Research Article

Endoplasmic reticulum–associated degradation regulates mitochondrial dynamics in brown adipocytes

See allHide authors and affiliations

Science  03 Apr 2020:
Vol. 368, Issue 6486, pp. 54-60
DOI: 10.1126/science.aay2494

Organelle cross-talk

Endoplasmic reticulum (ER)–associated degradation (ERAD) is a quality control mechanism that allows for targeted degradation of proteins in the ER. Zhou et al. found that a particular protein complex in ERAD, Sel1L-Hrd1, regulates the dynamics of another organelle, the mitochondrion, by altering ER-mitochondria contacts. Three-dimensional high-resolution imaging in brown adipocytes from cold-challenged mice revealed that defective ERAD led to the formation of enlarged and abnormally shaped mitochondria with perforating ER tubules. The authors explored the consequences of ERAD deficiency on mitochondrial function and thermogenesis, which provides insights into ERADmediated ER-mitochondrial cross-talk and advances our understanding of the physiological importance of interorganelle contact.

Science, this issue p. 54

Abstract

The endoplasmic reticulum (ER) engages mitochondria at specialized ER domains known as mitochondria-associated membranes (MAMs). Here, we used three-dimensional high-resolution imaging to investigate the formation of pleomorphic “megamitochondria” with altered MAMs in brown adipocytes lacking the Sel1L-Hrd1 protein complex of ER-associated protein degradation (ERAD). Mice with ERAD deficiency in brown adipocytes were cold sensitive and exhibited mitochondrial dysfunction. ERAD deficiency affected ER-mitochondria contacts and mitochondrial dynamics, at least in part, by regulating the turnover of the MAM protein, sigma receptor 1 (SigmaR1). Thus, our study provides molecular insights into ER-mitochondrial cross-talk and expands our understanding of the physiological importance of Sel1L-Hrd1 ERAD.

Membrane contact sites mediate interorganellar communication and are key to organelle homeostasis and organismal health (13). Endoplasmic reticulum (ER)–derived mitochondria-associated membranes (MAMs) are indispensable for mitochondrial dynamics and function. MAMs, characterized by intimate contact between the ER and mitochondria (~10 to 25 nm apart) (4), mark sites for mitochondrial DNA synthesis and fission (57) and for calcium exchange and lipid biosynthesis (1, 8, 9). During mitochondrial fission, ER tubules envelop and constrict mitochondria, and the activated dynamin-related guanosine triphosphatase (GTPase) protein (Drp1) aggregates to cleave the organelle (5, 1013). The mechanism underlying interorganellar communication and the physiological consequences of miscommunication remain largely unclear.

ER-associated degradation (ERAD) is a conserved quality control mechanism to recruit ER proteins for cytosolic proteasomal degradation (1419). The Sel1L-Hrd1 protein complex constitutes the most conserved form of ERAD from yeast to humans. Sel1L resides on the ER membrane and controls the stability of the E3 ligase Hrd1 (1618, 2022). Sel1L-Hrd1 ERAD is indispensable for fundamental physiological processes in vivo such as lipid metabolism, water balance, food intake, and systemic energy homeostasis (15, 19, 2331).

Formation of megamitochondria in Sel1L−/− brown adipocytes

Serendipitously, we observed elongated mitochondria in Sel1L−/− primary brown preadipocytes, resembling those in Drp1−/− preadipocytes but distinct from those in pro-fusion Opa1−/− cells (fig. S1, A and B). These findings prompted us to explore a possible role of Sel1L in mitochondria-rich brown adipose tissue (BAT), a potential therapeutic target for obesity (32). Acute cold challenge induces mitochondrial fission in BAT to enhance respiration and maintain body temperature (33), while having no impact on the expression of ERAD genes (fig. S2, A and B).

We generated adipocyte-specific Sel1L-deficient mice (Sel1LAdipCre) by crossing Sel1Lf/f mice to the adiponectin promoter–driven Cre mice (25). Sel1L was deleted specifically in both white adipose tissue (WAT) and BAT, resulting in lower Hrd1 expression (fig. S2C). We next performed transmission electron microscopy (TEM) to visualize mitochondria in BAT from mice housed at room temperature (22°C), or 4°C for 6 hours. There was no difference in mitochondrial morphology between Sel1Lf/f and Sel1LAdipCre BAT at 22°C (Fig. 1, A and B). Acute cold challenge led to smaller mitochondria in Sel1Lf/f BAT owing to increased fission (Fig. 1, A and C). By contrast, most mitochondria in Sel1LAdipCre brown adipocytes were enlarged and pleomorphic, with abnormal cristae architecture (yellow arrows, Fig. 1, B and C, and fig. S3, A and B). Consequently, mitochondrial density per cell was reduced in cold-stimulated Sel1LAdipCre mice compared to littermate controls (fig. S3C).

