Section of Digestive Diseases, Department of Internal Medicine, Yale University, New Haven, CT
Address reprint requests to: Michael H. Nathanson, M.D., Ph.D., Section of Digestive Diseases, Yale University School of Medicine, 333 Cedar Street, TAC S241D, New Haven, CT 06520-8019. E-mail: firstname.lastname@example.org; fax: 203-785-7273.
Potential conflict of interest: Nothing to report.
Insulin's metabolic effects in the liver are widely appreciated, but insulin's ability to act as a hepatic mitogen is less well understood. Because the insulin receptor (IR) can traffic to the nucleus, and Ca2+ signals within the nucleus regulate cell proliferation, we investigated whether insulin's mitogenic effects result from activation of Ca2+-signaling pathways by IRs within the nucleus. Insulin-induced increases in Ca2+ and cell proliferation depended upon clathrin- and caveolin-dependent translocation of the IR to the nucleus, as well as upon formation of inositol 1,4,5,-trisphosphate (InsP3) in the nucleus, whereas insulin's metabolic effects did not depend on either of these events. Moreover, liver regeneration after partial hepatectomy also depended upon the formation of InsP3 in the nucleus, but not the cytosol, whereas hepatic glucose metabolism was not affected by buffering InsP3 in the nucleus. Conclusion: These findings provide evidence that insulin's mitogenic effects are mediated by a subpopulation of IRs that traffic to the nucleus to locally activate InsP3-dependent Ca2+-signaling pathways. The steps along this signaling pathway reveal a number of potential targets for therapeutic modulation of liver growth in health and disease. (Hepatology 2014;58:274–283)
Liver regeneration is a carefully orchestrated process regulated by cytokines, growth factors, hormones, and neurotransmitters, which act in concert to restore liver mass and function within days after tissue loss.[1, 2] This process relies not only on proliferative cascades, in which hepatocytes switch from a quiescent to a proliferative phenotype,[1, 2] but also on metabolic pathways that help maintain cellular homeostasis after liver injury. Growth factors are particularly important for this process, and insulin specifically regulates both metabolism and proliferation in the liver.[4, 5] However, insulin's effects on liver regeneration are less well understood than those of other growth factors, such as epidermal growth factor (EGF) and hepatocyte growth factor (HGF).[6, 7] Insulin acts through the insulin receptor (IR), a heterotetrameric receptor tyrosine kinase (RTK) composed of two extracellular alpha subunits, which have ligand-binding activity, and two transmembrane beta subunits that possess tyrosine kinase activity. Once insulin binds to the IR, protein tyrosine kinase is activated, resulting in phosphorylation of the tyrosine residues within the beta subunit. This, in turn, leads to the recruitment of several adaptor proteins, including src-homology 2 domain (SH2)-containing proteins such as phosphatidylinositol 3-kinase (PI3K) and phospholipase C (PLC). PLC hydrolyzes phosphatidylinositol 4,5-bisphosphate (PIP2), resulting in the formation of diacylglycerol (DAG), which activates protein kinase C (PKC), and inositol-1,4,5-trisphosphate (InsP3), which promotes Ca2+ release from intracellular stores upon binding to the InsP3 receptor (InsP3R).
Several RTKs, including the IR and the receptors for EGF, HGF, and fibroblast growth factor (FGF), have been found in the nucleus.[11-15] Evidence suggests that the IR, like the HGF receptor, c-met, translocates to the nucleus upon ligand stimulation to selectively hydrolyze nuclear PIP2 and locally generate InsP3-dependent Ca2+ signals there. Additionally, nucleoplasmic, rather than cytosolic, Ca2+ is important for cellular proliferation and is necessary in particular for progression through early prophase. However, metabolic effects of insulin result from cytosolic, rather than intranuclear, events, typified by activation of protein kinase B/Akt (PKB). Therefore, we examined whether the cytosolic and nuclear effects of IR are mediated separately and whether the subpopulation of IRs reaching the nucleus upon insulin stimulation locally induces InsP3-dependent Ca2+ signals to regulate the proliferative effects of insulin.
