before cell harvesting. For C, the p-value was calculated by the Mann‒Whitney U test (two-tailed). For D–G, the data from three replicates (points), the mean values (bars), and the s.d.s (errors) are shown. The p-values were calculated by Student’s t test (two-tailed) (D, G) and by the Tukey‒Kramer test (two-tailed) (E, F). In box plots (C), the median (centerline), upper/lower quartiles (box limits), and 1.5× interquartile range (whiskers) are shown. See also Supplementary Fig. 2 and Supplementary Data 2.

Our on-gel mito-FUNCAT experiments showed that mitochondrial translation is a function of gravity. When the cell culture was returned to standard 1 × g (i.e., 24-h culture at simulated microgravity and then another 24-h culture at 1 × g for the recovery), the reduction in mitochondrial translation induced by microgravity was restored to the baseline level, suggesting the elasticity of the mechanism (Fig. 2F and Supplementary Fig. 2J). Moreover, centrifugation-induced 10 × g hypergravity resulted in the activation of mitochondrial translation (Fig. 2G and Supplementary Fig. 2K). Ultimately, we observed the correspondence between the mitochondrial translation activity and gravitational force (Supplementary Fig. 2L). The centrifugation itself did not explain our data, since centrifugation-generated 1 × g showed the limited impact on mitochondrial translation (Supplementary Fig. 2M, N).

Given that microgravity induces mitochondrial stress^2–4, we investigated the association between these stresses and in organello protein synthesis under microgravity. We did not observe a significant alteration in the copy number of mitochondrial DNA, which is susceptible to oxidative damage, under simulated microgravity (Supplementary Fig. 2O, P). Additionally, our microscopy analysis did not reveal mitochondrial fragmentation, which can be induced by mitochondrial oxidative stress^36,37 (Supplementary Fig. 2Q, R). In both space and clinostat samples, we could not find a signature of the mitochondrial unfolded protein response (mtUPR)³⁸ since the transcriptomic upregulation of the targets³⁹ was not observed (Supplementary Figs. 1H and 2S). Thus, our data indicated that the repression of mitochondrial translation is not caused by oxidative stress or morphological defects in the organelle.

We note that the “time-averaged” microgravity in the 3D clinostat recapitulates not all the responses to the loss of gravitational forces found in the ISS. A lesser magnitude of the mitochondrial translation efficiency changes was caused by simulated microgravity than by bona fide microgravity in the ISS (compare the y-axis scales in Figs. 1C and 2C). Mitochondrial mRNA reduction observed in the ISS microgravity was not seen in clinostat microgravity (Figs. 1C and 2C). Consequently, the adaptation of mitochondrial translation efficiency found at 48 h in the ISS microgravity (Fig. 1C) could not be observed in the simulated microgravity, even with extended incubation (at least 72 h) (Fig. 2E and Supplementary Fig. 2I). Translational responses in the cytosolic mRNAs showed limited overlaps in microgravity and simulated microgravity (Supplementary Fig. 2T). Nonetheless, the reduced mitochondrial translation was the common signature in both conditions (Supplementary Fig. 2T), suggesting the robustness of this phenotype.

The cell adhesion pathway relays gravitational force to mitochondrial translation

How do cells recognize gravity and transmit the information to organelle translation? The attenuation of cell adhesion and subsequent signal transduction through the focal adhesion kinase (FAK) pathway is a hallmark of microgravity^40–45. Actin stress fibers are thought to be gravity sensors to transmit the pN-order of force to FAK^46,47. Consistent with these reports, we observed that the phosphorylation of FAK, which is driven by cell adhesion, was reduced by simulated microgravity (Supplementary Fig. 3A). Thus, we reasoned that modulated cell adhesion may reduce mitochondrial translation in microgravity.

We first investigated the relationship between cell adhesion and organelle translation under regular gravity. Considering the translation flux of the α and β subunits of integrin, a receptor of the extracellular matrix (ECM), the laminin receptor α₆β₁ was an abundant integrin class in HEK293 cells (Supplementary Fig. 3B). Accordingly, pretreatment of cell culture dishes with laminin-511 (or α₅β₁γ₁), which has a high affinity for integrin α₆β₁^48,49, enhanced cell adhesion in a dose-dependent manner (Fig. 3A, B top and Supplementary Fig. 3C) and thus elevated FAK phosphorylation (Supplementary Fig. 3D). Through on-gel mito-FUNCAT, we found that mitochondrial protein synthesis was concomitantly activated by increasing the laminin concentration (Fig. 3A, B bottom left and Supplementary Fig. 3E) and correlated with cell adhesion activity (Fig. 3B bottom right). Similar mitochondrial translation activation by laminin was also observed in other cell types (human HAP1 and mouse C2C12, 3T3, and MEF) (Figs. 3C and 4I and Supplementary Figs. 3F–H, 4O, 5L, N). Again, the laminin-mediated enhancement of mitochondrial translation was mRNA non-selective rather global (Supplementary Fig. 3I) and also detected in the conventional metabolic labeling by ³⁵S Met (Supplementary Fig. 3J, K). Conversely, mitochondrial translation was inhibited by the RGDS peptide, an integrin inhibitor (Fig. 3D and Supplementary Fig. 3L), which led to FAK dephosphorylation (Supplementary Fig. 3M). Additionally, mitochondrial translation was influenced by the elasticity of the culture surface (Supplementary Fig. 3N, O), highlighting its dependency on mechanical stress. In contrast to in organello translation, changes in total protein synthesis, which predominantly originates from cytosolic translation, were limited by the treatment of laminin/RGDS peptide or the alteration in cell surface elasticity (Fig. 3B–D and Supplementary Fig. 3E–H, L, N, O). These results showed that cell adhesion specifically enhances mitochondrial translation.

Fig. 3. The attenuation of the laminin–integrin signaling pathway represses mitochondrial translation under microgravity.

A Schematic representation of the experiments to investigate the impacts of laminin–integrin-mediated cell adhesion on mitochondrial translation. B Relationships among laminin concentration, cell adhesion activity, and translation activity. For cell adhesion activity, the mean value without laminin precoating was set to “1”. Total and mitochondrial protein synthesis were separately assayed. ρ, Spearman’s rank correlation coefficient. C, D Quantification of on-gel FUNCAT experiments to monitor total and mitochondrial translation in the indicated cell lines and conditions. For D, the cells were treated with 10 μg/ml RGDS peptide. E Box plots for ribosome footprint change (left), RNA abundance change (middle), and translation efficiency change (right) in mitochondrial genome-encoded mRNAs (n = 13) over the course of simulated microgravity culture with or without laminin precoating. Note that the data without laminin precoating are the same as those in Fig. 2C. F Oxygen consumption rate (OCR) of HEK293 cells at basal respiration and after treatment with oligomycin, carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP), or rotenone/antimycin. G The relative mitochondrial ATP production rate was measured over the course of simulated microgravity culture. For C, D, the data from three replicates (points), the mean values (bars), and the s.d.s (errors) are shown. The p-values were calculated by Student’s t test (two-tailed). For E, the p-value was calculated by the Mann‒Whitney U test (two-tailed). For F, the mean values (points) from three replicates and the s.d.s (errors) are shown. For G, the data from three replicates (points) and the mean values (bars) are shown. The p-values were calculated by the Tukey-Kramer test (two-tailed). In box plots (E), the median (centerline), upper/lower quartiles (box limits), and 1.5× interquartile range (whiskers) are shown. The experiments were conducted in HEK293 cells unless otherwise noted. See also Supplementary Fig. 3 and Supplementary Data 2.

Fig. 4. The cellular pathway that transduces the signal from laminin–integrin cell adhesion to mitochondrial translation activation.

A, B, D, F Quantification of on-gel FUNCAT experiments to monitor mitochondrial translation in the indicated conditions. For A, cells were analyzed after siRNA transfection to knock down FAK. For B, D, F, cells were treated with the indicated compounds at the following concentrations: NSC23766, 200 μM; ZCL278, 100 μM; Rhosin, 50 μM; BAY-293, 1 μM; LY294002, 10 μM; SP600125, 20 μM; IPA-3, 5 μM; CK666, 25 μM; obatoclax, 2 μM; and TW-37, 10 μM. C Western blotting of RAC1 protein in the total cell lysate and in the GTP-bound fraction pulled down with the GST-fused PBD of PAK1. GST-tagged PAK1 PDB was detected as a loading control. The quantified relative amount of RAC1 in the GTP-bound, activated form is shown. E Western blotting of total PAK1 protein and the phosphorylated form (at S144 and S141). The quantified relative amount of phosphorylated PAK1 is shown. G Western blotting of BAD protein in the total lysate and in the mitochondrial immunoprecipitation (MitoIP) fraction with an anti-TOMM22 antibody. H Representative distribution of Cy3-conjugated HPG signals normalized to AF647-labeled TOMM20 signals in the indicated cell lines (left). The dashed line represents the mean of the distribution. The quantification is shown on the right. The distribution of unnormalized Cy3-conjugated HPG signals and AF647-labeled TOMM20 signals is shown in Supplementary Fig. 4N. I On-gel FUNCAT experiments to monitor mitochondrial translation in naïve and BAD KO HAP1 cells with a titrated concentration of laminin for precoating. For A–H, the data from replicates (points, n = 3 for A, C, E, G, H; n = 6 for B, D, F), the mean values (bars), and the s.d.s (errors) are shown. The p-values were calculated by Student’s t test (two-tailed) (C, E, G, H) and by the Tukey‒Kramer test (two-tailed) (A, B, D, F). For I, the data from three replicates and the mean values (bars) are shown. The experiments were conducted in HEK293 cells unless otherwise noted. See also Supplementary Fig. 4.

