ERAD deficiency promotes mitochondrial dysfunction and transcriptional rewiring in human hepatic cells

Mitochondrial dysfunction is associated with a variety of human diseases including neurodegeneration, diabetes, nonalcohol fatty liver disease (NAFLD), and cancer, but its underlying causes are incompletely understood. Using the human hepatic cell line HepG2 as a model, we show here that endoplasmic reticulum-associated degradation (ERAD), an ER protein quality control process, is critically required for mitochondrial function in mammalian cells. Pharmacological inhibition or genetic ablation of key proteins involved in ERAD increased cell death under both basal conditions and in response to proinflammatory cytokines, a situation frequently found in NAFLD. Decreased viability of ERAD-deficient HepG2 cells was traced to impaired mitochondrial functions including reduced ATP production, enhanced reactive oxygen species (ROS) accumulation, and increased mitochondrial outer membrane permeability. Transcriptome profiling revealed widespread down-regulation of genes underpinning mitochondrial functions, and up-regulation of genes associated with tumor growth and aggression. These results highlight a critical role for ERAD in maintaining mitochondrial functional and structural integrity and raise the possibility of improving cellular and organismal mitochondrial function via enhancing cellular ERAD capacity.

Mitochondrial dysfunction is associated with a variety of human diseases including neurodegeneration, diabetes, nonalcohol fatty liver disease (NAFLD), and cancer, but its underlying causes are incompletely understood. Using the human hepatic cell line HepG2 as a model, we show here that endoplasmic reticulum-associated degradation (ERAD), an ER protein quality control process, is critically required for mitochondrial function in mammalian cells. Pharmacological inhibition or genetic ablation of key proteins involved in ERAD increased cell death under both basal conditions and in response to proinflammatory cytokines, a situation frequently found in NAFLD. Decreased viability of ERAD-deficient HepG2 cells was traced to impaired mitochondrial functions including reduced ATP production, enhanced reactive oxygen species (ROS) accumulation, and increased mitochondrial outer membrane permeability. Transcriptome profiling revealed widespread down-regulation of genes underpinning mitochondrial functions, and up-regulation of genes associated with tumor growth and aggression. These results highlight a critical role for ERAD in maintaining mitochondrial functional and structural integrity and raise the possibility of improving cellular and organismal mitochondrial function via enhancing cellular ERAD capacity.
The endoplasmic reticulum (ER) serves several critical cellular functions, including Ca 21 storage, lipid synthesis, and the folding of protein molecules destined for the secretory pathway. Impaired ER function or ER stress has been implicated in an increasing number of chronic human diseases, including obesity, diabetes, neurodegeneration, nonalcohol fatty liver disease (NAFLD), and cancer (1)(2)(3). ER stress is an abnormal homeotic condition characterized by the accumulation of unfolded or misfolded proteins in the ER lumen (4). The classical view is that ER stress activates the unfolded protein response (UPR), which is coordinated by the ER stress sensors, PERK, IRE1a, and ATF6 (5). Activation of these sensors reduces protein translation and enhances protein folding and ER-associated degradation (ERAD).
Failure of these actions to restore homeostasis results in sustained ER stress, activation of the CCAAT-enhancer-binding protein homologous protein (CHOP) pathway and apoptosis (6).
The ER is a well-characterized "functional partner" of many organelles and dysregulation of these interactions may impair cellular function (7). The interactions between the ER and mitochondria, an organelle essential for life and death of eukaryotic cells (8)(9)(10), provide a good example of the importance of this partnership. The ER and mitochondria form tight physical contacts at the mitochondria-associated membranes (MAMs) to regulate several key cellular processes, including Ca 21 homeostasis, lipid exchange, steroid biosynthesis, and apoptosis (11). The MAMs were also found to function as a signaling hub for mitochondria fission as well as autophagy, key events involved in mitochondrial quality control (12). Over the last few years, biochemical and genetic approaches have provided evidence how these two organelles interact physically and functionally to coordinate various cellular functions (13,14). However, less is known about the potential pathophysiological consequences of an impaired UPR pathway in the ER on mitochondria function and vice versa.
We have recently reported that pharmacological and genetic disruption of ERAD, an ER protein quality control process, impairs glucose-stimulated insulin secretion in cultured rat pancreatic b-cell line and in mice (15). By utilizing an ERAD-deficient hepatic cell line model, we demonstrate here that ERAD deficiency causes structural and functional damages to mitochondria. Consequently, ERAD-deficient cells show impaired ATP production, increased mitochondrial outer membrane permeability (MOMP), and enhanced sensitivity to tumor necrosis factor a (TNFa)-induced cell death. These consequences of ERAD deficiency were associated with an altered gene expression profile characterized by the selective down-regulation of mitochondria-related and up-regulation of tumor growth and aggression-related genes. These results indicate that ERAD is critically required for maintaining normal mitochondrial function and ERAD deficiency may potential mammalian cells to adopt a tumorigenic molecular signature by activating the mitochondrial retrograde signaling pathway.

