SREBP-2-deficient and hypomorphic mice reveal roles for SREBP-2 in embryonic development and SREBP-1c expression[S]

Cholesterol and fatty acid biosynthesis are regulated by the sterol regulatory element-binding proteins (SREBPs), encoded by Srebf1 and Srebf2. We generated mice that were either deficient or hypomorphic for SREBP-2. SREBP-2 deficiency generally caused death during embryonic development. Analyses of Srebf2−/− embryos revealed a requirement for SREBP-2 in limb development and expression of morphogenic genes. We encountered only one viable Srebf2−/− mouse, which displayed alopecia, attenuated growth, and reduced adipose tissue stores. Hypomorphic SREBP-2 mice (expressing low levels of SREBP-2) survived development, but the female mice exhibited reduced body weight and died between 8 and 12 weeks of age. Male hypomorphic mice were viable but had reduced cholesterol stores in the liver and lower expression of SREBP target genes. Reduced SREBP-2 expression affected SREBP-1 isoforms in a tissue-specific manner. In the liver, reduced SREBP-2 expression nearly abolished Srebf1c transcripts and reduced Srebf1a mRNA levels. In contrast, adipose tissue displayed normal expression of SREBP target genes, likely due to a compensatory increase in Srebf1a expression. Our results establish that SREBP-2 is critical for survival and limb patterning during development. Reduced expression of SREBP-2 from the hypomorphic allele leads to early death in females and reduced cholesterol content in the liver, but not in adipose tissue.

(now http://www.genetrap.org) ( 21,22 ). Insertion of the genetrap vector pGT1Lxf into Srebf2 was verifi ed by direct sequencing of cDNA obtained by 5 ′ rapid amplifi cation of cDNA ends (5 ′ -RACE). Male chimeric mice from blastocyst microinjections were then bred with C57BL/6J females to generate mice carrying the mutant Srebf2 allele. The site of the gene-trap insertion within intron 1 of the Srebf2 gene was determined by inverse PCR. Offspring were genotyped by PCR of genomic DNA with primers specifi c for the WT Srebf2 allele (5 ′ -tgaaactgcggatcaggctgg-3 ′ , 5 ′ -gctggatcccctggagttac-3 ′ ; 483 bp) and for the mutant allele carrying the insertional mutation (5 ′ -gatgccagagactcagtgaag-3 ′ , 5 ′ -gctgcaaggcgattaagttgg-3 ′ ; 482 bp). To generate Srebf2 hypomorphic mice, Srebf2 +/ Ϫ mice were crossed with an Hprt -Cre transgenic mice ( 23 ). The Cre-mediated recombination event eliminated the fl oxed splice acceptor site in the gene-trapped allele; recombination was detected by a shift in the size of the PCR product generated from the mutant allele (from 482 to 136 bp). The Cre transgene was detected by PCR (primers 5 ′ -accagccagctatcaactcg-3 ′ , 5 ′ -ttacattggtccagccacc-3 ′ ; 199 bp). Mice were housed in a 12 h light/ dark cycle and fed a Purina 5001 chow diet. Samples were collected around 1:00 PM, after 5 h fasting. Mouse studies were performed under approval from the University of California at Los Angeles Institutional Animal Care and Use Committee.

Tissue staining and histology
Fresh tissues were fi xed in 4% paraformaldehyde and then processed as described ( 24 ). Tissues and embryos were stained for ␤ -galactosidase expression, as described ( 25 ) with minor modifi cations. Tissue samples were directly embedded in OCT (Sakura Finetek) and frozen. Fresh 10 m sections were fi xed with 2% formaldehyde and 0.2% glutaraldehyde in PBS for 5 min and incubated in X-gal solution at 37°C overnight. For whole embryos, samples were fi xed in 4% paraformaldehyde for 5 min, washed three times in PBS, and stained. Skeletal staining was performed with Alizarin red and Alcian blue as described (www. hhmi.ucla.edu/derobertis/).

