Outer mitochondrial membrane E3 Ub ligase MARCH5 controls de novo peroxisome biogenesis

Summary

We report that the outer mitochondrial membrane (OMM)-associated E3 Ub ligase MARCH5 is vital for generating mitochondria-derived pre-peroxisomes. In human immortalized cells, MARCH5 knockout leads to the accumulation of immature peroxisomes, reduced fatty acid-induced peroxisomal biogenesis, and abnormal peroxisome biogenesis in MARCH5/Pex14 and MARCH5/Pex3 dko cells. Upon fatty acid-induced peroxisomal biogenesis, MARCH5 redistributes to peroxisomes, and ubiquitination activity-deficient mutants of MARCH5 accumulate on peroxisomes containing high levels of the OMM protein Tom20 (mitochondria-derived pre-peroxisomes). Similarly, depletion of peroxisome biogenesis factor Pex14 leads to accumulation of MARCH5- and Tom20-positive pre-peroxisomes, while no peroxisomes are detected in MARCH5/Pex14 dko cells. Inconsistent with MARCH5 merely acting as a quality factor, mitochondrial decline is not evident in tested models. Furthermore, reduced expression of peroxisomal proteins is detected in MARCH5^−/− cells, whereas some of these proteins are stabilized in peroxisome biogenesis deficiency models lacking MARCH5 expression. Thus, MARCH5 is central for mitochondria-dependent peroxisome biogenesis.

eTOC blurb:

Verhoeven et al. found that mitochondrial E3 Ub ligase MARCH5 controls the formation of mitochondria-derived pre-peroxisomes. The data support the hybrid, mitochondria-dependent model of peroxisome biogenesis and show that MARCH5 is an essential mitochondrial protein in this process.

Graphical Abstract

Introduction

MARCH5 (Mitol) is an integral E3 Ub ligase of the outer mitochondrial membrane (OMM) implicated in many mitochondrial pathways, including regulation of fission and fusion, mitochondrial steps in apoptosis, and removal of disease-linked toxic proteins^1–12. MARCH5 expression is regulated by peroxisome proliferator-activated receptor-γ (PPAR γ) and negatively correlates with fat mass across a panel of genetically diverse mouse strains, including a common obesity model (ob/ob) in mice¹³. This relationship has also borne out in people suffering from obesity and metabolic syndrome¹³, linking MARCH5 with regulating energy metabolism. Consistent with the metabolic function of MARCH5, the Genome-wide associated loci prioritization (Open Targets Genetics) shows a high-confidence prediction of MARCH5 mutations as a likely causal gene for obesity and type II diabetes. However, the role and mechanism of MARCH5 in cellular energy metabolism are poorly understood. Zheng et al.¹⁴ reported that, in addition to the OMM, a subset of MARCH5 also localizes to the peroxisomes. Peroxisomes are abundant dynamic single-membrane organelles that host diverse biochemical pathways such as ether-phospholipid biosynthesis, bile acid synthesis, glyoxylate metabolism, and amino acid catabolism. Peroxisomes also play a central role in fatty acid β-oxidation^15,16 with preference for chain-shortening of very long-chain fatty acids, but can also metabolize medium- and long-chain fatty acids such as palmitate¹⁷ that are typically oxidized by mitochondria, providing energy intermediate donors for mitochondrial ATP generation¹⁸. While the function of MARCH5 in peroxisome-specific autophagy (pexophagy) was also proposed¹⁴, the scope and mechanism of MARCH5, and potential crosstalk in the control of mitochondrial and peroxisomal pathways, are currently not known.

The functional and molecular connection between mitochondria and peroxisomes was initially suggested by studies which found several proteins on peroxisomes that were previously thought to uniquely localize to the mitochondria, including fission factors dynamin-related protein 1 (Drp1), Fis1 and mitochondrial fission factor (Mff), AAA-ATPase ATAD1/Msp1, MARCH5^14,19–21. On the other hand, several peroxisomal proteins, such as Pex26, Pex14, and Pex3, localize to the OMM, especially when the peroxisomal biogenesis machinery is dysfunctional^21–24. This localization was considered a mis-targeting, but it also provided the early support for the emerging view by which mitochondria play a role in peroxisomal biogenesis^25,26. New peroxisomes can form by the growth and division of pre-existing organelles but can also emerge de novo. In the de novo biogenesis, a subset of pre-peroxisomes develops from the mitochondria, and some peroxisomal biogenesis factors, including Pex3 and Pex14, could have mitochondrial functions in this process²⁵. The metabolically active peroxisomes are then formed by the fusion of ER- and mitochondria-derived pre-peroxisomes (hybrid model)^25,26. The mitochondrial proteins required for mitochondria-dependent steps in peroxisome biogenesis remain unknown.

Here, we provide evidence supporting the central regulatory role of MARCH5 in peroxisome biogenesis and, thereby, in cellular metabolic adaptation to changes in the type of energy substrates available to the cell.

Results

Activity-dependent peroxisomal localization of MARCH5.

We tested subcellular localization of MYC-tagged WT MARCH5 (MYC-WT MARCH5), RING domain activity-deficient MARCH5 (MYC-MARCH5^H43W; ^2,3) and the C-terminal truncation mutant (MYC-MARCH5^ΔCT;²) in MARCH5 knockout HeLa cells (MARCH5^−/−). While MYC-WT MARCH5 localized mainly to the mitochondria, with occasional peroxisomal colocalization (Fig. 1A, D), MYC-MARCH5^H43W and MYC-MARCH5^ΔC were associated with peroxisomes to a much higher degree (Fig. 1B-D). Since these mutants have reduced activity toward MARCH5 substrates and self-ubiquitination^2,7, it is likely that inhibition of MARCH5 E3 Ub ligase activity increases association of MARCH5 with peroxisomes. Furthermore, MYC-WT MARCH5-expressing cells show minor colocalization of the OMM marker Tom20 with Pex14 (Fig. 1A, D), confirming a low-level colocalization of Tom20 and a peroxisome-cytosol shuttling import receptor, Pex5^27,28. On the other hand, cells expressing MARCH5 mutants show a high degree of Tom20-Pex14 colocalization (Fig. 1B-D). Furthermore, ~80% of MARCH5^H43W vesicular structures contain the peroxisome matrix marker Catalase (Fig. S1A, B) but only 1.5% contain the inner mitochondrial membrane (IMM) protein ATP5A1 (Fig. S1C, D), indicating that MARCH5^H43W preferentially localizes to the hybrid Tom20-enriched peroxisomes, but not mitochondrial fragments.

Figure 1: Peroxisomal localization of MARCH5.

Typical images of MARCH5^−/− (A-C) and Drp1^−/− (E-F) HeLa cells transfected with MYC-WT MARCH5 (A, E), MYC-MARCH5^H43W (B, F), or MYC-MARCH5^ΔCT (C) immunostained to detect peroxisomes (Pex14; red), MARCH5 (anti-MYC; green), and outer mitochondrial membrane (Tom20; blue). Detail images are from the areas marked with yellow rectangles. Fluorescence profiles are shown on the right sides of the images in A-C. Red arrowheads in the images and profiles indicate Pex14, and corresponding MARCH5 and Tom20 localizations. Blue arrowheads in fluorescence profiles indicate examples of mitochondrial, Pex14-negative, Tom20, and MARCH5 localizations. Bars: 10μm and 1μm (detail images). (D) Pearson’s Correlation Coefficients represent the colocalization of Pex14 with MARCH5 (orange) or Pex14 with Tom20 (blue). Mean±SEM. N=30 cells/condition. One-way ANOVA. **** p<0.0001. (G) Expression levels of exogenous MARCH5 and representative peroxisomal proteins were analyzed by Western blot, as indicated in the figure. Tom20 and actin are loading controls. *- background or signal remaining after membrane reblotting.

