Hepatic steatosis: a role for de novo lipogenesis and the transcription factor SREBP-1c


F. Foufelle, INSERM UMR-S 872, Centre de Recherche des Cordeliers, 15 rue de l’école de médecine, 75006 Paris, France.
E-mail: fabienne.foufelle@crc.jussieu.fr


Steatosis is an accumulation of triglycerides in the liver. Although an excessive availability of plasma fatty acids is an important determinant of steatosis, lipid synthesis from glucose (lipogenesis) is now also considered as an important contributing factor. Lipogenesis is an insulin- and glucose-dependent process that is under the control of specific transcription factors, sterol regulatory element binding protein 1c (SREBP-1c), activated by insulin and carbohydrate response element binding protein (ChREBP) activated by glucose. Insulin induces the maturation of SREBP-1c by a proteolytic mechanism initiated in the endoplasmic reticulum (ER). SREBP-1c in turn activates glycolytic gene expression, allowing glucose metabolism, and lipogenic genes in conjunction with ChREBP. Lipogenesis activation in the liver of obese markedly insulin-resistant steatotic rodents is then paradoxical. Recent data suggest that the activation of SREBP-1c and thus of lipogenesis is secondary in the steatotic liver to an ER stress. The ER stress activates the cleavage of SREBP-1c independent of insulin, thus explaining the paradoxical stimulation of lipogenesis in an insulin-resistant liver. Inhibition of the ER stress in obese rodents decreases SREBP-1c activation and lipogenesis and improves markedly hepatic steatosis and insulin sensitivity. ER is thus a new partner in steatosis and metabolic syndrome which is worth considering as a potential therapeutic target.


Non-alcoholic fatty liver disease (NAFLD) is currently the most common form of chronic liver disease and is strongly associated with obesity, type 2 diabetes and insulin resistance [1,2]. As such, NAFLD is now considered as the hepatic manifestation of the metabolic syndrome. NAFLD encompasses a wide spectrum ranging from simple triglyceride (TG) accumulation in hepatocytes (hepatic steatosis) to hepatic steatosis with inflammation (steatohepatitis), fibrosis, cirrhosis and in severe cases, hepatocarcinoma [3]. Although previously thought to be benign, it is now clear that steatosis can evolve to more severe liver damage. Furthermore, TG accumulation, which is the main characteristic of hepatic steatosis, is strongly associated with the development of insulin resistance. However, despite the close correlation between TG accumulation and insulin resistance, it remains unclear whether insulin resistance is responsible for the excessive fat deposition in the liver or whether the increase in TG content is a prerequisite for the development of insulin resistance.

Currently, there is no effective treatment for steatosis. Lifestyle modifications, similar to those recommended for obesity, remain the best therapeutic option. In order to develop appropriate treatment, the knowledge of the cellular and molecular mechanisms leading to hepatic steatosis is essential.

Mechanisms Leading to Hepatic Lipid Accumulation

Steatosis occurs when there is an imbalance between lipid availability and lipid disposal (via fatty acid oxidation or VLDL secretion) (figure 1). In physiological conditions, the potential sources of fatty acids that contribute to fatty liver development are non-esterified fatty acids coming from the hydrolysis of TGs stored in the adipose tissue, dietary fatty acids arising from uptake and metabolism of intestinal chylomicrons and newly fatty acids synthesized through de novo lipogenesis in the liver. Once in the liver, fatty acids are esterified into TGs, which can be stored in lipid droplets into the hepatocytes or secreted as TGs-enriched lipoproteins (VLDL) into the bloodstream. Hepatic TGs can also be hydrolysed and the released fatty acids oxidized by the mitochondrial β oxidation. Studies in rodents and humans have revealed that the excessive accumulation of TG observed in hepatic steatosis is mainly because of an overflow of fatty acids coming from the lipolysis of hypertrophied (and insulin resistant) adipose tissue and from the lipogenic pathway while lipid disposal through β oxidation or export has only minor contributions [4]. Quantitative analysis using isotopic methods in various nutritional and physiopathological conditions has shown that the fractional contribution of lipogenesis to VLDL TGs is in the 2–5% range in normal subjects eating a typical western-diet but that high-carbohydrate, low-fat, or simple-sugar enriched diets, obesity, alcohol consumption and infectious states can strongly increase this proportion up to 25–30% [5–8] (figure 1). Donnelly et al. showed in patients with NAFLD that 59% of the TG labelled in the liver arose from circulating fatty acid flux (mainly from adipose tissue lipolysis), 26% from de novo lipogenesis and 15% from the diet [9]. Notably, the lipogenic flux, which is normally inhibited in fasting conditions, is already high in the fasted state and failed to further increase post-prandially in NAFLD subjects.