Pharmacological inhibition of ERAD perturbs HepG2 cell mitochondrial function and reduces cell viability
We recently showed that ERAD disruption reduces ATP production in rat pancreatic b-cell line INS-1, indicating that ERAD deficiency impairs mitochondrial respiratory function (15). To determine whether this phenomena extends to other mammalian cells, we treated human hepatic HepG2 cells with the ERAD inhibitor eeyarestatin I (EerI). EerI treatment caused dose-and time-dependent decreases in ATP levels (Fig. 1, A and B) and cell viability (Fig. 1, C and D). These dose-and time-dependent EerI effects were associated with reciprocal increases in mitochondrial levels of calcium (Ca 21 ) (Fig. 1, E and F) and ROS (Fig. 1, G and H).
Next we tested the roles of increased of mitochondrial Ca 21 and ROS in mediating the effects of EerI on ATP levels and cell viability. HepG2 cells were treated with EerI in the absence or presence of the mitochondrial Ca 21 uptake inhibitor ruthenium red (RR) and the ROS scavenger N-acetylcysteine (NAC). As expected, the stimulatory effects of EerI were reversed by RR in the case of mitochondrial Ca 21 and by both RR and NAC in the case of ROS (Fig. 1, I and J). More importantly, both RR and NAC attenuated the negative effects of EerI on ATP levels and cell viability (Fig. 1, K and L). These results indicate that an effective ERAD is required to maintain normal mitochondrial function and cell viability in mammalian cells.
Genetic ablation of the ERAD core protein SEL1L sensitizes HepG2 cells to pro-inflammatory cytokine-induced cell death Mammalian SEL1L is essential for the formation of the ERAD complex on the ER membrane. To further assess the role of ERAD for normal mitochondrial function, we used CRISPR/Cas9-based gene editing to generate 3 independent SEL1L-deficient (Sel1l 2/2 ) HepG2 cell lines (Fig. 2, A and B). Although the SEL1L-deficient cell lines were viable and morphologically undistinguishable from the parental HepG2 cell line, they showed markedly higher rate of cell death during long-term culture as determined by real-time monitoring of propidium iodide (PI) staining (Fig. 2C). HepG2 cells were seeded into 6-well-plates at an initial density of 1 3 10 5 /well and allowed to recover for 12 h before treating with EerI and other chemical reagents (RR, 2-APB, and NAC). Following the treatments, HepG2 cells were analyzed for: A and B, cellular ATP concentration; C and D, cell viability; E and F, mitochondrial Ca 21 ; and G and H, mitochondrial ROS levels. I-L, antagonizing effect of RR or NAC on EerI-induced increase of mitochondrial Ca 21 (I) and ROS (J); and decrease of ATP production (K), and cell viability (L). All data are mean 6 S.E. (n = 3). *, p , 0.05; **, p , 0.01 by Student's t test.
A number of pro-inflammatory cytokines promote cell death in diseases affecting the liver and other tissues. To assess whether ERAD deficiency exacerbates cytokine-mediated cell death, WT and SEL1L-deficient HepG2 cells were treated with the pro-inflammatory cytokines TNFa, IL-1a, IL-6, and IL-12) and followed by an assessment of cell viability by the MTT assay. SEL1L-deficient HepG2 cells suffered a higher level of cell death than WT HepG2 cells when incubated with doses as low as 10 ng/ml of either TNFa or IL-12 (Fig. 2, D and E, respectively). Incubation with both TNF-a and IL-12 did not yield a synergistic effect (Fig. 2F). In contrast, no difference in viability was observed between WT and SEL1L-deficient cells incubated with either IL-1a or IL-6 and IL-22 (Fig. S1).
Next we investigated possible mechanisms accounting for the increased susceptibility of SEL1L-deficient cells to the cell death promoting effects of TNFa and IL-12. We first monitored fluorescence in cells transfected with the ER stress reporter plasmid pCAX-F-XBPDBD-venus. As we have previously shown in other cells (15), SEL1L deficiency increased ER stress; more importantly, TNFa and IL-12 did not increase fluorescence in either WT or SEL1L-deficient cells (Fig. 2, G and H, respectively), ruling out a role of ER stress in mediating the effects of TNFa or IL-12. In contrast, incubation with the necrosis inhibitor necrostatin-1 (Nec-1) partially abolished the increased susceptibility of Sel1l 2/2 cells to the death-promoting effects of TNFa (Fig. 2I). Finally we used annexin V (FITC) and propidium iodide (PI) double dye-staining followed by FACS to ask whether NAC and Ru360 could correct the increased susceptibility of Sel1l 2/2 cells to TNFa. As shown in Fig. 2J, TNFa incubation induced a higher percentage of FITC/ PE dual positive cells in Sel1l 2/2 than in WT population; more importantly, NAC or Ru360 significantly attenuated the deathpromoting effects of TNFa. Overall, these results indicate that ERAD-deficient cells are more susceptible to the necrosis-promoting effects of TNFa and a portion of this increased susceptibility is accounted for by increased mitochondrial Ca 21 and ROS levels.