Gene expression analyses
Total RNA was isolated from mouse tissues and MEFs by extraction with TRIzol (Invitrogen). cDNA synthesis, RT-PCR, and quantitative (q)PCR were performed as described ( 24 ). Gene expression was normalized to ␤ 2 microglobulin and 36b4. All primer sequences used in this study are presented in supplementary Table 1. Relative miR-33a expression was determined by Taq-Man RT-PCR using a predesigned probe set (Applied Biosystems) SREBP activity is regulated by the nutritional status of the cell ( 10,11 ). When cellular sterol concentrations are high, SREBP precursors are bound to SREBP cleavageactivating protein (SCAP) in the endoplasmic reticulum membrane. SCAP, in turn, interacts with insulin-induced proteins (Insigs), which serve to retain the SCAP-SREBP complex within the endoplasmic reticulum membrane . When sterol concentrations are low, SCAP-bound SREBPs are translocated to the Golgi where they undergo a twostep cleavage process (by S1P and S2P proteases), thereby releasing the N-terminal active domain (nuclear SREBP).
Studies with genetically modifi ed mice have provided insight into the physiological roles of the three SREBP isoforms [reviewed in ( 3,5 )]. The characterization of transgenic mice with expression of constitutively active nuclear forms of SREBP-1a, -1c, or -2 led to the identifi cation of preferential target genes for SREBP-1 (lipogenesis genes) and SREBP-2 (cholesterol synthetic genes) ( 9,12,13 ). However, the physiological roles of the SREBP proteins cannot be fully delineated with constitutively active transgenes that lack the capacity to be regulated by sterols. Studies with gene knockout models have largely confi rmed the gene targets for SREBP-1 and have provided additional insights into the physiological roles of this protein. Because of the critical functions of the SREBP pathway, SREBP-1 defi ciency is lethal in utero in 50-85% of mice ( 14 ). Although no analysis of SREBP-2-defi cient mice has been published, Shimano et al. ( 14 ) did comment that SREBP-2 defi ciency is lethal during embryogenesis. Also, defi ciency in S1P, which is required for activation of all SREBP isoforms, is lethal in utero ( 15 ). SCAP and S1P function has been examined with conditional gene knockouts in liver, which revealed marked (70-80%) reductions in hepatic cholesterol and fatty acid levels ( 15,16 ).
Although the SREBP-1-, S1P-, and SCAP-defi cient mice have revealed the importance of the SREBP pathway during embryogenesis, the role of each protein during development has not been studied in detail. It is intuitive that lipid homeostasis should be critical for normal development, because upregulation of cholesterol and phospholipid synthesis is required to satisfy the demand for membrane biogenesis during periods of rapid cell proliferation. Also, it is well-documented that impaired sterol synthesis results in developmental defects in humans and animals (17)(18)(19)(20). However, the activities of SREBP-1 and SREBP-2 that make them indispensable for development have not been identifi ed. In the current study, we set out to elucidate the physiological role of SREBP-2 during embryonic development and in the liver and adipose tissue of adult mice. Both of these objectives were approachable after identifying a gene-trap allele for Srebf2. That mutant allele made it possible to generate both SREBP-2-defi cient mice and hypomorphic SREBP-2 mice.

Generation of mutant mice
Mouse embryonic stem cells containing a gene-trap insertion in Srebf2 (cell line XD155) were obtained from BayGenomics To assess the effect of SREBP-2 defi ciency during development, we analyzed embryos at multiple developmental stages. Srebf2 Ϫ / Ϫ embryos died between E12.5 and E14.5 ( Fig. 1D ). At E12.5, the Srebf2 Ϫ / Ϫ embryos harbored a limb phenotype, with malformed fore-and hindlimbs and impaired digit formation ( Fig. 1D ). The blood vessels within the limbs appeared dilated, and these regions became necrotic prior to embryonic death.