Figure 1.

Respective contributions of various metabolic pathways to steatosis in obese humans. Reductions in fatty acid oxidation and triglyceride (TG) export seem to have only minor roles in hepatic TG deposition. By contrast, the increased availability of plasma fatty acids arising from the adipose tissue through unabated lipolysis and de novo lipogenesis from glucose are major providers of lipids in steatotic livers.

Lipogenesis and its Transcriptional Control by SREBP-1c and ChREBP

Lipogenesis is the metabolic pathway allowing the conversion of an excess of carbohydrates into fatty acids, which are ultimately esterified with glycerol 3-phosphate to form TGs. The activity of the lipogenic pathway is strongly dependent upon the nutritional conditions. A diet rich in carbohydrates stimulates the lipogenic pathway, whereas starvation decreases its activity. Lipogenic enzyme activities are controlled by post-translational mechanisms but the main control is at the transcriptional level. It is clearly established that the transcription of lipogenic enzymes requires to be fully induced by both high insulin and glucose concentrations [10,11]. SREBP-1c (sterol regulatory element binding protein-1c) and ChREBP (carbohydrate response element binding protein) have been identified as, respectively, mediators of the transcriptional effects of insulin and glucose on glycolytic and lipogenic gene expression [11,12]. It has been shown that glucokinase, the first enzyme of the glycolytic pathway, is a target gene of SREBP-1c alone [13], whereas lipogenic genes such as fatty acid synthase and acetyl-CoA carboxylase require both SREBP-1c and ChREBP to be fully activated [11,14].

SREBP-1c, a Master Regulator of the Lipogenic Pathway

SREBP-1c belongs to the basic-helix-loop-helix-leucine zipper (bHLH-LZ) family of transcription factors. Two other isoforms, SREBP-2 and SREBP-1a, are involved in the regulation of genes involved in cholesterol synthesis [15]. SREBPs are synthesized as inactive precursors bound to the membranes of the endoplasmic reticulum (ER) and thus must undergo proteolytic cleavage to liberate their N-terminal domain, which constitutes the mature transcription factor.

Transcriptional Regulation of SREBP-1c. SREBP-1c expression is transcriptionally controlled by various nutritional and hormonal factors, insulin being one of the most potent activator. In cultured hepatocytes, insulin activates the transcription of SREBP-1c whereas glucagon is inhibitory [16,17]. The induction of SREBP-1c transcription leads to a parallel increase in the expression of both the ER membrane-bound precursor and the nuclear mature form of the transcription factor [18]. The effect of insulin on SREBP-1c has been corroborated by in vivo studies showing that SREBP-1c expression and nuclear abundance are low in the liver of diabetic rats and increase markedly after an insulin treatment [17]. The effect of insulin on SREBP-1c expression is mediated by the phosphoinositide 3-kinase (PI3K)-dependent pathway [18,19] but the nature of the downstream effectors remains unclear. Several studies using kinase inhibitors in cultured cells and/or gene knockout experiments in mice indicated that the PI3K/PDK1/protein kinase B (PKB) axis is required for induction of SREBP-1c expression by insulin in the liver [18–22]. By contrast, other genetic models show a role for atypical protein kinase Cλ in the control of SREBP-1c expression by insulin [23–25]. A recent study pointed out a role of PKC β in the insulin-induced expression of SREBP-1c [26]. Finally, it has been recently reported that mammalian target of rapamycin complex 1 (mTORC1) is an essential component in the insulin-regulated expression of SREBP-1c in the liver [27]. The mechanism by which insulin enhances transcription of SREBP-1c is unknown, but the stimulation requires the participation of liver X receptors (LXR) (see below) and SREBP-1c itself producing a feed-forward stimulation [28,29].