Inhibition of dynamin-related protein 1 (Drp1) hinders peroxisomal fission, causing the formation of elongated peroxisomes^29,30 and allowing a straightforward visual evaluation of peroxisomal protein content. In Drp1^−/− HeLa cells, markedly higher levels of the MYC-MARCH5^H43W localized to the peroxisomes compared to MYC-WT MARCH5 (Figs. 1E, F, and S1E-J). The shift in the MARCH5^H43W localization was associated with reduced levels on the OMM, while WT MARCH5 localized mainly to the OMM. This pattern was apparent in all tested cells (~50 cells/group), regardless of MYC-MARCH5 expression levels (Fig. 1E, F). Moreover, the MARCH5^H43W-positive structures detected in Drp1^−/− cells were deficient in IMM and mitochondrial intermembrane space (IMS) markers ATP5A1 and cytochrome c, respectively (Fig. S1G-J) but contained Catalase (Fig. S1E, F), indicating that inactive MARCH5 mutant preferentially localizes to the peroxisomes, but not elongated mitochondria apparent in Drp1^−/− cells. MARCH5 E3 Ub ligase activity could mediate early steps in mitochondrial pre-peroxisome formation, and its inhibition induces abnormal accumulation of immature peroxisomes containing the OMM protein Tom20, associated with retention of inactive MARCH5 on Pex14-positive peroxisomes.

The mechanism by which mutant MARCH5 accumulates on peroxisomes remains to be determined. One possibility is that a reduced self-ubiquitination and degradation of E3 Ub ligase activity-deficient mutants leads to their abnormal retention on developing peroxisomes. At ~20hr after transfection, there were no apparent changes in the expression of representative peroxisomal markers (Fig. 1G), including Pex19, a peroxisome import receptor distributed between the cytoplasm and peroxisomes³¹. Pex19 has been reported to bind the transmembrane region on MARCH5 and target it to peroxisomes¹⁴. These data indicate no direct relationship between Pex19 expression and MARCH5 mutant localization.

MARCH5 deficiency affects peroxisome homeostasis.

Zheng et al.¹⁴ showed peroxisomal localization of WT MARCH5 and proposed the role of MARCH5 in peroxisome-specific autophagy (pexophagy). However, a detailed peroxisome status in MARCH5 deficient cells has yet to be determined. We used the quantitative imaging to explore the effect of MARCH5 knockout on peroxisome number, size, and maturation. Complete colocalization of peroxisomal fatty acid transporter PMP70 with Catalase, often used as a marker of “mature” peroxisomes^32–34 was apparent in WT cells (Fig. 2A). Conversely, in MARCH5^−/− cells, many PMP70-positive peroxisomes were deficient of Catalase staining (Fig. 2B; blue arrowheads). Quantifying peroxisome number also indicates reduced number of Catalase-positive peroxisomes, compared to PMP70- or peroxisome biogenesis factor Pex14-positive peroxisomes, in MARCH5^−/− HeLa and HCT116 cells⁷ (Fig. 2C). On the other hand, no difference in abundance of PMP70- and Catalase-positive peroxisomes was detected in WT HeLa and WT HCT116 cells (Fig. 2C). Furthermore, peroxisomes in MARCH5^−/− cells are smaller than in WT cells (Fig. 2D). Since similar changes occur in two MARCH5^−/− cell models, the CRISPR/Cas9 generated MARCH5^−/− HeLa and homologous recombination generated MARCH5^−/− HCT116 cells⁷, one can conclude that MARCH5 controls peroxisome size and number. The fact that many peroxisomes in MARCH5^−/− do not contain detectable levels of Catalase (Fig. 2B) also supports the possibility that peroxisome biogenesis stalled before the Catalase import. Fittingly, Catalase accumulates in cytosolic fractions from MARCH5^−/− cells, compared to WT cells (Fig. 2E; red arrowheads).

Figure 2: Defective peroxisomes in MARCH5^−/− cells.

WT (A) and MARCH5^−/− (B) HeLa cells were immunostained to detect PMP70 (purple) and Catalase (green). Arrowheads in detail images in B indicate PMP70-positive, Catalase-negative peroxisomes. Bars: 10μm and 2μm (detail images). (C) Number of Pex14-, PMP70- and Catalase-positive peroxisomes/cell in WT and MARCH5^−/− HeLa (left) and HCT116 (right) cells. Mean±SEM. N=62 (WT HeLa; PMP70), 52 (MARCH5^−/− HeLa: PMP70), 29 (WT HeLa, Catalase), 33 (WT HeLa; Pex14). Mean±SEM. One-way ANOVA. ** p<0.01, **** p<0.0001 (D) Size of Pex14-, PMP70- and Catalase-positive peroxisomes in WT and MARCH5^−/− HeLa (left) and HCT116 (right) cells. Mean±SEM. One-way ANOVA. **** p<0.0001. Data were pulled from 3 independent experiments. (E) WT and MARCH5^−/− HeLa (left) and HCT116 (right) cells were fractionated into mitochondria-enriched heavy membrane (HM), light membrane (LM), and cytosolic fractions (C) followed by Western blot to detect proteins indicated in the figure. TCL: total cell lysate. *-signal remaining after membrane reblotting. Red arrowheads indicate Catalase levels in cytosolic fractions. (F) TCLs from WT and MARCH5^−/− HeLa and HCT116 cells were analyzed by Western blot for expression levels of peroxisomal proteins. Actin and Tom20 are loading controls. (G) Densitometric evaluation of the expression of peroxisomal proteins from cells treated as shown in F. Levels of respective proteins in WT cells were taken as 1. Mean±SEM. N=4–8 independent experiments. One Way ANOVA with Sidak Correction. **** p<0.0001, ** p<0.01, * p<0.05, n.s.= no significant difference.

Depletion of one peroxisome biogenesis factor can lead to reduced steady-state levels of other peroxisomal proteins^35,36. We tested the effect of MARCH5 deficiency on the expression of peroxisomal proteins. We found a consistent reduction in the expression of PMP70, Pex5, Pex14, and Catalase, unaltered levels of Pex19, and slightly increased levels of Pex3 in both HeLa and HCT116 MARCH5^−/− as compared to WT cells (Fig. 2F, G). On the other hand, Pex16 was only reduced in MARCH5^−/− HeLa cells (Fig. 2F, G). Thus, depletion of MARCH5 leads to reduced levels of some peroxisomal proteins, resembling effects of the deficiencies in established peroxisome biogenesis factors^35,36. The cycloheximide (CHX) chase experiments (Fig. S1K) revealed that the stabilities of Catalase and Pex26 in WT and MARCH5^−/− cells are comparable. At the same time, no changes in Pex14, Pex16, and PMP70 were detected within 8hr of treatment (Fig. S1K). Together with the accumulation of Catalase-deficient peroxisomes (Fig. 2B) and reduced peroxisome size in MARCH5^−/− cells (Fig. 2D), these data suggest that MARCH5 is vital for peroxisome biogenesis, in a mechanism that does not rely on direct control of stabilities of tested proteins. The likely scenario is that the loss of MARCH5 affects other peroxisomal proteins indirectly, in a similar manner as deficiencies of certain peroxisomal factors lead to a reduction in levels of other peroxisomal proteins (Figs. S2J and S4F;^35,36).

MARCH5 controls peroxisome biogenesis under fatty acid- and OXPHOS-dependent growth.