Altered limb patterning gene expression in SREBP-2defi cient embryos
The malformed limbs in Srebf2 Ϫ / Ϫ embryos suggested that SREBP-2 is required for limb patterning. There are two primary signaling centers in the early limb bud: the zone of polarizing activity (ZPA) and the apical ectodermal ridge (AER) ( 30,31 ). They are the source of diffusible morphogens that control spatial patterning. Activities of the ZPA and AER are coordinated through a feedback regulatory circuit dependent on sonic hedgehog (SHH) and fi broblast growth factors (FGFs) (32)(33)(34)(35)(36). Additional morphogens, such as gremlin 1 (Grem1) and bone morphogenetic proteins (BMPs), are critical for propagating the SHH signal from the ZPA and for inducing FGF signaling in the AER ( 31 ). SHH signaling, along with the reciprocal positive interaction between the AER and ZPA, are crucial for regulating limb bud outgrowth, axial patterning, and digit number ( 32 ).
We assessed the expression of ZPA and AER patterning genes with whole-mount in situ hybridization on embryos. We selected embryos at E11.5, at which time the ZPA/AER pathways are active, but limb developmental defects in SREBP-2-defi cient embryos are not yet visible. As expected, Shh expression was present exclusively in the ZPA and Shh was expressed at higher levels in Srebf2 Ϫ / Ϫ compared with WT embryos ( Fig. 2 ). Expression of the SHH receptor, patched 1 ( Ptch1 ), was reduced in forelimbs and hindlimbs of Srebf2 Ϫ / Ϫ embryos (compared with WT embryos). During development, activation of Ptch1 by SHH leads to the release of the smoothened (Smo) receptor and induction of the Gli1 transcription factor , which activates target genes such as Bmp4 . The Srebf2 Ϫ / Ϫ embryos exhibited alterations in some, but not all, of these genes in the SHH signaling axis. Although levels of Gli1 mRNA were not substantially altered in Srebf2 Ϫ / Ϫ embryos, the Gli1 target gene, Bmp4, was expressed at reduced levels in Srebf2 Ϫ / Ϫ limb buds, but increased expression in the developing brain and spinal cord. Bmp5 expression was elevated and Fgf8 expression was unchanged in Srebf2 Ϫ / Ϫ embryos. These results indicate that lack of SREBP-2 in the developing embryo leads to abnormal regulation of some genes involved in limb bud morphogenesis.
To further explore the role of SREBP-2 and cholesterol homeostasis during embryonic development, we generated MEFs from E11.5 WT and Srebp2 Ϫ / Ϫ embryos and assessed expression of ZPA genes. Examination of marker genes for the AER signaling cascade in MEFs revealed striking effects of Srebf2 defi ciency on the expression of key morphogens ( Grem1 , Bmp2 , Bmp4 , Fgf8 ). Srebf2 Ϫ / Ϫ caused marked reductions in the expression of Grem1 , from cDNA synthesized from 100 ng total RNA (Applied Biosystems) and normalized to SnoRNA 202.

Western blot analyses
Livers were rinsed in cold PBS and homogenized in buffer A [10 mM HEPES (pH 7.4), 0.42 M NaCl, 2.5% glycerol, 1.5 mM MgCl 2 , 0.5 mM Na-EDTA, 0.5 mM EGTA]. Nonhomogenized tissue was removed with a 100 m cell strainer, and the homogenate was centrifuged at 4,000 rpm for 15 min at 4°C. The supernatant fl uid was centrifuged again at 13,000 rpm for 10 min at 4°C. The resulting pellet was resuspended in buffer B (buffer A containing 100 mM NaCl) to generate the cytosolic fraction. The original pellet was resuspended in buffer C [10 mM HEPES (pH 7.4), 0.42 M NaCl, 2.5% glycerol, 1.5 mM MgCl 2 , 0.5 mM Na-EDTA, 0.5 mM EGTA]. Twenty micrograms of these fractions were used for Western blot studies. Protein visualization was performed with rabbit anti-mouse SREBP-2 antibodies [1:5,000 ( 29 )] followed by an HRP-conjugated goat anti-rabbit IgG (1:20,000), and detected with an ECL+ kit (GE Healthcare).

Lipid analysis
For tissue lipid analysis, samples were processed as described ( 24 ). Total proteins were quantifi ed for normalization between samples (BCA kit, Pierce).

Steroid hormone quantifi cation
Plasma hormones were determined by the Endocrine Support Core Lab at Oregon National Primate Research Center.