To test whether microgravity-mediated repression of mitochondrial translation is caused by reduced cell adhesion, we pretreated a cell culture flask with laminin, cultivated the cells under simulated microgravity, and then conducted ribosome profiling and RNA-Seq²⁵. Strikingly, laminin counteracted the effect of microgravity on mitochondrial protein synthesis (Fig. 3E). We concluded that cell adhesion is a mediator of gravitational forces for mitochondrial translation.

The loss of in organello translation homeostasis results in physiological dysfunction of the organelle. The attenuation of cell adhesion by RGDS peptides, which reduced mitochondrial translation (Fig. 3D and Supplementary Fig. 3L), hampered the oxygen consumption rate of the mitochondria (Fig. 3F), as did the mitochondrial translation inhibitor chloramphenicol. Moreover, under simulated microgravity, the relative contribution of mitochondria to ATP production was decreased (Fig. 3G), possibly compensated by glycolysis (Supplementary Fig. 3P). Consistently, metabolites associated with the tricarboxylic acid (TCA) cycle were reduced by simulated microgravity (Supplementary Fig. 3Q). These data suggested that gravity is required to maintain the metabolic integrity of the organelle.

The FAK–RAC1–PAK1–BAD–Bcl-2 family protein axis conveys a signal from cell adhesion to mitochondrial protein synthesis

Given the phosphorylation and activation of FAK upon laminin–integrin interaction^50,51 (Supplementary Fig. 3D), we investigated the importance of this factor for mitochondrial translation. Indeed, FAK knockdown (Supplementary Fig. 4A) resulted in the loss of laminin-mediated activation of in organello translation (Fig. 4A and Supplementary Fig. 4B).

We next asked how the downstream signaling pathway(s) of FAK connect to mitochondria. Since FAK influences widespread enzymes such as small GTPases (RAC1, CDC42, RHOA, and KRAS), phosphatidylinositol 3-kinase (PI3K), and Jun N-terminal kinase (JNK)^50,51 (Supplementary Fig. 4C), these activities were pharmacologically inhibited (Supplementary Fig. 4D), and the potential for laminin-mediated mitochondrial translation activation was evaluated via on-gel mito-FUNCAT. Screening showed that RAC1 inhibition specifically hampered laminin-mediated mitochondrial translation activation (Fig. 4B and Supplementary Fig. 4E, F). Since phosphorylated FAK increases the active GTP-bound form of RAC1^50,51, we probed this fraction by the pulldown with GST-tagged PAK1 p21-binding domain (PBD) (Fig. 4C). As expected, laminin treatment led to the accumulation of the active form of RAC1 (Fig. 4C).

Further compound treatment narrowed the downstream pathway of RAC1⁵² to PAK1 but not to ARP2/3 (Fig. 4D and Supplementary Fig. 4G–I). We observed elevated phosphorylation of PAK1, which is mediated by GTP-activated RAC1⁵², in laminin-treated cells (Fig. 4E).

Given that PAK1 couples with the pro-survival cascade driven by Bcl-2 family proteins on the mitochondrial outer membrane^53–55, we reasoned that the Bcl-2 family pathway may participate in cell adhesion-mediated mitochondrial translation activation. We treated cells with pan-Bcl-2 family inhibitors (such as obatoclax and TW-37) and found that the effect of laminin on mitochondrial protein synthesis was counteracted (Fig. 4F and Supplementary Fig. 4J–L).

To regulate Bcl-2 family proteins, PAK1 phosphorylates BAD⁵⁶, one of the pro-apoptotic Bcl-2 homology 3 (BH3)-only proteins that bind to Bcl-2 family proteins on the mitochondrial outer membrane and inhibit the pro-survival effect^57–59. BAD phosphorylation reduces its affinity for Bcl-2 family proteins, dissociating BAD from Bcl-2 proteins and thus from the mitochondrial outer membrane^57–59. Indeed, laminin treatment decreased the level of the mitochondria-associated fraction of BAD (Fig. 4G).

Therefore, we examined the role of BAD in mitochondrial translation. During our analysis of the knockout of BAD in HAP1 cells (Supplementary Fig. 4M), we noticed a reduction in mitochondrial mass in this cell line (Supplementary Fig. 4N bottom). To normalize the mitochondrial volume and simultaneously determine the extent of protein synthesis in the organelle, we harnessed mito-FUNCAT coupled with fluorescence-activated cell sorting (FACS)³⁴. The mito-FUNCAT FACS revealed that the deletion of BAD resulted in a high level of net mitochondrial translation due to the loss of the suppressive mechanism for mitochondrial translation (Fig. 4H and Supplementary Fig. 4N). Ultimately, BAD deficiency led to an attenuated response to laminin in mitochondrial translation (Fig. 4I and Supplementary Fig. 4O).

Taken together, our results revealed the signal transduction pathway from the extracellular matrix laminin to the Bcl-2 family protein on the mitochondrial outer membrane for in organello translation activation.

Laminin-mediated mitochondrial translation activation is distinct from known mechanisms

These findings prompted us to investigate the key molecular events in the matrix where mitochondrial translation occurs. For this purpose, we tested the possibility of the involvement of known regulatory mechanisms of mitochondrial translation.

The membrane potential is a prerequisite for mitochondrial protein synthesis^60–62 (Supplementary Fig. 5A, B). However, alterations in the membrane potential could not explain the dynamics of mitochondrial translation upon cell adhesion modulation (Supplementary Fig. 5C–F).

Mitochondrial translation has been reported to occur at cristae membranes³³, where OXPHOS complexes are enriched^63–65. Indeed, we observed that more cristae formed with laminin treatment in HEK293 cells through electron microscopy (Supplementary Fig. 5G–I); both at 3 h (Supplementary Fig. 5J–K) and 12 h (Fig. 3C and Supplementary Fig. 3F) when mitochondrial translation enhancement was observed. Moreover, the loss of the inner membrane-located OPA1, which leads to oligomerization and subsequent tubulation of the inner membrane to form cristae structure⁶⁶, in MEF cells⁶⁷ reduced the basal level of mitochondrial translation (Supplementary Fig. 5L–M). However, the OPA1 KO cells still retained the potential to activate mitochondrial translation upon laminin treatment (Supplementary Fig. 5L, N), indicating that the two translation regulatory mechanisms are independent. Since we observed the laminin-mediated enhancement of cristae formation and mitochondrial translation were found simultaneously, we could not explicitly conclude the causality between them (Fig. 3C, Supplementary Figs. 3F and 5G–K). However, it would be a probable scenario that the activated mitochondrial translation increases the formation of the respiratory chain super complex and subsequent inner-membrane curvature and tubulation⁶⁸.

In yeast, rapid communication between the cytosolic and mitochondrial translation systems has been reported⁶⁹; cytosolic translation synchronizes its output with mitochondrial protein synthesis to maintain the synthesis rate of OXPHOS subunits. However, our ribosome profiling did not detect a large alteration in OXPHOS complex subunits encoded in the nuclear genome (Supplementary Figs. 1F, O, and 2C). Moreover, a recent report suggested that the synchronization mechanism is not conserved in humans⁷⁰.

mtFAS-coupled mitochondrial translation activation

We subsequently tested whether mtFAS (Fig. 5A) is involved in laminin-induced mitochondrial translation. mtFAS exploits manifold enzymes to covalently extend acyl chains on acyl carrier proteins (ACPs) (Fig. 5A)⁷¹. The eight-carbon fatty acid octanoate on ACP is converted into lipoic acid, generating lipoylated proteins (Fig. 5A)⁷¹. Since the loss of this metabolic pathway suppresses in organello translation in yeasts^71–73, we reasoned that the involvement of mtFAS in laminin-mediated mitochondrial translation activation. Indeed, we found that laminin treatment increased mtFAS activity, assessed by lipoylation (Fig. 5B). Conversely, blocking the downstream signal transduction pathway with the Bcl-2 family inhibitor TW-37 was associated with low mtFAS efficiency (Fig. 5C).

Fig. 5. mtFAS balances manlonyl-CoA for malonylation of the mitochondrial translation machinery and thus for translation in organello.

