Mitochondrial control of amino acid catabolism by a fasting-inducible mitochondrial carrier

Abstract

Metabolic adaptation to nutrient deprivation requires coordinated control of mitochondrial anaplerosis and cataplerosis; however, how metabolite flux across the mitochondrial membrane is regulated during fasting remains less defined. Here, we report SLC25A34 as a fasting-inducible mitochondrial carrier that is highly expressed in oxidative skeletal muscle. Using bacterial reconstitution, proteo-liposomes, and tracer studies, we showed that SLC25A34 mediates the import of phosphoenolpyruvate (PEP) into the mitochondrial matrix. Loss of SLC25A34 impaired glutamine-supported anaplerosis under nutrient-deprived conditions, while glucose and pyruvate utilization remained largely intact. Muscle-specific deletion of Slc25a34 resulted in reduced fasting-induced amino acid catabolism and the accumulation of amino acids, leading to activation of mTORC1 signaling even under fasted conditions. Consequently, SLC25A34-deficient soleus muscle exhibited hypertrophy and myopathic features, accompanied by mTORC1-dependent increase in protein synthesis. Together, these results highlight a unique biological role for the inducible mitochondrial carrier SLC25A34, which couples PEP import to amino acid catabolism and proteostasis to preserve skeletal muscle integrity in response to metabolic stress.

Mitochondrial metabolite transport is tailored to meet muscle-specific needs for utilizing amino acids during fasting.

INTRODUCTION

Mitochondria serve far broader roles than ATP production through integrating nutrient availability with fuel selection to support adaptation to physiological and pathological stress (1–3). Under conditions of limited glucose availability, such as prolonged fasting, peripheral tissues, including skeletal muscle and brown adipose tissue (BAT), shift away from carbohydrate use and increase the oxidation of fatty acids released by adipose tissue lipolysis (4–6). Skeletal muscle also undergoes proteolysis, supplying amino acids for hepatic gluconeogenesis, while sustaining its own bioenergetic needs to preserve glucose for the brain, an obligate glucose-dependent organ (7, 8). Similarly, endurance exercise requires flexible substrate utilization in muscle to balance carbohydrate sparing with sustained energy production (4, 9, 10). In turn, under pathological conditions, such as heart failure, the failing heart shifts away from fatty acid oxidation and relies on glucose metabolism (11).

In this context, a key regulator is the balance between anaplerosis and cataplerosis of the tricarboxylic acid (TCA) cycle. Anaplerosis replenishes TCA cycle intermediates to maintain oxidative capacity, whereas cataplerosis enables the exit of carbon from the cycle to support biosynthesis, signaling, and redox homeostasis, as well as to prevent excessive accumulation of intermediates in the matrix (12–15). For example, acetyl-CoA derived from pyruvate or fatty acid oxidation condenses with oxaloacetate (OAA) to form citrate; glutamine contributes anaplerotic carbon via conversion to α-ketoglutarate (αKG); and branched-chain amino acids feed the cycle through acetyl-CoA (Leu, Ile) or succinyl-CoA (Val). On the other hand, cataplerotic flux, such as the export of citrate, supports lipid synthesis and acetylation pathways (16, 17). Additionally, the export of malate and OAA maintains redox balance and serves as substrate shuttles (13–15). Despite the importance of this balance, the regulatory mechanisms that coordinate anaplerosis and cataplerosis during metabolic adaptation remain less explored.

A knowledge barrier has been the limited understanding of the molecular determinants controlling metabolite flux across the inner mitochondrial membrane (IMM). Since the IMM is impermeable to most metabolites, anaplerosis and cataplerosis depend on dedicated metabolite carriers localized to the IMM (18). For instance, the mitochondrial pyruvate carrier (MPC) is essential for importing pyruvate into the mitochondrial matrix (19, 20), while SLC25A1 (CIC) and SLC25A10 (DIC) facilitate the transport of citrate and dicarboxylic acids (e.g., succinate), respectively (21–24). SLC25A11 (OGC) and SLC25A12/13 (AGC1/2) mediate the malate-aspartate shuttle that supports NADH shuttling across the IMM (25–28). Of note, many of these IMM carrier proteins are ubiquitously and constitutively expressed; however, some carriers are tissue-selective and uniquely regulated by external cues. An example is uncoupling protein 1 (UCP1, also known as SLC25A7 for H⁺ import), which is highly induced by cold exposure and the beta-adrenergic signaling pathway in brown and beige adipocytes (29, 30). Thus, it is likely that stress-inducible IMM carriers expressed in a tissue-selective manner serve as an additional regulatory layer of metabolic adaptation by modulating anaplerotic and cataplerotic fluxes depending on the cellular context.

The present study reports SLC25A34 as an example of such: SLC25A34 shares the highest sequence homology with SLC25A35 among the SLC25A family, which we recently identified as a mitochondrial carrier for phosphoenolpyruvate (PEP) (31). PEP is a unique metabolite from a view of bioenergetics, as PEP contains one of the highest energy phosphate bonds (−61.9 kJ mol ⁻¹), higher than ATP (−30.5 kJ mol ⁻¹), and has diverse biological roles, including glycolysis, gluconeogenesis, glyceroneogenesis, and insulin secretion (32–35). We found that SLC25A34 is highly expressed in skeletal muscle, particularly in the mitochondria-rich soleus muscle at a higher level than SLC25A35 (fig. S1, A and B). Notably, unlike SLC25A35, SLC25A34 expression is induced by metabolic stressors, including fasting and exercise. Hence, the present work investigated the biological role of SLC25A34 in regulating the balance of mitochondrial anaplerosis and cataplerosis in muscle in response to metabolic stress.

RESULTS

SLC25A34 transports phosphoenolpyruvate (PEP)

SLC25A34 belongs to the SLC25A protein family, many of which are expressed in the mitochondrial membrane (24, 36), and shares a predicted structural similarity with SLC25A35, exhibiting 63.7% amino acid sequence similarity. We first validated the cellular localization of the SLC25A34 protein by immunostaining, confirming its presence in the IMM (fig. S1C). The predicted structure of SLC25A34, as generated by AlphaFold (37), suggests a highly similar overall architecture to SLC25A35, which consists of six transmembrane α-helices, a characteristic of SLC25 carriers (Fig. 1A). Our recent study in SLC25A35 identified critical amino acids responsible for substrate recognition, including Y72, Q73, M76, R80, Y124, K127, R175, R276, and H280 (31). Notably, these residues are well-conserved between SLC25A34 and SLC25A35 (Fig. 1, B and C), suggesting that they share common substrates.

Fig. 1. SLC25A34 transports phosphoenolpyruvate (PEP).

(A) Side view of the predicted SLC25A34 structure (green) aligned with the predicted SLC25A35 structure (purple), created using UCSF ChimeraX. (B) Top view of the aligned SLC25A34 (green) and SLC25A35 (purple) structures. Key amino acid residues involved in phosphoenolpyruvate (PEP) binding within the central channel-like cavity were highlighted. (C) Close-up of the key amino acid residues of SLC25A34 (green) and SLC25A35 (purple) in the central channel-like cavity. (D) Schematic of PEP uptake assay with bacteria. Created in BioRender. Oikawa, S. (2026) https://BioRender.com/hom68x4. (E) Immunoblot of MBP-SLC25A34 protein in E. coli. Note that E. coli expresses the endogenous MBP. (F) PEP uptake in control and MBP-SLC25A34-expressing E. coli. n = 5 per group. (G) Time course of PEP uptake in control and MBP-SLC25A34-expressing E. coli. n = 5 per group. (H) Competition assay. An excess amount of unlabeled PEP (50 mM) was added to the bacterial uptake assay. n = 5. (I) PEP uptake in control or MBP-SLC25A45-expressing E. coli. n = 5 per group. (J) CBB staining and immunoblot of affinity-purified SLC25A34 protein. Arrowhead indicates the MBP-SLC25A34-TwinStrep protein. (K) PEP transport assay in proteo-liposomes reconstituted with purified SLC25A34 or control-eluates, with or without excess unlabeled PEP (50 mM). n = 4 per group. (L) Competition assay. An excess amount of indicated metabolites (50 mM) was added to the assay. n = 4. (M) PEP transport assay in proteo-liposomes reconstituted with wild-type or mutant (Y125A/R176A/R277A) SLC25A34, or control-eluates. n = 4 per group. Bars represent mean ± s.e.m. P values were calculated by unpaired t-test [(F) and (I)] and two-way ANOVA with Tukey’s multiple comparisons test [(G, H), (K), (L), and (M)].

To test the hypothesis that SLC25A34 also transports PEP, we employed a bacterial reconstitution system in which human SLC25A34 protein was fused to maltose-binding protein (MBP) at the N terminus in the E. coli strain C43(DE3). This system has reduced Lon and OmpT protease activity, making it suitable for overexpressing membrane proteins and ensuring their correct orientation in the bacterial inner membrane (38) (Fig. 1D). Although a previous study reported that the human SLC25A34 protein was not expressed in E. coli (39), the recombinant MBP-tagged SLC25A34 protein was successfully expressed in this system (Fig. 1E). Note that E. coli expresses the endogenous MBP. Subsequently, the control E. coli expressing the endogenous MBP alone, as well as E. coli expressing MBP-SLC25A34, were incubated with ¹³C-labeled PEP, washed, and subjected to liquid chromatography-mass spectrometry (LC-MS) to evaluate their PEP import activity. We found that ¹³C-PEP uptake was significantly higher in E. coli expressing MBP-SLC25A34 than in the control (Fig. 1F). Time-course PEP uptake assays found that ¹³C-PEP was rapidly imported into E. coli expressing MBP-SLC25A34 and reached a plateau state within 15 min, whereas no uptake was detected in the control group (Fig. 1G). To validate the specificity of SLC25A34-mediated PEP transport, we next evaluated ¹³C-PEP import in the presence of excess unlabeled PEP as a competitor (50 mM). We found that ¹³C-PEP uptake was completely blocked by the addition of excess unlabeled PEP (Fig. 1H). To verify whether the observed PEP uptake activity is specific to SLC25A34, we expressed MBP-SLC25A45 as an additional control. SLC25A45 is a mitochondrial carrier for trimethyl-lysine (TML) that we and others have recently characterized (40–42). Under the same experimental conditions as MBP-SLC25A34, the assay found no PEP uptake in MBP-SLC25A45 (Fig. 1I). To test its substrate specificity, we tested whether SLC25A34 transports oxaloacetate (OAA), using ¹³C₄-OAA as a tracer. Since conventional LC-MS assays were unable to detect ¹³C₄-OAA due to its instability, we quantified ¹³C₄-malate derivatized with 3-nitrophenylhydrazine (3-NPH), which served as a proxy for ¹³C₄-OAA (43). In contrast to PEP, however, we did not find active OAA transport in the system (fig. S1D).

