AMC-F1 regulates mitochondria-autophagy crosstalk independent of nutrient stress Authors: Yuqin Wang, Raksha K Rao, Trung Vu, Ayano Sekine, Zhengmei Mao, Nami McCarty Published: 2026 ABSTRACT AbstractMitochondria and autophagy are fundamental yet distinct regulators of cellular homeostasis. Here, we identify AMC-F1 (Autophagy-Mitochondria Coupling Factor 1; formerly TRIM44) as a central integrator of mitochondrial bioenergetics and autophagy. Using Amcf1 knockout and knock-in mouse models, we demonstrate that AMC-F1 bidirectionally regulates these pathways: its loss reduces mitochondrial respiration and autophagic flux, whereas its overexpression promotes mitochondrial elongation and increases autophagy independently of nutrient stress. Transcriptomic analyses reveal AMC-F1-dependent regulation of mitochondrial biogenesis programs that engage autophagy, involving mitochondrial respiratory chain complex genes under basal conditions and mitochondrial organization factors under starvation-induced autophagy. Although dispensable under homeostasis, this coupling becomes essential during stress adaptation. In an acute liver-injury model, Amcf1 knock-in mice were fully protected, exhibiting elevated OPA1, reduced caspase-3 and PARP activation, and preserved Beclin 1. This functional duality reflects AMC-F1’s ability to modulate the mitochondrial integrated stress response (mtISR), enabling adaptive ATF4 signaling while preventing maladaptive responses when stress exceeds a threshold. Autophagy upregulation by AMC-F1 is critical for fine-tuning the ISR and preserving cellular resilience. Together, our findings position AMC-F1 as a stress-responsive gatekeeper and a novel coordinator of mitochondrial-autophagy crosstalk, defining a cellular state primed for stress adaptation.Subject terms: Mitochondria, Autophagy, Cancer epidemiology Introduction Mitochondria are central regulators of cellular metabolism and energy production, but they also serve as signaling hubs for inter-organelle communication and stress coordination1,2. A key stress pathway linked to mitochondrial function is the mitochondrial integrated stress response (mtISR), which coordinates protein synthesis and gene expression to determine cell fate under stress3–5. The outcome of this response is shaped by multiple factors, including stress intensity and duration, mitochondrial integrity, genetic context, and the activation of compensatory survival mechanisms6,7. Another major stress-adaptive pathway is macroautophagy (hereafter autophagy). Autophagy is a conserved catabolic process that maintains homeostasis by degrading damaged organelles, misfolded proteins, and other cytoplasmic components, and has traditionally been considered distinct from the anabolic functions of mitochondria8–11. While crosstalk between mitochondria and autophagy has been observed in Saccharomyces cerevisiae and mammalian cells under nutrient-deprivation12,13, direct molecular integrators of these pathways remain unidentified. Moreover, it remains unclear whether mitochondrial elongation during nutrient deprivation is merely a passive consequence of autophagy activation, or whether mitochondria actively regulate their own fate and whether these two processes are coordinated under homeostatic or stress conditions. Here, we identify AMC-F1/TRIM44 is a novel coordinator of mitochondrial functions and autophagy pathways. Our findings reveal that AMC-F1 bidirectionally regulates mitochondria-autophagy crosstalk under both homeostatic and stress conditions. Its activation enhances basal autophagic flux even in the absence of nutrient deprivation, accompanied by increased expression of mitochondrial fusion proteins, elevated complex activity, and oxidative phosphorylation, thereby uncovering a previously unrecognized communication axis in cells. Results AMC-F1 modulates mitochondrial morphology AMC-F1/TRIM44 is a deubiquitinase originally identified as upregulated in non-proliferative, stem-like multiple myeloma (MM) cells14. To investigate the role of AMC-F1 in normal mammalian development, we generated conditional Amcf1−/− (knockout, KO) (Supplementary Fig. 1a) and Amcf1KI/KI (knock-in, KI) mice (Supplementary Fig. 1b). Whole-body deletion or overexpression was achieved using Sox2-Cre (Supplementary Fig. 1c, d). Amcf1 expression was examined across multiple tissues (Supplementary Fig. 1e). Both Amcf1 KO and KI mice were viable and exhibited no significant differences in body weight, daily food intake, or litter size compared to littermate controls (Supplementary Fig. 1f-h). Despite normal organismal phenotypes, transmission electron microscopy (TEM) revealed striking AMC-F1-dependent alterations in mitochondrial ultrastructure. Mitochondria in Amcf1 KO mouse embryonic fibroblasts (MEFs) were rounder and exhibited significantly reduced length and area relative to wild-type (WT) controls (Fig. 1a, Supplementary Fig. 2a). In contrast, Amcf1 KI MEFs displayed elongated mitochondria, effectively reversing the KO-associated phenotype (Fig. 1b, Supplementary Fig. 2b). Re-expression of Amcf1 in Amcf1 KO MEFs restored mitochondrial morphology, confirming a direct role for AMC-F1 in regulating mitochondrial dynamics (Fig. 1c, Supplementary Fig. 2c). In addition, three-dimensional (3D) super-resolution imaging of mitochondria was performed using Multiplex 4Y Airyscan-2 mode on a Zeiss LSM980 microscope, with whole-cell scanning following TOMM20 immunostaining. This analysis confirmed AMC-F1-dependent mitochondrial remodeling (Supplementary Fig. 2d, e). AMC-F1 levels modulate mitochondrial morphology. a–f Representative transmission electron microscopy (TEM) images of mitochondria from Amcf1−/− (knock-out, KO, dark green) and littermate Amcf1+/+ (wild-type, WT, light green) MEFs under fed (a) or fasted (d) conditions; Amcf1KI/KI (knock-in, KI, dark pink) and littermate Amcf1WT/WT (WT, light pink) MEFs under fed (b) or fasted (e) conditions; and Amcf1−/− MEFs with doxycycline-inducible rescue expression of Amcf1 (KO + R, dark yellow) or without induction (KO, light yellow) under fed (c) or fasted (f) conditions. Boxed areas are enlarged. Violin plots showing the quantification of mitochondrial length and area are presented below for each corresponding group (a, n = 112 for WT, n = 114 for KO; b, n = 115 for WT, n = 110 for KI); c, n = 110 for KO, n = 171 for KO + R; d, n = 98 for WT, n = 80 for KO; e, n = 123 for WT, n = 119 for KI; f, n = 90 for KO, n = 135 for KO + R). Each dot represents an individual mitochondrion. Solid and dashed lines indicate the median and interquartile range, respectively. Statistical significance was determined using a two-tailed unpaired Student’s t test; P values are shown. Scale bars, 600 nm. Given that mitochondria elongate during starvation-induced autophagy to evade sequestration12, we next examined whether AMC-F1 modulates this adaptive response. Following 2.5 h of nutrient deprivation, Amcf1 WT MEFs exhibited marked mitochondrial elongation and increased mitochondrial area, whereas Amcf1 KO MEFs failed to mount this response (Fig. 1d). Notably, mitochondrial shortening in Amcf1 KO MEFs during starvation was more pronounced than that observed in Amcf1 WT MEFs under nutrient-replete conditions (Supplementary Fig. 2f). In contrast, Amcf1 KI MEFs displayed robust mitochondrial elongation under starvation (Fig. 1e, Supplementary Fig. 2g). Re-expression of AMC-F1 in KO MEFs under starvation conditions restored mitochondrial morphology, further confirming a direct role for AMC-F1 in regulating stress-induced mitochondrial remodeling (Fig. 1f). As Amcf1 KO and KI mice exhibited comparable feeding behavior, these findings suggest that AMC-F1-depedent regulation of mitochondrial morphology is uncoupled from systemic energy status, implicating AMC-F1 as an intrinsic sensor of mitochondrial state. AMC-F1 controls basal and starvation-induced autophagy in vivo Having elucidated that AMC-F1 deficiency disrupts mitochondrial elongation under both basal and nutrient-deprived conditions, we next assessed whether AMC-F1 regulates autophagic flux, an integrated measure of autophagic degradation15. To directly assess the role of AMC-F1 in autophagy in vivo, Amcf1−/− and Amcf1KI/KI mice were each crossed with the autophagy reporter CAG-mRFP-EGFP-MAP1LC3B mice16, the gold standard reporter for monitoring autophagic flux. This strategy generated KO (Amcf1−/−;RFP-EGFP-LC3B) and KI (Amcf1KI/KI;RFP-EGFP-LC3B) cohorts, respectively (Supplementary Fig. 3a). Fasting condition and status were confirmed across reporter cohorts (Supplementary Fig. 3b). Confocal imaging of tissues from Amcf1-/-;RFP-EGFP-LC3B mice revealed significantly fewer GFP-RFP+ autolysosome puncta in the liver, heart, gastrocnemius muscle, and kidney under both basal (Fig. 2a, b) and starvation conditions (Fig. 2e), indicating impaired autophagic flux in vivo. Loss of AMC-F1 impairs autophagic flux under both fed and fasted conditions. a Schematic of mouse treatments and tissue collection. Created in BioRender. https://BioRender.com/fy991jc. b, e Representative confocal images of RFP-EGFP-LC3 puncta in the indicated tissues from KO (Amcf1−/−; RFP-EGFP-LC3B) or littermate WT (Amcf1+/+; RFP-EGFP-LC3B) mice under fed conditions with chloroquine (CQ; 50 mg kg−1, 6 h) treatment (b) or after 24 h fasting (e). CQ was used to inhibit lysosomal degradation and amplify basal autophagy signals. Boxed areas are enlarged. Immunoblot analysis of LC3B (RFP-EGFP-LC3B; top, long exposure; bottom, short exposure) and p62 levels in the indicated tissues under basal conditions (c) or after fasting (f). d, g Quantification of total RFP-EGFP-LC3B and p62 levels in tissues corresponding to the immunoblots shown in (c) or (f). Filled circles indicate the tissue analyzed in each comparison. KO, light green; WT, gray. h Representative confocal images showing LC3 puncta in MEFs expressing GFP-LC3-RFP, including Amcf1+/+ (WT, top) and Amcf1−/− (KO, bottom), under fed, fasted, or fasted with bafilomycin A1 (BafA1,100 nM) conditions for 2 h. i Quantitative analysis of neutral autophagosomes (GFP+RFP+, yellow) and acidified autolysosomes (GFP-RFP+, red) per cell from (h). For WT groups, n = 19 cells for fed, n = 21 cells for fasted, and n = 17 cells for fasted + BafA1; For KO groups, n = 26 cells for fed, n = 25 cells for fasted, and n = 15 cells for fasted + BafA1. In (c, f), each lane represents an individual mouse. In (d, g), Data are presented as mean ± SD; n = 4 mice per group. In (i), each dot represents an individual cell. Solid and dashed lines indicate the median and interquartile range, respectively. One-sided unpaired Student’s t test (d, g), or two-way ANOVA with Tukey’s multiple comparisons test (i). P values are indicated. Scale bars, 10 μm. Source data are provided as a Source Data file. We further assessed autophagy biochemically by analyzing the total levels of RFP-EGFP-LC3B and the levels of the autophagy substrate p62 in tissues from KO and WT (Amcf1+/+;RFP-EGFP-LC3) mice. As previously reported, autophagy levels varied across organs, and these markers did not always change concordantly17, reflecting the inherent challenges of assessing autophagy in vivo. Nonetheless, KO tissues exhibited significantly increased total LC3B and accumulation of p62 in most organs, consistent with reduced autophagic flux under both fed (Fig. 2c, d) and fasted conditions (Fig. 2f, g). In contrast, tissues from KI (Amcf1KI/KI;RFP-EGFP-LC3B) mice showed enhanced autophagic flux across multiple organs under both conditions (Supplementary Fig. 3c). We also used KO MEFs transfected with the tandem mRFP-GFP-LC3 reporter to distinguish autophagosomes and autolysosomes, enabling stage-specific analysis across the autophagy pathway in vitro. Quantification of autophagosomes (GFP+RFP+ puncta) and autolysosomes (GFP-RFP+ puncta) revealed a significant reduction in autolysosome formation in KO MEFs compared with littermate WT controls (Fig. 2h, i; P = 0.0416 under fed condition and P= 0.0006 under fasted condition). Treatment with bafilomycin A1 (BafA1), an inhibitor of lysosomal acidification18, reduced starvation-induced autophagic flux in WT cells, as shown by accumulation of autophagosomes (P = 0.0009) and a decrease in autolysosomes (P = 0.0408). In contrast, BafA1 had minimal additional effect in KO MEFs, with continued accumulation of autophagosomes (P = 0.0030) and no further reduction in autolysosome formation (P = 0.5705). These findings indicate that AMC-F1 is required for efficient progression from autophagosomes to autolysosomes. Consistently, immunoblot analysis of LC3B-II accumulation demonstrated reduced autophagic flux under basal and fasted conditions in KO MEFs relative to WT MEFs, while KI MEFs exhibited enhanced autophagy (Supplementary Fig. 3d). To independently validate these findings, we employed a doxycycline-inducible Tet-On system, which demonstratred that inducible AMC-F1 expression promotes autophagic flux under both basal and fasted conditions (Supplementary Fig. 3e). Collectively, our data support that AMC-F1 is a key regulator of autophagic flux both in vivo and in vitro, independently of external nutrient signals. AMC-F1 requires MFN1 and synergizes with MFN2 to promote mitochondrial elongation and prevents stress-induced mitochondrial shortening Previous studies have shown that Mfn2-/- MEFs, unlike deletion of the dynamin-related GTPases optic atrophy 1 (Opa1−/−) or double knockout of mitofusin 1 and 2 (Mfn1−/−;Mfn2−/−), retain the ability to undergo mitochondrial elongation in response to starvation-induced autophagy12. We therefore investigated whether AMC-F1 participates in this process. TEM imaging showed silencing Amcf1 significantly impaired starvation-induced mitochondrial elongation in Mfn2−/− MEFs (Supplementary Fig. 4a, b). On the other hand, 3D super-resolution imaging of mitochondria with whole-cell volumetric scanning following TOMM20 immunostaining revealed that induction of Amcf1 in Mfn2−/− MEFs enhanced mitochondrial elongation upon starvation (Supplementary Fig. 4c, d). These data indicate that AMC-F1-driven mitochondrial elongation or shortening is independent of MFN2, while MFN2 can cooperate with AMC-F1 to influence mitochondrial morphology under nutrient stress. In contrast, induction of Amcf1 in Mfn1−/− MEFs did not restore mitochondrial elongation during starvation (Supplementary Fig. 4e), indicating that AMC-F1 requires MFN1 to promote mitochondrial elongation. Notably, silencing Amcf1 in Mfn1−/− MEFs resulted in further mitochondrial shortening (Supplementary Fig. 4g), suggesting that AMC-F1 also contributes to mitochondrial morphology through an MFN1-independent mechanism, potentially by supporting OPA1-dependent inner membrane integrity or limiting pro-fission activity. Together, these findings suggested that AMC-F1 regulates mitochondrial morphology at multiple levels, extending beyond outer membrane fusion to involve inner membrane dynamics. Given the role of AMC-F1 in coordinating autophagy-mitochondria crosstalk, we next examined its subcellular localization. Co-localization analysis with the mitochondrial marker TOMM20 revealed that AMC-F1 is distributed in both mitochondria and the cytoplasm (Supplementary Fig. 5a). This mitochondrial association was consistently observed across multiple independent approaches, including analysis of AMC-F1 using a smaller epitope tag (Myc-tagged AMC-F1; Supplementary Fig. 5b), and re-expression of untagged AMC-F1 in Amcf1−/− MEFs (Supplementary Fig. 5c), supporting a dual localization compatible with roles in both autophagy regulation and mitochondrial remodeling. AMC-F1 did not co-localize with the endoplasmic reticulum (ER); however, induction of ER stress with thapsigargin promoted a redistribution of AMC-F1 towards the ER in HeLa cells, indicating that AMC-F1 localization is dynamically regulated in response to cellular stress and may participate in context-dependent organelle communication (Supplementary Fig. 5d). Collectively, our data support a model in which AMC-F1 functions upstream of the mitochondrial fusion machinery by engaging MFN1 to promote mitochondrial elongation, whereas MFN2 is dispensable, but may cooperate with AMC-F1 to modulate the extent of mitochondrial elongation. AMC-F1 facilitates OPA1-dependent mitochondrial elongation and limits FIS1-mediated fission The above findings prompted us to further investigate whether AMC-F1 additionally regulates other components of the mitochondrial dynamic machinery. We therefore analyzed the expression of key regulators of mitochondrial fusion and fission, including OPA1, fission protein 1 (FIS1), and dynamin-related protein 1 (DRP1). OPA1, a keymediator of starvation-induced mitochondrial elongation19, was also required for AMC-F1 function. Inducible expression of AMC-F1 using a Tet-On system failed to restore mitochondrial elongation during starvation in Opa1−/− MEFs (Supplementary Fig. 6a), indicating that AMC-F1-mediated mitochondrial remodeling depends on OPA1. Consistent with this finding, AMC-F1 overexpression (KI) increased OPA1 levels while concomitantly reducing FIS1 abundance in MEFs, whereas silencing Amcf1 (KO) produced the opposite effects. In contrast, DRP1 protein levels were not altered under either condition (Supplementary Fig. 6b). To further investigate how AMC-F1 regulates OPA1, we analyzed whether this effect involves OMA1, a mitochondrial inner membrane metalloprotease that mediates stress-induced cleavage of OPA1. OMA1-dependent processing of OPA1 reduces the long isoform (L-OPA1) and generates short isoforms (S-OPA1), thereby impairing OPA1-mediated mitochondrial fusion. Consistent with this model, we observed increased levels of the long isoform (L2) under basal conditions and accumulation of short OPA1 isoform (S5) upon CCCP treatedKI MEFs, whereas the opposite pattern was observed in KO MEFs (Supplementary Fig. 6c). Consistent with these findings, levels of OMA1 were increased in livers from Amcf1 KO mice (Supplementary Fig. 6d). Because AMC-F1-dependent mitochondria elongation persisted in Mfn2−/− MEFs, we next assessed whether this effect was associated with changes in fission machinery. Inducible expression of Amcf1 in Mfn2−/− MEFs reduced FIS1 levels in an AMC-F1-dependent manner (Supplementary Fig. 6e). Similarly, altered expression of FIS1 and OPA1 was observed in Amcf1 KO MEFs, whereas DRP1 levels remained unchanged under both basal and fasted conditions (Supplementary Fig. 6f). We also expressed OPA1 in Amcf1 KO or WT MEFs. Opa1-overexpressing Amcf1 KO cells exhibited increased