SCAP/SREBP pathway is required for the full steroidogenic response to cyclic AMP

Significance

Luteinizing hormone stimulates production of testosterone and other steroids largely through a surge in the second messenger cAMP and subsequent activation of protein kinase A (PKA) in target cells. Rates of steroidogenesis are also dependent on the availability of cholesterol, a steroid building block. We propose, based on our results, that cAMP/PKA coordinates the functions of multiple pathways to regulate cellular cholesterol handling and synthesis and downstream steroid output. Activation of the cholesterol-sensing SCAP–SREBP2 pathway plays an important role in cAMP/PKA coordination of steroidogenesis. These cAMP/PKA-induced pathways are likely to be major regulators of sterol biosynthesis and cholesterol recharging in steroid hormone synthetic and other tissues. Cyclic nucleotide phosphodiesterases can be targeted to promote steroidogenesis and cholesterol metabolism.

Abstract

Luteinizing hormone (LH) stimulates steroidogenesis largely through a surge in cyclic AMP (cAMP). Steroidogenic rates are also critically dependent on the availability of cholesterol at mitochondrial sites of synthesis. This cholesterol is provided by cellular uptake of lipoproteins, mobilization of intracellular lipid, and de novo synthesis. Whether and how these pathways are coordinated by cAMP are poorly understood. Recent phosphoproteomic analyses of cAMP-dependent phosphorylation sites in MA10 Leydig cells suggested that cAMP regulates multiple steps in these processes, including activation of the SCAP/SREBP pathway. SCAP [sterol-regulatory element-binding protein (SREBP) cleavage-activating protein] acts as a cholesterol sensor responsible for regulating intracellular cholesterol balance. Its role in cAMP-mediated control of steroidogenesis has not been explored. We used two CRISPR (clustered regularly interspaced short palindromic repeat)-Cas9 (CRISPR associated protein 9) knockout approaches to test the role of SCAP in steroidogenesis. Our results demonstrate that SCAP is required for progesterone production induced by concurrent inhibition of the cAMP phosphodiesterases PDE4 and PDE8. These inhibitors increased SCAP phosphorylation, SREBP2 activation, and subsequent expression of cholesterol biosynthetic genes, whereas SCAP deficiency largely prevented these effects. Reexpression of SCAP in SCAP-deficient cells restored SREBP2 protein expression and partially restored steroidogenic responses, confirming the requirement of SCAP–SREBP2 in steroidogenesis. Inhibitors of 3-hydroxy-3-methylglutaryl-Coenzyme A reductase and isoprenylation attenuated, whereas exogenously provided cholesterol augmented, PDE inhibitor-induced steroidogenesis, suggesting that the cholesterol substrate needed for steroidogenesis is provided by both de novo synthesis and isoprenylation-dependent mechanisms. Overall, these results demonstrate a novel role for LH/cAMP in SCAP/SREBP activation and subsequent regulation of steroidogenesis.

Luteinizing hormone (LH) binding to its lutropin-choriogonadotropic hormone receptor on Leydig cells initiates a cascade of signaling events, including a surge in levels of the second messenger 3′,5′-cyclic adenosine monophosphate (cAMP) and subsequent activation of protein kinase A (PKA). A large number of cyclic nucleotide phosphodiesterases (PDEs) serve to limit the intracellular temporal and spatial effects of cAMP by hydrolyzing cAMP to 5′-AMP (1, 2). Among the different PDE enzymes, inhibition of PDE4 and PDE8 is known to stimulate steroidogenesis (3–6). Within these gene families, PDE8A and PDE8B both regulate steroidogenesis (5), whereas the role for specific PDE4 family gene products is less well understood. PDE4A, PDE4B, and PDE4C, but not PDE4D, have been shown to be expressed in Leydig cells (5, 7, 8), but the role for specific PDE4s in regulating steroidogenesis is unknown. We have recently demonstrated that simultaneous inhibition of PDE4 and PDE8 in MA10 Leydig cells results in an increase of a large number of phosphorylated PKA consensus sites in multiple proteins (8) concurrent with a very large increase in steroid production (5).

Cellular cholesterol levels are controlled in part by several transcription factors, including sterol-regulatory element-binding proteins (SREBPs) 2 and 1a, that promote cholesterol biosynthetic gene expression when cellular cholesterol levels are too low to meet demand (9, 10). The activities of the SREBPs are precisely controlled by an escort protein, SREBP cleavage-activating protein (SCAP), and the insulin-inducible gene product (Insig) (11–13). When cellular cholesterol levels are high and cholesterol binds to the sterol-binding domain of SCAP, SCAP is tethered by Insig, and the inactive SCAP–SREBP complex is retained in the endoplasmic reticulum (ER) membrane. Once cellular cholesterol levels become lower or demand for cholesterol supply becomes higher, the cholesterol-free form of SCAP is released from Insig, and the SCAP–SREBP complex is transported from the ER to the Golgi in coatomer protein II (COPII) vesicles. In the Golgi, SREBPs are sequentially cleaved by two proteases, site-1 protease and site-2 protease, which allows release of the active N-terminal mature forms of SREBPs and translocation into the nucleus, where they bind to sterol-regulatory elements (SREs) in target genes (14–16). This process has been studied largely in the context of regulation of cholesterol biosynthesis by the liver. A possible role(s) for cAMP in SCAP/SREBP regulation of steroid hormone production has not been extensively explored and is not well-understood.

In this study, we demonstrate using two different CRISPR-Cas9 gene-ablation approaches that cAMP elevation by inhibition of PDE4 and PDE8 can regulate SCAP/SREBP function in a manner consistent with activation of this pathway being required for maximal steroid hormone biosynthesis. More generally, the data suggest that cAMP/PKA/PDE4+8 coordinate the functions of multiple pathways to regulate steroid output and may also act as a major regulator of cholesterol biosynthesis in Leydig cells and possibly many other tissues.

Results

Activation of PKA by PDE4 and PDE8 Inhibitors Causes Increased Phosphorylation of Many Proteins Likely to Be Regulators of Cholesterol Handling and Steroidogenesis.

Based on our previous global screen of phosphorylation events in MA10 Leydig cells in response to PDE inhibition (8), we curated a list of several sites that were the most highly phosphorylated PKA consensus sites that also might directly influence steroid hormone production, cholesterol synthesis, and/or trafficking (Table 1). Nearly all identified phosphorylation sites showed more than a twofold increase after 1 h of treatment with specific PDE4 and PDE8 inhibitors (rolipram and PF-04957325, respectively) (increases were noted in at least three of five experiments). Many phosphorylation sites were increased more than fourfold. PKA consensus site phosphorylation in response to cAMP elevation occurred in a large number of proteins involved in cholesterol synthesis and delivery pathways, cholesterol and fatty acid liberation from lipid stores, and LDL receptor-, endosome-, and vesicle-trafficking proteins (Table 1). Among the identified phosphorylation sites was S821 of SCAP. Two other proteins also known to be important to the SCAP/COPII-dependent vesicle-trafficking pathway were even more highly phosphorylated (Sec22 homolog B vesicle-trafficking protein and Sec23-interacting protein) (Table 1). COPII vesicles transport a large number of cargo proteins from the ER to other membrane compartments in the cell and regulate a plethora of cellular functions in addition to cholesterol homeostasis. Therefore, the primary goal of this study was to begin to test the role for this pathway as a regulator of cAMP-stimulated steroid hormone production by examining the possible role of SCAP/SREBP in this process.

Table 1.