A Schematic representation of acetyl-CoA, malonyl-CoA, and the downstream pathways in the mitochondrial matrix. B, C Western blotting of DLAT and lipoylated DLAT proteins under the indicated conditions. The quantified relative amount of lipoylated DLAT is shown. D, H–J Quantification of on-gel FUNCAT experiments to monitor mitochondrial translation under treatment with the indicated compounds at the following concentrations: C75, 50 μM; nicotinamide, 10 mM; NRD167, 10 μM; and 3-TYP, 10 µM. E On-gel FUNCAT experiments to monitor mitochondrial translation in C75-treated cells with a titrated concentration of laminin for precoating. F Western blotting of proteins with malonylated lysine (Mal-K) in the mitochondrial IP (MitoIP) fraction with an anti-TOMM22 antibody, and then the pellet of the sucrose cushion ultracentrifugation fraction. MRPS22 (mS22) was used as the loading control. The quantified relative amount of Mal-K signal is shown. G Malonylated lysine residues (Mal-K) found in the compendium of protein lysine modifications (CPLM) database⁷⁶ were mapped on the mitoribosome complexed with mtEFG1 (6VLZ)¹⁰³, using UCSF ChimeraX^123–127. KIn vitro translation of mammalian mitochondria reconstituted with recombinant factors. CoA, acetyl-CoA, or malonyl-CoA was added to the reaction mixture. A representative nLuc reporter mRNA was used to probe basal translation activity in the system. Supplementary Fig. 6S shows the effect of CoA on the reaction. L High-resolution measurement of mitochondrial translation (n = 13) with MitoIP-Thor-Ribo-Seq. The fold change in mitoribosome footprints by laminin precoating was measured in the presence or absence of C75. M Mitochondrial translation elongation rate measured by the mitoribosome run-off assay with retapamulin and MitoIP-Thor-Ribo-Seq. The regression line for the expansion of the mitoribosome-free area over time determines the mitochondrial translation elongation rates in the presence or absence of laminin precoating. For B–D, F, H–J, the data from replicates (points, n = 5 for B; n = 3 for C, D, F, H–J), the mean values (bars), and the s.d.s (errors) are shown. The p-values were calculated by Student’s t test (two-tailed). For E, K, the data from three replicates (E) and nine replicates (K) (points) and the mean values (bars) are shown. The p-values in K were calculated by the Tukey‒Kramer test (two-tailed). For L, the p-value was calculated by the Mann‒Whitney U test (two-tailed). For M, the p-value was calculated by two-way ANOVA. In box plots (L), the median (centerline), upper/lower quartiles (box limits), and 1.5× interquartile range (whiskers) are shown. All the experiments except for G, K were conducted in HEK293 cells. See also Supplementary Figs. 5–6, Supplementary Data 1, and Supplementary Data 2.

Harnessing C75, an inhibitor of the 3-ketoacyl-ACP synthase OXSM (Fig. 5A)⁷⁴, we found that the inactivation of mtFAS (Supplementary Fig. 6A) led to reduced basal mitochondrial translation (Fig. 5D and Supplementary Fig. 6B). Strikingly, the C75 treatment attenuated the translational response to laminin (Fig. 5E and Supplementary Fig. 6B). Thus, mtFAS activity is coupled with mitochondrial translation.

Protein demalonylation in mitochondria leads to mitochondrial translation

Despite the reported linkage between mtFAS and mitochondrial translation in yeasts⁷¹, the underlying mechanism has remained largely unknown. In the cytosol, the counterpart fatty acid synthase (FASN) balances the abundance of malonyl-CoA; otherwise, excess malonyl-CoA generated by FASN impairment induces malonylation on lysine residues⁷⁵. Given that, we envisioned a similar scenario in mitochondria (Fig. 5A). We observed that laminin-mediated mtFAS activation (Fig. 5B) occurred without alteration of protein levels of related enzymes (Fig. 5A and Supplementary Fig. 6C–E). The increased mtFAS activity was associated with the demalonylation of mitochondrial proteins, especially mitochondrial translational machinery (such as mitoribosomal proteins) (Supplementary Fig. 6F, MitoIP + sucrose cushion). Conversely, mtFAS inhibition by C75 increased the malonylated fraction of the translational machinery proteins (Fig. 5F). Through the lysine modifications assessed by mass spectrometry⁷⁶, diverse solvent-accessible lysines in mitochondrial ribosome proteins, elongation factors, and recycling factors have been identified as malonylation sites (Fig. 5G, Supplementary Fig. 6G–I, and Supplementary Data 1), suggesting their functional implications in translation regulation.

To test the importance of mitochondrial protein malonylation in the regulation of protein synthesis, we directly modulated the protein malonylation status in mitochondria. Sirt5, a sirtuin family protein, is a major demalonylase in mitochondria^77–79. Chemical perturbation of this protein by the pan-sirtuin inhibitor nicotinamide or by the Sirt5-specific inhibitor NRD167 suppressed mitochondrial translation (Fig. 5H, I and Supplementary Fig. 6J–M), leading to hyper-malonylation of the translation machinery (Supplementary Fig. 6N) independent of mtFAS alteration (Supplementary Fig. 6O, P). Notably, in combination with laminin treatment, Sirt5 inhibition had limited effects on mitochondrial translation (Supplementary Fig. 6J–M), probably due to the unavailability of malonyl-CoA for further hyper-malonylation by activated mtFAS.

In addition to malonylation, lysines are often subjected to acetylation by acetyl-CoA donors in mitochondria⁷⁹. However, the inhibition of Sirt3, a mitochondrial deacetylase^79–81, by 3-TYP did not hamper mitochondrial translation but rather increased mitochondrial translation (Fig. 5J and Supplementary Fig. 6Q, R), suggesting the acyl group-dependent function of the lysine modifications for in organello translation.

A high concentration (millimolar range) of acetyl-CoA/malonyl-CoA in the matrix leads to nonenzymatic lysine acylation^79,82. To investigate the direct effect of acetyl-CoA and malonyl-CoA on mitochondrial protein synthesis, we employed a reconstituted translation system with recombinant factors (Fig. 5K left)^83,84. We observed that, compared to the CoA control, malonyl-CoA inhibited mitochondrial translation (Fig. 5K right and Supplementary Fig. 6S), whereas acetyl-CoA had a relatively weaker impact. Taken together, these data led us to conclude that a low malonyl-CoA supply resulting from activated mtFAS and the subsequent reduction in lysine malonylations in the translation machinery leads to enhanced mitochondrial protein synthesis.

Both initiation and elongation are controlled by protein malonylation

Given that malonylated lysines were found in mitoribosomal proteins close to initiation/elongation factors or those factors per se (Fig. 5G and Supplementary Fig. 6G–I, and Supplementary Data 1), we investigated the impact of this regulatory system on mitochondrial translation initiation and elongation. For this purpose, we employed MitoIP-Thor-Ribo-Seq, a high-resolution ribosome profiling derivative tailored for mitochondrial translation⁸⁵. Since translation initiation is a general rate-limiting step of mitochondrial protein synthesis⁸⁵, the overall number of mitoribosome footprint reads represents a proxy for initiation speed. We observed a global increase in the mitoribosome loading on mRNAs following laminin treatment (Fig. 5L), whereas C75 blocked this phenotype.

We also applied a retapamulin-assisted mitoribosome run-off assay with MitoIP-Thor-Ribo-Seq⁸⁵ to monitor the elongation rate. Retapamulin traps mitoribosomes only at the start codons and allows active elongation of mitoribosomes to complete protein synthesis. Thus, a time-course chase with chloramphenicol after retapamulin treatment generates a mitoribosome-free area downstream of the start codon, measuring the translation elongation rate⁸⁵. This assay revealed that the elongation of mitochondrial ribosomes was enhanced by laminin treatment (Fig. 5M). Thus, both the initiation and elongation steps of mitochondrial translation were regulated through the cell adhesion-mtFAS axis.

Mechanical stress enhances mitochondrial protein synthesis in vivo

The correspondence between mitochondrial translation and laminin–integrin-mediated cell adhesion led us to study the physiological importance of this cascade. Considering that integrin plays a critical role as a mechanosensor^86–89, we hypothesized that mechanical stress promotes mitochondrial translation. Given that mechanical stress and gravity maintain skeletal muscle homeostasis and FAK phosphorylation^{43–45,90,91}, we focused on translation in the tissues of mice. In a minimized mechanical stress (MMS) model, mice underwent hindlimb unloading and immobilization, and the soleus was subjected to ribosome profiling and RNA-Seq (Fig. 6A)⁹². During MMS, we found reduced mitochondrial translation at 14 d after the start of minimized mechanical stress (Fig. 6B, C), the time at which we detected reduced muscle weight (Supplementary Fig. 7A). Our results indicated the importance of mechanical stress for mitochondrial protein synthesis.

Fig. 6. Minimized mechanical stress reduces mitochondrial translation.

A Schematic representation of experiments with minimized mechanical stress (MMS) in mouse hindlimbs. B MA plots for ribosome footprint change (left), RNA abundance change (middle), and translation efficiency change (right) after 14-d MMS in the mouse soleus. Significantly altered transcripts (false discovery rate [FDR] <0.05) and mitochondrial genome-encoded mRNAs are highlighted. C Box plots for ribosome footprint change (left), RNA abundance change (middle), and translation efficiency change (right) in mitochondrial genome-encoded mRNAs (n = 13) over the course of MMS. D Schematic representation of the model of gravitational and mechanical force-mediated mitochondrial translation activation. For C, the p-value was calculated by the Mann‒Whitney U test (two-tailed). In box plots (C), the median (centerline), upper/lower quartiles (box limits), and 1.5× interquartile range (whiskers) are shown. See also Supplementary Fig. 7 and Supplementary Data 2.