The current understanding of the role and mechanisms by which the Ub proteasome system-mediated control of protein stability and/or activity regulates mitochondrial and peroxisomal function is mainly based on studies of cells grown in 25mM glucose (high glucose) media. Under high glucose growth, cells use mitochondria as biosynthetic hubs but rely primarily on glycolysis to generate ATP, and mitochondria display low bioenergetic activity³⁷. Even primary, normally OXPHOS-dependent cells adapt to glycolysis as a major source of ATP when high glucose levels are available³⁸. Switching cells to low or glucose-deficient growth and providing OXPHOS-stimulating substrates is predicted to increase the utilization of mitochondria for ATP generation and activation of peroxisomal metabolic pathways, especially when fatty acids are primary OXPHOS substrates. We set up growth conditions in which cells predominantly rely on OXPHOS as the source of ATP generation. This model consists of a glucose-free medium, supplemented with 4.5mM galactose and 25μM palmitoyl-L-carnitine (pLcar), as a donor of OXPHOS substrates. In addition to glucose exclusion, pyruvate, a final metabolite in the glycolysis pathway, is removed from this cell culture medium (OXPHOS medium) to promote mitochondrial energy reliance on fatty acid oxidation. Since MARCH5 is a peroxisome proliferator-activated receptorγ (PPARγ) target gene and appears to regulate adipocyte metabolism¹³, applying pLcar as an OXPHOS substrate donor is relevant.

WT HeLa cells shifted from high glucose to the OXPHOS medium for 6hr (Fig. 3A) show increased levels of OXPHOS proteins, including SDHB (respiratory complex II), COX-2 (complex IV), and NDUFB8 (complex I) (Fig. 3B, C), and peroxisomal proteins PMP70, Pex14 and Catalase (Fig. 3B, C). MARCH5^−/− cells have similarly elevated expression of OXPHOS proteins (Fig. 3B, C), suggesting that growth in the OXPHOS medium induces MARCH5-independent mitochondrial biogenesis. Conversely, the increases in expression of PMP70, Catalase, and Pex14 were significantly lower in MARCH5^−/− cells compared to WT HeLa cells (Fig. 3B, C). The quantification of images also revealed increased peroxisome number in OXPHOS media-treated WT but not in MARCH5^−/− cells (Fig. 3D). We also explored the subcellular localization of MARCH5 in glucose and OXPHOS media-grown cells using cell fractionation. The protocol we use results in the detection of peroxisomal proteins in the mitochondria-enriched heavy membrane (HM) and light membrane (LM) fractions (Fig. 2E, 3E, and Fig. S3J). This pattern could be due to mitochondrial tethering to the peroxisomes³⁹ and/or the high density of some peroxisomes. We generated Pex3^−/−, Pex16^−/−, and Pex26^−/− HeLa cells (Fig. S2). Pex26^−/− cells showed phenotypic mosaicism, with ~60% of the cells lacking Catalase-positive mature peroxisomes (Fig. S3D, H). Consistent with earlier reports^34,40, Pex3^−/− HeLa cells were deficient in peroxisomes (Fig. S2B, F). A similar phenotype was also apparent in Pex16^−/− cells (Fig. S2C, G). The published data indicate the phenotypic mosaicism of Pex16^−/− HeLa cells, with clones ranging from low to high number of cells containing peroxisomes³⁴ indicating that loss of peroxisomes in all Pex16-deficient cells can occur stochastically. Compared to WT cells, the levels of peroxisomal proteins in LM fractions were dramatically reduced in peroxisome-deficient Pex3^−/− and Pex16^−/− cells and only partly affected by Pex26 deficiency (Fig. S2J), validating the applicability of the cell fractionation method for the detection of peroxisomes. Supporting the specificity of these observations, levels of ER protein Calnexin were comparable in LM fractions obtained from all tested cells (Fig. S2J). Cell fractionations also show redistribution of MARCH5 and peroxisomal markers into LM fractions in WT cells cultured in the OXPHOS medium, while consistent with peroxisome deficiency, the shift of Pex14 and MARCH5 into LM fractions was not detected in OXPHOS medium-cultured peroxisome-deficient Pex16^−/− cells (Fig. 3E). Thus, in the OXPHOS medium-treated WT HeLa cells, MARCH5 likely localizes to the newly formed peroxisomes.

Figure 3. MARCH5 controls lipid-induced peroxisome biogenesis.

(A) Experimental protocol. (B) Total cell lysates from WT and MARCH5−/− HeLa cells cultured for 6hr in the high glucose, or OXPHOS media, were analyzed by Western blot, as indicated in the figure. Actin is a loading control. Actin is a loading control. G=Glucose medium, O=OXPHOS medium, (C) Densitometric evaluation of the expression of representative peroxisomal proteins and components of the OXPHOS complexes in cells treated as in B. Values obtained in Glucose medium were taken as 1. Mean±SEM. N=6–10. One-Sample t-test. **** p<0.0001, *** p<0.001, ** p<0.01, n.s.= no significant difference. (D) Number of Pex14-, PMP70- and Catalase-positive peroxisomes/cell in WT and MARCH5^−/− cells treated as in B. In WT cells n= 56 (PMP70), 25 (Pex14), 30 (Catalase) cells, in MARCH5^−/− cells n= 52 (PMP70), 25 (Pex14), and 29 (Catalase). Mean±SEM. One-way ANOVA. **** p<0.0001. n.s.= no significant difference. (E) WT (left) and Pex16^−/− (right) HeLa cells were fractionated into heavy membrane (HM), light membrane (LM), and cytosolic fractions (C), followed by Western blot to detect proteins indicated in the figure. TCL: total cell lysates. Consistent data were observed in 3 independent experiments. Red arrowheads indicate LM fractions.

ATP generation is the primary target and outcome of cellular metabolism. To assess the effect of MARCH5 deficiency on ATP generation in glucose and OXPHOS media, we measured total ATP levels (luciferase-based assay; Fig. S3A,B), maximum ATP generation capacity of the mitochondria (luciferase-based assay⁴¹; Fig. S3C-E), and cytosolic ATP levels (time-lapse imaging of cytosolic ATP sensor cyto-iATPSnFR1.0⁴²; Fig. S3F-J). The cytosolic and total ATP levels were reduced in MARCH5^−/− cells compared to WT HeLa cells cultured in the OXPHOS medium but not in the glucose medium. In addition, the maximum rates of mitochondrial ATP generation were ~25% lower in mitochondria isolated from MARCH5^−/− cells than in WT HeLa cells (Fig. S3D). However, the ATP generation rates in MARCH5^−/− mitochondria from glucose- and OXPHOS-dependent growth conditions show no differences, suggesting that reduced whole-cell ATP in OXPHOS media is likely due to extramitochondrial factor. While we cannot directly attribute this effect to MARCH5’s role in peroxisomes per se, MARCH5 knockout-induced deficiency in peroxisome function likely results in reduced fatty acid processing and thereby decrease in OXPHOS substrates available to the mitochondria. Another possibility is that MARCH5 deficiency could indirectly lead to peroxisome biogenesis inhibition through reduced ATP generation and mitochondrial dysfunction. Indeed, in yeast, de novo peroxisome division and proliferation are inhibited by OXPHOS deficiency⁴³. However, since peroxisome defects are detectable in glucose medium cultured MARCH5^−/− cells (Fig. 2), and under these conditions, total cellular ATP levels are comparable in WT and MARCH5^−/− cells (Fig. S3B), the second scenario is less likely. Consistent with this notion, mitochondrial fusion rates (Fig. S4A,B,C,H,I,L) and proteolytic processing of optic atrophy 1 (Opa1) (Fig. S4N), were not affected by MARCH5 deficiency. Mitochondrial fusion is highly sensitive to changes in mitochondrial activity, including mitochondrial dysfunction^44–46, but also metabolic adaptations to OXPHOS and lipid-fueled ATP generation^47–49. Accumulation of short form (s-Opa1) due to cleavage of the long form of Opa1 (l-Opa1) is another sensitive, non-intrusive marker of mitochondrial functional status^50–52. The MARCH5-independent reduction in mitochondrial fusion rates and Opa1 processing observed in WT and MARCH5^−/− cells cultured in the OPXPHOS medium (Fig. S4I,L,N) is likely an indication of mitochondrial adaptation to the lipid-fueled OXPHOS-dependent ATP generation, which is consistent with mitochondrial fragmentation in model cells relying on lipids for energy production^49,53.

