Functional Peroxisomes Are Essential for Efficient Cholesterol Sensing and Synthesis

Cholesterol biosynthesis is a multi-step process involving several subcellular compartments, including peroxisomes. Cells adjust their sterol content by both transcriptional and post-transcriptional feedback regulation, for which sterol regulatory element-binding proteins (SREBPs) are essential; such homeostasis is dysregulated in peroxisome-deficient Pex2 knockout mice. Here, we compared the regulation of cholesterol biosynthesis in Chinese hamster ovary (CHO-K1) cells and in three isogenic peroxisome-deficient CHO cell lines harboring Pex2 gene mutations. Peroxisome deficiency activated expression of cholesterogenic genes, however, cholesterol levels were unchanged. 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) protein levels were increased in mutant cells, whereas HMGCR activity was significantly decreased, resulting in reduced cholesterol synthesis. U18666A, an inhibitor of lysosomal cholesterol export, induced cholesterol biosynthetic enzymes; yet, cholesterol synthesis was still reduced. Interestingly, peroxisome deficiency promoted ER-to-Golgi SREBP cleavage-activating protein (SCAP) trafficking even when cells were cholesterol-loaded. Restoration of functional peroxisomes normalized regulation of cholesterol synthesis and SCAP trafficking. These results highlight the importance of functional peroxisomes for maintaining cholesterol homeostasis and efficient cholesterol synthesis.


INTRODUCTION
Cholesterol is an essential lipid constituent of cellular membranes, in particular the plasma membrane, and an obligatory precursor for synthesis of steroid hormones, bile acids, and regulatory oxysterols, as well as a requisite ligand for activation of Hedgehog family proteins. Its synthesis involves more than 20 enzymes distributed over several subcellular compartments. Numerous studies assign the pre-squalene steps of cholesterol biosynthesis to peroxisomes ( Figure 1A) [reviewed in Kovacs et al. (2002), Faust and Kovacs (2014)]. With the exception of HMGCR, the enzymes of the pre-squalene segment contain functional peroxisomal targeting signals (PTS) that mediate their import into the peroxisomal matrix (Kovacs et al., 2002;Faust and Kovacs, 2014). We showed that acetyl-CoA derived from peroxisomal β-oxidation of very long-chain fatty acids (i.e., [1,2,3,4-13 C 4 ]docosanoate) and dicarboxylic acids (i.e., [U-13 C 12 ]dodecanedioate) is channeled to cholesterol synthesis inside peroxisomes (Kovacs et al., 2007). The enzymes of the pre-squalene segment of the isoprenoid biosynthetic pathway are also localized to peroxisomes in plants [reviewed in Faust and Kovacs (2014), Henry et al. (2018)]. Furthermore, several enzymes of the isoprenoid biosynthetic pathway are localized within peroxisome-related microbodies called glycosomes in trypanosomatid parasites (Concepcion et al., 1998;Urbina et al., 2002;Colasante et al., 2006;Carrero-Lérida et al., 2009;Güther et al., 2014;Ferreira et al., 2016;Acosta et al., 2019).
Using the peroxisome-deficient Pex2 knockout (Pex2 −/− ) mouse we have shown the importance of peroxisomes for the maintenance of normal cholesterol homeostasis (Kovacs et al., 2004(Kovacs et al., , 2009(Kovacs et al., , 2012. Total cholesterol levels were similar in the brain, kidney, spleen, heart, and lung of control (Pex2 +/+ and Pex2 +/− , hereafter referred to as Pex2 +/ ) and Pex2 −/− mice, however, hepatic cholesterol levels in 9-day-old (P9) Pex2 −/− mice were decreased by 40%, relative to controls, while bile acid (BA) feeding normalized their hepatic cholesterol levels. HMG-CoA reductase (HMGCR), isopentenyl diphosphate isomerase 1 (IDI1), farnesyl diphosphate synthase (FDPS), and squalene synthase (FDFT1) activities and protein levels were significantly increased in the liver of P9-P10 Pex2 −/− mice, while BA feeding attenuated these enzyme activities and protein levels. Interestingly, the increase in protein levels was significantly greater than the increase in the activities of these enzymes, and the "catalytic efficiency" of HMGCR and IDI1 was significantly decreased in newborn (P0) and untreated as well as BA-fed postnatal Pex2 −/− mice (Kovacs et al., 2004(Kovacs et al., , 2009(Kovacs et al., , 2012. Notably, a different regulatory pattern of cholesterol enzyme activities was observed in kidneys from Pex2 −/− vs. control mice: the activity of HMGCR was significantly decreased, whereas activities of other cholesterol biosynthetic enzymes (i.e., IDI1, FDPS, FDFT1) were increased (Kovacs et al., 2004(Kovacs et al., , 2009(Kovacs et al., , 2012. Using [ 3 H]acetate as substrate, we showed that the rate of cholesterol synthesis in the liver, spleen, lung, and heart of P9 Pex2 −/− mice was significantly increased compared to controls, which is probably due to the highly increased FIGURE 1 | (A) Subcellular localization of the cholesterol biosynthetic pathway in mammalian cells. Blue fonts indicate SREBP-2-regulated cholesterol biosynthetic enzymes. With the exception of HMGCR, all enzymes for the conversion of acetyl-CoA to farnesyl diphosphate contain functional peroxisomal targeting signals. Acetyl-CoA derived from peroxisomal β-oxidation of very long-chain fatty acids and dicarboxylic acids is channeled to cholesterol synthesis inside peroxisomes (Weinhofer et al., 2006;Kovacs et al., 2007). Farnesyl diphosphate leaves the peroxisomes and is converted either to cholesterol in the ER or to nonsterol isoprenoids. The conversion of acetyl-CoA to HMG-CoA also occurs in the cytoplasm. The mitochondrial and peroxisomal acetoacetyl-CoA thiolase and the cytosolic and peroxisomal HMG-CoA synthase are encoded by single genes, respectively. In hepatocytes, the conversion of cholesterol to bile acids also partly takes place in peroxisomes. The primary C 24 -bile acids are formed from C 27 -bile acid intermediates by peroxisomal β-oxidation of the side chain, followed by conjugation of the C 24 -bile acids to glycine or taurine by the peroxisomal enzyme BAAT. Figure modified from Figure 1 in Faust and Kovacs (2014). (B) Regulation of the SREBP-2 pathway. The precursor SREBP-2 is embedded as inactive transcription factor in ER membranes. In the presence of cholesterol and hydroxysterols (e.g., 25-hydroxycholesterol, 27-hydroxycholesterol), the SCAP-SREBP-2 complex is retained in the ER by interaction with INSIG proteins. Whereas cholesterol binds to SCAP and induces a conformational change in SCAP through which it binds to INSIGs, hydroxysterols bind to INSIGs and cause INSIGs to bind to SCAP. In the absence of sterols, INSIGs dissociate from SCAP, whereupon the INSIGs are ubiquitinated and degraded in proteasomes. The SCAP-SREBP-2 complex is loaded into COPII-coated vesicles and transported to the Golgi. In the Golgi SREBPs are processed sequentially by Site-1 and Site-2 proteases to release the N-terminal transcription factor domain that then enters the nucleus and activates SRE (sterol regulatory element)-containing genes. protein levels and activities of cholesterol biosynthetic enzymes in the tissues of Pex2 −/− mice (Kovacs et al., 2004(Kovacs et al., , 2009. Interestingly, the rate of cholesterol synthesis was significantly decreased in the brain and kidneys of P9 Pex2 −/− mice (Kovacs et al., 2004(Kovacs et al., , 2009. The hepatic expression of SREBP-2 target genes was significantly increased both in P0 and postnatal Pex2 −/− mice. BA feeding reduced the expression of these genes significantly; yet, the mRNA levels were still increased (Kovacs et al., 2004(Kovacs et al., , 2009(Kovacs et al., , 2012. In addition, we showed that peroxisome deficiency activates hepatic ER stress pathways in Pex2 −/− mice, especially the integrated stress response which is mediated by PKR-like endoplasmic reticulum kinase (PERK) and activating transcription factor-4 (ATF4) signaling; hence, we hypothesized that ER stress leads to dysregulation of the endogenous sterol response mechanism and SREBP-2 pathway induction (Kovacs et al., 2009(Kovacs et al., , 2012Faust and Kovacs, 2014).
Conflicting reports on the role of peroxisomes in cholesterol biosynthesis have been published from in vitro studies using fibroblasts from patients with peroxisomal biogenesis disorders as well as peroxisome-deficient rodent cells [reviewed in Kovacs et al. (2002)]. In summary, activities of peroxisomal cholesterol biosynthetic enzymes were either normal or decreased in fibroblasts of patients with Zellweger spectrum disorders. Three studies employing a total of 24 fibroblast cell lines from patients with disorders of the Zellweger spectrum demonstrated a significantly reduced rate of cholesterol biosynthesis compared to control cells, whereas two studies using a total of seven fibroblast cell lines found that cholesterol biosynthesis rates were similar or higher than those in control fibroblasts. HMGCR activity and the rates of both cholesterol and non-sterol (dolichol) biosynthesis were found to be significantly lower in a study using peroxisome-deficient Pex2 −/− hamster cells (ZR-78.1C, ZR-82) (Aboushadi and Krisans, 1998), while in another study using ZR-82 cells, both HMGCR activity and the incorporation of [ 3 H]acetate into sterols (cholesterol + lanosterol) were increased compared to wild-type CHO-K1 cells (van Heusden et al., 1992). However, in vitro studies have not investigated the regulation of cholesterol biosynthesis in peroxisome-deficient cells.
In this study, we investigated the transcriptional regulation of cholesterol biosynthesis in wild-type CHO-K1 cells, three isogenic peroxisome-deficient CHO cell lines (ZR-78.1C, ZR-82, ZR-87) with mutations in the Pex2 gene, and ZR cells with restored functional peroxisomes after complementation with wild-type Pex2 cDNA. We also determined the rate of cholesterol biosynthesis as well as activities and protein levels of cholesterol biosynthetic enzymes in these cell lines. Finally, we explored mechanisms that might lead to a dysregulation of the endogenous sterol response in peroxisome-deficient CHO cells.