As a complementary approach, we developed a cell-free proteo-liposome system with purified SLC25A34 protein. In brief, C43(DE3) bacteria expressing either an MBP-SLC25A34-TwinStrep construct (SLC25A34) or a control vector expressing the endogenous MBP (control) were subjected to affinity purification using StrepTactin resin (Fig. 1J). Subsequently, the purified eluents (SLC25A34 protein or the corresponding control fraction) were reconstituted into liposomes. Consistent with the bacterial system, we found that ¹³C-PEP transport was significantly higher in the SLC25A34-liposomes preloaded with PEP than in control-liposomes. Background signals obtained from protein-free empty control liposomes (i.e., liposome lipids only) were subtracted, as these signals represent non-specific association with the liposomes (fig. S1E). The PEP transport was significantly reduced in the presence of an excess amount (50 mM) of unlabeled PEP (Fig. 1K).

Given the high structural similarity between SLC25A34 and SLC25A35 (see Fig. 1, A to C), we asked whether SLC25A34 exhibits comparable transport activity and substrate specificity. To this end, we purified SLC25A34 and SLC25A35 proteins and reconstituted them into liposomes. The assay showed active ¹³C-PEP transport at comparable levels in liposomes containing either SLC25A35 or SLC25A34 (fig. S1F). To determine the substrate specificity of SLC25A34, we tested ¹³C-PEP transport in the presence of excess unlabeled PEP, malate, OAA, citrate, α-ketoisocaproic acid (KIC), and α-ketoisovaleric acid (KIV), according to our recent study of SLC25A35 (31). We found that PEP transport was substantially blunted by PEP, whereas OAA, citrate, KIC, and KIV did not compete with PEP transport via SLC25A34 (Fig. 1L and fig. S1G). Malate showed modest yet significant inhibition, albeit to a lesser extent than PEP. The results are consistent with the substrate specificity of SLC25A35 (31).

Next, we assessed the functional role of the conserved residues identified in our structural analysis of SLC25A34 by generating a recombinant mutant SLC25A34 protein carrying Y125A/R176A/R277A. We then performed PEP transport assays in liposomes containing the wild-type or mutant form of SLC25A34. Unexpectedly, the assay found higher PEP transport in the SLC25A34 mutant (Y125A/R176A/R277A) than in the wild-type form (Fig. 1M). It is worth noting that we obtained a similar result with SLC25A35, in which mutations at the corresponding substrate-binding residues increased PEP transport activity compared with the wild-type protein (31). These results are in alignment with a report on the ADP/ATP carrier showing that mutations weakening the salt-bridge gate network make the gate leaky, thereby increasing the apparent transport rate (44). Our proteo-liposome assay was performed under saturating substrate conditions, and the readout reflects V_max; hence, gate-loosening mutations could lead to higher apparent transport activity. Nonetheless, the mutation analysis supports the structural prediction that Y125, R176, and R277 are functionally important for SLC25A34-mediated PEP transport.

SLC25A34 expression is induced by fasting and exercise

We next asked how the SLC25A family members are regulated by metabolic stress, such as fasting and exercise, in skeletal muscle. First, data mining of publicly available data (45) found that Slc25a34 was the most upregulated Slc25a gene in response to fasting (Fig. 2A). In contrast, Slc25a35 mRNA levels were reduced by fasting. We validated this observation using independent samples: Slc25a34 mRNA levels in the soleus muscle were upregulated following 24 hours of fasting (Fig. 2B). A similar trend was also observed in the plantaris muscle, although the basal expression levels in the plantaris muscle were lower than those in the soleus muscle (fig. S2, A and B). To examine whether such regulation occurs in a cell-autonomous manner, we investigated Slc25a34 expression in cultured myocytes in which Slc25a34 mRNA levels were increased during myoblast differentiation (Fig. 2C). We found that Slc25a34 mRNA expression was significantly induced when cells were cultured in a starvation medium (Hanks’ Balanced Salt Solution (HBSS) supplemented with 5.5 mM glucose) for 6 hours or longer, compared to cells cultured in a medium containing 25 mM glucose, 2% horse serum, and amino acids, including 4 mM glutamine (Fig. 2D). Furthermore, data analysis of publicly available RNA-seq data (46) showed that Slc25a34 is the most upregulated Slc25a gene in mouse skeletal muscle following exercise (Fig. 2E) and is also upregulated in BAT upon cold exposure (fig. S2C). On the other hand, Slc25a35 expression did not show such an increase in response to exercise and cold exposure. Consistent with the results in mice, transcriptomics data analysis of human skeletal muscle (GSE120862) showed that SLC25A34 mRNA levels were significantly upregulated 4 hours after one-legged knee extension exercise in untrained males (Fig. 2F). Together, these results suggest that SLC25A34 is highly expressed in oxidative muscle and its expression is inducible by fasting, exercise, and cold exposure; in contrast, Slc25a35 expression was distinctly regulated despite both sharing PEP as a specific substrate (Fig. 2G).

Fig. 2. SLC25A34 expression is induced by fasting and exercise.

(A) Transcriptomics of SLC25A genes in the gastrocnemius from fed and 24 hours-fasted mice (GEO: GSE210904). n = 4 per group. (B) Relative mRNA levels of Slc25a34 in soleus from wild-type mice after 24 hours of fasting. n = 3 for fed and n = 4 for fasted. (C) Relative mRNA levels of Slc25a34 in primary myoblasts and myotubes. n = 3 per group. (D) Relative mRNA levels of Slc25a34 in C2C12 myotubes cultured in nutrition-deprived media (HBSS) for indicated times. n = 5 for 0, 6, 12 hours and n = 4 for 24 hours. (E) Transcriptomics of SLC25A genes in the gastrocnemius of mice after 4 weeks of voluntary running (GEO: GSE123879). n = 5 per group. (F) Relative mRNA levels of SLC25A34 in human vastus lateralis from exercised and non-exercised legs after one-legged knee extension (GEO: GSE120862). n = 7 per time point. (G) Summary of SLC25A34 and SLC25A35 expression and regulation. Dots and arrows indicate relative basal expression and regulatory changes. (H) Relative mRNA levels of Slc25a34 in C2C12 myotubes treated with PPAR agonists for 48 hours. n = 4 per group. (I) The recruitment of PGC-1α to the Slc25a34 gene locus was visualized from PGC-1α ChIP seq (GEO: GSE51178), and ATAC-seq (GEO: GSE134962). (J) TPM values of Slc25a34 in the quadriceps from wild-type and muscle-specific PGC-1α knockout mice after exercise training (GEO: GSE221210). n = 5 per group. Bars represent mean ± s.e.m. P values were calculated by unpaired t-test with Benjamini-Hochberg FDR correction [(A) and (E)], unpaired t-test [(B) and (C)], one-way ANOVA with Dunnett’s multiple comparisons test [(D) and (H)], two-way repeated-measures ANOVA with Tukey’s multiple comparisons test (F) and two-way ANOVA with Tukey’s multiple comparisons test (J).

Next, we examined the regulatory mechanisms of Slc25a34 expression. One of the key transcriptional regulatory axes mediating the adaptive responses to fasting, exercise, and cold exposure involves peroxisome proliferator-activated receptor-delta (PPARδ) and peroxisome proliferator-activated receptor-gamma coactivator 1-alpha (PGC-1α), as well as AMP-activated protein kinase (AMPK), among other upstream regulators (47–49). Notably, treatment with a synthetic agonist for PPARδ (GW501516 at 1 μM), but not agonists for PPARα (fenofibrate, 100 μM) and for PPARγ (rosiglitazone at 10 μM), significantly increased the levels of Slc25a34 in cultured C2C12 myotubes (Fig. 2H). GW501516 (a PPARδ agonist) treatment led to a dose- and time-dependent increase in Slc25a34 in C2C12 myotubes (fig. S2, D and E). Furthermore, data analysis of chromatin immunoprecipitation sequencing (ChIP-seq) of PGC-1α (50) showed active recruitment of PGC-1α onto the proximal promoter region as well as the intron of the Slc25a34 gene locus in C2C12 myotubes (Fig. 2I). Importantly, transcriptome data analysis of muscle-specific PGC-1α knockout (KO) mice (51) showed that Slc25a34 mRNA levels were significantly elevated in skeletal muscle of mice following exercise, whereas such induction by exercise was not seen in muscle-specific PGC-1α KO mice (Fig. 2J). These data indicate that SLC25A34 expression is regulated by the PPARδ/PGC-1α transcriptional cascade in skeletal muscle.