mitochondrial shortening (P = 1.098e-006), indicating that AMC-F1 required to maintain OPA1-driven mitochondrial elongation (Supplementary Fig. 6g). Silencing Fis1 in these cells partially restore elongation, suggesting that AMC-F1 prevents mitohcondrial fragmentation by restraining FIS1-dependent fission (Supplementary Fig. 6g). We note that, due to the transient nature of induced plasmid-based expression, OPA1 levels may vary between cells, and additional genetic approaches will be required to further validate these mechanisms. Together, these findings indicate that AMC-F1 promotes OPA1-dependent inner membrane fusion while simultaneously restraining fission through selective modulation of FIS1, thereby coordinating opposing mitochondrial dynamics programs under both basal and starvation conditions. AMC-F1 regulates mitogenesis and mitochondrial respiratory chain programs during basal autophagy conditions To define transcriptional pathways regulated by Amcf1, we first performed RNA-seq profiling of Amcf1 KO versus WT MEFs under basal autophagy conditions. Differential expression analysis identified 3648 significantly altered genes (adjusted P < 0.05), comprising 1777 downregulated and 1871 upregulated transcripts in Amcf1 KO cells (Supplementary Fig. 7a). Gene set enrichment analysis (GSEA) using the HALLMARK database revealed MYC target gene signature as the most significantly enriched pathway (Supplementary Fig. 7b, c). Although MYC is best known for its oncogenic functions, it also serves as a central regulator of mitochondrial biogenesis and bioenergetic capacity in non-transformed cells20. Because the HALLMARK MYC target V1 gene set captures only a subset of known MYC-regulated genes, we expanded our analysis using a published compendium of 7054 direct MYC targets defined by ChIP-seq profiling21. Among these, 576 encoded mitochondrial-associated factors, hereafter referred to as mitobiogenesis genes20. Silencing of AMC-F1 significantly downregulated 124 MYC-induced mitobiogenesis genes (Fig. 3a, b, Supplementary Fig. 7d), supporting a role for AMC-F1 in sustaining the MYC-driven mitochondrial gene expression program in non-transformed cells. Gene Ontology (GO) analysis revealed that these genes were enriched across multiple mitochondrial functional categories, including mitochondrial respiratory chain complex assembly (26 genes), mitochondrial translation (72 genes), mitochondrial transport (22 genes), and mitochondrial organization (49 genes), and other mitochondrial-related processes (407 genes). Under basal autophagy conditions, Amcf1 silencing primarily affected genes involved in the mitochondrial respiratory chain complex assembly (8 genes out of 26, 30%). Although respiratory chain complex genes account for 20% of all regulated genes, similar to other functional categories - normalized analysis revealed a clear enrichment. Specifically, 30% of mitochondrial respiratory chain complex genes were regulated by Amcf1 (8 regulated out of 26 genes), compared with 20% of mitochondrial translation (15 genes regulated out of 72), 40% of mitochondrial transport genes (9 genes regulated out of 22), and only 2% of mitochondrial organization genes (8 regulated out of 407) (Fig. 3c). These results indicate that Amcf1 selectively regulates mitochondrial chain complex genes under basal conditions, which is confirmed by qRT-PCR (Fig. 3f). In contrast, the association between Amcf1 and the mitogenesis gene signature was lost upon nutrient starvation, a condition that induces autophagy (Fig. 3d). Notably, under fasted conditions, both the total number of mitogenesis genes affected (18 vs. 40) and their distribution across categories were markedly changed (Fig. 3e). The proportion of mitochondrial respiratory chain complex genes regulated was markedly reduced, with only 8% (2 of 26), in sharp contrast to basal autophagy conditions. AMC-F1 regulates MYC-dependent mitochondrial biogenesis genes under basal conditions. RNA-seq analysis of MEFs derived from Amcf1−/− (KO) and littermate Amcf1+/+ (WT) mice under fed (a–c) or fasted conditions (d, e). a Heatmap of direct MYC targets involved in mitochondrial biogenesis (mitogenesis)20. Genes are grouped into four major functional categories based on Gene Ontology (GO) enrichment analysis: mitochondrial respiratory chain complex (light green), mitochondrial translation (dark green), mitochondrial transportation (beige), and mitochondrial organization (purple). b, d Gene set enrichment analysis (GSEA) revealed significant negative enrichment of MYC target genes involved in mitogenesis in KO versus WT MEFs under fed conditions (b; normalized enrichment score (NES) = −2.0014; false discovery rate (FDR) q < 0.0001), but not under fasted conditions (d; NES = 1.0140; FDR q = 0.4149). Pie chart showing the distribution of mitogenesis genes regulated by Amcf1 under fed conditions (c) and fasted conditions (e). Percentages are shown with the corresponding absolute number of genes indicated in parentheses. f qRT-PCR analysis of selected mitogenesis genes confirming downregulation of multiple transcripts in KO (light blue) MEFs compared with WT (light gray) MEFs under fed conditions. In (f), Data are presented as mean ± SD; n = 3 biological replicates for each group. Statistical significance was determined using an unpaired two-tailed Student’s t test. P values are indicated. AMC-F1 links oxidative phosphorylation to autophagy In addition to MYC target gene signature, oxidative phosphorylation (OXPHOS) signaling ranked among the most markedly repressed HALLMARK pathways in Amcf1 KO cells (Supplementary Fig. 7b). Given the role of AMC-F1 in regulating MYC-driven mitogenesis program, we hypothesized that AMC-F1 also controls mitochondrial respiratory function. Transcriptomic analysis revealed broad and significant downregulation of genes encoding mitochondrial respiratory cahin subunits following Amcf1 deletion (Fig. 4a). To functionally validate these transcriptional changes, we assessed mitochondrial respiration by measuring oxygen consumption rate (OCR) in MEFs using Seahorse assays. Consistent with their elongated mitochondrial morphology, Amcf1 KI MEFs displayed increased ATP production, elevated basal and maximal respiration, and enhanced OXPHOS capacity (Fig. 4b). In contrast, Amcf1 KO MEFs exhibited fragmented, spherical mitochondria accompanied by reduced ATP production and diminished OXPHOS activity (Fig. 4c). To extend these observations and assess their physiological relevance in vivo, we examined OXPHOS across multiple tissues from Amcf1 KO and WT littermate mice. Immunoblotting and ELISA analyses revealed reduced expression of mitochondrial complex I in the heart and liver (Fig. 4d–f), which were accompanied by decreased complex I enzymatic activity (Fig. 4g). However, kidney tissue did not exhibit a significant reduction in either complex I subunit expression or enzymatic activity, indicating that AMC-F1-dependent regulation of mitochondrial respiration could display organ-specific sensitivity in vivo. Given accumulating evidence linking OXPHOS to cancer progression22, we next examined the role of AMC-F1 in HeLa cells. Lentiviral-mediated overexpression of Amcf1 significantly increased ATP production and enhanced both basal and maximal respiration (Supplementary Fig. 7e), closely recapitulating the metabolic phenotype observed in Amcf1 KI MEF cells. Consistently, immunoblot analysis showed increased levels of complex I in Amcf1 KI MEFs and decreased expression in Amcf1 KO MEFs (Supplementary Fig. 7f). Taken together, these findings establish AMC-F1 as a regulator of MYC-driven mitogenesis and demonstrate that loss of AMC-F1 impairs mitochondrial respiratory function, primarily through disruption of complex I activity in vivo. AMC-F1 silencing suppresses oxidative phosphorylation by inhibiting complex I activity in vivo. a RNA-seq analysis of MEFs derived from Amcf1−/− (KO) and the corresponding Amcf1+/+ (WT) controls under fed conditions. The volcano plot depicts differential expression of genes encoding components of the mitochondrial electron transport chain. Genes with statistically significant changes are indicated (|log2 fold change| > 1 and adjusted P value < 0.05). The x-axis represents log2 fold change, and the y-axis represents -log10 (adjusted p-value). b, c Oxygen consumption rate (OCR) measurements in MEFs from Amcf1 KI (KI, light pink) and control Amcf1 WT (WT, dark gray) (b), Amcf1 KO (KO, green) and its control Amcf1 WT (WT, gray) (c). Right panels, quantification of basal respiration (Basal), maximal respiration (Max), spare respiratory capacity (Spare), and ATP production (ATP). Immunoblot analysis of mitochondrial respiratory complex subunits in heart (d), liver (e), and kidney (f) tissues from Amcf1−/− (KO, green) and control Amcf1WT (WT, gray) mice. Subunits analyzed include mitochondrial complex I (CI, NDUFB8), complex II (CII, SDHB), complex III (CIII, UQCRC2), complex IV (CIV, MTCO1), and complex V (CV, ATP5A). VDAC and β-actin were used as loading controls. Each lane represents an individual mouse. Similar results were obtained in three biological replicates. g Relative mitochondrial complex I enzymatic activity in the indicated tissues (WT, gray, n = 5; KO, green, n = 3). Data are presented as mean ± SD (b, c, g). In (b, c), Data points indicate biological replicates (b, n = 3; c, n = 5). A two-sided statistical test with adjustment for multiple comparisons using the Benjamini-Hochberg FDR method (a), two-tailed unpaired Student’s t tests (b, c), and two-way ANOVA with Sidak’s multiple-comparisons test (g). P values are shown. Source data are provided as a Source Data file. AMC-F1 remodels mitochondrial transcriptional program during starvation-induced autophagy Although AMC-F1 regulates mitochondria-autophagy crosstalk under both basal and starvation conditions, the mitochondrial gene programs influenced by Amcf1 deletion differed markedly between these states. During starvation-induced autophagy, significantly fewer genes associated with mitochondrial bioenergetic processes were downregulated compared with basal autophagy conditions (68 vs. 124) (Supplementary Figs. 7d, 8a). Under basal autophagy conditions, AMC-F1 primarily regulated genes involved in mitochondrial respiratory chain function. In contrast, during starvation-induced autophagy, mitochondrial organization genes were mostly affected, indicating a functional reprogramming of AMC-F1-dependent mitochondrial regulation in response to nutrient stress. To assess whether Amcf1 loss influences transcriptional responses during starvation-induced autophagy, we performed GSEA. In Amcf1 WT MEFs, nutrient deprivation robustly induced an autophagy-related gene signature (Supplementary Fig. 8c, left). In contrast, this transcriptional program was markedly attenuated in Amcf1 KO MEFs, indicating that AMC-F1 remains required for proper induction of autophagy genes activation during starvation (Supplementary Fig. 8c, right). Notably, under both basal and starvation conditions, Amcf1status altered complex I activity (Supplementary Fig. 8d). To further delineate this signaling axis, we analyzed multiple mTORC1 downstream targets in Amcf1 KI and KO MEFs. While modest changes were observed in KO MEFs, Amcf1 KI MEFs exhibited reduced basal mTORC1 activity, as evidenced by decreased phosphorylation of S6K (T389), ULK (S757), and 4E-BP1 (T37/46) (Supplementary Fig. 8e). Upon starvation, mTORC1 activity was markedly suppressed in WT, consistent with global inhibition of mTORC1 signaling under nutrient deprivation. Notably, phosphorylation levels of these targets were further reduced in Amcf1 KI MEFs compared with WT controls, indicating enhanced suppression of mTORC1 signaling. Together, our data support that AMC-F1 promotes mitochondrial elongation under basal and starvation conditions, while engaging distinct downstream transcriptional and signaling outputs depending on nutrient availability. Amcf1 KI mice are completely protected from Fas-induced apoptotic liver injury Although both Amcf1 KO and KI mice exhibited altered mitochondrial morphology and basal autophagy, they remained phenotypically normal under unstressed conditions, indicating that AMC-F1 is dispensable for homeostatic maintenance. However, similar to Parkin-PINK1 and OMA1 knock out models23, AMC-F1 may function primarily in stress adaptation. To explore this possibility, we employed in vivo stress models known to dependent on mitochondrial bioenergetics or autophagy. The liver constitutively expresses Fas on hepatocytes, and Fas-induced hepatocyte apoptosis engages both mitochondrial pathways24,25 and autophagy26. However, this model has not been previously examined in the context of coordinated regulation of mitochondrial function and autophagy in vivo. Strikingly, following administration of the anti-Fas antibody Jo2, Amcf1 KI mice exhibited marked protection against acute liver injury. Compared with littermate WT controls, Amcf1 KI livers preserved lobular architecture and displayed significantly reduced hemorrhagic infiltration after Jo2 challenge (Fig. 5a). Of five control mice, three succumbed within 24 h post-injection, whereas all five Amcf1 KI mice survived without histological evidence of liver damage. TUNEL staining confirmed a significant reduction in apoptotic cell death in Amcf1 KI livers (Fig. 5b, c). Serum levels of liver injury markers, alanine aminotransferase (ALT) and aspartate aminotransferase (AST), were significantly lower in Amcf1 KI mice (Fig. 5d). In contrast, Amcf1 KO mice exhibited markedly elevated ALT and AST levels than WT littermates following Jo2 injection (Supplementary Fig. 9a). Immunoblotting analysis of liver lysates revealed robust induction of cleaved caspase-3 and PARP in WT mice, but not in KI mice, following injury (Fig. 5e). In addition, cleaved Beclin 1, a caspase-3 substrate that promotes apoptosis under stress, was similarly upregulated in WT mice. In contrast, Amcf1 KI livers retained elevated levels of full-length Beclin 1, a key initiator of autophagy. Notably, Amcf1 KI livers also exhibited elevated OPA1 expression, particularly following Jo2 challenge (Fig. 5e). To determine whether the protective effect of AMC-F1 is mediated through enhanced autophagy and preservation of mitochondrial integrity, and is intrinsic to AMC-F1 expression, primary hepatocytes were isolated from WT and Amcf1 KI mice and treated with a combination of Jo2 antibody and actinomycin D (Act.D), an established in vitro method for Fas activation in hepatocytes27. Fas-induced apoptosis, as assessed by levels of cleaved caspase-3 and cleaved PARP, was markedly attenuated in the Amcf1 KI hepatocytes (Supplementary Fig. 9b). Additionally, Amcf1 KI hepatocytes exhibited increased autophagic activity, as indicated by reduced LC3B-II accumulation. Pharmacological inhibition of autophagy with CQ significantly increased apoptosis in WT hepatocytes, whereas this pro-apoptotic effect was markedly reduced in Amcf1 KI hepatocytes. In parallel, disruption of mitochondrial integrity by depolarization of the mitochondrial membrane potential using CCCP induced apoptosis in WT hepatocytes, yet CCCP-induced apoptosis was also attenuated in Amcf1 KI hepatocytes. Collectively, our data support that AMC-F1 confers protection against Fas-induced liver injury by preserving mitochondrial integrity and increasing autophagic flux during apoptotic stress. Consistent with this enhanced mitochondrial and autophagic state, Amcf1 KI hepatocytes displayed increased resistance to both autophagy inhibition and mitochondrial depolarization induced cell death. Amcf1 KI mice are resistant to Fas-induced apoptotic liver injury. The extent of liver injury in Amcf1WT (WT) and Amcf1KI (KI) mice following Fas activation in vivo was assessed by H&E staining (a), TUNEL labeling and quantification (b, c), and measurement of serum ALT and AST levels (d). Amcf1 KI and WT male mice were injected with saline or Jo2 (0.25 μg g−1; n = 5 per group) and euthanized at 24-h post-injection. e Immunoblot analysis were conducted to assess the cleavage of caspase 3 and PARP, autophagy marker LC3B, and mitochondrial fusion protein OPA1 in total liver lysates from Amcf1 KI and its control WT mice. β-actin was applied as a loading control. Similar results were obtained in three biological replicates. In (c), Quantification of TUNEL-positive hepatocytes in WT and KI mice based on multiple TUNEL-stained sections. Data are presented as mean ± s.e.m (b, d). Two-tailed unpaired Student’s t test (c, d). P values are indicated. Scale bars, 50 μm (a) and 20 μm (b). Source data are provided as a Source Data file. AMC-F1 modulates the mitochondrial integrated stress response in a stress threshold-dependent manner Because AMC-F1 conferred complete protection against Fas-induced apoptotic liver injury, we next investigated whether this effect involves regulation of the mitochondrial integrated stress response (mtISR), a pathway commonly activated in response to electron transport chain dysfunction3–5,28,29. Analysis of our RNA-seq datasets revealed that Amcf1 silencing significantly suppressed ISR-related gene expression under both basal and starvation conditions, with suppression further amplified during starvation-induced stress (Fig. 6a, b). RNA-seq analysis of liver tissue recapitulated the ISR regulatory trend observed in MEFs; however, fewer genes were significantly altered, likely reflecting dilution of cell-intrinsic stress responses within heterogeneous liver tissue in vivo (Supplementary Fig. 10a). Among the ISR pathways, activating transcription factor 4 (ATF4), a central regulator of stress-adaptive transcriptional networks30, was particularly affected. ATF4 target genes were markedly downregulated in Amcf1 KO MEFs under basal and starvation conditions (Fig. 6c). To further delineate the relationship between AMC-F1 and mtISR signaling dynamics, we analyzed liver tissues from WT and KI mice at 4 h (early ISR phase) and 24 h (delayed ISR phase) following Jo2 injection (Supplementary Fig. 10b, c). During early ISR activation, Amcf1 KI livers showed reduced caspase activation and a remodeled ISRcharacterized by early ATF4 induction with reduced phosphorylation of eIF2α (Ser51) and CHOP expression. By 24 h, ATF4 levels returned to baseline while CHOP suppression persisted, consistent with suppression of the maladaptive/pro-apoptotic ISR. Our data support that AMC-F1 promotes survival by remodeling mtISR dynamics, enabling early adaptive ATF4 signaling while suppressing CHOP and apoptosis, and maintaining CHOP