If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
* This minireview will be reprinted in the 2000 Minireview Compendium, which will be available in December, 2000. This is the first article of three in the “Nutrient Control of Gene Transcription Minireview Series.” This work was supported by grants from INSERM, CNRS, and “Fondation pour la Recherche Médicale.” ‡ Recipients of a fellowship from the French “Ministère de l'Education Nationale et de la Recherche.” These two authors contributed equally to this work. § Recipient of the 2000 Award from the “Institut Danone.” ¶ Recipient of the 1999 Award from the “Association Française de Nutrition-Société de Nutrition de Langue Française.”
Fatty acids (FAs)1 are energy-rich molecules, which play important metabolic roles. They are also an integral part of cells as membrane components, which can influence fluidity and receptor or channel function. Over the past 10 years, it has become evident that FAs can also act as signaling molecules involved in regulating gene expression. For the most part, these target genes encode proteins with roles in FA transport or metabolism. The corresponding change in the amount of specific proteins is an adaptative process that the cells develop in response to variations in FA concentration in the vicinity of the target tissue. Although interesting progress has been made recently, the mechanism(s) by which FAs modulate gene transcription still remains largely unresolved. The purpose of the present review is to address this important issue.
Fatty Acid Structure and Metabolism
A comprehensive description of FA regulation of gene expression requires the understanding that 1) FA molecules have a common basic structure with specific diversity determined by chain length and degree of unsaturation and 2) FAs are rapidly metabolized.
Long chain FAs (C16 and above) (LCFAs) can be either saturated or mono- or polyunsaturated (PUFA) depending upon the presence of one or more double bonds in the polycarbon chain. The most abundant monounsaturated FA is oleate, in which the chain has 18 carbons and the double bond is between C9 and C10 from the methyl end (C18:1n-9). The two major classes of PUFAs are n-3 (or ω3) and n-6 (or ω6), named for the carbon involved in the first double bond. The precursors of these two classes of FAs cannot be synthesized in humans and must be provided by the diet. Interestingly, n-3 and n-6 PUFAs may have biologically opposite properties, probably because they give rise to different eicosanoid products. For instance, when model animals with a propensity to develop tumors are fed diets containing a large proportion of n-6 PUFAs, tumor formation is favored, whereas diets with a similar proportion ofn-3 PUFAs are somewhat protective (see Ref.
Long chain FAs are insoluble in water and are carried in plasma either esterified in triacylglycerols arranged in complex structures, the lipoproteins, or in a non-esterified form (NEFAs) loosely bound to albumin (Fig. 1). Blood lipoproteins are synthesized from dietary lipids after absorption and re-esterification in the intestine (chylomicrons) or the liver (very low density lipoproteins). Plasma lipoproteins are hydrolyzed by hepatic lipase or by lipoprotein lipase to produce NEFAs that are locally taken up by the liver or by muscles and adipose tissue. In contrast, circulating NEFAs are produced almost exclusively by white adipose tissue (WAT) as a consequence of lipolysis from the stored triacylglycerols during periods of starvation. In lipogenic tissues like liver and WAT, FAs can also be synthetized de novo from glucose and esterified.
Although butyrate, a short chain (four carbon, C4) FA, has been shown in some cases to exert regulation of specific genes besides its well known effect on chromatin, most of the reported actions of FAs on gene expression have been associated with LCFAs. Clearly in the cell, the signaling molecule is the free FA (not bound to albumin), which is transported in and out cells with the help of a membrane protein, the FA transporter (FAT). Six potential FAT candidates, the FA translocase (FAT-CD36), the FA transport protein, the mitochondrial aspartate aminotransferase, caveolin, the adipose differentiation-related protein, and the FA-binding protein (FABP, a cytosolic protein that can bind to membranes) have been cloned and characterized. In the cytoplasm, FAs are taken up by a cell-specific FABP and have alternate destinies (Fig. 1). They can be elongated, desaturated, β-oxidized in mitochondria or peroxisomes for energy production, submitted to a peroxidative process or to ω-oxidation in microsomes, exchanged with membrane phospholipids, and participate in or interfere with eicosanoid synthesis. It is, therefore, important to remember that whatever FA effect is investigated, not only FAs per se but also products of FA metabolism or FA-sensitive signal transduction cascade can act as a relay.