Fig. 1 Sel1L regulates mitochondrial morphology in BAT during cold exposure.

(A to C) Representative TEM images of BAT from Sel1Lf/f (A) and Sel1LAdipCre mice (B) at 22°C (left) or 4°C (right) for 6 hours, with quantitation of mitochondrial size shown in (C) (n = 870, 601 mitochondria from three Sel1Lf/f mice each at 22°C and 4°C; 624, 821 from three Sel1LAdipCre mice each at 22°C and 4°C, one-way analysis of variance (ANOVA). Yellow arrows, megamitochondria. M, mitochondrion; LD, lipid droplet. (D and E) Representative TEM images of BAT from Sel1LUcp1Cre mice at 4°C for 6 hours, with quantitation of mitochondrial size shown in (E) (n = 676, 618 mitochondria for three Sel1Lf/f mice each at 22°C and 4°C; 582, 674 for three Sel1LUcp1Cre mice each at 22°C and 4°C, one-way ANOVA). (F to H) Representative SBF-SEM images (F) and 3D reconstruction (G) of mitochondria from BAT of Sel1Lf/f and Sel1LUcp1Cre mice at 4°C for 6 hours. Mitochondria in pseudocolors were reconstructed from a set of 150 SBF-SEM image stacks (65 nm/slice). (H) Quantitation of 3D volume of mitochondria (n = 61 and 63, Sel1Lf/f and Sel1LUcp1Cre mice, Student’s t test). All experiments were repeated three times. Data are mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; n.s., not significant.

To exclude a possible contribution from WAT, we generated brown adipocyte–specific Sel1L-deficient mice (Sel1LUcp1Cre) using Ucp1 promoter–driven Cre mice (fig. S2D). Consistent with Sel1LAdipCre mice, mitochondria in Sel1LUcp1Cre BAT were markedly enlarged and pleomorphic upon acute cold exposure (yellow arrows, Fig. 1, D and E, and fig. S3D). This observation was confirmed using confocal microscopy, with BAT immunolabeled for translocase of outer mitochondrial membrane 20 (Tomm20) and matrix protein pyruvate dehydrogenase (PDH) (fig. S3E).

To gain further insight into mitochondrial morphology at the three-dimensional (3D) level, we performed serial block-face scanning electron microscopy (SBF-SEM) (fig. S4A), which allows imaging of a large field (>10 cells) at 3D planes with nanometer resolution (fig. S4, B and C) (34). In cold-exposed Sel1Lf/f BAT, mitochondria were predominantly spherical or ovoid, with an average volume of ~1.7 μm3 (fig. S4B and Fig. 1, F to H). Mitochondrial volume was increased fivefold in Sel1L−/− BAT (Fig. 1, F to H, and fig. S4C). Moreover, 3D reconstruction analyses revealed pleomorphic megamitochondria, e.g., branched, or chair- or cup-shaped (Fig. 1G and movies S1 and 2). Thus, Sel1L deficiency in brown adipocytes triggers the formation of pleomorphic megamitochondria within hours of cold exposure.

Sel1L controls ER-mitochondria contacts

We next explored whether Sel1L deficiency affected MAMs. In Sel1Lf/f BAT, MAMs were identified as thin sheets aligned with mitochondria, with the distance between the two organelles averaging 10 to 20 nm (cyan arrows, Fig. 2, A, C, and D, and fig. S5A). By contrast, MAMs of Sel1L−/− BAT formed tubules located closer to mitochondria, with an average distance of 5 to 10 nm (Fig. 2, B to D, and fig. S5B). Moreover, the number of MAMs per mitochondrion was increased in Sel1L/− BAT after cold exposure (Fig. 2E). These perimitochondrial tubular structures were verified as ER by binding-immunoglobulin protein (BiP)-specific immunoelectron microscopy (Fig. 2F): Pleomorphic megamitochondria appeared to “grow” around ER tubules in cold-stimulated Sel1L/− BAT.