Phosphorylation of PKA consensus sites of proteins implicated in cholesterol/steroid synthesis or handling that are increased in response to PDE4 plus PDE8 inhibition

Fold PO4 increase	Protein name	Site	Gene name	Processes regulated
Cholesterol synthesis and delivery
5.2	Oxysterol-binding protein-like 11	S194	Osbpl11	Sterol binding and delivery
2.1	SREBP cleavage-activating protein	S821	Scap	SREBP pathway
COPII vesicle trafficking (ER to Golgi)
7.7	Sec22 homolog B, vesicle-trafficking protein	S137*	Sec22b	COPII/SREBP pathways
2.9	Sec23-interacting protein	S748	Sec23ip	COPII/SREBP pathways
COPI vesicle trafficking (Golgi to ER)
11.7	Ral GEF with pleckstrin homology and SH3-binding motif 2	S315	Ralgps2	Ral GEF activity
5.7	Golgi autoantigen, golgin subfamily a, 5	S116	Golga5	COPI pathways
4.8	ARF GAP1	S360	Arfgap1	COPI pathways
4.0	Autophagy 16 like 1	S269	Atg16l1	Rab33b interaction? Autophagy
3.7	ARF GEF/Tyr kinase adapter protein 1	S85	Nck1	Arf activation
2.1	ARF GEF2	S621	Arfgef2	COPI pathways
Cholesteryl ester/lipid droplet hydrolysis
7.1	Hormone-sensitive lipase	S651†	Lipe	Cholesteryl ester hydrolysis
4.9	Perilipin 1	S81‡	Plin1	Lipid droplet hydrolysis
LDL receptor and endosome vesicle trafficking
9.5	G protein-coupled receptor 107	S537	Gpr107	Golgi-to-ER retrograde transport
4.7	Pleckstrin homology domain-containing, family F (with FYVE domain) member 2	S16‡	Plekhf2	Rab-dependent vesicle trafficking?
4.4	TBC1 domain family, member 10B	S673	Tbc1d10b	GAP for Rab family proteins?
4.3	Ras and Rab interactor 2	S510	Rin2	Rab-dependent vesicle trafficking
3.0	Amyotrophic lateral sclerosis 2 (juvenile)	S477	Als2	Rab-dependent vesicle trafficking
2.3	DENN/MADD domain-containing 1A	S520	Dennd1a	Clathrin-mediated endocytosis
Microtubule formation, vesicle trafficking
13.1	Regulator of microtubular dynamics 2 (FAM82A1)	S51	Rmdn2	Microtubular dynamics
12.1	CAP-GLY domain linker protein 1 or 2	S347/353	Clip1/Clip2	Microtubule elongation/stability
11.7	Cytoskeleton-associated protein 5	S1861	Ckap5	Microtubule elongation/stability; binding of clathrin
8.7	Neuron navigator 1	S651‡	Nav1	Microtubule elongation/stability
3.7	SLAIN motif family, member 2	S392	Slain2	Microtubule elongation/stability
Rho family GTPase activation/inactivation, vesicle trafficking
2.9	Rho GAP21	S917	Arhgap21	Rho inhibition/vesicle trafficking
2.6	Rho GAP17	S698‡	Arhgap17	Rho inhibition/vesicle trafficking
2.4	Pleckstrin homology domain-containing, family G (with RhoGef domain) member 3	S502	Plekhg3	Rho activation/vesicle trafficking
2.4	Rho GEF 17	S1324	Arhgef17	Rho activation/vesicle trafficking
2.1	Rho GEF 2	S885	Arhgef2	Rho activation/vesicle trafficking
1.9	Rho GAP23	S607	Arhgap23	Rho inhibition/Golgi trafficking
1.6	Rho GEF 11	S1353	Arhgef11	Rho activation/vesicle trafficking
Others
13.6	PI 4-kinase, catalytic, beta polypeptide	S511	Pi4kb	Phosphatidylinositol/vesicle trafficking?
8.9	A kinase anchoring protein 1	S55	Akap1	Steroidogenesis (48, 49)
5.8	Carbohydrate response element-binding protein	S626	Mlxipl (ChREBP)	Glucose to acyl CoA synthesis
5.5	DDHD domain-containing 2	S447	Ddhd2	Vesicle trafficking

Protein functions are based on data provided by PhosphoSitePlus (50) and GeneCards (version 4.1, build 29; www.genecards.org), unless otherwise specified. Proteins investigated in the present study are highlighted in bold. Original proteomics data are available at the public MS data repositories MassIVE (ID MSV000079412; massive.ucsd.edu) and ProteomeXchange (accession no. PXD003280; www.proteomexchange.org). The fold increases shown have been calculated from primary data reported in ref. 8. All values are the average fold increases seen from multiple identifications and quantification at the 1-h time point (usually in three to six of six runs) with the exception of the following.

^*The Sec22b site (S137) is not a classic PKA consensus site (RRNLGS).

^†For Lipe, no quantification was seen at the 1-h time point but an average increase of 7.1-fold was seen at earlier time points, and the phosphorylation increase was verified by a phospho-specific antibody.

^‡In these cases, the values reported are the average increase seen in one or two runs at the 1-h time point but are verified by data from other time points.

SCAP Deficiency Reduces Leydig Cell Steroidogenesis and Cholesterol Biosynthetic Gene Expression.

We initially examined whether SCAP/SREBP is required for hormone-stimulated steroidogenesis by knocking down SCAP in the MA10 Leydig cell line because reduction of SCAP is known to also reduce SREBP levels. We used two distinct CRISPR-Cas9 gene-editing techniques. First, SCAP activity and protein were knocked-down in pools of cells by electroporation of Cas9, guide RNA, and a repair construct containing a puromycin-resistance gene. After selection by puromycin, these cell pools had a dramatic reduction in SCAP as well as SREBP2 protein expression compared with wild-type (WT) cells (Fig. 1A). We have previously reported that simultaneous inhibition of PDE4 and PDE8 (rolipram at 10 µM and PF-04957325 at 200 nM, respectively) synergistically stimulates steroidogenesis and induction of the steroidogenic acute regulatory (StAR) protein, a protein that can be a rate-limiting factor for steroidogenesis (5). Further, this study demonstrated that both PDEs cooperatively control levels of cAMP-dependent phosphorylation and steroidogenesis (5). We therefore used coinhibition of PDE4 and PDE8 as well as LH stimulation to investigate the role of SCAP/SREBP in the cAMP-dependent component of hormone-stimulated steroidogenesis.

Fig. 1.

SCAP deficiency generated by the CRISPR-Cas9 system elicited reductions in SREBP2, steroidogenesis, and cholesterol biosynthetic genes. (A) MA10 cell extracts collected from a SCAP knockdown (KD) cell pool and a WT cell pool were used to determine SCAP and SREBP2 protein levels. For determination of p-HSL (Ser660 rat, Ser651 mouse) and StAR, both cell groups were treated with vehicle or PDE4+8 inhibitors (PDE4+8 in; 10 µM rolipram and 200 nM PF-04957325) for 15 min (p-HSL) or 2 h (StAR) before harvest. (B) A SCAP KD cell pool (red bars) and WT cell pool (blue bars) were serum-starved for 3 h and then treated with either vehicle, PDE inhibitors, LH (20 ng/mL), or LH plus PDE inhibitors for an additional 2 h. Progesterone released into the medium was quantified by ELISA. Each value represents mean ± SD (n = 4). Representative results from four repeated experiments are shown. (C) Messenger RNA levels were determined by quantitative real-time PCR using total RNA collected from SCAP KD cell pools and WT cell pools. Both groups were treated with vehicle or PDE inhibitors for 18 h under serum-starved conditions. Each value represents mean ± SD (n = 4). Data represent one of two sets of analyses. Statistical significance is shown as *P < 0.05 and **P < 0.01 vs. (-) inhibitors; ^‡P < 0.01 vs. WT.

Steroidogenesis induced by PDE4+8 inhibitors in the presence or absence of LH was significantly reduced in the pools of SCAP-deficient cells compared with WT cells (Fig. 1B). A similar fold decrease was observed in both LH-stimulated cells and unstimulated cells, although total progesterone production was markedly lower in the latter condition (Fig. 1B). However, SCAP-deficient cell pools exhibited equivalent levels of StAR protein induction in response to the PDE inhibitors, indicating that neither the overall viability of the cells nor the mechanism leading to StAR induction was impaired by SCAP deficiency (Fig. 1A). Furthermore, phosphorylation of hormone-sensitive lipase (HSL) on S660 (S651 in mouse HSL) was stimulated by PDE inhibition to the same extent in WT and SCAP-deficient cells (Fig. 1A), confirming a lack of nonspecific effect of the CRISPR-Cas9 gene targeting of SCAP. These results indicate that SCAP/SREBP is required for steroidogenesis through mechanisms that are distinct from StAR expression and HSL phosphorylation.

In other cells, SREBP2 is well-known to control the level of several cholesterol biosynthetic genes including 3-hydroxy-3-methylglutaryl-CoA reductase (Hmgcr), 3-hydroxy-3-methylglutaryl-CoA synthase (Hmgcs), farnesyl diphosphate synthetase (Fdps), and cytochrome P450, family 51 (Cyp51), as well as genes required for cholesterol uptake, such as the low-density lipoprotein receptor (Ldlr) (17). Messenger RNA levels for each of these genes and SREBP2 (Srebf2) itself were significantly reduced in the SCAP-deficient cell pool (Fig. 1C), confirming that SCAP deficiency resulted in reduced SREBP2 activity. This was expected, as it has been reported previously in SCAP knockout mice (18). Interestingly, mRNA levels of Hmgcr, Fdps, Cyp51, and Ldlr were up-regulated by PDE4+8 inhibition in both WT and SCAP-deficient cell pools, although the absolute levels were lower in SCAP-deficient cells. PDE inhibitor-inducible up-regulation in mRNA levels was also seen for steroidogenic acute regulatory protein (Star), cytochrome P450, family 11, subfamily a, polypeptide 1 (Cyp11a1), and acyl-CoA synthetase long-chain family member 1 (Acsl1) in both WT and SCAP-deficient cells (Fig. S1). HSL (Lipe) and acyl-CoA synthetase long-chain family member 4 (Acsl4) were significantly induced by PDE4+8 inhibition in WT cells (Fig. S1). ACSL enzymes act to convert free fatty acids into acyl-CoAs for further processing, including cholesteryl ester formation and inhibition of cholesterol efflux (19, 20). ACSL4 has been reported to contribute to steroidogenesis through increasing arachidonic acid delivery to mitochondria, where this fatty acid could enhance cholesterol transport (21, 22). Thus, PDE4+8 inhibition is likely to increase cholesterol biosynthetic gene products through multiple mechanisms, including both SCAP-dependent and SCAP-independent mechanisms.

