A growing amount of evidence suggests the involvement of ER (endoplasmic reticulum) stress in lipid metabolism and in the development of some liver diseases such as steatosis. The transcription factor SREBP-1 (sterol-regulatory-element-binding protein 1) modulates the expression of several enzymes involved in lipid synthesis. Previously, we showed that ER stress increased the SREBP-1a protein level in HepG2 cells, by inducing a cap-independent translation of SREBP-1a mRNA, through an IRES (internal ribosome entry site), located in its leader region. In the present paper, we report that the hnRNP A1 (heterogeneous nuclear ribonucleoprotein A1) interacts with 5′-UTR (untranslated region) of SREBP-1a mRNA, as an ITAF (IRES trans-acting factor), regulating SREBP-1a expression in HepG2 cells and in primary rat hepatocytes. Overexpression of hnRNP A1 in HepG2 cells and in rat hepatocytes increased both the SREBP-1a IRES activity and SREBP-1a protein level. Knockdown of hnRNP A1 by small interfering RNA reduced either the SREBP-1a IRES activity or SREBP-1a protein level. hnRNP A1 mediates the increase of SREBP-1a protein level and SREBP-1a IRES activity in Hep G2 cells and in rat hepatocytes upon tunicamycin- and thapsigargin-induced ER stress. The induced ER stress triggered the cytosolic relocation of hnRNP A1 and caused the increase in hnRNP A1 bound to the SREBP-1a 5′-UTR. These data indicate that hnRNP A1 participates in the IRES-dependent translation of SREBP-1a mRNA through RNA–protein interaction. A different content of hnRNP A1 was found in the nuclei from high-fat-diet-fed mice liver compared with standard-diet-fed mice liver, suggesting an involvement of ER stress-mediated hnRNP A1 subcellular redistribution on the onset of metabolic disorders.
The liver plays a central role in whole-body lipid homoeostasis. In this organ, dietary carbohydrates and fatty acids, and pancreatic hormones, i.e. insulin and glucagon, participate in the regulation of lipogenesis and lipid oxidation. Previous reports have suggested that ER (endoplasmic reticulum) stress plays crucial roles in lipid metabolism [1–4]. ER stress occurs owing to the disruption in ER protein-folding capacity and leads to the activation of an evolutionarily conserved UPR (unfolded protein response) signalling system in order to restore ER homoeostasis . Accumulating evidence suggests that activation of the UPR pathway can increase lipid metabolism by modulating the transcription of genes coding for lipogenic enzymes . The effect of ER stress on lipogenesis has been observed in various tissues from obese mice and humans [6–9]. Furthermore, ER stress and the UPR pathway play a fundamental role in the development of non-alcoholic fatty liver disease where lipid droplets accumulate in hepatocytes .
A number of transcription factors that regulate lipid homoeostasis have been identified, including the SREBPs (sterol-regulatory-element-binding proteins), in cells from vertebrates [11–13]. The SREBP family consists of three different proteins, SREBP-1a, SREBP-1c and SREBP-2, which are encoded in the mammalian genome by two genes, SREBF1 and SREBF2. SREBP-1a is constitutively expressed at low levels in liver and in most tissues of adult animals, and it is the predominant isoform in most cultured cell lines. This isoform is a potent activator of genes involved in the synthesis of cholesterol, fatty acids and triacylglycerol. SREBP-1c predominates in liver and in most other tissues and it is the major mediator of insulin lipogenic action in liver. SREBP-1c preferentially enhances transcription of genes required for fatty acid, but not for cholesterol, synthesis . It contributes to the regulation of glucose uptake and glucose synthesis, and also activates the expression of target genes involved in glycolysis . SREBPs are synthesized as large precursor proteins (pSREBP) that are inserted into the ER through two membrane-spanning domains [11,16,17]. In the ER, the C-terminal regulatory domain of SREBPs interacts with SCAP (SREBP cleavage-activating protein), which functions as a sensor of the membrane cholesterol level. In sterol-replete cells, the SREBP–SCAP complex remains in the ER owing to the interaction of this complex with the ER-embedded INSIG (insulin-induced gene) [11,16,17]. When cells become depleted in cholesterol, the SREBP–SCAP complex translocates from the ER to the Golgi through COPII (coat protein complex II)-coated membrane vesicles [11,16,17]. In the Golgi, a two-step proteolytic cleavage releases the nSREBP (N-terminal half of SREBP), allowing its entry into the nucleus [11,16,17]. Here, the nSREBP binds to the sterol regulatory element and E-box sequences in the promoter region of genes involved in cholesterol and fatty acid biosynthesis [16–19].
It is well known that SREBP-1 expression is tightly regulated by transcriptional, post-transcriptional and post-translational mechanisms. Oxysterols activate the expression of SREBP-1c through the binding of LXR (liver X receptor)/RXR (retinoid X receptor) heterodimers to its promoter . By contrast, PUFAs (polyunsaturated fatty acids) inhibit SREBP-1c expression blocking LXR activation, competitively binding its endogenous ligands. Furthermore, PUFAs lower SREBP-1c levels by accelerating degradation of its mRNA . In isolated rat hepatocytes, insulin treatment increases the amount of mRNA for SREBP-1c in parallel with the mRNA of SREBP target genes. Conversely, incubating primary hepatocytes with glucagon decreases the mRNAs for SREBP-1c and for its lipogenic target genes . SREBP-1 is targeted by various post-translational modifications, including phosphorylation, acetylation, SUMOylation and ubiquitination .
