SETD3 is a mechanosensitive enzyme that methylates actin on His73 to regulate mitochondrial dynamics and function

Mitochondria are functionally diverse, dynamic organelles that not only generate ATP but also serve as the primary sources of metabolites required for cell growth and signaling (Martínez-Reyes and Chandel, 2020). Mitochondrial dynamics is a term collectively used for mitochondrial fission, fusion, maintenance of shape, motility, transportation and degradation (Mishra and Chan, 2016). The dynamic nature of mitochondria is crucial for maintaining mitochondrial health and distribution during cell division, creating new mitochondria, the transmission of mitochondrial DNA to daughter mitochondria, and facilitating protein, lipid and metabolite exchange (Lewis et al., 2016; Kraus et al., 2021). Mitochondrial dynamics is also key to the interaction between mitochondria and other organelles, such as the endoplasmic reticulum, lysosomes and Golgi-derived vesicles (Lee et al., 2016; Ji et al., 2015; Wong et al., 2018). Mitochondria are also required at sites of high transcriptional demand to maintain the homeostasis of specialized cells, such as neurons (Rangaraju et al., 2019). A delay or disruption in the spatial production and availability of ATP in cells with high energy requirements, such as neurons and cardiomyocytes, can lead to impaired cellular function and disease (Verstreken et al., 2005). Mitochondrial dynamics can respond to multiple stimuli, including reactive oxygen species (ROS), nutrient changes, growth factors, mitochondrial damage and mechanical stimuli (Debattisti et al., 2017; Pekkurnaz et al., 2014; Romani et al., 2022; Fung et al., 2019; Chakrabarti et al., 2022).

Mitochondrial dynamics are regulated by the actin cytoskeleton and microtubules (MTs). Spontaneous actin nucleation is not thermodynamically favorable. There are three classes of proteins that facilitate the bypassing of spontaneous actin nucleation in cells, namely the actin-related protein 2/3 (ARP2/3) complex, SPIRE proteins and formins (Goley and Welch, 2006; Burtnick et al., 1997; Tomchick et al., 2005; Baum and Kunda, 2005). Consistent with this, the association of the actin cytoskeleton with mitochondria is dependent on these nucleators or actin-binding proteins (ABP), and various mechanisms are thought to dominate in different conditions (Basu et al., 2021; Fung et al., 2022, 2019b; Moore et al., 2021; Moore and Holzbaur, 2018).

Post-translational modifications on actin or regulators of actin polymerization, including phosphorylation, acetylation and methylation, represent another mechanism through which cells fine-tune actin polymerization, and is referred to as the ‘actin code’ (Varland et al., 2019; Mu et al., 2020; Wilkinson et al., 2019a). Recent studies have highlighted the importance of actin post-translational modifications in regulating mitochondrial dynamics (Basu et al., 2021; Fung et al., 2022). Although advancements have been made in our understanding of actin code-mediated regulation of actin polymerization around mitochondria, a significant knowledge gap remains.

In this study, we describe the mitochondrial enrichment and function of SET domain protein 3 (SETD3), a recently identified histidine methyltransferase. SETD3 belongs to a family of enzymes that utilize the cofactor S-adenosyl-L-methionine to methylate substrates. SETD3 methylates histidine 73 (His73) on actin (Wilkinson et al., 2019), which aids in the polymerization of actin filaments. In ischemia-reperfusion injury (IR), there is a reduction in SETD3, which leads to defective mitochondrial function and an increase in reactive oxygen species (ROS) (Xu et al., 2021). Our findings here reveal that SETD3 localizes on the outer mitochondrial membrane (OMM) and promotes actin polymerization around mitochondria. Loss of SETD3 leads to diminished mitochondrial actin polymerization and reductions in mitochondrial length, network size and velocity. We also observed that SETD3 is required for the maintenance of oxidative phosphorylation and complex I function. Finally, we reveal SETD3 as a mechano-responsive enzyme, which regulates mitochondrial shape in response to changes in extracellular matrix (ECM) stiffness.

To investigate the cellular function of SETD3, we searched the DepMap public Dep database to determine which genes are co-essential or have similar functions to SETD3 (Wang et al., 2017; Meyers et al., 2017). The DepMap database uses CRISPR loss-of-function screening of more than 1000 cell lines of different lineages and a computational model to predict gene sets that function together (Meyers et al., 2017; Wang et al., 2017). We conducted KEGG pathway and Gene Ontology (GO) analyses for the top 500 genes that were positively and negatively correlated with SETD3 (Fig. 1A). Unexpectedly, KEGG pathway and GO analysis of the top 500 genes that had a negative correlation with SETD3 showed a correlation with the categories ‘oxidative phosphorylation with intermediates’ and ‘mitochondrial translation and metabolism’ (Fig. 1B,C; Fig. S1A–F).