Fig. S1.

Effect of SCAP deficiency on mRNA levels of proteins likely to be involved in steroidogenesis. Messenger RNA levels were determined by quantitative real-time PCR using total RNA collected from the SCAP KD pool and the WT cell pool. Both groups were treated with vehicle or PDE4+8 inhibitors (10 µM rolipram and 200 nM PF-04957325) for 18 h under serum-starved conditions. Each value represents mean ± SD (n = 4). Data represent one of two sets of analyses. Statistical significance is shown as *P < 0.05 and **P < 0.01 vs. (-) inhibitors; ^‡P < 0.01 vs. WT. (Lower) Sequences of each primer set used for quantitative real-time PCR are shown.

To confirm that the effects of SCAP deficiency were not due to off-target effects of Cas9, we took advantage of a more specific CRISPR approach to generate individual SCAP-deficient cell clones. We used the FokI/dCas9 CRISPR gene-editing system (23) to delete SCAP. The FokI/dCas9 dimeric system requires two guide RNAs for FokI to cleave the target DNA, and therefore is likely to have much higher specificity than what can be achieved with the canonical Cas9 system (23). We used the clonal daughter cell line ms-16, derived from the original MA10 cells, for these experiments because we found that individual clonal cells derived from the original MA10 cell WT population varied substantially in their ability to produce progesterone. For example, in the ms-16 cell clone line, PDE4+8 inhibition resulted in at least sixfold higher induction of steroidogenesis, compared with the mixed-parent MA10 cells (e.g., compare the response in MA10 cells in Fig. 1 with the response in ms-16 cells in Fig. 2). Some other clonal cell lines showed much lower responses than the ms-16 cells. Following electroporation with the FokI/dCas9 and multiplex guide RNA plasmids, individual SCAP-deficient clonal cell lines were first screened by PCR amplification of the CRISPR target region (SCAP intron/exon 16) (Fig. S2). In selected clones showing multiple PCR products, SCAP protein levels were determined by Western blot (Fig. 2C) and genomic mutagenesis was confirmed by sequencing (Fig. S2). All clones with successful SCAP deletions exhibited significantly reduced progesterone production, compared with either the parent ms-16 cells or two other WT clonal lines, c-203 and c-206 (Fig. 2A). The average reduction in progesterone production in SCAP-deficient cells reached 54% under serum-starved conditions (Fig. 2A) and 60% in the presence of serum (Fig. S3A). These results suggested that SCAP contributes significantly to steroidogenesis even when cells have access to cholesterol from extracellular sources. Free cholesterol levels were determined in whole-cell extracts and mitochondrial fractions both in WT and SCAP-deficient cells. Mitochondrial free cholesterol as well as whole cellular free cholesterol were significantly lower in SCAP-deficient cells than in WT cells (Fig. 2D). This reduced free cholesterol availability in mitochondria is consistent with the lower rates of steroidogenesis in SCAP-deficient cells.

Fig. 2.

Steroidogenesis and cholesterol/isoprenoid biosynthetic pathway in SCAP-deficient cell clones generated by the CRISPR-FokI/dCas9 system. (A and B) SCAP-deficient clonal cells (s-138, s-139, s-140, s-141, s-142, and s-143) (A), HSL-deficient clonal cells (h-222, h-225, and h-227) (B), and WT clonal cells (ms-16, c-202, c-203, and c-206) were serum-starved for 3 h and treated with vehicle (Lower) or PDE4+8 inhibitors (10 µM rolipram and 200 nM PF-04957325; Upper) for 2 h under serum-starved conditions. Each value represents mean ± SD (n = 4). The numbers are represented relative to WT, mean ± SD (n = 3–6). *P < 0.05 and **P < 0.01 vs. WT. (C) SCAP and HSL levels were determined by Western blot analysis using whole-cell extracts of each clonal cell line. GAPDH was used as a loading control. (D) Whole-cell extracts and mitochondrial fractions were isolated from ms-16 and s-143 cells. Free cholesterol was measured in each fraction. Results are presented as relative to ms-16 (WT). Each value represents mean ± SD (n = 3) obtained from three repeated experiments. Free cholesterol levels were 11.4 ± 1.1 (n = 4) and 120.0 ± 8.5 (n = 4; pmol/µg protein) in whole-cell extract and mitochondrial fraction, respectively. *P < 0.05 and **P < 0.01 denote statistical significance. (E) ms-16 cells were pretreated with methyl-β-cyclodextrin cholesterol (Chol; 1 mM) or mouse HDL (25 µg/mL) with or without lovastatin (0.5 µM) for 30 min. Cells were then stimulated with PDE4+8 inhibitors for 2 h under serum-starved conditions. Each value represents the relative changes to PDE inhibitor-induced steroidogenesis without lovastatin as 100%; mean ± SD (n = 3). Progesterone levels in cells exposed to PDE inhibitors but not to exogenous cholesterol (indicated as “none” in the figure) were 791 ± 60 (n = 3; ng/mg protein). **P < 0.01 vs. none; ^‡P < 0.01 vs. (-) lovastatin. (F) ms-16 cells were pretreated with the indicated concentrations of lovastatin for 30 min and then treated with vehicle or PDE4+8 inhibitors for an additional 2 h under serum-starved conditions. Progesterone production in the absence of lovastatin is indicated by the black bar. The SCAP-deficient s-138 cell clone (red bar) was treated with PDE4+8 inhibitors in the absence of lovastatin. Each value represents mean ± SD (n = 4) of PDE inhibitor-stimulated progesterone levels after subtraction of basal levels. Representative results from three repeated experiments are shown. **P < 0.01 vs. cells in the absence of lovastatin (-). (G) ms-16 cells were pretreated with GGTI (2 and 5 µM) or FTI (2 and 5 µM) for 30 min and then treated with vehicle or PDE4+8 inhibitors for an additional 2 h under serum-starved conditions. Each value represents mean ± SD (n = 4) of PDE inhibitor-stimulated progesterone levels after subtraction of basal levels. Representative results from three repeated experiments are shown. **P < 0.01 vs. (-). (H) ms-16 cells were pretreated with mevalonate (10 mM) or GGPP (10, 30, and 50 µM) with or without lovastatin (0.5 µM) for 30 min. s-138 cells were pretreated with GGPP (10, 30, and 50 µM) for 30 min, and then both cell clones were stimulated with PDE4+8 inhibitors under serum-starved conditions. Each value represents the relative change to the level of PDE inhibitor-induced steroidogenesis in ms-16 cells; mean ± SD (n = 4). Representative results from three repeated experiments are shown. *P < 0.05 and **P < 0.01 denote statistical significance.

Fig. S2.

Guide sequences for CRISPR-FokI/dCas9 gene editing and screening. For editing the mouse SCAP and HSL genes, two guide RNA sequences (guide A, blue; guide B, green) were chosen using ZiFiT Targeter software (zifit.partners.org). Clonal cell lines were first screened by PCR/4% gel analysis. One representative gel image is shown (Bottom). We selected several positive clones showing multiple PCR amplicons (yellow asterisks). The efficiency for SCAP and HSL editing was 42 clones among 52 (80.8%) and 7 among 17 (41.2%), respectively. We then performed Western blot analysis of protein levels (for example, Fig. 2C). Clones with no protein expression were further analyzed by TA cloning and sequencing. The detected gene mutations were different-sized deletions occurring between the two guide sequences as shown (Top). PAM, protospacer adjacent motif.

Fig. S3.

