Feedback loop between hepatocyte nuclear factor 1α and endoplasmic reticulum stress mitigates liver injury by downregulating hepatocyte apoptosis

Hepatocyte nuclear factor alpha (HNF1α), endoplasmic reticulum (ER) stress, and hepatocyte apoptosis contribute to severe acute exacerbation (SAE) of liver injury. Here, we explore HNF1α–ER stress-hepatocyte apoptosis interaction in liver injury. LO2, HepG2 and SK-Hep1 cells were treated with thapsigargin (TG) or tunicamycin (TM) to induce ER stress. Carbon tetrachloride (CCl4) was used to induce acute liver injury in mice. Low-dose lipopolysaccharide (LPS) exacerbated liver injury in CCl4-induced mice. Significant apoptosis, HNF1α upregulation, and nuclear factor kappa B (NF-κB) activation were observed in human-derived hepatocytes during ER stress. Knockdown of Rela, NF-κB p65, inhibited the HNF1α upregulation. Following CCl4 treatment ER stress, apoptosis, HNF1α expression and RelA phosphorylation were significantly increased in mice. HNF1α knockdown reduced activating transcription factor 4 (ATF4) expression, and aggravated ER stress as well as hepatocyte apoptosis in vivo and in vitro. The double fluorescent reporter gene assay confirmed that HNF1α regulated the transcription of ATF4 promoter. LPS aggravated CCl4-induced liver injury and reduced HNF1α, and ATF4 expression. Therefore, in combination, HNF1α and ER stress could be mutually regulated forming a feedback loop, which helps in protecting the injured liver by down-regulating hepatocyte apoptosis. Low-dose LPS aggravates hepatocyte apoptosis and promotes the SAE of liver injury by interfering with the feedback regulation of HNF1α and ER stress in acute liver injury.

Knockdown of HNF1A aggravates apoptosis and reduces ATF4 as well as GRP78 expression during ER stress in LO2 cells. To determine the role of HNF1α in hepatocyte apoptosis and ER stress in vitro, LO2 cells were transfected with HNF1A short hairpin RNA (shRNA) and related protein expression was analyzed 48 h later. Transfection of HNF1A shRNA significantly reduced the expression of HNF1α protein ( Fig. 2A, p < 0.05). HNF1A shRNA reduced the viability of LO2 cells (Fig. 2B, p < 0.05), which was more obvious after 36-h TG treatment. In addition, the expression of HNF1α, ATF4, and GRP78 proteins decreased with or without TG (Fig. 2C, p < 0.05). However, it significantly increased the expression of ATF6, XBP1s, and cleaved caspase-3, and RelA phosphorylation.
A dual-luciferase reporter gene was used to detect the transcriptional regulation activity of HNF1α on ATF4 promoter. Compared with the control group, HNF1α had an enhanced transcriptional regulation on either the wild-type ATF4 promoter or mutant ATF4 promoter (p < 0.01). Of interest, the transcriptional activity of HNF1α on wild-type ATF4 promoter was stronger than that on mutant ATF4 promoter (p < 0.01, Table 1).

Knockdown of ATF4 aggravates apoptosis and downregulates the expression of GRP78 in TG-treated LO2 cells.
The downregulation of ATF4 protein was confirmed 48 h post-ATF4 shRNA transfection in LO2 cells (Fig. 3A, p < 0.01). The knockdown of ATF4 significantly reduced LO2 viability (Fig. 3B, p < 0.01), downregulated the expression of ATF4, GRP78 proteins, and increased the expression of cleaved caspase-3 in LO2 cells with or without TG (Fig. 3C, p < 0.01). However, knockdown of ATF4 did not alter HNF1α expression and RelA phosphorylation.