Fig. 2 Sel1L controls ER-mitochondria contacts in cold-stimulated brown adipocytes.

(A and B) Representative TEM images of BAT from Sel1Lf/f (A) and Sel1LUcp1Cre (B) mice at 4°C for 6 hours. M, mitochondrion; cyan arrows, MAMs; red lines, mitochondrial membranes; green, ER. The area shown by white dotted lines is enlarged in the inset. (C to E) Representative TEM images of the MAMs in BAT from Sel1Lf/f and Sel1LUcp1Cre mice at 4°C for 6 hours, with quantitation of ER-mitochondrion distance (D) and abundance of MAM per mitochondrion (E). n = 35 and 40 MAMs (D) and n = 450 and 572 mitochondria (E) for Sel1Lf/f and Sel1LUcp1Cre. (F) Representative TEM images of BiP-specific immunogold labeling in BAT from Sel1LAdipCre mice at 4°C for 6 hours. Red dotted line outlines one megamitochondrion; cyan arrows, BiP-positive ER tubule(s). (G) Representative FIB-SEM (left) and the 3D tomography images (300 and 170 slices, 5 nm/slice) of mitochondria in BAT from Sel1Lf/f and Sel1LUcp1Cre mice at 4°C for 6 hours. Magenta, mitochondria; green, ER. Cyan arrows, MAMs; orange arrows, MAMs going through the concave surface of a mitochondrion. All experiments were repeated two to three times. Data are mean ± SEM. ***p < 0.001 by Student’s t test.

We next performed focused ion-beam SEM (FIB-SEM) to reconstruct mitochondria with MAMs at the 3D level. In comparison to SBF-SEM, FIB-SEM offers higher resolution at the z plane (35) with a resolution of 5-nm isotropic voxels (fig. S6). Three-dimensional reconstruction revealed more intimate interactions between the ER (green) and mitochondria (purple) in Sel1LUcp1Cre BAT versus a Sel1Lf/f mitochondria (pink) (Fig. 2G). In many U-shaped mitochondria, several ER tubules were found along the entire concavity of the organelle (orange arrows, Fig. 2G and movie S3).

Formation of mitochondria-perforating ER tubules in the absence of Sel1L

In many U- or dumbbell-shaped megamitochondria, the outer membrane folded back on itself, with less than 10 nm between membranes (green arrows, Fig. 3A and fig. S7A), which we speculated may fuse to embed ER tubules within the mitochondria. Indeed, there were many tubular structures embedded within mitochondrial profiles in Sel1L−/− BAT (red arrows, Fig. 3, B and C, and fig. S7, B and C). In some, double- or triple- membrane structures were noted. Moreover, these tubular structures served as a “hub” from which cristae folds radiated (Fig. 3, B and C, and fig. S7, B to D). To determine whether these peculiar structures inside mitochondria were ER, we performed BiP-specific immunoelectron microscopy. Clusters of BiP-positive signal were detected both perimitochondrially (cyan arrows) and within mitochondrial profiles (red arrows, Fig. 3D and fig. S7E). We then performed SBF- and FIB-SEM to reconstruct whole mitochondria with perforating ER tubules. For example, we present 12 consecutive images of one megamitochondrion with two parallel penetrating ER tubules (fig. S8 and Fig. 3, E and F). Two ER tubules (arrows) perforated the mitochondrial profiles in parallel, at a distance of 1 to 1.3 μm apart, with radiating and interconnecting cristae folds, resembling wagon wheel spokes. Similarly, independent FIB-SEM analyses showed two parallel ER tubules (0.7 to 1 μm apart) penetrating another megamitochondrion in Sel1LUcp1Cre BAT upon cold exposure (fig. S9 and movie S4). Thus, Sel1L deficiency leads to the formation of pleomorphic megamitochondria with perforating ER tubule(s) in brown adipocytes upon cold challenge.

Fig. 3 Sel1L deficiency leads to the formation of megamitochondria with perforating ER tubules.

