Laboratory of Genetics and Physiology, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD
Department of Nanobiomedical Science and WCU Research Center of Nanobiomedical Science, Dankook University, Cheonan, Chungnam, Republic of Korea
Laboratory of Genetics and Physiology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, 8 Center Drive, Room 101, Bethesda, MD 20892-0822, fax: 301-480-7312
Potential conflict of interest: Nothing to report.
Supported by the Intramural Research Program of the National Institute of Diabetes and Digestive and Kidney Diseases and in part by the World Class University Program, Ministry of Education, Science and Technology, through the National Research Foundation of Korea (grant R31-10069).
L. H. is an adjunct member of the Department of Nanobiomedical Science and WCU Research Center for Nanobiomedical Science.
Loss of signal transducer and activator of transcription 5 (STAT5) from liver tissue results in steatosis and enhanced cell proliferation. This study demonstrates that liver-specific Stat5-null mice develop severe hepatic steatosis as well as hepatocellular carcinomas at 17 months of age, even in the absence of chemical insults. To understand STAT5′s role as a tumor suppressor, we identified and investigated new STAT5 target genes. Expression of Nox4, the gene encoding the reactive oxygen species (ROS)-generating enzyme NOX4, was induced by growth hormone through STAT5. In addition, the genes encoding the proapoptotic proteins PUMA and BIM were induced by growth hormone through STAT5, which bound to GAS motifs in the promoter regions of all three genes. We further show that STAT5-induced activation of Puma and Bim was dependent on NOX4. Treatment of mice with transforming growth factor-β, an inducer of apoptosis, resulted in cleaved caspase-3 in control but not in liver-specific Stat5-null mice. This study demonstrates for the first time that cytokines through STAT5 regulate the expression of the ROS-generating enzyme NOX4 and key proapoptotic proteins. Conclusion: STAT5 harnesses several distinct signaling pathways in the liver and thereby functions as a tumor suppressor. Besides suppressing the activation of STAT3, STAT5 induces the expression of proapoptotic genes and the production of ROS. (HEPATOLOGY 2012;56:2375–2386)
Signal transducers and activators of transcription (STAT) 5A and 5B are latent transcription factors that are induced by a plethora of cytokines, including growth hormone, prolactin and several interleukins.1 Recently, context-specific tumor suppressor functions have been associated with STAT5, such as inhibiting expression of NPM1-ALK2 and suppressing STAT3 and transforming growth factor-β (TGF-β) activity in the liver.3 Although active STAT5 has been detected in many human tumors, constitutively active STAT5A induces senescence in normal cells.4 In particular, SOCS1 expression induced by aberrant STAT5 signaling can facilitate the process of cellular senescence, which is an important tumor suppressor mechanism.5
Mice from which the Stat5a/b locus has been deleted specifically in liver tissue displayed altered metabolic pathways and developed fatty liver (nonalcoholic steatohepatitis).6, 7 Treatment of these mice with CCl4 led to liver fibrosis and hepatocellular carcinoma (HCC), suggesting that STAT5 is a tumor suppressor.3 Aberrant activation of the TGF-β and STAT3 pathways in these mice appears to contribute to the CCl4-induced fibrosis and HCC.3
Defects in apoptosis can be pivotal contributors to the development of cancer and the impaired response of tumor cells to therapy.8 The extent to which STAT5 regulates apoptotic mechanism in liver tissue is unclear. The proapoptotic BH3-only proteins PUMA, BIM, and BID are essential for the activation of BAX- and BAK-dependent cell death programs.9 PUMA expression is reduced in melanoma tumor tissue,10 and loss of PUMA dramatically accelerated myc-induced lymphomagenesis in vivo.11 Concomitant loss of PUMA and BIM in respective knockout mice exacerbated hyperplasia of lymphatic organs and promoted spontaneous malignancies.12 Loss of PUMA- and BAX/BAK-dependent apoptosis also enhanced tumorigenesis in a hypoxia-induced tumor model.13 In the liver, JNK1-dependent PUMA expression induced hepatocyte lipoapoptosis.14 Moreover, BIM and PUMA induction and BAX activation by palmitate induced apoptosis in hepatocytes.15 BIM and BID are critical contributors in hepatocyte apoptosis caused by TNF-β in vivo.16 TNF-β can cooperate with FasL to induce hepatocyte apoptosis by activating BIM and BID.17 These results demonstrate that PUMA and BIM can function as tumor suppressors in mice.
Recent studies have demonstrated that NOX4 as a source of oxidative stress promotes apoptosis in vascular endothelial cells18 and hepatocytes,19 mitochondrial dysfunction in cardiac myocytes,20, 21 and cellular senescence in hepatocytes.22
To further understand STAT5′s role as a liver-specific tumor suppressor, we identified novel STAT5 target genes in liver and mouse embryonic fibroblasts. This study explores for the first time the link between STAT5 and NOX4 and the apoptotic proteins PUMA and BIM.
Materials and Methods
Stat5f/f;Alb-Cre mice were generated by breeding Stat5f/f mice with Alb-Cre transgenic mice.23Stat5f/f and Alb-Cre transgenic mice were on a mixed background. Only 8- to 68-week-old male mice were used in the experiments unless indicated otherwise. Animals were treated humanely, and experiments and procedures were performed according to the protocol approved by the Animal Use and Care Committee at the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK).
Liver Induced by CCl4 or Growth Hormone.
Hepatic fibrosis in mice was induced by intraperitoneal injection with 2 mL/kg body weight of 10% CCl4 (Sigma, St. Louis, MO) dissolved in olive oil (Sigma, St. Louis, MO) three times per week for 12 weeks. For growth hormone (GH) stimulation, mice were injected intraperitoneally with GH (2 μg/g body weight) (mouse GH, National Hormone and Peptide Program, NIDDK). Mice were euthanized 4 hours after injection, and livers were harvested for analyses.