Discussion

Starting from the translatome analysis of space-flown samples, we elucidated a previously unexplored cell signaling pathway that connects cell adhesion sensing, the mitochondrial malonyl-CoA balance, and in organello translation (Fig. 6D). Although translational reprogramming mediated by integrins and mechanical forces has been suggested for decades^93–95, the landscape has remained elusive. Our study revealed mitochondrial translation as a subject of regulation. In muscle, which is constantly subjected to mechanical stresses through loading, postural maintenance, and exercise, this system enables rational regulation of energy production in mitochondria; increased mechanical stresses lead to increased mitochondrial protein synthesis and thus modulate the organelle’s functions.

Our results closely aligned with the multiomics approach utilizing data from NASA’s GeneLab platform^3,7,8. This dataset, which included an astronaut cohort and space-flown samples, illustrated mitochondrial damage as a hallmark of spaceflight. Consistent with our data, space-flown samples show the reduced metabolic activity of mitochondria³. Although reactive oxygen species (ROS) have been suggested to be a key mediator of the stress response in the study³, our study provided another (not mutually exclusive) source of mitochondrial dysfunction; the attenuation of mitochondrial translation by microgravity may prime the damage processes in space. In addition to the basal attenuation of cell adhesion by microgravity^40–45, space irradiation could damage the ECM, such as laminin⁹⁶, further mitigating mitochondrial translation.

Ultimately, the unbalanced subunit supply from the mitochondria matrix and cytoplasm for the OXPHOS complex (Supplementary Fig. 1F, O) may reduce ATP production of mitochondria and ultimately provide stress to the organelle in space. Although the cytosolic flux of OXPHOS subunits was not largely affected under microgravity in human cells (Supplementary Fig. 1F, left) and C. elengans (Supplementary Fig. 1O, left), whole body analysis of C. elengans showed a reduction at the RNA level (Supplementary Fig. 1O, middle), indicating compensation by translation efficiency (Supplementary Fig. 1O, right). Because this phenomenon was not observed in cultured cells (Supplementary Fig. 1O, middle and right), the complex response to microgravity may exist at the tissue level or may be specific to certain species.

We also observed other alterations in cytoplasmic mRNAs under microgravity (Supplementary Fig. 1E). While attenuated mRNA expression of actin filament-related genes was often found in the space experiments^97–100, our analysis instead showed the reduction of mRNA related to intermediate filaments, another cytoskeleton (Supplementary Fig. 1E). This decreased RNA expression may explain, at least partially, the looser structure of intermediate filaments observed during microgravity¹⁰¹.

By small-scale chemical screening, we outlined the signaling cascade involved in cell adhesion to the mitochondrial outer membrane (Fig. 6D). Earlier studies indicated that PAK1 phosphorylates BAD in two pathways: directly or indirectly through Raf1⁵⁶. Since we could not detect a significant effect of the Raf1 inhibitor (Supplementary Fig. 7B–E), direct phosphorylation by PAK1 may be a major pathway under our conditions, although our data did not exclude a variable balance of the direct and indirect pathways across cells. Although the pan-Bcl-2 family protein inhibitors obatoclax and TW-37 blocked laminin-activated mitochondrial translation (Fig. 4F), the same phenotype was not observed for the compounds specific for each Bcl-2 family protein (Supplementary Fig. 7F–I), suggesting that Bcl-2 family proteins perform redundant functions in enhancing mitochondrial translation.

This study provides a framework for mitochondrial translation activation primed by laminin–integrin but simultaneously raises new questions. It remains unclear how the Bcl-2 family proteins on the mitochondrial outer membrane transmit signals to mtFAS in the mitochondrial matrix. The pro-survival effect of Bcl-2 family proteins relies on their binding to BAX to suppress the oligomerization of BAX, a driver of the apoptotic cascade⁵⁹. However, the function of the Bcl-2 family in activating mitochondrial translation may not be based on this conventional mechanism, since neither the BAX activator nor the inhibitor impacted mitochondrial translation (Supplementary Fig. 7J–M). Consistent with this, we did not observe apoptosis induction under simulated microgravity (Supplementary Fig. 7N), which leads to mitochondrial translation repression through the proposed cascade inhibition.

Furthermore, how malonylation in the translation machinery affects the process of initiation and elongation should be addressed in future studies. Malonylation may affect the interaction between mitoribosomes and other factors. For example, malonylated lysines in uL12m, which consists of the L7/L12 stalk¹⁰², were located at the interface with mtIF2, mtEFG1, and mtEFG2 (Fig. 5G and Supplementary Figs. 6G, I). Similarly, the interaction between K191 of uL11m and E562 of mtEGF1¹⁰³ could be interfered with malonylation of K191 and K192 of uL11m (Fig. 5G).

Limitations of the study

Our results did not exclude the possibility that multiple mechanisms other than protein malonylation play an important role in mitochondrial translation response to gravity, cell adhesion, and mtFAS. Also, our work did not address the regulatory processes of cytosolic translation that occurred in a subset of mRNAs (Fig. 1B and Supplementary Fig. 1E, L, M). Although we observed microgravity-mediated repression of mitochondrial translation efficiency in C. elegans, the effects were relatively subtle compared to those in human tissue culture. This may stem from the protective/counteractive effects of the extracellular matrix at the organism level. To fill the gap in the responses at the cell level and the whole animal level, 3D organoids will serve as a useful model. At the molecular level, although our in vitro reconstitution system showed that malonyl-CoA, rather than acetyl-CoA, led to the reaction repressive, this artificial setup might not reflect the physiological situation. Further study will clarify the whole picture of translational control mediated by external gravitational and mechanical forces.

Methods

Cell culture

HEK293 cells (ATCC, CRL-1573, female), C2C12 cells (ATCC, CRL-1772, female), 3T3 cells (kind gifts from Shinichi Nakagawa’s laboratory, male), naïve HAP1 cells (Horizon Discovery, C631, male), BAD KO HAP1 cells (Horizon Discovery, 29-bp deletion, HZGHC005959c002, male), naïve MEF cells (ATCC, CRL-2991, gender unspecified), and OPA1 KO MEF cells (ATCC, CRL-2995, gender unspecified) were cultured in DMEM (high glucose, GlutaMAX Supplement, Thermo Fisher Scientific, 10566016) supplemented with 10% FBS (Sigma‒Aldrich) (for HEK293, C2C12, and 3T3), DMEM supplemented with 10% FBS, 1 × Na-pyruvate (Nakalai, 06977-34) and 0.05 g/L uridine (TCI, U0020) (for MEF), or IMDM (GlutaMAX Supplement, Thermo Fisher Scientific, 31980030) supplemented with 10% FBS (Sigma‒Aldrich) (for HAP1) at 37 °C with 5% CO₂, according to the manufacturer’s instructions. The absence of Mycoplasma-free culture was confirmed by an e-Myco VALiD Mycoplasma PCR Detection Kit (iNtRON Biotechnology). A disposable cultivation chamber (DCC) (JAXA and Chiyoda Corporation)¹⁰⁴ was used to culture the cells on a 3D clinostat (5.5 rpm on the X axis and 6.5 rpm on the Y axis; AES, the acceleration was verified empirically) and a centrifuge (AES) in the CO₂ incubator. The DCC was filled with medium to prevent shear stress caused by air bubbles.

Compounds

Cells were treated with the following compounds at the indicated concentrations: cycloheximide (Sigma‒Aldrich, C4859, 100 μg/ml), chloramphenicol (FUJIFILM Wako Pure Chemical Corporation, 030-19452, 100 μg/ml), anisomycin (Alomone Labs, A-520, 100 μg/ml), RGDS peptide (Tocris, 3498, 10 μg/ml), NSC23766 (Tocris, 2161, 200 μM), ZCL278 (Tocris, 4794, 100 μM), Rhosin (Tocris, 5003, 50 μM), BAY-293 (Tocris, 6857, 1 μM), LY294002 (Sigma‒Aldrich, L9908, 10 μM), SP600125 (Sigma‒Aldrich, S5567, 20 μM), IPA-3 (Tocris, 3622, 5 μM), CK666 (Tocris, 3950, 25 μM), obatoclax (Selleckchem, S1057, 2 μM), TW-37 (Selleckchem, S1121, 10 μM), valinomycin (Tocris, 3373, 2 μM), sorafenib (Selleckchem, S1040, 1 μM), GW5074 (Tocris, 1381, 5 μM), ZM336372 (Tocris, 1321, 10 μM), venetoclax (Selleckchem, S8048, 10 μM), A-1331852 (Selleckchem, S7801, 10 μM), S63845 (Selleckchem, S8383, 5 μM), BAI1 (Tocris, 2160, 1 μM), BTSA1 (Selleckchem, S8650, 20 μM), BAM7 (Tocris, 4810, 20 μM), C75 (Tocris, 4810, 50 μM), nicotinamide (Sigma‒Aldrich, N3376, 10 mM), NRD167 (Selleckchem, S9903, 10 μM), and 3-TYP (Selleckchem, S8628, 10 μM). Chloramphenicol was dissolved in 70% ethanol, nicotinamide was dissolved directly in the medium, and the other compounds were dissolved in dimethyl sulfoxide (DMSO).