(A) Representative TEM images of BAT in Sel1LUcp1Cre mice housed at 4°C for 6 hours, showing a megamitochondrion wrapping around the tubular structures (cyan arrows). Green arrows, two opposite sides of a mitochondrion. (B and C) Representative TEM images of BAT from Sel1LUcp1Cre (B) and Sel1LAdipCre (C) mice at 4°C for 6 hours, showing megamitochondria with tubular structures (red arrows). (D) Representative BiP-immunogold TEM images of BAT in Sel1LAdipCre mice at 4°C for 6 hours. Red and cyan arrows, mitochondria-perforating ER tubules and perimitochondria ER tubules. (E and F) Representative SBF-SEM images (E) and 3D reconstruction (F) in BAT of Sel1LUcp1Cre mice at 4°C for 6 hours, showing four different slices of a megamitochondrion with two parallel perforating ER tubules (red and magenta arrows). All 12 slices (65 nm/slice) are shown in fig. S8. All experiments were repeated two to three times.

Impaired mitochondrial function and thermogenic response in Sel1L−/− mice

Next, we asked how Sel1L deficiency affected mitochondrial function and thermogenesis. Purified mitochondria from cold-exposed Sel1LAdipCre mice had a reduced oxygen consumption rate (OCR) from oxidation of pyruvate and malate compared to those from Sel1Lf/f mice (Fig. 4A). In addition, differentiated Sel1L−/− brown adipocytes showed defective respiration in response to adrenergic receptor agonist norepinephrine (NE) stimulation (Fig. 4B). Mitochondrial functional defects were further confirmed using a targeted metabolomics analysis of 119 intracellular metabolites in NE-treated differentiated Sel1Lf/f versus Sel1L−/− brown adipocytes (fig. S10). Pathway analysis of the 30 significantly altered metabolites revealed three main pathways altered in Sel1L−/− adipocytes: tricarboxylic acid (TCA) cycle, and pyrimidine and purine metabolism (Fig. 4C), all of which are associated with mitochondrial function.

Fig. 4 Sel1L in brown adipocytes regulates mitochondrial function and cold-induced thermogenesis.

(A and B) Oxygen consumption rate (OCR) of purified mitochondria from BAT of cold-exposed mice (A) and differentiated brown adipocytes (B) with the addition of various stimuli as indicated. NE, norepinephrine; Oligo, oligomycin; Pyr/Mal, pyruvate/malate; FCCP, carbonyl cyanide p-trifluoromethoxyphenylhydrazone; Anti, antimycin; Rot, Rotenone. (C) Metaboanalyst pathway enrichment analysis for the metabolites in Sel1Lf/f and Sel1L−/− brown adipocytes treated with 1 μM NE for 1 hour (n = 3 per group). (D) Rectal temperature of 8- to 10- week-old mice housed at 22°C or 4°C for 6 hours (n = 7 to 10 mice each, one-way ANOVA). (E and F) Representative H&E (hematoxylin and eosin) (E) and Perilipin1 (F) staining of BAT from Sel1Lf/f and Sel1LAdipCre mice housed at 22°C or 4°C for 6 hours. All experiments were repeated three times except panel (C) (three samples per group). Data are mean ± SEM. ***p < 0.001.

Both Sel1LAdipCre and Sel1LUcp1Cre mice appeared normal and grew at a rate comparable to that of their Sel1Lf/f littermates up to 22 weeks on a low-fat diet at room temperature (fig. S11A). Although core body temperatures (~37°C) were similar between cohorts when housed at room temperature, Sel1LAdipCre mice were cold sensitive and dropped core body temperature to 26°C within 6 hours of cold exposure (Fig. 4D, versus 33°C in Sel1Lf/f). Histological examination of BAT revealed that, although largely indistinguishable at room temperature, Sel1LAdipCre BAT exhibited larger lipid droplets upon cold exposure compared to Sel1Lf/f mice (Fig. 4E), pointing to defects in lipid mobilization. This was confirmed using immunostaining of lipid droplet binding protein Perilipin 1 (Fig. 4F). Similar observations were made with Sel1LUcp1Cre mice (Fig. 4D and fig. S11, B and C). Phosphorylation of hormone-sensitive lipase (HSL) was not affected in response to β3-adrenergic receptor agonists (fig. S11, D and E), thereby uncoupling β-adrenergic signaling from the thermogenic defect of Sel1L−/− mice.

