The role of X-box binding protein 1 in the hepatic response to refeeding in mice

Refeeding mice after a prolonged fast is a potent stimulus of hepatic lipogenesis, but is also associated with induction of the hepatic unfolded protein response (UPR). The X-box binding protein 1 (Xbp1), a key regulator of the adaptive UPR, transcriptionally activates hepatic lipogenesis genes. We therefore determined whether hepatic Xbp1 mediates the hepatic lipogenic response to refeeding. Mice bearing a hepatocyte-specific deletion of Xbp1 and littermate controls were fasted overnight, followed by refeeding for up to 6 h. Among control mice, refeeding induced hepatic expression of activated Xbp1 and, as expected, induced hepatic expression of genes controlling de novo lipogenesis of fatty acids. Unexpectedly, deletion of hepatic Xbp1 allowed for normal induction of hepatic lipogenesis genes, yet impaired translation of SREBP1c and its targets in response to refeeding. Impaired protein translation was associated with enhanced postprandial activation of the global translational arrest protein, eukaryotic initiation factor 2α, among mice lacking hepatic Xbp1. Deletion of hepatic Xbp1 prevented postprandial induction of genes regulating protein folding and processing and shifted the pattern of postprandial UPR activation to favor proapoptotic signals. We conclude that activation of hepatic Xbp1 in the postprandial states serves the dual roles of restoring postprandial hepatic lipogenesis and proteostasis.

acid -oxidation is induced with fasting and declines upon refeeding (2)(3)(4), whereas the hepatic expression of lipogenesis genes declines with fasting, followed by marked induction upon refeeding (5,6). The shifts in fuel metabolism that occur in the postprandial state are well-established; however, the underlying molecular signals that trigger these shifts are complex and remain incompletely understood.
In addition to well-characterized shifts in fuel metabolism, refeeding is also associated with induction of the hepatic unfolded protein response (UPR) (7,8), a highly conserved signaling cascade activated in response to endoplasmic reticulum (ER) stress. A major function of the UPR is to activate adaptive mechanisms to cope with the accumulation of excess or misfolded proteins in the ER (9,10). In addition to its well-characterized function in maintaining protein homeostasis (i.e., "proteostasis"), the UPR is increasingly recognized for its role in mediating key metabolic processes. It is unclear, however, whether induction of the UPR in response to refeeding regulates the metabolic shifts that occur in the transition from fasting to refeeding.
The inositol requiring enzyme 1 (IRE1)-X-box binding protein 1 (XBP1) branch of the UPR plays a major role in mediating the adaptive phase of the UPR (11)(12)(13)(14). In response to ER stress, activated IRE1 promotes nonconventional splicing of XBP1 to produce a potent transcription factor (XBP1s), which, in turn, activates genes involved in protein folding, processing, and ER-associated degradation (ERAD) (11,15,16). Additionally, however, the IRE1-XBP1 branch of the UPR is increasingly recognized for its role in regulating hepatic lipid and glucose metabolism, a function that is distinct from its role in maintaining proteostasis. In particular, XBP1 has been shown to transcriptionally activate genes involved in hepatic lipogenesis Abstract Refeeding mice after a prolonged fast is a potent stimulus of hepatic lipogenesis, but is also associated with induction of the hepatic unfolded protein response (UPR). The X-box binding protein 1 (Xbp1), a key regulator of the adaptive UPR, transcriptionally activates hepatic lipogenesis genes. We therefore determined whether hepatic Xbp1 mediates the hepatic lipogenic response to refeeding. Mice bearing a hepatocyte-specific deletion of Xbp1 and littermate controls were fasted overnight, followed by refeeding for up to 6 h. Among control mice, refeeding induced hepatic expression of activated Xbp1 and, as expected, induced hepatic expression of genes controlling de novo lipogenesis of fatty acids. Unexpectedly, deletion of hepatic Xbp1 allowed for normal induction of hepatic lipogenesis genes, yet impaired translation of SREBP1c and its targets in response to refeeding. Impaired protein translation was associated with enhanced postprandial activation of the global translational arrest protein, eukaryotic initiation factor 2, among mice lacking hepatic Xbp1. Deletion of hepatic Xbp1 prevented postprandial induction of genes regulating protein folding and processing and shifted the pattern of postprandial UPR activation to favor proapoptotic signals. We conclude that activation of hepatic Xbp1 in the postprandial states serves the dual roles of restoring postprandial hepatic lipogenesis and proteostasis.-Olivares, S., and A. S. Henkel. The role of X-box binding protein 1 in the hepatic response to refeeding in mice. J. Lipid Res. 2019. 60: 353-359.
Supplementary key words unfolded protein response • fatty acid synthesis • fasting • lipogenesis • proteostasis Refeeding after a prolonged fast induces well-characterized changes in hepatic lipid metabolism that are critical to maintaining energy balance within an organism. The transition from fasting to refeeding is characterized by a shift from fatty acid -oxidation to fatty acid synthesis (1). Accordingly, the hepatic expression of genes controlling fatty (17)(18)(19). We therefore considered whether activation of hepatic XBP1 upon refeeding promotes the physiologic induction of lipogenesis genes that occurs in response to refeeding. Using mice bearing a hepatocyte-specific deletion of Xbp1 (Xbp1 LKO ), we directly determine the role of hepatic Xbp1 in mediating the hepatic response to refeeding.