A potent activator of SREBP-1c transcription is LXRα. LXRα is a nuclear hormone receptor with high hepatic expression that is activated by oxysterols (metabolites of cholesterol) [30,31]. LXRα induces the expression of a range of genes involved in cholesterol efflux and clearance [32]. In vivo studies have revealed a role for LXRα in the control of SREBP-1c expression and of its lipogenic target genes. Animals lacking LXRα exhibit reduced basal expression of SREBP-1c, FAS, ACC and SCD-1 [33,34]. In contrast, animals fed on a high-cholesterol diet or synthetic LXR agonists show a selective increase in SREBP-1c mRNA and nuclear protein, induced expression of lipogenic target genes and elevated rates of lipogenesis [34,35]. It has been proposed that LXRα induces SREBP-1c in order to generate fatty acids needed for the formation of cholesterol esters, which buffer the free cholesterol concentration. As we will see below, the sole activation by LXR of SREBP-1c transcription is not sufficient to generate an active from of SREBP-1c. Indeed, insulin is absolutely necessary to release the SREBP-1c active form from the ER membranes [36].

SREBPs' Activation by Proteolytic Cleavage. The limiting step in the activation of SREBP-1c is the release of a mature transcriptionally active form from the ER membranes. Indeed, SREBP-1c is synthesized as a precursor protein bound to the membranes of the ER and must be liberated by a cleavage process (figure 2). Two proteins are essential for this cleavage: SREBP cleavage-activating protein (SCAP) and insulin-induced gene (INSIG). SCAP is a large integral membrane protein of the ER that interacts with newly synthesized SREBP precursor and escorts it from the ER to the Golgi apparatus [15,37]. Once in the Golgi, SREBP is released from the membranes through a sequential two-step proteolytic process involving two resident proteases: S1P and S2P [15]. SCAP interacts also with INSIG, an ER protein that is deeply embedded in the membranes and retains the SCAP–SREBP complex within the ER [15]. Two mammalian INSIG proteins (INSIG1 and 2) encoded by two different genes have been identified. The expression of INSIG1 is strongly induced by insulin through SREBP-1c. By contrast, the mRNA encoding the INSIG2 protein (INSIG2a mRNA) is repressed by insulin in the liver [38]. The cellular events leading to proteolytic cleavage of the SREBP precursor have been elucidated for SREBP-2 and SREBP-1a, two isoforms that are processed in response to cholesterol depletion [15]. The association between the SREBP/SCAP complex and INSIG is tightly regulated by the binding of regulatory sterols directly to INSIG or to ‘the sterol sensing domain’ of SCAP [39,40]. When cholesterol levels fall in the ER, the conformation of SCAP changes leading to its dissociation from INSIG. This unmasks a critical interaction motif of SCAP, the sequence MELADL, which interacts with the Sec24 subunit of the coat protein II (COPII) trafficking complex and facilitates movement of the SCAP–SREBP cargo to the Golgi apparatus [40]. After its sterol-dependent dissociation from SCAP, INSIG-1 interacts with the membrane-bound E3 ubiquitin ligase GP78, resulting in its ubiquitination and proteasomal degradation [41].

Figure 2.

Mechanism of SREBP-1c (sterol regulatory element binding 1c) activation by proteolytic cleavage. The mature form of SREBP-1c is embedded in the endoplasmic reticulum (ER) membranes where it interacts with the protein SREBP cleavage-activating protein (SCAP). SCAP itself interacts with INSIG (insulin-induced gene) protein which retains the complex SREBP-1c/SCAP in the ER. In the presence of an adequate signal (insulin, ER stress), SCAP dissociates from INSIG and the complex SCAP–SREBP-1c is transferred to the Golgi apparatus in coat protein II (COPII) vesicles. SREBP-1c undergoes in the Golgi apparatus a double cleavage process by the proteases S1P and S2P which finally liberates the mature active form of the transcription factor.

