Liver-specific loss of glucose-regulated protein 78 perturbs the unfolded protein response and exacerbates a spectrum of liver diseases in mice§


  • Potential conflict of interest: Nothing to report.

  • This work was supported by the National Institutes of Health [mainly through grants R01AA018846 and R01AA018612 (to Cheng Ji) and partly through grant R01CA27607 (to Amy S. Lee) and grants P50AA11999, P30DK48522, and R01AA014428 (to Neil Kaplowitz and Cheng Ji)].

  • §

    Cheng Ji is the principal investigator of this study. In Ji's laboratory, Mo Yin Lau is a graduate student, and Eddy Kao is a research associate.


The endoplasmic reticulum (ER) chaperone protein glucose-regulated protein 78 (GRP78)/binding immunoglobulin protein is a master regulator of ER homeostasis and stress responses, which have been implicated in the pathogenesis of metabolic disorders. By applying the locus of X-over P1–cyclization recombination strategy, we generated mice with liver-specific GRP78 loss. Our studies using this novel mouse model revealed that liver GRP78 was required for neonatal survival, and a loss of GRP78 in the adult liver greater than 50% caused an ER stress response and dilation of the ER compartment, which was accompanied by the onset of apoptosis. This suggested the critical involvement of GRP78 in maintaining hepatocyte ER homeostasis and viability. Furthermore, these mice exhibited elevations of serum alanine aminotransferase and fat accumulation in the liver, and they were sensitized to a variety of acute and chronic hepatic disorders by alcohol, a high-fat diet, drugs, and toxins. These disorders were alleviated by the simultaneous administration of the molecular chaperone 4-phenylbutyrate. A microarray analysis and a two-dimensional protein profile revealed major perturbations of unfolded protein response targets, common enzymes/factors in lipogenesis, and new factors possibly contributing to liver steatosis or fibrosis under ER stress (e.g., major urinary proteins in the liver, fatty acid binding proteins, adipose differentiation-related protein, cysteine-rich with epidermal growth factor–like domains 2, nuclear protein 1, and growth differentiation factor 15). Conclusion: Our findings underscore the importance of GRP78 in managing the physiological client protein load and suppressing apoptosis in hepatocytes, and they support the pathological role of ER stress in the evolution of fatty liver disease under adverse conditions (i.e., drugs, diet, toxins, and alcohol). (HEPATOLOGY 2011;)

The endoplasmic reticulum (ER) is an essential organelle for protein synthesis, folding and posttranslational modifications, the biosynthesis of lipids and sterols, the metabolism of drugs, and the maintenance of Ca2+ homeostasis. Molecular chaperones in the ER ensure the proper folding and targeting of nascent proteins. Physiological or pathological conditions can stress the ER and trigger an adaptive unfolded protein response (UPR).1-4 The UPR signaling pathways are associated with a variety of disorders in both animal models and patients.1-5

The liver plays a central role in the homeostasis of glucose and lipids. Hepatocytes are rich in ER, which is the site of the synthesis of a large number of secretory and complex lipoproteins. This high level of secretory function renders the liver particularly susceptible to ER stress. The UPR plays pivotal roles in the liver: the maintenance of ER homeostasis under basal conditions and adaptation to fluctuations in nutrient availability. Mounting evidence indicates that ER stress plays an integral role in the pathogenesis of the most commonly encountered liver diseases.1, 3-5 Studies using animal models lacking or overexpressing factors involved in ER stress signaling have revealed that one common feature of these diseases mediated by ER stress is impaired lipid metabolism.1-5 Aberrant lipid accumulation is often an early stage leading to advanced and more irreversible liver injuries such as fibrosis, cirrhosis, and even liver cancer. However, the cause-versus-effect linkage of ER stress to all stages of liver injury, including early and advanced liver disorders, remains poorly understood.

The chaperone protein glucose-regulated protein 78 (GRP78)/binding immunoglobulin protein is a master regulator of ER homeostasis; in addition to facilitating protein folding and assembly, it interacts with all three major UPR sensors: protein kinase-like endoplasmic reticulum kinase (PERK), inositol-requiring kinase 1α (IRE1α), and activating transcription factor 6 (ATF6).6 This interaction allows GRP78 to act as a repressor of activation of the UPR transducers that influence the outcomes of the ER stress response. The induction of GRP78 upon ER stress is a UPR hallmark. The overexpression of GRP78 in the liver protected mice against insulin-induced and ER stress–induced sterol regulatory element binding protein 1c (SREBP1c) activation and hepatic steatosis.7 Thus, GRP78 represents an ideal target for directly testing the role of ER stress in various diseases. However, the global deletion of the Grp78 gene is lethal because it is required for embryonic development.8 In this study, we generated mice with a liver-specific deletion of Grp78 with the locus of X-over P1 (LoxP)–cyclization recombination (Cre) system, and we found that this liver-specific deletion resulted in an ER stress response and a spectrum of spontaneous or stress-aggravated pathological consequences, including fatty liver, insulin resistance, and increased susceptibility to drug-, toxin-, diet-, and alcohol-induced injury and fibrosis.