Peroxisome-Deficient CHO Cells
The Pex2 mutant and peroxisome-deficient CHO cell lines  were isolated from the CHO-K1 cell line used as control in this study (Zoeller and Raetz, 1986;Zoeller et al., 1989). PEX2 is anchored to the peroxisomal membrane by two membrane-spanning segments, with its N-and C-terminal regions exposed to the cytosol (Harano et al., 1999). The C-terminus of PEX2 contains a RING finger (C 3 HC 4 ) motif. The point mutations of Pex2 in the ZR-78.1C and ZR-82 cell lines have been identified, whereas the mutation of Pex2 in the ZR-87 cell line is not known (Thieringer and Raetz, 1993). In ZR-78.1C nucleotide G at position 737 was mutated to A, resulting in the conversion of a cysteine residue into a tyrosine residue in the RING finger motif. The mutation in ZR-82 cells introduces a stop codon which leads to the translation of a truncated form of the PEX2 protein. This N-terminal fragment constitutes only one-fifth of the entire protein and lacks both membranespanning regions.
To assess if a complete or partial absence of peroxisomes exist in our cell lines, an immunofluorescence analysis was performed. The immunofluorescence pattern obtained for acyl-CoA oxidase 1 (ACOX1), a peroxisomal matrix protein involved in peroxisomal fatty acid β-oxidation, showed the characteristic punctuate peroxisomal distribution in CHO-K1 cells (Figure 2A) and a diffuse, cytoplasmic fluorescence in peroxisome-deficient CHO cells (Figure 2A), consistent with mislocalization of ACOX1 to the cytoplasm. A punctuate peroxisomal staining pattern for the peroxisomal membrane proteins PEX14 and ACBD5 was observed in CHO-K1 cells (Figures 2B,C). In peroxisome-deficient CHO cells, PEX14 and ACBD5 were present in less abundant cellular vesicles, consistent with peroxisome membrane ghosts (Figures 2B,C). These findings are consistent with the established function of PEX2 in the import of peroxisomal matrix proteins, but not peroxisomal membrane proteins.
Acyl-CoA oxidase 1 is synthesized as a precursor protein of 72 kDa and undergoes proteolytic cleavage after import into peroxisomes, resulting in subunits of 52 and 20.5 kDa (Miyazawa et al., 1987). In peroxisomedeficient CHO cells only a slight band of the uncleaved precursor protein was seen ( Figure 2D), consistent with lack of peroxisomal protein import and rapid degradation outside of peroxisomes (Schram et al., 1986;Zoeller et al., 1989). Interestingly, the protein levels of ACBD5 were not significantly altered in peroxisome-deficient cells (Figure 2D), whereas PEX14 levels were decreased in ZR-78.1C and ZR-87 cells ( Figure 2D).
In order to restore functional peroxisomes, we transfected the peroxisome-deficient CHO cells with wild-type rat Pex2 cDNA and stably selected transfected cells with G418. Transfection with Pex2 restored peroxisomal matrix protein import, since a normal punctate fluorescence pattern was observed with ACOX1 and catalase staining (Figures 2E,F; cf. also Supplementary  Figures S1A,B). The peroxisomes of the Pex2-transfected mutant cell lines were morphologically indistinguishable from peroxisomes found in the wild-type CHO-K1 and Pex2transfected CHO-K1 cells. Hereafter we refer to these transfected cells as ZR-78.1C-Pex2, ZR-82-Pex2, and ZR-87-Pex2. Frontiers in Cell and Developmental Biology | www.frontiersin.org FIGURE 2 | Continued to peroxisome membrane ghosts in peroxisome-deficient CHO cells. (D) Immunoblots of total cell lysates with antibodies against peroxisomal matrix and membrane proteins. Numbers at the bottom of the blots indicate the fold change in protein levels in peroxisome-deficient cells relative to that in CHO-K1 cells, which were arbitrarily defined as 1. (E,F) Functional peroxisomes are restored in peroxisome-deficient CHO mutants (ZR-78.1C, ZR-82, ZR-87) upon complementation with rat Pex2 cDNA. Cells were immunostained with antibodies against the peroxisomal matrix proteins ACOX1 (E) and catalase (F). The nuclei were stained with DAPI (blue). An asterisk indicates non-transfected cells without import-competent peroxisomes. The scale bars represent 10 µm. The indicated plasmids were transfected into cells as described in Materials and Methods. 24 hours after transfection, the medium was switched to medium containing either 10% FCS or 5% LPDS and incubated for 24 h, after which the cells were harvested and assayed for dual luciferase activities as described in section "Materials and Methods." (B) The indicated plasmids were transfected into cells and 24 hours after transfection the medium was switched to medium supplemented with 5% LPDS and 0, 1, or 10 µg/ml 25-hydroxycholesterol (25-HC). After 24 h incubation, cells were harvested and assayed for dual luciferase activities. Data are presented as firefly luciferase activity normalized to Renilla luciferase activity and each value represents the mean ± SD of three independent transfection experiments done in quadruplicate. Statistical analysis was performed using Student's t-test or Student's t-test with Welch's correction or one-way ANOVA followed by Dunnett's multiple comparisons test. *** p < 0.001 vs. FCS-cultured CHO-K1. ## p < 0.01; ### p < 0.001 vs. LPDS-cultured CHO-K1. § p < 0.05; § § p < 0.01; § § § p < 0.001 vs. corresponding FCS-cultured cell line. ÇÇ p < 0.01; ÇÇÇ p < 0.001 vs. corresponding cell line incubated in medium containing 5% LPDS without 25-HC.

Transcriptional Activity of Cholesterogenic Promoter-Reporters Is Increased in Peroxisome-Deficient CHO Cells
In order to determine whether sterol-regulated transcription of cholesterogenic genes is affected in peroxisome-deficient CHO cells, luciferase reporter constructs containing the sterol response element (SRE) of SREBP-2 (Srebf2), HMG-CoA reductase (Hmgcr), farnesyl diphosphate synthase (Fdps), or squalene synthase (Fdft1) were transiently transfected into wild-type CHO-KI and peroxisome-deficient ZR-82 and ZR-87 cells. For this cells were incubated in medium supplemented with either 10% FCS or 5% lipoprotein-deficient serum (LPDS), the latter a condition known to upregulate the SREBP-2 pathway. Expression of the Srebf2 and Fdps luciferase reporter genes was increased significantly in all cell lines when incubated in medium with LPDS ( Figure 3A), indicating the responsiveness of SREBP to low cholesterol levels. However, Fdps luciferase activity was higher in peroxisome-deficient cells compared with CHO-K1 cells. It is interesting to note that the expression of the Hmgcr, Fdps, and Fdft1 luciferase reporter constructs in peroxisome-deficient cells was significantly increased compared with CHO-K1 even when the cells were incubated in medium containing 10% FCS and therefore sufficient amount of extracellular cholesterol was available ( Figure 3A). The expression of the Hmgcr luciferase reporter constructs was only slightly enhanced further in peroxisome-deficient cells incubated in medium with LPDS compared to FCS-cultured cells, whereas the expression was not increased at all in LPDS-cultured CHO-K1 cells. It has been shown that LDL in FCS provides enough cholesterol to suppress SREBP-2 cleavage in wild-type CHO and HEK-293 cells (Hannah et al., 2001;Lee and Ye, 2004). Whereas the expression of the Fdft1 luciferase reporter gene was increased in FCS-cultured peroxisome-deficient cells compared with CHO-K1, culturing all cell lines in medium with 5% LPDS did not increase the expression of the Fdft1 luciferase reporter ( Figure 3A). We then tested whether cholesterogenic promoter activity in peroxisome-deficient cells can be suppressed by incubation with 5% LPDS supplemented with 1 and 10 µg/ml 25hydroxycholesterol (25-HC) ( Figure 3B). Unlike LDL-borne cholesterol, 25-HC added to the culture medium in solvent enters the cell and reaches the ER without first traversing lysosomes . 25-HC suppressed Srebf2, Hmgcr, and Fdps promoter activity both in CHO-K1 and peroxisomedeficient ZR-82 and ZR-87 cells ( Figure 3B) indicating the responsiveness of SREBP to hydroxycholesterols. Taken together, these results show that peroxisome deficiency activates expression of cholesterogenic promoter-reporter genes, even though total cholesterol levels are unchanged when compared to wild-type CHO-K1 cells ( Table 1).