MARCH5 controls mitochondrial pre-peroxisome formation.

Until recently, it was thought that two distinct pathways form peroxisomes: the growth and fission of mature peroxisomes and de novo synthesis at the endoplasmic reticulum (ER)⁵⁴. However, Sugiura et al.²⁵ reported that a subset of pre-peroxisomes (immature peroxisomes) originates from the mitochondria. Peroxisome biogenesis is a multi-step process comprising several biogenesis factors. Mutations or deficiencies of peroxisome biogenesis proteins lead to distinct peroxisomal deficiencies (complete lack of peroxisomes vs partial reduction and defects in maturation)^33–35,40. We reasoned that inactivation of discrete biogenesis factors results in stalling peroxisome biogenesis at diverse steps. MARCH5 knockout in a background of defective peroxisomal biogenesis could unmask biogenesis steps requiring MARCH5 activity and show the epistatic relationship between MARCH5 and specific peroxisomal biogenesis factors. We focused on Pex3 and Pex14, which were proposed to control mitochondrial steps in peroxisome biogenesis²⁵. Pex3^−/− HeLa cells were depleted of peroxisomes, with Pex14 localizing to the mitochondria, Catalase to the cytosol and low levels of cytosolic PMP70 (Figs. S5B, F, J and S5A, B, E). The MARCH5/Pex3 dko cells showed similar peroxisome deficiency, Pex14 and Catalase localization (Fig. S5C-E). If MARCH5 controls peroxisome biogenesis, then the expression of Pex3 should rescue peroxisome deficiency in Pex3^−/− cells but have a lesser effect in MARCH5/Pex3 dko cells. Yellow fluorescent protein (YFP) tagged Pex3 (Pex3-YFP) was transfected into Pex3^−/− and MARCH5/Pex3 dko cells to test this notion. Consistent with Sugiura at al.²⁵, expression of Pex3-YFP in Pex3^−/− cells resulted in gradual recovery of peroxisomes (Figs. 4A, C, D, and S5I). On the other hand, Pex3-YFP was retained on the OMM and failed to rescue peroxisomes in MARCH5/Pex3 dko cells (Fig. 4B, C, D). Cell fractionation studies confirm the imaging data. HM localization of Pex3-YFP and other tested endogenous peroxins was detected in mock-transfected Pex3^−/− and MARCH5/Pex3 dko cells, and Pex3-YFP-transfected MARCH5/Pex3 dko cells, while in Pex3-YFP-transfected Pex3^−/− cells these proteins also accumulate in LM fractions (Fig. 4D). At 72hr after transfection, Catalase recovery was not evident in cell fractionation studies (Fig. 4D). At the same time, Catalase-positive peroxisomes were detected in some Pex3-YFP-expressing Pex3^−/− cells (Fig. S5G), suggesting that Catalase import, and formation of “mature” peroxisomes takes more time. Expression of Pex3-YFP rescues peroxisomal proteins in both Pex3^−/− and MARCH5/Pex3 dko cells (Fig. S5F, J, K), while peroxisomes are only formed in Pex3-YFP-expressing Pex3^−/− cells (Fig. 4C) suggesting that MARCH5 initiates peroxisome formation independently of its potential role in controlling peroxisomal protein stability and their ubiquitination, especially when they accumulate on the mitochondria.

Figure 4. MARCH5 controls the formation of mitochondria-derived pre-peroxisomes.

Pex3^−/− (A) and MARCH5/Pex3 dko (B) HeLa cells transfected with Pex3-YFP were immunostained to detect PMP70 (green) and Pex14 (red) at 24hr (top panels) or 72hr (bottom panels) after transfection. Bars: 10μm, and 1μm (detail images). (C) Subcellular localization of Pex3-YFP in Pex3^−/− and MARCH5/Pex3 dko cells was blindly counted at different time points after transfection, as indicated in the figure. N=4, n=100 cells/condition. Mean±SEM. One-way ANOVA. ****p<0.0001. (D) Pex3^−/− and MARCH5/Pex3 dko cells were subjected to cell fractionation and Western blot at 72hr after transfection with Pex3-YFP, as indicated in the figure. Red arrowheads indicate light membrane (LM) fractions. (E-G) WT (E), Pex14^−/− (F), and MARCH5/Pex14 dko (G) HeLa cells immunostained to detect PMP70 (green), Tom20 (red), and Catalase (single channel image in the right panels). Bars: 10μm, and 1μm (detail images). Fluorescence profiles along the lines drawn between light-blue arrowheads are shown on the right sides of respective detail images. (H) MARCH5/Pex14 dko HeLa cells transfected with MYC-WT MARCH5, immunostained to detect MARCH5 (anti-MYC; red), PMP70 (green), and Tom20 (blue). Bars: 10μm, and 1μm (detail images). Arrowheads indicate PMP70-positive peroxisomes and respective localization of MARCH5 and Tom20. (I) Number of PMP70- positive peroxisomes/cell in WT, Pex14^−/−, MARCH5/Pex14 dko and MARCH5/Pex14 dko HeLa cells reexpressing WT MYC-MARCH5. n=45 (pooled from 3 experiments), and n=19 (pooled from 2 experiments; MARCH5/Pex14 dko+WT MYC-MARCH5). Mean±SEM. One-way ANOVA. ****p<0.0001. (J) Size of PMP70-positive peroxisomes in WT, Pex14^−/− and MARCH5/Pex14 dko HeLa cells reexpressing WT MYC-MARCH5 HeLa cells. n=43 (pooled from 3 experiments), and n=19 (pooled from 2 experiments; MARCH5/Pex14 dko + WT MYC-MARCH5). Mean±SEM. Two-tailed unpaired t-test. **p-value<0.01. (K) Number of cells containing PMP70-labeled peroxisomes in Pex14^−/− cells and MARCH5/Pex14 dko cells, untransfected or transfected with MYC-MARCH5 constructs as indicated in the figure. N=3, n=300. Mean±SEM. One-way ANOVA. **** p<0.0001. n.s.= no significant difference. (L) WT, Pex14^−/−, and MARCH5/Pex14 dko cells were subjected to cell fractionation and Western blot. Red arrowheads indicate LM fractions. (M) Pearson’s colocalization coefficients represent the colocalization of PMP70 with Tom20, or PMP70, as indicated in the figure. Mean±SEM. n=25 cells/condition. One-way ANOVA. **** p<0.0001.

Similar mitochondrial fusion rates (Fig. S4F,G,K,M) and unaltered Opa1 status in untransfected Pex3−/− and MARCH5/Pex3 dko cells (Fig. S4O), and a similar Δψ_m in Pex3-YFP expressing or Pex3YFP-negative MARCH5/Pex3 dko cells (Fig. S4P,Q), at 72hr after transfection, indicates unchanged mitochondrial fitness and suggests a regulatory role of MARCH5 in peroxisome biogenesis.