SREBP-2 is required for viability and for limb bud development during mouse embryogenesis
To generate mouse models with absent or reduced SREBP-2, we used a gene-trap allele containing an insertional mutation in intron 1 of Srebf2 . The transcript from this allele contains exon 1 of Srebf2 joined in-frame to a ␤ -geo cassette (a fusion of a lacZ and a neomycin phosphotransferase gene) ( Fig. 1A ). The resulting SREBP-2-␤geo fusion protein contains only the fi rst 29 amino acids of SREBP-2; it lacks 97% of the SREBP-2 sequences, including those encoding the DNA-binding and transmembrane domains ( Fig. 1B ). Srebf2 +/ Ϫ mice appeared normal; Srebf2 Ϫ / Ϫ mice died in utero, aside from a single live birth (described later). This is in agreement with Shimano et al. ( 14 ), who commented that Srebf2 knockout mice die during embryogenesis (although the embryos were not further characterized).
To assess SREBP-2 expression during embryogenesis, we examined Srebf2 +/ Ϫ embryos for ␤ -galactosidase expression, which is controlled by the endogenous Srebf2 regulatory elements. High levels of Srebf2 expression were observed throughout embryogenesis ( Fig. 1C ). At embryonic day (E)8.5, the strongest staining was observed in the neural tube. At E11.5 and E13.5, staining was detectable in all parts of the embryo and was particularly notable in the dorsal root ganglia and the inter-digit region of limb buds. At E18.5, staining was evident in nearly all tissues, with skin, bones, and the gastrointestinal tract exhibiting high levels of staining. serum (to reduce exogenous cholesterol in the medium). Sterol deprivation in MEFs had similar effects as Srebf2 defi ciency: Bmp2 gene expression was induced, whereas Bmp4 and Fgf8 were downregulated ( Fig. 3 ). These results indicate that SREBP-2 and cellular cholesterol homoeostasis have a role in regulating limb development-related genes.

SREBP-2 is critical for embryonic hepatic lipid homeostasis
To assess the role of SREBP-2 on cholesterol homeostasis in the liver during embryonic development, we collected embryos at E11.5 and assessed expression of Srebf Bmp4 , and Fgf8 and a 4-fold increase in Bmp2 expression ( Fig. 3 ). The reduction of Fgf8 levels in MEFs was different from the normal levels of expression in Srebf2 Ϫ / Ϫ embryos ( Fig. 2 ), suggesting that some compensatory adjustments in gene expression may occur in vivo. Expression of another developmental gene, Hoxd13 , was not affected by Srebf2 defi ciency ( Fig. 3 ). These fi ndings implicate SREBP-2 in the regulation of AER morphogen gene expression, either directly or indirectly. We next assessed the effects of cholesterol deprivation on ZPA and AER gene expression. MEFs were treated with a combination of mevinolin (to suppress cholesterol biosynthesis), ␤ -cyclodextrin (to deplete cellular cholesterol content), and charcoal-treated

Reduced fat reserves in an adult SREBP-2-defi cient mouse
We recovered one viable Srebf2 Ϫ / Ϫ mouse from a total of ‫ف‬ 200 progeny from Srebf2 +/ Ϫ × Srebf2 +/ Ϫ intercrosses. At 2 months of age, the SREBP-2-defi cient mouse was hairless and weighed 40% less than its littermates ( Fig. 5A ). Srebf2 defi ciency was confi rmed by an absence of SREBP-2 protein in the liver ( Fig. 5B ). The size of the liver was similar to that of heterozygous littermates when normalized to body weight ( Fig. 5C ). The Srebf2 Ϫ / Ϫ mouse had dramatically reduced fat reserves, with nearly undetectable gonadal white adipose tissue (WAT) and markedly reduced inguinal subcutaneous WAT mass ( Fig. 5C ). Histological examination of tissue harvested from the perigonadal fat pad revealed smaller adipocytes with reduced lipid accumulation ( Fig. 5D , top). Sections through the skin showed strata of dermal, adipose, and and its target genes in the liver; we also extracted lipids from the remainder of the embryo. Whole body cholesterol and triglyceride levels were similar in WT and Srebf2 Ϫ / Ϫ embryos ( Fig. 4A ). As expected, Srebf2 mRNA levels in liver were nearly undetectable, and target genes involved in cholesterol synthesis [HMG-CoA synthase ( Hmgcs ), HMG-CoA reductase ( Hmgcr )], SREBP protein processing ( Insig1 ), and cellular cholesterol uptake [LDL receptor ( Ldlr )] were signifi cantly reduced ( Fig. 4B ). Also reduced by 70-80% were Srebf1a and Srebf1c transcript levels ( Fig. 4C ), consistent with a feed-forward regulatory loop between SREBP-2 and SREBP-1c ( 37,38 ). We also found reduced transcripts for SREBP1-regulated fatty acid synthesis genes [acetyl-CoA carboxylase ( Acc ), Fasn , stearoyl-CoA desaturase 1 ( Scd1 )] ( Fig. 4B ). Thus, Srebf2 -defi cient embryos displayed reduced expression of both SREBP-1c and SREBP-2 target genes.  The single adult Srebf2 Ϫ / Ϫ mouse provided an opportunity to assess effects of SREBP-2 defi ciency on liver and adipose tissue lipids and on gene expression levels. The effects of Srebf2 defi ciency in adult liver mirrored those observed in Srebf2 Ϫ / Ϫ embryos. Although statistical analysis was not possible due to availability of only a single muscle tissue in WT mice, but subdermal adipose tissue was virtually undetectable in the Srebf2 Ϫ / Ϫ mouse ( Fig. 5D , bottom). Even though limb bud abnormalities were encountered in Srebf2 Ϫ / Ϫ embryos, the skeleton of the sole surviving Srebf2 Ϫ / Ϫ mouse appeared normal ( Fig. 5E ).  were compound heterozygotes for the hypomorphic and null alleles ( Srebf2 -/ hyp ). Hepatic SREBP-2 protein levels in the Srebf2 -/ hyp mice were 80-90% lower than in WT mice ( Fig. 7B ). Srebf2 -/hyp mice appeared grossly normal during the neonatal period, but the females had reduced body weight beginning at 4 weeks of age ( Fig. 7C ), and 50% of these animals died between 8 and 12 weeks of age ( Fig.  7D ). Male Srebf2 -/ hyp mice exhibited normal body weight and survived until the study was terminated at 12 weeks of age. Compared with WT mice, male Srebf2 -/ hyp mice had 20-40% lower fat pad tissue mass and 15% lower liver mass (relative to total body weight) ( Fig. 7E ). We hypothesized that impaired steroid hormone synthesis might contribute to lethality in female Srebf2 -/ hyp mice. However, our analyses of hormone levels did not reveal a likely cause of the premature death in females. Plasma estradiol and progesterone levels were normal in females. Male Srebf2 -/ hyp mice actually had elevated levels of aldosterone and corticosterone in the plasma ( Table 1 ).