Expression of Cholesterol Biosynthetic Genes Is Increased in Peroxisome-Deficient CHO Cells
Next, we determined the mRNA expression of genes involved in cholesterol biosynthesis and their regulation, in cholesterol efflux, and in fatty acid synthesis in CHO-K1 and peroxisomedeficient CHO cells cultured in medium containing 10% FCS or 5% LPDS for 24 h. Again, the expression of Srebf2, Insig1, cholesterol biosynthetic genes (Hmgcs, Hmgcr, Idi1, Fdft1, Sqle, Lss, Dhcr24), Ldlr, and Fasn was significantly increased in all cell lines when incubated in medium containing LPDS ( Figure 4A). However, the expression of these genes was much more strongly induced in peroxisome-deficient cells, especially in ZR-82 cells, than in CHO-K1 cells. Importantly, the expression of cholesterol biosynthetic genes and its regulators Srebf2 and Insig1 also was significantly increased in peroxisomedeficient cells compared with CHO-K1 when the cells were incubated in medium containing 10% FCS (Figure 4A; cf. also Supplementary Figure S2). These data corroborate the promoter studies shown in Figure 3. Next, exit of cholesterol from lysosomes requires the membrane-bound Niemann-Pick Each value represents the average ± range (n = 2). C1 (NPC1) protein (Kwon et al., 2009); however, we did not detect any changes in Npc1 expression at least in peroxisomedeficient cells ( Figure 4A). Hence, although it is well-known that all cholesterol biosynthetic genes are regulated by SREBP-2 (Horton et al., 2002), the magnitude of response in their expression varied significantly among the genes tested, indicating that their levels do not depend on the expression level of SREBP-2 alone.

Activities of Cholesterol Biosynthetic Enzymes and Protein Levels Are Altered in Peroxisome-Deficient CHO Cells
To examine the effect of peroxisome deficiency on the activities of cholesterol biosynthetic enzymes, we measured the activities of HMGCR, IDI1, and FDPS in CHO-K1 and peroxisomedeficient cells ( Figure 4B). HMGCR, which catalyzes the ratelimiting step of sterol biosynthesis, is localized mainly in the ER, whereas IDI1 and FDPS are localized predominantly in peroxisomes (Kovacs et al., 2002;Faust and Kovacs, 2014). HMGCR activity was decreased 50% in ZR-78.1C and ZR-87 cells and 75% in ZR-82 cells, corroborating previously published data (Aboushadi and Krisans, 1998). The activity of IDI1 and FDPS was similar or increased in FCS-cultured peroxisome-deficient cells compared with CHO-K1 cells. IDI1 and FDPS activities were significantly increased in all cell lines when incubated in medium containing LPDS, however, the activities were somewhat lower in peroxisome-deficient cells compared with CHO-K1 cells.
Western blot analysis of proteins involved in cholesterol biosynthesis was performed to determine whether the measured activities are a reflection of the protein levels. The protein levels of HMGCR, mevalonate kinase (MVK), IDI1, FDPS, and FDFT1 were significantly increased in peroxisome-deficient cells compared with CHO-K1 ( Figure 4C). Thus, decreased HMGCR and IDI1 activities in the peroxisomal-deficient cells are not a reflection of their protein levels, however, gene expression data gave reason to expect even higher protein levels of cholesterol biosynthetic genes in peroxisome-deficient cells.

Rates of Cholesterol and Dolichol Biosynthesis Are Decreased in Peroxisome-Deficient Cells
To evaluate the effects of the decrease in HMGCR, IDI1, and FDPS activities in peroxisome-deficient cells, we measured the rate of sterol (cholesterol) and non-sterol (dolichols) biosynthesis in CHO-K1, ZR-82, and ZR-87 cells after incubating the cells with [ 14 C]acetate. The rates of cholesterol and dolichol biosynthesis were significantly reduced in the peroxisomal-deficient cells as compared to the CHO-K1 cells ( Figure 4D and Supplementary  Table S1). These data corroborate a previously published study showing that cholesterol and dolichol biosynthesis rates were decreased in ZR-78.1C and ZR-82 cells when incubated with either [ 3 H]acetate or [ 3 H]mevalonate (Aboushadi and Krisans, 1998). However, despite decreased biosynthesis rates, total cholesterol levels were similar in CHO-K1 and peroxisomedeficient cells cultured in medium containing either 10% FCS or 5% LPDS for 24 and 48 h ( Table 1).

FIGURE 4 | (A)
Expression of genes involved in cholesterol biosynthesis and its regulation, cholesterol efflux, and fatty acid synthesis in CHO-K1 and peroxisome-deficient CHO cells cultured in medium containing 10% FCS or 5% LPDS for 24 h. Each value represents the amount of mRNA relative to that in FCS-cultured CHO-K1, which was arbitrarily defined as 1. Data are mean ± SD (n = 3). (B) Activities of the cholesterol biosynthetic enzymes HMGCR, IDI1, and FDPS. Each value represents the average of 10-17 experiments, performed in duplicate. Data are mean ± SD. (C) Immunoblot analysis of cholesterol biosynthetic enzymes HMGCR, MVK, IDI1, FDPS, and squalene synthase (FDFT1) in cells after incubation in medium containing 5% LPDS for 24 h. Numbers at the bottom of the blots indicate the fold change in protein levels in peroxisome-deficient cells relative to that in CHO-K1 cells, which were arbitrarily defined as 1. l.e., long U18666A Treatment Activates the SREBP-2 Pathway in CHO-K1 and Peroxisome-Deficient CHO Cells A recent study suggested that peroxisomes play a critical role in the transport of LDL-derived cholesterol from the lysosome to the plasma membrane and other intracellular compartments via a lysosome-peroxisome membrane contact site (Chu et al., 2015). Knockdown of ABCD1, PEX1, and PEX26 in HeLa cells using shRNAs led to accumulation of cholesterol in lysosomes to a similar extent as in cells from Niemann-Pick disease type C (NPC) patients. Therefore, we aimed to determine the effect of U18666A on the SREBP-2 pathway in CHO-K1 and peroxisome-deficient cells. The amphiphilic compound U18666A has been shown to bind to Niemann-Pick C1 (NPC1) and inhibit cholesterol export from lysosomes (Liscum and Faust, 1989;Liscum, 1990;Lu et al., 2016), leading to SREBP-2 activation (Colgan et al., 2007). If peroxisome deficiency leads to an accumulation of cholesterol in lysosomes (Chu et al., 2015), U18666A should activate the SREBP-2 pathway only in CHO-K1 cells, but should not lead to further activation in ZR-78.1C, ZR-82, and ZR-87 cells. The expression of SREBP-2 target genes was determined in cells that were grown in cholesterol-containing medium (10% FCS) supplemented with 10 µM U18666A for 24 h. U18666A significantly increased the mRNA levels of Srebf2, Insig1, and cholesterol biosynthetic genes both in CHO-K1 and peroxisome-deficient cells compared with vehicle-treated cells ( Figure 5A). The increase was similar in CHO-K1, ZR-78.1C, and ZR-87 cells, whereas ZR-82 cells again showed a much stronger increase, even though with a larger standard deviation. U18666A treatment significantly decreased the expression of the cholesterol efflux transporter Abca1 (ATP-binding cassette transporter A1) in all cell lines ( Figure 5A). The addition of 10 µg/ml 25-HC significantly suppressed the activation of SREBP-2 target genes in U18666A-treated cells (Figure 5A), indicating that the sensitivity towards 25-HC is functional in peroxisome-deficient cells. U18666A treatment increased the protein levels of the cholesterol biosynthetic enzymes MVK, FDPS, and FDFT1 in all cell lines compared with vehicle-treated cells ( Figure 5B). Interestingly, despite highly elevated mRNA levels, HMGCR protein levels in U18666A-treated ZR-82 cells were lower compared with CHO-K1, ZR-78.1C, and ZR-87 cells, whereas the FDPS and FDFT1 protein levels were higher in ZR-82 cells. The protein levels of catalase and the ER proteins GRP78 and GRP94 were similar in CHO-K1 and peroxisome-deficient cells.
To assess levels of de novo lipogenesis upon U18666A treatment we measured total de novo lipid production by quantifying incorporation of radioactively labeled [ 14 C]acetate into lipids. Cholesterol synthesis was significantly decreased in vehicle-treated peroxisome-deficient cells compared with CHO-K1 cells ( Figure 5C). U18666A treatment significantly increased the rate of cholesterol biosynthesis in all cell lines, however, cholesterol biosynthesis was significantly lower in peroxisome-deficient cells compared with CHO-K1 cells ( Figure 5C). U18666A treatment significantly decreased as expected the rate of cholesteryl ester synthesis in all cell lines, whereas the rates of fatty acid and triacylglycerol synthesis were increased ( Figure 5C). Interestingly, the rate of cholesteryl ester synthesis was similar in all cell lines cultivated in FCS, suggesting that they have a similar supply of cholesterol in the ER. In summary, U18666A induced mRNA and protein levels of cholesterol biosynthetic enzymes in peroxisome-deficient cells, however, again the rates of cholesterol synthesis were significantly lower than in CHO-K1 cells.