Steroidogenesis under serum-fed conditions in SCAP-deficient cell clones generated by the CRISPR-FokI/dCas9 system. (A) SCAP clonal cells (s-138 and s-143), HSL clonal cells (h-222, h-225, and h-227), and WT clonal cells (ms-16, c-202, and c-206) were treated with vehicle (light columns) or PDE4+8 inhibitors (solid columns) for 2 h in the presence of serum. Progesterone released into the medium was quantified by ELISA. The results represent values relative to WT clones, mean ± SD. The experiment was repeated several times for different clones. The experiment shown was performed in triplicates. *P < 0.05 vs. WT. (B and C) The effects of lovastatin on steroidogenesis. MA10 cells were pretreated with or without lovastatin (0.5 µM) for 30 min. Cells were further treated with vehicle or PDE4+8 inhibitors for 16 h under serum-starved (B) or serum-fed conditions (C) in the presence or absence of lovastatin for measurement of steroidogenesis. Each value represents mean ± SD (n = 4). Representative results from three repeated experiments are shown. **P < 0.01 vs. (-) lovastatin. (D) MA10 cells pretreated with the HSL inhibitor CAY10499 for 30 min after 3 h of serum starvation. The cells were further treated with vehicle, PDE4+8 inhibitors, or 8Br-cAMP (500 µM) for 2 h for measurement of steroidogenesis. Each value represents mean ± SD (n = 4). Representative results from two repeated experiments are shown.

To further verify the specificity of the FokI/dCas9 approach and to investigate the possible contribution of newly imported and stored cholesteryl esters to steroid hormone production, we generated cell clones deficient in HSL (Lipe) by using the same method (Fig. S2). The HSL-deficient cell clones h-222, h-225, and h-227 showed no HSL protein expression (Fig. 2C), but also showed no reduction in progesterone production in either the presence or absence of serum, compared with the ms-16 cells or WT clones (Fig. 2B and Fig. S3A). Furthermore, treatment of MA10 cells with an HSL-selective inhibitor, CAY10499, did not affect steroid production (Fig. S3D), consistent with the results of HSL-deficient clones.

Both Reduced de Novo Cholesterol Synthesis and Isoprenylation Contribute to Impaired Steroidogenesis in SCAP-Deficient Cells.

Our results demonstrated that PDE4+8 inhibition resulted in phosphorylation of a large number of proteins, including several guanine nucleotide exchange factors (GEFs) and GTPase-activating proteins (GAPs) for Rab- and Rho-type GTPases. Those small GTPases are thought to be involved in vesicle trafficking, lipoprotein uptake, and processing in MA10 cells (Table 1) (8). We therefore investigated the relative contribution of cholesterol synthesis versus mobilization through these pathways by using the HMG-CoA reductase inhibitor lovastatin and inhibitors of farnesylation (FTIs) and geranylgeranylation (GGTIs) that inhibit isoprenylation of a large number of proteins, including those involved in the uptake and intracellular vesicular transport of cholesterol. Lovastatin dose-dependently reduced steroidogenesis induced by PDE4+8 inhibitors in ms-16 cells (Fig. 2F). The maximum inhibitory effect of this statin was reached at 1 µM, which produced an inhibitory effect on progesterone production equivalent to that of SCAP deficiency (Fig. 2F). Incubation with lovastatin for 18 h caused a more marked inhibition of steroidogenesis both in the presence and absence of serum, compared with the acute 2-h incubation (compare Fig. 2F with Fig. S3 B and C). Addition of mevalonate completely rescued the attenuated steroidogenesis in lovastatin-treated cells, demonstrating that lovastatin did not exert off-target effects (Fig. 2H). We next treated the cells with GGTIs or FTIs. Each of these inhibitors significantly reduced steroidogenesis (Fig. 2G). Furthermore, treatment of the cells with geranylgeranyl pyrophosphate (GGPP) at 50 µM partially reversed the inhibitory effect of lovastatin on steroidogenesis in WT cells, suggesting that part, but not all, of the effect of lovastatin on steroidogenesis is mediated by de novo cholesterol biosynthesis and part by isoprenylation (lovastatin inhibits both). In SCAP-deficient cells, GGPP at 30–50 µM significantly increased steroidogenesis (Fig. 2H), demonstrating that in this respect SCAP-deficient cells behave like WT cells in which cholesterol biosynthesis has been inhibited by lovastatin. Furthermore, exogenous cholesterol (methyl-β-cyclodextrin cholesterol and high-density lipoprotein; HDL) augmented steroidogenesis stimulated by PDE4+8 inhibition (Fig. 2E). In cells treated with lovastatin, exogenous cholesterol increased steroidogenesis to a lesser extent than in cells incubated in the absence of lovastatin (Fig. 2E). Together, these results suggest that a significant portion of cholesterol used in SCAP-dependent steroidogenesis is provided through de novo cholesterol biosynthesis but that the SCAP-dependent pathway also regulates isoprenylation that in turn is needed for optimal steroid hormone production. It should be noted that several of the small GTPases that are regulated by the GEFs and GAPs phosphorylated in response to the PDE4 + PDE8 inhibitors likely require isoprenylation to be fully active in the cell.

Reexpression of SCAP in SCAP-Deficient Clonal Cell Lines Restores SREBP2 Expression and Steroidogenesis.

As noted previously in the livers of SCAP-deficient mice (18), SCAP deficiency generated by the Cas9 and FokI/dCas9 systems in MA10 cells was characterized by a marked reduction in SREBP2 protein levels (Figs. 1A and 3A). This likely resulted from reduced SREBP2 mRNA and accelerated SREBP2 degradation. Expression of GFP-SCAP restored SREBP2 levels to wild-type levels (Fig. 3A). Furthermore, GFP-SCAP dose-dependently increased progesterone production (Fig. 3B), suggesting that a more complete transfection efficiency would also restore steroidogenesis. A SCAP-dependent restoration of SREBP2 protein levels was also demonstrated by immunocytochemistry in GFP-SCAP–expressing cells showing increased SREBP2 staining compared with the “no GFP-expressing” cells within the same microscopy field of vision (Fig. 3C). As a control, the ER marker PDI (protein disulfide isomerase) showed no differences in ER staining in cells expressing GFP-SCAP and cells that did not (Fig. 3C).

Fig. 3.

Restoration of SREBP2 and steroidogenesis by reexpression of SCAP in a SCAP-deficient cell clone. The cell clones s-138 and ms-16 were transfected with GFP-SCAP (2.5, 5, or 7.5 µg DNA in 2 × 10⁶ cells) or mock transfected together with a pIRS-GFP/Puro plasmid. Twenty-four hours later, puromycin was added to the medium and the cells were cultured for another 24 h. (A) Cell extracts were used to determine SCAP, GFP-SCAP, and SREBP2 by Western blot analysis. (B) The cells were treated with vehicle (unfilled column) or PDE4+8 inhibitors (filled column) for 2 h under serum-starved conditions for measurement of steroidogenesis. Each value represents mean ± SEM (n = 4). Representative results from three repeated experiments are shown. *P < 0.05 and **P < 0.01 denote statistical significance. (C) SCAP-deficient s-138 cells transfected with GFP-SCAP without the pIRS-GFP-Puro plasmid were immunostained with either an anti-SREBP2 antibody or an anti-PDI antibody (ER marker), as described in SI Materials and Methods.

Combined Inhibition of PDE4 and PDE8 Increases SREBP2 Activity.

To examine whether PDE4+8 inhibition might regulate SCAP–SREBP2 activity, the cleaved mature form of SREBP2 in nuclear fractions was measured by using a rabbit anti-SREBP2 antibody recognizing the active SREBP2 N terminus. Inhibition of PDE4+8 significantly (∼1.8-fold) increased the relative abundance of the mature form of SREBP2, compared with vehicle treatment (Fig. 4A). The full-length cytosolic SREBP2 may have been slightly (nonsignificantly) reduced by PDE4+8 inhibition. To clarify whether full-length SREBP2 in complex with SCAP is reduced by PDE inhibition, MA10 cells were transfected with GFP-tagged SCAP and treated with the combined PDE inhibitors. SREBP2–GFP-SCAP complexes were then immunoprecipitated with an anti-GFP antibody, and full-length SREBP2 in the immunoprecipitates was detected by Western blot analysis. The results demonstrated a significant reduction of full-length SREBP2 immunoprecipitated with GFP-SCAP in cells treated with PDE inhibitors, suggesting that active PDE4 and PDE8 retard processing and subsequent activation of SREBP2 (Fig. 4B). In addition, endogenous SREBP2–SCAP complexes were immunoprecipitated with a mouse anti-SREBP2 antibody that recognizes both the full-length and cleaved C-terminal portion of SREBP2 (the remaining inactive portion of SREBP2 after cleavage). Total SCAP bound to SREBP2 (both the precursor and the cleaved C-terminal portion, as detected with the mouse SREBP2 C-terminal antibody) was increased in the cells treated with PDE4+8 inhibitors compared with nontreated cells (Fig. 4C), suggesting that more of the SREBP2 C-terminal peptide is bound to SCAP in response to PDE inhibition. Overall, these results indicate that PDE4+8 inhibition facilitates SREBP2 processing possibly by increasing SREBP2–SCAP binding.