α-SMA, α-smooth muscle actin; ADRP, adipose differentiation-related protein; Alb, albumin; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ATF, activating transcription factor; CCl4, carbon tetrachloride; C/EBPα, CCAAT/enhancer-binding protein α; ChREBP, carbohydrate responsive element binding protein; Cre, cyclization recombination; CREBH, cyclic adenosine monophosphate responsive element binding protein H; CRELD2, cysteine-rich with epidermal growth factor–like domains 2; Ct, control injected with phosphate-buffered saline; CYP2E1, cytochrome P450 2E1; Derl3, derlin 3; EDEM2, endoplasmic reticulum degradation enhancer mannosidase alpha-like 2; eIF2, eukaryotic translation initiation factor 2; ER, endoplasmic reticulum; ERp57, endoplasmic reticulum protein 57; FABP, fatty acid binding protein; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; Gdf15, growth differentiation factor 15; Gl, glucose injection; GRP, glucose-regulated protein; GST, glutathione S-transferase; H&E, hematoxylin and eosin; HFD, high-fat diet; HIV, human immunodeficiency virus; HOMA-IR, homeostasis model assessment of insulin resistance; In, insulin injection; INSIG, insulin-induced gene; IR, insulin receptor; IRE, inositol-requiring kinase; IRS, insulin receptor substrate; IRS1-ser, phosphorylated serine of insulin receptor substrate 1; IRS1-tyr, phosphorylated tyrosine of insulin receptor substrate 1; JNK, c-Jun N-terminal kinase; L-FABP, liver fatty acid binding protein; LGKO, liver-specific glucose-regulated protein 78 knockout; LoxP, locus of X-over P1; MMP2, matrix metalloproteinase 2; mRNA, messenger RNA; MUP, major urinary protein; NF-κB, nuclear factor kappa B; Nupr1, nuclear protein 1; ORP150, oxygen-regulated protein 150; p-eIF2α, phosphorylated eukaryotic translation initiation factor 2α; p-IRE, phosphorylated inositol-requiring kinase; p-JNK, phosphorylated c-Jun N-terminal kinase; p-PERK, phosphorylated protein kinase-like endoplasmic reticulum kinase; PBA, 4-phenylbutyrate; PCR, polymerase chain reaction; PDI, protein disulfide isomerase; PERK, protein kinase-like endoplasmic reticulum kinase; PF, pair-fed control; PI, protease inhibitor; PPAR, peroxisome proliferator-activated receptor; RFU, relative fluorescent unit; SREBP, sterol regulatory element binding protein; sXBP1, spliced X-box binding protein 1; TGF-β, transforming growth factor β; tLGKO, transgene liver-specific glucose-regulated protein 78 knockout; TUNEL, terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling; UPR; unfolded protein response; USP, ubiquitin-specific peptidase; uXBP1, unspliced X-box binding protein 1; WT, wild type; XBP1, X-box binding protein 1.

Materials and Methods

Animal Breeding and Experiments.

Mice with a liver-specific Grp78 deletion were created with the LoxP-Cre strategy (see the supporting information). Polymerase chain reaction (PCR) genotyping with tail or liver genomic DNA was performed to distinguish Grp78 alleles of wild-type (WT), floxed, and deleted mice.9 The presence and quantity of the albumin (Alb)-Cre transgene were determined by duplex quantitative PCR with Cre-specific primers (Supporting Table 1). Male mice between the ages of 4 and 12 weeks were selected for acute or chronic alcohol or drug administration. A liquid diet that was 4.3% (vol/vol) alcohol (#710301, Dyets, Inc., Bethlehem, PA) or an isocaloric control diet was pair-fed for 6 to 8 weeks. For high-fat diet (HFD) experiments, animals were fed the American Institute of Nutrition 93G formula modified with a purified HFD (#180529, Dyets) for 6 weeks. The human immunodeficiency virus (HIV) protease inhibitors (PIs) ritonavir and lopinavir (10-20 mg/kg of body weight), which were mixed with the liquid diet from Dyets containing alcohol (4.8 g/kg of body weight), were gavaged into the animals after 10 hours of fasting. In some experiments, 4-phenylbutyrate (PBA; 1 g/kg of body weight) was mixed with the control liquid diet, and the mixture was gavaged into the animals 1 hour before the alcohol and drug treatment. Carbon tetrachloride (CCl4; 0.8 μL/g of body weight) that was dissolved in corn oil (1:5) with or without PBA was intraperitoneally injected into the animals twice per week for 6 weeks. For an examination of insulin signaling, mice were fasted for 6 hours after the start of the light cycle and were intraperitoneally injected with glucose (2.5 g/kg) or insulin (1.0 U/kg). Blood glucose levels were determined with a diabetes monitoring kit (Roche Diagnostics, IN). Insulin resistance was assessed with the homeostasis model assessment of insulin resistance (HOMA-IR) as follows10:

equation image

Littermate controls were used for all experiments, and age matching was performed for each experiment whenever it was possible. All animals were treated in accordance with Guide for the Care and Use of Laboratory Animals, and their treatment was approved by a local committee.

Pathological Parameters, Molecular Analysis, and Liver Histology.