Expression of SREBP-2 Target Genes in Extrahepatic Tissues of SW/129 Pex2 −/− Mice
We reported previously that total cholesterol levels were similar in the lung, kidney, heart, and spleen of P10 SW/129 Pex2 −/− mice when compared with control mice, whereas the rate of cholesterol synthesis was significantly increased in the lung, heart, and spleen and decreased in the kidney (Kovacs et al., 2004(Kovacs et al., , 2009). As total cholesterol levels were similar in P10 SW/129 Pex2 −/− and control mice, we examined the expression of Srebf2, Insig1, Insig2, and cholesterol biosynthetic genes (Hmgcr, Pmvk, Mvd, Idi1, Fdps, Fdft1, Sqle, Sc4mol, Lss) in these tissues as well as skeletal muscle. Indeed, the mRNA levels of all these genes were significantly increased in Pex2 −/− mice (Figures 6A-E). In summary, there is a very similar mRNA up-regulation of Srebf2 and its target genes in extrahepatic tissues compared with livers of P0 and postnatal Pex2 −/− mice (Kovacs et al., 2004(Kovacs et al., , 2009(Kovacs et al., , 2012. Taken together, these data corroborate our findings in CHO cell lines, which indicate that functional peroxisomes are necessary for efficient cholesterol biosynthesis.

ER Stress Does Not Activate the SREBP-2 Pathway in CHO-K1 Cells and Peroxisome-Deficient CHO Cells
We showed that peroxisome deficiency activates hepatic ER stress pathways, especially the integrated stress response mediated by PERK and ATF4 signaling, in livers of newborn and postnatal Pex2 −/− mice (Kovacs et al., 2009(Kovacs et al., , 2012. Studies have shown that ER stress induces SREBP-2 activation independently of intracellular cholesterol concentration in various cell lines (Werstuck et al., 2001;Lee and Ye, 2004;Colgan et al., 2007;Lhoták et al., 2012). The analysis of the mRNA expression FIGURE 5 | (A) Expression of genes involved in cholesterol biosynthesis and its regulation, cholesterol efflux, and fatty acid synthesis in CHO-K1 and peroxisome-deficient CHO cells cultured in medium containing 10% FCS and 0.1% EtOH as solvent control or 10 µM U18666A or 10 µM U18666A and 10 µg/ml 25-hydroxycholesterol (25-HC) for 24 h. Each value represents the amount of mRNA relative to that in control CHO-K1, which was arbitrarily defined as 1. Data are mean ± SD (n = 3). (B) Immunoblots of total cell lysates with antibodies against cholesterol biosynthetic enzymes (HMGCR, MVK, FDPS, and FDFT1), peroxisomal (catalase, ACOX1) and ER (GRP78, GRP94) proteins after treatment with 10 µM U18666A for 24 h. Shown are immunoblots from three experiments. (C) Rate of (Continued) of several unfolded protein response (UPR) target genes in extrahepatic tissues of P10 Pex2 −/− mice showed that ER stress is also present in the lung, kidney, skeletal muscle, heart, and spleen (Figures 6F-J). To determine whether ER stress is also present in peroxisome-deficient CHO cells, we examined the mRNA expression of several UPR target genes in cells cultured in medium containing 10% FCS or 5% LPDS for 24 h (Figure 7A). The expression of most UPR target genes was, with the exception of Chop (C/EBP homologous protein), somewhat similar in CHO-K1 and peroxisome-deficient cells cultured in FCS-or LPDS-containing medium. Chop was significantly increased in all peroxisome-deficient cells compared with CHO-K1 when grown in medium containing 10% FCS. Culturing cell in 5% LPDS increased the expression of Xbp1t (total) and Xbp1u (unspliced) in all cell lines, however, the expression levels of the active transcription factor Xbp1s (spliced) were very low (C T > 38) and could not be analyzed. Interestingly, Xbp1t and Xbp1u mRNA levels were increased in peroxisome-deficient CHO cells, but the expression of Xbp1s was again too low to be evaluated. Next, we investigated whether induction of ER stress activates the SREBP-2 pathway in CHO-K1 cells. CHO-K1 cells were treated with various concentrations of the ER stress inducers thapsigargin (Tg), an inhibitor of the ER Ca 2+dependent ATPase, or tunicamycin (Tm), an inhibitor of protein N-glycosylation (the latter resulting in misfolding of proteins that normally would be glycosylated). Quantitative real-time PCR analysis revealed that Tg and Tm induced a significant dose-dependent increase in mRNA levels of various ER stress markers (Grp94, Dnajb9, Calr, Canx, Atf4, Chop, Gadd34, Xbp1t), however, the expression of SREBP-2 target genes was in general not induced in Tg-or Tm-treated CHO-K1 cells (Figures 7B,C). The mRNA levels of Srebf2 and Sqle were decreased in Tg-and Tm-treated cells, and only Insig1 expression was increased.
In summary, the increased expression of cholesterol biosynthetic genes in peroxisome-deficient CHO cells cannot be ascribed to ER stress, and activation of ER stress in CHO-K1 cells does not lead to an activation of the SREBP-2 pathway. mTORC1 and MAPK Pathways Do Not Activate the SREBP-2 Pathway in Peroxisome-Deficient CHO Cell Next, we investigated whether activated mTORC1 (mechanistic target of rapamycin complex 1) triggers activation of the SREBP-2 pathway in peroxisome-deficient cells. mTORC1 has been shown to activate SREBP-2 in mammalian cells through several mechanisms, including, at least in some cell types, through p70 ribosomal S6 kinase (p70 S6K) (Peterson et al., 2011;Wang et al., 2011;Eid et al., 2017). The only kinase known to be activated directly by mTORC1 is p70 S6K, which subsequently activates ribosomal S6 protein (S6) (Laplante and Sabatini, 2012). We assessed mTORC1 activity by analyzing the phosphorylation of S6 and p70 S6K in lysates from CHO-K1 and peroxisome-deficient cells treated with U18666A or cultured in the presence of FCS or LPDS for 24 h (Figures 8A-C). Phosphorylation of S6 on Ser235/236 was decreased in peroxisome-deficient cells compared with CHO-K1 in all three culture conditions. Phosphorylation of p70 S6K on Thr421/Ser424 was decreased in ZR-82 cells, while it tended to be decreased in ZR-78.1C and increased in ZR-87 cells. These data suggest that mTORC1 activity is inherently lower in peroxisome-deficient cell lines. Importantly, in accordance with our data, a recent study suggested that acetyl-CoA derived from peroxisomal β-oxidation promotes Raptor acetylation and mTORC1 activation in hepatocytes, and hepatic Acox1 deficiency resulted in inhibition of mTORC1 (He et al., 2020).
The phosphatidylinositol 3-kinase (PI3K)/AKT pathway is involved in SREBP-2 transport to the Golgi and thereby contributes to the control of SREBP-2 activation (Du et al., 2006). To compare AKT phosphorylation in CHO-K1 and peroxisomedeficient cells, immunoblot analyses were performed on lysates from cells treated with U18666A or cultured in FCS and LPDS for 24 h (Figures 8A-C). AKT phosphorylation on Ser473 was similar in FCS-cultured cells and slightly decreased in LPDS-or U18666A-treated ZR-82 and ZR-87 cells compared with CHO-K1 cells. Phosphorylation of AKT was only increased in LPDScultured ZR-78.1C cells.
Major signaling pathways that couple transcription factors to extracellular stimuli include the mitogen-activated protein (MAP) kinase cascades through extracellular signal-regulated kinase (ERK1/2). It has been shown that the mature form of SREBP-2 is a substrate of ERK-MAPK in vitro, affecting its trans-activity (Kotzka et al., 2000(Kotzka et al., , 2004. In order to examine a possible link between the MAPK pathway and increased SREBP-2 activity and expression of cholesterol biosynthetic genes, we determined whether the phosphorylation of p44/42 ERK was affected in peroxisome-deficient CHO cells. We measured the amounts of total and phosphorylated ERK proteins in cells treated with U18666A or cultured in the presence of FCS or LPDS for 24 h (Figures 8A-C). Total ERK levels were similar in CHO-K1 and peroxisome-deficient cells; however, phosphorylated ERK levels were equally increased in ZR-82 cells under all three culture conditions.
We conclude that the activities of the PI3K/AKT and mTORC1 pathways are not, in general, increased in peroxisomedeficient cells, and also are not the cause of the observed increased expression of cholesterol biosynthetic genes. That said, however, phosphorylated ERK could contribute to higher activity of SREBP-2 in ZR-82 cells. Each value represents the amount of mRNA relative to that in control mice, which was arbitrarily defined as 1. Data are mean ± SD (n = 6 for control and Pex2 −/− mice). Statistical analysis was performed using Student's t-test or Student's t-test with Welch's correction. *p < 0.05; **p < 0.01; ***p < 0.001 vs. control mice. nd, not detected.