repression during the late ISR phase. AMC-F1 promotes adaptive ISR signaling while suppressing maladaptive ISR responses. a–c RNA-seq analysis of MEFs derived from Amcf1−/− (KO) and littermate WT mice under basal (fed) and starvation (fasted) conditions: heatmap showing the expression profile of integrated stress response (ISR) signature genes (a), GSEA of ISR signature genes in KO vs WT MEFs under fed conditions (top; NES = −2.41; FDR q < 0.0001), and fasted conditions (bottom; NES = −2.86; FDR q < 0.0001) (b), and heatmap showing the expression profile of ATF4 target genes predicted by iRegulon (Cytoscape plug-in)29,57,58 (c). In (a) and (c), genes are grouped by hierarchical clustering, and expression values are presented as z-scores, where red indicates higher expression and blue indicates lower expression. Each column represents a biological replicate. The top annotation bars denote experimental conditions (fed, light blue; fasted, teal), and samples are further grouped as WT (lavender/blue) and KO (salmon/orange) under each condition. AMC-F1 is amplified and upregulated, and predicts poor prognosis in hepatocellular carcinoma and across pan-cancer cohorts Amcf1 is a driver amplicon located on chromosome 11p13 and is amplified in approximately ~12% of all cancers, with higher frequencies observed in breast (19.8%), melanoma (18%), lung (16%), colorectal (16.8%), and esophageal cancers (15.9%)31. Given its role in coordinating mitochondrial function and autophagy, we hypothesized that Amcf1 upregulation facilitates stress adaptation in cancer cells, thereby promoting tumor progression. AMC-F1 expression was elevated by ~40% in Amcf1 KI mice (Supplementary Fig. 11a). To evaluate the clinical relevance of this upregulation, we analyzed published studies and a pan-cancer cohort comprising 19,131 samples, which revealed that Amcf1 levels were increased by ~15–50% across multiple human cancers compared to normal tissues (Supplementary Fig. 11b)32, indicating that our transgenic KI model likely recapitulate the upregulation observed in patients. To determine whether AMC-F1 expression has prognostic significance, we examined the Human Protein Atlas database, which indicates that high AMC-F1 expression is associated with unfavorable prognosis in liver cancer (LIHC) (Supplementary Fig. 11c). Further analysis of ISR signature genes revealed a strong positive correlation with Amcf1 expression, with heatmap analysis showing marked upregulation of ISR genes in tumors with elevated Amcf1 levels (Supplementary Fig. 11d), suggesting a role in stress adaption. We next evaluated the prognostic relevance of Amcf1 across pan-cancer cohorts. Kaplan-Meier survival analyses, stratifying patients into high- and low- Amcf1 expression groups using optimal, median, or quartile thresholds, consistently showed that elevated Amcf1 expression was associated with reduced overall survival (OS) across the pan-cancer cohort, as well as in pan-lympho-hematopoietic (Ly-hem), pan-stem cell-like (SC), and pan-squamous cell carcinomas (Squamous) subtypes33, irrespective of cutoff methods (Supplementary Fig. 12a–d). Moreover, Amcf1 expression was significantly higher in advanced versus early-stage tumors (Supplementary Fig. 12e). Additional clinical outcome endpoints, including progression-free interval (PFI), disease-specific survival (DSS), and disease-free interval (DFI), were also significantly worse in patients with high-Amcf1 groups, supporting a pan-cancer prognostic role for Amcf1 (Supplementary Fig. 12f). Discussion In this study, we identify AMC-F1/TRIM44 as a previously unrecognized regulator of mitochondria-autophagy crosstalk that modulates the ISR response in a stress-threshold-dependent manner. While this regulatory axis is largely dispensable under basal conditions, it becomes essential for effective stress adaptation in both normal and malignant cells. Although Fas activation engages extensive apoptotic and stress signaling cascades, apoptotic outcome is determined by a limited set of downstream regulatory nodes, including caspase amplification, mitochondrial commitment, Beclin 1 cleavage, and CHOP induction. Our data indicate that AMC-F1 coordinates communication across these nodes by enhancing mitochondria-autophagy crosstalk in AMC-F1-expressing hepatocytes, thereby limiting caspase amplification and suppressing CHOP-dependent apoptotic commitment. Though this coordinated control, AMC-F1 elevates the threshold for irreversible apoptosis, converting otherwise lethal death-receptor signaling into a survivable stress response. AMC-F1 is encoded by TRIM44; however, unlike most TRIM proteins, it lacks a RING E3 ligase domain and instead functions as a deubiquitinase. Therefore, we refer to TRIM44 by its functional alias, AMC-F1, while retaining the approved gene symbol. Although mitochondria and autophagy have independently been studied in the context of cellular stress responses, our findings identify AMC-F1 as a novel integrator that functionally couples these pathways and links them to organismal stress adaptation via ISR regulation. The ISR is activated in response to mitochondrial ETC dysfunction and other cellular stresses3,4,28,29,34–39, leading to global suppression of protein synthesis through phosphorylation of eIF2α at Ser5140 while selectively enhancing translation of adaptive transcription factors such as ATF4, thereby enhancing stress tolerance. On the other hand, autophagy alleviates cellular stress by clearing protein aggregates that trigger ER stress and by supplying intracellular amino acids11,41. Thus, both pathways have important roles in cellular adaptation, yet how they communicate under homeostatic or stress conditions remains unclear. Our work positions AMC-F1 at the intersection of these networks, providing a molecular mechanism for coupling mitochondrial signals to autophagic flux. Enhancing this axis may represent a more effective way to cope with stress. Notably, Amcf1 KI mice were protected from liver damage to a greater extent than Opa1 KI mice, indicating that the protective effect of AMC-F1 extends beyond OPA1-mediated mitochondrial fusion. Unlike essential mitochondrial complex proteins or core fusion factors, AMC-F1 is not required for basal mitochondrial function, suggesting that silencing or overexpressing AMCF1 will have a limited impact on homeostasis and could therefore represent a more clinically tractable strategy. Stress adaptation is similarly critical for tumor progression and therapy resistance. In cancer, autophagy pathways display a functional duality: they suppress tumorigenesis in normal cells but promote survival and drug resistance in established cancers11. A similar pattern is seen with the ISR, where transient activation supports adaptation, but chronic signaling drives cell death. Collectively, our findings identify AMC-F1 as a novel regulator that bidirectionally coordinates mitochondrial and autophagic programs, thereby coupling mitohcondiral adaptation to autophagic activation. By enhancing mitochondrial complex activities and OXPHOS, AMC-F1 reinforce the proper engagement of autophagic flux and thereby efficiently supports cell survival under stress through modulating of ATF4-dependent ISR programs. Methods Mouse strains C57BL/6J mice (C57BL/6J, 000664), Sox2-Cre mice (B6.Cg-Edil3Tg(Sox2-cre)1Amc/J, 008454), which express Cre recombinase under the control of the SRY-box transcription factor 2 (Sox2) promoter, and CAG-RFP-EGFP-LC3 reporter mice (C57BL/6-Tg(CAG-RFP/EGFP/Map1lc3b)1Hill/J, 027139) mice16,42 were obtained from The Jackson Laboratory. Conditional Amcf1 knock-in (Amcf1KI-fl/+) and knockout (Amcf1KO-fl/+) were generated by Ingenious Targeting Laboratory (Ronkonkoma, NY). These mice were crossed with Sox2-Cre mice to generate whole-body Amcf1 knock-in (Amcf1KI/KI) or knockout (Amcf1−/−), respectively, and subsequently backcrossed onto C57BL/6J background. Genotyping of Amcf1 allele was performed according to protocols provided by Ingenious Targeting Laboratory. For Amcf1 knock-in allele identification, the following primers were used: forward primer 5′-TGAACTT CAAGATCCGCCACA-3′ and reverse primer 5′-TGCTTACATAGTCTAACTCGCGAC-3′, yielding a 904-bp fragment. Identification of knock-in littermate wild-type alleles was performed using forward primer 5′-CTGCAACTCCAGTCTTTCTAGAAGATG-3′ and reverse primer 5′-CCAGCTACAGCCTCGATTTGTGGTG-3’, generating a 620-bp fragment. For Amcf1 knockout allele detection, primers used were forward primer 5′-CCCATTTCTATACATGATCACCGTTGT-3′ and reverse primer 5′-TCATGCTATTGCTTCCCGTATGGC-3′, producing a 912-bp fragment. Knockout littermate wild-type alleles were identified using forward primer 5′-CCAAGAG GAAGTGTCCAGACCATG-3′ and reverse primer 5′-CTCCTTGGTTTTGTC AAAATCACCGTT-3′, yielding a 550-bp fragment. Genotyping of the Map1lc3b (CAG-RFP-EGFP-LC3) transgene was performed according to the Jackson Laboratory using primers 24935, 24936, oIMR7338, and oIMR7339. Animals were housed under a 12/12 h light/dark cycle at 20–22 °C and 30–70% relative humidity at the University of Texas Health Science Center at Houston animal facility. All animal procedures were approved by the UTHealth Institutional Animal Care and Use Committee. Reagents The following primary antibodies were used: LC3B (Cell Signaling