Effects of reduced Srebf2 gene expression in liver and adipose tissue
We studied adult Srebf2 -/ hyp mice to assess effects of reduced SREBP-2 expression on cholesterol homeostasis. At 12 weeks of age, Srebf2 -/ hyp males had reduced cholesterol levels in the liver, but not in adipose tissue; tissue triacylglycerol levels were normal ( Fig. 8A, C ). Reduced SREBP-2 expression also led to lower expression of cholesterol biosynthetic and lipogenic genes in the liver ( Fig. 8B ), but not in adipose tissue ( Fig. 8D ). Srebf2 -/ hyp mice had reduced Srebf2 Ϫ / Ϫ mouse, hepatic cholesterol levels in the Srebf2 Ϫ / Ϫ mouse were reduced by 40% and triacylglycerol levels were reduced by 60% ( Fig. 6A ). Transcript levels for Srebf1a and Srebf1c in the liver were reduced, as were the transcripts for most SREBP target genes analyzed ( Fig.  6B ). In inguinal WAT, cholesterol levels were reduced by approximately one-half compared with heterozygous littermates, but triacylglycerol levels (normalized to protein) were similar ( Fig. 6C ). In contrast to the liver of the Srebf2 Ϫ / Ϫ mouse, most SREBP target genes in WAT were expressed at levels similar to those in WT mice ( Fig. 6D ). Interestingly, Srebf1a mRNA levels in WAT were elevated 5-fold, whereas Srebf1c mRNA levels were normal. In addition, the expression levels for Acc , Fasn , and Scd1 were equal to (or exceeded) those in Srebf2 +/ Ϫ mice. Thus, in the adipose tissue of the Srebf2 Ϫ / Ϫ mouse, higher levels of Srebf1a expression may have partially compensated for the loss of SREBP-2.