Fig. 4.

PDE4+8 inhibition increases SREBP2 activity. (A) MA10 cells were serum-starved for 3 h and then treated with vehicle or PDE4+8 inhibitors. Nuclear and cytosolic fractions were isolated and used to determine mature SREBP2 and precursor SREBP2, respectively, by using a rabbit anti-SREBP2 antibody detecting both full-length and cleaved SREBP2. The values in the bar graph represent mean ± SEM of densitometric data obtained from three separate experiments. (B) MA10 cells transfected with GFP-SCAP were treated with vehicle or PDE4+8 inhibitors for 2 h, and cell extracts were used for immunoprecipitation with an anti-GFP antibody. Immunoprecipitated samples and the lysate (6%) were used to determine SREBP2 levels (rabbit SREBP2 antibody) and expressed GFP-SCAP. 14-3-3 protein was used as a loading control. The values in the bar graph represent mean ± SEM of densitometric data obtained from three separate experiments. IB, immunoblotting; IP, immunoprecipitation. (C) MA10 cells were treated with vehicle or PDE4+8 inhibitors for 2 h, and cell extracts were then used for immunoprecipitation with a mouse anti-SREBP2 antibody (anti-SR) recognizing full-length SREBP2 and the C-terminal fragment of cleaved SREBP2, or control IgG (cont IgG). Immunoprecipitated samples and the lysate (6%) were used to determine SCAP levels. The values in the bar graph represent mean ± SEM of densitometric data obtained from two separate experiments each performed in triplicates. *P < 0.05 and **P < 0.01 denote statistical significance.

PKA-Dependent Phosphorylation of SCAP in Response to PDE4+8 Inhibition.

Because simultaneous inhibition of PDE4 and PDE8 resulted in increased SREBP2 processing, we hypothesized that cAMP/PKA-dependent phosphorylation events promote SREBP2/SCAP signaling as part of the response leading to increased steroidogenesis. S821 in SCAP is located within the WD domain, which is known to associate with the SREBP2 C-terminal domain (24, 25). To investigate the potential role of this phosphorylation event, a custom-made S821 phospho-specific antibody was used to confirm that the increase of S821 phosphorylation noted in the mass spectrometry data could be seen by another method. The specificity of the phospho-antibody was confirmed by Western blot (Fig. S4). A large increase in the S821 band was induced by the combination of PDE4+8 inhibitors. A smaller increase was seen with the PDE8 inhibitor alone (Fig. 5A) and essentially no effect was seen with the PDE4 inhibitor alone, mimicking the effect on steroidogenesis previously reported (5) and the mass spectrometry data (8). Furthermore, stimulation of the cells with either LH (10 ng/mL) or 8Br-cAMP (300 µM) also increased SCAP S821 phosphorylation. The PKA inhibitors H89 (10 µM) and Rp-CPT-cAMPS (0.5 mM) both partially blocked the effect of PDE4+8 inhibition (Fig. 5A), indicating that this phosphorylation event was likely to be largely PKA-dependent. However, we cannot rule out the possibility of some contribution of another kinase. Immunostaining using the phospho-S821 SCAP antibody detected signals only in cells treated with PDE inhibitors (Fig. 5B).

Fig. S4.

PDE4+8 inhibitors increase Ser821 phosphorylation of SCAP. (A) MA10 cells were serum-starved for 3 h and then treated with vehicle or a combination of PDE8 inhibitor (200 nM) and PDE4 inhibitor (10 µM) for the indicated incubation times. The cell extracts were used to determine SCAP S821 phosphorylation by Western blot. Note the single band detected by the antibody. (B) Antibody specificity was further confirmed in a SCAP-deficient cell clone. A WT cell clone (ms-16) and a SCAP-deficient cell clone (s-143) were serum-starved for 3 h and then treated with PDE8 and PDE4 inhibitors for 30 min. The cell extracts were used to determine phospho-S821 SCAP, total SCAP, or GAPDH. Note the absence of bands in the SCAP-deficient cell clone.

Fig. 5.

PDE4+8 inhibitors and PKA activators increase Ser821 phosphorylation of SCAP. (A) MA10 cells were serum-starved for 3 h and then treated with vehicle, PDE8 inhibitor (200 nM), PDE4 inhibitor (10 µM), or PDE4+8 inhibitors for the indicated incubation times. H89 (10 µM) or Rp-8-CPT-cAMPS (Rp; 0.5 mM) was added to the cells 30 min before addition of PDE4+8 inhibitors (20-min stimulation). The intensity of p-SCAP was densitometrically measured and is shown as an average (SD) (n = 3). Serum-starved MA10 cells were also treated with LH (20 ng/mL) or 8Br-cAMP (300 µM) for 10–40 min. The cell extracts were used to determine the phosphorylated SCAP at S821. Total SCAP or GAPDH was used as a loading control. (B) MA10 cells transfected with GFP-SCAP were used for immunostaining with the phospho-SCAP–specific antibody (Ser821). Phospho-SCAP immunoreactivity was labeled with an anti-rabbit secondary antibody conjugated with Alexa 546 (red). TOPRO3 was used as a nuclear counterstain (blue).

Discussion

The data in our previous phosphoproteomic study (8) and this study strongly suggest that cAMP/PKA coordinates not just one but rather a large number of different processes that operate together to provide sufficient cholesterol for maximal cAMP-stimulated steroidogenesis. This conclusion is based mostly on the identity and known roles of the large number of different proteins phosphorylated on PKA consensus sites in response to what should be a relatively pure cAMP signal (i.e., a signal caused by inhibition of PDEs 4 and 8). Two of the phosphoproteins seen were SCAP and HSL/cholesteryl ester hydrolase (Table 1).

In this study, we have used CRISPR-Cas9–mediated gene inactivation to investigate in some detail the SCAP/SREBP pathway, and found a heretofore unrecognized and important role for cAMP/PDE4+8/PKA in the SCAP/SREBP pathway that likely contributes to the large increase in steroidogenic response to cAMP and LH.

We decided to knock down SCAP by using two different CRISPR gene-editing protocols [CRISPR-Cas9 and FokI/dCas9 (23)] because standard RNAi and shRNA gene-depletion protocols usually are quite difficult in these cells, as they are hard to fully transfect. As previously seen in the liver of SCAP knockout mice (18), MA10 cells with SCAP deficiency had greatly reduced levels of SREBP2. Importantly, these cells showed markedly attenuated steroid production in response to either PDE4+8 inhibition or to LH. SCAP deficiency also attenuated mRNA levels of the SREBP2 targets Hmgcr, Hmgcs, Fdps, Cyp51, and Ldlr, as previously reported in different tissues and cell types (18). Furthermore, overexpression of exogenous SCAP restored both attenuated SREBP2 and steroidogenesis, suggesting that SCAP is necessary for optimal steroidogenesis in these cells.

We then focused on examining whether cAMP/PKA, through inhibition of PDE4 and PDE8, could directly modulate SREBP2 processing. The PDE inhibitors increased the presence of cleaved mature SREBP2 in the nuclear fraction while reducing the amount of full-length SREBP2 bound to SCAP, clearly indicating that PDE4+8 inhibition caused SREBP2 activation. One explanation for this effect could be that PDE inhibition increased formation of SCAP–SREBP2 complexes and/or SCAP recycling after SREBP cleavage (26), thereby increasing the pool of SREBP2 available for subsequent cleavage and activation. Thus, immunoprecipitation with a C-terminal anti-SREBP2 antibody resulted in more SCAP bound in cells treated with the PDE inhibitors, compared with untreated cells. It also is possible that PKA reduces Insig availability (27) and/or results in recruitment of unknown adapter proteins to either SCAP or SREBP2. Such processes might allow formation of more SREBP2–SCAP complexes.

Our phosphoproteomic study demonstrated an increased phosphorylation of SCAP S821 in response to PDE4+8 inhibition (Table 1). Interestingly, S821 is located in SCAP’s WD domain, which directly binds to the C terminus of SREBP2. We confirmed that PKA activation through PDE4+8 inhibition increased the phosphorylation of S821 in SCAP by developing a phospho-S821–specific antibody. S821 phosphorylation was clearly increased by cAMP agonists as well as the PDE inhibitors and was sensitive to PKA inhibitors (H89 and Rp-8-CPT-cAMPS) in MA10 cells. Interestingly, our recent phosphoproteomic study (8) identified only 54 phosphorylation sites as significant regulatory sites in the response to PDE8 inhibition alone, compared with 749 sites regulated by the combination of PDE4 and PDE8 inhibitors. S821 was one of the 54 regulatory sites, which might be characterized as low-threshold for cAMP and/or be location-specific, possibly Golgi-specific in the case of SCAP, because PDE8B also is localized in the Golgi apparatus (5).