Sel1L effect on mitochondria is mediated through Hrd1 ERAD

Sel1L may have an Hrd1-independent function, and vice versa (25, 36, 37). To determine whether Sel1L and Hrd1 act similarly to regulate mitochondria in BAT, we generated brown adipocyte–specific Hrd1-deficient (Hrd1Ucp1Cre) mice (fig. S12A). Deletion of Hrd1 stabilized Sel1L protein (fig. S12A), in line with the notion that Sel1L is a substrate of Hrd1. Hrd1Ucp1Cre mice grew normally (fig. S12B) but were cold sensitive (Fig. 5A), with more lipid droplets in cold-challenged BAT compared to Hrd1f/f mice (fig. S12C). Moreover, Hrd1Ucp1Cre BAT demonstrated enlarged mitochondria, some with penetrating ER tubule(s), upon cold stimulation (Fig. 5, B to D, and fig. S12D).

Fig. 5 Hrd1, but not Ire1α of UPR, controls mitochondrial morphology and thermogenesis in brown adipocytes.

(A) Rectal temperature of 8- to 10- week-old Hrd1f/f and Hrd1Ucp1Cre mice housed at 22°C or 4°C for 6 hours, n = 4 per group (one-way ANOVA). (B to D) Representative TEM images of BAT from Hrd1f/f and Hrd1Ucp1Cre mice housed at 22°C or 4°C for 6 hours with quantitation shown in (C). n = 543, 581 mitochondria for 22°C and 4°C, Student’s t test. Yellow arrows, megamitochondria; red arrows, ER tubules within mitochondrial profiles. (E) Reverse transcription–polymerase chain reaction analysis of Xbp1 mRNA splicing in BAT at 22°C or 4°C for 6 hours with quantitation of the ratio of spliced Xbp1 (s) to total Xbp1 [spliced + unspliced (u)] shown below the gel (n = 6 mice per group). BAT from wild-type mice injected with vehicle or tunicamycin (Tuni, 1 mg/kg) included as controls. (F and G) Representative TEM images of BAT from Ire1aAdipCre mice at 4°C for 6 hours with quantitation shown in (G) (n = 500, 665 from two mice each, Student’s t test). All experiments were repeated two to three times. Data are mean ± SEM. **p < 0.01; ***p < 0.001.

Because ER stress has been linked to mild mitochondrial elongation (38, 39), we next asked whether the effect of ERAD deficiency on mitochondria was mediated through ER stress. Basal Xbp1 mRNA splicing was slightly higher in Sel1LAdipCre BAT compared to Sel1Lf/f BAT at room temperature; however, cold exposure reduced Xbp1 mRNA splicing in Sel1LAdipCre BAT (Fig. 5E). There was no detectable cell death in Sel1LAdipCre BAT (fig. S13A). To study a possible impact of ER stress signaling on mitochondria in vivo, we generated mice lacking Ire1α, a key sensor of the unfolded protein response (UPR), specifically in adipocytes (Ire1aAdipCre). These knockout mice grew normally and were not cold sensitive (fig. S13, B to D). Mitochondria from Ire1aAdipCre exhibited normal morphology, with reduced size at 4°C (Fig. 5, F and G). Thus, the effect of Sel1L on mitochondria is mediated by ERAD, independent of ER stress or cell death.

Sel1L-Hrd1 ERAD regulates ER-mitochondria contacts and dynamics through SigmaR1

We next explored how mitochondrial dynamics were affected by ERAD dysfunction. Whereas total protein amounts of mitochondrial respiratory proteins and Ucp1 were largely comparable between the cohorts (fig. S14A), cold-induced phosphorylation of Drp1 at Ser616, a key activating event in mitochondrial division (11, 40), was reduced in Sel1LAdipCre BAT compared to Sel1Lf/f BAT (Fig. 6A). Opa1 processing to its shorter form upon cold exposure did not occur in Sel1LAdipCre BAT (Fig. 6A). Protein amounts of outer mitochondrial membrane Drp1 receptors Fis1 and Mff were unchanged (fig. S14A). Similar observations were obtained in Hrd1−/− BAT (fig. S14B), but not in Ire1α−/− BAT (fig. S13E). Furthermore, in Sel1L−/− BAT, cold exposure increased oligomerization of key fusion factor Mfn2 into high molecular weight (HMW) complexes as revealed by blue-native gels and sucrose gradient fractionation (fig. S14, C and D). Indeed, the Mfn2 HMW complex promotes mitochondrial fusion (4143). We then tested the protein abundances of several known MAM proteins, including Mfn2, Vapb, Bap31, and SigmaR1. Only SigmaR1 amounts were increased in Sel1L- and Hrd1-deficient BAT (Fig. 6B and fig. S15, A and B), but not in Ire1α−/− BAT (fig. S13E). However, SigmaR1 mRNA abundance was comparable in Sel1L−/− versus WT BAT in response to cold (fig. S15C), pointing to a posttranscriptional regulation of SigmaR1 protein.