Srebf2 hypomorphic mice exhibit reduced adiposity and sex-specifi c lethality as young adults
While we were able to study a single viable Srebf2 Ϫ / Ϫ mouse, embryonic lethality prevented us from thoroughly investigating SREBP-2 function in adult mice. To circumvent this issue, we used Cre recombinase to remove the splice acceptor in the gene-trap allele, with the goal of restoring limited amounts of normal mRNA splicing and creating mice with a hypomorphic Srebf2 allele ( Srebf2 hyp ). Intercrosses of Srebf2 +/ Ϫ and Srebf2 +/ hyp mice generated 127 live offspring, 19% of which (compared with 25% expected)  ( Fig. 9A ), mirroring the gene expression fi ndings in the liver of Srebf2 -/ hyp mice ( Fig. 8B ). SREBP-2 defi ciency in MEFs led to nearly undetectable Srebf1c mRNA levels, but to increased levels of Srebf1a transcripts ( Fig. 9B ). SREBP-2 defi ciency also Srebf1 gene expression, with 90% lower Srebf1c mRNA levels in liver and 70% lower Srebf1c levels in adipose tissue. Srebf1a levels were modestly altered in both tissues ( Fig.  8B, D ). The adipose tissue of Srebf2 -/ hyp mice displayed a 6-fold increase in transcripts for the LDL receptor ( Fig. 8D ). Given that the gene for miR-33a resides within an intron of Srebf2 , we assessed whether the Srebf2 gene-trap allele leads to altered miR33a mRNA expression. miR-33a RNA levels and mRNA levels for several miR-33a target genes were similar in livers of Srebf2 -/hyp and WT mice ( Fig. 8E, F ). Glucose-6-phosphatase ( G6pc ) mRNA levels were decreased in the hypomorphic mice, but because miR-33a and other target genes tested were expressed at WT levels, this is unlikely a result of altered miR-33a regulation.   ( 7 ). However, the activities of the two transcription factors overlap, making it challenging to decipher their unique physiological functions. Previously published studies characterized effects of SREBP-1 defi ciency in the mouse ( 14 ), but the impact of SREBP-2 defi ciency was not characterized. Here, we used a combination of two mutant Srebf2 alleles to investigate the roles of SREBP-2 in mouse development and in adult mice. We identifi ed a crucial role for SREBP-2 in survival and in limb patterning through effects on the SHH pathway. We also demonstrated a key role for SREBP-2 in maintaining hepatic cholesterol content and in regulating Srebf1 gene expression, in addition to regulating well-established SREBP-2 target genes.
impaired the regulation of gene expression in response to sterol deprivation, with a failure to induce SREBP target genes, with the exception of Srebf1a ( Fig. 9 , the two bars at the right of each panel). Thus, SREBP-2 is required for basal expression of Srebf1c , but not Srebf1a , and also is essential for sterol regulation of many SREBP target genes.