Animals and treatments
C57BL/6 -Xbp1 fl/fl mice with loxP sites flanking exon 2 of the Xbp1 gene were kindly provided by Dr. Laurie H. Glimcher. The generation of the Xbp1 fl/fl strain has been previously reported (17). Xbp1 fl/fl mice were bred with C57BL/6-Albumin-Cre mice (Jackson Laboratory, ME) that express Cre-recombinase in albumin-producing hepatocytes as previously described (12,20). Xbp1 fl/fl mice expressing Cre recombinase were confirmed to be liver-specific Xbp1-KO mice (Xbp1 LKO ) by Western blot for XBP1 and real-time PCR using primers targeting a deleted region of the transcript in exon 2. Littermate Xbp1 fl/fl mice negative for the expression of Cre recombinase were used as control mice. To determine the effect of hepatic Xbp1 deletion on the hepatic response to refeeding, female Xbp1 LKO mice and Xbp1 fl/fl controls (6 weeks of age) were fasted overnight (18 h from 4 PM to 10 AM) followed by ad libitum refeeding of a standard chow diet for 2 or 6 h prior to being euthanized. All mice underwent 14/10 h light/dark cycling before and during the treatment protocol and were given free access to water during dietary manipulation. Mice were euthanized by CO 2 inhalation followed by cardiac puncture. The cardiac blood was immediately centrifuged to collect the plasma. The livers were rapidly excised, flushed with ice-cold saline, and sectioned. An aliquot of liver was fixed in formalin, and the remaining liver was snap-frozen in liquid nitrogen. All animal protocols were approved by the Northwestern University IACUC.

Histologic analysis
Liver sections were stained with Oil Red O at the Northwestern University Mouse Histology and Phenotyping Laboratory (Chicago, IL). Slides were blindly evaluated for the presence of steatosis.

Liver and plasma chemistries
Liver samples were homogenized in Dulbecco's PBS for hepatic lipid analysis (100 mg of liver tissue/1 ml). Triglyceride levels were measured in liver homogenate and fresh plasma using an Infinity spectrophotometric assay per the manufacturer's protocol (Thermo Scientific, Melbourne, Australia). Plasma -hydroxybutyrate levels were measured using a biochemical assay (Stanbio Laboratory, Boerne, TX). Plasma insulin was measured using an Insulin ELISA assay kit (Thermo Scientific, Frederick, MD).

Xbp1
LKO and Xbp1 fl/fl control mice were fasted for 18 h prior to injection of poloxamer 407 (1 mg/g ip), a lipoprotein lipase inhibitor. Blood was collected by tail venipuncture at baseline followed by 1, 2, and 4 h postinjection. Triglyceride concentration was measured in fresh plasma using an Infinity spectrophotometric assay (Thermo Scientific).

Analysis of gene and protein expression
Total RNA from frozen liver was isolated using TRIzol reagent, and real-time quantitative PCR was performed as previously described (21,22). Cellular nuclei were isolated from 100 mg of fresh liver tissue using a nuclear extraction kit (Cayman Chemical, Ann Arbor, MI) per protocol. Total protein was isolated from frozen liver samples and nuclear extracts followed by Western blotting as previously described (21). Protein detection was performed using polyclonal rabbit Abs to SREBP1c, FAS, acetyl-CoA carboxylase (ACC), stearoyl-CoA desaturase 1 (SCD1), C/EBP homologous protein (CHOP), total and phosphorylated eukaryotic initiation factor 2 (eIF2), and GAPDH (Cell Signaling Technology, Danvers, MA). Bound Ab was detected using goat anti-rabbit polyclonal HRP Ab (Cell Signaling Technology). Representative Western blots of pooled samples are shown.