SLC25A34 is required for amino acid utilization in myocytes

We next examined the metabolic consequence of SLC25A34 loss in myocytes. To this end, we generated primary myoblasts that lacked SLC25A34 (Slc25a34 KO cells) and control cells (control) by the Easi-CRISPR (Efficient additions with ssDNA inserts-CRISPR) system (52) (Fig. 3A). There were no differences in myoblast growth and differentiation between control and Slc25a34 KO cells in culture (fig. S3, A to C). To isolate mitochondria from control and Slc25a34 KO myotubes rapidly, we expressed the Mito-Tag in these cells (53) and analyzed their whole-cell metabolites as well as mitochondrial metabolites by LC-MS (Fig. 3B and fig. S3D). Under standard cultured conditions with 25 mM glucose, 5% horse serum, and amino acids, there was no significant difference in mitochondrial metabolites between control and Slc25a34 KO myotubes (Fig. 3C and table S1). Similarly, at the whole-cell level, there was no major change in intracellular PEP contents between the two groups (fig. S3E and table S2). However, under a starved culture condition with HBSS supplemented with 5.5 mM glucose for 6 hours, we found dynamic differences in mitochondrial metabolites between control and Slc25a34 KO myotubes. This includes significant reductions in PEP and several TCA intermediates, such as aspartate, malate, fumarate, and αKG; on the other hand, several amino acids (lysine, arginine, glutamine, and threonine) and choline were significantly elevated in Slc25a34 KO mitochondria (Fig. 3D and table S3). Such changes were not observed at the whole-cell level (fig. S3F and table S4). Accordingly, mitochondrial enrichment of PEP in Slc25a34 KO myotubes was significantly lower than that in control myotubes under this condition (Fig. 3E). Similarly, mitochondrial enrichment of several metabolites, including malate, fumarate, aspartate, choline, lysine, and arginine, was altered in Slc25a34 KO cells (fig. S3G).

Fig. 3. SLC25A34 is required for amino acid utilization in myocytes.

(A) Relative mRNA levels of Slc25a34 in control and Slc25a34 KO primary myotubes. n = 3 per group. (B) Schematic of mitochondrial and whole-cell metabolomics experiments using MITO-Tag. (C) Mitochondrial metabolomics under a nutrient-replete condition. n = 5 per group. (D) Mitochondrial metabolomics under a nutrient-depleted condition. n = 5 per group. (E) Mitochondrial PEP enrichment under a nutrient-depleted condition was calculated as the ratio of mitochondrial to whole-cell PEP intensity. n = 5 per group. (F) Mitochondrial PEP uptake assay. n = 3 per group. (G) Mitochondrial uptake of the indicated metabolites. n = 4 per group for citrate, n = 3 per group for other metabolites. (H) Schematic of [U-¹³C]-glucose tracing under a fasted condition. Cells were cultured in a nutrient-depleted condition (green arrow, no change). Created in BioRender. Oikawa, S. (2026) https://BioRender.com/y38ws2q. (I) Labeling of the indicated ¹³C-labeled PEP (%) in (H). n = 5 per group. (J) Schematic of [U-¹³C]-glutamine tracing. Control and KO primary myotubes were cultured in a nutrient-depleted condition (red arrow, increased; blue arrow, decreased; green arrow, no change). Created in BioRender. Oikawa, S. (2026) https://BioRender.com/waw2lpb. (K) M + 5 glutamine and glutamate labeling (%) in (J). n = 5 per group. (L) M + 5 proline labeling (%) in (J). n = 5 per group. (M) Labeling of the indicated ¹³C-labeled PEP (%) in (J). n = 5 per group. (N) M + 5 α-ketoglutarate (αKG) and citrate labeling (%) in (J). n = 5 per group. Bars represent mean ± s.e.m. P values were calculated by unpaired t-test [(A), (E), (F), (G), (K), (L), and N)], unpaired t-test with Benjamini-Hochberg FDR correction [(C) and (D)] and two-way ANOVA with Tukey’s multiple comparisons test [(I) and (M)].

Since the liposome-based assay suggests that SLC25A34 transports PEP, we next asked whether mitochondrial PEP uptake was altered in Slc25a34 KO myotubes. To this end, mitochondria isolated from control and Slc25a34 KO myotubes were incubated with ¹³C-labeled PEP for 5 min, quickly rinsed, and their tracer contents were measured by LC-MS. In alignment with the above results, mitochondrial PEP uptake was significantly reduced in Slc25a34 KO myotubes compared to that in control cells (Fig. 3F). To determine the selectivity, we also incubated Slc25a34 KO and control mitochondria with ¹⁴C-labeled citrate, αKG, malate, aspartate, leucine, as well as ¹³C-labeled KIC, KIV, and α-keto-β-methylvaleric acid (KMV). However, we did not find any difference in their mitochondrial uptake between the two groups (Fig. 3G). These results suggest that SLC25A34 loss resulted in the selective impairment of mitochondrial PEP import, whereas changes in mitochondrial enrichment of malate, aspartate, αKG, and other metabolites observed in Slc25a34 KO myotubes were consequences of Slc25a34 deletion, rather than reduced mitochondrial import per se.

Considering the diverse biological roles of PEP, we next examined the extent to which SLC25A34 loss affects the utilization of glucose, pyruvate, and amino acids by performing tracer studies. First, [U-¹³C]-glucose tracing in differentiated myotubes found no significant difference in M + 3, M + 2, and M + 1 PEP contents between control and Slc25a34 KO cells (Fig. 3, H and I). Similarly, there was no difference in ¹³C-labeled contents of citrate, αKG, succinate, aspartate, and glutamate between the two groups (fig. S4A). When we used [U-¹³C]-pyruvate as a tracer, we did not find significant differences in M + 3, M + 2, and M + 1 PEP contents, as well as ¹³C-labeled TCA cycle intermediates, between control and Slc25a34 KO cells (fig. S4, B to D). These results suggest that SLC25A34 loss does not affect glycolysis or the cellular utilization of glucose and pyruvate in myotubes. This finding is consistent with the cellular respiration data showing that the oxygen consumption rate (OCR) of Slc25a34 KO myotubes was comparable to that of control cells in cultured media containing high glucose (fig. S4E).

On the other hand, we found that amino acid utilization was significantly altered in Slc25a34 KO cells, as summarized in Fig. 3J. When [U-¹³C]-glutamine was used as a tracer, M + 5 glutamate levels were significantly lower in Slc25a34 KO cells than in control cells, even though M + 5 glutamine contents were comparable between the two groups (Fig. 3K). Data for M + 1 to M + 4 glutamine and glutamate are shown in fig. S4, F and G. Instead, we found that Slc25a34 KO cells contained significantly higher levels of M + 5 proline than control cells (Fig. 3L). No change was seen in M + 1 to M + 4 proline levels between KO and control cells (fig. S4H). These results suggest that glutamine-driven anaplerosis was impaired, whereas the conversion to proline was enhanced in Slc25a34 KO cells. Notably, M + 2 PEP levels, i.e., PEP regeneration from OAA, but not M + 1 and M + 3 PEP levels, were higher in Slc25a34 KO cells than in control cells (Fig. 3M). Since there was no difference in M + 5 αKG and citrate levels between the two groups (Fig. 3N), the data suggest that glutamine-derived PEP (M + 2) is accumulated in Slc25a34 KO cells due to impaired PEP import into the mitochondria (pathways highlighted in red lines in Fig. 3J).

SLC25A34 loss results in muscle hypertrophy and myopathy

The above results suggest that SLC25A34 is essential for amino acid utilization in skeletal muscle under conditions of nutritional deprivation. Of note, SLC25A34 expression was downregulated in muscle diseases. For example, Duchenne muscular dystrophy (DMD), caused by loss-of-function mutations in the dystrophin gene, is characterized by chronic muscle degeneration and regeneration, leading to alterations in muscle amino acid metabolism (54, 55). In a mouse model of DMD (GSE162455), Slc25a34 mRNA levels in the soleus muscle were lower than in control mice (Fig. 4A). Furthermore, SLC25A34 mRNA levels in the vastus lateralis muscle of DMD patients were significantly lower compared to healthy subjects (56) (Fig. 4B). Friedreich’s ataxia (FRDA) is an inherited mitochondrial disease caused by frataxin deficiency, resulting in metabolic skeletal muscle impairments, including downregulation of mitochondrial enzymes involved in amino acid catabolism (57, 58). Transcriptomic analysis of Friedreich’s ataxia subjects (59) found lower levels of SLC25A34 mRNA in the gastrocnemius muscle compared to those in healthy control subjects (Fig. 4C). These results motivated us to investigate how the loss of SLC25A34 in skeletal muscle affects skeletal muscle function in vivo.

Fig. 4. SLC25A34 loss results in hypertrophy and myopathy in slow-twitch muscle.

(A) RPKM values for Slc25a34 in soleus from mdx mice (GEO: GSE162455). n = 4 per group. (B) Relative SLC25A34 mRNA levels in human vastus lateralis from healthy and Duchenne muscular dystrophy (DMD) subjects (GEO: GSE3307). n = 20 for healthy and n = 10 for DMD subjects. (C) SLC25A34 transcript levels in human gastrocnemius from subjects with Friedreich’s ataxia (FRDA) (GEO: GSE226646). n = 6 for control and n = 7 for FRDA subjects. (D) Relative mRNA levels of Slc25a34 in the indicated tissues in male control and KO mice. n = 3–4 for control and KO. (E) Muscle weight of the soleus, plantaris, and extensor digitorum longus (EDL) in male control and KO mice at 16 weeks old. Muscle weight (mg) was normalized to the body weight (g). n = 8 per group. (F) Representative laminin and DAPI-stained soleus sections from male mice. Right: Distribution of the cross-sectional area. Scale bar, 100 μm. n = 7 for control and n = 6 for KO. (G) Representative laminin and DAPI-stained soleus sections from male mice. Central nuclei are indicated by arrowheads. Right: Quantification of centrally nucleated myofibers per soleus cross-section. Scale bar, 100 μm. n = 8 for control and n = 7 for KO mice. (H) Representative electron microscopy images of subsarcolemmal mitochondria in soleus of male mice. Right: Quantification of subsarcolemmal mitochondrial area. Scale bar, 0.5 μm. n = 101 for control and 75 for KO collected from n = 2 mice per group. Bars represent mean ± s.e.m. P values were calculated by unpaired t-test [(A), (B), (D) to (F), and H)], Wald test with Benjamini-Hochberg FDR correction (C) and Mann-Whitney test (G).