ERAD deficiency impairs mitochondrial morphology and cellular bioenergetics of HepG2 cells
Mitochondria were imaged using a green fluorescent reporter targeted to the mitochondrial outer membrane (15). Mitochondria of EerI-treated and SEL1L-deficient cells appeared more rounded than those in control cells (Fig. S2). To further determine the morphological changes of mitochondria induced by ERAD deficiency, we performed electron microscopic (EM) analysis for Sel1L 2/2 cells. Mitochondria in Sel1L 2/2 cells appeared more scarcely distributed with round and smaller shapes, compared with those in Sel1L 1/1 cells (Fig. 3

, A-D).
There were clear signs of mitochondrial fission (arrowheads in Fig. 3D) in Sel1L 2/2 cells. To assess whether these morphological changes were associated with altered function, we stained cells with the lipophilic cationic fluorescent dye JC-1 followed by dual channel FACS analysis. The ratio of high to low (H/L) fluorescence signals was substantially lower in Sel1L-deficient than in WT HepG2 cells indicating a reduction of mitochondrial membrane potential (MMP) in Sel1l 2/2 cells (Fig. S3A).
We next used Seahorse XF analysis to assess the effects of ERAD deficiency on cellular energetics. Sel1L-deficient cells showed an altered oxygen consumption rate (OCR) profile (Fig. S3B). The basal and maximal respiration rate, proton leak, and ATP production were significantly reduced (Fig. 3, E-H), whereas the spare respiratory capacity and nonmitochondrial respiration rate were not altered (Fig. 3, I and J). Conversely, extracellular acidification rate (ECAR), an index of glycolysis, did not differ between control and Sel1l 2/2 cells (Fig. S3C). Overall, these results suggest that ERAD-deficient cells suffer from a reduction of oxidative phosphorylation (OXPHOS), and as a consequence, reduced mitochondrial membrane potential.