The epistasis experiments in Pex14^−/− and MARCH5/Pex14 dko cells tested whether MARCH5 is also implicated in Pex14-dependent steps in peroxisome biogenesis. Knockout of Pex14 resulted in a marked reduction in peroxisome number, with a consistent presence of PMP70-positive ~100 pre-peroxisomal structures per cell (Fig. 4F, I). The pre-peroxisomal structures in Pex14^−/− cells also contained peroxisome biogenesis factor Pex26 (Fig. S6A, D), but they were deficient in Catalase, (Fig. 4F) supporting the direct function of Pex14 in Catalase import⁵⁵. Notably, these structures colocalized with the OMM marker Tom20 (Fig. 4F, M and Fig S6A, C, D) but not the IMM protein ATP5A1 (Fig. S6C, D), suggesting that they represent mitochondria-derived pre-peroxisomes, but not mitochondrial fragments per se. The pre-peroxisomes in Pex14^−/− cells were moderately larger than those in WT HeLa cells (Fig. 4J). Progression of z-sections with 0.18μm interval in, other than shown in Fig. 4F, examples of Pex14^−/− cells labeled to detect Tom20 and PMP70 further confirm a strong colocalization, but not mere organelle juxtaposition between these proteins (Fig. S6E). Consistent with MARCH5 acting upstream of Pex14 in mitochondrial pre-peroxisome formation, the data show almost complete absence of peroxisomes in MARCH5/Pex14 dko cells (Figs. 4G, K, and S6B). In these cells majority of PMP70 localized to the mitochondria, with occasional cytosolic puncta, much smaller than pre-peroxisomal structures detected in Pex14^−/− cells (Fig. 4F, G, and S6B). While they may represent other subsets of pre-peroxisomes, these structures do not colocalize with Pex26 (Fig. S6B, D) and require higher imaging power to be detected. The essential role of MARCH5 E3 Ub ligase activity in mitochondrial pre-peroxisome formation is supported by the data showing that ectopic expression of WT MARCH5, but not MARCH5 mutants MARCH5^H43W or MARCH5^ΔCT, rescued pre-peroxisomes in MARCH5/Pex14 dko cells (Figs. 4H, I, K, and S4L). We found that ~70% of the WT MARCH5-expressing MARCH5/Pex14 dko displayed a similar number of PMP70-positive peroxisomes as Pex14^−/− cells (Fig. 4I). The peroxisome size was also comparable (Fig. 4J). However, these larger pre-peroxisomal structures may be formed by membrane sheets that aggregate and represent a clustering of pre-peroxisomes. Therefore, these numbers may underrepresent peroxisome abundance. Furthermore, unlike in WT and MARCH5^−/− cells, where only sporadic WT MYC-MARCH5 colocalization with peroxisomes was detected (Figs. 1A, D), there is a high degree of colocalization of WT MYC-MARCH5 with peroxisomes in MARCH5/Pex14 dko cells (Fig. 4H, M) indicating that Pex14 deficiency leads to stalling of WT MARCH5 on mitochondria-derived pre-peroxisomes. In line with the imaging experiments (Fig. 4E-G, and S6A, B), reduction, or elimination of PMP70 and Pex26 was detected in LM fractions of Pex14^−/− and MARCH5/Pex14 dko cells, respectively (Fig. 4L). Reexpression of WT MARCH5, but not MARCH5^H43W, nor MARCH5^ΔC in MARCH5/Pex14 dko cells partially rescued PMP70 in LM fractions (Fig. S4L). On the other hand, Catalase was not affected by MARCH5 deficiency in Pex14^−/− (Fig. 4G) cells and reexpression in MARCH5/Pex14 dko cells (Fig. S5L). Thus, MARCH5 does not affect Pex14-dependent Catalase import. Like in other tested models, MARCH5 deficiency does not negatively affect mitochondrial functional integrity in Pex14^−/− cells (Fig. S4D,E,J,M,O).

Discussion

Recent evidence indicates that de novo peroxisome biogenesis requires mitochondria^25,26. However, until now, the mechanism and mitochondrial factors implicated in the formation of mitochondrial subsets of peroxisomes have not been determined. We report that the OMM E3 Ub ligase MARCH5 is vital for peroxisome homeostasis and biogenesis of mitochondria-derived pre-peroxisomes. Analysis of peroxisomes in two independent MARCH5 knockout cell models show a reduced number of mature peroxisomes. Supporting the idea that MARCH5 controls peroxisome biogenesis, we found that upon a shift from glycolysis-dependent into fatty acid-supported OXPHOS-dependent growth, peroxisome proliferation occurs in WT but not in MARCH5-deficient cells. Under these growth conditions, MARCH5 translocates to the newly formed peroxisomes. We also found that MARCH5 deficiency inhibits pre-peroxisome accumulation in Pex14^−/− cells and peroxisome formation in Pex3^−/− cells. Pex14 was proposed to control mitochondria-derived pre-peroxisome biogenesis^25,26. Our data support this notion and indicate that Pex14 acts downstream of MARCH5, and that both are likely to be regulated by Pex3.

Limitations of the Study:

Given multiple reported functions of MARCH5 on the mitochondria, including control of the OMM protein stability^7,56, regulation of mitochondrial membrane dynamics^2,57, mitophagy^1,58, and function in removal of exogenously expressed toxic proteins from the mitochondria⁵⁹, one can envision more than one function of this protein in control of peroxisomes, including the reported here role in de novo peroxisome biogenesis and proposed earlier control of pexophagy¹⁴. The critical point is that in all models we use in this work, reduced mitochondrial functional integrity does not appear to be the contributing factor, indicating that the effect of MARCH5 deficiency in peroxisome biogenesis is not merely due to mitochondrial decline but more likely the specific regulatory role of this protein. However, the mechanism, targets of MARCH5, and the scope of mitochondrial contribution to peroxisome biogenesis require further research.

Material and Methods

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Mariusz Karbowski (mkarbowski@som.umaryland.edu).

Material availability

New reagents generated from this study are available from the lead contact upon request with Material Transfer Agreement.

Data and code availability

Any additional information required to reanalyze all the data reported in this paper is available from the lead contact upon request. This paper does not report original code.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Cells: HeLa and HCT116 cells were maintained in Dulbecco Modified Eagle Medium (DMEM; Invitrogen) supplemented with 10% fetal bovine serum (FBS; Sigma), non-essential amino acids (Sigma), sodium pyruvate (Invitrogen), and penicillin/streptomycin (Invitrogen). Cells were maintained in 5% CO₂ at 37°C.

METHOD DETAILS

Knockout cells, DNA constructs, and transfections: MARCH5^−/− HCT116 and Drp1^−/− HeLa cells were reported previously^2,7,62. MARCH5^−/−, Pex14^−/−, Pex16^−/−, Pex26^−/−, MARCH5/Pex14 dko, Pex16/MARCH5 dko and Pex26/MARCH5 dko HeLa cells were generated using double nickase plasmids from Santa Cruz Biotechnology. This system consists of a pair of plasmids each encoding a D10A mutated Cas9 nuclease and a target-specific 20-nt guide RNA (gRNA). The gRNA sequences are as follows:

Pex26 Double Nickase Plasmid (h) (sc-408486-NIC) - A: AGCTCCGCATCCGTGACTGA; B: ATGCAGCCTTCCGGATCCAG

Pex16 Double Nickase Plasmid (h) (sc-411278-NIC)- A: GGATCCTACGGAAGGAGCTT; B: CGTCATTGAGCAGCACAAGC

Pex3 Double Nickase Plasmid (h) (sc-405639-NIC)- A: TCACAGCTCTGCTAAAAAAC; B: GGAATTCAGTTGCTGCATTA

Pex14 Double Nickase Plasmid (h) (sc-405075-NIC)- A: GCCGATGAGCCTTCGTCCTT; B: AGCAGTGCCCGACTGCTGGA

MARCH5 Double Nickase Plasmid (h) (sc-404655-NIC)- A: ACCAGGCCTGTCTACAACGC; B: GTAGATCCTCTGCACCTGCA