To this end, we developed Slc25a34^flox/flox mice and subsequently crossed them with human α-skeletal actin (HSA)-Cre mice (HSA-Cre x Slc25a34^flox/flox, herein muscle-Slc25a34 KO mice). We validated that HSA-Cre effectively deleted Slc25a34 in the soleus and plantaris muscles of muscle-Slc25a34 KO mice, but not in the heart and liver (Fig. 4D). We found no difference in body weight between muscle-Slc25a34 KO mice and littermate control mice (Slc25a34^flox/flox) when mice were fed a regular diet at room temperature (fig. S5A). However, the soleus muscle of muscle-Slc25a34 KO mice exhibited modest but significantly greater tissue mass than that of control mice (Fig. 4E). This phenotype was consistently observed both in male and female mice (fig. S5B). Note that the tissue weight of the soleus muscle was greater in muscle-Slc25a34 KO mice than in control mice without normalization to body weight (fig. S5C). The difference was attributed to myofiber hypertrophy, as the soleus muscle of muscle-Slc25a34 KO mice contained larger myofibers than those of control mice (Fig. 4F). On the other hand, we did not find changes in the composition of muscle fiber types (Type I, IIa, and IId/x) between muscle-Slc25a34 KO mice and littermate control mice (fig. S5D). In alignment with the results in cellular respiration (shown in fig. S4E), there was no significant difference in muscle tissue oxygen consumption rate (JO₂) between the two groups when pyruvate, malate, and succinate were provided (fig. S5E). Additionally, muscle-Slc25a34 KO mice showed endurance capacity and grip strength similar to control mice (fig. S5, F and G).

However, we found that the muscle-Slc25a34 KO mice exhibited two notable abnormalities in the soleus muscle. First, there was a marked increase in centronuclear myofibers, defined by the presence of centrally located myonuclei, rather than the normal peripheral localization beneath the sarcolemma (arrowheads in the left panel of Fig. 4G). In contrast to healthy adult muscle fibers in control mice, where nuclei resided at the periphery, muscle-Slc25a34 KO mice contained centronuclear fibers, which are recognized as a morphological hallmark of active muscle regeneration, also reflecting cycles of degeneration or chronic myofiber stress (right panel in Fig. 4G; quantification is shown on the right graph). Second, the muscle-Slc25a34 KO mice displayed enlarged subsarcolemmal mitochondria in the soleus (Fig. 4H). It is worth noting that this mitochondrial enlargement was uniquely found in the subsarcolemmal compartment, but was not observed in the intermyofibrillar mitochondrial network (fig. S5H). This compartment-selective phenotype aligns with prior studies showing that subsarcolemmal mitochondria exhibit distinct metabolic sensitivities to amino acid-related stimuli, such as increased ADP-stimulated respiration and ATP production following acute amino acid infusion in humans (60), and augmented glutamate-driven respiration after endurance exercise training (61). These results suggest that loss of SLC25A34 resulted in abnormalities in the ultrastructure of subsarcolemmal mitochondria in the soleus muscle, contributing to localized metabolic stress and regenerative responses.

Altered protein turnover in the soleus muscle of Slc25a34 KO mice

To determine the molecular changes caused by SLC25A34 loss in the soleus muscle, we next performed RNA-sequencing in muscle-Slc25a34 KO and littermate control mice. The transcriptome analysis identified 680 genes that were significantly altered between the control and muscle-Slc25a34 KO groups (131 genes were upregulated, 549 genes were downregulated). Among these changes, the MSigDB hallmark gene sets analysis found that the myogenesis/regenerative gene signature (e.g., Mylpf, Myod1, Myh8, and Tnnt2) and IL-6/JAK/STAT3 signaling (e.g., Socs3) were significantly upregulated in muscle-Slc25a34 KO mice relative to control mice (Fig. 5, A and B). On the other hand, several genes known for the regulation of PI3K/AKT/mechanistic target of rapamycin (mTOR) signaling (e.g., Pten and Tsc1, negative regulators of mTOR signaling), were significantly downregulated in muscle-Slc25a34 KO mice relative to control mice. These transcriptional changes appeared to be independent of general mitochondrial dysfunction per se, as many, if not all, of the mitochondrial TCA cycle genes and OXPHOS proteins were expressed at equivalent levels in both groups (fig. S6, A and B).

Fig. 5. Altered protein turnover in the soleus muscle of Slc25a34 KO mice.

(A) MSigDB hallmark gene set enrichment analysis (Enrichr) was performed using genes that were significantly (q < 0.01) upregulated or downregulated in the Slc25a34 KO soleus muscle compared to the control in RNA-seq analysis. (B) Relative expression levels of genes related to myogenesis and PI3K/AKT/mTOR signaling in the soleus muscle of control and muscle-Slc25a34 KO mice. Data are represented as Z-score heatmaps for the indicated genes in each sample, based on RNA-seq analysis. n = 6 per group. Red indicates elevated levels, and blue indicates decreased levels. (C) Nascent protein synthesis in the soleus muscle of female control and muscle-Slc25a34 KO mice, assessed by the SUnSET assay. Immunoblotting was performed to detect the puromycin. The intensity of puromycin was normalized to the total protein level (Ponceau S staining). The molecular weights (kDa) are indicated on the right. n = 7 for control and 6 for muscle-Slc25a34 KO. (D) Nascent protein synthesis in the plantaris muscle of female control and muscle-Slc25a34 KO mice, assessed by the SUnSET assay. Immunoblotting was performed to detect the puromycin. The intensity of puromycin was normalized to the total protein level (Ponceau S staining). The molecular weights (kDa) are indicated on the right. n = 7 for control and 6 for muscle-Slc25a34 KO. Bars represent mean ± s.e.m. P values were calculated by Fisher’s exact test (A), unpaired t-test [(C) and (D)], and q-values were calculated by Wald test with Benjamini-Hochberg FDR correction (B).

The PI3K/AKT/mTOR signaling pathway plays a central role in integrating nutrient and hormonal cues to drive anabolic protein synthesis (62). Down-regulation of negative regulators of mTOR signaling suggested that elevated muscle protein synthesis was accompanied by active muscle regeneration and tissue remodeling in muscle-Slc25a34 KO mice. To test this possibility, we performed the SUnSET assay, a puromycin-based method to assess nascent protein synthesis in vivo (63). Control and muscle-Slc25a34 KO mice received an intraperitoneal injection of puromycin (0.04 μmol/g body weight), and the soleus and plantaris muscles were harvested exactly 30 min after puromycin injection. We found that puromycin incorporation into nascent polypeptides was significantly elevated in the soleus muscle of muscle-Slc25a34 KO mice relative to control mice (Fig. 5C). This change was selective to the soleus muscle rather than a global change in protein synthesis because there was no difference in puromycin incorporation into nascent proteins in the plantaris muscle between the two groups (Fig. 5D). These results suggest that muscle hypertrophy in the soleus of muscle-Slc25a34 KO mice is due to an elevated regeneration program and protein synthesis.

SLC25A34 loss leads to increased protein synthesis via the mTORC1 pathway

We asked to what extent the muscle hypertrophy observed in muscle-Slc25a34 KO mice was due to cell-autonomous changes in myofibers. Under a nutrient-replete culture condition (25 mM glucose, 5% horse serum, and amino acids, including 4 mM glutamine), there was no statistically significant difference in myotube diameter between control and Slc25a34 KO myotubes (Fig. 6A, left panels). When cells were cultured in a nutrient-deprived medium with HBSS supplemented with 5.5 mM glucose for 6 hours, control myotubes became noticeably thinner, and their diameter was significantly reduced relative to the fed condition, consistent with increased protein degradation during fasting (Fig. 6A, right panels). Slc25a34 KO myotubes were significantly resistant to this starvation-induced reduction in myotube diameter, indicating a defect in proteostasis under a nutrient-deprived condition (Fig. 6B). Since SLC25A35 shares the same substrate as SLC25A34, we next asked whether SLC25A35 also plays a similar biological role in starvation-induced myotube atrophy and amino acid metabolism. Of note, Slc25a35 mRNA levels were much lower than those of Slc25a34 in skeletal muscle (see fig. S1A) and were not induced by fasting (see Fig. 2A). We isolated myotubes from Slc25a35 KO mice and cultured them under a nutrient-deprived medium (HBSS supplemented with 5.5 mM glucose) for 6 hours, under the same protocol for Slc25a34 KO myotubes (fig. S7A). In contrast to Slc25a34 KO myotubes, Slc25a35 deficiency did not affect atrophy in response to starvation (Fig. 6C).

Fig. 6. SLC25A34 loss leads to increased protein synthesis via the mTORC1 pathway.