Increased Transfer of GFP-SCAP From the ER to the Golgi in Peroxisome-Deficient CHO Cells
SCAP, the sterol sensor, plays an essential role in SREBP-2 activation by mediating its ER-to-Golgi transport. To test if peroxisome deficiency enhances ER-to-Golgi trafficking of SCAP/SREBP, we analyzed SCAP trafficking in CHO-K1 and peroxisome-deficient CHO cells using a GFP-SCAP fusion protein, which allows determination of the subcellular distribution of GFP-SCAP by confocal microscopy. Cells stably expressing GFP-SCAP were cultured in medium either containing 10% FCS or 5% LPDS or, alternatively, 10% FCS and 10 µM U18666A (Figures 9A-D). GFP-SCAP showed a reticular ER pattern in CHO-K1 cells grown in cholesterolcontaining medium. Incubating cells in medium with 5% LPDS for 24 h or treatment with U18666A for 12 h led to increased juxtanuclear staining and co-immunolocalization with giantin (also known as GOLGB1), a Golgi marker (Linstedt and Hauri, 1993), in addition to the reticular ER pattern (Figure 9A). Peroxisome-deficient ZR-78.1C and ZR-87 cells showed already an increased juxtanuclear staining when cultured in medium , and ZR-87 (D) cells. Cells were either cultured in 10% FCS or 5% LPDS for 24 h or treated with 10 µM U18666A for 12 h. GFP-SCAP was imaged using a chicken polyclonal anti-GFP antibody followed by Alexa488 secondary antibody; Golgi was visualized by rabbit anti-giantin followed by Alexa594 secondary antibody. The nuclei were stained with DAPI (blue). Representative fields are shown for each condition. The scale bars represent 10 µm. (E) The endoplasmic reticulum was detected with an antibody against GRP78. The scale bars represent 10 µm. (F) Cholesterol esterification in CHO-K1 and peroxisome-deficient cells. Cells cultured in 10% FCS were incubated for 4 h with [ 14 C]palmitate (∼8000 dpm/nmol). Cholesterol ester levels were analyzed by TLC. Data are mean ± SD (n = 3). Statistical analysis was performed using one-way ANOVA followed by Dunnett's multiple comparisons test. **p < 0.01 vs. FCS-cultured CHO-K1. containing 10% FCS; incubating cells in medium containing 5% LPDS or treatment with U18666A led to a strong colocalization of GFP-SCAP with giantin (Figures 9B,D). In ZR-82 cells GFP-SCAP localized independently from the culture conditions only to the Golgi (Figure 9C). In general, the data show that deficiency of functional peroxisomes promotes GFP-SCAP trafficking to the Golgi.
To exclude the possibility that the altered GFP-SCAP localization in ZR-82 cells cultured in medium containing 10% FCS was caused by general changes in ER morphology, we stained the ER with an antibody against the glucose-regulated protein 78 (GRP78). However, the morphology of the ER was comparable in CHO-K1 and ZR-82 cells (Figure 9E). Radhakrishnan et al. (2008) determined the concentration of cholesterol in the ER that is sufficient to retain the SCAP-SREBP-2 complex in the ER and suppress SREBP-2 processing. They showed that SCAP transported SREBP-2 from the ER to the Golgi when ER cholesterol was ≤5 mol% of ER lipids, and ER-to-Golgi transport of SCAP was inhibited when the cholesterol content of the ER rose above the sharp 5% threshold. An open question is whether the cholesterol level in the ER is reduced in peroxisome-deficient cells, which could explain the increased ER-to-Golgi trafficking of SCAP and SREBP-2 activation in peroxisome-deficient cells even when total cholesterol levels are comparable to control cells. Therefore, we determined the rate of cholesterol esterification as a surrogate marker for ER cholesterol levels ( Figure 9F). The enzyme acyl-coenzyme A:cholesterol acyltransferase (SOAT) is located in the ER and synthesizes cholesteryl esters, using cholesterol and fatty acyl-CoA as its substrates (Chang et al., 2006). Two Soat genes have been identified in mammals (Soat1 and Soat2), and SOAT1 is the major isoenzyme in CHO cells. Assuming that SOAT is located in the ER and that its activity is controlled by cholesterol availability, SOAT activity in intact cells has been used to monitor cholesterol arriving at the ER in cholesterol trafficking studies (Chang et al., 2006). We found that the esterification of cholesterol was similar in CHO-K1 cells and peroxisome-deficient cell lines (Figure 9F), suggesting that ER cholesterol levels are normal in peroxisomedeficient cells.
To exclude the possibility that the altered GFP-SCAP localization in ZR-82 cells cultured in medium containing 10% FCS was caused by general changes in ER morphology, we examined the localization of activating transcription factor 6 (ATF6) in CHO-K1 and ZR-82 cells. ATF6 resides as a transcriptionally inactive membrane-bound precursor protein in the ER by its interaction with GRP78. In response to ER stress and dissociation of GRP78, ATF6 is trafficked from the ER to the Golgi where it is sequentially cleaved by the proteases S1P and S2P to produce an active transcription factor, a process similar to that involving SREBP-2 (Ye et al., 2000;Chen et al., 2002). However, SCAP is not required for the ER-to-Golgi translocation of ATF6, but a similar molecular escort may bind ATF6 (Ye et al., 2000;Chen et al., 2002). We transfected CHO-K1 and ZR-82 cells with a GFP-ATF6 construct where the S1P site was mutated [GFP-ATF6(S1P-)] (Chen et al., 2002). The S1P cleavage blockade due to the S1P site mutation causes a longer retention of GFP-ATF6 in the Golgi after ER stress-induced ERto-Golgi transfer (Chen et al., 2002). The cells were analyzed by fluorescence microscopy and the staining of GFP-ATF6(S1P-) in the perinuclear and cytoplasmic regions of FCS-cultured CHO-K1 and ZR-82 cells was consistent with ER localization (Figure 10A). Treatment with thapsigargin ( Figure 10B) and tunicamycin ( Figure 10C) for 12 h caused the movement of GFP-ATF6(S1P-) to perinuclear locations consistent with a Golgi marker staining pattern. GFP-ATF6(S1P-) colocalized with the Golgi marker giantin in both cell lines in response to ER stress induction (Figures 10B,C), but GFP-ATF6(S1P-) was still detectable in the ER in both cell lines. These data show that the altered SCAP localization in peroxisome-deficient cells is specific for SCAP, while other proteins normally subject to ER-to-Golgi transport were not affected, indicating that defective ER-to-Golgi transport is not a generalized phenomenon.
Normally, SCAP is retrieved from the Golgi and moves back to the ER by COPI-mediated Golgi-to-ER retrograde trafficking after processing of SREBP under sterol-deficient conditions (Takashima et al., 2015). To provide further evidence that SCAP in peroxisome-deficient cells increasingly moves to the Golgi, we used brefeldin A, which causes the relocation of Golgi proteins into the ER (Lippincott-Schwartz et al., 1989;Klausner et al., 1992). GFP-SCAP-transfected ZR-82 cells cultured in medium containing 10% FCS were treated with brefeldin A for 30, 60, and 90 min. GFP-SCAP showed a Golgi-like staining pattern and co-localization with giantin in ZR-82 cells cultured in medium containing 10% FCS (Figure 11A). Even by 30 min, treatment with brefeldin A caused the pattern to disperse and return to a more ER-like pattern ( Figure 11A). However, as expected, the staining pattern of the ER marker GRP78 was not affected by brefeldin A treatment ( Figure 11B).
Next, we examined if treatment with 25-HC affects the GFP-SCAP localization in ZR-82 cells. We have shown that 25-HC suppressed both cholesterogenic promoter activity in LPDStreated ( Figure 3B) and activation of SREBP-2 target genes in U18666A-treated ( Figure 5A) CHO-K1 and peroxisomedeficient cells. GFP-SCAP-transfected ZR-82 cells cultured in medium containing 10% FCS were treated with 10 µg/ml 25-HC for 90 min, 12, 15, and 18 h. GFP-SCAP showed a Golgi-like staining pattern and co-localization with giantin in vehicletreated ZR-82 cells ( Figure 11C) and cells treated with 25-HC for 90 min (data not shown). Treatment with 25-HC for 12 and 15 h caused the pattern to disperse and return to a more ER-like pattern (data not shown), and GFP-SCAP showed a reticular ER pattern after an 18 h treatment with 25-HC ( Figure 11C).