DISCUSSION
It has long been appreciated that SREBP-2 is a crucial transcriptional regulator of lipid homeostasis. SREBP-1 and SREBP-2 preferentially target genes involved in fatty -/ hyp mice. mRNA levels in liver (B) and iWAT (D) of Srebf2 -/ hyp mice. Hepatic levels of miR-33a (E) and mRNA for several target genes (F). All WT values are set at 1.0. Data represent mean ± SD (n = 6 per group). * P < 0.05 versus WT. Nsf , N-ethylmaleimide-sensitive factor; Prkaa1 , AMP-activated protein kinase ␣ 1 catalytic subunit; Cpt1a , carnitine palmitoyltransferase 1a; Hadhb , hydroxyacyl-CoA dehydrogenase/3-ketoacyl-CoA thiolase/enoyl-CoA hydratase ␤ subunit; G6pc , glucose-6-phosphatase; Pck1 , phosphoenolpyruvate carboxykinase-1. reduced expression of downstream components of the signaling pathway ( Ptch1 , Bmp4 ). Given that both SHH and Ptch1 both contain sterol-sensing domains ( 41 ), it is interesting to speculate that SREBP-2 regulation of cholesterol levels in the developing embryo may play a role in the proper functioning of this pathway, but additional work is required.
Our results in the Srebf2 -/ hyp mice demonstrate that low levels of SREBP-2 expression are suffi cient for survival during embryogenesis. Srebf2 -/ hyp mice are born at a near-normal frequency; the viability of males appeared normal, but the females exhibited failure-to-thrive and most succumbed during the fi rst few months of life. An analysis of steroid hormones was not helpful in explaining the differences in the survival of males and females. Studies of adult male Srebf2 -/ hyp mice revealed a complex relationship between Srebf2 expression and the regulation of Srebf1 . Previous work established that loss of SREBP-1 leads to increased hepatic SREBP-2 transcript and protein levels, increased expression of cholesterol biosynthetic genes, and a 3-fold increase in hepatic cholesterol content ( 14 ). We found that reduced SREBP-2 expression in Srebf2 -/ hyp mice resulted in tissue-specifi c alterations in Srebf1 expression. In the liver, the Srebf2 -/ hyp mice had nearly undetectable levels of Srebf1c transcripts and an 80% reduction in Srebf1a transcripts. The low levels of SREBPs in these mice were accompanied by SREBP-2 defi ciency is incompatible with normal embryonic development; most Srebf2 Ϫ / Ϫ embryos died by 14.5 days post coitus . This fi nding likely explains why no one has encountered or studied a human subject with complete SREBP-2 defi ciency. In the mouse, a spontaneous Srebf2 missense mutation was associated with cataracts and persistent skin wounds ( 39 ), but the effect of the amino acid substitution on SREBP-2 activity is unknown, but unlikely to abolish activity. The cause of death in Srebf2 Ϫ / Ϫ embryos is unclear, given that Srebf2 expression is expressed widely and multiple organ systems could have been compromised by the absence of SREBP-2.
A striking phenotype of Srebf2 Ϫ / Ϫ embryos was abnormal digit formation. At 13.5 days post coitus, Srebf2 expression is detectable in the limb buds of WT embryos in the regions destined to undergo apoptosis during digit specification. The absence of digit formation in Srebf2 Ϫ / Ϫ embryos confi rmed a requirement for SREBP-2 in this process. This fi nding is in accord with previous work showing that in the developing limb bud, there is spatial concordance between cholesterol biosynthetic gene activity and the apoptotic process that defi nes the digits ( 40 ). Gene expression analyses in Srebf2 Ϫ / Ϫ embryos revealed alterations in pathways that regulate limb morphogenesis, including components of the SHH signaling axis. Limb buds of Srebf2 Ϫ / Ϫ embryos exhibited increased Shh transcript level, but RNA was isolated after an overnight incubation and mRNA levels were determined by qPCR. Data represent mean ± SD (n = 6 per group). * P < 0.05; *** P < 0.001 by two-way ANOVA .
40% lower stores of cholesterol in the liver. In contrast, the cholesterol content of adipose tissue and embryonic fi broblasts was maintained despite a 70% reduction in Srebf1c expression, perhaps because of slightly increased Srebf1a expression levels. Exon 16 of Srebf2 encodes a microRNA, miR-33a, which appears to be generated as a result of mRNA splicing following transcription of the Srebf2 gene (42)(43)(44)(45). miR-33a targets transcripts from multiple genes involved in cellular cholesterol effl ux, such as Abca1 and Abcg 1, and miR-33a inhibition has been proposed as an anti-atherosclerosis treatment ( 46,47 ). It is thought that the coordinate expression of Srebf2 and miR-33a represents a mechanism to maintain cellular cholesterol homeostasis by enhancing cholesterol synthesis while limiting sterol effl ux. Because miR-33a is present within the Srebf2 locus, it was possible that miR-33a levels would be reduced in our Srebf2 -deficient mice. However, the expression of miR-33a and several target genes were not perturbed in the liver, suggesting that the Srebf2 mutant allele did not disrupt miR-33a expression. Furthermore, the phenotypes of our mice did not recapitulate those of miR-33a-defi cient mice, which manifest hepatic steatosis and are prone to increased body weight on chow and high-fat diets ( 48 ). It is possible that miR-33a production is unaffected by our gene-trapped alleles or that there are compensatory changes in our mice that serve to maintain transcript levels for miR-33a target genes.
Finally, these studies provide a proof-of-principle for generation of hypomorphic alleles from gene-trap insertions. Gene trapping has been a valuable adjunct to standard homologous recombination approaches for deciphering gene function in mice ( 21,22 ). Regardless of which technique is used, it is often the case that knocking out a gene in mice results in lethality during embryonic development. In those cases, most investigators rely on the production of tissue-specifi c knockout mice to assess gene function. In the current study, we showed that the creation of a hypomorphic allele from the gene-trap insertion represents a useful approach for deciphering gene function. We used Cre recombinase to remove the slice acceptor site in the gene-trap allele, which restored a low level of Srebf2 transcripts and led to markedly reduced SREBP-2 levels throughout the body. Of note, this approach is not limited to creating a "global hypomorphic mouse;" it would have been possible to use a cell type-specifi c Cre transgene to create mice that express low amounts of SREBP-2 in a tissue.