A number of reports have highlighted the link between the UPR and SREBP transcriptional activity [1,3]. It has been shown that the homocysteine- or thapsigargin-induced UPR was able to activate SREBP-1c and, in turn, transcription of lipogenic genes [2,3,23]. Furthermore, we showed that ER stress induced by tunicamycin or thapsigargin treatment caused an increase in the level of SREBP-1a precursor and mature forms in HepG2 cells . We demonstrated that the increase in SREBP-1a content, observed after UPR induction in HepG2 cells, might be ascribed to an enhanced SREBP-1a mRNA cap-independent translation, through an IRES (internal ribosome entry site)-mediated mechanism . Similarly, treatment of HepG2 cells with tri-iodothyronine hormone increased the IRES-mediated expression of SREBP-1a .
The physiological role of IRES-mediated translation of some cellular mRNAs has been increasingly recognized in numerous biological processes, including differentiation, cell-cycle processes and apoptosis, and under stress conditions such as hypoxia, UV irradiation or treatment with ER stressors.
IRES activity is regulated by several proteins collectively termed ITAFs (IRES trans-acting factors) . Besides their implication in a variety of cellular activities (e.g. RNA splicing and/or export), ITAFs are generally believed to increase (or, in certain cases, decrease) the affinity of binding between IRESs and components of the translational apparatus (canonical initiation factors and ribosomes) . In fact, several ITAFs such as PTB (polypyrimidine tract-binding protein), Unr (upstream of N-Ras), hnRNP (heterogeneous nuclear ribonucleoprotein) C1/C2 and hnRNP A1 have been shown to enhance the IRES-mediated mRNA translation; in contrast, the ITAF HuR (human antigen R) inhibits p27Kip1 IRES activity [27–29]. In the present study, we investigated the molecular basis of the IRES-dependent translation of SREBP-1a mRNA. Furthermore, the putative role of the UPR on the regulation of SREBP-1 at the translational level has been demonstrated. The results of the present study demonstrate that hnRNP A1 represents an ITAF that is able to specifically interact with the 5′-UTR (untranslated region) of SREBP-1a mRNA. Experiments based on overexpression and/or knockdown of hnRNP A1 demonstrated that the binding of hnRNP A1 to the leader region of SREBP-1a mRNA increased the efficiency of the IRES-dependent translation of the corresponding mRNA. We also reported that tunicamycin-induced ER stress triggered the hnRNP A1 shuttling from nuclei to the cytosol, which in turn promoted the IRES-mediated translation of SREBP-1a mRNA. Finally, evidence for a different nuclear content of hnRNP A1 in liver from HFD (high-fat diet)-fed mice compared with SD (standard diet)-fed mice has been provided.
Plasmid pRS1aF was constructed by inserting the cDNA of the SREBP-1a 5′-UTR into the intercistronic site of the pRF vector . Plasmids pRc-mycF (formerly pGL3utrH) and pREF (formerly pCREL), have been described previously [30,31]. pcDNA-SV5-hnRNPA1 were kindly provided by Professor Ronald T. Hay (Wellcome Biocentre, University of Dundee, Dundee, U.K.) . The cDNA for hnRNP A1 was excised from pcDNA-hnRNP A1 and inserted into the bacterial expression vector pET28a. The cDNA for HuR was excised from pET21a-HuR  kindly provided by Dr Eleanor K. Spicer (Medical University of South Carolina, Charleston, SC, U.S.A.) and inserted into the pcDNA-SV5.
Cell culture, preparation of primary rat hepatocytes and transient transfection assay
HepG2 cells were maintained in DMEM (Dulbecco's modified Eagle's medium; Sigma) supplemented with 10% (v/v) heat-inactivated FBS (fetal bovine serum), penicillin G (100 units/ml) and streptomycin (100 μg/ml). Primary rat hepatocytes were cultured as described previously in . HepG2 cells and rat hepatocytes were kept at 37°C in a humidified atmosphere containing 5% CO2. For transient transfections, 3.5×104 cells were seeded into 12-well plates at 48 h before transfection. Cells were transfected using FuGENE® 6 (Roche Diagnostics) following the manufacturer's instructions. After an 8-h transfection period, the medium was changed to fresh DMEM supplemented with 10% (v/v) FBS and cells were incubated for 24 h. After cell lysis, RL (Renilla luciferase) and FL (firefly luciferase) activities were measured using the Dual-Luciferase® Reporter Assay System (Promega). The β-galactosidase activity was determined using a β-galactosidase assay.
RNA affinity chromatography
Nuclear proteins (500 μg) from HepG2 cells were precleared by incubating with 300 μl of Streptavidin MagneSphere® Paramagnetic Particles (Promega) for 1 h in 1 ml of binding buffer [20 mM Hepes/KOH (pH 7.9), 150 mM KCl, 5% (v/v) glycerol, 1 mM DTT (dithiothreitol), 0.5 mM EDTA, 25 μg/ml tRNA, 1.5 mM PMSF and 1.5 mM MgCl2]. Precleared proteins were then incubated with 30 μg of SREBP-1a or EMCV (Encephalomyocarditis virus) 5′-UTR biotin–RNA for 2 h. Fractions (300 μl) of paramagnetic particles were then added to the biotinylated RNA–protein complexes and incubated for 1 h. All incubations were performed at 4°C at constant rotation. The proteins retained on the paramagnetic particles were eluted by 50 μl of SDS/PAGE sample buffer at 95°C for 2 min, separated by SDS/10% PAGE and stained by Brilliant Blue G colloidal concentrate (Sigma).
Purification of recombinant proteins
The procedure used for purification of recombinant hnRNP A1 and HuR from Escherichia coli cells has been described previously .