Statistical analysis
Data are presented as mean ± SD. Comparisons between groups were performed using Student's t-test analysis.

Hepatic expression of spliced Xbp1 is induced upon refeeding
We began by examining the effects of refeeding on hepatic Xbp1 activation in hepatocyte-specific Xbp1-KO mice (Xbp1 LKO ) and littermate Xbp1 fl/fl control mice. Xbp1 LKO and Xbp1 fl/fl control mice were fasted overnight for 18 h, followed by refeeding for 2 or 6 h. Refeeding Xbp1 fl/fl control mice resulted in an 7-fold induction of hepatic Xbp1s expression by 2 h and >10-fold induction by 6 h relative to mice in the fasted state (Fig. 1A). As expected, Xbp1 LKO mice showed near absence of spliced Xbp1s mRNA in the fasted state and no induction of Xbp1s in response to refeeding.
Hepatic Xbp1 activates hepatic lipogenesis at a posttranscriptional level upon refeeding Having shown that hepatic Xbp1s is induced upon refeeding, we next considered whether activation of hepatic Xbp1 in the postprandial state serves to induce hepatic expression of fatty acid synthesis genes. As previously reported (6), refeeding was associated with marked induction of genes controlling de novo fatty acid synthesis, including Srebp1c, Fas, and Acc among control mice ( Fig. 1B-D). Surprisingly, Xbp1 LKO mice showed normal induction of Srebp1c, Fas, and Acc mRNA levels upon refeeding, indicating that Xbp1 is not required for transcriptional activation of de novo lipogenesis genes in the postprandial state. Postprandial induction of lipogenesis genes is mediated, in part, by insulin-induced activation of SREBP1c (23). Consistent with normal induction of Srebp1c and its target genes, Xbp1 LKO mice showed an expected surge in plasma insulin level in response to refeeding (Fig. 1G).
In addition to inducing fatty acid synthesis genes, prolonged refeeding has been associated with induction of Scd-1, the enzyme catalyzing the conversion of saturated fatty acids to MUFAs, and diacylglycerol acyl-transferase 2 (Dgat2), controlling the final step in triglyceride synthesis (5,(24)(25)(26). In contrast to the pattern observed with fatty acid synthesis genes, we found no significant induction of hepatic Scd-1 and Dgat2 by 6 h of refeeding among Xbp1 fl/fl control mice (Fig. 1E, F, H). Xbp1 LKO mice showed profound suppression of Scd-1 and Dgat2 in both the fasted and refed state.
Given the unexpected finding of normal induction in Srebp1c and its targets among Xbp1 LKO mice, we next examined whether Xbp1 regulates lipogenesis at a posttranscriptional level. In contrast to the gene-expression profile, Xbp1 LKO mice showed significantly attenuated induction of hepatic nuclear (active) SREBP1c, FAS, and ACC protein levels relative to Xbp1 fl/fl control mice in response to refeeding, indicating impaired protein translation in the absence of hepatic Xbp1 (Fig. 1H). A major mechanism of global translational arrest under conditions of stress is phosphorylation of eIF2 (27). We have previously shown that Xbp1 LKO mice demonstrate enhanced phosphorylation of eIF2 in response to ER stress (12). We therefore considered whether impaired protein translation among Xbp1 LKO mice in the refed state may be attributable to enhanced phosphorylation of eIF2. Among Xbp1 fl/fl control mice, refeeding for 2 h resulted in a marked increase in the hepatic level of phosphorylated eIF2 (p-eIF2) (Fig. 1H). By 6 h of refeeding, hepatic p-eIF2 had returned to baseline, indicating only transient phosphorylation of eIF2 in response to refeeding. In contrast, Xbp1 LKO mice showed persistent phosphorylation of eIF2 at 6 h of refeeding, suggesting ongoing suppression of global protein translation.