The proteolytic cleavage of SREBP-1c is not affected by cholesterol depletion but is rather stimulated by insulin treatment. In cultured rat hepatocytes, we showed that insulin led to a decrease of the microsomal form of SREBP-1c and to the rapid build-up of the SREBP-1c mature form [36]. By overexpressing the SREBP-1c precursor form in hepatocytes, Yellaturu et al. also showed that insulin treatment led to an enhanced post-translational processing of SREBP-1c. The mechanism by which insulin stimulates SREBP-1c processing is not entirely clear. It has been shown that insulin phosphorylates the ER-bound nascent SREBP-1c protein, increasing the affinity of the SCAP–SREBP-1c complex for the Sec23/24 proteins of the COPII vesicles. Chemical inhibition of the PI3kinase/PKB pathway prevented both insulin-mediated phosphorylation of nascent SREBP-1c protein and its post-translational processing [42]. A downregulation of INSIG2 mRNA by insulin has been also proposed as a potential mechanism leading to SREBP-1c activation [36,38]. It has been recently reported that insulin enhanced the turnover rate of INSIG2a mRNA via its 3′-untranslated region. The insulin-induced downregulation of INSIG2a mRNA promotes association of the SCAP/SREBP-1c complex with COPII vesicles and subsequent migration to the Golgi [43]. Finally, it has been recently reported that activation of the ER stress pathway results in the proteolytic cleavage of the SREBP-1c and SREBP-2 isoforms [44,45]. The underlying mechanisms are presently unknown but potential explanations are given in the following paragraphs.

Post-translational Modifications of SREBPs. Once in the nucleus, mature SREBP is subject to many post-translational modifications including phosphorylation [46–48], acetylation [49], sumoylation [50] and ubiquitination [47,48,51]. These modifications regulate the stability and/or transcriptional activity of the active transcription factor.

ChREBP, A Glucose-Responsive Transcription Factor

As mentioned before, SREBP-1c activity alone does not appear to fully account for the stimulation of lipogenic gene expression in response to carbohydrate because SREBP-1c gene deletion in mice only results in a 50% reduction in fatty acid synthesis [52]. In addition, SREBP-1c expression is not sufficient to account for the glucose/insulin induction of glycolytic and lipogenic genes in primary cultured hepatocytes [14]. Therefore, glucose metabolism via GK and SREBP-1c exerts a synergistic effect on the expression of glycolytic and lipogenic genes.

ChREBP has emerged as a glucose-responsive transcription factor in the liver [14,53]. Glucose activates ChREBP by stimulating its expression [14], by regulating its entry from the cytosol into the nucleus and by promoting its binding to carbohydrate responsive element (ChoRE) present in the promoter regions of both glycolytic and lipogenic genes [54]. ChREBP is located in the cytoplasm when the glucose concentration is low and enters into the nucleus when the glucose concentration is high [54]. This would be secondary to a dephosphorylation of a serine (Ser196) phosphorylated by protein kinase A (PKA) in conditions of high cAMP concentrations such as the fasting state. A second PKA phosphorylation near the DNA binding domain and which precludes ChREBP binding would also be dephosphorylated in the presence of high glucose [54]. Thus, the main effect of glucose would be to activate a phosphatase counteracting the effect of cAMP, inducing the translocation of ChREBP in the nucleus and stimulating its DNA binding activity. Glucose must be metabolized in order to activate ChREBP, and xylulose-5-phosphate has been described as the activator [55]. It implies that ChREBP activity is dependent secondarily on SREBP-1c because SREBP-1c activates expression of glucokinase, the enzyme that catalyses the first step in hepatic glucose metabolism. Like SREBP-1c, ChREBP is a key determinant of lipid synthesis [14].