To gain a deeper understanding of SETD3 loss of function and mitochondrial function, we generated SETD3 knockout (SETD3-KO) C2C12 cells (a mouse cell line) using CRISPR-Cas9 gene editing. We chose C2C12 cells as they can be differentiated from myoblasts into myotubes by serum withdrawal. To investigate mitochondrial-specific SETD3 function, we performed stable isotope labeling using amino acids in cell culture mass spectrometry (SILAC-MS) with the use of heavy methionine and mass spectrometry of the crude mitochondrial fraction (Fig. 1D). We looked at the ion score of the peptides with histidine methylation, and, apart from the multiple novel targets in our data, we found actin methylation at His73 was lost in the mitochondrial fraction of SETD3-KO cells (Fig. 1E; Fig. S1G), consistent with previously published data showing that SETD3 mediates His73 methylation on actin (Wilkinson et al., 2019; Kwiatkowski et al., 2018) and suggesting that SETD3 methylates actin associated with mitochondria. To validate our SILAC-MS findings, we performed western blot analysis using a pan-histidine methylation antibody. We observed a decrease in histidine methylation in SETD3-KO total cell lysates (Fig. 1F). We investigated our proteomics data to analyze the enrichment and depletion of protein from the crude mitochondrial fraction in the wild-type (WT) and SETD3-KO cells (Fig. 1G). The semiquantitative proteomics data showed a significant depletion of the mitochondrial proteins in SETD3-KO cells as compared to what was seen in WT cells (Fig. 1G). GO analysis of the depleted proteins revealed that these proteins were involved in mitochondrial translation, electron transport chain and oxidative phosphorylation (Fig. 1H, Fig. S1C–E). We examined our proteomics dataset to identify actin-related proteins. We observed a decrease in proteins related to Arp2/3-mediated actin nucleation, polymerization, and regulation of actin-based movement in the SETD3-KO crude mitochondrial fraction (Fig. 1I), which is consistent with a role for SETD3 in actin polymerization. However, we did not observe any difference in the levels of actin between the WT and SETD3-KO crude mitochondrial fraction (Fig. S1F). We also performed western blotting to examine levels of SETD3 targets and mitochondria (as assessed with a pan-mito antibody cocktail) (Fig. S1H,I). Consistent with previously published data, we found no difference in total β-actin levels between WT and SETD3-KO cells, but we noted an increase in levels of cytochrome c oxidase subunit I (MTCO1, complex IV protein), and succinate dehydrogenase (complex II protein) (Fig. S1H,I). These observations of the loss of methylated actin around the mitochondria and changes in the mitochondrial proteome in SETD3-KO cells suggest that SETD3 might be involved in regulating actin dynamics around mitochondria and consequent mitochondrial function.

The absence of mitochondrial actin methylation in SETD3-KO cells strongly indicates that SETD3 plays a crucial role in mitochondrial actin polymerization. To address this question, we used structured illumination microscopy (SIM) and phalloidin conjugated to Alexa Fluor 488 to visualize cytoplasmic actin in WT and SETD3-KO cells. Our data show a decrease in the cytoplasmic actin meshwork in SETD3-KO cells compared to that in WT C2C12 cells, which includes mitochondrial actin in addition to other pools of cytoplasmic F-actin (Fig. 2A). In addition, ultracentrifugation to segment actin revealed a greater relative proportion of F-actin sediments in WT cells compared to that seen in SETD3-KO cells (Fig. 2B). We observed a further decrease in F-actin sedimentation in SETD3-KO cells treated with cytochalasin D, an inhibitor of actin polymerization (Fig. 2B). These results indicate that SETD3-KO decreases actin polymerization without causing actin catastrophe and increases cell sensitivity to cytochalasin D treatment.

Next, we investigated whether SETD3 is required for actin polymerization around mitochondria. As most F-actin and mitochondrial contacts are lost during fixation (Schiavon et al., 2020), we performed live imaging for actin and mitochondria. To examine cytoplasmic actin, we stably expressed Lifeact–GFP and transiently transfected with the mitochondrial marker Tom20–mCherry in HeLa cells. To avoid any clone-specific increase in Lifeact–GFP signal, we used the cells derived from a single clone. SETD3 siRNA-treated cells displayed a significant decrease in cytoplasmic actin filaments and mitochondria colocalization (Fig. 2C–E; Fig. S2A). We then assessed mitochondria-specific actin using a nanobody-based mitochondrial actin probe (Schiavon et al., 2020). This probe utilizes an actin nanobody sequence attached to a mitochondrial localization signal and a HALO tag for visualization (Mito-Actin–HALO). As actin polymerization increases around the mitochondria, the concentration of the probe increases in that area, leading to enhanced fluorescence (Schiavon et al., 2020). To eliminate expression-specific bias due to differential expression of the probe in certain cells, we created stable C2C12 and HeLa cell lines expressing Mito-Actin–HALO and performed CRISPR-mediated KO of SETD3 in the C2C12s and siRNA knockdown of SETD3 in HeLa cells. A significant decrease in Mito-Actin–HALO colocalization with Fis1–GFP (a mitochondrial marker) in SETD3 KD HeLa and SETD3-KO C2C12 cells revealed decreased polymerization around mitochondria (Fig. 2F–I). Similarly, we observed a significant decrease in Mito-Actin–HALO colocalization with the control Fis1–GFP in HeLa cells treated with SETD3 siRNA compared to controls. Together, these data indicate that loss of SETD3 leads to a reduced actin polymerization around mitochondria (Fig. 2J).

Most organisms have distinct pools of motile and stationary mitochondria that rapidly change their shape and network in response to external stimuli that depend on the interaction of mitochondria with actin filaments and microtubules (Moore and Holzbaur, 2018). SETD3 mediates His73 methylation of mitochondrial actin, and SETD3-KO reduces mitochondrial actin. To determine whether mitochondrial shape and distribution are dependent on His73 methylation of actin, we performed live cell imaging of SETD3-KO cells using MitoTracker green (MTG) which forms a covalent bond with multiple cristae and membrane proteins (Presley et al., 2003). The mitochondrial footprint, an indicator of mitochondrial mass, did not differ between WT and SETD3-KO cells (Fig. 3A). We examined mitochondrial branching and elongation in live cells via ImageJ-based Mitochondrial Network Analysis (Valente et al., 2017). SETD3-KO cells displayed reduced mitochondrial branch length and network branching (Fig. 3B), indicating that SETD3 is important in maintaining mitochondrial dynamics and networks. We used Cytochalasin B, an actin polymerization inhibitor, to treat cells as a positive control. The treatment with Cytochalasin B produced the same mitochondrial shape phenotype that we observed in SETD3-KO cells. This indicates that the primary reason for the mitochondrial shape phenotype in SETD3-KO cells could be the actin polymerization defect. Next, we used mitochondrial photoactivatable GFP (mito-PAGFP) (Karbowski et al., 2004) to quantify mitochondrial interconnectivity. Compared with controls, SETD3-KO cells exhibited decreased mitochondrial interconnectivity and interchange of mitochondrial content (Fig. 3C,D). We investigated whether DRP1 (also known as DNM1L) mediates SETD3-mediated maintenance of mitochondrial shape by staining WT and SETD3-KO cells with anti-TOM20 and anti-DRP1 antibodies and performing a colocalization analysis. There was no change in DRP1 mitochondrial localization between WT and SETD3-KO cells (Fig. S2B–D), indicating that SETD3 might regulate mitochondrial shape independently of DRP1.