The serum alanine aminotransferase (ALT), aspartate aminotransferase (AST), and plasma homocysteine measurements, the liver lipid extraction and analysis, the primary hepatocyte isolation, the extractions and analysis of RNA and whole cell or nuclear proteins, and the liver histology by hematoxylin and eosin (H&E), Sirius red, and terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling (TUNEL) staining were described previously.11, 12 The primers are listed in Supporting Table 1. Histological changes were confirmed by a pathologist blinded to the genotypes. The quantitation of Sirius red staining was performed with ImageJ software from the National Institutes of Health.

Statistical Analysis.

Values are expressed as means and standard errors of the mean unless otherwise indicated. Statistical analyses were performed with the Student t test for paired data when each group of animals were from one litter and for unpaired data when each group of animals were from two or more litters or with an analysis of variance for the comparison of multiple groups. P values < 0.05 were considered statistically significant. The supporting information includes information on breeding, insulin detection, antibodies, immunoblotting, Phos-tag gel use, proteasome activity, electron microscopy, DNA microarrays, two-dimensional difference gel electrophoresis, and mass spectrometry.


Pathological Consequences of the Liver-Specific Deletion of Grp78.

Mice with a liver-specific Grp78 deletion [i.e., Grp78f/f Alb-CreTg/0 or liver-specific glucose-regulated protein 78 knockout (LGKO) mice] were generated (Supporting Fig. 1A,B). The liver-specific deletion was detected in genomic DNA from the livers of LGKO mice but not from their kidneys (Supporting Fig. 1C). The GRP78 protein level was reduced by 35% to 70% between the ages of 30 and 90 days in the LGKO mouse liver versus the WT mouse liver (Fig. 1A,B). The protein level was reduced by 15% to 25% in the GRP78 heterozygous [i.e., Grp78f/w Alb-CreTg/0 (WK)] mice in comparison with the WT mice between the ages of 30 and 90 days (Supporting Fig. 1D). The immunohistochemistry of liver tissue with anti-GRP78 antibodies confirmed the decrease in the liver GRP78 levels (Supporting Fig. 1E). Some of the remaining brown spots were identified as possible stromal cells in which Alb-Cre was not active. The viability rate for primary hepatocytes from LGKO mice was 68%, whereas the viability rate for primary hepatocytes from WT or WK mice was greater than 90%. Interestingly, a complete loss of GRP78 in the primary hepatocytes of LGKO mice was not detected, and the GRP78 protein level in the LGKO hepatocytes was 27% of the level in WT hepatocytes; however, the GRP78 protein level in the WK hepatocytes was 74% of the level in WT hepatocytes (Supporting Fig. 1F).

Figure 1.

Liver-specific loss of GRP78/binding immunoglobulin protein and liver injury. (A) Western blots showing the liver-specific loss of the GRP78 protein. (B) Graph of the levels of GRP78 in LGKO and WT mice. (C) H&E staining (×100) of fat accumulation in WT and LGKO mice. (D) Hepatic apoptosis (brown spots) according to immunohistochemistry (×100) with anti–active caspase-3 antibodies. (E) Serum ALT and AST levels. The values are shown as means and standard deviations.

The LGKO mice were viable and smaller than the WT and WK mice but otherwise appeared grossly normal (Supporting Figs. 1B and 2A,B). However, under unchallenged conditions, mild to moderate fatty livers, apoptosis, and necroinflammation were observed in 12 of 21 LGKO mice killed at 90 days (Fig. 1C,D and Supporting Fig. 2C). Serum ALT and AST levels in the LGKO mice were significantly elevated (Fig. 1E). In electromicrographs, ER in the LGKO hepatocytes was dilated, vesiculated, and accompanied by lipid inclusions (Supporting Fig. 2D); this indicated ER damage from the Grp78 deletion.

Newborn pups with homozygous Grp78 floxed alleles and full copies of the Cre transgene [i.e., Grp78f/f Alb-CreTg/Tg (tLGKO) mice] usually died within 1 week after birth. The liver GRP78 protein levels in the tLGKO mice (32% of the levels in the WT mice) were lower than those in the LGKO mice (83% of the levels in the WT mice), whereas the Cre levels were higher in the tLGKO mice (38% of the levels of the Cre transgenic adults) versus the LGKO mice (18% of the levels of the Cre transgenic adults; Supporting Fig. 3). Severe hepatic inflammation and massive cell death (as many as 30% of the hepatocytes) were observed in the tLGKO mice (Supporting Fig. 3).

Molecular Validation of the Hepatic ER Stress Response in LGKO Mice.