Knockdown of RELA inhibits HNF1α expression in TG-treated LO2 cells. Knockdown of ATF6
expression in LO2 cells by ATF6 shRNA did not alter the expression of HNF1α protein (Fig. 4A, p > 0.05). However, knockdown of RELA reduced the expression of HNF1α protein (Fig. 4B, p < 0.01). Of note, knockdown of RELA reduced the viability of LO2 cells with or without TG treatment (Fig. 4C, p < 0.01), and decreased TGinduced the expression of HNF1α, ATF4, and GRP78 proteins, but increased the expression of cleaved caspase-3 (Fig. 4D, p < 0.01). In addition, bioinformatic analysis using JASPAR (http:// jaspar. gener eg. net/) predicted the presence of fifteen binding sites in HNF1A promoter for human RELA, based on the relative profile score of ≥ 80% (Table 2).

Discussion
In this study, we examined the regulatory mechanism of HNF1α and ER stress and their impact on apoptosis in SAE of liver injury. Our results show that ER stress inducer TG or TM treatment induced apoptosis, and increased HNF1α expression in LO2, HepG2, and SK-Hep1 cells. Similarly, the upregulation of HNF1α expression, ER stress, and hepatocyte apoptosis were observed in the liver injury mouse model induced by CCl 4 . In addition, knockdown of RelA significantly inhibited the upregulation of HNF1α in vitro. Taken together, these results suggest that ER stress enhanced HNF1α expression in liver injury and might be involved in activating NF-κB signaling. Furthermore, the downregulation of HNF1α in vitro showed that apoptosis was aggravated. Of interest, the proteins involved in ER stress signaling were differentially expressed: the expression of ATF4 and GRP78 was significantly downregulated while ATF6 and XBP1 expression was upregulated. Similar results were observed in vivo, as well as enhanced ER stress-related apoptosis. In addition, the double fluorescent reporter gene assay confirmed that HNF1α regulated the transcription of ATF4 promoter; knockdown of ATF4 decreased GRP78 expression and aggravated apoptosis in TG-treated LO2 cells. This interesting result implied potential crosstalk between HNF1α and ER stress through a feedback loop to alleviate hepatocyte apoptosis in liver injury. Spontaneous peritonitis aggravated liver injury by LPS was partially simulated. The results demonstrated that LPS induced liver injury, hepatocyte apoptosis, and ER stress, but inhibited the expression of HNF1α in a dosedependent manner. Treatment with low dosage LPS (0.1 mg/kg) selectively decreased CCl 4 -induced HNF1α, ATF4, and GRP78 expression and further aggravated ER stress-related hepatocyte apoptosis and liver injury. The ability of low-dosage LPS in inducing apoptosis and hence promoting the SAE of liver injury could be possibly attributed to the interference with the feedback loop between HNF1α and ER stress.
Hepatocyte degeneration and necrosis are fundamental in the pathogenesis of various liver diseases. This can cause the leakage or release of ALT, and irregular bilirubin metabolism in hepatocytes 25,26 . In this study, liver injury was induced by CCl 4 or LPS. Liver injury was evaluated by the levels of serum ALT and TBil, and the necrotic area of the liver tissue. In the liver, CCl 4 is converted into carbon trichloride that causes oxidative stress, inflammation, and ER stress 27,28 . LPS causes liver injury by activating inflammatory pathways or directly damaging hepatocytes 29 . In this study, CCl 4 and LPS increased serum ALT and TBil levels, as well as the area of liver tissue necrosis, in a dose-dependent manner. This suggests the successful establishment of different severity degrees of liver injury models.
Hepatocyte apoptosis is closely related to liver injury 30 . Caspase plays a crucial role in apoptosis signal transduction 31 . Caspase-12 is related to ER-stress-mediated apoptosis 32 . Degradation of DNA into fragments of about 180-200 bp is a prominent morphological change marking apoptosis. In this study, hepatocyte apoptosis was comprehensively evaluated by analyzing the protein expression of caspase-12 and cleaved caspase-3. The apoptosis index was measured by TUNEL staining. Our results demonstrated that hepatocyte apoptosis significantly increased in CCl 4 -induced liver injury. Aggravation of liver injury by LPS increased hepatocyte apoptosis. This agreed with previous reports, where hepatocyte apoptosis was associated with the severity of liver injury 33 . Therefore, enhancing the resistance of hepatocytes towards apoptosis might be a feasible strategy to prevent liver injury.