Hepatic Xbp1 is not required for suppression of fatty acid -oxidation upon refeeding
Refeeding promotes suppression of fatty acid -oxidation, associated with a decline in hepatic expression of PPAR target genes (2)(3)(4). Accordingly, we found that the PPAR targets, PPAR- cofactor 1, carnitine palmitoyl acyl-CoA transferase 1 (Cpt1), and acyl-CoA-oxidase (Aco), declined upon refeeding Xbp1 fl/fl control mice ( Fig. 2A-C). Likewise, the plasma -hydroxybutyrate level normalized upon refeeding in control mice (Fig. 2D). Xbp1 LKO mice showed normal suppression of fatty acid -oxidation genes and an appropriate decline in plasma -hydroxybutyrate level upon refeeding (Fig. 2). These data indicate that hepatic Xbp1 is not required for normal suppression of fatty acid -oxidation in the immediate postprandial period.

Hepatic Xbp1 does not regulate fasting-induced hepatic steatosis
In the fasted state, excess free fatty acids are taken up by the liver and undergo esterification to neutral triglyceride, leading to a transient physiologic hepatic steatosis (28,29). As expected, Xbp1 fl/fl control mice showed mild hepatic lipid accumulation after an overnight fast, as seen on Oil Red O staining of liver sections (Fig. 3A). Xbp1 LKO mice showed a similar degree of fasting-induced lipid accumulation relative to Xbp1 fl/fl littermates. Consistent with the histologic findings, Xbp1 LKO and Xbp1 fl/fl mice showed an equivalent increase in hepatic triglyceride content after an overnight fast (Fig. 3B). These data indicate that hepatic Xbp1 is not required for development of the physiologic hepatic steatosis that occurs in response to prolonged fasting. In response to refeeding, there was no significant change in hepatic triglyceride content in either strain relative to the fasted state (Fig. 3A, B).
Deletion of hepatic Xbp1 is associated with inhibition of VLDL secretion and profoundly suppressed levels of plasma triglyceride (17). Similar to prior reports, we found that Xbp1 LKO mice showed decreased plasma triglycerides in the fed and fasted states associated with impaired hepatic triglyceride secretion (Fig. 3C). In response to refeeding, the plasma triglyceride level rose among Xbp1 fl/fl mice, but remained suppressed in Xbp1 LKO mice (Fig. 3D). These data suggest a persistent defect in VLDL secretion among Xbp1 LKO mice, regardless of nutritional state.

Hepatic Xbp1 is required for postprandial activation of genes controlling proteostasis
Hepatic Xbp1 is a critical mediator of the adaptive response to excess or misfolded proteins via induction of genes involved in protein folding, processing, and degradation (11-14, 30, 31). We therefore considered whether activation of hepatic Xbp1 by refeeding serves to restore proteostasis in response to the increased hepatic protein load associated with the postprandial state. We next examined the effect of hepatic Xbp1 deletion on the expression of adaptive genes of the UPR that function to maintain protein folding and processing. Specifically we measured the expression of Slc33a1, involved in the intraluminal acetylation of ER-associated proteins (32,33); ER degradation enhancing -mannoside (Edem), a key regulator of ERAD (34); ERdj5, an ER chaperone; Sec61, involved in protein translocation across the ER (35,36); and ER oxidase 1 (ERo1), a glycoprotein required for oxidative protein folding in the ER (37). Among Xbp1 fl/fl control mice, the hepatic expression of genes regulating protein folding and processing was induced by refeeding (Fig. 4). In contrast, Xbp1 LKO mice failed to induce Slc33a1, Edem, ERdj5, Sec61, or ERo1 in response to refeeding (Fig. 4).
CHOP is a major mediator of the proapoptotic phase of the UPR (38)(39)(40)(41)(42)(43). We found that hepatic Chop expression was transiently induced by refeeding in control mice (Fig. 5A, B). Specifically, Xbp1 fl/fl mice showed an 4-fold induction of CHOP protein and mRNA at 2 h of refeeding, followed by normalization by 6 h, suggesting successful resolution of ER stress. On the contrary, Xbp1 LKO mice showed persistent elevation in hepatic CHOP protein and mRNA expression at 6 h of refeeding. Bim, Tribbles-related protein 3 (Trb3), and death receptor 5 (Dr5) are CHOP targets that mediate ER-stress-induced apoptosis (44,45). Consistent with persistent CHOP activation at 6 h of refeeding, hepatic Bim, Trb3, and Dr5 expression were increased among Xbp1 LKO mice relative to Xbp1 fl/fl mice upon refeeding (Fig. 5C). These data suggest that deletion of hepatic Xbp1 results in preferential induction of proapoptotic elements of the UPR.