UV cross-linking, immunoprecipitation of UV-cross-linked proteins and competition experiments
DNA templates for synthesis of probes corresponding to the full-length SREBP-1a 5′-UTR or to its deletions and to c-myc IRES were generated by PCR using pRS1aF and pRc-mycF as a template, and the following primers: probe RNAI, forward 5′-CGGCCGGGGGAACCCAGTT-3′ and reverse 5′-CCATGGCGCAGCCGCCTCC-3′; probe RNAII, forward 5′-CGGCCGGGGGAACCCAGTT-3′ and reverse 5′-GCCGCCGCGGCCCCGGCTCT-3′; probe RNAIII, forward 5′-CGGCCGGGGGAACCCAGTT-3′ and reverse 5′-CCTGGCCTCAGAGGCGGCC-3′; probe RNAIV, for 5′-GCAGGACACGAACGCGCGGA-3′ and reverse 5′-GGCGCAGCCGCCTCCTCCGG-3′; probe RNAV, forward 5′-GCTCCCTAAGAAGGGCCGTA-3′ and reverse 5′-GGCGCAGCCGCCTCCTCCGG-3′; probe c-myc, forward 5′-TAATTCCAGCGAGAGGCAGA-3′ and reverse 5′-ATACCATGGTCGCGGGAGGCTGCT-3′ . All forward primers have anchored in the 5′-region of the T7 promoter sequence 5′-CGGAATTAATACGACTCACTATAGGG-3′. A BamHI/NcoI fragment containing the EMCV IRES with a T7 promoter sequence was excised from the bicistronic pREF construct and used as a template for the synthesis of the corresponding RNA probe. The 32P-labelled RNA probe corresponding to the SREBP-1a 5′-UTR was synthesized by in vitro transcription with T7 RNA polymerase (MAXIScript T7 RNA polymerase kit; Ambion) in the presence of [α-32P]GTP. 32P-labelled RNA probe (3×105 c.p.m.) was incubated with 20 μg of HepG2 nuclear extracts or 100 ng of purified His6-tagged hnRNP A1 in 30 μl of binding buffer. In competition experiments, 20× or 100× of unlabelled RNA competitor was incubated together with the 32P-labelled SREBP-1a 5′-UTR in binding buffer. After 20 min of incubation at 25°C, the samples were irradiated with UV on ice for 15 min with a UV Stratalinker (Stratagene). Unbound RNA was digested with 4 μl of RNase cocktail (2 μl of 10 mg/ml RNase A and 2 μl of 100 units/ml RNase T1) at 37°C for 45 min and then analysed by SDS/10% PAGE. For UV cross-linking and immunoprecipitation assays, after RNase cocktail treatment, samples were diluted in 1 ml of immunoprecipitation buffer [20 mM Hepes/KOH (pH 7.4), 125 mM KCl, 0.05% Nonidet P40, 0.5 mM DTT, 0.5 mM PMSF and 0.5 mM EDTA] and precleared by adding 30 μl of Protein G–agarose slurry (Roche). After 1 h of incubation at 4°C in a rotary mixer, samples were centrifuged for 5 min at 12000 g. Then 2 μg of anti-hnRNP A1 (sc32301, Santa Cruz Biotechnology) monoclonal antibody was added to the precleared samples. After 2 h of incubation at 4°C in a rotary mixer, 30 μl of Protein G–agarose slurry was added and further incubated for 2 h. After washing four times with 1 ml of immunoprecipitation buffer, resin-bound proteins were detached from the beads by adding 30 μl of Laemmli sample buffer and heating the mixture to 100°C for 5 min and then the proteins were resolved using SDS/12% PAGE.
RNA extraction and Northern blot analysis
Preparation of nuclear and cytosolic protein fractions from HepG2 cells
All procedures were carried out at 4°C. To prevent proteolysis, a mixture of protease inhibitors from Sigma was included in all of the buffers. HepG2 cells from 25 cm2 flasks were pooled and centrifuged at 900 g for 5 min at 4°C. The resulting cell pellet was resuspended in buffer 1 [20 mM Tris/HCl (pH 8.0), 420 mM NaCl, 2 mM EDTA, 2 mM Na3VO4, 0.2% Nonidet P40 and 10% (v/v) glycerol]. After 10 min of incubation on ice, cells were passed several times through a 20-gauge syringe needle and then sonicated until no cells remained intact. The homogenate was centrifuged at 1100 g for 10 min, and the supernatant was collected as the cytosolic fraction. The nuclear pellet was washed once in buffer 1 and then resuspended in high-salt buffer 2 [20 mM Tris/HCl (pH 7.9), 420 mM NaCl, 10 mM KCl, 0.1 mM Na3VO4, 1 mM EDTA, 1 mM EGTA and 20% (v/v) glycerol]. This suspension was rotated for 30 min and then centrifuged at 15000 g for 30 min. The resulting supernatant is designated as the nuclear extract fraction. The purity of fractions was tested by immunoblotting with an anti-lamin B polyclonal antibody. The protein concentration was determined using the Bio-Rad protein assay kit.
Preparation of nuclear extract from mice liver
Livers from lean and obese mice were kindly provided by Dr Silvana Gaetani (Sapienza, University of Rome, Rome, Italy). To obtain liver samples, male C57BL/6 mice were divided into two groups and fed on different diets for 24 days. The first group (lean mice, SD group) received a standard diet (Research Diet, D12450B) containing 4.3% fat, 19.2% protein and 67.3% carbohydrate. The second group (obese mice, HFD group) received a diet with a higher fat content (Research Diet, D12451), with 24% fat, 24% protein and 41% carbohydrate. Animals were allowed ad lib access to food and water. Mice were killed by decapitation, according to the guidelines for care and use of laboratory animals. All procedures were performed in conformity with the instructions of the Italian Committee for Experimental Animals. Livers were homogenized in a motor-driven Potter–Elvehjem homogenizer in 1 volume of homogenization buffer [10 mM Hepes (pH 7.9), 10 mM KCl, 2 mM Na3VO4, 0.1 mM EDTA, 0.74 mM spermidine, 0.25 M sucrose, 50 mM imidazole, 1 mM DTT, 4 mM PMSF and 1 mM benzamidine], and the homogenate was centrifuged at 1200 g for 15 min. Nuclei were further purified as reported in , and the nuclear fraction was obtained as described above.