Other phosphorylation sites likely to be of direct importance in the SCAP/SREBP activation pathway were also seen in the phosphoproteomic studies (Table 1). For example, Sec23-interacting protein (S748), oxysterol-binding protein-related protein 11 (S194), and Sec22b (S137) are likely to be involved in the production and transport of the SCAP/SREBP-containing vesicles (28). In fact, a PKA inhibitor has been reported to prevent the recruitment of Sar1 and Sec23/24 to the sites forming the COPII coat complex, thereby preventing ER export of cargo (29). These findings suggest that the phosphorylation on SCAP S821 may be just one of the modifications needed in order for cAMP/PKA to exert an effect on SCAP function. Similarly, the recycling of SCAP back to the ER is likely to involve ARF1-mediated pathways, and several small GTPase regulatory proteins important to these pathways were also phosphorylated on consensus PKA sites by the combination of PDE4 and PDE8 inhibitors (Table 1). The relative importance of these other sites remains to be elucidated in future work.

Our study also suggests that a major source of cholesterol used for steroidogenesis in serum-starved Leydig cells is provided through de novo synthesis, consistent with earlier studies demonstrating that primary Leydig cells and MA10 tumor Leydig cells do not depend heavily on extracellular cholesterol uptake (30, 31), at least acutely. Thus, the HMGCoA reductase inhibitor lovastatin inhibited steroidogenesis in WT cells to levels similar to those seen in SCAP-deficient cells. Isoprenylation of proteins provided by farnesyl pyrophosphate and geranylgeranyl pyrophosphate through the cholesterol synthesis pathway is likely to contribute as well, because addition of inhibitors of farnesyl or geranylgeranyl transferases led to significant impairment of cAMP-dependent steroidogenesis, which was partially reversed by geranylgeranyl pyrophosphate. Furthermore, addition of exogenous cholesterol did not fully restore progesterone production in lovastatin-treated cells. It is therefore likely that prenylation of proteins involved in trafficking of cholesterol from the plasma membrane or intracellular compartments by mechanisms that may rely on prenylated proteins also contribute to steroidogenesis (32). In addition, we found that levels of active RhoA were increased by the PDE4+8 inhibitor treatment in MA10 cells (Fig. S5), and significant phosphorylation events in Arhgap17 and Arhgef2, which are known regulators of Rho activity (33–35), were identified in the phosphoproteomic study (Table 1). Importantly, these findings strongly suggest that PKA activation could modify the activities of these GAPs and GEFs (in addition to regulating prenylation), thereby resulting in a faster activation cycle of Rho-type GTPases that are involved in vesicle transport and microfilament arrangement (36). These molecular events might also be expected to influence cholesterol transport to mitochondria for steroidogenesis (37). A similar argument can be made for several different GEFs and GAPs that regulate Rab proteins, which are also highly implicated in regulation of vesicle trafficking (Table 1).

Fig. S5.

Active RhoA is increased by PDE4+8 inhibition in MA10 cells. Cells were serum-starved for 3 h and treated with vehicle or PDE4+8 inhibitors for 15 or 30 min. GTP-bound active RhoA was immunoprecipitated with rhotekin-RBD beads and analyzed by Western blot using an anti-RhoA antibody. A similar level of total RhoA in each sample is detected. (A) A representative blot from three repeated experiments is shown. (B) The intensity of each GTP-bound RhoA and total RhoA was measured by ImageJ (NIH). Values of GTP-bound RhoA relative to total RhoA represent mean ± SEM of three repeated experiments. *P < 0.05 and **P < 0.01 denote statistical significance vs. vehicle.

Finally, the data strongly suggest that PKA has multiple known targets in both the cholesterol biosynthetic and uptake/mobilization pathways, some of which appear to be partly independent of SCAP. For example, the SREBP2-regulated genes Hmgcr, Fdps, Cyp51, and Ldlr were all significantly up-regulated by the PDE inhibitors in both WT and SCAP-deficient cells, although the absolute levels were lower in SCAP-deficient cells. Some portion of these effects could be due to PKA-mediated activation of the transcription factors CREB, SF-1 (Nr5a1), GATA4, and NF-γ/p300, because maximal transcriptional activation by SREBPs requires activation of multiple transcription factors (38–41). In addition to the facilitation of cholesterol biosynthesis, the up-regulation of Hmgcr and Fdps by PKA should also increase the availability of isoprenoids that are necessary to locate small GTPases to their proper targets.

cAMP elevation and PKA activation are known to stimulate several critical steps in steroid production (42). PKA activation leads to activation of HSL (5, 43), which releases cholesterol and free fatty acids from cholesteryl ester storage and perhaps other sites in the cell. However, we demonstrate using HSL-deficient cell clones that HSL appears not to contribute significantly to cAMP-dependent steroidogenesis under the experimental conditions studied. Perhaps stimulation of HSL activity is only important under certain conditions of cholesterol availability. Ultimately, the liberated free cholesterol from all sources is transported into mitochondria by StAR protein, which is also reported to be phosphorylated and activated by PKA (44, 45). Once in mitochondria, cholesterol is converted to pregnenolone by CYP11A1 and then further processed to progesterone and testosterone. Both StAR and CYP11A1 expression are induced by PKA, as demonstrated previously and in the present study, and are generally considered as rate-limiting steps in steroidogenesis (46, 47). However, the results of the present study suggest that other PKA substrates can also be regulatory with regard to cAMP activation of steroid synthesis, and add PKA-mediated regulation of SCAP function as a likely critical step for cAMP-dependent regulation of steroidogenesis.

In summary, we demonstrate that SCAP deficiency attenuates Leydig cell steroidogenesis in response to LH and cAMP elevation, and that simultaneous inhibition of PDE4 and PDE8 results in facilitated processing and activation of SREBP2. These results highlight an important role for SCAP and the cholesterol biosynthetic pathway in steroidogenesis in response to LH and cAMP elevation by PDE4 and PDE8 inhibition, and may open new avenues for modulation of SCAP-dependent processes by PDE inhibitors in other tissues.

Materials and Methods

A detailed description of experimental procedures and reagents is provided in SI Materials and Methods. In short, phosphorylation of proteins in response to cAMP elevation by combined inhibition of PDE4 and PDE8 in MA10 Leydig cells was determined by mass spectrometry. Two CRISPR-Cas9 approaches were used to knock down expression of SCAP, one of the proteins involved in cellular cholesterol handling identified by the phosphoproteomic screen. In some experiments, SCAP was reexpressed in SCAP-deficient cells by transfection by a GFP-SCAP plasmid. Activation of the SCAP–SREBP2 pathway in response to PDE4+8 inhibition was measured by immunoprecipitation of the SCAP–SREBP2 complex, detection of nuclear mature SREBP2, real-time PCR to evaluate downstream gene expression, and progesterone production. The relative contribution of substrate for steroidogenesis provided by de novo cholesterol synthesis and isoprenylation-dependent mechanisms was evaluated by inhibitors of cholesterol synthesis and prenylation and exogenous addition of substrates for these pathways.

SI Materials and Methods

Materials.

The PDE8 and PDE4 inhibitors PF-04957325 and rolipram, LH, 8Br-cAMP, the PKA inhibitor H89, protease inhibitor mixture, and culture medium were obtained as previously described (8). Lovastatin was purchased from Tocris Bioscience. Methyl-β-cyclodextrin-cholesterol was from MP Biomedicals. GGTI-298 and FTI-276 were purchased from Millipore. Mevalonate and geranylgeranyl pyrophosphate (in MeOH/NH₄OH) were purchased from Sigma. Rp-8-CPT-cAMPS was purchased from Biolog. CAY10499 was purchased from Cayman Chemical. Goat polyclonal antibodies against SCAP (C-20) and GFP (I-16) and the mouse monoclonal antibody against SREBP2 (1C6) were purchased from Santa Cruz. The rabbit polyclonal antibody against SREBP2 was purchased from Cayman Chemical. The rabbit polyclonal antibody against phospho-SCAP (Ser821) was custom-produced by LifeTein. We confirmed the specificity of this antibody by Western blot using SCAP-deficient cells. No signal was observed in these cells. Rabbit antibodies against HSL and phospho-HSL (Ser660) were purchased from Cell Signaling Technology. The mouse monoclonal antibody against PDI (RL90) was purchased from Thermo Scientific. The optimized Csy4 and FokI/dCas9 expression plasmid pSQT1601 and guide RNA expression plasmid pSQT1313 were purchased from Addgene. pIRES-hrGFPII/Neo was purchased from Agilent Technologies. The GFP-SCAP vector was kindly provided by P. J. Espenshade (Johns Hopkins University, Baltimore, MD). The mouse HDL isolated by ultracentrifugation was provided by the Quantitative and Functional Proteomics Core, Diabetes Research Center (University of Washington).