ERAD deficiency promotes MOMP and sensitizes cells to proinflammatory cytokine-induced cell death
Next, we analyzed the effect of ERAD deficiency and TNF-a on the abundance and intracellular distributions of proteins involved in necrosis and apoptosis. Under basal conditions, whereas there was no difference between control and Sel1L-deficient HepG2 cells in the abundance total c-Jun N-terminal kinase (JNK), total cytochrome c, and receptor-interacting serine/threonine protein kinase 1 (RIPK1), but the latter had higher levels of phosphorylated p54 JNK (p54/p-JNK) and cytoplasmic cytochrome c (c-cytoochrome c) (Fig. 4, A, left panels, and B and C). Upon incubation with TNFa, p54/p-JNK, and Ccytochrome c contents in either cell types, but to a greater extent in Sel1L-deficient than in WT cells (Fig. 4, A, middle panels, and B and C), and this effect was attenuated by co-incubation with the JNK inhibitor SP600125 (Fig. 4, A, right panels, and B and C). Given the increased level of cytochrome c in the cytoplasm, we measured the activity of caspase 3 and 8. Consistent with the Western blotting result showing increased cytoplasmic content of cytochrome c, enzymatic analysis showed increased caspase 3 but not caspase 8 activity in Sel1l 2/2 HepG2 cells, and this pattern was exacerbated by treatment with TNFa (Fig. 4, D and E). Together, these data indicate that ERAD deficiency promotes MOMP leading to increased release of cytochrome c, and this process is exacerbated in the presence of the proinflammatory cytokine TNFa. ERAD deficiency causes transcriptional rewiring that disrupts multiple mitochondrial activities but activates several pathways related to "stemness" property of cells Finally, we used RNA-Seq on WT and ERAD-deficient cells to obtain a global view of the consequences of ERAD deficiency on gene expression. This analysis identified 1216 differentially expressed genes (DEGs), including 199 up-regulated and 1017 down-regulated genes. By gene enrichment analysis for the Gene Ontology biological process and biological pathway, the down-regulated DEGs were found to be significantly enriched in the functional categories including mitochondrial structure and function (mitochondrial ATP synthesis coupled electron transport, mitochondrion organization, mitochondrial respiratory chain complex assembly, mitochondrial translation, and mitochondrial transmembrane transport), endoplasmic reticulum function (protein targeting to ER and establishment of  protein localization to endoplasmic reticulum), and metabolic process related to mitochondria or endoplasmic reticulum (oxidative phosphorylation, respiratory electron transport, oxi-dation2reduction process, and metabolic pathways) (Fig. 5, A,  left, and B). The up-regulated genes were overrepresented in epithelial cell adhesion, P53 signaling, MAPK signaling, mitochondrial structure and function (mitochondrial fragmentation, negative regulation of mitochondrial Ca 21 concentration, regulation of Ca 21 import, Ca 21 channels), endoplasmic reticulum function (protein exit from the ER, protein processing in the ER, unfolded protein response) (Fig. 5, A, right, and C). Overall, these results indicate that ERAD deficiency caused a global reorganization of mitochondrial activities.
To further understand the pathophysiological significance of the obtained molecular signature in ERAD-deficient HepG2 cells, we performed gene set enrichment analysis (GSEA) by comparing the identified DEGs to a priori-established gene modules. As expected, genes of a reported mitochondrial gene module (16) were overrepresented in the down-regulated DEGs in ERAD-deficient HepG2 cells (Fig. 5D). Intriguingly, the Jak2 signaling pathway-associated or the so-called stemness genes (17,18) were highly represented in the up-regulated DEGs (Fig. 5E). Together, these results indicate that ERAD deficiency caused a global reorganization of transcription that disrupts multiple mitochondrial activities and activates several pathways associated with the stemness or invasiveness property of eukaryotic cells.