Fig. 6 Sel1L-Hrd1 ERAD regulates MAM and mitochondrial dynamics through SigmaR1.

(A) Immunoblot analysis of mitochondrial dynamic proteins in BAT from Sel1Lf/f and Sel1LAdipCre mice at 22°C or 4°C for 6 hours with quantitation of phos-/total Drp1 and large (L)-/small (S)-Opa1 shown below the gel (n = 7 mice per group). (B) Immunoblot analysis of MAM proteins in BAT of Sel1Lf/f and Sel1LAdipCre mice with quantitation of SigmaR1/Hsp90 shown below the gel (n = 7 mice per group). (C) Immunoblot analysis of endogenous SigmaR1 in WT and Hrd1-deficient (Hrd1−/−) HEK293T cells with or without cycloheximide (CHX) and MG132 treatment. (D) Immunoblot analysis of GFP or endogenous SigmaR1 immunoprecipitates in HEK293T (with or without MG132 treatment), showing Hrd1-mediated SigmaR1 ubiquitination. (E) Immunoblot analysis of Flag immunoprecipitates in HEK293T cells transfected with a combination of plasmids, showing a dose-dependent interaction between SigmaR1-Flag and Mfn2-EGFP (enhanced GFP). (F) Immunoblot analysis of Mfn2 oligomers or Mfn2-containing high molecular weight (HMW) complexes in brown adipocytes treated with or without NE (1 μM, 1 hour) in blue native (BN)–polyacrylamide gel electrophoresis (PAGE) and regular SDS-PAGE. (G and H) Representative images of mito-DsRed– and mEmerald-Sec61b– expressing pre-adipocytes. DAPI (4′,6-diamidino-2-phenylindole) (blue) with quantitation of ER-mitochondrial colocalization using Manders’ overlap coefficient (50 cells each) shown in (H), one-way ANOVA. (I and J) Confocal images showing the spread and decay of mitochondria-targeted photoactivated GFP (mito-PAGFP) in brown adipocytes 60 min after NE stimulation and photoactivation. Active mitochondrial were stained with tetramethylrhodamine ethyl ester perchlorate (TMRE) (red). Quantitation of changes of GFP signal intensity over time shown in (J). (n = 13, 15, and 10 cells for Sel1Lf/f, Sel1L−/−, and Sel1L−/−;SigmaR1−/−, one-way ANOVA). All experiments were repeated two to three times. Data are mean ± SEM. **p < 0.01; ***p < 0.001.

SigmaR1 is a single-span ER-resident protein residing at the MAMs, where it may regulate ER-mitochondria contacts and mitochondrial dynamics (44, 45), although a detailed mechanism remains vague. SigmaR1 protein was degraded by the proteasome with a half-life of ~6 hours (Fig. 6C), interacted with the E3 ligase Hrd1 (fig. S15, D and E), and was ubiquitinated in a largely Hrd1-dependent manner (Fig. 6D). In the absence of Sel1L or Hrd1, SigmaR1 protein was stabilized (Fig. 6C) and accumulated in the ER (fig. S15F). SigmaR1 interacted with Mfn2 as well as other MAM proteins Bap31 and Vapb, but not Ip3r (Fig. 6E and fig. S15G), and was localized at very close proximity to mitochondria in Sel1L−/− cells (fig. S15H). Genetic deletion of SigmaR1 reduced cold-induced Mfn2 HMW complexes in Sel1L−/− brown adipocytes (Fig. 6F and fig. S16A). Deletion of SigmaR1 reversed mitochondrial elongation (fig. S16, B and C) and reduced ER-mitochondrial contacts in Sel1L−/− cells (Fig. 6, G and H). In a mitochondrial fusion assay (33) using mitochondria-targeted photoactivated green fluorescent protein (mito-PAGFP), deletion of SigmaR1 attenuated the spread and decay of GFP intensity seen in Sel1L−/− brown adipocytes upon NE treatment (Fig. 6, I and J). Thus, SigmaR1 mediates mitochondrial hyperfusion in Sel1L-deficient cells, thereby linking ERAD to ER-mitochondrial contacts and mitochondrial dynamics in brown adipocytes.