Using MTG in time-lapse imaging, we investigated whether SETD3 is important for mitochondrial motility. We observed a significant decrease in fast-traveling mitochondria and an increase in slow-traveling mitochondria in SETD3-KO cells (Fig. 3E). Together, these data indicate that SETD3 is required for maintaining mitochondrial shape and movement.

We cultured SETD3-KO cells in a medium lacking glutamine and pyruvate and containing glucose or galactose as a nutrient source. SETD3-KO cells grown in galactose had reduced viability (Fig. 4A). Cells deficient in oxidative phosphorylation (OXPHOS) or treated with inhibitors of OXPHOS survive in glucose-rich medium but are inviable in media supplemented with galactose, as OXPHOS-deficient cells primarily rely on cytosolic ATP production via glycolysis-derived glucose (Arroyo et al., 2016). To detect dying cells in glucose versus galactose medium, we stained cells with annexin V, an established marker of necrotic and apoptotic cells (Arroyo et al., 2016). Our results reveal that ∼90% of SETD3-KO cells were positive for annexin V in the galactose medium (Fig. 4B), indicating that SETD3-KO cells have deficient OXPHOS and ATP production, and are auxotrophic for glucose.

We performed a Seahorse flux analysis to investigate the OXPHOS status of SETD3-KO cells. Compared with WT cells, SETD3-KO cells exhibited a severe decrease in basal and maximal oxygen consumption rate (OCR) and ATP production (Fig. 4C,D). We did not observe any changes in mitochondrial content as measured through mitochondrial DNA mass (Fig. S2B). Consistent with this, the mitochondrial footprint, as measured by assessing MitoTracker Green intensity, did not change in SETD3-KO cells compared to WT cells (Fig. 3A,B). We observed a significant increase in basal glycolytic rate, total glycolysis, and glycolytic reserves extracellular acidification rate (ECAR) in SETD3-KO cells, indicating that glycolysis is increased (Fig. 4E,F). Next, we checked whether treating cells with DRP1 inhibitor can rescue the OCR in SETD3-KO cells. Consistent with our TOM20 and DRP1 colocalization data, there was no rescue of OCR after DRP1 inhibitor treatment (Fig. S2E). To investigate the effect of SETD3 on mitochondrial polarization, we treated cells with TMRE, a mitochondrial potential-dependent dye, and MitoTracker Green and performed a colocalization analysis. We did not observe any difference in TMRE uptake between WT and SETD3-KO cells, showing that there is no difference in mitochondrial potential (Fig. S2F). Re-expression of WT SETD3 in SETD3-KO cells partially rescued the reduced OCR (Fig. 4G). The partial recovery of OCR can be attributed to the use of SETD3-KO cells that express SETD3-WT stably at various levels across the cell population. In contrast, expression of catalytic-deficient SETD3 was unable to rescue the OCR (Fig. 4G). These data indicate that the enzymatic activity of SETD3 is required for optimal OXPHOS activity.

Defective mitochondrial respiration can arise from defective mitochondrial translation, structure or any defects in electron chain complex formation (Vafai and Mootha, 2012). OXPHOS is conducted in respiratory chain complexes I–V, which are macromolecular complexes that catalyze the electron transfer from reducing equivalents. Complex I, the largest macromolecular complex in mitochondria, comprises 45 subunits that exist as a super-complex with complexes II, III, IV and V. Proper assembly of complex I and super complex formation are required for ATP production (Lapuente-Brun et al., 2013; Carroll et al., 2006). Our proteomics data revealed that there is a reduction in the levels of complex I proteins, including NDUFS4, NDUFA12 and NDUFA10 (Fig. 4H), which are required for proper assembly of the complex (Stroud et al., 2016). To investigate the role of SETD3 in complex I assembly, we performed Blue Native PAGE western blot analysis of mitochondria isolated from WT and SETD3-KO C2C12 cells. We probed with antibodies against different parts of the N-module of complex I, which is the final part of complex I to be assembled. We probed for NDUFA13, which is present in the matrix-facing part of complex I and is partially embedded in the inner mitochondrial membrane, and for NDUFV1, which is in the soluble part of the N module (Stroud et al., 2016). Complex I in SETD3-KO cell mitochondria migrated faster than complex I from WT mitochondria. Strikingly, we observed SETD3-KO mitochondria had an increase in unassembled N-modules of complex I (Fig. 4I), even though there was no change in total protein levels of NDUFV1 and NDUFA13 (Fig. S1H,I), indicating a defect in complex I assembly.

We performed a Seahorse assay on permeabilized cells to further assess complex I function by introducing the substrate for complex I into the assay (Salabei et al., 2014). We observed that permeabilized cells with the WT SETD3 gene showed an increase in OCR upon the addition of the complex I substrates pyruvate and malate. However, this response was diminished after the addition of rotenone, which is a complex I inhibitor. By contrast, cells with the SETD3 gene knocked out failed to respond to these substrates (Fig. 4J). Complex I-mediated ROS are important for muscle differentiation (Lee et al., 2011). Consistent with this, SETD3-KO cells expressed myosin 4 and myogenin, which are differentiation markers, but failed to form multinucleated myotubes, marked by the absence of multinucleated myosin 4-positive and myogenin-positive cells in the SETD3-KO cells, whereas these were present in WT cells (Fig. 4K,L; Fig. S2G). Together, these data indicate that deficits in SETD3 cause mitochondrial dysfunction because impaired complex I assembly and function.