Overview of Fatty Acid Regulation of Gene Transcription in Various Organisms
FA regulation of gene transcription occurs in unicellular and complex organisms. In Escherichia coli, LCFAs are transported and activated as fatty acyl-CoAs catalyzed by FadD, the bacterial acyl-CoA synthetase (ACS). In this organism, CoA derivatives, not FAs, bind with high affinity to a transcription factor, the FaDR, preventing its binding to a response element, thereby allowing the gene involved in FA synthesis (fabA) to be repressed and genes encoding enzymes of FA transport and metabolism (fadL, fadD, fadE, fabBA, fadH) to be induced (reviewed in Ref.
). Studies usingSaccharomyces cerevisiae focused on regulation of the OLE1 gene, which encodes the Δ-9 desaturase, a membrane protein that converts saturated palmitoyl (C16:0) and stearoyl (C18:0) CoA to their monounsaturated counterparts. As expected from the biological role of the desaturase, the rate of OLE1 gene transcription is increased in response to exogenous saturated FAs, whereas exposure to unsaturated FAs sharply reduces transcription (
); however, the mechanisms of activation and of repression remain to be elucidated.
In mammals, the expression of many genes has been shown to be modulated by FAs in a positive or a negative manner. However, in only a few cases has the transcription rate been clearly demonstrated as the main site of control. Post-transcriptional regulation can also occur, as exemplified by the PUFA-induced decrease in stability of stearoyl-CoA desaturase (SCD1) and glucose transporter 4 (GLUT4) mRNAs in the adipocytes of the 3T3-L1 cell line (reviewed in Ref.
). This review focuses exclusively on the transcriptional aspects of FA control. Studies reporting mRNA variations in response to FAs without the demonstration, by run-on or transfection experiments in differentiated cells, that the rate of transcription is affected are mentioned only when pertinent.
The control of hepatic lipogenic enzymes is currently the best example of negative regulation. Thirty years ago, Allmann and Gibson (
) first observed that feeding mice with linoleate (C18:2n-6) greatly depressed hepatic lipogenesis and activities of fatty acid synthase (FAS), malic enzyme, and glucose-6-phosphate dehydrogenase, as would be expected from the physiology. Surprisingly, however, this effect appeared to be restricted to PUFAs, because neither palmitate (C16:0) nor oleate (C18:1n-9) was effective. Following this original observation, several studies showed that a PUFA-rich diet reduced the hepatic mRNA levels for FAS, acetyl-CoA carboxylase, L-PK, ATP citrate-lyase, malic enzyme, SCD1, apolipoprotein A-1 (apoA-1), the S14 protein (reviewed in Refs.