Discussion
The endoplasmic reticulum and mitochondria are physically and functionally integrated with each other to coordinate an array of metabolic and biosynthetic functions. It has been shown recently that ER protein quality control or ERAD affects mitochondrial dynamics in brown adipose cells by regulating the turnover of the MAM protein, sigma receptor 1 (SigmaR1) (19). Stress or dysfunction in one compartment is likely to affect the other and the overall function status of the cell. Here, we provide evidence that ERAD deficiency impacts mitochondrial function human hepatic cells via influencing their Ca 21 and ROS homeostasis. We show that pharmacological inhibition or genetic disruption of ERAD function decrease the viability of hepatic cell line HepG2 and sensitize them to TNF-ainduced cell death. ERAD deficiency induces Ca 21 overflow from the ER to mitochondria, causing over-production of ROS and subsequent damage to the mitochondrial outer membrane. ERAD-deficient cells had decreased ATP production, reduced mitochondrial transmembrane potential, and increased outer membrane permeability and most notably, transcriptome analysis indicates that ERAD deficiency caused a global reorganization of the transcription program. These results highlight an important role for ERAD in maintaining mitochondrial health and cell survival by, at least partially, supporting mitochondrial Ca 21 and ROS homeostasis.
Increasing evidence links ER stress to the development of chronic human diseases, such as type 2 diabetes mellitus, NAFLD, neurodegeneration, and cancer (1)(2)(3). However, the mechanisms whereby ER stress lead to the pathogenesis of vari-ous human diseases are incompletely understood. ER stress has been shown to activate the unfolded protein response (UPR) responses or the CHOP-dependent cell death program (5,6). However, this current UPR model alone cannot fully explain why ER stress preferentially affects metabolically active organs, such as neurons, the endocrine pancreas and liver. 4 In this study, we show that chronic ER stress due to defective ERAD causes structural and functional damages to mitochondria. Moreover, we demonstrate that blocking mitochondria Ca 21 uptake or removing excessive ROS partially rescues cellular defects of ERAD-deficient cells. Given the essential roles of mitochondria in regulating cellular physiology, metabolism, and cell death (20), our results suggest that mitochondrial dysfunction may be an under-appreciated driver of many ER stressinduced pathogenesis. These results also strongly support an increasingly recognized notion that normal cellular and organismal health depends on tightly regulated inter-organelle interaction and communication, particularly under stressed conditions, such as viral infection or obesity (21).
Our results provide some mechanistic evidence to explain how ERAD deficiency leads to mitochondrial dysfunction. The ER makes contact with mitochondria through the formation of MAMs, which are temporary structures required for exchange of Ca 21 and other molecules between these two organelles (22). We showed that, first, elevated levels of Ca 21 and ROS were detected in mitochondria of EerI-treated and SEL1L-depleted HepG2 cells. Second, blocking mitochondria Ca 21 uptake by treating ERAD-deficient HepG2 cells with Ru360 (a known MCU inhibitor), or removing excessive mitochondrial ROS with NAC (a mitochondria ROS scavenger), partially restores mitochondrial function of ERAD-deficient HepG2 cells. Furthermore, increased cytochrome c, a mitochondrial matrix protein, was detected in the cytoplasmic compartment of ERADdeficient HepG2 cells. These results suggest a model whereby ERAD deficiency triggers overflow of Ca 21 from the ER to mitochondria, which in turn enhances ROS production and progressive damage to the mitochondrial respiratory system and membrane structure.