Cell Culture.

MA10 cells and clonal cells derived from MA10 were grown in RPMI1640 medium containing 15% (vol/vol) horse serum and penicillin (100 U/mL)/streptomycin (100 µg/mL) in a humidified incubator at 37 °C in a 5% CO₂ atmosphere, as described previously (5). The cells tested negative for mycoplasma contamination by the MycoProbe Mycoplasma Detection Kit (R&D Systems). The cells were plated on 24-well plates 2 d before measurements of steroidogenesis. After the culture medium was removed, the cells were washed with RPMI1640 twice and incubated for 3 h under serum-starved conditions. The RPMI was then replaced with fresh RPMI containing either vehicle or test drugs such as lovastatin, H89, Rp-8-CPT-cAMPS, mevalonate, methyl-β-cyclodextrin-cholesterol, HDL, GGTI-298, FTI-276, geranylgeranyl pyrophosphate, or CAY10499 and preincubated for 30 min. A one-twentieth volume of 20× rolipram (final concentration 10 µM) and PF-04957325 (final concentration 200 nM) mixture or LH was added and the cells were then incubated for an additional 2 h. When steroidogenesis was measured under serum-fed conditions, the medium was replaced with fresh culture medium containing vehicle or a test drug without the washing steps. The PDE inhibitor mixture was added to stimulate cells as described above. The medium was collected for progesterone assay and kept at −20 °C until use. For preparation of samples for Western blot, the cells in a culture well were dissolved directly in a 1× SDS sample buffer and boiled for 5 min. All stock solutions of the test drugs were prepared in DMSO. Within an experiment, the total final concentration of DMSO (less than 0.1%) was kept constant between conditions.

Generation of a SCAP-Deficient Cell Pool by CRISPR-Cas9.

We generated SCAP-deficient MA10 cells by two different CRISPR-Cas9 approaches. The first approach was used to knock down SCAP in a population of MA10 cells by using a pool of three plasmids, each encoding the Cas9 nuclease and a commercially designed mouse SCAP-specific 20-nt guide RNA (SCAP CRISPR-Cas9 KO plasmid sc-433484; Santa Cruz). Cell transfection was performed using the Neon Transfection System (Invitrogen). MA10 cells at 70% confluency on 100-mm dishes were trypsinized, washed, and pelleted by centrifugation at 400 × g (1.5 × 10⁵ cells). The cell pellet was resuspended in 10 µL of buffer R provided by the manufacturer and mixed with 0.5 µg CRISPR-Cas9 KO plasmid and 0.5 µg homology-directed DNA repair (HDR) plasmid (SCAP sc-433484-HDR). The mouse SCAP HDR plasmid consists of a pool of plasmids, each containing a HDR template corresponding to the cut sites generated by the SCAP CRISPR-Cas9 KO plasmid, two 800-bp homology arms designed to specifically bind to the genomic DNA surrounding the corresponding Cas9-induced double-strand DNA break site, a puromycin-resistance gene for selection of stable knockout cells, and a red fluorescent protein gene. The cell suspension was put into a 10-µL syringe and electroporated (1,400 V, 20 ms, one pulse) according to the manufacturer’s protocol. The cells were immediately plated on six-well plates and cultured for 24 h. The transfected cells with the mutated SCAP gene along with an insertion of a puromycin-resistance gene were then selected in medium containing puromycin (20 µg/mL) and pooled (SCAP KD pool). A portion of the same batch of cells used for CRISPR gene editing was maintained with the CRISPR cells side by side and used as WT cells.

Generation of SCAP and HSL Knockout Clonal MA10 Cell Lines by the CRISPR-FokI/dCas9 System.

The dimeric CRISPR-FokI/dCas9 genome-editing system was alternatively used for mutagenesis of the SCAP and HSL genes, taking advantage of the method described by Tsai et al. (23). The guide RNA sequences were chosen by using ZiFiT Targeter software (zifit.partners.org) as follows: SCAP target site 1 (labeled guide A in Fig. S2), 5′-ACCACACAAGTCTGAGCTAA-3′; SCAP target site 2 (labeled guide B in Fig. S2), 5′-CCACGCCGGGATAGCTGCGG-3′; HSL target site 1, 5′-ACGTGTAGAGGGGCATGTGG-3′; HSL target site 2, 5′-CCCCTTCATGTCTCCTCTGC-3′, as described in Fig. S2. Complementary oligonucleotides with appropriate overhangs for two sites and spacer oligonucleotides were annealed and ligated into a multiplex expression plasmid, pSQT1313. The following oligonucleotides were used: SCAP target site 1, 5′-GCAGACCACACAAGTCTGAGCTAAGTTTTAG-3′ and 5′-AGCTCTAAAACTTAGCTCAGACTTGTGTGGT-3′; SCAP target site 2, 5′-GGCAGCCACGCCGGGATAGCTGCGG-3′ and 5′-AAACCCGCAGCTATCCCGGCGTGGC-3′; HSL target site 1, 5′-GCAGACGTGTAGAGGGGCATGTGGGTTTTAG-3′ and 5′-AGCTCTAAAACCCACATGCCCCTCTACACGT-3′; HSL target site 2, 5′-GGCAGCCCCTTCATGTCTCCTCTGC-3′ and 5′-AAACGCAGAGGAGACATGAAGGGGC-3′. The ms-16 cell line, a highly cAMP-responsive steroidogenic cell line clone isolated from the original MA10 cells, was used in this set of experiments. The cells were trypsinized, washed, and pelleted by centrifugation at 400 × g (1.5 × 10⁵ cells). The cell suspension in 10 µL of buffer R was mixed with pSQT1601 (0.75 µg), pSQT1313 (0.25 µg) containing guide sequences for either SCAP or HSL, and pIRES-hrGFPII/Neo (0.08 µg). Control cells were mixed with pSQT1601 and pIRES-hrGFPII/Neo plasmid. The cell suspension was electroporated (1,400 V, 20 ms, one pulse) with a 10-µL syringe as described above. The transfected cells were immediately plated, cultured for 24 h, and then selected by G418 (600 µg/mL) for 3 d. The selected cells were subcultured at low density (100–150 cells per 100-mm dish) and grown until visible colonies were formed. Each colony was removed onto a 12-well plate and grown until enough cells for screening and further subculturing were accumulated. To correctly interpret the gene-editing results, individual cell clones were karyotyped (Diagnostic Cytogenetics). Like most transformed cell lines, the ms-16 cells and resulting clones were found to have structural and numerical abnormalities. The clones analyzed were mostly tetraploid. The SCAP and HSL wild-type and mutant cell clones were verified by sequencing (Fig. S2). To screen clonal cells, cell pellets collected from each clone were directly resuspended in Tris-EDTA buffer, boiled for 5 min, and used for PCR amplification. The region containing the CRISPR target was amplified with primers as follows: for SCAP, forward (5′-GATCCCACGCCCAGGGTAAG-3′) and reverse (5′-GGAGGGCTGTCTCCAGGTTC-3′); for HSL, forward (5′-GGAAGAGGCTGAAGCCAAAG-3′) and reverse (5′-AGGTCGCCCAAGTCACATC-3′). The PCR products were separated on 4% (wt/vol) agarose gels (Fig. S2). Clones showing multiple amplicons were considered as carrying one or more mutated alleles (Fig. S2). The efficiency of gene editing was 80.8% (42 positive clones among 52 tested) for SCAP and 41.2% (7 positive clones among 17 tested) for HSL. TA cloning of the PCR products was performed using a TOPO TA Cloning Kit (with Topoisomerase I-activated pCR2.1-TOPO vector; Invitrogen), and subsequent sequencing confirmed gene editing as shown in Fig. S2.

GFP-SCAP Plasmid and Transient Transfection.

Transient transfections were performed using the Neon Transfection System as described above. MA10 cells or SCAP-deficient cell clones were trypsinized and pelleted by centrifugation at 400 × g (2 × 10⁶ cells). Cell suspensions in 100 µL of buffer R were mixed with GFP-SCAP (5 µg or as described in Fig. 3). In the experiments measuring steroidogenesis, pIRES-hrGFPII/Neo (1 µg) was added to select for transfected cells. Mock transfection was performed in the presence of Tris-EDTA buffer or pIRES-hrGFPII/Neo only. The cells were electroporated with a 100-µL syringe and immediately plated on 48-well plates and cultured for 24 h followed by treatment with G418 (600 µg/mL) for an additional 24–48 h. The cells were then used for measurements of steroidogenesis or processed for Western blot analysis of protein levels. For immunostaining, the cells were transfected with GFP-SCAP without pIRES-hrGFPII/Neo, plated on eight-well chamber slides, and used for staining 2 d after transfection.