Spaceflight experiments

DCCs were pretreated with laminin overnight as described below. Then, 5 × 10⁵ HEK293 cells were seeded in each DCC and incubated overnight. The medium was exchanged for CELLBANKER 1 plus (TaKaRa), and the cells were stored at −80 °C. The frozen cells were then sent to the ISS by Cygnus NG-14 in October 2020. In the Japanese Experiment Module KIBO in the ISS, the samples were thawed, exchanged for DMEM, high glucose, GlutaMAX Supplement (Thermo Fisher Scientific), supplemented with 10% FBS using a Pre-Fixation Kit-III (PFK-III, JAXA)¹⁰⁴ and cultured under centrifugation at 1 × g in the Cell Biology Experiment Facility (CBEF)¹⁰⁵ at 37 °C with 5% CO₂ and 80 ± 5% RH for 24 h. Subsequently, the cells were cultured under microgravity for 24 and 48 h, while the control cells were maintained at 1 × g. For chemical fixation, the cell culture media were replaced with CELLBANKER 1 plus containing 100 µg/ml cycloheximide and 100 µg/ml chloramphenicol by PFK-III. The samples were subsequently stored in a −80 °C Laboratory Freezer for ISS (MELFI) at −95 °C. The samples were returned to the ground by SpaceX CRS-22 (SpX-22) in July 2021 and shipped to the laboratory for ribosome profiling and RNA-Seq.

Space-flown C. elegans samples prepared for earlier experiments^16,29 were used in this study.

Animals

All animal experiments were authorized and approved by the Animal Care and Use Committee of The University of Tokyo. Adult male (7 weeks old) C57BL/6J mice were purchased from Sankyo Lab Service Corporation, housed in the facility under the supervision of the IACUC at 18–22 °C, 40–60% humidity with a 12-h:12-h light–dark cycle, and allowed free access to food and water.

Minimized mechanical stress (MMS)

The MMS model was developed by one of the coauthors and will be published elsewhere. First, we applied adhesive tape to the tails of 8-week-old mice. The knee joints of the mice were then kept immobilized using plastic cylinders. The feet of the mice were also taped to prevent the plastic cylinders from sliding off. Additional adhesive tape was applied between the rear limbs so that the mice could not bite the adhesive tape around their tail. Following joint fixation, the tails of the mice were suspended by attaching a spring clip to the adhesive tape. The spring clip was connected to a piano wire that was attached to the cage lid so that the mice could move in their cages. Food was left on the floor, and a water bottle was fixed to the wall with double-sided tape for easy access.

Mice were sacrificed at 1, 3, and 14 d after MMS started. For the control, age-matched mice were harvested (8 weeks old and 10 weeks old). The soleus muscles were dissected, and the wet weights were measured. The muscles were immediately flash-frozen in liquid nitrogen and stored at −80 °C.

Ribosome profiling and RNA-Seq

Lysate preparation

Spaceflight samples, HEK293

The cells in a DCC were thawed, washed with ice-cold PBS, lysed with 500 µl of lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 1 mM dithiothreitol [DTT], 1% Triton X-100, 100 µg/ml cycloheximide, and 100 µg/ml chloramphenicol), and triturated ten times in a syringe with a 30-gauge needle (NIPRO Corporation). The lysate was incubated with 25 U/ml TURBO DNase (Thermo Fisher Scientific) and clarified by centrifugation at 20,000 × g for 10 min at 4 °C.

Spaceflight samples, C. elegans

Frozen nematode cultures (1.5–4.5 ml) were thawed in 10 ml of wash buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 100 µg/ml cycloheximide, and 100 µg/ml chloramphenicol). The adult worms were filtered through a 30-µm ÜberStrainer (pluriSelect Life Science) and washed three times with wash buffer. Then, the nematodes were pelleted by centrifugation at 1600 × g for 1 min at 4 °C, resuspended in 600 µl of lysis buffer, and dripped into liquid nitrogen. After pulverization using a Multi-beads Shocker (Yasui Kikai) at 2800 rpm for 10 s, the lysate was thawed at 4 °C, incubated with 25 U/ml Turbo DNase (Thermo Fisher Scientific), and clarified by centrifugation at 20,000 × g for 10 min at 4 °C.

Simulated microgravity samples

Cells in a DCC were washed with PBS and lysed with 500 µl of lysis buffer. After 25 U/ml TURBO DNase treatment (Thermo Fisher Scientific), the lysate was clarified by centrifugation at 20,000 × g for 10 min at 4 °C.

Mouse soleus muscles

Frozen soleus muscles were pulverized with 600 µl of frozen lysis buffer containing cOmplete, EDTA-free Protease Inhibitor Cocktail (Roche) and 1000 U/ml RNase inhibitor, Murine (New England Biolabs [NEB]), using a Multi-beads Shocker (Yasui Kikai) at 3000 rpm for 30 s for 2 cycles. After thawing at 4 °C, the lysate was incubated with 25 U/ml Turbo DNase (Thermo Fisher Scientific) and clarified by centrifugation at 20,000 × g for 10 min at 4 °C.

Library preparation for ribosome profiling (standard Ribo-Seq and Thor-Ribo-Seq)

For the simulated microgravity samples, ribosome profiling library preparation was performed as previously described¹⁰⁶. The other libraries were generated by the Thor-Ribo-Seq protocol as described earlier²⁶.

The lysate was treated with 20 U of RNase I (LGC Biosearch Technologies) at 25 °C for 45 min. A sucrose cushion was used to collect ribosomes. The RNA was run on a 15% UREA PAGE gel. RNA fragments ranging from 17 to 34 nucleotides (nt) were excised from the gel, dephosphorylated, and ligated with linkers. rRNAs were depleted with the Ribo-Zero Gold rRNA Removal Kit (Human/Mouse/Rat) (Illumina) for human and mouse samples and with Caenorhabditis elegans Ribo-Seq riboPOOLs (siTOOLs Biotech) for nematode samples. For the simulated microgravity samples, cDNA was reverse-transcribed, circular-ligated, and PCR-amplified. For the other samples, after hybridization of an oligonucleotide to the T7 promoter region of the linker, complementary RNAs were transcribed by a T7-Scribe Standard RNA IVT Kit (CELLSCRIPT) and then ligated to the second linker. cDNA was reverse-transcribed and PCR-amplified. The DNA libraries were sequenced on a HiSeq 4000 platform in 50-bp single-read mode (simulated microgravity experiments) or on a HiSeq X Ten platform in 150-bp paired-end mode (the other experiments).

Library preparation for RNA-Seq

For RNA-Seq, total RNA was extracted from the same lysate used for ribosome profiling with TRIzol LS (Thermo Fisher Scientific) and a Direct-zol RNA Microprep Kit (Zymo Research). rRNA depletion was conducted with the Ribo-Zero Gold rRNA Removal Kit (Human/Mouse/Rat) (Illumina) for human and mouse samples and with Caenorhabditis elegans Ribo-Seq riboPOOLs (siTOOLs Biotech) for nematode samples.

For HEK293 cells in spaceflight experiments and mouse soleus muscles, libraries were prepared with a SMARTer Stranded Total RNA-Seq Kit v3 - Pico Input (TaKaRa). The other RNA-Seq libraries were generated with a TruSeq Stranded mRNA Library Prep Kit (Illumina). The DNA libraries were sequenced on a HiSeq 4000 platform in 50-bp single-read mode (s-µg experiments) or on a HiSeq X Ten platform in 150-bp paired-end mode (the other experiments).

MitoIP-Thor-Ribo-Seq

Lysate preparation

Lysate was prepared as previously reported⁸⁵. Cells in 15-cm dishes were washed with ice-cold PBS and lysed with 1200 µl of hypotonic buffer (10 mM HEPES-KOH pH 7.5, 10 mM KCl, 1.5 mM MgCl₂, 1 mM DTT, 100 µg/ml cycloheximide, and 100 µg/ml chloramphenicol). The mitochondrial fraction in the lysate was purified through immunoprecipitation with an anti-TOMM22 antibody from the Mitochondria Isolation Kit, Human (Miltenyi Biotec), according to the manufacturer’s instructions with minor modifications. Then, the eluted mitochondria were pelleted by centrifugation at 7000 × g for 10 min at 4 °C and resuspended in modified lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 15 mM MgCl₂, 1 mM DTT, 1% Triton X-100, 100 µg/ml cycloheximide, 100 µg/ml chloramphenicol, and 2.5 U/ml TURBO DNase). The lysate was clarified by centrifugation at 20,000 × g for 10 min at 4 °C for library preparation.

Library preparation

Library preparation was performed as previously described⁸⁵. The lysate was incubated with 40 U of RNase I (LGC Biosearch Technologies) in a 50-µl reaction (scaled up by the modified lysis buffer) at 25 °C for 45 min. After sucrose cushion ultracentrifugation and Ribo-FilterOut¹⁰⁷, RNA fragments ranging from 17 to 50 nt were collected.