(A) Representative α-actinin and DAPI-stained myotubes. Left: Cells were cultured in a nutrient-replete medium. Right: Cells were cultured in a nutrient-deprived medium. Scale bar, 100 μm. Insets show higher-magnification views of the boxed regions. Scale bar, 50 μm. (B) Starvation-induced myotube atrophy in control and Slc25a34 KO primary myotubes. n = 74 for control-fed, n = 83 for control-starved, n = 95 for KO-fed, n = 67 for KO-starved. (C) Starvation-induced myotube atrophy in control and Slc25a35 KO primary myotubes. n = 22 for control-fed, n = 25 for control-starved, n = 29 for KO-fed, n = 22 for KO-starved. (D) Relative amino acid levels in control and Slc25a34 KO myotubes in a nutrient-replete medium, and in a nutrient-deprived medium. n = 5 per group. (E) Relative amino acid levels in control and Slc25a35 KO myotubes in a nutrient-replete medium, and in a nutrient-deprived medium. n = 3 per group. (F) mTORC1 signaling after 24 hours of fasting. Immunoblotting was performed to detect p-S6K^Thr389, total S6K, p-rpS6^Ser240/244 and total rpS6. n = 7 per group. (G) Schematic of SUnSET assay following the inhibition of mTORC1 signaling. Created in BioRender. Oikawa, S. (2026) https://BioRender.com/2fqjaos. (H) mTORC1 signaling in the soleus muscle of male control and muscle-Slc25a34 KO mice following rapamycin treatment. Immunoblotting was performed to detect p-rpS6^Ser240/244 and total rpS6. I. Nascent protein synthesis in the soleus muscles of male control and muscle-Slc25a34 KO mice following rapamycin treatment. n = 6 per group. Bars represent mean ± s.e.m. P values were calculated by two-way ANOVA with Tukey’s multiple comparisons test [(B), (C), (I)] and unpaired t-test [(D) to (F)].

Consistent with these morphological changes, we found differences in intracellular amino acid levels between control and Slc25a34 KO myotubes in a nutritional state-dependent manner. Metabolomics found elevated cellular proline levels in Slc25a34 KO cells compared to control cells (Fig. 6D, left panel, and table S5). In addition, several amino acids, including serine, glutamine, threonine, aspartate, and glutamate, were lower in Slc25a34 KO cells than in control cells. Under a nutrient-deprived condition, several amino acid levels were significantly elevated in Slc25a34 KO cells compared to control cells, including proline and glutamine (Fig. 6D, right panel, and table S6). In contrast, intracellular amino acid levels were largely unchanged in Slc25a35 KO myotubes compared with control myotubes under both fed and starved conditions Fig. 6E and tables S7 and S8). Together, these results indicate that SLC25A34 plays a distinct functional role in maintaining muscle integrity during nutritional deprivation, whereas SLC25A35 is dispensable, despite the two sharing PEP as a substrate.

Since these amino acids are known activators of the mTOR complex (64), and our RNA-seq data showed that several genes known for negatively regulating mTOR signaling, including Pten and Tsc1, were down-regulated in muscle-Slc25a34 KO mice (see Fig. 5B), these data collectively led to the hypothesis that mTOR complex 1 (mTORC1) signaling is activated in the absence of SLC25A34 even under a nutrition-deprived state. To test this, we harvested the soleus muscle of control and muscle-Slc25a34 KO mice after 24 hours of fasting. Immunoblotting showed that phosphorylated S6K^Thr389 and phosphorylated rpS6^Ser240/244, both of which are established markers of mTORC1-S6K signaling, were significantly elevated in the soleus muscle of muscle-Slc25a34 KO mice compared with controls (Fig. 6F). Because AMPK serves as a key sensor of cellular energy status and an upstream negative regulator of mTORC1 signaling (49, 65), we next examined AMPK signaling in the soleus muscle of muscle-Slc25a34 KO mice. However, we found that phosphorylated AMPKα^Thr172 levels were comparable between control and muscle-Slc25a34 KO soleus muscles after 24 hours of fasting (fig. S7B).

Lastly, we examined the functional contribution of elevated mTOR signaling to the increased protein synthesis observed in muscle-Slc25a34 KO mice. To this end, we treated muscle-Slc25a34 KO mice and littermate control mice with rapamycin (1.5 mg/kg) for 7 days via i.p. injection (Fig. 6G). After the treatment, we examined mTOR signaling as well as nascent protein synthesis in the soleus muscle. As anticipated, rapamycin treatment effectively reduced the levels of phosphorylated rpS6^Ser240/244 in both control and muscle-Slc25a34 KO mice (Fig. 6H). Consistent with the above result (see Fig. 5C), the SUnSET assay showed that vehicle-treated muscle-Slc25a34 KO mice showed elevated protein synthesis in the soleus compared to control mice that were treated with vehicle. However, rapamycin treatment for 7 days blunted the difference in protein synthesis between muscle-Slc25a34 KO mice and control mice (Fig. 6I and fig. S7C). Rapamycin-treated muscle-Slc25a34 KO mice and control mice exhibited comparable endurance capacity and grip strength (fig. S7, D and E). These results suggest that elevated amino acids and subsequent activation of mTORC1 signaling mediate active protein synthesis observed in the soleus muscle of muscle-Slc25a34 KO mice.

DISCUSSION

Skeletal muscle is the largest protein reservoir in the body and plays a central role in metabolic adaptation during fasting. Under the conditions of nutrient deprivation, muscle increases proteolysis to supply amino acids for hepatic gluconeogenesis and for use by other metabolic organs, while simultaneously oxidizing amino acids via anaplerotic pathways to sustain mitochondrial ATP synthesis. These adaptive responses are accompanied by suppression of mTORC1 signaling, reduced protein synthesis, and enhanced protein breakdown (7, 8). In this study, we identify SLC25A34 as a fasting- and exercise-inducible mitochondrial carrier that is required for maintaining amino acid catabolism in skeletal muscle (Fig. 7). SLC25A34 expression is induced by the PPARδ/PGC-1α transcriptional axis, linking its regulation to established programs of oxidative and endurance adaptation. Consistent with our recent identification of SLC25A35 as a mitochondrial PEP carrier (31), we found that SLC25A34 also mediates PEP transport into the mitochondrial matrix. SLC25A34-dependent PEP import in muscle supports glutamine-driven anaplerosis and preserves the balance between TCA cycle anaplerosis and cataplerosis during fasting. Loss of SLC25A34 resulted in blunted amino acid catabolism, leading to amino acid accumulation and elevated mTORC1 signaling under the conditions of nutritional deprivation. Importantly, muscle-specific deletion of Slc25a34 resulted in elevated protein synthesis, hypertrophy, and myopathy in the soleus muscle, whereas the increased protein synthesis can be reversed by mTORC1 inhibition. Together, the present study identifies SLC25A34 as a mitochondrial metabolic hub that preserves amino acid utilization and proteostasis in skeletal muscle under conditions of nutrient deprivation.

Fig. 7. A model of SLC25A34-mediated PEP transport across the IMM regulates amino acid catabolism and maintains metabolic resilience in skeletal muscle during fasting.

Under conditions of limited glucose availability, such as prolonged fasting, skeletal muscle relies on amino acids to sustain anaplerotic flux, accompanied by suppression of nutrient-sensing mTORC1 signaling and reduced protein synthesis. SLC25A34 expression is strongly induced by fasting through the PPARδ/PGC-1α transcriptional pathway. SLC25A34-dependent PEP import into mitochondria is essential for the balance between TCA cycle anaplerosis and cataplerosis during fasting; loss of SLC25A34 reduces mitochondrial PEP availability and impairs amino acid catabolism (indicated by red lines). As a consequence, SLC25A34 deficiency leads to amino acid accumulation, inappropriate activation of mTORC1 signaling, and enhanced protein synthesis, leading to hypertrophy and myopathic remodeling in the soleus muscle. Pharmacological inhibition of mTORC1 fully reverses enhanced protein synthesis, demonstrating that aberrant mTORC1 activation is the key driver of the anabolic remodeling in muscle lacking SLC25A34. Created in BioRender. Oikawa, S. (2026) https://BioRender.com/j1sufsm.

Broadly, our findings underscore the importance of coordinated anaplerotic and cataplerotic balance in skeletal muscle metabolism. During prolonged fasting, precise regulation of these pathways via SLC25A34 is necessary to maintain the TCA cycle intermediate pool by utilizing amino acids as alternative substrates. Disruption of this balance leads to the depletion or inappropriate accumulation of TCA intermediates and their precursors, with deleterious consequences for muscle integrity. Consistent with this principle, mutations in enzymes or transporters that regulate anaplerotic or cataplerotic flux are associated with muscle pathology and human disease. For example, SUCLA2, which encodes the β-subunit of ADP-forming succinyl-CoA synthetase, is required for efficient utilization of anaplerotic carbon entering the TCA cycle via amino acid-derived propionyl-CoA. Mutations in SUCLA2, which is highly expressed in skeletal muscle and brain, cause Leigh-like encephalomyopathy accompanied by elevated methylmalonic acid levels (66, 67). Loss-of-function mutations in SLC25A1, a mitochondrial citrate carrier that mediates cataplerotic export of citrate from the TCA cycle (68), are associated with congenital myasthenic syndrome characterized by fatigable proximal muscle weakness, enlarged mitochondria, and mitochondrial abnormalities (69, 70). Additionally, biallelic mutations in SLC25A10, which impair anaplerotic replenishment of malate and succinate, lead to severe epileptic encephalopathy with mitochondrial dysfunction in skeletal muscle (71).

We note two limitations in the present study. First, it remains unknown which metabolites mediate mTORC1 activation in muscle-Slc25a34 KO muscle. Several amino acids, including Met and Gln that activate mTORC1 signaling (72, 73), were elevated in Slc25a34 KO myotubes, whereas canonical mTORC1-activating amino acids (Leu, Arg) were unchanged. It is conceivable that Met via SAMTOR signaling mediates mTORC1 activation in Slc25a34-deficient muscles. The second is that the liposome-based assays could not determine the precise affinity of SLC25A34 relative to SLC25A35 due to the methodology we used, which required a millimolar concentration of ¹³C-PEP, substantially higher than physiologically relevant PEP concentrations (50–60 μM in HeLa cells). Additionally, the use of ultracentrifugation for separation limits the temporal resolution of PEP transport kinetics.