Cell death is a common histological feature of many chronic diseases, and inflammatory cytokines have been shown to play an important role during this process (23). For example, the pro-inflammatory cytokine TNF-a contributes to the development and progression of inflammation-related diseases, such as nonalcohol steatohepatitis, type 2 diabetes, and cancer, by promoting necrotic cell death (24)(25)(26). Importantly, we show that TNF-a-induced cell death is increased in the nascence of ERAD and that this effect could be blocked with the necrosis inhibitor Nec-1. These results indicate that ERAD deficiency exacerbates the cell death-promoting effect of TNFa. Although further studies are needed to clarify the exact molecular mechanisms underlying the observed synergy between ERAD deficiency and TNF-a on cell death, we speculate that the lower intracellular ATP level, the impaired mitochondrial Ca 21 level and increased mitochondrial outer membrane permeability may be among the contributing factors.
ERAD deficiency had profound effects on gene expression in HepG2 cells. RNA-Seq revealed over 1200 DEGs in ERAD-deficient cells with the majority falling in the category of downregulated genes. Many of those genes are involved in OXPHOS, ATP synthesis, mitochondrial organization, and respiratory chain complex assembly (Fig. 5A). It is possible that down-regulation of these genes are a consequence of the activation of the mitochondrial unfolded protein response (UPR mt ) (27). A major outcome of the activation of the UPR mt pathway is suppression of the transcription of mitochondrial function-associated genes, so that the overall mitochondrial activity would be tuned down to reduce mitochondrial stress (28). However, many mitochondrial organization and respiratory chain complex assembly are also down-regulated in ERAD-deficient HepG2 cells. A possible explanation for the down-regulation of these genes is that the damage of mitochondria in ERAD-deficient cells has reached a threshold point of "no-return," so that a mitochondrial functional collapse had occurred.
Intriguingly, most of the 200 up-regulated genes in Sel1L-deficient cells are associated with either cell adhesion or proliferation. Gene set enrichment analysis showed that these genes overlap significantly with a group of so-called stemness genes in the IL-6/STAT3 signaling pathway (17,18). Although further studies are needed to ascertain whether the ERAD-deficient HepG2 cells have reverted to stem-like or more aggressive cells, these results provide evidence that ERAD deficiency may potentiate mammalian cells toward stem or cancer-like cells, perhaps through influencing their mitochondrial function.
In summary, we demonstrate here that ERAD deficiency decrease the viability of HepG2 cells and sensitizes them to TNFa-induced cell death. ERAD deficiency also enhances the stemness or metastasis property of mammalian cells. Mechanistically, ERAD deficiency induces structural and functional damage of mitochondria by impairing the ability of mitochondrial to maintain Ca 21 and ROS homeostasis. Our findings highlight the importance of a regulated ER-mitochondria interaction in supporting normal cellular and organismal physiology, and have important implications in understanding the molecular mechanisms of chronic diseases, such as T2D, nonalcohol steatohepatitis, neurodegeneration, and cancer.