) showed that the reduction in hepatic FAS and S14 mRNAs following the administration to rats of a PUFA-rich diet was caused primarily by the inhibition of gene transcription. In that case, n-3 and n-6 PUFAs were both effective. This effect was specific because transcription of the cytosolic phosphoenolpyruvate carboxykinase (PEPCK) (a gluconeogenic enzyme) and actin genes was not affected by the treatment. Primary hepatocytes were then used to ascertain the direct nature of PUFA action (
). Moreover, transfection of hepatocytes with a chimeric gene containing −4315 to +19 base pairs relative to the transcriptional start site of the S14 gene linked to the chloramphenicol acetyltransferase structural gene demonstrated that the 5′-flanking region of the S14 gene was involved (
). Because the effect is restricted to PUFAs and because PUFAs are very sensitive to peroxidation, it is possible that cytotoxic peroxidative products could be the active molecules, a still debated issue (
). This would be consistent with the fact that the effect of PUFA on transcription of genes coding for lipogenic enzymes is restricted to hepatocytes in which peroxidation occurs. Tissue-specific regulation by PUFA seems to occur in rat retroperitoneal, but not subcutaneous, WAT (
) that FAs are inducers of the adipocyte lipid-binding protein (aP2) gene transcription in pre-adipose cells, evidence has accumulated demonstrating that FAs are potent regulators of the adipose differentiation process (adipogenesis). From a physiological viewpoint, it makes sense that FAs induce the expansion of WAT mass after an excess of food is ingested. Here, in contrast to the situation described above, saturated and unsaturated long chain FAs (C16 and longer) are equipotent. Therefore, it seems likely that the mechanism differs from that involved in the negative regulation of the liver lipogenic enzymes. The non-metabolizable FA α-bromopalmitate is also a strong inducer, suggesting that the metabolism of FAs is not necessary (
). This result strongly argues against a direct action of FA on theaP2 gene but is consistent with the possibility that FAs act by altering the synthesis of either a transcription factor or of some other protein with a rapid turnover that is required for the effect (Fig. 2). Indeed, several other marker mRNAs of the adipocyte phenotype (ACS, FAT, lipoprotein lipase, GLUT4, the uncoupling protein 2, etc.) are clearly induced by long term treatment of preadipocytes with FAs.
The data discussed above support the idea that FAs stimulate the general adipogenic process but do not give clues to the mechanism. It is possible that a common response element(s) to a FA-responsive master transcription factor is present in the promoter regions of all these genes. One way to decipher the mechanism would be to study in detail the genes when transcription is stimulated by FAs in differentiated adipocytes at a stage when adipogenesis is not occurring. TheaP2 and the PEPCK genes meet this criterion (
). PEPCK activation also allows esterification of FAs in adipocytes after consumption of a glucose-free, lipid-rich diet. Therefore, a stimulatory action of FAs on PEPCK is physiologically relevant; in agreement with the physiology, glucose inhibits the FA stimulation (
) or on carnitine palmitoyltransferase 1 (CPT-1) and liver FABP (L-FABP) in hepatocytes (see below). The effect of FA is rapid, does not require protein synthesis, and acts directly on PEPCK gene transcription. Induction of PEPCK activity follows (
). This difference in behavior of the same gene in two tissues may be related to the difference in the function of PEPCK in WAT (glycerogenesis) from that in liver where it is glucogenic.
The expression of a number of genes in the liver is under positive FA control in a physiologically relevant manner. These include the acyl-CoA oxidase (AOX), CPT-1, the L-FABP, cytochrome P4504A1, ACS, 3-hydroxy-3-methylglutaryl-CoA synthase, and cholesterol 7α-hydroxylase. To date, the effect of FA on transcription has been determined for the AOX (
). In both cases, saturated and unsaturated LCFAs have equivalent potencies.
Is PPAR the FA-responsive Transcription Factor?
The main postulated mechanisms of FA regulation of gene transcription are summarized in Fig. 2. FAs, FA-CoAs, or FA metabolites can: 1) induce a cascade of events leading to a covalent modification of a transcription factor (TF), for instance phosphorylation, thereby altering its transactivation capacity, 2) directly bind to and activate a TF, 3) modify the mRNA stability or 4) influence the transcription rate of a TF, hence 5) changing its de novo synthesis. The FA-responsive TF binds to a recognition sequence, the FA-response element in the promoter-regulatory region of the target gene, either 6) as a monomer or 7) as a homo- or a heterodimer.
Following the cloning in 1990 by Issemann and Green (
) of a new member of the steroid/thyroid receptor superfamily, the so-called peroxisome proliferator-activated receptor (PPAR), the hypothesis that it might be the FA-activated receptor arose. This receptor is activated by the hypolipidemic agents fibrates, peroxisome proliferators (PP), and xenobiotics in a transactivation assay in which recipient cells (usually undifferentiated highly proliferative cells selected for their high transfection efficiency) are co-transfected with a PPAR expression vector and a reporter gene placed under the control of a transcription unit that contains a response element recognized and activated by the receptor. Using such an assay, Gustafsson's laboratory (
) first showed that FAs are indeed PPAR activators.