Materials and Methods
Cells and Cell Culture
The liver cancer cell line, SkHep-1, was obtained from the American Type Culture Collection (Manassas, VA). Cells were cultured at 37°C in 5% CO2 in Dulbecco's modified Eagle's medium (Gibco, Grand Island, NY), supplemented with 10% fetal bovine serum (FBS), 1 mM of sodium pyruvate, 50 units/mL of penicillin, and 50 g/mL of streptomycin (Gibco, Grand Island, NY).
Male Holtzmman rats (70 g), obtained from CEBIO (Federal University of Minas Gerais, Brazil) or from Harlan Laboratories (South Easton, MA), were used for all studies. Animals were maintained on a standard diet and housed under a 12-hour light/dark cycle. The investigation conformed to the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (publication 86-23, revised 1985).
SkHep-1 cells plated onto coverslips were fixed with 4% paraformaldehyde. Confocal immunofluorescence (IF) was performed as previously described.[18, 19]
SkHep-1 cell and Holtzman rat hepatocyte immunoblottings and separation of nuclear and non-nuclear protein extracts were carried out as previously described.
Cell-surface biotinylation and streptavidin pull-down were performed, with modifications, as previously described.
InsP3 Buffer Constructs
Plasmids were generated, and adenoviral constructs were amplified and purified as previously described.
Detection of Ca2+ Signals
Ca2+ signals were detected and measured by time lapse confocal microscopy as described.[14, 18, 19]
Transfection of Small Interfering RNA
Validated small interfering RNAs (siRNAs) for clathrin heavy chain (cla) and caveolin-1 (cav) were obtained from Ambion (Austin, TX). SkHep-1 cells were transfected with 5 nM of each siRNA using Lipofectamine 2000, according to the manufacturer's instructions (Gibco, Grand Island, NY). Cells were used 48 hours after transfection.
Measurement of Bromodeoxyuridine Incorporation
Cell proliferation was measured by bromodeoxyuridine (BrdU) incorporation using an enzyme-linked immunosorbent assay (Roche Applied Science, Indianapolis, IN), according to the manufacturer's instructions.
Partial Hepatectomy and In Vivo Proliferation Assay
Two-thirds (partial) hepatectomy (PH) was performed on adult male Holztman rats as previously described.
Immunohistochemistry (IHC) was performed following standard methods for microwave antigen retrieval.
Quantification of Plasmatic Glucose
Glucose content in the blood was measured using an enzymatic colorimetric assay method (Analisa, Belo Horizonte, Brazil), according to the manufacturer's instructions.
Quantification of Glycogen in Liver
Glycogen content from liver samples was determined by a phenol-sulfuric acid method, as described by Dubois et al., with modifications.
Results are expressed as mean values ± standard deviation (SD). PRISM software (GraphPad, La Jolla, CA) was used for data analysis. Groups of data were compared using the Student t test or one-way analysis of variance (ANOVA; which was used because data sets included only one independent variable), followed by Bonferroni's post-tests, and P < 0.05 was taken to indicate statistical significance.
Detailed and additional methods are available in the Supporting Materials and Methods.