Measurement of Cellular Cholesterol.

The wild-type ms-16 cell clone and the SCAP-deficient s-143 cell clone cultured on 100-mm dishes were serum-starved for 3 h and collected. Subcellular fractionation was performed by the method previously described by Poderoso et al. (51), with minor modifications as follows. The cells were collected by trypsinization, resuspended in a homogenization buffer (219 mM d-mannitol, 90 mM sucrose, 5 mM EDTA, 0.1% BSA, 1.8 mM Hepes, 1 mM Na₂VO₃, 2 mM NaF, and 1:100 proteinase inhibitor mixture), and mechanically digested by 40 strokes with an insulin syringe. One-tenth of the homogenate (whole-cell fraction) was removed and diluted in a detergent buffer (final 0.8% Na cholate and 0.1% Triton X-100 in PBS). The rest of the homogenate was centrifuged at 3,000 × g for 20 min and the supernatant was further centrifuged at 15,000 × g for 30 min. The pellet (crude mitochondrial fraction) was washed once and dissolved in a detergent buffer (final 0.8% Na cholate and 0.1% Triton X-100 in PBS). Fractions were subjected to free cholesterol measurement by using the Amplex Red Cholesterol Assay Kit (Invitrogen) as well as protein measurement (BCA Protein Assay Kit; Thermo Scientific). The mitochondrial fraction was validated by Western blot using a rabbit anti-VDAC1 antibody (Abcam; ab-15895).

Nuclear Fractionation for Mature SREBP2.

MA10 cells cultured on 60-mm dishes were treated for 1 h with vehicle or PDE inhibitors after a 3-h serum starvation (three dishes per condition). The cells were washed with cold PBS, scraped, and spun down. The cell pellets were processed to obtain a cytosol fraction and a nuclear fraction using NE-PER nuclear and cytoplasmic extraction reagents (Thermo Scientific) according to the manufacturer’s protocol. Isolated fractions were prepared for Western blot analysis. Protein samples were resolved by SDS/PAGE and transferred to nitrocellulose membranes. Blots were blocked in 5% (wt/vol) milk in Tris-buffered saline with 0.05% Tween 20 and incubated with a rabbit SREBP2 antibody for 5 h at room temperature. Antibody binding was visualized with a goat anti-rabbit peroxidase-conjugated secondary antibody and the enhanced chemiluminescence method (Thermo Scientific). Equal loading was verified by a glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody.

Immunoprecipitation.

Cells transfected with GFP-SCAP or untransfected cells were cultured on six-well plates or 100-mm dishes for 2 d. They were then treated for 1 h with vehicle or PDE inhibitors after a 3-h serum starvation. The cells were washed with cold PBS twice and dissolved in lysis buffer [20 mM Tris, 150 mM NaCl, 2 mM EDTA, 0.5% Triton X-100, 1 mM NaF, 1 mM Na₃PO₄, 1.5 mM PMSF, and protease inhibitor mixture (1:100), pH 7.3]. The cell lysate was gently agitated on a rotator for 1 h at 4 °C and centrifuged at 14,000 × g for 20 min at 4 °C. The supernatant was preabsorbed with protein A/G agarose for 1 h at 4 °C. The recovered supernatant was then incubated with control IgG, anti-GFP antibody, or anti-SREBP2 antibody at 4 °C overnight and further incubated with protein A/G agarose beads for 2 h at 4 °C. The beads were washed three times with lysis buffer, dissolved in an SDS buffer, and boiled for 5 min. Proteins were resolved by SDS/PAGE and transferred to nitrocellulose membranes. Western blotting was performed as described above.

Quantitative Real-Time PCR.

WT and SCAP KD cell pools cultured on 60-mm dishes were treated for 18 h with vehicle or PDE inhibitors under serum-starved conditions. The cells were rinsed with cold PBS, scraped off, and spun down. The cell pellets were processed to collect RNA by the NucleoSpin RNA isolation kit (Macherey-Nagel). cDNA was then synthesized with SuperScript II reverse transcriptase (Thermo Fisher Scientific) and random hexamers. Quantitative real-time PCR was performed with SYBR Green (Bio-Rad), and each primer set is listed in Fig. S1. Target gene mRNA levels were normalized to levels of Rn18s.

Immunofluorescence Microscopy.

Cells transfected with GFP-SCAP constructs were plated on eight-well chamber slides (BioCoat culture slide; BD Biosciences). Two days later, the cells were serum-starved for 3 h and then incubated in the presence or absence of PDE inhibitors for 1 h. The cells were rinsed with PBS, fixed with 4% (wt/vol) paraformaldehyde, and processed for immunostaining as described previously (5). Briefly, the cells were blocked with 5% (wt/vol) goat serum, 0.01% Triton X-100, and 1 mg/mL BSA in PBS for 1 h at room temperature and then incubated with either mouse anti-SREBP2 antibody, mouse anti-PDI antibody, or rabbit anti–phospho-SCAP antibody. The primary antibodies were labeled by a goat secondary antibody conjugated with Alexa 546 or Alexa 633 in sequential steps. Confocal microscopic imaging was performed with a Leica SP8 X (Leica Microsystems) at the Keck Microscopy Facility, University of Washington.

Measurement of Active RhoA.

MA10 cells cultured on six-well plates were treated for 15 or 30 min with vehicle or PDE inhibitors after a 3-h serum starvation. At the end of the incubation, the cells were rinsed with cold PBS and immediately lysed. Sample preparations and immunoprecipitation of active RhoA were performed by using a Rho Activation Assay Biochem Kit (BK036; Cytoskeleton). Briefly, the cells (three wells per sample) were lysed in cell lysis buffer containing proteinase inhibitor mixture (1:100) and 10 mM NaF and centrifuged at 10,000 × g for 2 min. An aliquot of the supernatant was separated for detecting total RhoA. The rest of the supernatant was incubated with rhotekin-RBD beads on a rotator for 1 h at 4 °C. The beads were pelleted by centrifugation at 5,000 × g for 1 min, washed twice, and dissolved in 6× SDS sample buffer to make a final dilution of 1× SDS, and the samples were then boiled for 2 min for Western blot analysis. All samples as well as positive (GTPγS-treated supernatant) and negative (GDP-treated supernatant) controls were resolved by SDS/PAGE and transferred to nitrocellulose membranes. Western blotting was performed with a mouse anti-RhoA monoclonal antibody provided in the kit.

Cell Labeling and LC-MS/MS for Phosphoproteomics.

Detailed procedures were described previously (8). For SILAC (stable isotope labeling with amino acids in cell culture) labeling, MA10 cells were cultured in custom RPMI medium supplemented with 15% dialyzed FBS, 292 µg/mL glutamine, 20 µg/mL proline, and the corresponding isotope-labeled Lys (20 µg/mL) and Arg (120 µg/mL) (light label, Lys0/Arg0; medium label, Lys4/Arg6; heavy label, Lys8/Arg10). Cells were split every 2–3 d and grown for seven cell doublings on 15-cm cell-culture dishes to achieve complete incorporation of isotope-labeled Lys/Arg. The cells were serum-starved in serum-free SILAC medium for 3 h and treated with either vehicle, PDE4 inhibitor (10 µM rolipram), PDE8 inhibitor (200 nM PF-04957325), or PDE4+8 inhibitors for 1 h (three conditions per analysis; four analyses were conducted). The medium was then aspirated and the cells were washed twice quickly with ice-cold PBS, immediately dissolved in Tris-buffered 8 M urea (pH 7.8) containing a protease and phosphatase inhibitor mixture, and sonicated. Individual SILAC-labeled lysate samples (light, medium, or heavy) were combined at a protein ratio of 1:1:1. Proteins were then digested by endoproteinase LysC and trypsin. Peptides were extracted using Oasis C18 cartridges, enriched using Fe-IMAC (immobilized metal ion affinity chromatography), and separated on a Thermo Scientific Dionex RSLCNano UHPLC. MS data were collected with a Thermo Orbitrap Elite. Raw files were analyzed by MaxQuant version 1.2.5.8 (Computational Systems Biochemistry) using protein, peptide, and site false discovery rates of 0.01 and a score minimum of 40 for modified peptides and 0 for unmodified peptides and a delta score minimum of 17 for modified peptides and 0 for unmodified peptides. MS/MS spectra were searched against the UniProt mouse database. MaxQuant search parameters were as follows: Variable modifications included oxidation (M), phosphorylation (STY), and carbamidomethyl (C).

Statistical Analysis.

Data are presented as the mean ± SD or the mean ± SEM. Statistical significance was determined by using a two-tailed t test, Mann-Whitney test, and one-way ANOVA followed by Tukey’s post hoc test. A value of P < 0.05 was considered to be statistically significant.