ER stress promotes intracellular homeostasis through UPR response, but excessive ER stress activates apoptosis signaling pathways 34 . Simultaneously, ER stress inhibits the overall protein synthesis through PERK/eIF2α signaling and upregulates the expression of molecular chaperones, such as GRP78 through ER stress-related transcription factors, such as ATF6, ATF4, and XBP1 21 . In this study, the expression of ATF4, ATF6, XBP1s, GRP78, and caspase-12 were analyzed to monitor ER stress. Results demonstrated that ER stress was associated with hepatocyte apoptosis and liver injury that was induced by CCl 4 or LPS. In addition, TG or TM induced ER stress and apoptosis in LO2, HepG2, and SK-Hep1 cells. Accumulating evidence suggests that targeted regulation of ER stress may change the progression of liver injury by altering hepatocyte apoptosis 35 . www.nature.com/scientificreports/ HNF1α is a transcription regulator that is essential for normal liver function 36 . In rats, downregulating HNF1α promoted the development of liver fibrosis and the overexpression of HNF1α significantly reduced liver fibrosis in rats 37 . Following acute inflammation, HNF1α regulates the repair of acute liver inflammation by promoting C-reactive protein expression 38,39 . In this study, it was demonstrated that HNF1α expression increased in TGinduced ER stress. Similarly, the upregulation of HNF1α expression in CCl 4 -induced liver injury in mice and HNF1α expression was positively correlated with the severity of liver injury. Immunohistochemistry indicated elevated HNF1α expression and ER stress. These results imply that ER stress can induce HNF1α expression, as well as hepatocyte apoptosis, in liver injury both in vivo and in vitro.   19 . ATF4 plays an essential role in stress signaling, which includes ER stress, hypoxia, amino acid deletion, and oxidative stress 40 . In particular, ATF4 is involved in the transcriptional regulation of amino acid synthesis, protein folding and degradation, redox balance, autophagy, and apoptosis 39,41,42 . Mutations in ATF4 significantly altered glucose homeostasis and energy consumption 43 .The activation of ATF6 upregulates the expression of ER stress-related proteins, which include XBP1 and GRP78, and enhances the ability of cells to eliminate misfolded proteins 44,45 . NF-κB signaling refers to a family of nuclear transcription factors, which includes RelA (NF-κB p65), RelB, c-Rel, NF-κB1/p50, and NF-κB2/p52 9 . In this study, ATF4, ATF6 and RelA were knocked down to analyze the impact of ER stress on HNF1α. The knockdown of RelA reduced the expression of HNF1α in TG-treated LO2 cells; however, ATF4 and ATF6 knockdown did not downregulate HNF1α expression. These results suggest that the activation of NF-κB might be one of the ways by which ER stress mediates the upregulation of HNF1α.
However, to the best of our knowledge, the regulatory role of HNF1α in ER stress and apoptosis in hepatocytes is unknown. In this study, the knockdown of HNF1α increased TG-induced apoptosis in vivo. On the other hand, HNF1α knockdown aggravated the CCl 4 -induced liver damage as indicated by increased TBil levels, apoptosis, and necrotic liver area. Of note, serum ALT levels did not increase, which might be attributed to the regulatory role of HNF1α on numerous liver-specific genes 46 . ALT is a metabolic enzyme; therefore, knocking down HNF1α might decrease liver ALT levels and decrease serum ALT These results suggest that HNF1α upregulation mitigates liver injury by reducing hepatocyte apoptosis. Further, knockdown of HNF1α differentially affects the expression of ER stress-related proteins. Specifically, it significantly downregulated the expression of ATF4 and GRP78, but increased ATF6, XBP1s, and caspase-12 expression in vivo and in vitro. In addition, this confirmed that HNF1α www.nature.com/scientificreports/ regulated the transcription of ATF4 promoter. These results suggest that HNF1α may reduce hepatocyte apoptosis through mitigating ER stress during acute phase liver injury. The knockdown of ATF4 reduced GRP78 levels and increased apoptosis in TG-treated LO2 cells. This suggested that ATF4-mediated GRP78 expression in ER stress is beneficial to alleviate apoptosis, which agrees with previous research [47][48][49][50] . These results suggest that HNF1α may reduce ER stress-mediated hepatocyte apoptosis through upregulating the expression of ATF4 and GRP78.