DISCUSSION
Hepatic Xbp1 is a potent transcription factor that has been shown to regulate hepatic lipogenesis distinct from its canonical function within the adaptive UPR. Refeeding mice after a prolonged fast activates hepatic Xbp1, but is also a potent stimulus of hepatic lipogenesis. Whether induction of hepatic Xbp1 in the postprandial state is required for restoration of hepatic lipogenesis was previously unknown. Here, we find that hepatic Xbp1s is induced by refeeding after a prolonged fast, yet is not required for the physiologic induction of hepatic lipogenesis genes that occurs in the immediate postprandial period. Instead, we find that hepatic Xbp1 is required for the translation of proteins that regulate de novo lipogenesis in response to refeeding. Furthermore, we find that mice lacking hepatic Xbp1 demonstrate sustained activation of eIF2, a major function of which is to promote global translational arrest. We therefore speculate that the role of hepatic Xbp1 in protein translation is mediated, at least in part, via modulation of eIF2 activation. As such, this work identifies an alternative or complementary mechanism by which hepatic Xbp1 regulates hepatic lipid metabolism distinct from its Mean (n = 7-9) ± SD. * P < 0.05 previously reported functions in direct transcriptional activation and modulation of regulated IRE1-dependent decay of mRNA (17,46).
It has been shown that refeeding induces the UPR; however, the pattern of UPR activation was previously undefined. We find that refeeding promotes sustained activation of a set of UPR-related genes that are specifically involved in protein folding and processing-that is, adaptive genes. Conversely, refeeding only transiently induces proapoptotic elements of the UPR. Refeeding is associated with an increase in protein synthesis in the liver (47,48), and, as such, it is highly plausible that activation of the adaptive UPR upon refeeding functions to maintain proteostasis in the face of a postprandial increase in hepatic protein burden. We find that refeeding promotes sustained activation of genes regulating proteostasis, and hepatic Xbp1 is required for postprandial induction of these key adaptive genes. As such, we speculate that activation of hepatic Xbp1 in the postprandial state serves to accommodate the increased protein load caused by refeeding, consistent with the canonical function of Xbp1 in promoting adaptation to protein accumulation.
Failure of the adaptive UPR to restore proteostasis in the face of cellular stress results in activation of proapoptotic signals (9,10). We have previously shown that Xbp1 LKO mice show sustained activation of proapoptotic elements of the UPR in response to pharmacologic ER stress (12). We now demonstrate that failure to induce adaptive UPR genes among mice lacking hepatic Xbp1 is associated with enhanced and sustained activation of proapoptotic elements of the UPR in response to refeeding. These data indicate that hepatic Xbp1 not only mediates the adaptive response to pathologic cellular stress (e.g., pharmacologic ER stress, experimental nonalcoholic steatohepatitis, etc.) (11-13, 16, 17), but also plays a critical function in maintaining cellular homeostasis in response to the physiologic stress associated with shifts in fuel metabolism. With the recent rise in popularity of intermittent or periodic fasting among humans, it is becoming increasingly important to understand the hepatic stress responses involved in the transition from fasting to refeeding. More long-term studies are warranted to determine whether failure of the adaptive UPR to restore homeostasis in response to chronic physiologic shifts in nutrient flux promotes liver injury.
In addition to regulating de novo lipogenesis and proteostasis, our data also support a critical function of Xbp1 in regulating the final stages of hepatic triglyceride synthesis and secretion. Moreover, it has been shown that the IRE1-XBP1 pathway regulates Dgat2 and Scd-1 mRNA levels and promotes VLDL secretion (17,18). Indeed, we find that mice lacking hepatic Xbp1 show impaired VLDL secretion and markedly suppressed mRNA levels of Dgat2 and Scd-1, defects that persist in the fasted and refed state. Of note, in contrast to Srebp1c and its target genes, we did not find induction of either Scd-1 or Dgat2 at 6 h of refeeding among control mice. Previous studies demonstrating induction of these genes upon refeeding required high-carbohydrate or high-sucrose diets and/or more prolonged timepoints of refeeding (e.g., 12-24 h) (5,(24)(25)(26).
In summary, our data demonstrate that refeeding after a prolonged fast activates an adaptive UPR, of which robust hepatic Xbp1 activation is a prominent feature. Furthermore, we find that activation of hepatic Xbp1 in response to refeeding serves the dual functions of restoring both hepatic lipid and protein homeostasis. C57BL/6 -Xbp1 fl/fl mice with loxP sites flanking exon 2 of the Xbp1 gene were kindly provided by Dr. Laurie H. Glimcher.