From linker-ligated RNA fragments, rRNAs were depleted with Human Ribo-Seq riboPOOL (siTOOLs Biotech). Then, an oligonucleotide was hybridized to the T7 promoter region of the linker to generate the dsDNA T7 promoter. Complementary RNAs were transcribed with a T7-Scribe Standard RNA IVT Kit (CELLSCRIPT). After ligation of the second linker, cDNA was reverse-transcribed and PCR-amplified. The DNA libraries were sequenced on a HiSeq X Ten in 150-bp paired-end mode.

Data analysis

The deep sequencing data were processed as described previously^28,108, with modifications. For ribosome profiling with paired-end sequencing, base correction was performed by Fastp (version 0.21.0)^109, and read 1 was used for downstream analysis. After read quality filtering and adapter trimming by Fastp, all reads aligned to noncoding RNAs (rRNAs, tRNAs, mt-rRNAs, mt-tRNAs, snRNAs, snoRNAs, and miRNAs) were removed using STAR (version 2.7.0a)¹¹⁰. The reads were subsequently mapped to the corresponding nuclear genomes (human, hg38; mouse, mm10; C. elegans, WBcel235) and custom-made mitochondrial transcript sequences by STAR. UMI suppression in Thor-Ribo-Seq was performed using UMItools (version 1.1.2)¹¹¹. For ribosome profiling, ribosomal A-site position offsets from the 5′ end of reads were empirically estimated along the read length, as summarized in Supplementary Data 2. For RNA-Seq, the offset was set to 15.

To count reads from CDSs, reads assigned to the first and last 5 amino acids were excluded. The relative enrichment of reads across CDSs in ribosome profiling and RNA-Seq was calculated by the DESeq2 package¹¹². Additionally, translation efficiency was calculated by a generalized linear model with the same package. Gene Ontology analysis was performed with DAVID^113,114. Mitocarta3 genes were defined at https://www.broadinstitute.org/files/shared/metabolism/mitocarta/human.mitocarta3.0.html)¹¹⁵. The mtUPR target genes were defined in an earlier study³⁹.

The global mitochondrial elongation rate was calculated as previously reported⁸⁵. Smoothed metagene profiles were generated from the mitochondrial transcripts, excluding MT-ATP6 and MT-ND4 due to the bicistronic nature. The mean read density from 100 to 200 codons was normalized to 1.

Laminin precoating

DCCs, 6-well plates, 12-well plates, and 96-well plates were precoated with 0.25 µg/cm² (or dilutions as indicated in the figures) laminin (Easy iMatrix-511 silk, MATRIXOME) overnight at 4 °C before cell seeding.

Cell viability assay

Cell viability was monitored as described previously¹¹⁶. Cells were seeded on 96-well plates, treated with the indicated compounds for 12 h, and then incubated with RealTime-Glo MT Cell Viability Assay reagent (Promega) for 30 min. The luminescence was quantified by the GloMax Navigator System (Promega).

Cell adhesion assay

The wells of a 96-well plate were precoated with laminin as described above. One hundred microliters of HEK293 cell suspension (5 × 10⁵ cells/ml) was seeded and incubated for 1 h in a CO₂ incubator. After washing out the unattached cells with PBS, the dish-attached cells were fixed with 4% (w/v) paraformaldehyde in PBS for 20 min, permeabilized with 0.1% Triton X-100 in PBS, and stained with 0.2 μM CellTag 700 (LI-COR Biosciences) for 1 h. Images were acquired by an Odyssey CLx (LI-COR Biosciences) with an IR 700-nm channel.

On-gel mito-FUNCAT

On-gel mito-FUNCAT was conducted as described previously³⁴. Unless otherwise noted, cells were seeded on dishes precoated with or without laminin and incubated for 12 h with the indicated compounds before cell harvesting. Nascent peptides were labeled with 50 μM HPG (Jena Bioscience) in methionine-free medium (DMEM, high glucose, no glutamine, no methionine, no cystine [Thermo Fisher Scientific], supplemented with 48 µg/ml L-cysteine and 4.08 mM L-alanyl-L-glutamine) containing 100 mg/ml anisomycin for 3 or 4 h. Cells were washed with ice-cold PBS and lysed with FUNCAT lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM MgCl₂, and 1% Triton X-100). After clarification by centrifugation at 20,000 × g for 10 min at 4 °C, the supernatants were labeled with 50 μM IRdye800CW Azide (LI-COR Biosciences) via a Click-it Cell Reaction Buffer Kit (Thermo Fisher Scientific), according to the manufacturer’s instructions. The proteins were separated on SDS–PAGE gels. Images were acquired with an Odyssey CLx (LI-COR Biosciences) with an IR 800-nm channel. For signal normalization, the gels were stained with EzStain AQua (ATTO) or GelCode Blue Safe Protein Stain (Thermo Fisher Scientific) to detect total proteins, and images were acquired with an Odyssey CLx with an IR 700-nm channel. The signals on the gels were quantified with Image Studio (LI-COR Biosciences, version 5.2).

For the measurement of total protein synthesis, anisomycin was omitted from the experiments.

For the measurement of protein synthesis on the dish with an elastically supported surface (ESS), cells were cultured in μ-Dish 35 mm, high ECC 1.5 kPa, 15 kPa, and 28 kPa (ibidi), respectively.

For knockdown experiments, cells were transfected with 5 μM siRNA targeting FAK (Horizon Discovery, L-003164-00-0005) or control siRNA (Horizon Discovery, D-001206-13-20) using TransIT-X2 Reagent (Mirus Bio, MIR6003) in 10-cm dishes and incubated for 24 h. Then, the cells were reseeded on 12-well dishes with or without laminin coating as described above, cultured for an additional 24 h, and used for on-gel mito-FUNCAT experiments.

Mito-FUNCAT FACS

Mito-FUNCAT FACS was performed as described previously³⁴. HAP1 cells were cultured in methionine-free medium supplemented with 100 µM HPG and 100 mg/ml anisomycin for 3 h before cell harvesting. After washing with PBS, the cells were dissociated from the dish with 0.05% trypsin. The collected cells were washed, pre-permeabilized with 0.0005% digitonin in mitochondrial protective buffer (10 mM HEPES-KOH pH 7.5, 300 mM sucrose, 10 mM NaCl, and 5 mM MgCl₂) for 5 min at room temperature, fixed in 4% paraformaldehyde (PFA) for 15 min at 4 °C, and permeabilized with 0.1% (v/v) Triton X-100 in PBS for 5 min at room temperature. Subsequently, nascent peptides within mitochondria were labeled with 1 μM azide-conjugated Cy3 (Jena Bioscience) with a Click-it Cell Reaction Buffer Kit (Thermo Fisher Scientific). Mitochondria were immunostained with Alexa Fluor 647 Anti-TOMM20 antibody (Abcam, ab209606, 1:100) in Intercept Blocking Buffer (LI-COR Biosciences) for 1 h at 4 °C. The cells were washed three times with PBS and then analyzed via flow cytometry (BD FACSAria II, BD Biosciences).

³⁵S Met-mediated metabolic labeling of newly synthesized proteins

HEK293 cells were seeded on a 12-well plate precoated with or without laminin and incubated for 12 h before cell harvesting. To label mitochondrial translation products, cells were pulsed with 2.96 MBq (80 µCi) of ³⁵S Met (EasyTag L-[³⁵S]-Methionine, Revvity) in methionine-free medium containing 100 µg/ml anisomycin and incubated at 37 °C for 30 min. Cells were then washed with PBS and lysed with lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM MgCl₂, and 1% Triton X-100). After SDS–PAGE, the gels were stained with GelCode Blue Safe Protein Stain (Thermo Fisher Scientific) and dried on a gel drier. The dried gel was exposed to an Amersham Typhoon Scanner IP system (GE Healthcare) to visualize the radiolabeled proteins.

Mitochondrial DNA content measurement

Total DNA was extracted from 1 × 10⁶ cells using NucleoSpin DNA RapidLyse (MACHEREY-NAGEL). qPCR was performed with primers targeting the ND2 locus in the mitochondrial genome (5′-TGTTGGTTATACCCTTCCCGTACTA-3′ and 5′-CCTGCAAAGATGGTAGAGTAGATGA-3′) and those targeting the Alu repeat sequence in the nuclear genome (5′-CTTGCAGTGAGCCGAGATT-3′ and 5′-GAGACGGAGTCTCGCTCTGTC-3′)¹¹⁷ with TB Green Premix Ex Taq II (TaKaRa) on a Thermal Cycler Dice Real Time System II (TaKaRa). The relative mitochondrial DNA content was normalized to the nuclear-genome content by the ∆∆Ct method.

Microscopic analysis

The cells were cultured in a glass-bottomed DCC (AES), washed with PBS, fixed with 4% (w/v) paraformaldehyde in PBS for 15 min at room temperature, and subsequently incubated with methanol for 10 min at −20 °C. After washing twice with PBS, the cells were incubated with Intercept Blocking Buffer (LI-COR Biosciences) containing 0.2% Triton X-100 for 60 min at room temperature. Then, the mitochondria were immunostained in Intercept Blocking Buffer with 0.2% Triton X-100 and Alexa Fluor 647 Anti-TOMM20 antibody (Abcam, ab209606, 1:100) for 1 h at room temperature. The cells were washed with TBS containing 0.2% Triton X-100 three times and then incubated with VECTASHIELD Mounting Medium (VECTOR LABORATORIES, INC., H-1500) overnight at 4 °C in the dark. Images were obtained using an FV3000 confocal microscope (Olympus) with a 60× objective lens (Olympus Japan, UPLXAPO60XO) via the processes described below.