To test for the subcellular localization of SETD3, we performed immunofluorescence studies using SETD3-specific antibodies in undifferentiated C2C12 myoblasts. SETD3 staining showed a mitochondrial staining pattern and co-localizes with MTCOI, suggesting it is a mitochondrial protein (Fig. 5A). This was surprising as SETD3 is not included in the extensive mitochondrial compendium of MitoCarta 3.0 and MitoCop (Rath et al., 2020; Morgenstern et al., 2021). To further confirm our observation, we then performed subcellular fractionation of differentiated C2C12 cells. In accordance with our immunofluorescence data, SETD3 was present in crude mitochondrial fractions (Fig. 5B). To define the membrane association of SETD3, we sonicated and ultracentrifuged isolated mitochondria. SETD3 was slightly enriched in the pellet, suggesting it is an integral membrane protein or, more likely, that it is interacting with an integral mitochondrial membrane protein (Fig. 5C). SETD3 sequence analysis revealed an absence of a transmembrane domain (data not shown). Hence, it is likely that SETD3 mitochondrial localization is mediated by interaction with other mitochondrial membrane proteins. To further investigate this, we performed digitonin extraction on the crude mitochondria from C2C12 cells. SETD3 was released in the supernatant at lower digitonin concentrations, indicating that SETD3 is an OMM protein (Fig. 5D). To determine whether SETD3 was localized on the inner or outer mitochondrial membrane, we performed a protease protection assay on isolated mitochondria. Treatment of mitochondria with pronase E partially degraded SETD3, and we detected a lower molecular mass band, further supporting the notion that SETD3 is localized on the OMM (Fig. 5E).

We then performed an in vitro mitochondrial transport reaction for SETD3. We used rabbit reticulocyte lysates to produce translated SETD3 labeled with [S35]methionine, which was then incubated with freshly isolated C2C12 mitochondria. With increasing time, SETD3 bound mitochondria, and after pronase E addition, we observed a faster migrating and partially degraded band (Fig. 5F). These data provide strong evidence that SETD3 associates with the OMM. In complementary live imaging studies, we overexpressed mouse SETD3 with a Halo Tag (SETD3–Halo) in HeLa cells stably expressing Lifeact–GFP and transfected with the mitochondrial marker Fis1–mCherry. Our imaging revealed that SETD3 localizes around mitochondria (Fig. 5G). Together, these data indicate that SETD3 is localized on the OMM.

Mitochondrial actin is regulated by ECM stiffness, and changes in ECM stiffness alters mitochondrial morphology and function (Urra et al., 2021; Romani et al., 2022). We hypothesized that SETD3 mediates ECM-induced regulation of mitochondrial dynamics via F-actin mediated mechanism(s). To test this, we, we plated C2C12 cells in soft and hard matrix stiffness plates and performed western blotting. We observed that SETD3 levels are downregulated in cells plated in the softer matrix and gradually increase with increased stiffness (Fig. 6A), indicating that SETD3 levels are regulated by matrix stiffness. We then analyzed the mitochondrial morphology of cells plated under different matrix stiffness via live imaging. Consistent with the finding that mitochondria of SETD3-KO cells had a reduced mitochondrial branch length and mitochondrial network (Fig. 3C,D), mitochondria of the cells plated in softer ECM have a significant reduction in mean length and a significant decrease in mitochondrial network size (Fig. 6B,C). We also determined whether SETD3 knockdown leads to a further decrease in mitochondrial morphology parameters. Surprisingly, we observed no further reduction in mitochondrial mean length and network size (Fig. 6B,C). To determine whether changes in mitochondrial morphology in softer matrix stiffness are specifically mediated by SETD3, we plated cells that stably overexpress SETD3–EGFP in soft ECM plates. SETD3 overexpression rescued the mitochondrial length and network size (Fig. 6D,E). Altogether, these data indicate that SETD3 is a mechanosensitive (mechanoresponsive) enzyme that regulates mitochondrial morphology in response to matrix stiffness changes.

Here, we have shown that SETD3 is a mechanoresponsive enzyme that mediates methylates histidine 73 of actin, which is required for F-actin formation around mitochondria to regulate mitochondrial shape, velocity and function. We identified novel SETD3-mediated histidine methylation targets that are involved in various cell processes, including metabolism, F-actin regulation and RNA processing. Given our data showing that SETD3 was enriched in the mitochondrial fractions, and that mitochondrial shape and function are regulated by F-actin formation around mitochondria, we prioritized investigating the mitochondrial function(s) of SETD3 in later experiments. Our SILAC-MS analysis revealed that actin in the enriched mitochondrial fraction lacks the His73 methylation actin polymerizing modification in SETD3-KO cells. We discovered that SETD3-KO cells have decreased levels of cytoplasmic actin fibers. Mitochondria-associated F-actin is a portion of the cytoplasmic F-actin; using a mitochondrial-specific actin reporter, we observed a significant decrease in actin polymerization around mitochondria in SETD3-deficient cells. Our findings show a clear decrease in proteins associated with Arp2/3-mediated actin nucleation, polymerization and regulation of actin-based movement in the SETD3-KO crude mitochondrial fraction. This indicates that SETD3-mediated actin polymerization is dependent on the Arp2/3. Likely, the actin-polymerizing modification of His73 by SETD3 is a fast acting and finetuning event. It would be interesting to look at the role of SETD3 in actin polymerization dynamics in SETD3-KO cells. As there are multiple subpopulations of actin, namely, actin cables, clouds and comets, in the future it would be interesting to explore whether SETD3 regulates actin polymerization in these subpopulations.

As movement depends on the actin cytoskeleton, we evaluated mitochondrial shape and movement and observed that SETD3-KO cells are unable to form an elongated and interconnected mitochondrial network. We also observed that loss of SETD3 leads to a decrease in mitochondrial motility, as evidenced by a decrease in fast-traveling mitochondria and an increase in slow-moving mitochondria in SETD3-KO cells. The dynamic nature of mitochondria is responsible for cell fate decisions, regulating local protein translation, signal transduction in neurons, removal of damaged mitochondria and Ca²⁺ buffering, which are critical to stress response in the cells (Eisner et al., 2018; Rangaraju et al., 2019). It will be interesting to study the role(s) of SETD3 in the aforementioned processes.