The IR Translocates to the Nucleus to Initiate Nuclear InsP3-dependent Ca2+ Signals
Translocation of the IR to the nucleus has been observed in primary rat hepatocytes. To investigate whether the IR translocates to the nucleus in the SkHep-1 human hepatoma cell line as well, cells were analyzed by confocal IF microscopy to monitor localization of the IR before and after insulin stimulation. This liver cell line was used because, as in primary hepatocytes, it contains Ca2+-signaling machinery in both the cytoplasm and the nucleus. Moreover, SkHep-1 cells functionally express particularly well-studied G-protein-coupled receptors in hepatocytes, such as the V1a vasopressin receptor[14, 24] and the purinergic P2Y receptor, as well as receptors for HGF, EGF, and the IR, which are important RTKs in the process of liver regeneration. The gamma-1 isoform of PLC, which is activated by growth factors, is also present in these cells. In addition, SkHep-1 cells express the type II InsP3R, which is the most abundant isoform of this Ca2+ release channel in hepatocytes. Furthermore, the use of this cell line would facilitate transient transfection studies. Before stimulation with insulin, the IR was preferentially localized to the plasma membrane (PM) and nearly absent from the nucleus. After 5 minutes of insulin (10-nM) exposure, the IR was diffusely distributed in the cytoplasm and in the nuclear interior, and nuclear labeling was further intensified after 10 minutes of stimulation (Fig. 1A,B). To confirm IF results, immunoblottings of non-nuclear and nuclear fractions of SkHep-1 cells were performed before and 10 minutes after insulin stimulation. The purity of the nuclear fractions was confirmed by the presence of the nuclear markers, lamin B1 and histone H3, and the absence of the non-nuclear markers, Na+/K+-ATPase, glyceraldehyde 3-phosphate dehydrogenase (GAPDH), and α-tubulin (Fig. 1C). Additionally, immunoelectron microscopy for GAPDH and alpha-tubulin revealed the expression of these markers exclusively in the cytosol of intact SkHep-1 cells and their absence from intact isolated nuclei (Supporting Fig. 1). Similar results were found in samples from isolated rat hepatocytes (not shown). IR was detected in non-nuclear fractions and at low levels in nuclear fractions before stimulation. However, there was an increase in IR expression in the nuclear fractions after 10 minutes of insulin treatment (Fig. 1C). To determine whether the receptors that reach the nucleus originate at the plasma membrane, cell-surface biotinylation and subsequent streptavidin pull-down of non-nuclear and nuclear SkHep-1 fractions were performed. Biotinylated IR was found in nuclear fractions only after stimulation with insulin for 10 minutes (Fig. 1D). Together, these results show that the IR translocates from the plasma membrane to the nucleus of SkHep-1 cells upon insulin stimulation, similar to what is observed in primary rat hepatocytes.
To verify the relative contribution of nuclear versus cytosolic InsP3 to insulin-induced Ca2+ signals, we used specific adenoviral monomeric red fluorescent protein (mRFP)-tagged nuclear or cytosolic-InsP3 buffers, which contain the ligand-binding domain (residues 224-605) of the human type 1 InsP3R with either a nuclear localization sequence (InsP3-Buffer-NLS) or a nuclear exclusion signal (InsP3-buffer-NES), respectively.[14, 20] SkHep-1 cells loaded with the Ca2+-sensitive dye, Fluo-4/AM, were examined by time-lapse confocal microscopy under control, nuclear, and cytosolic InsP3 buffering conditions. Nuclear and cytosolic Ca2+ signals were monitored during insulin (10-nM) stimulation. InsP3-Buffer-NLS and InsP3-Buffer-NES were correctly localized in the nucleus and in the cytosol, respectively (Fig. 2A). In control cells, insulin-induced Ca2+ signals occurred in the nucleus and in the cytosol. However, the Ca2+ increase occurred first in the nucleus (Fig. 2A,B). Both nuclear and cytosolic Ca2+ signals were nearly eliminated by buffering InsP3 in the nucleus (Fig. 2A,C,E); nuclear Ca2+ signals were not affected in the presence of the cytosolic InsP3 buffer, whereas cytosolic Ca2+ signals had a minimal decrease (Fig. 2A,D,E). These results are similar to previous findings in SkHep-1 cells. Collectively, these observations demonstrate that insulin promotes IR translocation to the nucleus and initiation of Ca2+ signals dependent on nuclear InsP3.