An analysis using complementary DNA microarrays revealed that the expression of 450 of 18,833 transcripts was altered in the LGKO liver. Molecular chaperones [GRP94, oxygen-regulated protein 150 (ORP150), protein disulfide isomerase (PDI), inhibitor of the interferon-induced double-stranded RNA-activated protein kinase (p58IPK), J-domain-containing PDI-like protein (ERdj5), and calreticulin], ubiquitin and protein degradation factors [ubiquitin-specific peptidase 4 (Usp4), Usp18, homocysteine-induced ER protein 1, ubiquitin protein ligase E3B, endoplasmic reticulum degradation enhancer mannosidase alpha-like 2 (EDEM2), and derlin 3 (derl3)], transcription factors regulating apoptosis [nuclear protein 1 (Nupr1), C/EBP homologous protein (CHOP), tribbles homolog 3, growth arrest and DNA damage-inducible gene 45 (Gadd45), and forkhead box O], and some nuclear factor kappa B (NF-κB)–targeted genes (interleukin-6 receptor α, complement component 1q/tumor necrosis factor–related protein 1, and tumor necrosis factor receptor 1) were among the genes with increased expression, whereas the expression levels of B cell lymphoma 2–interacting killer-like and the cyclic adenosine monophosphate responsive element binding protein H (CREBH)–targeted gene hepcidin 2 were reduced in the LGKO liver (Supporting Table 2).

Proteomic two-dimensional difference gel electrophoresis analysis identified alterations in 35 of 2350 protein spots in the LGKO liver (Supporting Fig. 3F). The proteins with altered expression included GRP94, ORP150, all isoforms of PDI, catalase, glutathione S-transferase μ1 (GSTμ1), GSTπ1, ubiquinol cytochrome C reductase, cytochrome b5, glyoxalase 1, major urinary proteins (MUPs), adipose differentiation-related protein (ADRP), farnesyl diphosphate synthase, and liver fatty acid binding protein (L-FABP; Supporting Table 3). This confirmed the constitutive UPR and impaired energy metabolism at protein levels in the LGKO liver.

Using immunoblotting, we detected the altered expression of GRP94, PDI, IRE, phosphorylated eukaryotic translation initiation factor 2α (p-eIF2α), ADRP, L-FABP1, and MUP1 and the slight activation of NF-κB and CREBH in the LGKO liver (Fig. 2A). The phosphorylation of IRE and PERK and the unconventional splicing of X-box binding protein 1 (Xbp1) were observed, and this was partially reduced by the administration of PBA (Fig. 2B,C).

Figure 2.

UPR and insulin resistance in LGKO mice. (A) Western blots of whole cell or nuclear proteins. (B) Western blots of Phos-tag gel used to detect phosphorylation. LGKO/PBA mice received an intraperitoneal injection of PBA after they were weaned. Each lane represents an individual mouse. (C) Unconventional splicing of Xbp1 mRNA by a reverse-transcription PCR assay. (D) Comparison of the HOMA-IR index in WT and LGKO mice (*P < 0.05, n = 6). (E) Immunoblots of mouse liver proteins. Abbreviations: Ct, control injected with phosphate-buffered saline; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; Gl, glucose injection; In, insulin injection; IRS1-ser, phosphorylated serine of insulin receptor substrate 1; IRS1-tyr, phosphorylated tyrosine of insulin receptor substrate 1; p-IRE, phosphorylated inositol-requiring kinase; p-PERK, phosphorylated protein kinase-like endoplasmic reticulum kinase; uXbp1, unspliced X-box binding protein 1.

Insulin Resistance in LGKO Mice.

The HOMA-IR index increased more than 2-fold in the LGKO mice versus the WT mice (Fig. 2D). The immunoblotting of select proteins involved in insulin signaling indicated that the protein levels of phosphorylated c-Jun N-terminal kinase 1 (p-JNK1), p-JNK2, and phosphorylated serine 307 for insulin receptor substrate 1 (IRS1) were increased in the LGKO mice versus the WT mice with or without a glucose or insulin injection (Fig. 2E). In contrast, the levels of phosphorylated tyrosine 989 for IRS1 were reduced in the livers of the LGKO mice versus the WT mice after an injection of glucose or insulin. The insulin receptor (IR) protein was not changed in either genotype. These results indicate that the liver-specific loss of GRP78 impairs insulin signaling.

Loss of Grp78 Exacerbates Alcohol-Induced or HFD–Induced Fatty Liver Injury.

Chronic intragastric alcohol infusions are associated with hyperhomocysteinemia, ER stress responses, and fatty liver injury.4, 5, 11 To test directly whether the ER stress response could contribute to alcohol-induced liver injury, we orally fed a liquid alcohol diet to LGKO and WT mice. No significant changes in the plasma homocysteine levels between pair-fed and alcohol-fed LGKO and WT mice were detected with a reduced alcohol dose for 45 days (data not shown). The ALT and liver triglyceride levels were 19.3 ± 2.1 U/L and 43 ± 4.6 mg/g of protein, respectively, for pair-fed WT mice and 37.9 ± 3.4 U/L and 97.5 ± 8.2 mg/g of protein, respectively, for pair-fed LGKO mice. In response to alcohol feeding, the ALT level increased by 100% in the LGKO mice versus the WT mice (Fig. 3A-D). The liver triglyceride level increased 3-fold in the LGKO mice versus the WT mice. Cell death, which was revealed by TUNEL-positive hepatocytes, was significantly increased in the alcohol-fed LGKO mice, but it was not increased in the alcohol-fed WT mice. These data suggest that a loss of GRP78 increases the susceptibility to alcohol-induced fatty liver and liver injury.