Lipogenesis and Selective Hepatic Insulin Resistance

The livers of obese insulin-resistant rodents, such as the ob/ob or lipodystrophic mice, are characterized by severe hepatic insulin resistance resulting in glucose overproduction and hyperglycaemia. Insulin-dependent lipogenesis is paradoxically very active in the same livers. Moreover, hepatic levels of SREBP-1c are over-induced in the livers of obese animals [45,56,57]. The failure of insulin to suppress gluconeogenesis while lipogenesis is strongly activated could reflect a differential sensitivity to insulin (or a selective insulin resistance) according to the metabolic pathway. In an insulin-sensitive liver (figure 3A), insulin activates the kinase PKB/Akt which phosphorylates the transcription factor FoxO1, resulting in its exclusion from the nucleus. As FoxO1 is a major activator of the transcription of glucose-6-phosphatase (G6Pase), a key gluconeogenic gene, insulin thus downregulates G6Pase transcription and reduces glucose production [58]. Insulin also activates the transcription of the SREBP-1c gene and the cleavage of the precursor protein into its mature form, thus leading to an activation of the lipogenic program. Selective insulin resistance of the gluconeogenic pathway when compared to the lipogenic pathway could mean that somewhere between the insulin receptor and these two effectors, FoxO-1 and SREBP-1c, insulin signalling pathway diverges. One early diverging step could be the insulin receptor substrates (IRS) IRS1 and IRS2. A distinct role for IRS-1 and IRS-2 in hepatic metabolism has been proposed. IRS-1 would transduce the activating effects of insulin on SREBP-1c and its target genes, glucokinase and lipogenic genes, whereas IRS-2 would be involved in the inhibitory effects of insulin on FoxO1 and thus hepatic glucose production [57,59–61].

Figure 3.

Selective insulin resistance in the liver: what are the current hypothesis? (A) Insulin-sensitive liver: In an insulin-sensitive liver, insulin signals through insulin receptor substrate 1 (IRS-1), protein kinase B (PKB)/Akt and mammalian target of rapamycin complex 1 (mTORC1) and stimulates sterol regulatory element binding protein 1c (SREBP-1c) cleavage and thus activates the expression of lipogenic genes. Insulin also signals through IRS-2, PKB/Akt (probably a different pool from the one mobilized in IRS-1 signalling) and FoxO1. FoxO1 is a transcription factor necessary for the transcription of gluconeogenic enzymes such as glucose-6-phosphatase (G6Pase). When phosphorylated by PKB/Akt, it is excluded from the nucleus, thus resulting in the downregulation of gluconeogenesis. (B) The IRS hypothesis: The hyperinsulinemia observed in insulin-resistant animals would decrease the expression of IRS-2, thus precluding the inhibitory effect of insulin on gluconeogenesis, but not that of IRS-1 thus allowing an ongoing lipogenesis. This hypothesis does not take into account, however, the fact that the signalling through IRS-1 is normal or even reduced because of the action of stress kinases whereas lipogenesis is overactivated. (C) The mTORC1 hypothesis: It has been recently shown (see the text) that mTORC1 could be implied in SREBP-1c expression. As mTORC1 integrates not only signals arising from insulin but also from nutrients, it has been suggested that nutrient signals such as amino acids could be plethoric in obesity and thus replace the deficient insulin signal. (D) The endoplasmic reticulum (ER) stress hypothesis: It has been shown that an ER stress is able to activate SREBP-1c cleavage, the expression of its target genes and lipogenesis. As an ER stess is present in the liver of obese insulin-resistant rodents, it could, on one hand, induce an insulin resistance through the activation of stress kinases [e.g. Jun Kinase (JUNK)] in conjunction with lipids and cytokines, and, on the other hand, overactivate the lipogenic pathway through SREBP-1c maturation and in an insulin-independent manner.

It was proposed that the hyperinsulinaemia, which prevails in insulin-resistant states, can selectively induce a decreased content of IRS-2. This in turn could account for the resistance to insulin-mediated repression of gluconeogenic genes, whereas insulin could still activate lipogenic genes by IRS-1 [57] (figure 3B). A recent interesting hypothesis pertaining to selective hepatic insulin resistance is the involvement for SREBP-1c activation of the mTORC and more specifically mTORC1 (figure 3C). In the insulin signalling pathway, mTORC1 is downstream of PKB/Akt and upstream of the S6 kinase, a major regulator of protein translation and cell growth. Porstmann et al. [62] have shown in retinal epithelial cells that mTORC1 activity is necessary for PKB/Akt-induced SREBP-1c maturation and transcription and lipogenesis. Li et al. [27] have shown in primary cultured hepatocytes that whereas the insulin-induced expression of SREBP-1c is downregulated by rapamycin, an inhibitor of mTORC1, this treatment has no effect on insulin suppression of gluconeogenic genes. A similar selective effect of rapamycin was found in livers of rats and mice submitted to a fasting-refeeding protocol [27].