Insulin-Induced Cell Proliferation Depends Upon Nuclear InsP3
Insulin regulates viability, growth, and proliferation of primary hepatocytes and hepatoma cell lines,[4, 27] and nuclear, rather than cytosolic, Ca2+ is required for cell proliferation. To verify whether nuclear InsP3 is the upstream regulator of insulin-induced cell proliferation, SkHep-1 cells were synchronized in G0 by serum withdrawal, transfected with InsP3-Buffer-NLS, and assayed for BrdU incorporation. Insulin, 10% FBS, and HGF each induced significant increases in BrdU uptake, when compared to unstimulated control cells, as expected. However, BrdU uptake was reduced in cells expressing InsP3-Buffer-NLS, relative to control cells treated with insulin. Nuclear InsP3-buffered cells treated with insulin also had significantly smaller BrdU uptake than control cells stimulated with insulin. BrdU uptake in InsP3-Buffer-NLS cells stimulated with insulin was not significantly higher than in untreated InsP3-Buffer-NLS cells (Fig. 2F). Together, these results indicate that formation of InsP3 in the nucleus is required for insulin-induced cell proliferation.
Insulin-Induced Proliferation Depends on IR Translocation to the Nucleus
Upon insulin stimulation, the IR undergoes endocytosis through the classic clathrin (cla)-dependent pathway, such as does other RTKs. However, a subpopulation of IRs on the PM is associated with caveolin (cav)-enriched membrane domains. To determine whether cla and/or cav are necessary to mediate IR translocation from the plasma membrane to the nucleus, we used specific siRNAs that allowed a knockdown of 97% in both cla and cav expression, compared to scrambled siRNA-transfected cells (Fig. 3A-D). Immunoblottings of non-nuclear and nuclear fractions showed that silencing of cav caused a decrease in nuclear IR by 46.5%, when compared to scrambled siRNA-transfected cells stimulated with 10 nM of insulin. Silencing of cla caused a 24.7% decrease in nuclear IR, as compared to scrambled siRNA-transfected cells stimulated with insulin (10 nM), which was marginally significant (P = 0.08). Furthermore, simultaneous silencing of both proteins had an additive effect, causing a decrease in nuclear IR by 65.8%, when compared to scrambled siRNA-transfected cells stimulated with insulin (Fig. 3E,F). These observations suggest that these two proteins act in concert to mediate the translocation of the IR to the nucleus upon insulin stimulation. To determine whether translocation of IR to the nucleus is necessary for insulin-induced cell proliferation, cells were assayed for BrdU uptake, as described above, in the presence of each or both siRNAs. The presence of either cla or cav siRNA decreased BrdU uptake, compared to scrambled siRNA-transfected cells treated with insulin (Fig. 3G). Cla or cav siRNA-transfected cells treated with insulin also had reduced BrdU uptake, compared to scrambled siRNA-transfected cells treated with insulin. Similarly, BrdU uptake was reduced in the presence of both siRNAs before or after insulin treatment, when compared to scrambled siRNA-transfected cells treated with insulin (Fig. 3G). Collectively, these results provide evidence that cla- and cav-dependent translocation of the IR to the nucleus is necessary for insulin-induced proliferation in vitro. The fact that there appeared to be a stepwise decrease in nuclear IR with knockdown of clathrin, then caveolin, then both (Fig. 3F), but a similar decrease in BrdU uptake under all three circumstances (Fig. 3G), may reflect that the actions of other RTKs may have been inhibited as well. To examine whether impaired IR translocation to the nucleus affects insulin-induced Ca2+ signals, cells were analyzed by time-lapse confocal microscopy in the presence of scrambled siRNA and each or both cla and cav siRNAs. Silencing of either protein caused a decrease in both nuclear and cytosolic Ca2+ signals. Both nuclear and cytosolic Ca2+ signals were further impaired after simultaneous cla and cav silencing, when compared to scrambled siRNA-transfected cells (Fig. 4A-C). These results provide evidence that cla- and cav-mediated translocation of the IR from the PM to the nucleus is required to initiate insulin-induced Ca2+ signals. To confirm the specificity of these effects for insulin's action as a mitogen, we examined Akt activation, a known cytosolic action of insulin and the IR. Silencing of either or both proteins had no effect on Akt phosphorylation, when compared to scrambled siRNA-transfected cells treated with insulin (Fig. 4D,E); this indicates that this metabolic effect of insulin does not depend on IR translocation to the nucleus, whereas nuclear Ca2+ signals and cell proliferation do. Collectively, these results demonstrate that cla- and cav-mediated translocation of IR from the PM to the nucleus regulates insulin-induced Ca2+ signals and cell proliferation.