Our findings also indicate that SETD3 is required for OXPHOS; SETD3 knockout leads to a decrease in OXPHOS-dependent ATP production and a compensatory increase in glycolysis-dependent cytosolic ATP production. Our Seahorse data of latrunculin-a-treated cells (Fig. S2H) reveal a similar trend of increased compensatory glycolysis, consistent with previous findings that changes in mitochondrial actin and shape lead to alterations in mitochondrial function. SETD3 deficiency led to a significant reduction in complex I assembly factors. As a result, the assembly and function of complex I in the electron transport chain are impaired in SETD3-KO cells. Compared to wild-type cells, SETD3-KO cells have more free-floating portions of the N-module. The enzymatic activity of complex I is dependent on the proper assembly of the N-module (Guerrero-Castillo et al., 2017). This indicates that SETD3 is required for complex I assembly and function.

OXPHOS-dependent energy production relies on proper folding and mitochondria inner membrane-cristae organization (Giacomello et al., 2020). Membrane tension generated by the mitochondrial contact site and cristae organizing system (MICOS) complex has been shown to be important in maintaining the membrane curvature and cristae shape (Rabl et al., 2009; Tarasenko et al., 2017). Cells lacking the MICOS complex have OXPHOS defects (Giacomello et al., 2020). Tension generated by the actin cytoskeleton is required for the sculpting of the inner mitochondrial membrane, cristae density and mitochondrial function (Shi et al., 2022). SETD3 regulates OXPHOS via aiding the mitochondrial complex I assembly. Given these observations, we hypothesize that SETD3 regulates mitochondrial function by promoting actin polymerization and maintaining the outer mitochondrial membrane tension, which in turn promotes cristae formation and complex I assembly by maintaining the cristae organization.

Finally, we have shown that SETD3 levels are upregulated in response to increased ECM stiffness, with cells in softer matrix plates having reduced levels of SETD3. Mitochondria of these cells a have reduced mitochondrial actin and a reduction in length and network size. Overexpression of SETD3 is sufficient to rescue the mitochondrial morphology of the cells plated in softer ECM, indicating that rescuing the SETD3 levels alone in cells plated in a softer matrix can increase the mitochondrial size and mitochondrial network size. Consistently, mitochondria of SETD3-KO cells phenocopy the fragmented and smaller mitochondria seen in the cells in soft ECM. The energetics of SETD3-KO cells, which have a reduced OCR, also align with previous findings showing that cells on softer ECM have reduced OCR (Romani et al., 2022). Loss of SETD3 results in the loss of cytoplasmic F-actin, which is essential for mechanosensing when the extracellular matrix is stiff. This observation is consistent with the previously published report that SETD3-knockout mice exhibit reduced smooth muscle contractility and dystocia, which necessitates induced contractions (Wilkinson et al., 2019). Together, these findings suggest that SETD3 is a mechanoresponsive enzyme that regulates mitochondrial dynamics and function by aiding in F-actin polymerization around mitochondria in response to mechanotransduction (Fig. 7).

SETD3 has also been linked to exercise-linked muscle hypertrophy after high-intensity training. To meet the higher energy demands related to high-intensity exercise, mitochondrial biogenesis and electron transport chain efficiency must be increased in skeletal muscle cells (Sousa-Victor et al., 2022a). Furthermore, high-intensity exercise leads to muscle fiber damage and activation and the asymmetric division of satellite cells. These satellite cells differentiate into myocytes to rebuild muscle fiber (Sousa-Victor et al., 2022). SETD3 levels increase during the differentiation of C2C12 cells from myoblast to myotubes. Consistent with the observation that SETD3 levels increase during differentiation, SETD3-KO C2C12 myoblast cells fail to differentiate into myotubes. The hallmark of differentiation of progenitor cells into specialized cell types, like neurons and cardiomyocytes, is the elongation of mitochondria and the switch of metabolism from glycolysis to oxidative phosphorylation (Dorn et al., 2015; Devine and Kittler, 2018). Longer mitochondria are more efficient with ATP production owing to the increased surface area of inner mitochondrial membrane cristae (Giacomello et al., 2020). This marked change is required to fulfill the energy demands of the differentiated cell types (Dorn et al., 2015; Devine and Kittler, 2018). SETD3-KO cells display smaller and toroidal mitochondria with smaller diameters than those in wild-type cells, and are deficient in oxidative phosphorylation. This suggests that SETD3 might contribute to the increase in energy demand during and post-differentiation, especially in cells like skeletal muscle, cardiomyocytes, and neurons.

Tables S1 and S2 list all plasmids and antibodies used in this study, respectively.

C2C12 cells (American Type Culture Collection; ATCC) were cultured in high-glucose Dulbecco's modified Eagle's medium (DMEM) (Hyclone) supplemented with 1 mM sodium pyruvate, 2 mM L-glutamine and 20% fetal bovine serum (FBS) (SH30243.FS, ATCC; complete medium). Cells were subcultured every 48 h and maintained below 70% confluency. For differentiation, plates where cells were 100% confluent were cultured in DMEM supplemented with 1 mM sodium pyruvate, 2 mM L-glutamine and 2% horse serum for 5 days. HeLa cells (Baylor College of Medicine's Tissue culture core, originally acquired from the ATCC) were grown in high-glucose DMEM supplemented with 1 mM sodium pyruvate, 2 mM L-glutamine and 10% FBS. C2C12 cells were transfected using the Cell Line 4D-Nucleofector™ X kit and an Amaxa™ 4D-Nucleofector™ (Lonza) or Lipofectamine 3000 according to the manufacturer's protocol. HeLa cells were transfected using Lipofectamine 2000. For RNAi knockdown, ON-TARGETplus siRNA sets targeting human SETD3 were purchased from Horizon Discovery, and 30 pmol was transfected per 10⁶ HeLa cells using Lipofectamine RNAiMAX. For live-cell imaging, cells were plated in a 35 mm glass bottom dish 48–72 h before imaging, depending on the treatment. For soft matrix imaging experiments, polyacrylamide-based gel-coated imaging dishes (Young's Elastic E≈0.2 kPa) were purchased from Matrigen.

