PERK-dependent regulation of lipogenesis during mouse mammary gland development and adipocyte differentiation

Deletion of perk in Mouse Mammary Epithelium Impairs Secretory Activation and Attenuates Pup Growth.

Examination of PERK expression at various stages of mammary gland development revealed PERK expression at all developmental stages (supporting information (SI) Fig. S1A), with activation and accumulation of phosphorylated eIF2α species during lactation (18, 19). We next generated a mammary epithelium-specific knockout of perk. Mammary glands isolated from these mice (cKO) revealed undetectable PERK protein at all developmental stages and a corresponding decrease in levels of p-eIF2α (Fig. S1A). Histological analysis revealed normal development of virgin and pregnant (Fig. S1 B and C) glands in cKO mice. By day 3 of lactation (secretory activation), glands of cKO mice exhibited distinct morphological deficiencies (Fig. 1 A and B). cKO glands were morphologically similar to wild-type pregnancy day 16 displaying reduced alveolar expansion with areas of uncollapsed adipocyte stroma (Fig. 1B).

Because PERK regulates protein translation, we assessed milk protein content in cKO glands. Significantly, the milk protein profile was unchanged (Fig. S2A). Because milk fat is a major milk component and energy source in neonates (12), we analyzed milk lipid content. Qualitative analysis of milk lipids by thin-layer chromatography (Fig. S2B), or quantitative analysis of free fatty acids and triglycerides in mouse milk, revealed a significant reduction in free fatty acid content and a trend toward reduced triglycerides in the milk of cKO dams (Fig. 1 C and D). Nile Red (binds triglycerides and free fatty acids; ref. 20) staining of mammary glands revealed decreased lipid droplet size and formation in cKO glands (Fig. 1E). Consistent with reduced fatty acids and triglyceride in milk, growth of pups nursed by cKO dams was significantly attenuated, and average litter weight was reduced by ≈30% by day 12 of lactation (Fig. 1F). Similar growth retardation was observed in wild-type pups nursed by cKO dams, whereas cKO pups exhibited normal growth when fostered by control dams (Fig. S2C).

Perk Deletion Inhibits the Sustained Induction of Lipogenic Enzymes.

Lactating mice rely on de novo synthesis of glycerol and medium chain fatty acids from glucose and amino acids (12, 21). Thus, the decrease in the amount of free fatty acids in the milk may serve as an indicator of reduced levels of precursors for de novo triacylglycerol synthesis. Because expression of key lipogenic enzymes, ATP citrate lyase (ACL), fatty acid synthase (FAS), stearyl-CoA desaturase-1 and -2 (SCD1/SCD2), is induced during secretory activation (21), we assessed expression of these enzymes in mammary gland samples from control and cKO mice. Increased ACL and SCD1 mRNA levels were apparent by lactation day 3 in control and cKO glands, whereas FAS mRNA levels were slightly increased only in control mice (Fig. 2A). Lipogenic enzyme mRNA levels declined to a much greater extent in the cKO glands than in control mice on lactation day 12 (Fig. 2A). ACL and SCD1 protein levels were lower in cKO mice at all stages examined, whereas FAS protein specifically declined in cKO mice on lactation day 12 (Fig. 2B).

Because SREBP transcription factors are major regulators of lipid metabolic genes and can be induced by ER stress (22, 23), we addressed whether PERK deletion inhibited SREBP activation. Indeed, loss of PERK attenuated SREBP1 expression (Fig. 2A), consistent with previous work revealing autoregulation of srebp1 gene expression (24). Additionally, an increased retention of SREBP1 precursor was detected in cKO samples (Fig. 2B), suggesting reduced processing without PERK.

SREBP Maturation Is Induced in a PERK and eIF2α-Dependent Fashion.