Western blot analysis
XBP-1 (X-box-binding protein 1), hnRNP A1 and SREBP-1 protein content was analysed as described in . A total of 50 μg of protein was separated by SDS/PAGE. After electrophoretic transfer to nitrocellulose, the blot was probed with antibodies directed against SREBP-1 (sc-13551, Santa Cruz Biotechnology), XBP-1 (sc-7160, Santa Cruz Biotechnology), hnRNP A1 (sc-32301, Santa Cruz Biotechnology), β-actin (sc-47778, Santa Cruz Biotechnology) and lamin B (sc-6216, Santa Cruz Biotechnology). The detection system employed was the ECL Plus™ Western blotting reagents (GE Healthcare).
RNA interference analysis
HnRNP A1 gene silencing in HepG2 cells and in primary rat hepatocytes was performed by RNA interference with synthetic siRNA (small interfering RNA) (Silencer® Select Pre-designed siRNA, Ambion Life Technologies; FlexiTube siRNA, Qiagen) directed at the sequences within the 3′-UTR of human and rat hnRNP A1. An siRNA with a scrambled sequence was used as a negative targeting control. The expression of hnRNP A1 was rescued by co-transfecting HepG2 cells or rat hepatocytes with hnRNP A1-directed siRNA together with pcDNA-SV5-hnRNPA1, which harbours the hnRNP A1 cDNA lacking its 3′-UTR target of the hnRNP A1 siRNA. Transfection was performed for 48 h by using Metafectene Pro (Biontex), according to the manufacturer's recommendations.
hnRNP A1 interacts with the SREBP-1a IRES
In order to elucidate the molecular basis of the IRES-dependent translation of SREBP-1a mRNA, we analysed the cellular factors interacting with the SREBP-1a IRES by RNA affinity chromatography. The nuclear extract from HepG2 cells was incubated with the biotinylated 5′-UTR of SREBP-1a mRNA and the retained proteins were resolved by SDS/12% PAGE. The pattern of proteins bound to the SREBP-1a leader was compared with that obtained by an analogous experiment performed using the 5′-UTR of EMCV mRNA that contains a well-characterized IRES [37,38]. Colloidal Blue staining of the gel revealed that several proteins were retained by the RNAs (Figure 1). We focused our attention on the 34-kDa protein, which is specifically bound to the SREBP-1a RNA but not to the EMCV RNA. This protein shares the molecular mass of hnRNP A1, which is a well-known ITAF . To define whether the 34-kDa protein corresponded to hnRNPA1, a UV cross-linking assay followed by immunoprecipitation with an hnRNP A1-specific monoclonal antibody was performed. In this experiment, nuclear proteins from HepG2 cells were incubated with the 32P-labelled SREBP-1a 5′-UTR or [32P]-labelled c-myc and EMCV IRESs used as positive and negative control respectively [39,40]. Results reported in Figure 2(A) showed that the 34-kDa protein was detected after immunoprecipitation of the UV cross-linked RNA–protein complexes from SREBP-1a (Figure 2A, lane 2) and c-myc (Figure 2A, lane 3) IRESs, but not from the EMCV IRES (Figure 2A, lane 4).
Analysis of proteins interacting with the SREBP-1a 5′-UTR
Identification of hnRNP A1 as an ITAF interacting with the SREBP-1a IRES
To evaluate the interaction of hnRNP A1 with the SREBP-1a 5′-UTR, a UV cross-linking assay was carried out by using the purified His6-tagged hnRNP A1. A protein migrating at the size corresponding to the hnRNP A1 molecular mass (34 kDa) was observed when 32P-labelled RNA corresponding to SREBP-1a 5′-UTR was used (Figure 2B, lane 1). A cross-linked RNA–protein complex was found at the same migration position of hnRNP A1 when the c-myc IRES was tested (Figure 2B, lane 2), whereas a weak signal of the cross-linked RNA–protein complex was found for the EMCV IRES (Figure 2B, lane 3). A parallel UV cross-linking experiment was performed with the purified His6-HuR, which is an IRES-interacting protein . When His6-HuR was used in the UV cross-linking assay with the SREBP-1a 5′-UTR probe, a weak signal of the cross-linked RNA–protein complex was detected (Figure 2B, lane 4).
To test the specificity of hnRNP A1 binding to the SREBP-1a IRES, a UV cross-linking experiment was carried out in the presence of excess unlabelled competitor SREBP-1a, c-myc and EMCV IRESs. Results showed that unlabelled SREBP-1a and c-myc IRESs, but not EMCV IRES, were competitive with 32P-labelled SREBP-1a IRES for binding the hnRNP A1 (Figure 2C).
In order to map the site of hnRNP A1 binding to the SREBP-1a 5′-UTR, the purified His6-hnRNP A1 protein and 32P-labelled RNA spanning from nucleotide −189 to −1 of the SREBP-1a mRNA were UV cross-linked in the presence of 100-fold unlabelled SREBP-1a 5′-UTR or its deleted fragments, as depicted in Figure 3(A). The addition of 100-fold molar excess of unlabelled SREBP-1a 5′-UTR (RNA I), or RNA II or RNA III corresponding to nucleotide −189/−70 and nucleotide −189/−126 respectively, strongly inhibited the binding of hnRNP A1 to the probe (Figure 3B, lanes 2–4). RNA IV and RNA V, corresponding to nucleotide −125/−1 and nucleotide −69/−1 respectively, did not compete for the binding with hnRNP A1 (Figure 3B, lanes 5 and 6). Taken together, these data indicate that the hnRNP A1 protein strongly binds to the −189/−126 region of the SREBP-1a 5′-UTR.