Nonetheless, our study provides insights into a broader principle of mitochondrial metabolite transport: several members of the SLC25A carrier family share overlapping substrate specificities, i.e., “redundant” carriers, yet fulfill distinct biological roles that arise from their tissue-specific expression patterns, regulatory mechanisms, and the unique metabolic demands in a given cell/tissue type. Despite SLC25A34 and SLC25A35 sharing PEP as a substrate, SLC25A34 performs a specialized, non-redundant function in skeletal muscle during nutritional stress, as SLC25A34, but not SLC25A35, is highly inducible by fasting, exercise, and cold exposure. Functionally, SLC25A35 is required for glycerolipid synthesis in lipogenic cells (31), whereas SLC25A34 is required for amino acid catabolism in the muscle in response to fasting. Similarly, the aspartate/glutamate carriers AGC1 (SLC25A12) and AGC2 (SLC25A13), which mediate exchange of aspartate for glutamate and protons across the IMM as part of the malate–aspartate shuttle (25), have distinct biological roles: AGC1 is highly expressed in skeletal and cardiac muscle and is essential for neuronal myelination (74), while AGC2 is ubiquitously expressed and regulates hepatic urea cycle function (75). Accordingly, mutations in SLC25A12 cause developmental and epileptic encephalopathies (76), whereas mutations in SLC25A13 result in citrullinemia type II and neonatal intrahepatic cholestasis (75). A similar distinction exists between ANT isoforms: ANT2 (SLC25A5) is ubiquitously expressed and essential for cellular viability, with deficiency causing embryonic lethality in mice (77), whereas ANT1 (SLC25A4) is enriched in energetically demanding tissues such as skeletal muscle and heart, where its loss leads to mitochondrial myopathy and cardiomyopathy in mice (78) and humans (79). Collectively, the present study reinforces the notion that mitochondrial carriers are not interchangeable conduits but are tailored to meet cell- and tissue-specific metabolic demands. Understanding such functional specialization will be crucial for deciphering mitochondrial contributions to metabolic adaptation, muscle disease, and systemic energy homeostasis, extending beyond ATP production.

MATERIALS AND METHODS

Animal study

All the animal experiments in this study were performed in compliance with protocols approved by the Institutional Animal Care and Use Committee (IACUC, Protocol# 028-2022-25) at Beth Israel Deaconess Medical Center. Mice were kept under a 12 hours:12 hour light-dark cycle at ambient temperature (22–23°C) and had free access to food and water. For the fasting experiments, the food was removed from cages of fasted animals for 24 hours. Slc25a34^flox/flox mice on a C57BL/6J background were generated by the Easi-CRISPR method at the BIDMC Transgenic Core Facility (52). Skeletal muscle-specific Slc25a34 knockout mice were obtained by crossing the Slc25a34^flox/flox mice with human α-skeletal actin (HSA)-Cre mice (B6.Cg-Tg (ACTA1-cre)79Jme/J, 006149 The Jackson Laboratory). A list of the ssDNA, gRNAs and genotyping primers sequences is provided in table S9.

Cells

C2C12 myoblasts were cultured with 20% FBS in DMEM at subconfluence. After myoblasts reached confluence, the medium was switched to 2% horse serum (heat-inactivated HS, Gibco) in DMEM for the differentiation. After 5–7 days of differentiation, myotubes were treated with PPARδ agonist (GW501516, SML1491, Sigma), PPARα agonist (fenofibrate, F6020, Sigma) or PPARγ agonist (rosiglitazone, 71740, Cayman Chemical) at indicated dose and time. For experiments that required starvation, differentiated myotubes were washed twice with PBS and incubated in Hanks’ Balanced Salt Solution (HBSS) medium (H8264, Sigma).

To isolate primary myoblasts, hindlimb muscles of Slc25a34^flox/flox mice or Slc25a35^flox/flox mice (1 week-old) were isolated and digested with digestion buffer containing Collagenase, type 2 (0.2%, Worthington Biochemical Corporation), Dispase II (2.4 U/ml, Millipore), and 2.5 mM CaCl₂ at 37°C for 45 min. The digested cell mixture was passed through a 100 μm and a 40 μm filter to remove cell debris. Cells were collected by centrifugation for 5 min at 350g and suspended in Ham’s F-10 nutrient mixture supplemented with 20% FBS and 2.5 ng/ml human recombinant bFGF (Stem Cell Technology). After 4 hours of incubation, the medium containing cells was transferred to collagen-coated plates and cultured overnight and then transferred to new collagen-coated plates again to purify primary myoblasts (pre-plating). At 60–80% confluence, muscle differentiation was induced by replacing the medium with 5% HS in DMEM. After 2 days of incubation, myotubes were infected with Ad-CMV-eGFP (1060, Vector Biolabs) or Ad-CMV-iCre (1045, Vector Biolabs) overnight and analyzed 2 days after the infection. For starvation experiments, cells were washed twice with PBS and cultured with HBSS (Sigma) for 6 hours. The fusion index was determined manually by calculating the ratio of nuclei in α-actinin-positive myotubes that contain more than two nuclei to the total number of nuclei.

DNA constructs and virus production

Mouse Slc25a34 cDNA was amplified from Mouse Untagged Clone (MC214787, ORIGENE) and ligated into a pcDNA3.1 plasmid (Thermo Fisher Scientific). For retrovirus production, HEK293T packaging cells were transfected with 10 μg of pMXs-3XHA-EGFP-OMP25, 5 μg of gag/pol, and 5 μg of VSV-G using the calcium phosphate method. After 48 hours of transfection, the virus-containing supernatant was collected, filtered through a 0.45 μm filter, and stored at −80°C for future use. Primary myoblasts were infected with the virus and polybrene (10 μg/ml) for 24 hours and then sorted with GFP using FACS.

Mitochondrial isolation and metabolomics

Primary Slc25a34^flox/flox myoblasts stably expressing a mitochondria-localized epitope-tag (HA-MITO) (53) were differentiated into myotubes for 2 days and then infected with adenovirus Ad-eGFP (control) or Ad-iCre (Slc25a34 KO) in a collagen-coated 15 cm dish. Mitochondrial metabolomics was performed 2 days after infection. Control and Slc25a34 KO myotubes were washed twice with ice-cold PBS and collected with KPBS (136 mM KCl, 10 mM KH2PO4, pH 7.25) into the tubes and then centrifuged for 2 min at 1000g at 4°C. The collected cells were resuspended in 1 ml KPBS, and 100 μl of the cell suspension was added to 500 μl of 100% MeOH for whole-cell metabolomic analysis. The cells were homogenized using a Potter-homogenizer with a PTFE pestle to obtain a mitochondrial fraction and then centrifuged for 2 min at 1000g at 4°C. The supernatant containing the mitochondrial fraction was mixed with 200 μl of Pierce Anti-HA Magnetic Beads (Thermo Scientific) on an end-over-end rotator for 4 min at 4°C. The beads were collected using a magnet rack for 1 min and carefully washed three times with KPBS. The bead suspension was collected using a magnet, and the supernatant was removed. Then 80% MeOH (80 μl) was added to the beads, and the samples were placed at −80°C overnight for metabolite extraction. After incubation, the samples were mixed well by vortexing and centrifuged for 10 min at 21,000g at 4°C. The supernatant was transferred to a new tube, and this step was repeated once, and stored at −80°C until LC-MS analysis. Metabolomics data were acquired using a UHPLC system (Vanquish Horizon, Thermo Scientific) coupled to an Orbitrap mass spectrometer (Exploris 240, Thermo Scientific) as described previously (6).

Structural prediction of SLC25A34

The structural models of SLC25A34 and SLC25A35 were obtained from the AlphaFold Protein Structure Database (https://alphafold.ebi.ac.uk/) (37) using the codes AF-Q6PIV7-F1-v6 for SLC25A34 and AF-Q3KQZ1-F1-v6 for SLC25A35. The predicted structures of SLC25A34 and SLC25A35 were aligned and visualized using UCSF ChimeraX-1.10.1 software (80). Protein sequence alignment and homology between human SLC25A34 and human SLC25A35 were analyzed using the EMBOSS Needle tool (81).

Purification of MBP- and TwinStrep-tagged SLC25A34

Codon-optimized N-terminally MBP-tagged and C-terminally Twin-Strep-tagged human wild-type SLC25A34 or mutant SLC25A34 or SLC25A35 cDNA was synthesized as gBlock gene fragments (IDT) and inserted into the pET-21b(+) expression vector (V011022, NovoPro) using the In-fusion Cloning Kit (638948, Takara). Bacteria (C43(DE3) transformed with pET-21b(+) empty or MBP-hSLC25A34-TwinStrep were cultured at 37°C for 6 hours in LB medium containing 100 μg/ml ampicillin, followed by a 1:100 dilution in 500 ml of LB medium containing 100 μg/ml ampicillin. After 1 hour of incubation at 37°C, isopropyl β-D-1-thiogalactopyranoside (IPTG, 0.1 mM, I56000, Research Products International) was added to induce protein expression, followed by overnight induction (20°C, 300 rpm, 14 hours). After induction, bacteria were collected and washed once with PBS, and then resuspended in W buffer (2–1003-100, IBA) containing a protease inhibitor cocktail (Roche). The resuspended pellets were lysed by high-pressure homogenization using a French Press (27 kpsi). The lysate was centrifuged at 8,000g for 10 min to remove the debris and supplemented with 2% Triton X-100 and rotated at 4°C for 2 hours. The lysate was centrifuged at 21,000g for 15 min, and the supernatant was rotated with a Strep-TactinXT 4Flow high-capacity resin (2–5030-010, IBA) at 4°C for 2 hours. The resin was washed five times with W buffer containing 2% Triton X-100 and a protease inhibitor cocktail, followed by four rounds of rotation in BXT buffer (2–1042-025, IBA) containing 1% Triton X-100 and a protease inhibitor cocktail at 4°C for a total of 2 hours to obtain purified proteins.