It remains however to explain in insulin-resistant obese rodents why a reduction in PI3-kinase-PKB/Akt activation could lead to a selective reduction in one of its downstream target, that is FoxO1, but into an activation of the other, mTORC1. As mTOR can be activated independent of insulin through an excess of cellular aminoacids, it has been suggested that it could be the case in obese rodents.

We have recently proposed that the ER stress response [or unfolded protein response (UPR)] could be involved in obese rodents in the high rate of lipogenesis despite the strong hepatic insulin resistance (figure 3D).

ER Stress, A Potential Actor Explaining a High Lipogenic Rate in Obese Insulin-Resistant Rodents

The ER is an organelle involved in the synthesis of secretory and membrane proteins, which are folded and assembled with the help of chaperones in the ER. Alterations of ER homeostasis such as disturbances in calcium concentration (the ER is the main calcium store of the cell), glycosylation potential or elevated synthesis of secretory proteins induce an adaptive coordinated response of ER called ER stress or unfolded protein response (UPR) to limit accumulation of unfolded protein and restore ER homeostasis. The UPR is regulated by three ER transmembrane proteins, the PERK kinase, the transcription factor ATF6 and the kinase and endonuclease IRE1 (inositol requiring enzyme 1). The activation of all three components of the UPR depends mainly on the dissociation of the luminal chaperone Bip/GRP78 from these signalling proteins. Bip/GRP78 dissociation from ATF6 results in the transit of ATF6 from the ER to the Golgi where it is cleaved (in a very similar manner as SREBP-1c and with the same S1P and S2P proteases), yielding an active transcription factor. The release of IRE1 and PERK from Bip/GRP78 also induces their homodimerization, autophosphorylation and activation. Once activated, these three effectors decrease protein translation and promote expression of foldases and chaperones or proteins involved in the degradation of misfolded proteins [63,64]. Two transcription factors, ATF4 and XBP1, are downstream targets of, respectively, PERK and IRE1. The ER stress response is also concomitant with the expansion of its membranes, thus increasing its capacity to perform correctly its protein folding functions.

Our interest in ER stress stems from observations linking UPR, lipogenesis, SREBP-1c and steatosis. The genetic deletion of XBP1 in liver leads to a decrease of the de novo lipid synthesis [65]. Similarly, sustained dephosphorylation of the elongation initiation factor eIF2α, a target of PERK, is associated with decreased hepatic lipogenesis and steatosis in mice fed on a high-fat diet [66]. PERK deletion itself induces a decrease of lipogenesis in mouse mammary gland [67]. Conversely, overexpression of a constitutively active form of ATF6 stimulated fatty acid synthesis in NIH-3T3 cells [68]. Patients suffering from severe hyperhomocysteinaemia develop hepatic steatosis. Homocysteine induces ER stress concomitant with the activation of SREBP in cultured human hepatocytes and in vascular endothelial cells [69]. Alcohol promotes an ER stress [70] and an increase in both SREBP-1c expression and its nuclear form [71,72]. It has also been suggested that in pancreatic beta cells, hyperglycaemia-induced ER stress could activate SREBP-1c [73], thus promoting lipid synthesis and inducing lipotoxicity in beta cells in type 2 diabetic subjects. The most important finding was the demonstration of the presence of a UPR in the liver and adipose tissue of insulin-resistant rodents, which when counteracted improves the insulin resistance of these animals [74–76]. Finally, it must be underlined that an expansion of the ER membrane during a UPR would require an additional provision of fatty acids in order to synthesize the necessary phospholipids. We then reasoned that in the liver of obese insulin-resistant animals, an ER stress, by inducing SREBP-1c cleavage, could be an alternative explanation for the ongoing lipid synthesis leading to hepatic steatosis.