Materials and methods
Cell culture, plasmid transfection, fluorescent imaging, and CRISPR/Cas9 mediated gene editing HepG2 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 15% fetal bovine serum, 100 units/ ml of penicilin,100 mg/ml of streptomycin at 37°C in an humidified CO 2 incubator. For visualization of mitochondria or monitoring ER stress, the expression plasmids p-mtGFP containing mitochondria-targeted GFP, or pCAX-F-XBPDBD-venus (29) were transfected into HepG2 cells using Lipofectamine 3000. GFP images were acquired with a Fluoview FV1000 Olympus confocal microscope analyzed by the software Image J (NIH Image).

Cell death and viability assay
WT and Sel1L-deficient HepG2 cells were seeded into a 6well-plate at an initial density of 1 3 10 5 /well and allowed to recover for 12 h. PI and annexin V were then added into the culture medium (10 ng/ml) and cell death was monitored in real-time over the next 72 h using an Incucyte Zoom system. Cell viability was assessed by MTT assay or by FACS using a Beckman Counter Flow Cytometer.

ROS, cytosolic and mitochondrial Ca 21 measurement
HepG2 cells were seeded into 4-well glass-bottomed chambers (Nunc, IL) at a density of 1 3 10 5 /well in Dulbecco's modified Eagle's medium and cultured for 24 h. Cytosolic and mitochondrial Ca 21 and intracellular ROS were measured by adding the freshly prepared fluorescent dye Fluo4/AM, Rhod2 AM (Invitrogen), and Mitosox into the culture medium followed by cell imaging with a Fluoview FV1000 Olympus confocal microscope.
Mitochondrial transmembrane potential, OXPHOS, and glycolytic functional analysis Mitochondrial transmembrane potential was determined using a JC-1 mitochondrial membrane potential assay kit (Abcam). Oxidative phosphorylation and glycolysis were determined using a Seahorse XFe24 extracellular flux analyzer (Seahorse Bioscience) according to the manufacturer's protocols and as previously described (31). Briefly, a total of 4 3 10 4 cells were seeded in each well of Seahorse XF24 cell culture microplate and cultured for 16 h in complete medium. The OCR and ECAR were assayed using the XF Cell Mito Stress Test Kit and Figure 5. ERAD deficiency causes overhauling of mitochondrial activities and enhances the stemness property of HepG2 cells. Sel1L 1/1 and Sel1L 2/2 cells were seeded at 1 3 10 5 /well in 6-well-plates and cultured in a humidified CO 2 incubator for 36 h before being subjected to total RNA extraction using the TRIzol kit. Three replicates of RNA sample for each genotype were used for RNA-Seq. A, Gene Ontology biological process (red) and biological pathways enrichment (green) analysis of down-regulated (left) and up-regulated (right) genes in Sel1L 2/2 cells. B and C, representative heatmaps of down-regulated (B) and up-regulated (C) genes in Sel1L 2/2 cells. Colors indicate scaled expression of individual genes. D and E, gene set enrichment plot of down-regulated (D) and up-regulated (E) genes in Sel1L 2/2 cells. The reference gene modules used in D and E are the Wong mitochondria gene module (https://www. gsea-msigdb.org/gsea/msigdb/cards/WONG_MITOCHONDRIA_GENE_MODULE) and the JAK2 pathway module (https://www.gsea-msigdb.org/gsea/ msigdb/cards/JAK2_DN.V1_DN), respectively. Each vertical bar on the x axis in D and E represents a denoted gene in the specific gene set.
XF Glycolysis Stress Test Kit, respectively, coupled with the XF24-3 Extracellular Flux Assay Kit. All OCR and ECAR measurements were normalized to protein content.

RNA-Seq and bioinformatics analyses
Total RNA was isolated from HepG2 cells grown under standard conditions using TRIzol reagent (Invitrogen). RNA-Seq was performed by GENEWIZ Co., Ltd. (Suzhou, China). In brief, libraries were prepared using the VAHTS mRNA-seq V2 Library Prep Kit for Illumina, according to the manufacturers' instructions. Sequencing libraries were run on an Illumina NovaSeq platform with the 23 150-bp paired-end protocol. These RNA-Seq data has been deposited onto public repository of Genome Sequence Archive (GSA) database in BIG Data Center (BioProject: PRJCA002400; GSA accession number: CRA002460).
Raw reads yielded by paired-end transcriptome sequencing were mapped to the human reference genome (GRCh37/hg19) by tophat2 (version 2.1.1) (32). Read counts were obtained from BAM files using samtools (version 1.9) and gfold (version 1.1.4) (33), and then converted into log2 counts/million (logCPM) values to quantify the gene expression level by the R package edgeR (version 3.26.8) (34). Statistical significances of differentially expressed genes were determined by the empirical Bayes method powered by the R package limma (version 3.40.6) (35) with p , 0.005 as the statistical cutoff. Programs written in R language (version 3.6.1) use library extension to pull in functions from noncore packages from the bioconductor repository (RRID:SCR_006442).
ToppGene Suite (RRID:SCR_005726) (36) was used to perform the enrichment analysis of differentially expressed genes for Gene Ontology (GO) biological process (Gene_Ontology_-Consortium, 2015), Reactome pathway (37), and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway (38). GSEA (version 4.0.3) (39) was performed to identify the a priori defined gene set associated with transcriptome differences between phenotype (SEL1L-deficient versus intact WT). The c6.all.v7.0.symbols.gmt (oncogenic signatures) and c2.cgp.v7.0. symbols.gmt (chemical and genetic perturbations) gene set libraries were used as the reference gene set collection for enrichment analysis. The statistical cutoff for this analysis was set at p , 0.05.

Statistical analyses
Data were analyzed by Student's t test or two-way repeated analysis of variance using GraphPad (7.0) with statistical significance set at p , 0.05. Data are present as mean 6 S.E.

Data availability
High-throughput sequencing data for HepG2 cells are stored in the BioProject Library (https://bigd.big.ac.cn/ bioproject/browse/PRJCA002400). All other data are contained within the manuscript.