The situation became more complex with the discovery of other members of the PPAR family. At present, three isoforms have been cloned (PPARα, -δ (= β = NUC1 = FAAR), and -γ) with tissue-specific expression, ligand-specific activation, and the ability to heterodimerize with retinoid X receptors (RXR), of which three isoforms (α, β, γ) have also been isolated (reviewed in Ref.
) is expressed selectively in adipocytes. Retrovirus-mediated ectopic expression of PPARγ2 in non-adipose NIH-3T3 cells directed adipocyte differentiation when PPAR activators, like anti-diabetic thiazolidinediones or polyunsaturated FAs, were provided (
). This observation suggested that PPARγ2 was the master transcription factor of the adipocyte lineage. However, PPARγ2 is not expressed in pre-adipose cells, and α-bromopalmitate, an inducer of differentiation, is an inefficient transactivator of PPARγ. Hence, the hypothesis that FAAR (PPARδ) could play such a role has been proposed and is still a matter of debate (
). Using various methods to estimate binding, some corroborative specificity was shown among FAs, PUFAs being much better ligands than saturated FAs except for PPARα, to which palmitate (C16:0) bound with good affinity. Of particular interest also was the observation that specific prostaglandins and leukotrienes are PPAR activators (reviewed in Ref.
). The 15-deoxy-Δ12,14-prostaglandin J2 (15d-PGJ2), an arachidonic acid (C20:4n-6) metabolite of the PGD2-J2 series, is a very potent and specific PPARγ ligand. The lipoxygenation product of arachidonate, (8S)-hydroxyeicosatetraenoic acid, is the most potent PPARα activator in a transactivation assay. The leukotriene B4 is a PPARα-specific activating ligand. Carbaprostacyclin, a stable analogue of prostacyclin, activates PPARα and PPARδ. Therefore, all these molecules could be second messengers for FA transcriptional action. This would imply, however, that they are produced by the cells in which FAs are directly acting, a debated issue.
, PP response elements (PPREs) have been identified in a number of genes. They consist of imperfect versions of a direct repeat of the AGGTCA consensus sequence separated by one nucleotide (DR1) to which the additional 5′-extended portion AACT appears to be important for polarity and selectivity of recognition (reviewed in Ref.
). Such an element is able to bind PPAR/RXR heterodimers in a gel retardation assay. A degenerate AGGTCA is also the half-site recognition sequence for the 9-cis- and all-trans-retinoic acids, thyroid hormone, vitamin D subclass of nuclear receptors and orphan receptors like hepatic nuclear factor 4 (HNF4), and chicken ovalbumin upstream promoter transcription factor (reviewed in Ref.
). A number of genes shown to be FA-responsive respond also to PPAR activators and present a PPRE in their promoter-regulatory regions. Hence, it has been assumed that FAs regulate gene transcription via a PPAR/PPRE-mediated process (reviewed in Ref.
). This is the case for instance with AOX, ACS, apoA-1, aP2, CPT-1, L-FABP, SCD1, S14, and PEPCK. This non-exhaustive list can be found in positively and negatively FA-regulated genes. However, recent evidence suggests that alteration of gene transcription by FAs and PPAR activators is often disconnected. Some genes that contain a PPRE do not respond to FAs, as is the case with, for instance, apoA-II in primary hepatocytes (
). It is responsive to FAs in adipocytes but not in hepatocytes or hepatoma cells (see above), which, however, contain large amounts of PPARα. In the S14 gene, thecis-regulatory negative element for the PPARα-specific ligand wy14643 is clearly distinct from that for PUFAs (
These conclusions were reinforced with the analysis of PPARα-deficient mice. In such animals, the administration of fish oilin vivo or the addition of PUFAs to cultured hepatocytes repressed S14 and FAS gene expression although induction of AOX and CYP4A2 mRNAs was abolished (
). Hence, some of the FA effects, but certainly not all, can be assigned to PPAR activation.