siRNAs targeting the following sequences were used: Setd3_sirna_1, 5′-GGTAAGAAGAGTCGAGTAA-3′; Setd3_sirna_2 5′-GCTAATGACTGTTGAATCT-3′; Setd3_sirna_3, 5′-GCACTGGCCTTTCATTTGC-3′; and Setd3_sirna_4, 5′-ACACTCCTCTCTACTTTGA-3′.

Lentiviral constructs were transfected into 90% confluent HEK293T cells (ATCC) in a 10-cm dish along with packaging plasmids (Table S1) using Lipofectamine 2000 (Thermo Fisher Scientific, 11668019). The medium was changed to fresh DMEM (Hyclone, SH30243) supplemented with 10% fetal bovine serum (FBS) (ATCC) at 16 h post-transfection. The medium was collected 48 to 72 h after transfection, centrifuged at 1000 g for 5 min, and the supernatant was passed through a 45-μm filter (Sigma SLHVR33RS). The filtered solution was then stored at −80°C for future use. We thawed 1 ml of lentiviral supernatant and added 3 ml of 10% FBS or 20% FBS DMEM high-glucose medium with 10 µg/ml polybrene. Then, we added the supernatant to a 60-mm dish containing a subconfluent culture of HeLa or C2C12 cells. The dish was incubated at 37°C for 24 h, and the medium was replaced with 10% or 20% FBS DMEM medium depending on the cell type. Cells were subjected to fluorescence-activated cell sorting (FACS) to isolate single clones. Clones expressing low levels were selected for the experiments.

Mouse SETD3 was deleted in C2C12 cells by using two guide RNAs against exon 6 cloned individually in pX458 (Addgene #48138). C2C12 cells (10⁶) were electroporated with 1 μg of each guide RNA plasmid using a Cell Line 4D-Nucleofector™ X kit and an Amaxa™ 4D-Nucleofector™ (Lonza). Cells were immediately plated in 20% FBS high-glucose DMEM in a 10-cm dish. At 16 h post-transfection, the medium was changed to fresh medium. At 24 h post-transfection, the cells were lifted by using 2 ml trypsin (HyClone, SH3004201), and GFP-positive cells underwent single-cell FACS analysis in 96-well plates containing 200 μl complete medium. We initially screened the cells for positive clones by using PCR-based genotyping using (primer Setd3_R, 5′-GAGTTGGGGCTGGCTCGCTC -3′ and Setd3_F: 5′-AGCGTGTTTTGAGTTGTGGCC-3′).

The mitochondrial fraction obtained as described below was dissolved in 50 mM ammonium bicarbonate solution and digested using 1 μg of trypsin overnight at 37°C followed by another 4 h digestion. The peptides were measured using the Pierce™ Quantitative Colorimetric Peptide Assay (Thermo Fisher Scientific, 23275). The tryptic peptides were subjected to a simple C18 cleanup using a C18 disk plug (3 M Empore C18) and dried in a speed vac. LC-MS/MS analysis was conducted using a nano-LC 1200 system (Thermo Fisher Scientific) coupled to Orbitrap Lumos mass spectrometer (Thermo Fisher Scientific). 1 μg peptide was loaded on a pre-column of 2 cm × 100 μm internal diameter (i.d.) switched in-line with an in-housed 5 cm × 150 μm i.d. column (Reprosil-Pur Basic C18, 1.9 μm, Dr. Maisch, Germany) and equilibrated in 0.1% formic acid in water. The peptides were eluted using a 75 min gradient of 5–28% acetonitrile and 0.1% formic acid at a flow rate of 750 nl/min. The mass spectrometer was operated in the data dependent acquisition mode with a 3 s cycle time. MS1 was acquired in Orbitrap (120,000 resolution, 300-1400 m/z) followed by MS2 in IonTrap (HCD 30%, AGC 5E3, 50 ms ion injection) with 18 s dynamic exclusion time. Obtained MS/MS spectra were searched against target-decoy Mus musculus NCBI RefSeq protein database in the Proteome Discoverer (PD1.4, Thermo Fisher Scientific) with the Mascot algorithm (Mascot 2.4, Matrix Science). The following dynamic modifications were used: mono (14.02 Da), di (28.03 Da), tri (42.05 Da) methyl on histidine; heavy mono (18.04 Da), di (36.08 Da), tri (54.11 Da) methyl on histidine; oxidation (15.99 Da) on methionine; and protein N-terminal acetylation. For label-free quantification of the proteome using iBAQ values, the MS data was analyzed using the MaxQuant algorithm (Sinitcyn et al., 2018). Using the following dynamic modifications were used: mono (14.02 Da), di (28.03 Da), tri (42.05 Da) methyl on histidine; heavy mono (18.04 Da), di (36.08 Da), tri (54.11 Da) methyl on histidine; oxidation (15.99 Da) on methionine; and protein N-terminal acetylation. The precursor mass tolerance was confined within 20 ppm with a fragment mass tolerance of 0.5 Da, and a maximum of two missed cleavages were allowed. The iBAQ values were further analyzed on the Perseus software platform to calculate proteins significantly changing between the groups using a t-test with permutation-based FDR of 0.05% (Tyanova et al., 2016). Data was imported into R and GO, and KEGG analysis was performed using the Clusterprofiler package (Yu et al., 2012).