Figure 3.

Liver-specific deletion of GRP78 exacerbates alcohol-induced ER stress and fatty liver injury. (A) Serum ALT levels, (B) liver triglyceride levels, and (C) cell death are compared for WT mice that were fed ethanol (WT_E) and LGKO mice that were fed ethanol (LGKO_E). Dashed lines represent the levels for pair-fed controls. The animals were 4 to 6 weeks old (*P < 0.05 and **P < 0.01 versus WT mice, n = 6). (D) TUNEL-stained liver sections (×200). (E) Western blots of nuclear protein extracts for activated nSREBP1c and nChREBP, sXbp1, activated ATF6, and C/EBPα with lamin A/C as a loading control and western blots of whole protein extracts for p-eIF2α with β-actin as a loading control; (F) Diagram of the impact of Grp78 deletion and ER stress factors on alcohol-induced lipogenesis.

We examined multiple transcription and lipogenic factors that are potentially involved in ER stress–induced lipogenesis, including SREBP1c, carbohydrate responsive element binding protein (ChREBP), Xbp1, ATF6, Gadd34, ATF4, CCAAT/enhancer-binding protein α (C/EBPα), CHOP, insulin-induced gene 1 (Insig1), Insig2, cytochrome P450 2E1 (CYP2E1), peroxisome proliferator-activated receptor α (PPARα), PPARγ, glucokinase, stearoyl coenzyme A desaturase 1, and fatty acid synthase (Fig. 3E and Supporting Fig. 4A). Only slight changes in the transcription factors and the phosphorylation of eukaryotic translation initiation factor 2α (eIF2α) were detected in the WT mice that were fed a low dose of alcohol, and this was consistent with the mild fat accumulation observed previously. However, in the LGKO liver, most of these factors were altered. Alcohol feeding had either enhancing or reducing effects on the tested ER stress factors (Fig. 3E and Supporting Fig. 4). In LGKO mice, alcohol further increased the expression of nuclear SREBP1c, ChREBP, and C/EBPα and increased messenger RNA (mRNA) levels of Insig2. Alcohol decreased the levels of ATF6, Gadd34, Insig1, the ATF6 transcriptional target endoplasmic reticulum protein 57 (ERp57), and Derl3. Alcohol had no further effects on spliced X-box binding protein 1 (sXbp1), CHOP, or ATF4, the levels of which were already increased after the GRP78 deletion. Alcohol had no effects on PPARα. CYP2E1 levels were increased in response to alcohol, but the GRP78 deletion had no apparent effect on CYP2E1 expression. The levels of stearoyl coenzyme A desaturase 1, fatty acid synthase, and PPARγ were increased by alcohol or the GRP78 deletion alone and were increased further in alcohol-fed LGKO mice. This indicates that interplay between ER stress and alcohol feeding aggravates fat accumulation in the liver (Fig. 3F).

To determine whether the liver Grp78 deletion worsens nonalcoholic steatosis, we fed the mice an HFD. The HFD induced moderate fatty liver injury (a 1.5-fold increase in LGKO mice versus pair-fed controls) within 6 weeks, and there were no significant differences in the HFD-induced liver injuries of WT and WK mice (Fig. 4A-C). GRP78 and GRP94 were increased in the WK mice in response to the HFD, which may have compensated for the heterozygous loss of Grp78 in the liver (Fig. 4D). A slight activation of ATF6 was detected in LGKO mice not fed the HFD, and an apparent activation was observed in both WT mice and LGKO mice that were fed the HFD (Fig. 4E). In comparison with HFD-fed WT mice, the HFD doubled both hepatic triglyceride and ALT levels in LGKO mice, and this was accompanied by greater GRP94 induction and eIF2α phosphorylation, which indicated a severe ER stress response in HFD-fed LGKO mice.

Figure 4.

Effects of HFD feeding on WT, WK, and LGKO mice. (A) Serum ALT levels. (B) Liver triglyceride levels. (C) H&E staining of liver tissues showing fatty livers (×200). (D) Western blots showing liver protein levels of GRP78 and GRP94. (E) Western blots of liver proteins showing the activation of ATF6α and the phosphorylation of eIF2α (*P < 0.05, **P < 0.01, and ***P < 0.005 versus PFs, n = 3). Abbreviation: PF, pair-fed control.

A Loss of Grp78 Sensitizes Mice to HIV PI–Induced ER Stress and CCl4-Induced Fibrosis.