Technology, 2775), GFP (Santa Cruz Biotechnology, sc-9996), β-actin (GenScript, A00702), OXPHOS (Abcam, ab110413), AMC-F1/TRIM44 (Proteintech, 11511-1-AP, 1:1000 for immunoblotting, 1:100 for immunofluorescence staining), Caspase 3 (Cell Signaling Technology, 9662), cleaved Caspase 3 (Cell Signaling Technology, 9661), cleaved PARP (Cell Signaling Technology, 9541), Beclin 1 (BD Biosciences, 612113), OPA1 (BD Biosciences, 612607), OMA1 (Santa Cruz Biotechnology, sc-515788), MFN1 (Proteintech, 13798-1-AP), MFN2 (Abcam, ab56889), FIS1 (Proteintech, 10956-1-AP), TOMM20 (Proteintech, 11802-1-AP, 1:1000 for immunoblotting, 1:200 for immunofluorescence staining), VDAC (Cell Signaling Technology, 4866), CHOP (Cell Signaling Technology, 2895), ATF-4 (Cell Signaling Technology, 11815), phospho-eIF2α (Ser51) (Cell Signaling Technology, 3398), eIF2 alpha (Cell Signaling Technology, 9722), phospho-p70 S6 kinase (Thr389) (Cell Signaling Technology, 9234), p70 S6 kinase (Cell Signaling Technology, 9202), phosphor-ULK1 (Ser757) (Cell Signaling Technology, 6888), ULK1 (Cell Signaling Technology, 8054), phosphor-4E-BP1 (Thr37/46) (Cell Signaling Technology, 2855), 4E-BP1 (Cell Signaling Technology, 9644), Myc-tag (Cell Signaling Technology, 2276; 1:200 for immunofluorescence staining). The following secondary antibodies were used: anti-Rabbit IgG (H + L) conjugated with horseradish peroxidase (HRP) (Invitrogen, 31460; 1:10,000), anti-Mouse IgG (H + L) conjugated with HRP (Invitrogen, 31430; 1:10,000) for immunoblotting and anti-Rabbit IgG (H + L) conjugated with Alexa FluorTM 546 (Invitrogen, A10040) for immunostaining. ER-TrackerTM dye (Invitrogen, E34250) was used to label the endoplasmic reticulum. The following drugs were used: bafilomycin A1 (MedChemExpress, HY-100558), chloroquine (Sigma-Aldrich, C6628), doxycycline (MedChemExpress, HY-N05658), carbonyl cyanide m-chlorophenyl hydrazone (CCCP) (MedChemExpress, HY-100941), Thapsigargin (Sigma-Aldrich, T9033), and actinomycin D (Selleck Chemicals, S8964). Constructs and siRNA The following constructs were used: pEGP-N1 (Addgene, 172281), pEGP-N1-AMC-F143, GFP-LC3-RFP (Addgene, 117413), GFP-LC3 (Addgene, 24920), pMSCV-OPA1 (Addgene, 26047). The following small interfering RNA oligonucleotides were used: mouse AMC-F1-siRNA (Santa Cruz Biotechnology, sc-154652), mouse Fis1 siRNA (Santa Cruz Biotechnology, sc-60644), control siRNA (Santa Cruz Biotechnology, sc-37007). Cells were transfected using LipofectamineTM 3000 transfection reagent (Invitrogen, L3000015) according to the manufacturer’s instruction. Cell culture HeLa (CCL-2), 293T (CRL-3216), Mfn1 null MEFs (CRL-2992), Mfn2 null MEFs (CRL-2993), and Opa1 null MEFs (CRL-2995) were obtained from the American Type Culture Collection (ATCC). Cells were cultured in Dulbecco’s Modified Eagle’s medium (DMEM) (Corning, 10-013-CM) supplemented with 10% fetal bovine serum (FBS) (GenDEPOT, F0601-050), 2 mM L-glutamine (Thermo Fisher Scientific, 25030081) and 1% penicillin–streptomycin (Thermo Fisher Scientific, 15140122). HEK293FT cells (Invitrogen, R70007), used for high-efficiency lentiviral production, and were cultured in DMEM/F12 supplemented with GlutaMAXTM (Thermo Fisher Scientific, 10565-042), 10% FBS, 1% penicillin-streptomycin and 100 µg mL–1 GeneticinTM (Thermo Fisher Scientific, 10131027). All cells were maintained at 37 °C in a humidified incubator with 5% CO2. For starvation-induced stress, cells were washed twice with pre-warmed PBS (GenDEPOT, CA008-050) and cultured in Earle’s Balanced Salt Solution (EBSS, Sigma-Aldrich, E2888) for 2.5 h unless otherwise indicated. All cell lines were routinely tested for mycoplasma contamination using the MycoAlert Mycoplasma Detection kit (Lonza) and were used only when a confirmed negative result was obtained. Lentivirus production Lentivirus particles were produced using the packaging plasmid (psPAX2, Addgene, #12260), viral envelope vector (pMD2.G, Addgene, #12259), and the indicated transfer vectors. For Amcf1 overexpression in HeLa cells, LentiORF Amcf1 (DharmaconTM Reagents, OHS5898-202620525) was used. For Amcf1 knockdown in HeLa cells, GIPZ lentiviral shRNA targeting Amcf1 (DharmaconTM Reagents, RHS4430-200177847) was used. For doxycycline (Dox)-inducible Amcf1 overexpression in MEFs, pLV[Exp]-Bsd-TRE3G and pLV[Exp]-CMV>Tet3G/Hygro (VectorBuilder) were used. To generate lentivirus, HEK293FT cells were cultured in 10-cm dishes and transfected with 10 µg lentiviral plasmid, 5 µg psPAX2 and 2.5 µg pMD2.G using LipofectamineTM 3000 (Thermo Fisher Scientific). Seventy-two hours after transfection, viral supernatants were collected, filtered through 0.45-μm filters, and concentrated by ultracentrifugation (Beckman Optima XL-100K Ultracentrifuge) at 85,000 × g for 2 h at 4 °C. Virus pellets were resuspended in cold DMEM/F12 for cell transduction. Generation of stable cell lines Cells were seeded in 6-well plates and incubated overnight. The following day, cells were pretreated with polybrene (5 µg mL–1; Santa Cruz Technology, sc-134220) for 1 h prior to lentivirus infection. Forty-eight hours after infection, cells were subjected to antibiotic selection for 2 weeks. For Amcf1 overexpression and corresponding control lines, cells were selected with blasticidin (5 µg mL⁻¹; InvivoGen, ant-bl-05). For Amcf1 knockdown and control lines, puromycin (2 µg mL⁻¹; InvivoGen, ant-pr-1) was used. For doxycycline-inducible Amcf1 overexpression, cells were selected with a combination of blasticidin (5 µg mL⁻¹) and hygromycin B (200 µg mL⁻¹; InvivoGen, ant-hg-1). Successfully infected, antibiotic-resistant populations were expanded and subsequently maintained in culture medium containing half of the original antibiotic concentrations. Amcf1 overexpression, knockdown, and inducible expression in these stable cell lines were validated by immunoblotting. Generation of MEFs Mouse embryonic fibroblasts (MEFs) were isolated from embryos at embryonic day 13.5 (E13.5). Amcf1 heterozygous mice were used as breeding pairs. The presence of a copulation plug the following morning was designated as embryonic day 0.5 (E0.5), and males were removed thereafter. At E13.5, pregnant females were euthanized, and embryos were harvested. Embryos were carefully dissected from the uterus, washed in PBS, and heads and internal organs (including heart and liver) were removed; heads were used for genotyping. The remaining tissues were transferred to 0.25% trypsin-EDTA, minced with a scalpel, and incubated at 37 °C for 10 min, followed by trituration and an additional 10 min incubation. The resulting cell suspension was transferred to a 50 mL tube containing 20 mL of DMEM supplemented with 10% FBS and penicillin (100 U mL−1) and streptomycin (100 µg mL−1). After allowing debris to settle for ~5 min, the supernatant was transferred to T75 flasks. Medium was replaced the following day to remove dead cells and debris. Cells were passaged upon reaching confluence. Quantitative real-time PCR For cell samples, total RNA was extracted using the FastPure Cell/Tissue Total RNA Isolation Kit (Vazyme, RC112). Reverse transcription was performed using the HiScript III All-in-one RT SuperMix (Vazyme, R333). Quantitative PCR was carried out using Taq Pro Universal SYBR qPCR Master Mix (Vazyme, Q712) on a 7900HT Fast Real-Time PCR System (Applied BiosystemsTM, 4351405). For tissue samples, snap-frozen specimens were ground in pre-chilled mortars (Sigma, BAH372600000) filled with liquid nitrogen and surrounded by liquid nitrogen to maintain cryogenic conditions. Relative gene expression levels were normalized to Actb and calculated using the 2−ΔΔCt method. Primers sequences are provided in Supplementary Table 1. SDS-polyacrylamide gel electrophoresis (PAGE) and immunoblotting Cells were lysed in RIPA buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM EDTA) supplemented with a protease inhibitor cocktail (Thermo Fisher Scientific, A32965) and phosphatase inhibitor cocktail (Thermo Fisher Scientific, A32957). Lysates were centrifuged at 18,000 × g for 20 min at 4 °C, and the supernatants were collected for analysis. For tissue samples, snap-frozen tissues were ground using pre-chilled mortars filled with liquid nitrogen, followed by lysis using the same protocol described above. Protein concentrations were determined using the bicinchoninic acid (BCA) assay (Thermo Fisher Scientific, 23225). Lysates were mixed with Laemmli LDS sample buffer (Thermo Fisher Scientific, J61337) and boiled at 95 °C for 10 min. For the detection of OXPHOS, samples were prepared at room temperature to preserve antigen integrity. Proteins were separated by SDS-PAGE and transferred onto membranes using the Trans-Blot Turbo RTA Transfer Kit (Bio-Rad, 1704274) and the Trans-Blot Turbo Transfer System (Bio-Rad, 1704150), following the manufacturer’s instructions. Membranes were blocked with EveryBlot Blocking Buffer (Bio-Rad, 12010020) for 5 min at room temperature, then incubated overnight at 4 °C with primary antibodies diluted in TBS containing 0.05% Tween (TBST), 5% Bovine Serum Albumin (BSA), and 0.02% NaN3. After washing with TBST, membranes were incubated with HRP-conjugated secondary antibodies at room temperature for 1 h. Immunoreactive bands were detected using enhanced chemiluminescence (ECL) reagents (Thermo