Viral and bacterial infections are common risk factors that are associated with severe liver injury 51,52 . Endotoxins, which are the main toxic effect component is LPS, are a cell wall component of Gram-negative bacteria. Under normal physiological conditions, a small amount of LPS produced by Gram-negative bacteria in the intestine can reach the liver via the portal circulation 53 . Most LPS is cleared by Kupffer cells without damaging hepatocytes. Compromised intestinal barrier function increases LPS leakage and this eventually leads to hepatocyte damage 54 . LPS causes liver inflammation, hepatocyte apoptosis, and aggravates liver cell injury 55,56 . Blood endotoxin levels are positively correlated with the degree of hepatocyte apoptosis 57,58 . In this study, the effect of spontaneous peritonitis aggravated liver injury by LPS injection was partially simulated. Our results demonstrated that LPS dose-dependently induced liver injury, hepatocyte apoptosis, and ER stress, but inhibited the expression of HNF1α during liver injury. In addition, low-dose LPS did not cause liver damage, and its preintervention could alleviate the liver injury 59 . However, low-dose LPS aggravates the liver injury, hepatocyte apoptosis and ER stress, inhibits the mutual regulation of HNF1α and ATF4, and reduces GRP78 expression in the CCl 4 -induced liver injury model. Therefore, LPS treatment can have different outcomes depending on the baseline status of the liver. Under physiological conditions, low-dose LPS is not enough to cause liver damage but www.nature.com/scientificreports/ it significantly increases the hepatocyte sensitivity towards apoptotic signaling mediated by ER stress after liver damage has occurred. Intestinal endotoxemia is common in severe liver injury 60 . However, the role of low-level LPS in promoting the SAE of the liver in future studies needs to be determined. It has been demonstrated that HNF1α expression was not correlated with chronic liver failure 61 . The results of this study agree with previous research and support the existence of multiple regulatory modes of HNF1α in liver injury. The expression level of HNF1α might not correlate with the degree of injury, but the effective upregulation of HNF1α in liver injury could be beneficial in controlling liver injury.
In conclusion, following liver injury, hepatocyte HNF1α and ER stress are mutually regulated to form a feedback loop, which can help reduce the severity of liver injury by regulating hepatocyte apoptosis. Whereas ER stress upregulates the expression of HNF1α by activating NF-κB signaling and the upregulated HNF1α reduces ER stress through selective regulation of ATF4 increasing GRP78 expression. In addition, high-dose LPS can directly cause liver injury, hepatocyte apoptosis as well as ER stress and inhibit the expression of HNF1α. Lowdose LPS alone cannot cause liver damage; however, it reduces the tolerance of hepatocytes to apoptosis in liver injury and leads to the deterioration of liver injury through interfering with the feedback regulation of HNF1α and ER stress (Fig. 9).