HIV-infected patients receiving anti-HIV therapeutics with HIV PIs often concomitantly consume or abuse alcohol.13 To address whether HIV PIs alone or in combination with alcohol trigger ER stress in the mouse liver, we treated mice with alcohol and/or ritonavir and lopinavir by gavage. Neither liver injury (according to serum ALT and AST levels) nor an ER stress response was detected when the animals were treated with the acute administration of alcohol alone or with a combination of ritonavir and lopinavir for 16 hours (data not shown). However, the combined treatment of alcohol and HIV PIs caused significant increases in plasma ALT levels (Fig. 5A) and the activation of CHOP, ATF4, and sXbp1 in both WT and LGKO mice (Fig. 5B); this was comparable to the response of WT mice injected with tunicamycin for 24 hours. In response to the combined treatment, the ALT values and ER stress responses were greater in LGKO mice versus WT mice. A pretreatment with PBA partially reduced the alcohol-induced and HIV PI–induced ER stress response and decreased the elevated ALT levels by more than 50% in both WT and LGKO mice. In addition, an accumulation of ubiquitinated proteins was detected in LGKO mice but not in WT mice treated with alcohol plus HIV PIs. Alcohol and HIV PIs reduced proteasome activity by 15% in WT mice and by more than 50% in LGKO mice. The PBA treatment restored proteasome activity in both WT and LGKO mice (Fig. 5C).

Figure 5.

Effects of Grp78 deletion on acute alcohol administration–induced and HIV PI–induced ER stress and liver injury. Control animals were gavaged with an isocaloric liquid diet only; treated animals were gavaged with a liquid diet containing alcohol and HIV PIs (ritonavir and lopinavir) with or without PBA. (A) Comparison of ALT levels in WT and LGKO mice. (B) Immunoblots of ER stress–related transcription factors in nuclear protein extracts and ubiquitinated proteins in whole cell protein extracts. WT mice treated with tunicamycin for 24 hours (WTu mice) served as positive controls for the ER stress response in vivo. (C) Comparison of the 20S proteasome activity in WT and LGKO mice. The data are presented as RFUs and were recorded with a fluorometer (*P < 0.05 and **P < 0.01 versus WT mice, #P < 0.05 and ##P < 0.01 versus WT mice or LGKO mice with PBA, n = 5). Abbreviation: RFU, relative fluorescent unit.

To determine the effects of the liver-specific Grp78 deletion on progressive stages of liver injury, we examined fibrotic changes in LGKO and WT mice. Spontaneous mild fibrotic changes were observed in Sirius red–stained liver tissues of 2 of 10 LGKO mice, but this was not detected in WT mice (Fig. 6A and Supporting Fig. 5A). A chronic CCl4 treatment induced fibrotic changes in both WT and LGKO mice. However, the fibrosis was greater in LGKO mice versus WT mice. Quantitatively, the red-stained collagen was increased 15-fold in LGKO mice versus WT mice without CCl4 (Fig. 6A). The collagen deposition was increased by 24-fold in WT mice and by 41-fold in LGKO mice after the chronic CCl4 treatment in comparison with WT mice without CCl4. The levels of type I collagen mRNA in WT and LGKO mice were increased 7.7- and 12.5-fold, respectively, in response to CCl4 (Fig. 6B). There were apparent differences in the expression of select markers of fibrosis between WT and LGKO mice. Without CCl4, the levels of transforming growth factor β (TGF-β), α-smooth muscle actin (α-SMA), and matrix metalloproteinase 2 (MMP2) were increased 1.5- to 2.5-fold in LGKO mice versus WT mice (Fig. 6C). With CCl4, the levels of these markers were increased 2- to 3.5-fold in WT mice with enhanced GRP78 and 3- to 5-fold in LGKO mice. This indicates that the GRP78 deletion worsened CCl4-induced fibrosis. The PBA treatment reduced CCl4-induced fibrosis by more than 50% in LGKO mice, and this was accompanied by the decreased expression of type I collagen mRNA and decreased protein levels of CHOP, TGF-β, α-SMA, and MMP2 (Fig. 6). In reducing CCl4-induced fibrosis, the PBA treatment of WT mice appeared to be not as effective as it was in LGKO mice, and this was indicative of an ER stress contribution. In addition, the mRNA levels of sXbp1 (Fig. 6D and Supporting Fig. 5B), cysteine-rich with epidermal growth factor–like domains 2 (Creld2), Derl3, growth differentiation factor 15 (Gdf15), and Nupr1 were increased in WT mice treated with CCl4 and were increased more in LGKO mice treated with CCl4 (Fig. 6D). Twenty-four hours after a single injection of CCl4, both the ALT level and the rate of cell death increased more in LGKO mice versus WT mice (Supporting Fig. 5C,D). This suggests that repeated and gradual hepatocellular injury led to greater fibrosis in LGKO mice over the course of the CCl4 administration.

Figure 6.

Liver-specific deletion of GRP78 exacerbates CCl4-induced ER stress and liver fibrosis. (A) Fibrotic/cirrhotic changes after CCl4 treatment. Control WT or LGKO mice were injected with corn oil only, and treated WT or LGKO mice were injected with CCl4 mixed with corn oil in a 1:5 ratio or with PBA and CCl4 mixed with corn oil. (B) Expression of collagen α1 according to quantitative PCR. (C) Immunoblots of protein samples from mice that were subjected to chronic CCl4 administration. (D) Reverse-transcription PCR of sXbp1, Derl3, Creld2, Gdf15, and Nupr1 mRNA (*P < 0.05, **P < 0.01, and ***P < 0.005 versus WT mice without CCl4, #P < 0.05 versus the PBA treatment, n = 4). Abbreviation, GAPDH, glyceraldehyde 3-phosphate dehydrogenase.