Cells grown on 6-well plates were transfected with either single or co-transfected with two plasmids (ratio 1:1). At 48hr after transfection, cells were treated with fresh media containing 3 μg/ml puromycin and grown in puromycin-containing media for 4 days, with daily media change. Then, cells were split into 25-cm cell culture plates and cultured in puromycin-deficient media until colonies were formed, followed by the additional 72-hr culture in puromycin-supplemented media. This treatment eliminated puromycin-sensitive cells carried over after the first puromycin selection, which usually represented ~70% of the colonies. The total time of puromycin selection was 7 days. Puromycin-resistant colonies were sequentially transferred into 24-well plates and then into duplicate 6-well plates. Cells from one 6-well plate were used for knockout verifications^2,7, and from another for propagation. Mammalian expression vectors encoding MYC-tagged dominant negative mutant of Drp1 (MYC-Drp1^K38A), MYC-tagged WT MARCH5, MARCH5^H43W, MARCH5^ΔCT were generated using pCMV-3Tag-7 mammalian expression vector (Agilent Technologies) as a backbone using Gibson assembly, and reported earlier ^2,7,62. For imaging experiments, they were used at 0.2μg DNA per well in 4-well Nunc Lab-Tek II Chambered Coverglass (culture area 1.7cm²; Invitrogen), resulting in low expression of proteins of interest. The cytosolic ATP sensor cyto-iATPSnFR1.0 was a gift from Baljit Khakh (Addgene plasmid# 102550; http://n2t.net/addgene:102550; RRID:Addgene_102550)⁴². Cells were transfected with Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer’s instructions. Cells were used at ~12–20 h after transfection. For Pex3^−/− and MARCH5/Pex3 dko rescue experiments, cells were analyzed at 16hr, 22hr, and 40hr after transfection with YFP-Pex3 plasmid provided by H. McBride (Montreal Neurological Institute) and described by Sugiura et al.²⁵

Cell lysates, cell fractionation, Western blot and Cycloheximide (CHX) chase:

For total-cell lysates, cells were collected by scraping into ice-cold PBS, washed, and suspended in ice-cold PBS. Cell suspensions (100–200μl) were lysed in the same volumes of 2 x SDS sample buffer (Thermo Fisher Scientific) supplemented with 5% β-mercaptoethanol (Millipore) and incubated at 100°C for 10 min, as described ^2,7. Cell fractionation was performed as previously described ^2,62. Cells were washed once with ice-cold PBS and scraped into 15-ml tubes in ice-cold PBS; this was followed by centrifugation at 500 × g for 5 min. The cell pellets were resuspended in ~3 volumes of fractionation buffer (10 mM HEPES, 10 mM NaCl, 1.5 mM MgCl₂, 5mM N-ethylmaleimide (NEM), and protease inhibitors (Sigma). Cells were then passed 15 times through a 25-G needle attached to a 1-ml syringe to disrupt cell membranes. This suspension was centrifuged at 2500 × g at 4°C for 5 min to remove unbroken cells and cell debris. The supernatant was centrifuged at 6000 × g at 4°C for 10 min to pellet the heavy membrane (HM) fraction. The supernatants were centrifuged at 21,000 × g at 4°C for 10 min to pellet the light membrane (LM) fraction. To reduce cytosolic and HM contaminations, the HM and LM fractions were washed in ice-cold PBS supplemented with protease inhibitors (Sigma) and recentrifuged at 8000 × g, and 21,000 × g, respectively. Protein concentrations were measured directly in the samples using a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific). 50μg of protein per sample was separated on 10–20% or 4–12% (Opa1 processing experiments) gradient Novex Tris-glycine polyacrylamide gels (Thermo Fisher Scientific) transferred onto polyvinylidene fluoride membranes (Bio-Rad Laboratories). Membranes were blocked in 5% blocking-grade nonfat dry milk (Bio-Rad Laboratories) in PBS-Tween20 and incubated with primary antibodies overnight at 4°C, followed by horseradish peroxidase-conjugated anti-mouse (Cell Signaling) or anti-rabbit (Cell Signaling) secondary antibodies for 60 min at RT. Blots were developed with Super Signal West Pico ECL (Thermo Fisher Scientific) and imaged using Amersham Imager 600 chemiluminescence imager (GE Healthcare Life Sciences). Primary antibodies used for Western blotting and their dilutions are listed in the Reagents and Resources table. Densitometric evaluations of protein expression were performed using the ImageJ64 image analysis software (NIH, Bethesda, MD), as reported^2,62. For reblotting, membranes were stripped using Restore^™ Western Blot Stripping Buffer (ThermoFisher Scientific) when necessary. For Cycloheximide (CHX) chase cells cultured on 10-cm cell culture plates were treated with fresh culture medium followed by addition of 100μg/ml of freshly prepared CHX (100mM) stock in ethanol. Cells were collected over time for up to 8hr; total cell lysates were prepared for Western blotting as described above.

Immunofluorescence:

Immunofluorescence was performed as previously described^2,63. Briefly, cells grown in 2- or 4-well chamber slides (model 1 German borosilicate; Lab-Tek; VWR) were fixed with freshly prepared 4% formaldehyde in PBS solution (using 16% Methanol-free Formaldehyde; Thermo Fisher Scientific) for 20 min at RT, then permeabilized with permeabilization buffer (PB; 0.15% Triton X-100 in PBS) for 20 min at RT. After blocking with blocking buffer (BB; PB supplemented with 7.5% BSA) for 45 min, samples were incubated with primary antibodies suspended in BB for 90 min at RT, followed by 3 washes with BB and incubation with secondary antibodies diluted in BB for 60 min at RT. Primary antibodies and their dilutions are listed in the Reagents and Resources table. Secondary antibodies were highly cross-absorbed goat anti-mouse Alexa Fluor-488 and goat anti-rabbit Alexa Fluor-546 (1:1000; both from Thermo Fisher Scientific). In triple labelling experiments, cells were immunostained, as above, followed by 3 washes with BB and incubation with anti-Tom20 antibodies conjugated with Alexa 647 fluorophore (Santa Cruz Biotechnology Inc.; 1:100) for 90 min at RT. After 3 washes with PBS, cells were subjected to Airyscan imaging. Immunolabeled cells on 4-well chamber slides were stored in PBS at 4°C and imaged within 10 days after processing.

Image acquisition, analysis, and processing:

Images were acquired with a Zeiss LSM 880 microscope (Zeiss MicroImaging) equipped with an Airyscan superresolution imaging module, using 63/1.40 Plan-Apochromat Oil DIC M27 objective lens (Zeiss MicroImaging), as described ^2,62. The 488-nm Argon laser line, 561-nm DPSS 561 laser, and 633-nm HeNe 633 laser were used to detect Alexa-488, Alexa-546, and Alexa-647, respectively. For image quantifications the z-stacks covering the entire depth of cells with intervals of 0.018μm were acquired, followed by Airyscan image processing (set at 6) and generation of maximum intensity projection images. The lateral and axial resolution of the resulting images was ~120nm and ~320nm, respectively. For presentation in the manuscript the images were processed using Zeiss MicroImaging Joint Deconvolution method, resulting in generation of 90nm lateral resolution images. Time-lase images were processed using Airyscan processing mode of ZEN software (XXXX). Image analyses were done using ZEN software (version 2.3SPL, Zeiss MicroImaging) and ImageJ (NIH, Bethesda, MD) image acquisition and processing software. Colocalization/overlap, indicated by Pearson’s correlation coefficient (R), was determined from maximum intensity projections of the Airyscan processed images using a “colocalization” function of the ZEN software. For quantifications of fluorescence profiles, single z-section Airyscan-processed images were used. The profiles along peroxisomes were generated from the 2-channel images using the “profile” option available in the ZEN software. For peroxisome number and size, single cells in maximum-intensity projection Airyscan images were outlined, cropped, and converted into binary images using ImageJ, and analyzed using the “particle analysis” module of the software. The data were tabularized and transferred to Microsoft Excel software (Microsoft) for further analyses. Image cropping and global brightness and contrast adjustments were performed for presentation using Adobe Photoshop CS6 software (Adobe Systems).