Restoring Functional Peroxisomes
Normalized the Transcriptional Regulation of the Cholesterol Biosynthetic Pathway, the Rate of Cholesterol Synthesis, and SCAP Trafficking To examine whether restoring functional peroxisomes normalizes the transcriptional regulation of the cholesterol FIGURE 10 | Peroxisome deficiency does not enhance ER-to-Golgi trafficking of GFP-ATF6(S1P-). Confocal microscopy images show GFP-ATF6(S1P-) trafficking related to the Golgi marker giantin in CHO-K1 and ZR-82 cells. Cells were either cultured in 10% FCS (Control) (A) or treated with 100 nM thapsigargin (B) or 0.5 µg/ml tunicamycin (C) for 12 h. GFP-ATF6 was visualized using a chicken polyclonal anti-GFP antibody; the Golgi was visualized by rabbit anti-giantin; DAPI was used for nuclear staining. Representative fields are shown for each condition. The scale bars represent 10 µm.
biosynthetic pathway, the expression of SREBP-2 target genes was determined in Pex2-transfected cells cultured in medium containing 10% FCS, 5% LPDS, or 10% FCS plus 10 µM U18666A for 24 h (Figure 12A). Incubation in medium containing 5% LPDS or 10 µM U18666A significantly increased the mRNA levels of SREBP-2 target genes in all cell lines compared with FCS-cultured cells. However, the expression levels in CHO-K1-Pex2 cells and Pex2-complemented ZR cells with restored functional peroxisomes were similar under all conditions. Next, the protein levels of HMGCR, MVK, and FDPS were determined in CHO-K1-Pex2 and Pex2-complemented ZR cell lines cultured in medium containing 10% FCS, 5% LPDS, or 10 µM U18666A for 24 h (Figure 12B). The cultivation of the cells in medium containing 5% LPDS and the treatment with U18666A increased the protein levels of HMGCR, MVK and FDPS in all cell lines as expected. MVK and FDPS protein levels tended to be lower in ZR-78.1C-Pex2 cells, whereas HMGCR protein levels tended to be lower in LPDS-cultured and U18666A-treated ZR-82-Pex2 cells ( Figure 12B). Interestingly, the protein levels of PEX14 were increased in Pex2-transfected ZR-82 cells. In summary, compared to peroxisome-deficient cells, no significant increase in protein levels of cholesterol biosynthesis enzymes compared to CHO-K1 cells was observed in Pex2-transfected ZR cells after restoring peroxisome functionality.
Finally, we examined if the enhanced localization of GFP-SCAP to the Golgi in peroxisome-deficient cells could be prevented by restoring functional peroxisomes. Since GFP-SCAP localized independently of culture conditions exclusively to the Golgi complex in ZR-82 cells, ZR-82-Pex2 cells were transiently transfected with GFP-SCAP. GFP-SCAP showed a reticular ER pattern both in CHO-K1-Pex2 and ZR-82-Pex2 cells grown in cholesterol-containing medium (Figure 13). Importantly, none of the transfected ZR-82-Pex2 cells showed localization of GFP-SCAP to the Golgi complex.
Taken together, restoring functional peroxisomes in these cells resulted in a coordinated transcriptional and posttranscriptional regulation of the cholesterol biosynthetic pathway, demonstrating the importance of peroxisome functionality for this pathway.