By deleting GRP78 specifically in the mouse liver, we observed liver injury, which was indicated by elevated serum ALT levels. The LGKO mice with the liver-specific Grp78 deletion developed ER dilatation, hepatic apoptosis, necroinflammation, fatty liver, insulin resistance, and mild fibrosis. In agreement with the literature on the predominant role of GRP78 in the UPR, the loss of GRP78 activated at the molecular level the three branches of the UPR. This was indicated by the increased phosphorylation of IRE1α, PERK, eIF2, c-Jun N-terminal kinase (JNK), and IRS serine and the altered expression of GRP94, ORP150, PDI, CHOP, ATF4, tribbles homolog 3, Gadd34, forkhead box O, interleukin-6 receptor α, complement component 1q, tumor necrosis factor receptor 1, and hepcidin 2, which were involved in the UPR or ER stress response. The loss of GRP78 also affected the ubiquitin pathway and protein degradation because alterations of Usp4, Usp18, ubiquitin protein ligase E3B, EDEM2, and derl3 were detected. Therefore, the pathogenic mechanisms occurring with GRP78 loss could include the following: hepatic cell death mediated by CHOP and JNK; oxidative stress resulting from the altered expression of catalase, GSTμ1, and GSTπ1; inflammation resulting from NF-κB and CREBH activation; impaired insulin signaling due to the abnormal phosphorylation of IRS1; and impaired energy metabolism mediated by ubiquinol cytochrome C reductase, cytochrome b5, and glyoxalase 1. The exact contribution of each of these pathways is not certain at this time. The cell death resulting from the GRP78 deletion may or may not be dependent on ER stress–induced lipogenesis because the early sequence of the two events has been difficult to determine in vivo. However, it is likely that there is interplay between lipogenesis and cell death as the stress continues. In addition, the broad impact of the GRP78 deletion on the UPR and ER stress signaling pathways without any pharmacological ER stress challenge confirms that the liver is sensitive to ER stress, which accompanies and contributes to most forms of liver injury, and adequate levels of GRP78 may be essential for maintaining ER homeostasis and cell health in the liver.

The global deletion of Grp78 is lethal to embryos.8 However, mice with a heterozygous Grp78 deficiency (Grp78W/−) survived; this suggests that at least 50% of the GRP78 protein is required for the early development of animals. Is GRP78 required for liver development and normal function in the adult liver? In embryos, Grp78 expression starts at 3 days after fertilization (E3), and hepatoblasts form at E8.5 when hepatocyte-specific Alb is being expressed.14 In our LoxP-Cre system, the cre gene driven by the Alb promoter should be expressed at E8.5. If we assume that Cre is fully functional and the turnover time for GRP78 is 3 days,15 a loss of the GRP78 protein greater than 90% should occur around the time of delivery (i.e., E21). Live pups would not be delivered if at least 50% of the GRP78 protein is required for survival. However, we obtained live animals with an incomplete deletion of GRP78 because of the variable efficiency of the Cre function, which has been reported.16 LGKO mice gradually lost the GRP78 protein after birth, and pathological consequences were observed when the GRP78 protein loss was greater than 50%. The GRP78 protein levels in the livers of tLGKO mice (Grp78f/f Alb-CreTg/Tg) were less than 30% of the levels in the livers of WT mice at birth, and this resulted in massive hepatic cell death and neonatal lethality. In addition, during the course of this study, a few LGKO mice died of hypoglycemia 4 to 8 months after birth. An increased death rate for LGKO mice was observed 12 months after birth when the GRP78 levels in the dying LGKO mice were usually less than 30% of the levels in the WT littermates; this suggests that the reduction of GRP78 may shorten the life span. Thus, at least 30% of the GRP78 protein is required for liver development, and more than 50% is required for normal function of the adult liver if we assume that the possible adverse effects of a continuous accumulation of the Cre protein are minimal. With respect to the incomplete deletion of GRP78, it is possible that Grp78 is essential for the viability of hepatocytes (this is the case for HeLa cells17) and forces the outgrowth of WT and Grp78 heterozygous cells. This could account for the portion of the GRP78 protein in all hepatocytes rather than the homogeneous reduction of Grp78 in all hepatocytes in the adult liver. It is also likely that the LGKO hepatocytes were sensitized by the substantial reduction of GRP78, and this caused the pathological changes without a complete loss of GRP78. Nevertheless, we were able to generate viable mouse lines (LGKO) with reduced liver GRP78 expression, and this allowed us to use these mice for phenotypic analysis.