Identification of the minimal region of the SREBP-1a 5′-UTR interacting with hnRNP A1
hnRNP A1 increases the activity of the SREBP-1a IRES
The binding of hnRNP A1 to the SREBP-1a 5′-UTR led us to hypothesize that this protein could play a role on SREBP-1a IRES activity. Therefore we performed co-transfection experiments in HepG2 cells by using the bicistronic construct pRS1aF , together with either the pcDNA3-hnRNP A1 effector plasmid harbouring the cDNA coding for SV5-tagged hnRNP A1 or the pcDNA3 empty vector. pRS1aF contains the human SREBP-1a 5′-UTR inserted between two reporter genes coding for RL and FL  (Figure 4A). After transcription in cells, pRS1aF generates a bicistronic mRNA in which the first cistron RL is translated via a cap-dependent mechanism, whereas the second cistron FL is translated independently from the cap structure, through the IRES of the SREBP-1a 5′-UTR . The analogous experiment was performed by using pRc-mycF and pREF bicistronic constructs, containing the c-myc and EMCV 5′-UTR respectively (Figure 4A). As shown in Figure 4(B), overexpression of hnRNP A1 in Hep G2 cells increased the IRES activity of SREBP-1a mRNA by approximately 2.8-fold. Analogously, overexpression of hnRNP A1 caused a similar increase in c-myc IRES activity (Figure 4B). By contrast, the IRES activity of the EMCV 5′-UTR assayed with the pREF construct was not affected by the overexpression of hnRNP A1 (Figure 4B). In all the experiments, RL activity was not affected by hnRNP A1 overexpression. We also performed co-transfection experiments in HepG2 cells by using pRS1aF together with pcDNA-HuR or pcDNA-SSB harbouring cDNA coding for HuR and for La autoantigen respectively, being both ribonucleoproteins able to induce IRES activity [41,42]. Results reported in Figure 4(C) showed that neither HuR nor La autoantigen are able to transactivate SREBP-1a IRES even though these effector proteins were expressed at detectable levels in the transfected cells (results not shown). Overexpression of hnRNP A1 did not affect either the amount or the integrity of the bicistronic mRNA (Figure 4D). As shown by Northern blot analysis, no monocistronic mRNA corresponding to the FL ORF (open reading frame), which might be produced by a putative cryptic promoter or cryptic splicing site, has been detected. The knockdown of hnRNP A1 gene expression was carried out in Hep G2 cells by siRNA. As shown in Figure 4(E), a drastic decrease in both hnRNP A1 mRNA and protein levels was observed in Hep G2 cells transfected with an hnRNP A1-specific siRNA when compared with cells transfected with scrambled siRNA. siRNA-mediated hnRNP A1 knockdown caused a reduction of FL activity, but not RL activity in Hep G2 cells transfected with pRS1aF (Figure 4F). Taken together, these results indicate that hnRNP A1 specifically enhances the SREBP-1a IRES activity.
hnRNP A1 enhances the IRES activity of SREBP-1a mRNA
Changes in hnRNP A1 expression alters the SREBP-1 protein level in HepG2 cells and in rat hepatocytes
Next, the existence of a link between the hnRNP A1 expression and the SREBP-1 protein level was evaluated by overexpressing or interfering with hnRNP A1 gene expression in HepG2 cells. We observed an increase in the content of both the precursor and nuclear forms of SREBP-1 in HepG2 cells transfected with pcDNA-SV5-hnRNP A1 with respect to control cells transfected with the pcDNA3 empty vector (Figure 5A). Conversely, the SREBP-1 content decreased after siRNA-induced hnRNP A1 gene knockdown when compared with that measured in scrambled siRNA-transfected control cells (Figure 5A). The SREBP-1 mRNA abundance was not affected by either hnRNP A1 overexpression or by siRNA-induced hnRNP A1 gene silencing (Figure 5B). The effect of hnRNP A1 protein content on the SREBP-1 expression in primary rat hepatocytes was also analysed. Similar results described for HepG2 cells have been observed in primary rat hepatocytes (Figures 5C and 5D).
Effect of overexpression or knockdown of hnRNP A1 on SREBP-1a expression in HepG2 cells and in primary rat hepatocytes
Role of hnRNP A1 on the activation of SREBP-1a IRES-dependent translation under ER stress conditions
It has been demonstrated that ER stress increased the SREBP-1a IRES activity in HepG2 cells . Thus we evaluated whether in ER-stressed HepG2 cells and primary hepatocytes hnRNP A1 could play a role on the activation of cap-independent translation of SREBP-1a, through the IRES activity. Treatment with 1 μg/ml tunicamycin, a well-known ER stressor, strongly enhanced SREBP-1a protein levels in scrambled siRNA, but not in hnRNP A1 siRNA in transfected HepG2 cells (Figure 6A, lanes 2 and 3) and rat hepatocytes (Figure 6D, lanes 2 and 3) compared with the respective control cells (Figures 6A, lane 1, and 6D, lane 1). The tunicamycin-mediated induction of SREBP-1 expression in hnRNP A1-knockdown cells was rescued after hnRNP A1 overexpression (lane 4 in Figures 6A and 6D). To rule out the possibility that changes in the SREBP-1a protein level could be due to a modification of the corresponding mRNA level, the SREBP-1a mRNA abundance was monitored by qRT (quantitative real-time) PCR. In agreement with previous results , the SREBP-1a mRNA abundance was not affected by tunicamycin treatment (Figure 6B). No significant change in SREBP-1a mRNA abundance was observed in tunicamycin-treated hnRNP A1-knockdown and in hnRNP A1-rescued HepG2 cells with respect to control cells (Figure 6B). Similar results were observed in rat hepatocytes (Figure 6E).