Liver Regeneration, but Not Insulin's Metabolic Effects, Depends on Nuclear InsP3
To determine the physiological relevance of observations in SkHep-1 cells, BrdU uptake experiments were performed in vivo. Cell proliferation was measured in Holtzman rats after partial (70%) hepatectomy (PH), under nuclear (InsP3-Buffer-NLS; Fig 5A) or cytosolic (InsP3-Buffer-NES) InsP3 buffering conditions. Holtzman is an outbred stock of the widely used Sprague-Dawley rat strain and has been used in other liver regeneration studies as well. The level of infection was monitored by confocal imaging of liver sections to detect mRFP-positive cells. InsP3-Buffer-NLS was efficiently delivered and expressed in nearly 100% of hepatocyte nuclei (Fig 5A, insert). A comparable efficiency of infection was observed in InsP3-Buffer-NES animals (not shown). BrdU uptake was impaired in animals expressing InsP3-Buffer-NLS, compared to control hepatectomized (PH) animals (Fig. 5B). Of note, expression of InsP3-Buffer-NES did not significantly alter BrdU uptake, when compared to control PH animals, although this value was significantly higher than in InsP3-Buffer-NLS animals (Fig. 5B). Additionally, liver/body weight ratio after PH was reduced in InsP3-Buffer-NLS animals, when compared to sham or PH animals (Fig. 5C). InsP3-Buffer-NES animals had a smaller liver/body weight ratio, when compared to sham animals, although this value was not significantly different from control PH animals. Indeed, IR levels in the nucleus were increased 24 hours after PH in PH animals, compared to sham animals, as evidenced by IHC (Fig. 5D) and immunoblotting (Supporting Fig. 2). The 24-hour time point was chosen because it is the time at which the rate of DNA synthesis reaches its peak in hepatocytes after PH. Positive proliferating cell nuclear antigen labeling in PH animals confirms that hepatocytes are undergoing cell proliferation under these conditions (Supporting Fig. 2). These results show that liver regeneration after PH depends on nuclear InsP3, and increased nuclear IR may contribute, at least in part, to this process, in accord with our observations in vitro.
To investigate whether either cytosolic or nuclear InsP3 participate in insulin's metabolic actions, we analyzed blood glucose levels and liver glycogen content under control, nuclear (InsP3-Buffer-NLS), and cytosolic (InsP3-Buffer-NES) InsP3 buffering. Using one-way ANOVA with Bonferroni's post-tests, cytosolic, but not nuclear InsP3 buffering significantly reduced blood glucose levels (Fig. 5E) and increased liver glycogen content, compared to control animals (Fig. 5F). These results are consistent with the idea that nuclear InsP3 mediates insulin's effects on liver regeneration, but is unrelated to insulin's metabolic actions.
The effects of Ca2+ on hepatocyte proliferation are closely related to the subcellular compartments where it is released. For instance, buffering mitochondrial Ca2+ inhibits apoptosis and accelerates liver regeneration on that basis. On the other hand, buffering cytosolic Ca2+ retards liver regeneration and progression through the cell cycle after PH, although there are a number of mechanisms by which cytosolic Ca2+ can increase, and different sources of cytosolic Ca2+ may have different effects. Here, we found that liver regeneration and insulin-induced cell proliferation both depend on nuclear, but not cytosolic, InsP3, which suggests that the only InsP3-mediated Ca2+ signaling that is relevant for liver regeneration occurs in the nucleus, rather than the cytosol. Although buffering nuclear Ca2+ inhibits cell proliferation, inhibiting InsP3-mediated Ca2+ signals in the nucleus is more specific than solely buffering nuclear Ca2+. Of note, because nuclear InsP3 was buffered in our PH studies, this likely attenuated the proliferative effects not only of IR, but also of c-met and EGF receptor (EGFR) as well. Therefore, these findings serve to provide the first in vivo evidence for the general importance of RTK-induced nuclear InsP3 formation in liver regeneration.