Live cell imaging:

Cells were grown on 2- or 4-well chamber slides (model 1 German borosilicate; Lab-Tek; VWR) to 30–50% confluency. In all time-lapse experiments chamber slides were mounted on the environmental control chamber (Stagetop TIZW Series, Neco Incubation System with sensor feedback system) attached to Zeiss LSM 880 microscope (Zeiss MicroImaging) set at 5% CO₂ and 37°C and imaged in Phenol Red-free DMEM, 10% fetal bovine serum (FBS; Sigma), non-essential amino acids (Sigma), penicillin/streptomycin (Invitrogen), supplemented with 0–25mM glucose, galactose, or 25μm Palmitoyl-DL-carnitine, as described in the text. To avoid z-section shift, focus was maintained using “definitive focus” autofocusing system.

Cytosolic ATP:

Cells were transfected with cytosolic ATP sensor cyto-iATPSnFR1.0 <14hr, before analyses. Cells were imaged for up to 6hr, with 5min interval between images.

mito-PAGFP:

Cells were transfected 18–24hr, before analyses. The imaging was set that two pre-activation images were followed by photoactivation of the ~1-μm-diameter circular region of interest with 405-nm laser (100%; 5 iterations; Coherent Enterprise Ion Laser 80.0mW), followed by time-lapse imaging of five z-sections (interval 0.18μm) with a 488-nm excitation light for ~24min, every 20sec. For quantification of the mitochondrial fusion (reduction of mito-PAGFP fluorescence due to its transfer from “photoactivated” to “non-photoactivated mitochondria^7,60,61), the orthogonal projections of Airyscan-processed images were used.

Mitochondrial membrane potential (Δψ_m).

Cells were grown on 2-well chamber slides (model 1 German borosilicate; Lab-Tek; VWR) to 60–80% confluency. They were incubated with 50nM TMRM-containing media for 30 min, washed with phenol red-free complete DMEM media, and imaged, as described above. Five z-sections with an interval of 0.18μm were acquired and processed using standard Airyscan processing mode (Zeiss MicroImaging). Images were projected using “orthogonal projection” mode and fluorescence of TMRM within 4μm diameter circular ROIs in cells expressing Pex3-YFP (detected with 488-nm), and untransfected cells were measured. The highest fluorescence value in untransfected cells within each image was taken as 1, fluorescence from other cells in the same image was calculated accordingly.

Total and mitochondrial ATP measurements:

For total ATP measurements, cells were grown overnight in 10-cm cell culture plates in 25mM glucose medium to 80% confluence. Cells were washed and incubated in the starvation medium (DPBS) for 60min to reduce cellular ATP ⁶⁴ followed by growth for 6hr in 5% CO₂ at 37°C in DPBS anew, 4.5mM galactose medium, 25mM glucose medium, and 4.5mM galactose media supplemented with 25μM pLcar, and 25mM glucose supplemented with 25μM pLcar. To inhibit mitochondrial ATP generation, each condition (except the DPBS condition) was also subjected to treatment with 5μM oligomycin (Sigma). ATP extractions were performed using a protocol described by Yang et al. ⁶⁵, with modification. Cells were collected, transferred to 2ml Eppendorf tubes, and exposed to boiling deionized water containing 0.5% Triton-X 100 (Sigma). Denatured protein clumps were broken down by 10 passages through a 25 G x 5/8 needle attached to 1 ml syringe. Samples were centrifuged at 20,000 x g for 10 min, at 4°C. The supernatants were then transferred to a new tube and used in measurements. Pellets were solubilized in 2 x SDS PAGE sample buffer supplemented with 5% β-mercaptoethanol, incubated at 100°C for 10 min, and centrifuged at 20,000 x g for 5 minutes at 4°C. The supernatants were transferred to new tubes and used for protein measurements. Mitochondrial ATP measurements were performed as described⁶⁶. Cells were grown to confluency on 15-cm plates, the growth media was discarded, and the plates were washed twice with ice-cold PBS supplemented with 1 mM EGTA (pH 7.4). The cells were then harvested using a cell scraper, transferred to 50 mL conical tubes, and centrifuged at 500 x g to pellet the cells and discard the PBS solution. Cells were then resuspended in ice-cold isolation buffer (IB) containing (in mM): KCl 100, MOPS 50, MgSO₄ 5, EGTA 2, K₂HPO₄ 10. The cell suspension was washed twice with IB by centrifugation at 500 x g to pellet the cells and discard the supernatant. The remainder of the preparation was conducted in a cold room (4°C). 10 ml of IB containing cells underwent 15 repetitive homogenizations with a 1-μm clearance pestle at low speed. The homogenate was centrifuged for 8 min at 600 x g at 4°C, after which the supernatant was transferred to a new centrifuge tube. These homogenizations of the cell pellet and centrifugation to pellet cells were repeated 3 more times. The supernatant from four homogenizations and centrifugation was pooled together, transferred to new centrifuge tubes, and spun at 600 x g for 8 min at 4°C to pellet cell debris The supernatant was transferred to a clean centrifuge tube and spun at 600 x g again to pellet cell debris for 8 min. The supernatant was transferred to a clean centrifuge tube and spun at 3200 x g to pellet mitochondria for 12 min. The supernatant was discarded, and the pellet (the mitochondria sample) was resuspended in IB warmed to 30°C and supplemented with Na-Pyruvate (10 mM). After 10 min incubation at room temperature, the suspension was spun at 3200 x g for 12 min to pellet the mitochondria. The supernatant was discarded, and the pellet (the mitochondria sample) was resuspended in ice-cold IB supplemented with Na-Pyruvate (10 mM) and kept on ice for 30 min and centrifuged at 3200 x g for 12 min to pellet mitochondria. The pellet was resuspended in ice cold resuspension buffer (RB1) base solution containing (in mM): KCl 100, MOPS 50, K₂HPO₄ 1, supplemented with Na-Pyruvate (10 mM), EGTA (40 μM). The mitochondria were centrifuged at 3200 x g at 4°C for 12 min and resuspended in RB2, which is RB supplemented with Na-Pyruvate (1 mM) and EGTA (40 μM). The mitochondria were centrifuged at 3200 x g for 12 min, and a final resuspension and pelleting were done using RB3, consisting of RB and EGTA (40 μM). The concentration of mitochondria was quantified by Lowry assay. The high purity of mitochondria isolated via this procedure was previously shown ⁶⁶. Mitochondria were used within 4 hours of isolation. Measurements of total cellular ATP and mitochondrial ATP production rate were carried out using a BMG LABTECH CLARIOstar plate reader. Mitochondria (0.1 mg per mL) were mixed in ATP production assay buffer (AB) consisting of (in mM): K-Gluconate 130, KCl 5, K₂HPO₄ 1 or 10, MgCl₂ 1, HEPES 20, EGTA 0.04, BSA 0.5 mg/mL, D-Luciferin (Sigma) 0.005, Luciferase (Sigma) 0.001 mg/mL, pH 7.2. A luminescence standard curve was performed daily over a range of 100nM to 1 mM ATP with Oligomycin A (15 μM) treated mitochondria. The mitochondria were supplemented for 2 minutes prior to the start of the assay in 1 mM Pyruvate and 0.5 mM Malate, 1 mM Glutamate and 0.5 mM Malate, or 0.1 mM palmitoylcarnitine + 2.5 mM malate. Assays were initiated by injection of 20 μL ADP (5.0 mM) and 80 μL luciferin/luciferase in AB to bring the final volume to 200 μL. Luminescence signal was recorded for 20 seconds with 1 second integration. ATP production rates were scaled to nMol ATP per sec per mg mitochondrial protein (nMol s⁻¹ mg⁻¹).