DISCUSSION
Herein, we have provided evidence that functional peroxisomes are necessary for efficient cholesterol synthesis and that peroxisome deficiency dysregulates the SREBP-2 pathway. Cells maintain cholesterol homeostasis by multiple feedback controls that act through transcriptional and post-transcriptional mechanisms, especially transcriptionally through SREBPs (Goldstein et al., 2006) and sterol-accelerated ubiquitination and degradation of HMGCR (DeBose-Boyd, 2008). Cholesterol synthesis and uptake are tightly regulated by SREBP-2 (Horton et al., 2003), a membrane-bound transcription factor that undergoes sterol-regulated export from the ER to the Golgi apparatus for proteolytic processing. The transcriptionally inactive SREBP-2 precursor protein is retained in the ER membrane through the interaction with SCAP and INSIG proteins. In sterol-depleted cells, such as would occur under conditions of starvation or where the extracellular supply of cholesterol is diminished substantially, SCAP undergoes a conformational change and it escorts SREBP-2 from the ER to the Golgi in COPII vesicles, where it is processed to its mature form. The ER contains relatively low levels of sterols compared to the whole cell or plasma membrane, making it a highly effective environment for the high-affinity cholesterol sensor SCAP to reside (Brown et al., 2018). Radhakrishnan et al. (2008) showed that SCAP transports SREBP-2 to the Golgi when the ER cholesterol content was less than 5 mol% of the total ER lipids. SCAP transport was inhibited when cholesterol content of ER membranes rose above this minimum threshold. The SREBP-2 pathway is activated in peroxisome-deficient CHO cells even when grown in cholesterol-replete medium. Accordingly, stronger activation of the SREBP-2 pathway is seen when peroxisome-deficient CHO cells are grown in cholesterol-deficient medium or if cholesterol is sequestered in the lysosomes, e.g., upon U18666A treatment. However, peroxisome-deficient cells are still sensitive towards 25-HC, indicating that the sterol-dependent mechanism for activation of the SREBP-2 pathway is still intact.
explanation for why no ER stress and UPR activation was observed in peroxisome-deficient CHO cells, in contrast to, e.g., postnatal Pex2 −/− mouse tissues or biopsies obtained from patients with peroxisomal disorders (Kovacs et al., 2009(Kovacs et al., , 2012Launay et al., 2017), might be the lack of accumulation of ER stress-inducing metabolites. Indeed, it has been shown that ER stress and UPR are induced in human primary skin fibroblast cell lines from X-linked adrenoleukodystrophy patients with ABCD1 mutations and impaired peroxisomal β-oxidation only after incubation with very long-chain fatty acids (van de Beek et al., 2017). All three parallel branches of the UPR− governed by the ER-stress sensors IRE1, PERK, and ATF6, FIGURE 13 | Restoring functional peroxisomes in ZR-82 cells prevents enhanced ER-to-Golgi trafficking of GFP-SCAP. CHO-K1-Pex2 and ZR-82-Pex2 cells were transiently transfected with GFP-SCAP and cultured in 10% FCS. GFP-SCAP was imaged using a chicken polyclonal anti-GFP antibody followed by Alexa488 secondary antibody; peroxisomes were visualized by rabbit anti-ACOX1 and rabbit anti-catalase followed by Alexa594 secondary antibody; DAPI was used for nuclear staining. The scale bars represent 10 µm.
respectively, are activated upon pharmacologically induced ER stress (Lin et al., 2007;Zhang et al., 2015). However, also acute ER stress, which was chemically induced in the in vitro system used in the present study, did not activate SREBP-2 in CHO-K1 cells. Mechanistic target of rapamycin complex 1 has emerged as a regulator of lipid biosynthesis (Laplante and Sabatini, 2012). Porstmann et al. (2005) showed that activation of Akt, an upstream activator of mTOR, induced the expression of SREBP-1 and its target genes, but the effect on SREBP-2 was less pronounced. The mTOR inhibitor rapamycin has been shown to have opposing effects on SREBP-2 processing and the expression of HMGCR in different cell types (Gueguen et al., 2007;Sharpe and Brown, 2008;Ma et al., 2010). It has been shown that SREBP-2 processing and cholesterol biosynthetic gene expression are coordinately regulated by the eukaryotic initiation factor 4E-binding protein 1 (4E-BP1) and p70 S6K, and that the relative contribution of each mTORC1 effector varies between cell types (Wang et al., 2011). It has been reported that an increase in mTORC1 activity is accompanied by a decrease in ER cholesterol and by SREBP-2 activation (Eid et al., 2017). mTORC1, through its regulation of autophagy and endosomal membrane trafficking, activates SREBP-2 by suppressing cholesterol trafficking to lysosomes in mammalian cells (Eid et al., 2017). On the other hand, cholesterol has been shown to drive mTORC1 recruitment and activation at the lysosomal surface through a mechanism that is, at least in part, distinct from growth factor signaling (Castellano et al., 2017). Our data suggest that mTORC1 does not play a role in SREBP-2 activation or in the increased expression of cholesterol biosynthetic genes in peroxisome-deficient CHO cells, since the mTORC1 activity is inherently lower in peroxisome-deficient cell lines (i.e., decreased phosphorylation of S6 and p70 S6K). However, we cannot exclude the possibility that another target of mTORC1 could be involved in the activation of SREBP-2 in CHO cells instead of p70 S6K. Importantly, in accordance with our data, a recent study suggested that acetyl-CoA derived from peroxisomal β-oxidation promotes Raptor acetylation and mTORC1 activation in hepatocytes, and hepatic Acox1 deficiency resulted in inhibition of mTORC1 (He et al., 2020).
Despite the sensitivity of the SREBP-2 pathway to ER cholesterol levels, the rate of cholesterol synthesis is lower in peroxisome-deficient cells compared with CHO-K1 cells, illustrating the fact that functional peroxisomes and probably the compartmentalization of cholesterol biosynthetic enzymes in peroxisomes are necessary to ensure optimal biosynthesis of cholesterol. Consequently, plasma cholesterol levels are reduced both in patients with peroxisome biogenesis disorders and in 10-day-old Pex2 −/− mice (Kovacs et al., 2002(Kovacs et al., , 2004. Indeed, restoring functional peroxisomes by complementing peroxisome-deficient CHO cells with rat Pex2 cDNA enabled efficient cholesterol synthesis, at levels comparable to those of control (CHO-K1) cells.
Interestingly, loss of intact peroxisomes leads to enhanced ER-to-Golgi trafficking of GFP-SCAP. SCAP trafficking was normalized after restoration of functional peroxisomes, which goes also hand in hand with a normalization of the expression of SREBP-2 target genes and the rate of cholesterol synthesis. Thus, intact peroxisomes are a prerequisite for efficient cholesterol synthesis and their loss leads to defective cholesterol homeostasis.
Peroxisome-deficient ZR-78.1C, ZR-82, and ZR-87 cells have mutations in Pex2, a gene encoding an integral peroxisomal membrane protein, resulting in nonfunctional PEX2 (Zoeller et al., 1989;Thieringer and Raetz, 1993). PEX2 is involved in the translocation of peroxisomal matrix proteins following the docking of PTS receptors to the peroxisome membrane. These mutant cells exhibit typical peroxisomal membrane "ghosts"− abnormal membrane-delimited compartments that contain integral peroxisomal membrane proteins, but lack peroxisomal matrix proteins (Santos et al., 1988(Santos et al., , 2000. In such cases (e.g., Zellwegers Syndrome, and other peroxisomal deficiency diseases), peroxisomal matrix proteins such as catalase and ACOX1 are not appropriately compartmentalized in peroxisomes; instead, they are diffusely distributed throughout the cytoplasm, consistent with the inability of these mutant cells to import peroxisomal matrix proteins (Figure 2). The number of peroxisomal ghosts in the peroxisome-deficient cells is lower compared with the number of functional peroxisomes in CHO-K1 cells, however, the number of ghosts also varies between the peroxisome-deficient cells Figures 2B,C). Recently, tethers between the ER and peroxisomes have been identified in mammalian cells, involving the peroxisomal tailanchored proteins ACBD4 and ACBD5 and the ER-resident vesicle-associated membrane protein-associated proteins (VAPs) (Costello et al., 2017a,b;Hua et al., 2017). In immunostaining experiments in peroxisome-deficient cells, ACBD5 and PEX14 were present in cellular vesicles, suggesting that membrane contact sites between peroxisomal membrane ghosts and the ER might exist in peroxisome-deficient CHO cells (Figures 2B,  C). Hua et al. (2017) showed that tethering of peroxisomes to the ER via ACBD5-VAP-mediated contact sites is necessary for the maintenance of cellular cholesterol and plasmalogen levels: depletion of these tethers resulted in a decrease of total cholesterol levels. While total cholesterol levels in peroxisome-deficient CHO cells were normal, it is tempting to speculate that a transfer of farnesyl diphosphate to the ER in order to be used as substrate of FDFT1 might occur via ACBD5-VAP-mediated contact sites.
The cholesterol transfer from endocytic organelles to the ER likely involves several parallel pathways (Pfisterer et al., 2016). Recently, it has been suggested that LDL-derived cholesterol traffics from lysosomes via peroxisomes to the ER and that membrane contact sites facilitate this cholesterol transfer (Chu et al., 2015;Hu et al., 2018;Xiao et al., 2019). Tethering between the lysosomes and peroxisomes is facilitated by peroxisomal phosphatidylinositol-4,5-biphosphate (PI(4,5)P 2 ) and synaptotagmin 7 on lysosomes, and disruption of the lysosome-peroxisome membrane contacts caused cholesterol accumulation in lysosomes (Chu et al., 2015). Since ZR-78.1C, ZR-82, and ZR-87 cells lack functional peroxisomes, but contain abundant peroxisomal membrane ghosts, lysosomeperoxisome membrane contact sites might be able to form in such cells. This raises a fundamental biological question: if cholesterol traffics from the lysosome to peroxisomes for further distribution in the cell, are peroxisomal matrix proteins necessary (i.e., must the cholesterol go into the matrix) or does the transfer occur only via the peroxisomal membrane? Here, we demonstrated that total cholesterol levels were similar in CHO-K1 and peroxisome-deficient CHO cells. Interestingly, we and others have reported that cholesterol levels are either unaffected or decreased in peroxisome biogenesis-defective human fibroblasts, mouse tissues or serum (Kovacs et al., 2002;Faust and Kovacs, 2014). These findings are at odds with those of Chu et al. (2015); the reasons for this disparity are not yet clear. Cholesterol is distributed among multiple cellular pools, and any newly appearing cholesterol (arising either by de novo synthesis or by extracellular uptake) is rapidly incorporated into many types of cellular membranes by both vesicular and non-vesicular means (Lange and Steck, 2016). It might be that peroxisomal dysfunction results in subcellular cholesterol redistribution without a net increase in cholesterol content. Furthermore, we cannot exclude the possibility that peroxisome deficiency alters the equilibrium among different cellular cholesterol pools. In particular, cholesterol represents only about 5% of the total lipid mass of ER membranes, making them exquisitely sensitive to changes in the levels of cholesterol in cells (Radhakrishnan et al., 2008). However, peroxisomedeficient CHO cells are sensitive to inhibition of SREBP-2 activation by 25-HC, which enters cells and reaches the ER without traversing the lysosomes (and by a mechanism that does not involve membrane vesicles). This suggest that the regulation of SREBP-2 processing via hydroxysterols is still responsive in peroxisome-deficient cells. However, peroxisome deficiency leads to relocalization of SCAP to the Golgi, thereby altering the SCAP-Insig-SREBP-2 cholesterol sensing complex. In conclusion, functional peroxisomes are essential for efficient cholesterol sensing and synthesis.
The current experiments were performed not only in medium containing LPDS, but also in the presence of 10% serum (FCS), which provides an abundant source of cholesterol and fatty acids and usually suppresses the activation of all three SREBPs (Hannah et al., 2001). Despite the presence of lipid-replete medium, peroxisome deficiency led to activation of SREBP-specific reporter genes and increased expression of SREBP target genes. However, as mentioned above, it is not known whether cholesterol obligatorily enters the peroxisomal matrix in order to be distributed to other intracellular compartments or if the redistribution only occurs via the peroxisomal membrane. Blocking the exit of cholesterol from the lysosomes by treatment with U18666A resulted in activation of the SREBP-2 pathway, both in CHO-K1 and peroxisome-deficient cells, however, the increase of SREBP-2 target genes was more pronounced in peroxisome-deficient CHO cells, whereas cholesterol synthesis was lower in these cell lines.
Taken together, our findings demonstrate that disruption of peroxisomal function directly leads to a disarray in cholesterol metabolism and regulation, despite normal cholesterol content and sensitivity to external stimuli via 25-HC. Thus, intact (functional) peroxisomes are of utmost importance for a balance in the cholesterol homeostatic pathways.