Morphological quantification of mitochondria

Image analysis was performed using MitoGraph (version 3.0), a fully automated C++ program for 3D mitochondrial structure analysis (https://github.com/vianamp/MitoGraph)^118,119. We first isolated 3D images of individual cells from the raw images using the FIJI macro to extract cellular trends and then applied them as inputs to the MitoGraph software. For the MitoGraph parameter, we used 0.414 for the flag-xy and 0.3 for the flag-z.

Growth assay

Cells in a DCC were dissociated from the dish with 0.05% trypsin and neutralized in DMEM supplemented with 5% FBS. Then, cells were stained with trypan blue, and live cells were quantified with Countess (Thermo Fisher Scientific).

Assessment of mitochondrial ATP production

Mitochondrial ATP production was monitored with Mitochondrial ToxGlo Assay (Promega) according to the manufacturer’s instructions. Cells in a DCC were dissociated from the dish with 0.05% trypsin and resuspended in glucose or galactose-supplemented medium. After cell counting by Countess (Thermo Fisher Scientific), cells were reseeded on 96-well plates and incubated with ATP detection reagent (Promega) for 10 min. The luminescence was measured by the GloMax Navigator System (Promega).

Glucose uptake assessment

Glucose uptake was assessed with Glucose Uptake-Glo Assay (Promega) according to the manufacturer’s instructions. Cells in a DCC were dissociated from the dish with 0.05% trypsin and resuspended in glucose-free medium. After cell counting by Countess (Thermo Fisher Scientific), cells were reseeded on 96-well plates and incubated with 2-deoxyglucose (Promega) for 10 min. Then, cells were treated with stop buffer (Promega), neutralization buffer (Promega), and 2DG6P reagent (Promega) for 60 min. The luminescence was quantified by the GloMax Navigator System (Promega).

Oxygen consumption rate (OCR)

After the cells were cultured on Seahorse XFe96 cell culture microplates (Agilent), the medium was replaced with Seahorse XF DMEM (Agilent) containing 1 mM pyruvate, 2 mM glutamine, and 10 mM glucose and then incubated at 37 °C for 1 h. The OCR was monitored on a Seahorse XFe96 Analyzer (Agilent) throughout sequential injections of 0.5 μM oligomycin, 0.5 μM FCCP, and 0.5 μM rotenone/antimycin A into the cell medium, according to the manufacturer’s instructions.

Metabolomic analysis

The cells on DCCs were cultured under simulated microgravity or standard gravity (at 1 × g) for 24 h, washed with 5% (w/w) mannitol (Fujifilm Wako Pure Chemical Corporation), and incubated with methanol (Fujifilm Wako Pure Chemical Corporation) for 30 s. Subsequently, the cells were incubated with 1 mM internal standard solution (Human Metabolome Technologies [HMT]) and collected in 1.5-ml tubes. The samples were centrifuged at 2300 × g for 5 min at 4 °C. The proteins were removed from the supernatant by ultrafiltration filters (HMT). Ionic metabolites were comprehensively measured at HMT using capillary electrophoresis time-of-flight mass spectrometry (CE-TOFMS) and capillary electrophoresis triple-quadrupole mass spectrometry (CE-QqQMS).

Apoptosis measurement

Apoptosis was monitored using the RealTime-Glo Annexin V Apoptosis Assay (Promega) according to the manufacturer’s instructions. Cells in a DCC were dissociated from the dish with 0.05% trypsin, resuspended in the medium, and 1.2 × 10⁴ cells were seeded into each well of a 96-well plate. After cell seeding, the detection reagents were added, and luminescence was measured using the GloMax Navigator System (Promega) to quantify apoptotic activity.

Mitochondrial membrane potential assessment

The mitochondrial membrane potential was measured by flow cytometry as described previously¹²⁰. The cells were incubated with 200 nM MitoTracker Red CMXRos (Thermo Fisher Scientific) and 150 nM MitoTracker Green FM (Thermo Fisher Scientific) at 37 °C for 25 min and then washed with PBS supplemented with 5% FBS. The cells were dissociated from the dish with 0.05% trypsin, neutralized, collected by centrifugation at 300 × g for 3 min, resuspended in PBS supplemented with 5% FBS, and then subjected to flow cytometry (BD FACSAria II, BD Biosciences).

Western blotting

Anti-β-actin (MEDICAL & BIOLOGICAL LABORATORIES [MBL], M177-3, 1:1000), anti-GAPDH (Cell Signaling Technology [CST], 2118, 1:1000), anti-FAK (CST, 3285, 1:1000; Thermo Fisher Scientific, AHO0502, 1:1000), anti-p-FAK^Y397 (CST, 3283, 1:200), anti-TOMM20 (CST, 42406, 1:1000), anti-RAC1 (CST, 8631, 1:200), anti-GST (CST, 2625, 1:1000), anti-PAK1 (CST, 2602, 1:1000), anti-p-PAK1^S144/p-PAK2^S141 (CST, 2606, 1:200), anti-BAD (CST, 9292, 1:200), anti-lipoic acid (Abcam, ab58724, 1:1000), anti-DLAT (Abcam, ab172617, 1:1000), anti-MRPS22/mS22 (Thermo Fisher Scientific, PA5-52249, 1:1000), anti-Mal-K (PTM Bio, PTM-201, 1:1000), MCAT (Santa Cruz Biotechnology, sc-390858, 1:1000), OXSM (Thermo Fisher Scientific, PA5-32132, 1:1000), and anti-MECR (PROTEINTECH, 51027-2-AP, 1:1000) were used as primary antibodies. IRDye800CW anti-rabbit IgG (LI-COR Biosciences, 926-32211, 1:10,000), IRDye680RD anti-rabbit IgG (LI-COR Biosciences, 925-68071, 1:10,000), and IRDye680RD anti-mouse IgG (LI-COR Biosciences, 925-68070, 1:10,000) were used as secondary antibodies. Images were obtained with an Odyssey CLx (LI-COR Biosciences) and quantified with Image Studio (LI-COR Biosciences, version 5.2).

Isolation of RAC1 in the GTP-bound form

RAC1 in the GTP-bound form was isolated with an Active Rac1 Detection Kit (CST, 8815) according to the manufacturer’s instructions. Anti-RAC1 Western blotting was performed as described above.

Mitochondrial purification

Mitochondria were isolated with a Mitochondria Isolation Kit, Human (Miltenyi Biotec), according to the manufacturer’s instructions with minor modifications. The cells were washed with ice-cold PBS and lysed with hypotonic buffer (10 mM HEPES-KOH, pH 7.5, 10 mM KCl, 1.5 mM MgCl₂, and 1 mM DTT). The cell lysate was incubated with TOMM22 beads for 60 min at 4 °C. After washing, the mitochondrial fraction was eluted from the beads, pelleted by centrifugation at 7000 × g for 10 min at 4 °C, resuspended in modified lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 15 mM MgCl₂, 1 mM DTT, and 1% Triton X-100), and then centrifuged at 20,000 × g for 10 min at 4 °C. The supernatant was subjected to Western blotting.

Electron microscopy

The cells were fixed with 2.5% glutaraldehyde (Electron Microscopy Sciences) in DMEM for 1 h at room temperature. After washing with 0.1 M phosphate buffer (0.02 M sodium dihydrogenphosphate dihydrate and 0.08 M disodium hydrogenphosphate), the cells were scraped and collected with 0.2% BSA/0.1 M phosphate buffer, followed by centrifugation at 820 × g. After being embedded in low-melting agarose (2% in 0.1 M phosphate buffer, MP Biomedicals), the cell pellets were sectioned at 200-µm thickness with a Leica VT1000S vibratome. The sections were postfixed with 1% OsO₄ (Electron Microscopy Sciences) and 1.5% potassium ferrocyanide (FUJIFILM Wako Pure Chemical Corporation) in 0.05 M phosphate buffer for 30 min. After being rinsed 3 times with H₂O, the cells were stained with 1% thiocarbohydrazide (Sigma‒Aldrich) for 5 min. After being rinsed with H₂O 3 times, the cells were stained with 1% OsO₄ in H₂O for 30 min. After being rinsed with H₂O 2 times at room temperature and 3 times with H₂O at 50 °C, the cells were treated with Walton’s lead aspartate (0.635% lead nitrate [Sigma‒Aldrich] and 0.4% aspartic acid [pH 5.2, Sigma‒Aldrich]) at 50 °C for 20 min. After incubating with an ascending ethanol series (10 min each in 50% on ice, 70% on ice, and 10 min each in 90%, 95% ethanol/H₂O at room temperature), the sections were rinsed for 10 min with 99.5% ethanol 4 times. Then, the sections were infiltrated with a 1:1 mixture of EPON812 (TAAB) and ethanol for 22 h. After incubating with 100% EPON812 resin for 4 h, the resin was cured at 65 °C for 6 d. EPON812 resin was made by mixing 7.5 g of MNA (TAAB), 13.7 g of Epok812 (TAAB), 3.8 g of DDSA (TAAB), and 0.2 g of DMP-30 (TAAB). Resin blocks were trimmed with a TrimTool diamond knife (Trim 45, DiATOME). Fifty-micron-thick ultrathin sections made with a diamond knife (Ultra 45, DiATOME) were collected on a cleaned silicon wafer strip using a Leica Ultramicrotome (UC7). The ultrathin sections were imaged with a scanning electron microscope (JSM-IT800SHL, JEOL). Imaging was done at 1 kV accelerating voltage, 5 kV specimen voltage, 1280 × 960 frame size, 6-mm working distance, 6.40 × 4.80-μm field of view, and 14.1-μs dwell time, using a Scintillator Backscattered Electron Detector in Beam Deceleration mode. The final pixel size was 5 nm square.