Effect of hnRNP A1 knockdown on the ER stress-mediated induction of SREBP-1a expression in HepG2 cells
The role of hnRNP A1 on the tunicamycin-mediated induction of SREBP-1 IRES activity was also analysed. HepG2 cells and rat hepatocytes were co-transfected with the bicistronic pRS1aF construct together with either scrambled siRNA or hnRNP A1-specific siRNA for 48 h. After this period, ER stress was induced either in scrambled siRNA-transfected cells or in hnRNP A1-knockdown cells by adding 1 μg/ml tunicamycin in culture medium for a further 8 h. After ER stress induction, cells were harvested and RL and FL activities were compared with those measured in control cells transfected with scrambled siRNA and incubated in the absence of tunicamycin. As reported in Figure 6(C), ER stress induction increased the SREBP-1a IRES activity by approximately 6- and 5-fold in Hep G2 cells (Figure 6C, lane 2) and rat hepatocytes (Figure 6F, lane 2) respectively, with respect to control cells (lane 1 in Figures 6C and 6F). By contrast, tunicamycin treatment did not cause significant change in IRES activity of SREBP-1a mRNA in hnRNP A1-knockdown cells (lane 3 in Figures 6C and 6F). Furthermore, the induction of IRES activity of SREBP-1a mRNA in the same cells was rescued after overexpression of hnRNP A1 (lane 4 in Figures 6C and 6F).
Cytosolic shuttling of hnRNP A1 and its binding to the SREBP-1a IRES increase in ER-stressed cells
We hypothesized that induction of ER stress in Hep G2 cells by ER stressors, such as tunicamycin and thapsigargin, could favour the shuttling of hnRNP A1 from the nuclei to the cytosol. Therefore the nuclear and the cytosolic fractions of hnRNP A1 were analysed in tunicamycin- and in thapsigargin-treated Hep G2 cells by Western blotting experiments (Figures 7A and 7C). Results showed that hnRNP A1 levels in the nuclei decreased after 3 h of ER stress, reaching the minimum level at 24 and 12 h in tunicamycin- and thapsigargin-treated cells (Figures 7A and 7C, upper panel) respectively. However, only a slight increase in the cytosolic counterpart of hnRNP A1 was observed after stressor treatments (Figures 7A and 7C, lower panel). It has been shown that hnRNP A1 can be degraded through the ubiquitin–26S proteasome system . Therefore the effect of ER stressors on the nuclear and cytosolic distribution of hnRNP A1 in HepG2 cells treated with MG-132, an inhibitor of the ubiquitin–26S proteasome system was analysed.
Effect of tunicamycin- and thapsigargin-induced ER stress on the nucleo–cytosolic shuttling of hnRNP A1 and on the hnRNP A1 binding to the SREBP-1a IRES in Hep G2 cells
In the presence of 20 μM MG-132, treatment with 1 μg/ml tunicamycin caused a strong and time-dependent increase in cytosolic hnRNP A1 content, which was paralleled by the reduction of the nuclear fraction counterpart (Figure 7B). A similar result was obtained in 300 nM thapsigargin-treated cells, incubated with the ubiquitin–26S proteasome inhibitor, where the maximum level of cytosolic hnRNP A1 was observed at 6 h (Figure 7D). We also evaluated whether both tunicamycin and thapsigargin treatment could affect the nuclear and cytosolic distribution of hnRNP A1 in primary rat hepatocytes. Results confirmed that the induction of ER stress by tunicamycin and thapsigargin treatment triggered the translocation of hnRNP A1 from nuclei to the cytosol in rat hepatocytes (Supplementary Figures S1A and S1B, Lower panel at http://www.biochemj.org/bj/449/bj4490543add.htm). To investigate whether the amount of hnRNP A1 bound to the SREBP-1a IRES increases under tunicamycin- or thapsigargin-induced ER stress, affinity chromatography experiments were performed by using the biotinylated RNA corresponding to the IRES of SREBP-1a mRNA as a probe, and cytoplasmic extracts from ER-stressed HepG2 cells or rat hepatocytes. The cytoplasmic extracts from the respective untreated cells were used as controls. The RNA–protein complexes were then pooled with streptavidin-coated paramagnetic beads. The amount of hnRNP A1 bound to the SREBP-1a IRES in control and in ER-stressed cells was analysed by Western blotting using the specific monoclonal antibody. The biotinylated EMCV RNA probe was used as negative control. Results showed that the amount of cytosolic hnRNP A1 bound to the SREBP-1a probe was higher in both tunicamycin- and thapsigargin-treated cells with respect to control cells (Figure 7E, lanes 3 and 4 and Supplementary Figure S1C, lanes 3 and 4). In contrast, hnRNP A1 was not detected when the EMCV RNA probe was used (Figure 7E, lanes 1 and 2 and Supplementary Figure S1C, lanes 1 and 2). Thus all these data indicated that ER stress triggers hnRNP A1 shuttling from the nuclei to the cytosol where the protein can bind to the SREBP-1a 5′-UTR inducing SREBP-1a IRES activity.
HFD affects the subcellular distribution of hnRNP A1 in mice
ER stress is a tract commonly found in metabolic disorders, such as obesity . Therefore we addressed the question of whether the ER stress could affect the nuclear and cytosolic hnRNP A1 distribution in liver from lean compared with obese animals. Mice fed on a diet containing a low or high content of fat were chosen as lean or obese animal models. For this purpose, the mice were divided into two groups; the SD group received a diet with 4.3% fat, whereas the HFD group received a diet containing 24% fat. After 24 days, livers from SD and HFD mice were collected, and the nuclear extracts were prepared. To test whether ER stress was induced in liver of mice fed on a HFD, the level of XBP-1s was quantified by Western blotting in both SD and HFD nuclear extracts. In fact, the active isoform of the transcription factor XBP-1 (XBP-1s) represents a well-known marker of ER stress. Results showed that XBP-1s was found in the nuclear extract from the HFD group, but not in that of the SD group (Figure 8). Next, the levels of SREBP-1 and hnRNP A1 were analysed in nuclear extracts from the SD and HFD groups. Results showed that the level of the precursor and the mature form of SREBP-1 was higher in HFD mouse nuclei than in the SD mouse nuclei. By contrast, the level of hnRNP A1 decreased in HFD mouse nuclei with respect to SD mouse nuclei (Figure 8).