Actin segmentation was performed by ultracentrifugation as previously described (Qiao et al., 2017). WT and SETD3-KO C2C12 cells were lysed in prewarmed actin stabilization buffer [50 mM PIPES pH 6.9, 50 mM NaCl, 5 mM MgCl₂, 5 mM EGTA, 2 mM ATP, 5% glycerol, 0.1% Nonidet P-40, 0.1% Triton X-100, 0.1% Tween 20, 0.1% β-mercaptoethanol, protease inhibitor cocktail and phosphatase inhibitor (Sigma-Aldrich)] directly in the dish at 37°C for 10 min. Cells were collected and centrifuged at 300 g for 5 min to remove insoluble particles. An aliquot of the cell lysate was saved as total input protein. An equal amount of cell lysate was transferred for ultracentrifugation at 100,000 g at 37°C for 1 h in a TLA 100.3 (Beckman Coulter) rotor. The supernatant fraction was collected as the G-actin, and the pellet was collected as F-actin. The F-actin fraction was resuspended in 1 μM cytochalasin D (Sigma-Aldrich) and kept on ice for 45 min to dissolve F-actin. Samples were boiled in Laemmli buffer for 10 min and were used in western blot analyses.

We plated 100,000 low passage WT and SETD3-KO cells in 6-well plates in triplicate for each time point and treatment analysis. The cells were cultured in DMEM, 20% dialyzed FBS and 2 mM glutamine supplemented with 4.5 g/l of glucose or galactose. After 4, 24 or 48 h of culture in glucose or galactose medium, cell death was assessed by staining with annexin V and Alexa fluor 647 conjugate (Invitrogen) following the manufacturer's protocol, and DAPI. Cells were analyzed by flow cytometry using an ARIA FACS.

HeLa cells were washed twice in ice-cold 1× phosphate-buffered saline (PBS) and collected in lysis buffer [20 mM HEPES, 150 mM NaCl, 1 mM MgCl₂, 0.1% NP-40, 1% SDS and 5% sucrose supplemented with protease cOmplete™, EDTA-free Protease Inhibitor Cocktail (Roche) and phosphatase inhibitor PhosSTOP™ (Roche)]. Cells were lysed via freeze thaw cycles, by freezing them at −80°C and thawing on ice. The lysate was sonicated for 3 cycles of 10 s on and 10 s off with a probe sonicator on ice and then centrifuged at 21,000 g for 15 min. The supernatant was collected, and protein concentration was determined by using the BCA Protein Assay Kit (Sigma-Aldrich). 30–50 μg of protein per sample were run on a 10% SDS-PAGE gel and then transferred the proteins to a PVDF membrane (Immobilon-P, Millipore IPVH00010). Membranes were stained with Ponceau S (Sigma P7170).  Membranes were blocked with 1% BSA in PBST (PBS with 0.1% Tween 20) and incubated with specified primary antibodies (1:1000 dilution) at 4°C overnight. The next day, membranes were washed with PBST three times for 15 min each. Then, membranes were incubated with secondary antibodies at room temperature for 2 h and washed with PBST three times for 15 min each. Signal was detected using either the Immobilon Western chemiluminescent substrate (Millipore P90720) or SuperSignal West Femto Chemiluminescent substrate (Pierce PI34095) using X-ray film or Amersham™ Imager 680.

We plated 7500 WT or SETD3-KO C2C12 cells in a Seahorse XF96 cell culture microplate (Agilent). After 24 h, the cells were washed with XF base medium (Agilent) supplemented with 1 mM sodium pyruvate, 2 mM L-glutamine, and 10 mM glucose. The cells were incubated at 37°C in a non-CO₂ incubator. OCR was measured by using a Seahorse XFe96; the cells were perturbated sequentially with 1 µM oligomycin, 500 nM FCCP, and 500 µM rotenone and antimycin A using a Seahorse XF Cell Mito Stress Test Kit. For ECAR measurement, the cells were washed twice with Seahorse XF Base medium with 5 mM HEPES (Agilent) supplemented with 1 mM sodium pyruvate and 2 mM L-glutamine. The cells were incubated at 37°C in a non-CO₂ incubator. ECAR was measured by using the XFe96 analyzer for cells that were treated sequentially with glucose, oligomycin and 2-deoxy-glucose. After each experiment, we removed the medium and washed the plates with PBS, followed by lysing with PBS supplemented with 1% SDS. Protein concentration was determined using the BCA Protein Assay Kit (Sigma-Aldrich). The data were normalized to the total protein concentration and analyzed by using Wave software (Agilent).

We seeded 1.5×10⁶ WT and SETD3-KO cells onto a 150-mm culture dish. At ∼100% confluency, the cells were washed with ice-cold PBS and scrapped in mitochondria isolation buffer (MB) containing 225 mM mannitol, 75 mM sucrose, 30 mM Tris-HCl pH 7.4 and 0.5 mM EGTA. A dounce tissue grinder with a tight-fitting pestle (DWK Life Sciences Kimble Kontes) (Suski et al., 2014) was used to homogenize the cells. The lysate was centrifuged at 800 g for 5 min to remove unbroken cells, and the nuclear fraction pellet was labelled as P1. The supernatant was centrifuged at 800 g for 5 min, and the pellet was discarded. Then, the supernatant was centrifuged at 10,000 g for 20 min; the pellet was washed once with MB and centrifuged at 10,000 g for 10 min. Finally, the pellet was collected as the mitochondrial fraction. Then, the supernatant was centrifuged at 20,000 g for 20 min, and the supernatant from that was collected as the cytoplasmic fraction. The pellet was washed once with MB and centrifuged at 20,000 g for 20 min; this pellet was collected as the plasma membrane-associated membrane fraction. P1 was resuspended in hypotonic buffer (20 mM HEPES pH 7.4, 20 mM NaCl, 1% NP-40 supplemented with protease and phosphatase inhibitor), passed through a 26-gauge needle twice and centrifuged at 800 g for 5 min; this pellet was collected as the nuclear fraction.