Discussion

Here, we found a critical role for Sel1L-Hrd1 ERAD in regulating ER-mitochondria contacts and mitochondrial function. We observed profound changes in mitochondrial morphology, beyond mere elongation or swelling that is commonly associated with mitochondrial dysfunction or ER stress. Using 3D imaging techniques, we reconstructed these pleomorphic mitochondria and their associated MAMs. We speculate that the ER tubules embedded within profiles of pleomorphic megamitochondria represent halted fission and/or accelerated fusion intermediates. Our mechanistic studies identified MAM protein SigmaR1 as an ERAD substrate, which, when deleted, reduced ER-mitochondria contacts and rescued abnormal mitochondrial dynamics and morphology in Sel1L−/− cells. SigmaR1 may regulate mitochondrial dynamics by interacting with Mfn2 as well as other MAM proteins and promoting Mfn2 oligomerization in an unknown manner.

MAMs are disturbed in many diseases such as obesity (46), cardiac disease (47), and neurodegeneration (48); however, to date, the nature of this disruption in disease pathogenesis remains vague. Our data suggest that Sel1L-Hrd1 ERAD may regulate the dynamics of ER-mitochondria interactions through modulation of MAMs. Indeed, an inverse correlation between Sel1L and SigmaR1 expression levels has been reported in patients with neurodegeneration (4951). Considering the recent identification of Sel1L mutants in progressive early-onset cerebellar ataxia in canines (52) and various cancers (53, 54), this study may have implications not only for interorganelle communication but also for the pathophysiological role of Sel1L-Hrd1 ERAD.

Supplementary Materials

science.sciencemag.org/content/368/6486/54/suppl/DC1

Materials and Methods

Figs. S1 to S16

References (5564)

Movies S1 to S4

References and Notes

Acknowledgments: We thank D. Fang, L. M. Hendershot, D. Lombard, J. Lin, S. Soleimanpour, T.-P. Su, and R. Wojcikiewicz for providing reagents; F. Mao for data analysis; P. Arvan, L. Rui, Y. Shi, S. Sun, and O. MacDougald for constructive comments on the manuscript; and other members in the Arvan and Qi laboratories for technical assistance and insightful discussions. Funding: This work is supported by 1R01GM123266 and R01GM130695 (H.Se.); HL144657 (D.A.B.); 1R01DK107583 (J.W.); R01DK110439 and P20GM121176 (M.L.); R01NS086819 and R01NS091242 (T.H.S.); UL1TR000433, F32CA228328, and P30DK034933 (C.J.H.); and 1R01GM113188, 1R01DK105393, 1R01DK120047, and 1R35GM130292 (L.Q.). Metabolomics studies performed at the University of Michigan were supported by NIH grant DK097153. Z.Z. is supported by ADA Postdoctoral Fellowship (1-19-PDF-093). M.T. was supported in part by the Pew Latin American Postdoctoral Fellowship. R.B.R. is supported by the Training Program in Endocrinology and Metabolism (5T32DK007245). C.A.L. was supported by a 2017 AACR NextGen Grant for Transformative Cancer Research (17-20-01-LYSS) and an ACS Research Scholar Grant (RSG-18-186-01). Author contributions: Z.Z. and M.T. designed and performed most of the experiments; H.Sh., S.K., C.J.H., F.V.d.B., C.A.L., T.Y., S.W., Y.L., C.W., M.L., H.Se., and A.H.H. performed the experiments; A.T., J.W., T.H.S., and D.A.B. provided reagents and insightful discussion; R.B.R. edited the manuscript and provided insightful discussion; L.Q. directed the study and wrote the manuscript; Z.Z. and M.T. wrote the methods and figure legends; all authors commented on and approved the manuscript. Competing interests: The authors declare no conflicts of interest. Data and materials availability: The microarray data have been deposited in Gene Expression Omnibus with accession number GSE145895.

Stay Connected to Science

Navigate This Article