Upon activation, the IR is internalized by endosomes. The acidic environment (pH, ∼6) within endosomes promotes ligand-receptor dissociation, as well as inactivation of the IR. Insulin is degraded within the endosome, and the IR is either sent to lysosomes for degradation or recycled to the plasma membrane for another round of binding, activation, and internalization. These trafficking steps are crucial for insulin signal transduction, because phosphorylated internalized receptors can bind to intracellular substrates to achieve insulin's biological actions. After insulin binding, the IR undergoes endocytosis through the classic clathrin-dependent pathway, as do other RTKs. It has been observed that a subpopulation of IR on the PM is associated with caveolin-enriched membrane domains,[29, 34] although the functional significance of this has not been clear. Our results suggest that these endocytic pathways act together to mediate the translocation of the IR to the nucleus. This is consistent with the observation that clathrin is involved in EGFR nuclear translocation in SkHep-1 cells. Movement of plasma membrane receptors to the nucleus has also been described for other RTKs, for some G-protein-coupled receptors, and in various cell types.[13, 15, 35, 36] The steps that follow endocytosis as the IR moves to the nucleus have not yet been elucidated. However, nuclear translocation of FGF is mediated by importin β, and movement of c-met to the nucleus depends on importin β and the chaperone, GRB2-associated binding protein 1 (Gab1), which can bind to other RTKs as well. However, whether these proteins also participate in IR translocation to the nucleus remains to be determined.
The metabolic actions of insulin in the liver are largely mediated by the PI3K-PKB pathway. Akt is activated at the plasma membrane upon IR-mediated phosphorylation of PI3K. We found that Akt activation is not inhibited by blocking IR movement to the nucleus, suggesting that insulin's effects on cell proliferation are mediated by nuclear IR and are independent of IR in the cytosol. Interestingly, insulin's metabolic effects are enhanced in the liver-specific Gab1 knockout mouse. Because Gab1 may mediate nuclear translocation of RTKs,[14, 37] the metabolic effects of insulin may be enhanced in the absence of Gab1 because of an increased amount of IR retained in the cytosol. In line with this hypothesis, our studies show that nuclear InsP3 buffering decreased cell proliferation and liver weight after PH, but did not affect insulin's metabolic actions, whereas cytosolic InsP3 buffering had the opposite effects, suggesting that insulin's proliferative effects depend on nuclear InsP3, whereas its metabolic effects depend on cytosolic signaling events. Cytosolic InsP3 buffering caused a decrease in blood glucose levels and an increase in liver glycogen content, although this might be the result, in part, of inhibition of glucagon, which can increase cytosolic InsP3/Ca2+.
Taken together, our observations provide evidence that insulin's effects on hepatocyte proliferation are the result of nuclear, and not cytosolic, InsP3/Ca2+. Furthermore, nuclear InsP3 formation is important for liver regeneration, although not only insulin, but also other growth factors, such as HGF and EGF, likely contribute to this process in vivo. Indeed, the signaling steps identified here, in which the IR trafficks from the PM to the nucleus to selectively and locally activate InsP3/Ca2+, reveal a series of potential targets to regulate cell growth in the liver to likely enhance cell proliferation after resection/living donor transplant. Therefore, this nuclear signaling pathway for insulin has broad, widespread clinical implications.
The authors acknowledge Gilson Nogueira, Douglas L. Almeida, Xinran Liu, Morven M. Graham and Kathy Harry for their technical support and Ana M. de Paula for her scientific support.