UPR, SREBP-1c and Steatosis

Using cultured primary rodent hepatocytes, we have shown that regardless of its source (disturbance of calcium metabolism, glycosylation or the redox state), ER stress rapidly induces cleavage of the precursor form of SREBP-1c and expression of SREBP-1c target genes independent of insulin [45]. The expression of the other lipogenesis-related transcription factor ChREBP, which transduces more specifically the effects of glucose, is also enhanced by UPR, probably through the fact that SREBP-1c enhances the glycolytic flux through glucokinase [13,14].

If UPR-induced SREBP-1c cleavage contributes to steatosis in obese rodents, then UPR inhibition should reduce lipogenesis and steatosis by downregulating overexpressed glycolytic and lipogenic genes. We have inhibited the UPR response in vivo by overexpressing the chaperone Bip/GRP78 using an adenoviral vector. This inhibition was concomitant with a drastic reduction of the mature form of SREBP-1c in the nucleus, of the expression of ChREBP and with a massive decrease in glycolytic and lipogenic gene expression, leading to a markedly reduced hepatic steatosis [45]. An increased fatty acid oxidation could also be a contributing factor by de-inhibition of carnitine palmitoyl transferase I through a decrease in malonyl-CoA concentration.

Interestingly enough, inhibiting the UPR in obese mice also leads to reduced expression of cholesterol synthesis genes and decreased hepatic cholesterol concentration. This observation suggests that the other isoform of SREBP, SREBP-2, is activated in obese rodents by a UPR-dependent mechanism despite the absence of the usual trigger of SREBP-2 activation, that is cholesterol depletion. A concomitant activation of lipid and cholesterol synthesis would fit with the hypothesis that one reason for the activation of the SREBP pathway by UPR is the provision of lipid substrates for ER membrane synthesis. Interestingly, the UPR would bypass the usual regulators of SREBPs activation, insulin for SREBP-1 and cholesterol depletion for SREBP-2.

Several studies in humans support a role of SREBP in the pathogenesis of steatosis [77–79] but the exact role of ER stress requires to be further documented.

Finally, another mechanism that could contribute to hepatic steatosis is a decrease of VLDL synthesis and secretion. The packaging of TGs into VLDL is a complex process that takes place in ER lumen and requires synthesis of apolipoproteins such as apoB100, the major apolipoprotein of VLDL. It has been shown that ER stress leads also to a decrease of apoB100 secretion by inducing its co-translational proteasomal degradation but without affecting its mRNA [80]. This result was confirmed by another study showing that ER stress decreases apoB100 content by two different ways: degradation of apoB100 and attenuation of ApoB translation via the PERK branch of UPR [81].

SREBP-1c Processing: A Role for Bip/GRP78?

The mechanism by which the UPR could induce the cleavage and activation of SREBP-1c (and probably SREBP-2) is intriguing. Indeed, it cannot be solely related to a decreased content of INSIG proteins as suggested in some models [67,82] as the inhibitory effect of Bip/GRP78 overexpression on SREBP-1c cleavage occurs in mice without increase of INSIG protein expression [45]. An interesting possibility is that—similar to other UPR components like PERK, ATF6, IRE1—SREBP-1c is kept inactive in the ER by the binding of Bip/GRP78 to SREBP-1c itself or one of the proteins of the complex, SCAP or INSIG. We have indeed found an association between Bip/GRP78 and the SREBP-1c complex in lean mice, which was apparently disturbed in the liver of obese mice [45]. In conditions of ER stress, Bip/GRP78 dissociation would allow the export of SREBP1c to the Golgi for proteolytic cleavage as showed previously for ATF6. The precise involvement of GRP78 in SREBP-1c cleavage remains however to be addressed.

What Could be the Inducers of an ER Stress in the Liver of Obese Rodents?

