Most mutant mice lacking both SREBP-1 isoforms die during embryonic development, and the small percentage that survive express elevated levels of SREBP-2 (17
). Mice lacking only the 1c isoform are completely viable (11
), which leads us to suspect that SREBP-1a might be absolutely required for development. However, the total viability of mice with a deficiency of SREBP-1a in our study showed that this was not the case. There are two potential explanations for this finding. The first and most likely is that either one of the SREBP-1 isoforms is sufficient for normal embryonic development. The second possibility is that trace amounts of SREBP-1a are expressed in critical cells during development and that the “leaky expression” from our gene-trap allele is sufficient for normal embryogenesis. Although we cannot completely discount the latter possibility, it seems unlikely to us, given that levels of SREBP-1a mRNA expression in the 1aDf mice were extremely low in each of 13 different adult mouse tissues tested (see Table S1 in the supplemental material).
SREBPs are well known to regulate the core pathways for the synthesis of cholesterol and fatty acids, but the current observations have uncovered a special role for the 1a isoform in balancing the choice between fatty acid oxidation and triglyceride synthesis in the liver. Because the two SREBP-1 isoforms have identical DNA binding and dimerization domains, they are predicted to bind DNA identically. However, their activation domains are distinct, and prior studies show they activate target genes with different potencies (20
). SREBP-1c is by far the more abundant isoform in liver, so any functions that are equally performed by both isoforms would not be affected by the loss of SREBP-1a. Indeed, the fact that we identified only a few differentially expressed genes in wild-type versus 1aDf mice indicates that most functions of SREBP-1 in the liver can be carried out by SREBP-1c. A key finding of the current study is that Acacb
expression was significantly perturbed in the 1aDf mice, indicating that SREBP-1c is not sufficient for controlling its expression.
Genes such as Acacb
and their promoters, which are sensitive to the loss of SREBP-1a, must have a combination and organization of transcription factor binding sites that depend on the stronger activation domain of SREBP-1a to recruit the required coactivator proteins for efficient gene expression. In support of this concept, our previous work revealed that SREBPs are weak transcription factors by themselves and that without additional coregulatory binding sites and the corresponding activator proteins, SREBPs are ineffective in activating gene expression (15
). We also showed that SREBPs synergize with different transcription factors through activation domains that are promoter specific (2
) and in our studies of Fasn
expression, we showed that SREBP-1a functions efficiently with just NF-Y as a coregulator, whereas SREBBP-1c requires both NF-Y and Sp1 for high-level activation (12
). These previous studies were based largely on cotransfection studies with plasmid-borne promoter templates, and the present results provide direct in vivo evidence to support the molecular mechanisms suggested by these earlier transfection studies.
To address the mechanism by which Acacb gene expression might be uniquely dependent on SREBP-1a in vivo, we showed that efficient recruitment of the p300 coactivator protein was reduced at the Acacb promoter in the 1aDf mice despite wild-type levels of total SREBP-1 binding (Fig. ). The reduced recruitment of p300 by the SREBP-1c-bound promoter complex is sufficient to account for the decreased Acacb expression observed in 1aDf mice.
ACC2 is highly expressed in skeletal muscle and heart, two tissues with a high rate of fatty acid oxidation but a low rate of fatty acid synthesis (22
). Hepatic expression of ACC2 presents an interesting metabolic challenge because the liver has a high capacity for both fatty acid synthesis and oxidation and must regulate these opposing activities according to physiologic demand (Fig. ) and to prevent a futile metabolic cycle. ACC2 is critical for this process, particularly during fasting. We show that reduced hepatic ACC2 activity from a deficiency in SREBP-1a leads to an unbalanced partitioning of fatty acyl-CoAs between the mitochondrion and the cytosol. The reduction in malonyl-CoA levels close to the mitochondrial membrane increases the activity of the CPT-1 shuttle system. This leads to uncontrolled transport of fatty acyl-CoAs into the mitochondrion, where they are oxidized and converted to excess ketone bodies at the expense of storing the surplus acyl-CoAs in the form of cytosolic triglycerides. It is noteworthy that the phenotype of SREBP-1a deficiency is quite similar to the hepatic phenotype displayed by mice with a knockout of Acacb
Model for ACC1 and 2 in regulation of fatty acid partitioning in liver.
A potential physiological explanation for why SREBP-1c cannot functionally substitute for SREBP-1a in the activation of ACC2 is suggested by the fact that Acacb is not a simple lipogenic target gene required for fatty acid biosynthesis. If SREBP-1c expression levels controlled ACC2, then ACC2 levels would fall precipitously during fasting, along with other SREBP-1c-controlled lipogenic genes. However, ACC2 levels need to be maintained during fasting at a sufficiently high level to regulate CPT-1 activity for fatty acyl-CoA partitioning as fatty acids are mobilized from adipose tissue and delivered to the liver. It makes sense that ACC2 would be controlled by SREBP-1a, because that isoform, in contrast to SREBP-1c, does not fall significantly during fasting.
The metabolic decision of whether to store or oxidize acyl-CoAs is fundamental to the development of nonalcoholic fatty liver disease, which is often associated with metabolic syndrome and hepatic insulin resistance (5
). By establishing a role for SREBP-1a in this process, our studies suggest a novel avenue for further investigation and therapeutic intervention in nonalcoholic fatty liver disease.