Bacterial PEP uptake by SLC25A34

Bacterial expression of SLC25A34 proteins and uptake assays were performed as described previously, with modification (38, 82, 83). MBP-tagged fusion human SLC25A34 was generated by combining E. coli MBP containing the MalE signal peptide and the SLC25A34 gene. Each protein was fused via a thrombin-cleavage site. MBP and codon-optimized hSLC25A34 coding sequences were synthesized as gBlock gene fragments (IDT) and cloned into the pET-21b(+) expression vector. Expression of MBP-hSLC25A34 was achieved in E. coli C43(DE3) cells (CMC0019-20X40UL; Biosearch Technologies) (84, 85). According to the manufacturer’s protocol, single aliquots of competent bacterial cells were transformed with pET-21b(+) containing MBP-hSLC25A34 and selected on LB agar plates containing 100 μg/ml ampicillin (37°C, overnight). Bacteria transformed with pET-21b(+) empty or MBP-hSLC25A34 plasmids were cultured overnight in LB medium containing 100 μg/ml ampicillin, followed by a 1:100 dilution in 300 ml of LB medium containing 100 μg/ml ampicillin. Refreshed bacteria were incubated (37°C, 200 rpm) until OD600 reached 0.3–0.4, and the cultures were removed and cooled on ice. The prepared bacteria were split into 50 ml aliquots in 250 ml flasks. IPTG (0.1 mM) was added to the chilled bacteria to induce protein expression, followed by overnight induction (20°C, 300 rpm, 14 hours). MBP-SLC25A45-expressing bacteria were prepared as previously described (40). The protein expression of SLC25A34 or SLC25A45 in bacteria was confirmed by immunoblotting.

Induced bacteria were collected by centrifugation (4000g, 10 min, 4°C), washed once with ice-cold potassium phosphate buffer (KPi), pH 7.4, and centrifuged again (4000g, 15 min, 4°C). The bacteria were resuspended in ice-cold KPi buffer, and the OD600 was measured. Thirty OD600*ml units of bacteria transformed with empty vector or overexpressing MBP-hSLC25A34 were resuspended in 1 ml of KPi buffer. For the uptake assay, 100 μM ¹³C₂-PEP (CLM-3398, Cambridge Isotope Laboratories) or 1 mM ¹³C₄-OAA (O845032, Toronto Research Chemicals) was added to the sample and shaken at 37°C for 30 min. For time-course experiments, the samples were incubated with 100 μM ¹³C₂-PEP, and aliquots were taken at 5, 15, 30, and 60 min. For competition assays, the samples were incubated with 100 μM ¹³C₂-PEP and unlabeled (¹²C) 50 mM PEP at 37°C for 30 min. Samples were centrifuged at 20,000g for 1 min and washed twice with KPi buffer. Metabolites were extracted with ice-cold 80% MeOH (PEP) or 50% Acetonitrile/30% MeOH (OAA) for LC-MS analysis. Bacterial uptake of ¹³C-labeled metabolites was analyzed by LC-MS and normalized to the total protein content of each sample.

Proteo-liposome assay

To prepare liposomes, 100 mg of the lipids (Egg PC, E. coli Polar lipids, 18:1 Cardiolipin at a 4:4.2:9 ratio, Avanti Polar Lipids) in 10 ml of chloroform was incubated in a rotary evaporator at 100 rpm at 50°C overnight. The lipid film on the internal surface of the flask was rehydrated with 2 ml of PIPES buffer (10 mM PIPES, 50 mM NaCl, pH 7.0) containing 20 mM unlabeled PEP. The liposomes were extruded 15 times using a mini-extruder with a 200-nm pore membrane at 60°C. Extruded liposomes were rotated with purified SLC25A34 protein or background proteins (purified from empty vector-transformed bacteria) at 4°C for 1 hour. Proteo-liposomes were then rotated with Bio-Beads SM-2 five times to completely remove Triton X-100. The resultant proteo-liposomes were isolated on a PD-10 desalting column (17–0851-01, Cytiva) to remove the external substrates and centrifuged at 100,000 rpm for 10 min. Proteo-liposomes were incubated with 5 mM ¹³C₂-labeled PEP in PIPES buffer at 37°C for 20 min and centrifuged at 100,000 rpm for 5 min. The pellet was washed twice with PIPES buffer, followed by centrifugation at 100,000 rpm for 5 min. The final pellet was lysed in 80% MeOH for LC-MS analysis to detect ¹³C₂-labeled PEP.

Mitochondrial uptake assays

Control and Slc25a34 KO myotubes were prepared by adenovirus infection in 10 cm dishes. The cells were collected in KPBS and centrifuged for 10 min at 600g at 4°C. The pellets were resuspended in KPBS and homogenized using a Potter-homogenizer with a PTFE pestle. Homogenates were centrifuged for 10 min at 600g at 4°C to remove cell debris and the mitochondria fraction containing supernatant was transferred into a new tube and then centrifuged for 10 min at 7000g at 4°C. The pellet was incubated in KPBS containing 0.32 μCi/ml of 1,5-¹⁴C-Citrate, [U-¹⁴C]-Malic acid, 1-¹⁴C-α-ketoglutarate, [U-¹⁴C]-Aspartic acid or [U-¹⁴C]-Leucine (Moravek) for 1 hour at 4°C, or 1 mM of [U-¹³C]-α-ketoisocaproic acid (KIC), [U-¹³C]-α-ketoisovaleric acid (KIV), [U-¹³C]-α-keto-β-methylvaleric acid (KMV) (CIL) and 10 μM of ¹³C₂-PEP for 5 min at 4°C. After incubation, mitochondria were washed twice with ice-cold KPBS. For ¹⁴C-labeled samples, mitochondria were lysed with 500 μl of RIPA buffer, and radioactivity was measured using a scintillation counter and normalized to the total protein content. For ¹³C-labeled samples, metabolites were extracted using 80% MeOH and analyzed by LC-MS with signals normalized to the total protein content.

Derivatization

To quantify ¹³C₄-OAA and ¹³C₄-malate, the extracted solvent containing ¹³C₄-OAA and ¹³C₄-malate was derivatized with 3-nitrophenylhydrazine (3-NPH, N21804, Sigma) following a published method (43). The solvent was mixed with 200 mM 3-NPH in 50% acetonitrile and 120 mM N-(3-Dimethylaminopropyl)-N´-ethylcarbodiimide (EDC, E1769, Sigma) in 50% acetonitrile/6% pyridine (1.09728, Sigma) and incubated at 4°C for 45 min with gentle shaking. After incubation, the samples were centrifuged for 5 min at 20,000g at 4°C and used for LC-MS analysis. All reagents were freshly prepared.

Whole-cell tracing experiments

For glucose tracing, differentiated primary myotubes were incubated with 5.5 mM [U-¹³C]-Glucose (Cambridge Isotope Laboratories, CLM-1396-1) in HBSS (137 mM NaCl, 5.4 mM KCl, 0.34 mM Na2HPO4, 0.44 mM KH2PO4, 4.2 mM NaHCO3, 1.26 mM CaCl₂ and 0.4 mM MgSO₄) for 6 hours. For pyruvate and glutamine tracing, the cells were incubated for 6 hours in HBSS containing 5.5 mM glucose supplemented with either 1 mM [U-¹³C]-Pyruvate (Cambridge Isotope Laboratories, CLM-2440) or 2 mM [U-¹³C]-Glutamine (Cambridge Isotope Laboratories, CLM-1822). After incubation, the cells were washed twice with ice-cold PBS and collected for metabolite extraction with ice-cold 80% MeOH and stored at −80°C until LC-MS analysis. The labeling rate of each metabolite was calculated by dividing the level of labeled metabolites by the total metabolite level.

Electron microscopy and quantification

Soleus muscle was immersion fixed in 2.5% glutaraldehyde (Electron Microscopy Sciences, Hatfield, PA), 2% paraformaldehyde (Electron Microscopy Sciences), and 0.3% picric acid (Sigma-Aldrich, St. Louis, MO) in 0.1 M sodium cacodylate buffer (Sigma-Aldrich) pH 7.4 for 1 hour at room temperature then at 4°C overnight. Tissues were washed with 0.1 M sodium cacodylate buffer, and then post-fixed for 1 hour at 4°C in 1% osmium tetroxide (Electron Microscopy Sciences) in 0.1 M sodium cacodylate buffer. Tissues were washed in DI water and incubated in 2% aqueous uranyl acetate (Electron Microscopy Sciences) overnight at 4°C. The following day, tissues were washed with DI water and then dehydrated at 4°C in a graded ethanol series (Thermo Fisher Scientific, Waltham, MA). The tissues were then brought to room temperature and dehydrated with 100% ethanol (Thermo Fisher Scientific) followed by propylene oxide (Electron Microscopy Sciences). Infiltration in LX112 resin (Ladd Research Industries, Williston, VT), was followed by embedding in flat bottom Beem capsules (Electron Microscopy Sciences). The resulting blocks were sectioned using a Leica Ultracut E ultramicrotome (Leica Biosystems Nussloch, Germany), and sections were placed on formvar (Electron Microscopy Sciences) and carbon-coated grids. The sections were contrast stained with 2% uranyl acetate (Electron Microscopy Sciences) followed by lead citrate (Sigma-Aldrich) and imaged in a JEOL 1400 transmission electron microscope (JEOL, Peabody, MA) equipped with a Gatan Orius SC1000 digital CCD camera (Gatan, Pleasanton, CA). The mitochondrial area was quantified by tracing the outer membrane using a free-hand tool and normalizing to scale.