Although it is now established that hepatic steatosis is associated with the appearance of ER stress, the mechanisms and the factors involved in this process are currently unknown. Obesity is characterized by multiple metabolic alterations such as dyslipidaemia, hyperinsulinaemia, hyperglycaemia, low-grade inflammation and oxidative stress, which could individually or together participate in the activation of the UPR pathway in the liver. For instance, saturated fatty acids such as palmitate are excellent activators of an ER stress in several cellular models including hepatocytes [83]. In relation with the chronic low-grade inflammation observed in obese patients, an effect of IFNγ on ER stress markers has also been described in primary hepatocytes [84]. Hyperglycaemia itself, by activating the hexosamine biosynthesis pathway (HBP), could promote ER stress as well as induction of lipogenesis leading to lipid accumulation [85].

Another explanation for the induction of an ER stress in obese rodents could be that the ER is the place where TGs are synthesized and packaged as membrane-embedded neutral lipid particles thus requiring de novo ER membrane formation [45] which could be sensed like a stress by the ER. TG synthesis is particularly active if excessive insulin-mediated lipogenic rate occurs (e.g. high intake of a carbohydrate diet in leptin deficient ob/ob mice) or in case of an excessive fatty acid availability (e.g. consumption of a high-fat diet).

ER Stress and Hepatic Insulin Resistance

Hepatic insulin resistance is a hallmark of obesity. Many different factors have been involved in order to explain this resistance. The present view is that lipids and inflammatory cytokines will activate stress kinases such as Jun Kinase (JUNK) and IKK. They will in turn interfere with insulin signalling by phosphorylating IRS-1 and -2 at specific serines, thus precluding their phosphorylation at tyrosines by the insulin receptor kinase. As pointed out by Hotamisligil in seminal papers [86,87], ER stress could also be an important player in this phenomenon. Indeed, ER stress is able to activate stress kinases and specifically JUNK [88]. Inhibition of the ER stress either by chemical chaperones or by overexpression of molecular chaperones attenuates JUNK activation and improves insulin sensitivity. In addition to a decreased tyrosine phosphorylation per IRS unit, IRS-2 expression is also markedly decreased in the liver of obese insulin-resistant rodents. It was convincingly shown that a hyperinsulinaemia per se could mimic the decreased IRS-2 expression [89]. We showed that in a situation of hyperinsulinaemia, inhibition of ER stress reverses the decreased IRS-2 expression [45]. This raises the intriguing possibility that the ER stress somehow mediates the effects of hyperinsulinaemia for the decreased IRS-2 expression. How an ER stress is then able to specifically decrease IRS-2 expression is presently unknown.


The reason why lipogenesis, an insulin-sensitive pathway, is overactivated in the liver of insulin-resistant obese subjects is still a matter of debates. An activation of lipogenesis through the UPR of the ER and SREBP-1c stimulation is certainly an interesting possibility. It explains at least in part the paradox of having to admit that one side of the insulin signalling pathway is still sensitive whereas the other one is resistant. In addition, it confers to the transcription factor SREBP-1c (and probably SREBP-2) a strategic place in the UPR in the liver if one considers that the membrane network represented by the ER is an organelle extensively requiring lipids.

Some important questions still remain. First what are the reasons for the development of an ER stress in the liver of obese insulin-resistant rodents? Is it directly linked to an initial fat storage which by activating an ER stress will activate lipogenesis and lead to a vicious circle or is it a consequence of insulin resistance (including hyperinsulinaemia) per se? Is it secondary to the inflammatory state which is a hallmark of obesity? Is it a physiological response which if attenuated would still worsen on the long term the metabolic homeostasis or could it be an interesting therapeutic target in order to improve liver steatosis and insulin sensitivity? Is SREBP-1c regulated by similar mechanisms as the other UPR actors, that is a chaperone-dependent mechanism? And finally are all these mechanisms also relevant to human physiopathology?


The authors are supported by grants from INSERM, from the Agence Nationale de la Recherche (ANR-2005-PCOD-035), from the EXGENESIS European Commission Integrated Project Grant LSHM-CT- 2004-005272, from the European Commission's Seventh Framework programme under grant agreement No. 241913 (FLORINASH) and from Alfediam Takeda.

Conflict of Interests

The authors do not declare any conflict of interest relevant to this manuscript.