Fisher Scientific, 32106) or SuperSignalTM West Pico PLUS Chemiluminescent Substrates (Thermo Fisher Scientific, 34578) and visualized on X-ray film. Analysis of autophagy in vivo Amcf1KI/KI and Amcf1−/− mice were crossed with RFP-EGFP-LC3 transgenic reporter mice to enable analysis of autophagic flux in vivo, respectively. Tissues from the resulting offspring were used for autophagy analyses. To assess autophagy levels across tissues, eight- to twelve-week-old male mice Amcf1 KI (Amcf1KI/KI; RFP-EGFP-LC3) and littermate WT (Amcf1WT/WT; RFP-EGFP-LC3), or Amcf1 KO (Amcf1−/−; RFP-EGFP-LC3) and corresponding WT (Amcf1+/+; RFP-EGFP-LC3) were treated with chloroquine (CQ, 50 mg kg−1) or PBS for 6 h to assess autophagic flux, as previously described17. In parallel experiments, mice were fasted for 24 h with free access to drinking water to induce autophagy. For starvation-induced autophagy analysis, as previously described17, mice were fasted overnight and subsequently re-fed for 3 h prior to tissue collection to minimize variability due to differences in food intake. Measuring oxygen consumption rate (OCR) Cells were counted using a TC20 Automated Cell Counter (Bio-Rad, 1450101) and seeded into Seahorse XF24 cell culture microplates (Agilent, 102342-100) at a density of 4.5 × 104 cells per well for MEFs or 3.0 × 104 cells per well for HeLa cells. Prior to analysis, cells were washed twice with Seahorse XF assay medium (Agilent, 103680-100) supplemented with 1 mM pyruvate, 2 mM L-glutamine, and 10 mM D-glucose, and 500 μl of supplemented medium was added to each well. Plates were incubated for 1 h at 37 °C in a non-CO2 chamber before analysis using a Seahorse XFe24 Analyzer (Agilent). OCR was measured using the Seahorse XF cell Mito Stress Test Kit (Agilent, 103015-100). For HeLa cells, 1.5 µM oligomycin was used; for MEFs, 2.5 µM oligomycin was applied. In both cell types, 1 µM FCCP and 0.5 µM rotenone/antimycin A were sequentially added. A total of three OCR measurements were recorded following the addition of each compound. Data were collected using Wave software (v.2.6.1) and experiments were repeated at least three times unless otherwise specified. OCR values were normalized to cell number per well, determined by post-assay nuclear staining and imaging using the Cytation 5 system (BioTek). Confocal imaging Images were acquired using a Nikon AX-R Confocal Laser Microscope System (Nikon Instruments, Melville, NY) equipped with a 60 × 1.42 numerical aperture (NA) oil-immersion objective, unless otherwise indicated. For cell samples, cells cultured on glass coverslips were fixed at room temperature with 4% paraformaldehyde (Thermo Fischer scientific, AAJ61899AK) for 20 min. After three washes with 1× phosphate-buffered saline (PBS; Corning, 46–013-CM), cells were permeabilized with 0.2% Triton X-100 (Sigma-Aldrich, 93,443) or 100 μM digitonin (MedChemExpress, HY-N4000) for 20 min, washed, and then blocked in 5% bovine serum albumin in PBS for 1 h at room temperature. Cells were then incubated overnight at 4 °C in a humidified chamber with primary antibodies in blocking solution. After three washes with PBS, secondary antibody (1:200 dilution) was applied for 1 h at 37 °C in blocking solution. Coverslips were mounted onto glass slides (Epredia, 4941WFX-003) using ProLongTM Diamond Antifade Mountant with DAPI (Invitrogen, P36966). Tissue samples for imaging were prepared as described previously44. Briefly, mice were euthanized by CO2 asphyxiation, and tissues were collected and fixed in 4% paraformaldehyde (PFA) in 0.1 M sodium phosphate buffer (pH 7.4) for overnight. Samples were then incubated in 15% sucrose in PBS for 4 h, followed by incubation in 30% sucrose until tissues fully sank. Samples were embedded in Tissue-Tek O.C.T Compound (Fisher Scientific, 23-730-571), quickly frozen on dry ice, and stored at −80 °C. Sections (7 µm) were cut using a cryostat (Leica CM3050S) and mounted onto SuperFrost Plus slides (Fisher Scientific, 1518848). Coverslips were mounted using ProLong Diamond Antifade Mountant (Invitrogen P36961). For analysis of LC3 puncta, the “Top Hat” algorithm in Amira software (version 2024.2, Thermo Fisher Scientific) was used to enhance individual dot signal extraction45. To analyze the colocalization between AMC-F1 and TOMM20, plasmids encoding GFP-AMC-F1 or GFP control were transfected into Amcf1−/− MEFs. To assess whether AMC-F1 is recruited to mitochondria, its co-localization with TOMM20 was quantified using Mander’s Colocalization Coefficient (MCC)46, as implemented in the JACoP plugin47. To minimize bias, Costes automatic thresholding was applied. For imaging mitochondrial morphology, seven z-stacks (each 0.2 µm thick) were acquired using a 60× oil-immersion objective with a 2048 × 2048-pixel resolution in resonant scanning mode. Images were then compiled using maximum intensity projections (MIPs) in Nikon NIS-Elements software. Mitochondrial 3D imaging Mitochondrial three-dimensional (3D) morphology was visualized using a Plan-Apochromat 63×/1.40 Oil DIC M27 objective on a Zeiss LSM980 confocal microsope equipped with Airyscan-2. Z-stacks spanning the entire cell were acquired at a sampling rate finer than the Nyquist criterion (step size 0.08 μm vs. software-recommended optimal of 0.17 μm) to achieve improved accuracy of 3D reconstructions. Raw image data were then processed using the Airyscan Processing module (3D Standard) in ZEN Blue software (version 3.9, Zeiss). Processed Airyscan images were analyzed using Amira software (version 2025.1.1; Thermo Fisher Scientific). Briefly, all images were passed through a ‘Non-Local Means’ filter’ module to remove noise. For visualizations, the denoised images were first passed through an ‘Unsharp Masking’ filter. The resulting images were then visualized using a ‘Volume Rendering’ module (‘Volume Rendering’ in figures), or a ‘Generate Surface’ module (‘Surface View’ in figures). For quantification, the denoised images were sequentially processed with ‘Nagao’ filter module followed by a ‘Structure Enhancement’ filter module to enhance and extract the boundary of the mitochondria, which was then binarized using an ‘Interactive Top-Hat’ module. Separate mitochondria were identified using ‘Connected Components’ module. The 3D-Feret length and the 3D Feret breadth of each segmented mitochondria were measured, and the ratio of these parameters (Feret L/B) was calculated to quantify mitochondrial aspect ratio. A Feret L/B value approaching 1 indicates shortened mitochondria, whereas higher values correspond to elongated mitochondrial morphologies. Transmission electron microscopy (TEM) MEFs were seeded in 6-well plates at an approximately 80% confluency and incubated overnight. Cells were then cultured in complete medium or starvation medium (EBSS) for 2.5 h. Cells were fixed in fixative (15960-01, Electron Microscopy Sciences) and stored at 4 °C until for further processing. Sample processing and imaging were performed at the MD Anderson High Resolution Electron Microscopy Facility. In detail, the samples were then washed in 0.1 M cacodylate buffer. Post-fixation steps included treatment with 0.1% Millipore-filtered cacodylate buffered tannic acid, followed by 1% buffered osmium tetroxide, and en bloc staining with 0.1% Millipore-filtered uranyl acetate. Samples were dehydrated through a graded ethanol series, infiltrated, and embedded in LX-112 resin. Polymerization was carried out at 60 °C for ~3 days. Ultrathin sections were prepared using a Leica Ultracut microtome (Leica, Deerfield, IL), then stained with uranyl acetate and lead citrate in a Leica EM Stainer. Sections were examined using a JEM-1010 transmission electron microscope (JEOL USA, Inc., Peabody, MA) at an accelerating voltage of 80 kV. Digital images were acquired using an AMT imaging system (Advanced Microscopy Techniques Corp., Danvers, MA). Images were acquired at three magnifications (15,000×, 25,000×, 50,000×). For quantification of mitochondrial morphology, including length and area, scale calibration was applied, and measurements were performed using ImageJ software (v1.54 f)48. Mitochondrial Complex I activity Mitochondrial Complex I activity was measured using the Complex I Enzyme Activity Assay Kit (Abcam, ab109721) according to the manufacturer’s instructions. Briefly, adherent cells were scraped from confluent culture flasks and washed with PBS, or mouse tissue (100–200 mg) was prepared as described above. Protein concentrations were determined using a BCA assay on a small aliquot of each sample, and the remaining samples were adjusted to 5.5 mg/mL with PBS. Proteins were then extracted by adding 1/10 volume of detergent solution to achieve a final concentration of 5 mg/mL, followed by centrifugation at 16,000 × g for 20 min at 4 °C. Samples were further diluted to the recommended concentration, and 200 µL of each sample was loaded into the wells of the assay plate. The plates were incubated at room temperature for 3 h, followed by washing. The assay solution was then added, and the absorbance was recorded at 450 nm in kinetic mode. Enzymatic activity was calculated as the change in absorbance per minute per amount of sample loaded. Fas-induced apoptosis Four-month-old male mice of the indicated genotypes were used for this