We resuspended 500 μg of intact mitochondria in 250 μl of mitochondria isolation buffer and treated them with 150 μg/ml pronase E (Sigma-Aldrich) for 30 min on ice with and without 0.1% Triton X-100. The reaction was stopped by adding 2 mM PMSF for 10 min. The protein was precipitated by using trichloroacetic acid (TCA) and 0.1% sodium deoxycholate overnight at 4°C and centrifuged at 21,000 g for 30 min. The pellets were washed with ice-cold acetone, dried, resuspended in 1:1 2× Laemmli buffer with 0.1 M NaOH, boiled for 10 min, and analyzed using SDS-PAGE followed by immunoblotting.

Radiolabeled SETD3 was generated using in vitro transcription followed by in vitro translation using rabbit reticulocyte lysates (Promega) in the presence of [35S]methionine (NEG009T001MC, Perkin-Elmer). Product was centrifuged at 100,000 g for 30 min at 4°C to pellet the ribosomes. Translated product was incubated with freshly isolated mitochondria from differentiated C2C12 cells (see above) in 20 mM HEPES-KOH pH 7.4, 250 mM sucrose, 80 mM potassium acetate, 5 mM magnesium acetate, 10 mM sodium succinate and 5 mM methionine at 37°C for the indicated times. 10 volumes of reaction buffer was added to the sample and centrifuged at 5000 g for 5 min to remove the unincorporated SETD3. Samples were subsequently subjected to the pronase-E on ice for 30 min and then 1 mM PMSF for 10 mins. Protein was precipitated using TCA and 0.1% sodium deoxycholate overnight at 4°C and analyzed using western blotting.

Airyscan videos were acquired at 10.13 seconds per frame for 30 steps. Airyscan-processed images were imported in Fiji software, and the background was subtracted using a rolling ball radius of 20 pixels. Mitochondria trajectories were generated using TrackMate 1 in Fiji (Tinevez et al., 2017). A Laplacian of Gaussian detector was used to identify mitochondria with an estimated blob diameter of 2 μm, and a threshold value of 2, using subpixel localization. Spurious localizations were manually corrected using the add or delete spot tool.

C2C12 and HeLa cells were seeded in 35-mm imaging dishes were fixed in cytoskeleton buffer (10 mM MES, pH 6.1, 150 mM NaCl, 5 mM EGTA, 5 mM glucose, 5 mM MgCl₂) with 4% PFA for 10 min. The cells were permeabilized by using 0.5% Triton X-100 in PBS for 10 min, followed by blocking in 10% donkey serum in PBS with 0.1% Tween20 (PBST) for 1 h at room temperature. Cells were incubated in primary antibody (1:200) in 10% donkey serum in PBST overnight at 4°C. Then, they were washed with PBST, incubated with secondary antibodies (1:200) in 10% donkey serum PBST for 1 h at room temperature, washed again with PBST and stored in PBS. For phalloidin staining, the cells were incubated with 1:40 phalloidin 488 (Alexa Fluor™ 488 Phalloidin, A12379) for 1 h at room temperature and imaged immediately. We used a Zeiss LSM880 confocal microscope in Airyscan mode with a 63× 1.4 NA Plan-Apochromat oil to image the cells in PBS. Airyscan images were processed in Zen blue (Carl Zeiss) using default values; these images were further processed and analyzed in Fiji software (NIH).

Airyscan microscopy was performed on a Zeiss LSM 880 with an Airyscan module equipped with 63×1.4 NA oil immersion objective lens. The samples were excited with 405 nm, 488 nm, 561 nm and 642 nm, lasers and fluorescence images were collected using appropriate filters. Image size and z-stacks were taken according to the Nyquist optimized values. Live imaging was conducted at 37°C and 5% CO₂ in FluoroBrite DMEM supplemented with 10% FBS and sodium pyruvate and glutamine. The medium was prewarmed and saturated with CO₂ in 37°C and placed in a 5% CO₂ incubator for 1 h. All the imaging data in the manuscript is from at least two independent experiments.

Structured illumination microscopy was performed on a Zeiss Elyra equipped with a 63×1.4 NA oil immersion objective lens. Excitation wavelengths were 405 nm, 488 nm, 561 nm and 642 nm with exposure time between 50 and 200 ms depending on the signal strength. Fluorescence images were collected using appropriate filters. Z-stacks with a size of 2560×2560 pixels with a slice-to-slice distance of 0.1 μm were acquired. SIM processing and channel alignment were performed in Zen black (Zeiss), with 3D rendering and standard settings. Reconstructed images were further processed in Fiji software (NIH).

Actin segmentation was performed as previously described (Moore et al., 2021), with modifications. Airyscan processed images were imported in Fiji, and actin cables were segmented using the Ridge detection plugin (https://zenodo.org/record/845874#.YfHi799MGx8) with sigma values between 1 and 5 depending on the signal-to-noise ratio, with a minimum line length of 6 pixels; a binary map was generated. The map was filtered using a 2-pixel gaussian filter to stimulate the cable width. Colocalization analysis was performed between the gaussian filter-detected actin cable segment and mitochondria using JACoP to calculate Manders' and Pearson's coefficients (Bolte and Cordelières, 2006) in a custom Fiji macro (available upon request).

Airyscan-processed images were imported in Fiji software, and the threshold was decided manually depending on the signal-to-noise ratio. The threshold was kept constant for mitochondrial actin and mitochondria, and a colocalization analysis was performed using JACoP to calculate Manders' and Pearson's coefficients (Bolte and Cordelières, 2006) in a custom Fiji macro (available upon request).

We thank Rebecca Bartow, PhD, of the Department of Scientific Publications at the Texas Heart Institute for editorial support. We are grateful to Uri Manor and Cara R. Schiavon (Salk Institute for Biological Studies) for providing Mito-Actin probes and Jason Kirk at the BCM OIVM core for his insights and suggestion in microscopy. We thank the Mass Spectrometry Proteomics Core at the Baylor College of Medicine for processing the proteomics samples.