In Search of the FA-responsive Transcription Factors: Other Candidates
Several lines of evidence indicate that transcription factors different from the PPARs could mediate FA effect. As mentioned above, other proteins recognize DR1-like sequences. Among these, the orphan hepatic receptor HNF4 was shown to bind FA-CoA (
). In transient transfection in recipient cells, oleate and PUFAs inhibited HNF4 transactivation potential of the human apolipoprotein CIII gene promoter governing chloramphenicol acetyltransferase expression. Among saturated FAs, medium and short chain had no effect, palmitate was stimulatory, and stearate (C18:0) was inhibitory. Whether this observation has any physiological relevance remains an open question. Last but not least, in chick hepatocytes, it was shown that medium chain FAs can alter triiodothyronine or estrogen action without interfering with the binding of the respective receptors to their cognate DNA elements (
Another route that FA could follow to regulate gene transcription would be to alter the de novo synthesis of a TF as shown in step 5 of Fig. 2. There are three recent examples of this possibility. One is related to the PUFA inhibition of lipogenic genes. Because of similarities between the negative feedback control by FA and cholesterol of their respective biosynthetic pathways, Worgall et al. (
) analyzed the effects of PUFA on genes having sterol regulatory elements (SRE) in their promoter-regulatory regions. In transfected cells they found that oleate and PUFAs reduced expression of SRE-containing genes by decreasing the level of SRE-binding protein (SREBP). Saturated FAs were ineffective, a situation reminiscent of what was shown for the PEPCK gene regulation in adipocytes. It was later reported that PUFAs decrease SREBP1 gene expression and nuclear content of SREBP1 in liver and in primary hepatocytes but not in 3T3-L1 adipocytes (
). Because SREBP1 is a positive effector of the FAS and S14 lipogenic genes, the reduction in quantity of SREBP1 accounts for the FA inhibition. This observation is corroborated by results of experiments with transgenic mice either overexpressing SREBP1 (
). Interestingly, the action of PUFAs on SREBP1 is post-transcriptional as exemplified by step 3 in Fig. 2, but the mechanism has not been alleviated yet. Another report describes the induction of the immediate-early response genes c-fos andnur-77 in the pancreatic β-cell line INS-1 by palmitate and oleate but not by PUFAs (
). Treatment of rat hepatoma cells or primary hepatocytes with unsaturated FAs induces an increase in LXRα protein, mRNA, and gene transcription. Because LXRα is a nuclear receptor thought to be an important regulator of cholesterol, steroid hormone, and bile acid catabolic pathways, the observation that it is FA-responsive suggests that it plays a role in the cross-talk between FA and cholesterol regulation of lipid metabolism. In the case of SREBP, c-fos, nur-77, and LXRα, FAs modulate the amount of transcription factors potentially acting as regulators of the expression of genes, which code for proteins having a metabolic role.
It seems clear that depending upon the specific cell context and target gene, FAs can use very different routes to alter transcription. Although the details of exact mechanisms are still needed, rapid progress have been made recently, particularly with the help of mutant mice in which specific transcription factors or nuclear receptors were either overexpressed or deleted by knockout. In vivo studies using such animal models and analysis of specific cell types derived from these mice would permit the identification of the targeted protein as a FA-responsive regulator. One reason that the exact mechanism has been so difficult to decipher may be related to the complexity of chromatin structure that adds to the task of integrating selective signals in the rapidly growing network of recently discovered coregulators acting on histone acetylation and altering the transcriptional activity of specific nuclear proteins. Because it is now recognized that plasma FA concentration and identity have profound effects on metabolism, linking nutrition to obesity, diabetes, cardiovascular diseases, and cancer, it is of great value to actively elucidate the mechanisms by which these molecules alter gene transcription.