Respiration in isolated muscle fiber

A portion of the soleus and plantaris muscles was dissected and placed in Buffer Z (1 mM EGTA, 5 mM MgCl₂, 105 mM K-MES, 30 mM KCl, 10 mM KH2PO4, 5 mg/ml fatty acid-free BSA, pH 7.2). Muscle fibers were separated under a microscope and permeabilized in Buffer Z with saponin (30 μg/ml) at 4°C for 30 min. The muscle fibers were incubated in ice-cold Buffer Z with pyruvate (5 mM)/malate (2 mM) and then transferred to the chamber of an Oroboros O2k respirometer (Oroboros Instruments) in Buffer Z. Respiration was measured in Buffer Z with 5 mM pyruvate, 2 mM malate, 0.2 mM ADP, 5 mM succinate, and 5 μM rotenone. After the experiments, the muscle fibers were washed, and the protein concentration was measured to normalize the respiration data.

Cellular respiration assay

Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured using the Seahorse XFe Extracellular Flux Analyzer (Agilent) in a 24-well plate. Control and Slc25a34 KO primary myotubes were cultured in standard assay medium containing 25 mM glucose, 2 mM glutamine and 1 mM pyruvate. For the measurement of uncoupled respiration, myotubes were treated with 2 μM oligomycin, 1.5 μM carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) and 2 μM antimycin A.

Immunofluorescent imaging

HEK293T cells were placed in a glass-bottom dish (10810–054, VWR) and cultured for 24 hours. The cells were then transfected with 100 ng of pcDNA3.1-SLC25A34 plasmid using Lipofectamine 3000 (Thermo Fisher Scientific). 24 hours after transfection, the cells were washed twice with PBS and fixed with 4% paraformaldehyde (PFA) at 37°C for 30 min. The cells were washed three times with PBS and permeabilized with 0.3% NP-40, 0.05% Triton X-100, and 0.1% bovine serum albumin (BSA) in PBS for 3 min. After three rinses with wash buffer (0.05% NP-40, 0.05% Triton X-100, and 0.2% BSA in PBS), the samples were blocked for 1 hour at room temperature with SuperBlock Blocking Buffer (37515, Thermo Fisher Scientific). The samples were then incubated with primary antibodies in wash buffer overnight at 4°C, washed three times with wash buffer. The primary antibodies used targeted TOMM20 (11802-AP, Proteintech) and FLAG (8146S, Cell Signaling Technology). Secondary antibodies were conjugated with Alexa Fluor 488 (ab150117, abcam) and Alexa Fluor 647 (A21245, Invitrogen). Images were acquired using Zeiss LSM900 confocal microscope with Airyscan and processed using Zeiss Zen software.

Control, Slc25a34 KO, and Slc25a35 KO myotubes were fixed with 4% PFA at 37°C for 10 min, washed with PBS, and then permeabilized with 0.1% Triton X-100 in PBS at room temperature for 15 min. The cells were then washed with PBS and blocked with 2% BSA in PBS at room temperature for 1 hour and incubated with the α-actinin antibody (A7811, Sigma) at room temperature for 1 hour. After washing with PBS, the cells were incubated with Alexa Fluor 594-conjugated anti-mouse IgG1 secondary antibody and DAPI at room temperature for 1 hour. Images of the myotubes were acquired using the Revolve microscope (ECHO Laboratories).

Frozen cross-sections from OCT-embedded muscle tissues were fixed with 4% PFA in PBS and washed once with PBS. The samples were permeabilized with 0.3% Triton X-100 in PBS, washed twice with PBS and then blocked with 5% normal goat serum in PBS. The tissue sections were then incubated at 4°C overnight with the following primary antibodies: Laminin (L9393, Sigma), myosin heavy chain (MyHC) I (BA-F8, DSHB), MyHC IIa (SC-71, DSHB), or dystrophin (D8043, Sigma). The secondary antibodies were Alexa Fluor 647-conjugated anti-rabbit IgG for laminin, DyLight 405-conjugated anti-mouse IgG2b for MyHC I, Alexa Fluor 488-conjugated anti-mouse IgG1 for MyHC IIa and dystrophin (Jackson ImmunoResearch Lab). Images were acquired using the Revolve microscope (ECHO Laboratories).

Immunoblotting

Skeletal muscle tissues were homogenized in complete protein loading buffer containing 50 mM Tris-HCl (pH 6.8), 1% SDS, 10% Glycerol, 20 mM dithiothreitol (DTT), 127 mM 2-mercaptoethanol, and 0.01% bromophenol blue, supplemented with protease inhibitors and phosphatase inhibitors (Sigma). The protein contents were measured with an RC DC Protein Assay Kit (Bio-Rad) according to the manufacturer’s instructions. The proteins were electrophoresed on SDS-PAGE precast gels (Bio-Rad) and transferred to a PVDF membrane (Millipore), and the signals were immunodetected with Clarity Western ECL Substrate (Bio-Rad) using the ChemiDoc Imaging System (Bio-Rad). The following antibodies were used: Puromycin (MABE343, Sigma), S6K (#2708, CST), p-S6K (#9205, CST), p-rp-S6 (#5364, CST), rp-S6 (#2217, CST), Strep-Tactin-HRP (2-1502-001, IBA), MBP tag (66003-1-Ig, Proteintech), ATP5A (ab14748, Abcam), Total OXPHOS (ab110413, Abcam), AMPK alpha (#2532, CST), p-AMPK alpha (#2535, CST) and GAPDH (2118S, CST). Anti-Rabbit IgG, HRP-Linked antibody (ab6721, Abcam) and anti-mouse IgG HRP (#31430, Thermo Fisher Scientific) were used as secondary antibodies.

Animal physiology

The endurance exercise capacity test was performed as described previously (86). Mice were acclimatized to the treadmill for 5 min at 7 m/min on a 10% incline on two consecutive days. The test began at 8 m/min for 30 min, followed by 9 m/min for 15 min and 10 m/min for 15 min. The speed was then increased by 1 m/min every 10 min until the mice were exhausted. Exhaustion was defined as the point at which the mice were unable to return to the treadmill belt after 10 sec of encouragement. Grip strength was measured using a grip strength meter (Ametek) by placing all four limbs of the mice on a metal grid. The peak force was recorded in each of five trials, and the median value was used for analysis.

SUnSET assay

In vivo SUnSET assay was performed to evaluate nascent protein synthesis as described previously (63). Mice were intraperitoneally injected with puromycin (0.04 μmol/g body weight) dissolved in 100 μl PBS. At precisely 30 min after injection, soleus and plantaris muscles were dissected and immediately frozen in liquid nitrogen for immunoblotting. To inhibit mTORC1 signaling, mice were intraperitoneally injected with rapamycin (1.5 mg/kg) for 7 consecutive days.

RNA-sequencing and Data Analysis

Total RNA was purified from the soleus using the RNeasy Fibrous Tissue Mini Kit (Qiagen) with DNase treatment on the column. Total RNA samples were quantified using a Qubit 2.0 Fluorometer (Life Technologies), and RNA integrity was checked using a 4200 TapeStation (Agilent Technologies). The samples were initially treated with TURBO DNase (Thermo Fisher Scientific) to remove DNA contaminants. ERCC RNA Spike-In Mix (Thermo Fisher) was added to normalized total RNA prior to library preparation, following the manufacturer’s protocol. The next steps included rRNA depletion using the QIAGEN FastSelect rRNA HMR Kit (Qiagen). RNA sequencing libraries were multiplexed and clustered on the flowcell. After clustering, the flowcell was loaded onto an Illumina NovaSeq Instrument. The samples were sequenced using a 2x150 Paired-End (PE) configuration. After demultiplexing, the sequence data were checked for overall quality and yield. Then, raw sequence reads were trimmed to remove possible adapter sequences and nucleotides with poor quality using Trimmomatic v.0.36. The reads were then mapped to the reference genome (mm10) using the STAR aligner v.2.5.2b. Unique gene hit counts were calculated by using featureCounts from the subread package v.1.5.2. Only unique reads that fell within exon regions were counted. Downstream analysis was performed in R. Differential expression analysis was performed using DESeq2. P values were calculated using the Wald test and adjusted for multiple testing using the Benjamini-Hochberg procedure. Genes with Benjamini-Hochberg-adjusted p value (false discovery rate; q value) < 0.05 were considered differentially expressed. Gene enrichment analysis was performed using Enrichr (87).

RT-qPCR

Total RNA was isolated from cells and non-muscle tissues using the Direct-zol RNA Purification Kit (Zymo Research), and from skeletal muscle using the RNeasy Fibrous Tissue Mini Kit (Qiagen), according to the manufacturer’s protocols. For the measurement of mRNA expression, 1 μg of total RNA was reverse transcribed with Oligo-dT(20) primer using SuperScript IV Reverse Transcriptase (Thermo Fisher Scientific). The synthesized cDNA and SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) were used to quantify mRNA levels. GAPDH and 18S rRNA were used as the internal controls. The primer sequences are presented in table S10.

Statistics

Statistical analyses were conducted using GraphPad Prism version 10 (GraphPad). All data are represented as mean ± s.e.m. Data were obtained from biologically independent samples. P < 0.05 was considered to be significant throughout the study. For comparison between two groups, unpaired Student’s t-tests or the Mann-Whitney test were employed. To compare three or more groups, one-way ANOVA followed by suitable post-hoc tests was applied. To assess the effects of two independent factors, two-way ANOVA or two-way repeated-measures ANOVA was applied with Tukey’s multiple comparisons test. For metabolomics and SLC25A family transcriptomics analyses, unpaired t-tests followed by Benjamini-Hochberg FDR correction were used.

Supplementary Materials

This PDF file includes:

Figs. S1 to S7

Tables S1 to S10