study. Mice were injected intraperitoneally (i.p.) with anti-Fas antibody (Jo2, 0.25 µg g−1 body weight, in sterile saline; BD Pharmingen) or sterile saline solution as a control. Mice were euthanized at the indicated time points post-injection in accordance with institutional ethical guidelines. Liver tissues were immediately processed for histological evaluation and apoptosis detection, while blood samples were used for serum analysis of liver enzymes. Liver H&E staining and TUNEL assay Livers were dissected and cut into small fragments, then immediately fixed overnight at 4 °C in 4% PFA in PBS. Fixed tissues were dehydrated through a graded ethanol series, cleared with xylene, and embedded in paraffin. Paraffin-embedded liver sections (5 μm) were mounted onto glass slides for subsequent analysis. Sections were deparaffinized and rehydrated through a graded ethanol series prior to staining. For histological assessment, sections were stained with hematoxylin and eosin (H&E) to evaluate tissue architecture and pathological alterations. Apoptosis was assessed by TUNEL assay using ApopTag Peroxidase In Situ Apoptosis Detection Kit (Sigma-Aldrich, S7101), following the manufacturer’s instructions. Sections were treated with terminal deoxynucleotidyl transferase (TdT) and labeled nucleotides to incorporate dUTPs at sites of DNA fragmentation. Apoptotic cells were visualized by peroxidase-based chromogenic detection, and nuclei were counterstained with methyl green. Serum ALT and AST analysis Blood samples were collected from mice at two time points: baseline (prior to Jo2 injection) and indicated time points post-injection. Samples were allowed to clot at room temperature and then centrifuged at 4000 × g for 10 min at 4 °C to isolate serum. Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) activities were measured using the Pointe Scientific assay kit (Pointe Scientific, A7526-150, A7561-150). Absorbance was measured at 340 nm, and enzyme activity was expressed in international units per liter (IU/L). Isolation of primary hepatocytes Primary hepatocytes were isolated from 10-week-old male mice as previously described49. Mice were euthanized by CO2 asphyxiation, and the liver was perfused with 25 mL Liver Perfusion Medium (Thermo Fisher Scientific, 17701038), followed by 25 mL Liver Digest Medium (Thermo Fisher Scientific, 17703034). The liver was then transferred to a tissue culture dish with cold DMEM. Hepatocytes were released by gentle dissociation using a cell lifter (Corning, CLS3008) and passed through a 70-μm cell strainer (Corning, CLS431751-50EA) to remove tissue debris. The resulting cell suspension was centrifuged at 50 × g for 3 min at 4 °C, and then the pellet was resuspended in 12.5 mL DMEM mixed with 12.5 mL Percoll buffer (prepared by diluting Percoll with 10× PBS at a 9:1 ratio; Cytiva, 17544502). Cells were further purified by centrifugation at 100 × g for 5 min at 4 °C. The supernatant containing dead cells was removed by aspiration. The pellet was resuspended in cold PBS, and cell number and viability were determined. RNA-seq library preparation and sequencing Total RNA was extracted using the RNeasy Mini Kit (Qiagen, 74104) with on-column DNase digestion (RNase-Free DNase Set; Qiagen, 79254) according to the manufacturer’s protocol. Samples with RNA integrity number (RIN) > 7 were used for library preparation. Poly(A) + RNA was isolated using the NEBNext Ploy(A) mRNA Magnetic Isolation Module (New England Biolabs, E7490L) and libraries were constructed using the NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (New England Biolabs, E7760L) along with NEBNext Multiplex Oligos for Illumina (New England Biolabs, E6609S), following the manufacturer’s instructions. Library quality was assessed using the Agilent High Sensitive DNA Kit (Agilent, 5067-4626) on a Bioanalyzer 2100 (Agilent Technologies, Santa Clara, USA), and concentrations were quantified by qPCR using Collibri Library Quantification kit (Thermo Fisher Scientific, A38524500). Equimolar libraries were pooled and subjected to paired-end 150-cycle sequencing on an Illumina NovaSeq X Plus platform (Illumina, Inc., USA). RNA-seq libraries were prepared and sequenced by the Cancer Genomics Center at The University of Texas at Houston. RNA-seq processing and analysis RNA-Seq data were processed using the Galaxy platform (https://usegalaxy.org/). Quality control was performed using FastQC (v0.11.8), followed by adapter and low-quality base trimming with Trim Galore (v0.4.5)50. High-quality reads were aligned to the Mus musculus reference genome (GRCm38/mm10) using HISAT2 (v2.1.0)51. Gene-level quantification was performed using the FeatureCounts function from the Subread package (v1.6.4)52. Differential expression analysis was carried out using DESeq2 package (v1.22.2)53. Gene ontology (GO) enrichment or pathway analysis were performed using the list of significant differentially expressed genes (DEGs). Heatmaps of gene expression across genotypes and conditions were generated using the pheatmap package (http://github.com/raivokolde/pheatmap). The list of mouse mitochondrial complex subunits was obtained from MitoCarta3.0 database. Gene set enrichment analysis (GSEA) was performed using the GSEA software (v4.2.1)54 with either MSigDB hallmark gene sets55 or curated lists, including MYC-related mitogenesis genes and ISR genes56. Clinical data acquisition and analysis across 33 tumor types from TCGA cohorts mRNA expression datasets and corresponding clinical data (including outcome endpoints, tumor stage, etc.) from 33 tumor types were obtained from The Cancer Genome Atlas (TCGA) via the NCI Cancer Genomic Data Commons (GDC: http://gdc.cancer.gov) using the GDCquery function from the R package (v4.3.1) TCGAbiolinks. Clinical outcome endpoints, including overall survival (OS), progression-free interval (PFI), disease-specific survival (DSS), and disease-free interval (DFI), were analyzed using the R package survival (v3.5.8). Kaplan-Meier survival curves were generated using the suvfit function to compare outcomes between groups stratified by AMCF1 expression. Optimal expression cutpoints were determined using the suv_cutpoint function and applied to stratify patients. Statistical significance between survival curves was assessed using the log-rank test (survdiff) as implemented in the R package survminer (v0.4.9). Statistical analysis Statistical analyses were performed using GraphPad Prism software (version 10.0.02). Sample sizes were not predetermined by statistical methods. All experiments were conducted with at least three independent biological replicates unless otherwise indicated in the figure legends. No randomization was performed, as experimental group were defined by genotype using age-matched mice. For comparisons between two groups, unpaired two-tailed Student’s t test were used. Comparisons among multiple groups were performed using one-way or two-way ANOVA followed by Tukey’s multiple-comparison test, unless otherwise specified. Survival analyses were conducted using two-sided log-rank tests. P values are reported where applicable; a 95% confidence interval was used to assess statistical significance, and P < 0.05 was considered statistically significant. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Reporting Summary Transparent Peer Review file Supplementary information Source data Source Data Acknowledgements We thank the following core facilities at the Texas Medical Center for their support: the Cancer Genomics Center at the UTHealth (CPRIT RP240610), the UTHealth Center for Advanced Microscopy, the UTHealth Histopathology Core, and the High Resolution Electron Microscopy Facility at MD Anderson Cancer Center. Mitochondrial 3D imaging was done using resources of the Shared Equipment Authority at Rice University (https://research.rice.edu/sea/). We thank Dr. Alloysius J. Budi Utama for technical help with the Zeiss LSM980 Airyscan-2 confocal microscope. This work was supported by the National Institute of Health (NIH/NCI) grant R01CA181319 (to N.M.) and Annie and Bob Graham Endowment (to N.M). Author contributions N.M. initiated, designed and supervised the study and wrote the manuscript with contributions from the co-authors. Y.W., R.R., T.V. and N.M. analyzed and interpreted the data. Y.W., R.R., and A.S. performed mitochondrial TEM experiments and interpreted the data. Z.M. and Y.W. performed mitochondrial 3D imaging and interpreted the data. Y.W. and R.R. performed autophagy analyses and Fas-induced liver injury studies. Y.W. analyzed RNA-seq data. Peer review Peer review information Nature Communications thanks Takeshi Noda, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Data availability RNA sequencing data (raw fastq files) and sample metadata have been deposited to the NCBI Sequence Read Archive under BioProject accession code PRJNA1446969. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-73841-3. References Associated Data Supplementary Materials Reporting Summary Transparent Peer Review file Supplementary information Source Data Data Availability Statement RNA sequencing data (raw fastq files) and sample metadata have been deposited to the NCBI Sequence Read Archive under BioProject accession code PRJNA1446969. Source data are provided with this paper.