Effect of HFD-induced ER stress on the nuclear level of hnRNP A1 in mice
In previous years, increasing amounts of evidence suggest that ER stress can play a crucial role in lipid metabolism. A link between liver diseases such as steatosis and steatohepatitis and the ER stress response has been demonstrated [1–3,10]. It has been shown that UPR activation triggers the proteolytic cleavage of SREBP-1c and SREBP-2 [3,44].
In a previous study, an IRES element has been found in the 5′-UTR of the SREBP-1a mRNA . We showed that this IRES mediates the cap-independent translation of SREBP-1a mRNA, and its activity increased in response to either serum starvation or tunicamycin-induced ER stress . Results the present study clearly indicate that the ITAF hnRNP A1 specifically interacts with the 5′-UTR of SREBP-1a mRNA, increasing the SREBP-1a IRES activity. RNA affinity chromatography experiments carried out by using two biotinylated RNAs, corresponding to the leader region of SREBP-1a and EMCV mRNA respectively, resulted in a different pattern of bound proteins, suggesting that each RNA was able to bind to a specific class of proteins (Figure 1). Among the proteins bound to the SREBP-1a RNA, a 34-kDa protein was identified as hnRNP A1. hnRNP A1 belongs to the hnRNP family and it is involved in several aspects of nucleic acid metabolism . hnRNP A1 is a nucleo–cytoplasmic shuttling protein associated with poly(A)+ mRNA in the cytoplasm . hnRNP A1 belongs to the class of ITAFs that act as translational regulators of specific mRNAs . Data from UV cross-linking of HepG2 nuclear extracts or purified hnRNP A1 with 32P-labelled SREBP-1a IRES RNA, followed by immunoprecipitation with an anti-hnRNP A1 antibody, supported the hypothesis that hnRNP A1 is an ITAF interacting with the SREBP-1a 5′-UTR (Figures 2A and 2B). Competition with unlabelled RNAs corresponding to either the SREBP-1a or c-myc 5′-UTR (Figure 2C, lanes 2–5) or to nucleotide −189/−70 and nucleotide −189/−126 of the SREBP-1a 5′-UTR (Figure 3B, lanes 2–4) strongly impaired the binding of hnRNP A1 to the 32P-labelled SREBP-1a RNA. By contrast, unlabelled RNA competitors corresponding to the EMCV 5′-UTR (Figure 2C, lanes 6 and 7) or to nucleotide −125/−1 and nucleotide −69/−1 of the SREBP-1a 5′-UTR (Figure 3B, lanes 5 and 6) were not able to affect the binding of hnRNP A1 to the 32P-labelled SREBP-1a RNA. Overexpression of hnRNP A1 led to an increase in FL activity but not in RL activity in HepG2 cells transfected with the pRS1aF, suggesting a role for hnRNP A1 in the induction of the SREBP-1a IRES activity (Figure 4B). The specificity of this effect was supported by analogous transfection experiments performed with pRc-MycF and pREF constructs (Figure 4B) used as positive and negative controls respectively [39,40]. Furthermore, SREBP-1a IRES activity was not induced by the overexpression of HuR and La protein (Figure 4C), which are well-known ITAFs [41,42]. Conversely, siRNA-mediated knockdown of the hnRNP A1 gene inhibited the SREBP-1a IRES activity, as shown by co-transfection experiments with the pRS1aF construct together with hnRNP A1 siRNA (Figure 4F). The observed increase in FL activity in hnRNP A1-overexpressing cells was not due to a FL monocistronic mRNA, which could be produced either from a cryptic promoter or an aberrant splicing of the bicistronic mRNA, as demonstrated by Northern blot analysis performed with a 32P-labelled probe corresponding to the FL ORF (Figure 4D).
The endogenous SREBP-1a level was increased or reduced in hnRNP A1-overexpressing or hnRNP A1-knockdown cells respectively, with respect to control cells (Figure 5A). By contrast, SREBP-1a mRNA abundance was not affected in the same conditions (Figure 5B). Similar results were obtained when the same experiments were performed with primary rat hepatocytes (Figures 5C and 5D). Thus all these data indicated that hnRNP A1 bound specifically to the first 64 nucleotides of the leader region of SREBP-1a mRNA, inducing SREBP-1a translation in Hep G2 cells and in rat hepatocytes. By in silico analysis of the leader sequence of SREBP-1a mRNA, we observed that the −189/−70 region is highly conserved among the mammalian species (results not shown). This might support a physiological function of this region, such as the IRES-mediated translation of SREBP-1 mRNA.
In a previous study, we demonstrated that SREBP-1a expression increased after UPR induction in tunicamycin-treated HepG2 cells, through the cap-independent translation of SREBP-1a mRNA . In the present study, we investigated the role of hnRNP A1 on the tunicamycin-induced activation of SREBP-1 expression. In agreement with previous results , both pSREBP-1 and nSREBP-1 content increased in ER-stressed HepG2 cells and in rat hepatocytes (Figures 6A and 6D) and this effect was ascribed to an enhanced SREBP-1a mRNA translation through a cap-independent mechanism, as demonstrated by transfection experiments with the bicistronic construct pRS1aF (Figures 6C and 6F).