QUANTIFICATION AND STATISTICAL ANALYSIS

Unless otherwise indicated, data was analyzed using unpaired, two-tail Student’s test; for multiple-group comparisons, data were analyzed using one-way ANOVA with Bonferroni correction (α=0.05) or Sidak correction (confirmed by Bonferroni) to access significance. All error bars are expressed as SEM. P-values < 0.05 were considered significant. “N” indicates independent experiments; “n” indicates repeats within each experiment.

Key resources table.

REAGENT/RESOURCE	SOURCE	IDENTIFIER
Antibodies
Primary:
anti-MARCH5 polyclonal Ab (WB: 1:2500)	Cell Signaling	Cat# 19168
anti-Pex14 polyclonal Ab (WB: 1:4000; IF: 1:500)	Protein Technology	Cat# 10594-1-AP
anti-Pex16 polyclonal Ab (WB: 1:2500)	Protein Technology	Cat# 14816-1-AP
anti-Pex26 polyclonal Ab (WB: 1:2500; IF: 1:250)	Protein Technology	Cat# 27472-1-AP
anti-Pex3 polyclonal Ab (WB: 1:2500)	Sino Biological Inc.	Cat#200628-T44
anti-Pex19 polyclonal Ab (WB: 1:2000)	Protein Technology	Cat# 14713-1-AP
anti-Catalase polyclonal Ab (WB: 1:2500)	Cell Signaling	Cat# 12980
anti-Catalase mAb (IF: 1:500)	Invitrogen	Cat# 702955
anti-Pmp70 mAb (cl2524) (WB: 1:2500; IF: 1:500)	Sigma	Cat# SAB4200181
anti-Calnexin polyclonal Ab (WB: 1:20,000)	Protein Technology	Cat# 10427-2-AP
anti-b-actin polyclonal Ab (WB: 1: 5000)	Cell Signaling	Cat# 3700
anti-Tom20 polyclonal Ab (WB: 1: 10,000)	Protein Technology	Cat#11802-1-AP
anti-Tom20-Alexa647 (IF: 1: 100)	Santa Cruz Biotechnology	Cat# 17764 AF647
anti-human OXPHOS mAb (WB: 1: 2500)	Invitrogen	Cat# 458199
anti-Mcl1 polyclonal Ab (WB: 1: 10,000)	Protein Technology	Cat# 16225-1-AP
anti-ATP5A1 polyclonal Ab (IF: 1:1000)	Protein Technology	Cat# 14676-1-AP
anti-cytochrome c mAb (IF: 1:1000)	BD Biosciences	Cat# 556432
anti-Opa1 (clone 18) mAb (WB: 1: 2500)	BD Biosciences	Cat# 612606
Secondary:
anti-mouse IgG, HRP-linked	Cell Signaling	Cat# 7076S
anti-rabbit IgG, HRP-linked	Cell Signaling	Cat# 7074S
Goat anti-mouse Igg (h+l) highly cross-adsorbed, Alexa Fluor 488	Invitrogen	Cat# A11029
Goat anti-rabbit Igg (h+l) cross-adsorbed, Alexa Fluor 488	Invitrogen	Cat# A11008
Goat anti-mouse Igg (h+l) Highly Cross-adsorbed, Alexa Fluor 546	Invitrogen	Cat# A11030
Goat anti-rabbit Igg (h+l) highly cross-adsorbed, Alexa Fluor 546	Invitrogen	Cat# A11035
Chemicals and recombinant proteins
ADP, Adenosine 5′-diphosphate monopotassium salt dihydrate	Sigma	Cat# A5285
ATP, Adenosine 5’-triphosphate Disodium Salt Hydrate	Sigma	Cat# A2383
Tetramethylrhodamine, Methyl Ester, Perchlorate (TMRM)	Invitrogen	Cat# T668
Luciferase	Sigma	Cat# SRE0045
Luciferin	Sigma	Cat# L9504
Palmitoyl-DL-carnitine (chloride)	Cayman Chemical	Cat# 11095
Oligomycin	Sigma	Cat# 495455
MG132	Sigma	Cat# 474787
Bortezomib	Sigma	Cat# 5043140001
NMS-873	Selleck Chemical	Cat# S7285
Cycloheximide	Sigma	Cat# C7698
16% Formaldehyde (w/v)	Invitrogen	Cat# 28908
Triton X-100	Sigma	Cat# T8787
Critical commercial assays and supplies
Novex Tris-Glycine Mini Protein Gels, 10–20%, 1.0 mm	Invitrogen	Cat# XP10200
Novex Tris-Glycine Mini Protein Gels, 4–12%, 1.0 mm	Invitrogen	Cat# XP04120
Complete Mini EDTA-free Protease Inhibitor Cocktail Tablets	Sigma	Cat# 11836170001
Supersignal West Pico Plus	Invitrogen	Cat# 34580
Blotting-Grade Blocker	Bio-Rad	Cat# 1706404
Restore™ Western Blot Stripping Buffer	ThermoFisher Scientific	Cat#21059
Immuno-Blot PVDF Membrane	Bio-Rad	Cat# 1620177
DMEM, High Glucose	Invitrogen	Cat# 11965118
DMEM, -Glucose, -Sodium Pyruvate	Invitrogen	Cat# A1443001
FBS	Sigma	Cat# F4135
Sodium Pyruvate	Invitrogen	Cat# 11360070
Non-essential AAs	Sigma	Cat# SIG-M7145
Penicillin Streptomycin Solution	Invitrogen	Cat# 15140122
Puromycin Dihydrochloride	Sigma	Cat# P8833
Dulbecco’s Phosphate Buffered Saline (with Ca2+ and Mg2+)	Genesee Scientific	Cat# QBI 114-059-101
Experimental models
Cell lines
HeLa	ATCC	N/A
Drp1−/− HeLa	Karbowski lab; Cherok et al.2	N/A
HCT116	ATCC	N/A
MARCH5−/− HCT116	Karbowski lab; Xu et al.7	N/A
Recombinant DNA
MARCH5 Double Nickase Plasmid	Santa Cruz Biotechnology	Cat# 404655-NIC
Pex14 Double Nickase Plasmid	Santa Cruz Biotechnology	Cat# 405075-NIC
Pex16 Double Nickase Plasmid	Santa Cruz Biotechnology	Cat# 411278-NIC
Pex3 Double Nickase Plasmid	Santa Cruz Biotechnology	Cat# 405639-NIC
Pex26 Double Nickase Plasmid	Santa Cruz Biotechnology	Cat# 408486-NIC
Pex3-YFP	Provided by H. Mc Bride (McGill University, Canada); Sugiura et al.25	N/A
cyto-iATPSnFR1.0	Addgene; Lobas et al.42	Addgene plasmid# 102550
MYC-WT MARCH5	Karbowski lab; Cherok et al.2	N/A
MYC-WT MARCH5H43W	Karbowski lab; Cherok et al.2	N/A
MYC-WT MARCH5ΔCT	Karbowski lab; Cherok et al.2	N/A
mito-PAGFP	Addgene; Karbowski et al.60,61	Addgene plasmid #23348
Software
Zen version 2.3SPL (regular Airyscan image processing)	Zeiss MicroImaging	N/A
Zen version 3.9 (90-nm joint deconvolution processing)	Zeiss MicroImaging	N/A
ImageJ	NIH, Bethesda, MD	N/A
Photoshop v 23.5	Adobe	N/A
Excel	Microsoft	N/A
Prism	GraphPad	N/A
Other
4-well Nunc Lab-Tek II Chambered Coverglass	Invitrogen	Cat# 155382
Immuno-Blot PVDF Membrane	Bio-Rad	Cat# 1620177

HIGHLIGHTS:

• Mitochondrial proteins required for de novo peroxisome biogenesis investigated

• E3 Ub ligase MARCH5 controls peroxisome homeostasis

• MARCH5 is vital for peroxisome biogenesis independently of its role in quality control

• Pex3 is critical for MARCH5 function in peroxisomes