The overexpression of GRP78 inhibited steatosis in the livers of obese (ob/ob) mice.7 The GRP78 deletion led to liver fat accumulation in this study. How ER stress regulates lipid metabolism is not fully understood. Emerging evidence indicates that each of the three branches of the UPR signaling pathways has direct molecular effects on lipid synthesis.1-5 Although previous studies collectively revealed crucial roles of the UPR pathways in lipogenesis, no single animal model of ER stress has led to a spontaneous fatty liver under physiological conditions. The fat accumulation in our LGKO model, which is similar to circumstances in human nonalcoholic steatohepatitis/nonalcoholic fatty liver disease, is likely linked to multiple mechanisms. In addition to commonly recognized factors such as sXBP1, active SREBPs, ATF6, Gadd34, C/EBPα, and PPARα, we have found that the altered expression of MUP1, L-FABP, and ADRP might also contribute to ER stress–induced steatosis in LGKO mice. Noticeably, the direct linking of MUPs, fatty acid binding protein (FABP), or ADRP to ER stress–caused steatosis has not been observed in other knockout mouse models of the UPR. FABP and ADRP are factors known to be involved in lipid transport and lipogenesis.18, 19 MUPs are secreted by the liver and excreted into the urine, and recent evidence indicates that circulating MUPs serve as metabolic signals that regulate glucose and lipid metabolism.20 Therefore, the role of these new factors in ER stress–induced steatosis warrants further investigation.

Previous studies by us and other researchers have suggested that alcohol-induced ER stress involves increased levels of homocysteine, which lead to increased levels of S-adenosyl-L-homocysteine in the liver.5, 11 In the present study, no increases in homocysteine were detected with low-level oral alcohol feeding, so the enhanced ER stress and liver injury in the alcohol-fed LGKO mice probably represent the unmasking of a distinct mechanism by which alcohol induces ER stress. This mechanism normally is largely obscured by compensatory changes that are suppressed in LGKO mice. Furthermore, we observed enhanced ER stress and severely fatty livers in LGKO mice that were orally fed low doses of alcohol, whereas the effects were minimal in WT mice that were orally fed low doses of alcohol. With respect to the role of ER stress in alcohol-induced liver injury, our observations imply that alcohol feeding not only enhanced ER stress but also affected ER stress signaling pathways in the LGKO mice. Alcohol enhanced the expression of SREBP and sXbp1 but decreased the expression of Insig1 and ATF6; this was supported by downstream reductions of ERp57, Derl3, and Gadd34, which appeared to be independent of CHOP. All of these may contribute to and/or aggravate lipid accumulation in the liver (Fig. 3F). As for the question of the differential activation of Ire1α, PERK, and ATF6α, we speculate that alcohol metabolites such as acetaldehyde might form adducts differentially with the ER sensors or that unknown epigenetic changes due to alcohol might alter the responses by the sensors.

The liver-specific deletion of GRP78 also led to sensitization to liver injury by drugs such as HIV PIs. HIV PIs are used in highly active antiretroviral therapy. However, the chronic use of HIV PIs is associated with HIV PI–induced ER stress and injury.21 Considering that a significant proportion of HIV-infected patients consume or even abuse alcohol, we tested the effects of alcohol combined with HIV PIs on liver injury. The combination induced more severe ER stress and injury in LGKO mice versus WT mice. PBA ameliorated both ER stress and liver injury, and this confirmed the direct involvement of GRP78 and ER stress in the injurious effects potentiated by alcohol and HIV PIs.

There is no direct evidence linking ER stress to liver fibrosis/cirrhosis, although cirrhotic livers exhibited partial UPR activation in the basal state and full UPR activation after an lipopolysaccharide challenge.22 We observed some increases in fibrosis in LGKO mice under basal conditions, and this was accompanied by increased levels of sXbp1 and CHOP, which were enhanced with a CCl4 challenge. Thus, severe fibrosis developed in LGKO mice but not in WT mice with GRP78 enhancement. The acute administration of CCl4 resulted in greater increases in serum ALT levels and liver necrosis in LGKO mice versus WT mice, and this indicated that the continuously augmented injury in LGKO mice that were chronically challenged with CCl4 promoted the fibrotic changes. The accelerated fibrotic changes in LGKO mice treated with CCl4 were associated with the altered expression of CHOP and Nupr1 (stress response factors),23 Creld2 and Derl3 (emerging mediators in protein quality control in the ER and in the regulation of the onset and progression of various ER stress–associated diseases),24, 25 and Gdf15 (a protein belonging to the TGF-β superfamily with a role in regulating inflammatory and apoptotic pathways in injured tissues and during disease processes).26 In addition, the levels of α-SMA and TGF-β were decreased by the simultaneous injection of PBA. The evidence thus individually or collectively supports a mechanistic role for ER stress in promoting fibrotic/cirrhotic changes in the liver.

In conclusion, the loss of the key molecular chaperone Grp78 directly disturbs ER homeostasis in the liver and causes or sensitizes mice to a variety of acute and chronic hepatic disorders. These findings underscore the importance of the UPR and GRP78 with respect to the physiological client protein load and hepatocyte viability and the potential pathological role of ER stress in the evolution of drug-induced, toxin-induced, alcoholic-induced, and nonalcoholic fatty liver diseases. The LGKO mouse represents a model of impaired ER defense that unmasks an important role for ER stress in these causes of liver disease.


The authors thank the Cell and Tissue Imaging Core, the Cell Culture Core, and the Proteomics Core (University of Southern California Research Center for Liver Diseases) as well as the Doheny Eye Institute Specialized Microscopy Core for technical services. They also thank Ms. Miao Wang for her helpful assistance with the genotyping of the Grp78 floxed mice.