Cell Culture
The CHO cell line, CHO-K1, and the peroxisome-deficient mutant CHO lines ZR-78.1C, ZR-82, and ZR-87 were obtained from Dr. Raphael A. Zoeller (Boston University School of Medicine, Boston, MA, United States) (Zoeller and Raetz, 1986). The mutant CHO cell lines were isolated from the CHO-K1 cell line used as control in this study. Cells were maintained in monolayer culture at 37 • C in 5% CO 2 in Ham's F-12 medium (Cat. no. 21765-029; Life Technologies, United Kingdom) containing 100 units/ml penicillin and 100 µg/ml streptomycin sulfate supplemented with 10% (v/v) fetal calf serum (FCS).

Animals
Homozygous Pex2 −/− mice were obtained by breeding Pex2 heterozygotes (Pex2 +/− ) on a hybrid Swiss Webster-129 (SW/129) background (Kovacs et al., 2004(Kovacs et al., , 2009. Mice had access to food and water ad libitum and were exposed to a 12:12hour light-dark cycle. For the purposes of this study, control mice consisted of Pex2 +/+ and Pex2 +/− genotypes (hereafter referred to as Pex2 +/ ), as their biochemical characteristics were comparable to one another (Kovacs et al., 2004(Kovacs et al., , 2009). All protocols for animal use and experiments were reviewed and approved by the Institutional Animal Care and Use Committee of San Diego State University.
Transient transfections were performed in 12-well plates using the Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA, United States) according to the manufacturer's instructions. Cells were seeded at a density of 5 × 10 4 cells/well. Cells were transfected with 500 ng of the luciferase reporter plasmids. Transfection efficiencies were normalized by co-transfecting in the ratio of 1:10 with a control plasmid encoding Renilla luciferase driven by the HSV-thymidine kinase promoter (pRL-TK, Cat. no. E2241, Promega, Madison, WI, United States). After transfection cells were cultured in medium with either 10% FCS or 5% LPDS for 24 h. After incubation cells were washed twice with PBS and harvested in 1× Passive Lysis Buffer (Promega). Firefly and Renilla luciferase activities in the cell lysates were measured with the Dual-Luciferase R reporter assay system (Promega) according to the manufacturer's protocol. Photon production was detected as relative light units by using an Analytical Luminescence Laboratory Monolight TM 2010 Luminometer. The values presented are the means of at least three separate transfections done in quadruplicate. The amount of Firefly luciferase activity in relative light units was normalized to the amount of Renilla luciferase activity from the same test tube.

Quantitative Real-Time RT-PCR
Total RNA was prepared from cells and frozen tissues with RNeasy Mini Kit (QIAGEN, Hilden, Germany) and treated with DNase I. Quantitative RT-PCR (qRT-PCR) assays were performed as described previously (Kovacs et al., 2012). First-strand cDNA was synthesized with random hexamer primers using Ready-To-Go You-Prime First-Strand Beads (GE Healthcare, Glattbrugg, Switzerland) and the High-Capacity RNA-to-cDNA Kit (No. 4368813; Applied Biosystems). Realtime RT-PCR was performed on a Roche LightCycler LC480 instrument in duplicates using 10 ng cDNA, 7.5 pmol forward and reverse primers, and the 2x KAPA SYBR FAST qPCR Mastermix (No. KK4601; KAPA Biosystems). Thermal cycling was carried out with a 5 min denaturation step at 95 • C, followed by 45 three-step cycles: 10 s at 95 • C, 10 s at 60 • C, and 10 s at 72 • C. Melt curve analysis was carried out to confirm the specific amplification of a target gene and absence of primer dimers. Primer sequences are listed in Supplementary Table S3. Relative mRNA amount was calculated using the comparative threshold cycle (C T ) method (Livak and Schmittgen, 2001). Gapdh and actin were used as the invariant control for CHO cells, and 18S rRNA and cyclophilin (Ppia) were used as the invariant control for mouse tissues.

Determination of Rate of Biosynthesis of Lipids
Cells were incubated with 5 µCi [1-14 C]acetate (58 mCi/mmol; PerkinElmer, Boston, MA, United States) for 4 h. After incubation, the cells were rinsed three times with PBS, and lipids were extracted with hexane:isopropanol (3:2). Manipulative losses of lipids were accounted for by addition of a known amount of [ 3 H]cholesterol as internal standard. After lipid extraction cells were lysed in 0.1 M NaOH for protein determination using the BCA method. Acetate incorporation into specific lipids was analyzed after separation of lipids by thin-layer chromatography (TLC). Therefore, organic phases were evaporated to dryness under a nitrogen stream. Lipids were resuspended in chloroform and spotted together with appropriate lipid standards on silica gel 60 plates (Merck, Darmstadt, Germany). For separation of neutral lipids, plates were developed in heptanes:diethylether:acetic acid (90:30:1) as solvent. Lipid samples and standards were visualized by iodine vapor. The lipid fractions were scraped from the plate and their radioactive content was determined by liquid scintillation counting. Values were normalized for sample protein content.
After incubation with 5 µCi [1-14 C]acetate incorporation into cholesterol was also analyzed by reverse-phase radio-HPLC as previously described (Kovacs et al., 2004). Briefly, following saponification and petroleum ether extraction, the nonsaponifiable lipids were analyzed by reverse-phase radio-HPLC, and the specific activities of the radiolabeled products (e.g., sterols, squalene) were determined.

Cholesterol Esterification
[1-14 C]palmitic acid (56.1 mCi/mmol, NEC075H050UC) was purchased from PerkinElmer. A 2.5 mM stock solution of [1-14 C]palmitic acid was prepared by complexing labeled and unlabeled palmitate to fatty acid-free albumin as described previously (Knobloch et al., 2017). The reaction mixture was prepared by diluting the palmitate/BSA stock with cell culture medium and contained 0.4 µCi [1-14 C]palmitic acid and 100 µM sodium palmitate in 1 ml medium. Cells were seeded in 6well plates in medium containing 10% FCS (v/v). The reaction was initiated by replacing the culture medium with 1 ml of the reaction mixture. The cells were incubated for 4 h at 37 • C. After incubation, lipids were extracted with hexane:isopropanol (3:2) and cholesterol ester levels were analyzed by TLC as described above for the determination of the rate of lipid biosynthesis.

Immunofluorescence Staining
Immunofluorescence staining was performed as described previously (Kovacs et al., 2007). As secondary antibodies, Alexa Fluor 488 donkey anti-chicken IgY (#703-545-155; Jackson ImmunoResearch), Alexa Fluor 488 goat anti-rabbit IgG (A11070; Invitrogen), Alexa Fluor 594 goat anti-rabbit IgG (A11072; Invitrogen), and Alexa Fluor 594 donkey antigoat IgG (A11058; Invitrogen) were used in a dilution of 1:400. Images were taken with a Leica SP8-AOBS confocal laser scanning microscope (Leica Microsystems GmbH, Wetzlar, Germany) using a 63× oil immersion objective and the Leica LAS Lite Software. Acquirement settings were optimized for each experiment, and identical settings were used for all images throughout one experiment. Fluorescent dyes were imaged sequentially in frame interlace mode to eliminate crosstalk between the channels. Image processing was performed off-line using Fiji software (Schindelin et al., 2012).

Statistical Analyses
Data are expressed as mean ± SD. When two groups where compared, statistical significance was evaluated by an unpaired, two-tailed Student's t-test or an unpaired, twotailed Student's t-test with Welch's correction when variances where significantly different. For multiple group analysis oneway ANOVA followed by Dunnett's multiple comparisons test was performed. Data were assumed to be normally distributed. Statistical analyses were performed using GraphPad Prism version 8.2.0.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusion of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by Institutional Animal Care and Use Committee of San Diego State University.

AUTHOR CONTRIBUTIONS
HS and WK: conception and design. KC, JS, PF, SF, HS, and WK: acquisition, analysis, and interpretation of data. PF, SF, and WK: resources. WK: writing -original draft and study supervision. PF, SF, HS, and WK: writing -review and editing. All authors approved the submitted manuscript.