Tunicamycin-induced activation of both the SREBP-1a protein content and SREBP-1a IRES activity did not occur in hnRNP A1-knockdown cells (Figures 6A and 6D). Tunicamycin induction of SREBP-1a protein content and SREBP-1a IRES activity were rescued in Hep G2 cells (Figures 6A and 6C) and in rat hepatocytes (Figures 6D and 6F) transfected with both hnRNPA1 siRNA and pcDNA-hnRNP A1.
How does ER stressor treatment enhance the hnRNP A1-mediated SREBP-1a protein expression through IRES-dependent translation? hnRNP A1 is a nucleo–cytoplasmic shuttling protein that regulates gene expression, acting on mRNA metabolism and translation. The cytoplasmic redistribution of hnRNP A1 has been observed during viral infection and cellular stress, such as osmotic shock or UV-C irradiation [40,47], and it is promoted by activation of the p38–Mnk1/2 pathway [48,49]. Thus we hypothesized that a specific interaction of hnRNP A1 with the SREBP-1a 5′-UTR is crucial for the increase of SREBP-1a IRES activity in both HepG2 cells and in primary rat hepatocytes treated with ER stressors. Indeed, Western blotting experiments showed that both tunicamycin- and thapsigargin-induced ER stress enhanced the nucleo–cytoplasmic relocation of hnRNP A1 both in HepG2 cells (Figure 7A–D) and in rat hepatocytes (Supplementary Figure S1), being more evident after treatment of the cells with MG-132, an inhibitor of the ubiquitin–26S proteasome system. Moreover, the binding of hnRNP A1 to the SREBP-1a 5′-UTR probe increased in ER-stressed cells with respect to control cells (Figure 7E and Supplementary Figure S1).
It is noteworthy that the endogenous SREBP-1 content decreased in hnRNP A1-knockdown cells (Figure 5A), but did not in hnRNP A1-knockdown cells treated with tunicamycin (lane 3 in Figures 6A and 6D), when compared with the corresponding control cells (lane 1 in Figures 6A and 6D). This discrepancy could be explained by supposing that, besides hnRNP A1, other protein factors might bind to the 5′-UTR of SREBP-1 mRNA, regulating its expression. Further studies will be performed to investigate the role of other ITAFs on the regulation of SREBP-1a expression.
Multiple mechanisms for ER stress-induced steatosis have been proposed, such as the decreased expression of Insig-1 protein, a negative regulator of lipid synthesis that retains the SREBP–SCAP complex in the ER . Tumour necrosis factor-α, a cytokine that induces UPR, provokes cleavage and activation of SREBP-1 in ethanol-exposed cells via a caspase-dependent pathway . In the present paper, we show that the ER stress-induced SREBP-1a expression can be mediated by an increased binding of hnRNP A1 to the SREBP-1a IRES element.
It is well known that ER stress is implicated in several chronic metabolic diseases, including insulin resistance, diabetes and obesity, in which a high expression of SREBP-1 has been observed . Moreover, SREBP-1 is abundant in nuclei from liver of obese animal models, such as ob/ob mice characterized also by ER stress pathway activation . Thus an interesting question that can be addressed is whether ER stress, in liver from obese animals, triggers an efficient translation of SREBP-1a mRNA by inducing hnRNP A1 translocation from nuclei to the cytosol. To this aim, the nuclear level of SREBP-1 and the subcellular localization of hnRNP A1 has been analysed in liver from SD- and HFD-fed mice. Results showed that the level of SREBP-1 increased whereas hnRNP A1 decreased in nuclei from liver of HFD-fed mice with respect to SD-fed mice (Figure 8).
Of note, SREBP-1c represents the main isoform expressed in liver, whereas SREBP-1a is predominant in cultured cell lines . Taking into account these differences, the findings obtained in ER-stressed primary rat hepatocytes (Figures 5C and 5D) as well as in mouse liver (Figure 8) led us to speculate that the increase in SREBP-1 expression is likely to be mainly caused by an increased translation of SREBP-1c, rather than of SREBP-1a. Thus an interesting question is whether translation of mRNA for SREBP-1c as well as for SREBP-2 could be induced by ER stress through a similar mechanism described for SREBP-1a mRNA. Even though a cryptic promoter makes characterization of the SREBP-1c 5′-UTR difficult , the presence of an IRES in the leader region of SREBP-1c mRNA cannot be ruled out. Further investigations in vivo can shed light on the mechanism of ER stress-mediated induction of SREBP-1 expression, providing a possible therapeutic target for metabolic disorders.
Dulbecco’s modified Eagle’s medium
fetal bovine serum
heterogeneous nuclear ribonucleoprotein
human antigen R
internal ribosome entry site
IRES trans-acting factor
liver X receptor
N-terminal half of SREBP
open reading frame
polyunsaturated fatty acid
SREBP cleavage-activating protein
small interfering RNA
unfolded protein response
X-box-binding protein 1
Fabrizio Damiano and Luisa Siculella conceived and designed the experiments, analysed the data and wrote the paper. Alessio Rochira, Romina Tocci, Antonio Gnoni and Simone Alemanno helped to perform the experiments. All authors discussed the results and commented on the paper.
We thank A. Willis (MRC Toxicology Unit, University of Leicester, Leicester, U.K.) for providing the pRc-mycF, R.T. Hay (Wellcome Biocentre, University of Dundee, Dundee, U.K.) for pcDNA-SV5-hnRNPA1, E.K. Spicer (Medical University of South Carolina, Charleston, SC, U.S.A.) for pET21a-HuR; and M. Serino and C. Touriol (Institut National de la Santé et de la Recherche Médicale, Toulouse, France) for pCREL. We thank S. Gaetani (Sapienza, University of Rome, Rome, Italy) for providing the liver samples from lean and obese mice.
The research discussed in the present study was supported by grants from Apulia Region PS 101 to F.D. and L.S.