The PERK dependence of ER stress-mediated SREBP processing was further tested through utilization of an acute model of PERK deletion in murine embryonic fibroblasts (MEFs); perk LoxP MEFs were transduced with a retrovirus expressing Cre recombinase (PERK Cre) or empty virus (PERK Mock) and treated with thapsigargin (Fig. 2C). To facilitate detection of SREBP1, the cells were engineered to stably express Myc-tagged SREBP1. Acute perk deletion eliminated processing of Myc-SREBP1 (Fig. 2C) and endogenous SREBP1 (Fig. S3A). Although SREBP1 precursor levels rapidly declined upon ER stress, no processed isoform was detected. Because the processed SREBPs are rapidly degraded by the 26S proteasome (25), we inhibited the proteasome with MG132. In the presence of MG132, thapsigargin induced SREBP1 processing with accumulation of the processed SREBP1 in control cells while processing was significantly attenuated in perk^−/− cells (Fig. 2C). Because MG132 induces activation of the cytosolic eIF2α kinase, GCN2 (26), incomplete abrogation of SREBP1 processing in PERK Cre MEFs likely reflects compensatory GCN2 action. Indeed, MG132 alone resulted in a small accumulation of processed SREBP1, suggesting basal processing. Importantly, endogenous SREBP1 precursor levels declined independently of MG132 presence, ruling out SREBP1 degradation during ER stress (Fig. S3B). Essentially no processing of SREBP1 was observed in eIF2α S51A MEFs (27) demonstrating that PERK-dependent SREBP1 activation occurs in an eIF2α-dependent manner (Fig. 2D).

SREBP Maturation During ER Stress Is Mediated by Insig1 Depletion.

SREBP1 processing can be regulated via direct depletion of Insig1, an ER-localized protein that anchors the SCAP–SREBP complex in the ER membrane (22). Because Insig1 is a short-lived protein, analogously to cyclin D1 (28), PERK-mediated translation inhibition should result in the rapid depletion of Insig1. We therefore generated a Myc-tagged Insig1 (due to the lack of suitable antibodies for detecting endogenous protein) and transduced wild-type and perk^−/− MEFs. Thapsigargin treatment triggered a rapid loss of Insig1 in a PERK- (Fig. 3A) and phospho-eIF2α-dependent manner (Fig. S3C).

Robust induction of SREBP1 target genes mRNA (FAS, ACL, and SCD1) was observed in wild-type but not perk^−/− MEFs (Fig. 3B). Consistent with target induction, a 2-fold increase in SREBP1 DNA binding was noted in thapsigargin-treated wild type but not perk^−/− MEFs (Fig. 3C).

PERK Deficiency Inhibits the Lipogenic Pathway During Differentiation of MEFs into Adipocytes.

Given the ability of PERK to regulate SREBP1 activation and lipogenic enzymes expression in mouse mammary epithelium, we reasoned that PERK might also contribute to the initiation of a lipogenic program during adipocyte differentiation. To evaluate this possibility, we cultured PERK Mock MEFs and PERK Cre MEFs, engineered to express Myc-tagged SREBP1, in medium favoring adipocytic differentiation and monitored induction of lipogenic enzymes. Increased SREBP1c, FAS, and SCD1 mRNA levels were observed in control cells by day 7, and ACL mRNA levels were induced by day 9 of treatment; this was significantly attenuated in PERK Cre MEFs (Fig. 4B). Increases in FAS and ACL protein levels were also detected by day 9 of treatment in PERK Mock, but not in PERK Cre MEFs (Fig. 4A). Although transient activation of PERK is difficult to detect, and phosphorylation of eIF2α is not completely eliminated by PERK excision (compensation by GCN2), the depletion of Myc-tagged SREBP1 precursor could be clearly observed in PERK Mock, but not in PERK Cre MEFs by day 7 of treatment (Fig. 4A). Consistently, the accumulation of Oil Red O-positive droplets was significantly attenuated in the absence of PERK (Fig. 4C) and was essentially abolished in perk/Gcn2 double knockout (DKO) MEFs (data not shown). Similar results were observed after knockdown of PERK in 3T3-L1 cells (Fig. S4 A and B).

Materials.

Tissue culture media and medium supplements were purchased from Invitrogen and HyClone. All chemicals and reagents were purchased from Fisher Scientific, Sigma, and Invitrogen.

Animals and Tissue.