Reconstituted mammalian mitochondrial translation system

In vitro translation of mammalian mitochondria was conducted as reported previously^83,84. Unless otherwise specified, the standard translation mixtures (5 µl) contained 50 mM HEPES-KOH pH 7.5, 100 mM potassium glutamate, 11 mM Mg(OAc)₂, 0.1 mM spermine, 1 mM DTT, 0.15 mM each amino acid (except for methionine and cysteine), 0.05 mM methionine, 0.1 mM cysteine, 1 mM ATP, 1 mM GTP, 20 mM creatine phosphate, 0.01 µg/µL 10-formyl-5,6,7,8-tetrahydrofolic acid, 100 nM creatine kinase, 20 nM myokinase, 15 nM nucleoside-diphosphate kinase, 15 nM pyrophosphatase, 0.5 µM IF-2mt, 1.0 µM IF-3mt, 5 µM EF-Tumt, 1 µM EF-Tsmt, 0.5 µM EF-G1mt, 0.5 µM EF-G2mt, 0.5 µM RF-1Lmt, 0.5 µM RRFmt, 5 μM methionyl-tRNA transformylase (E. coli MTF), 0.2 µM 55S ribosome, 0.0225 A₂₆₀ units of yeast aminoacyl-tRNA mixture, and 0.1 µM mRNA. CoA, acetyl-CoA, and malonyl-CoA were dissolved in 20 mM HEPES-KOH pH 7.5, and 100 mM KOAc pH 7.5, and added to the reactions. The reporter mRNA 3×FLAG-Pgk-nLuc^83,84 was used. The reaction mixture was incubated at 37 °C for 3 h.

The enzymatic activity of the products was assessed with a Nano-Glo Luciferase Assay System (Promega). The 2-µl reaction was stopped by adding 18 µl of stop solution (20 mM HEPES-KOH pH 7.5, 100 mM KOAc pH 7.5, 2 mM Mg(OAc)₂, and 0.1 mg/ml RNase A) and then mixed with 20 µl of the substrate in a white 96-well half-area plate. After incubating for 17 min at room temperature, the luminescence was detected with a GloMax Multi Detection System (Promega).

Statistics and reproducibility

Statistical analyses were conducted with R software version 4.1.1 operated within the RStudio interface version 2021.09.0 + 351. In the analysis of deep sequencing data, fold-change significance was evaluated by likelihood ratio testing based on a generalized linear model, as implemented in DESeq2¹¹². We used the Mann‒Whitney U test (two-tailed), Student’s t test (two-tailed), the Tukey‒Kramer test (two-tailed), and the modified Fisher’s exact test (EASE score) as described in the figure legends.

In deep sequencing experiments, reproducibility was assessed using two biological replicates. The number of replicates is consistent with common practice in the field and was sufficient for dispersion estimation in the statistical analysis performed with DESeq2. Cell viability assay, cell adhesion assay, on-gel mito-FUNCAT, mito-FUNCAT FACS, ³⁵S Met-mediated metabolic labeling of newly synthesized proteins, mitochondrial DNA content measurement, assessment of mitochondrial ATP production, glucose uptake assessment, oxygen consumption rate (OCR), metabolomic analysis, apoptosis measurement, mitochondrial membrane potential assessment, Western blotting, and reconstituted mammalian mitochondrial translation system were conducted with triplicates or more as described in the figure legends. No statistical method was used to predetermine sample size. No data were excluded from the analyses. The experiments were not randomized. The Investigators were not blinded to allocation during experiments and outcome assessment.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conceptualization, T.W., Y.K., T. Yamamori, T. Yamazaki, A. Higashibata, and S.I.; Methodology, T.W., Y.K., M.M., T. Tsubaki, M.L., K.N., A.H.K., H.S., T. Yamamori, T. Yamazaki, A. Higashibata, T. Tsuboi, Y.H., N.T.-T., T. Saito, A. Higashitani, Y.S., and S.I.; Formal analysis, T.W., Y.K., M.M., T. Tsubaki, M.L., K.N., A.H.K., H.S., and T. Tsuboi; Investigation, T.W., Y.K., M.M., T. Tsubaki, M.L., K.N., and H.S.; Resources, A. Higashibata and A. Higashitani; Writing – Original Draft, T.W., Y.K., and S.I.; Writing – Review & Editing, T.W., Y.K., M.M., T. Tsubaki, M.L., K.N., A.H.K., H.S., T. Yamamori, T. Yamazaki, A. Higashibata, T. Tsuboi, Y.H., N.T.-T., T. Saito, A. Higashitani, Y.S., and S.I.; Visualization, T.W. and S.I.; Supervision, T. Yamamori, T. Yamazaki, A. Higashibata, T. Tsuboi, Y.H., N.T.-T., T. Saito, A. Higashitani, Y.S., and S.I.; Funding Acquisition, T.W., Y.K., K.N., T. Tsuboi, Y.H., Y.S., and S.I.

Peer review

Peer review information

Nature Communications thanks Sylvain Costes, Johannes Herrmann, and the other anonymous reviewer for their contribution to the peer review of this work. A peer review file is available.

Funding

S.I. was supported by the Japan Society for the Promotion of Science (JSPS) (JP19K22406, JP23H02415, and JP23H00095), the Ministry of Education, Culture, Sports, Science and Technology (MEXT) (JP20H05784 and JP24H02307), the Japan Agency for Medical Research and Development (AMED) (JP20gm1410001), the Japan Science and Technology Agency (JST) (JPMJCR25T2), the Gushinkai Foundation, and RIKEN (Pioneering Project and RIKEN TRIP initiative “TRIP-AGIS”). T.W. was supported by The Graduate School of Frontier Sciences, The University of Tokyo (C2205 and C2306) and by JSPS (JP23KJ0444). Y.K. was supported by JSPS (JP20J10665). K.N. was supported by The Graduate School of Frontier Sciences, The University of Tokyo (C2205) and JSPS (JP22J23099 and JP22KJ1154). Y.S. was supported by JSPS (JP21K15023), MEXT (JP21H05734 and JP23H04268), AMED (JP23gm6910005), the Exploratory Research Center on Life and Living Systems (ExCELLS, 23EX601), and RIKEN (Pioneering Project). T. Tsuboi acknowledges support from the Key Research and Development Program of the Ministry of Science and Technology (2024YFE0102700 and 2023YFA0914303), the Jilin Fuyuan Guan Food Group Co., Ltd, the Science, Technology, Innovation Commission of Shenzhen Municipality (WDZC20220811144737001), startup funds (OD2021031C) from Tsinghua SIGS. Y.H. was supported by AMED (JP19dm0207082 and JP21wm0525015) and MEXT (JP22H05532). We also thank the following fellowships: the RIKEN Junior Research Associate Program for Y.K. and H.S.; the JSPS Research Fellow (DC1) for K.N.; the JSPS Research Fellow (DC2) for Y.K., T.W., and T. Tsubaki; JST SPRING (JPMJSP2108) for T.W. and T. Tsubaki; and the ANRI fellowship for T.W.

Data availability

The standard Ribo-Seq, RNA-Seq, and MitoIP-Thor-Ribo-Seq data (GSE222998) [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE222998] obtained in this study were deposited in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database. This study also used the reported data for the retapamulin-assisted mitoribosome run-off assay (GSE237154) [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE237154]⁸⁵. The details of the experiments on the ISS and on the ground (OSD-425 [https://osdr.nasa.gov/bio/repo/data/studies/OSD-425], OSD-936, OSD-955, and OSD-956)^25,29,30,92 were also found in the NASA Open Science Data Repository (OSDR). Source data are provided with this paper. The metabolomic data obtained in this study were deposited in the Metabolomics Workbench (PR003099 [10.21228/M83K2F]). Source data are provided with this paper.

Code availability

For the data analysis for Ribo-Seq, RNA-Seq, and MitoIP-Thor-Ribo-Seq data, we used deposited codes in Zenodo (https://zenodo.org/records/5675739 and https://zenodo.org/records/11348425)^121,122, which used a reported custom script (https://github.com/ingolia-lab/RiboSeq).

Competing interests

S.I. is a member of the Scientific Reports editorial board, an associate editor of The Journal of Biochemistry, and a paid consultant of Eisai. The remaining authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-74493-z.