Methods
Cell culture and UPR induction. Human-derived hepatocytes LO2, HepG2, and SK-Hep1 cells were purchased from The Cell Bank of Type Culture Collection of Chinese Academy of Sciences (Shanghai, China) and were maintained in RPMI-1640 that was complemented with 10% fetal bovine serum and 1% antibiotics. To induce UPR, LO2, HepG2, or SK-Hep1 cells were treated for 12, 24, and 48 h with either dimethyl sulfoxide (DMSO; TG solvent), phosphate buffer saline (PBS; TM solvent), TG (Sigma), or TM (Sigma). Alternatively, LO2 cells were treated with different concentrations of TG (0.5, 1.0, and 2.0 μmol/L) and samples were analyzed after 24 h. To analyze the impact of HNF1α downregulation in vitro, LO2 cells were transfected with the target shRNA or control shRNA using a plasmid vector according to the manufacturer protocols and protein expression was analyzed 48 h later (Table 3). Similarly, we analyzed the impact of ATF4, ATF6 and RelA downregulation using their respective target shRNA (Table 3). Then, cells were treated with TG (1.0 μmol/L) for 36 h to induce hepatic ER stress. Previous studies demonstrated that SK-Hep1 cells were more sensitive to ER stress inducers; therefore, an induction dose of 0.25 μmol/L TG was used 62 .  . Schematic diagram representing the mechanism of suggested hepatocyte HNF1α-ER stress feedback loop on apoptosis in acute liver injury. Following liver injury, ER stress upregulates the expression of HNF1α by activating NF-κB signaling; the upregulated HNF1α reduces ER stress through positive feedback regulation with ATF4, which results in upregulation of GRP78 expression and reduction of hepatocyte apoptosis and liver injury. Low-dose LPS aggravates the CCl 4 induced hepatocyte apoptosis and liver injury by inhibiting the mutual regulation of HNF1α and ER stress. Table 3. shRNA sequences used in LO2 cells. www.nature.com/scientificreports/ Mice were placed in a 4000-mL euthanasia box and the box chamber air was replaced with 100% CO 2 at a rate of 30% of chamber volume/min 64 . Following the loss of consciousness, tissue and blood were harvested when blood circulation was maintained.
Liver function indexes. Serum ALT and TBil levels were measured using the rate method and diazonium, respectively according to the standard protocols (Beckman Coulter auto-analyzer, AU5800, USA) 19, 67 . Dual luciferase reporter assay. The search engine of the University of California, Santa Cruz Genomics Institute was used to find the promoter sequence of ATF4 (https:// genome. ucsc. edu/). The binding site of HNF1α to the ATF4 promoter sequence was predicted through JASPAR (http:// jaspar. gener eg. net/). Then, the ATF4 promoter sequence was synthesized. Next and cloning it on the firefly luciferase gene, constructing as an ATF4-promoter-wt plasmid. In addition, the HNF1α binding site on the ATF4 promoter sequence was mutated to serve as the ATF4-promoter-mt plasmid. The nucleotide sequence that corresponded to HNF1A was cloned into the pcDNA3.1 vector. The pcDNA3.1 empty vector was used as a control. HEK 293FT cells (ATCC) were cultured and seeded into 24-well plates and grown for 10-24 h (80% confluence). Cells were cotransfected with the reporter gene plasmid, and the transcription factor expression plasmid, and the Renilla luciferase plasmid Table 4. Sequence of the control and Hnf1a shRNA.

Insert content Sequence
Hnf1a shRNA 5 '-GCG ATG AGC TGC CAA CTA AGA  CGAA TCT TAG TTG GCA GCT CAT CGC -3'   Control shRNA  5'-AAA CGT GAC ACG TTC GGA GAA  CGA ATT CTC CGA ACG TGT CAC GTT T-3 www.nature.com/scientificreports/ (an internal reference). Following cell lysis and protein extraction, luciferase activity was detected by firefly luciferase and Renilla luciferase detection reagents (Biyuntian Biotechnology, Shanghai, China) to determine the relative light unit (RLU). The RLU value (Fluc) obtained by the firefly luciferase assay was divided by the RLU (Rluc) value obtained by the Renilla luciferase assay as an internal reference. According to the obtained ratio, the activation degree of the target reporter gene between the different groups was compared 68 .
Statistical analysis. One sample Kolmogorov-Smirnow test was used to test whether the continuous variables satisfied a normal distribution. Quantitative data that satisfied a normal distribution were shown as mean ± standard deviation ( χ ± SD). Differences between groups were evaluated using one-way ANOVA. If the difference was statistically significant, the Student-Newman-Keuls test was used for further comparison between the two groups. A p-value < 0.05 was statistically significant.
Ethics approval. This study was reviewed and approved by the Animal Experimental Ethics Committee, Zunyi Medical University (ZMC-LS [2020] No. 2-321) according to the animal care and research guidelines. All procedures were performed following the relevant guidelines and regulations.

Data availability
The datasets generated and analyzed during the current study are not publicly available due to none of the data types requiring uploading to a public repository but are available from the corresponding author on reasonable request.