Experiments were conducted in accordance with the Animal Welfare Act and the Department of Health and Human Services Guide. Mice homozygous for the LoxP allele of perk (7) were mated to MMTV-CRE transgenic mice (38). The Cre transgene-bearing offspring were bred to homozygocity for the LoxP allele of perk thus generating mammary gland-specific perk“null.” Littermates not inheriting the Cre transgene were bred to homozygocity for the LoxP allele of perk and used as controls. Mammary glands were harvested from virgin mice, pregnant mice at day 8 and 16, lactating mice at day 0, 3, 7, 12, and involuting glands were harvested at day 3 postweaning. Samples were snap-frozen and stored at −80°C. For analysis of lipid composition, 10 μl of milk was extracted and triglyceride (Cayman Chemical) and free fatty acid (BioVision Research Products) content was measured after lipid extraction by using the method of Bligh and Dyer (39).

Cell Culture.

MEFs were cultured in DMEM supplemented with 4 mM l-glutamine, 10% (vol/vol) FBS, 100 units/ml penicillin/streptomycin, and 55 μM β-mercaptoethanol. Passage immortalized perk LoxP MEFs were derived as previously described (40). Perk LoxP pBabe and perk LoxP Cre MEFs were generated as described in ref. 41 and selected with puromycin. For differentiation, cells were grown to confluence and treated with differentiation mixture (10 μg/ml insulin, 0.5 mM isobutylmethylxanthine, 1 μM dexamethasone, and 5 μM troglitazone) for 7 days. On day 5, 7, and 9 of the differentiation protocol, MEFs were refed fresh differentiation medium (day 5) or 5 μM troglitazone (days 7 and 9).

Plasmids.

The N-terminally Myc-tagged human SREBP1 (BC057388; Open Biosystems) and C-terminally Myc-tagged rat Insig1 (BC078827; Open Biosystems) was cloned into the pBabe retroviral vector. All primer sequences and cloning schemes are available on request. Retroviruses were produced and used as described in ref. 28. The anti-mouse shPERK was from Open Biosystems (clone ID V2MM_51300).

RNA analysis.

RNA was collected with TRIzol (Invitrogen). Five micrograms of total RNA was used for first-strand cDNA synthesis by using SuperScript II Reverse Transcription Kit (Invitrogen). For RT-PCR, input cDNA was analyzed in triplicate. All reactions were performed by using SYBR Green (SuperArray) and 500 nM of the forward/reverse primers (ABI Prizm 7200 Sequence Detector, Applied Biosystems). The following primers were used: mFAS, mAcl, mSCD1, SREBP1c (34), and mXOR, mKeratin18 (42), m18S rRNA (F) 5′-AAATCAGTTATGGTTCCTTTGGTC-3′; (R) 5′-GCTCTAGAATTACCACAGTTATCCAA-3′. In cultured cells, mRNA was normalized to 18S rRNA and each target was determined relative to signal from untreated cells (1-fold). For tissue, target gene levels were normalized to cytokeratin18. The induction of each target expression on day 3 or day 12 of lactation was determined relative to the signal in samples of the corresponding genotype on pregnancy day 16 (1-fold).

EMSA.

An oligonucleotide containing the sterol regulatory element (SRE) from the human insulin receptor substrate 2 (IRS2) promoter (43) was used. Binding assays were performed at room temperature and the DNA–protein complexes were separated by electrophoresis on 4% TBE PAGE gel and visualized by using STORM PhosphorImager/Image Quant software (Molecular Dynamics). Where indicated, 2 μg of SREBP1 (Santa Cruz Biotechnology) antibody was included.

Oil Red O and Nile Red Staining.

Cells were fixed for 2 min in 3.7% formaldehyde, washed with water, and stained with Oil Red O. 10 μM OCT-embedded mammary gland sections were fixed in 4% paraformaldehyde, stained with Hoechst 33258, and coverslipped with Nile red.

Western Analysis.

Cultured cells were lysed in 0.15 M NaCl/0.05 mM Tris·HCl, pH 7.2/1% Triton X-100/1% sodium deoxycholate/0.1% SDS (RIPA buffer). Slices of mammary gland were homogenized in EBC buffer (28). Immunoblot was performed by using following antibodies: phospho-Ser51eIF2α, eIF4E, SCD1 (Cell Signaling); eIF2α (BioSource); PERK (Rockland Immunochemicals); FAS (BD Biosciences); ACL (44); SREBP1 (Thermo Scientific); c-Myc (9E10), XOR (H-110) (Santa Cruz Biotechnology).