Mouse hepatocyte AML12 cells were obtained from American Type Culture Collection (Manassas, VA) and cultured in a 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F12 medium supplemented with 10% fetal bovine serum, 5 μg/mL insulin, 5 μg/mL transferrin, 5 ng/mL selenium, and 40 ng/mL dexamethasone at 37°C with 5% CO2.
Antibodies, Immunoblotting, and Immunostaining.
In brief, liver tissue was lysed by adding NuPAGE LDS Sample buffer (Invitrogen, Carlsbad, CA). Western blotting was performed according to the manufacturer's instructions (Invitrogen). The rabbit polyclonal anti-STAT5 (C-17), anti-STAT3 (C-20), anti–β-actin antibodies (Santa Cruz Biotechnology, Santa Cruz, CA), anti–phospho-STAT5, anti–phospho-STAT3 (Cell Signaling Technology, Beverly, MA), anti-NOX4 (Novus Biologicals, Littleton, CO), anti-PUMA (Abcam, Cambridge, MA), and anti-BIM (Cell Signaling Technology) were used for probing western blots. Immunohistochemistry was performed using standard procedures. In short, liver tissues were removed and fixed in 10% neutral buffered formalin and embedded in paraffin wax. Five-micrometer sections were prepared for hematoxylin and eosin staining and immunofluorescence analyses. After deparaffinization, antigen unmasking was performed in a decloaking chamber (Biocare Medical, San Diego, CA) using BORG Decloaker Solution (Biocare Medical, San Diego, CA) for 5 minutes at 125°C. The sections were blocked for 30 minutes in Tris-buffered saline/Tween 20 buffer containing 3% goat serum. Primary antibodies used in this study included rabbit anti–phospho-STAT5 (Tyr694), anti-cleaved caspase-3 (Cell Signaling Technology), rabbit anti-NOX4 (Novus Biologicals, Littleton, CO), rabbit anti-PUMA (Abcam, Cambridge, MA), anti-BIM (Cell Signaling Technology), anti–phospho-histone H3 (Upstate Biotechnology, Lake Placid, NY), and anti-Ki 67 (Santa Cruz Biotechnology) in addition to mouse anti-β-catenin (BD Transduction Laboratories, San Jose, CA). For double-labeling immunofluorescence analyses, sections exposed to a pair of primary antibodies were incubated in a 1:400 dilution of goat anti-rabbit immunoglobulin G (IgG) conjugated with a red fluorophore (Alexa Fluor 594; Molecular Probes, Eugene, OR) and goat anti-mouse IgG conjugated with a green fluorophore (Alexa Fluor 488; Molecular Probes, Eugene, OR) for 30 min at room temperature. Images were obtained with a Retiga Exi camera on a Olympus BX51 microscope (Olympus America, Center Valley, PA) using Image-Pro 5.1 software.
Chromatin Immunoprecipitation Assay.
For GH stimulation, mice were injected with 2 μg/g body weight of GH intraperitoneally. They were sacrificed 45 minutes after injection, and liver tissue was harvested. Noninjected mice were used as controls. Liver tissue was cross-linked in 1.5% formaldehyde for 15 minutes at 37°C and sonicated using the Misonix Sonicator 3000 (Misonix, Farmingdale, NY). Immunoprecipitation was carried out in TE buffer containing protease inhibitors (Sigma, St. Louis, MO). Chromatin was incubated with protein A Dynabeads (Invitrogen, Carlsbad, CA), which were preincubated with STAT5A or IgG antibody (R&D Systems, Minneapolis, MN). Immunoprecipitated DNA was eluted and amplified by real-time polymerase chain reaction (PCR) using a 7900 HT fast real-time PCR system (Applied Biosystems, Foster City, CA) and analyzed using SDS2.3 Software (Applied Biosystems, Foster City, CA). Sequence-specific primers used for amplification of the putative STAT5 binding sites (GAS sites) within the Socs2, Nox4, Puma, and Bim genes were as follows: Socs2 GAS sequence, 5′-GGAGGGCGGAGTCGCAGGC-3′ (forward), 5′-GACTTGGCAAGAGTTAACCGTC-3′ (reverse); Nox4 gene GAS1, 5′-AGGCTACTTCCGGCTCAAAT-3′ (forward), 5′-GCGCATACACCCTACTTCCT-3′ (reverse); Nox4 gene GAS2, 5′-CCCAATCAGGGCATACATTT-3′ (forward), 5′-TTTCCCATTCCTAGCACAGC-3′ (reverse); Puma gene GAS1, 5′-AGCAGGAACCTGTCTCAGGA-3′ (forward), 5′-TAAAGGCTGACCCCTTCTCA-3′ (reverse); Bim gene GAS1, 5′-GAAGAGGGGTGAGCATCTTG-3′ (forward), 5′-CAGTTGGAAGCCTCAGAAGG-3′ (reverse); Bim gene GAS2, 5′-GGGTCGGTACTGGCATCTAA-3′ (forward), 5′-GCTCGGCGTTAATCACTTTC-3′ (reverse).
RNA Isolation and Quantitative Real-Time PCR Analysis.
Total RNA was isolated from liver tissue of Stat5f/f, Stat5f/f;Alb-Cre mice and hepatocytes using an RNeasy mini kit (Qiagen, Valencia, CA) and 1 μg of RNA was reverse-transcribed (complementary DNA reverse-transcription kit; Applied Biosystems, Foster City, CA). Real-time quantification of messenger RNA (mRNA) transcript levels was performed using the TaqMan Gene Expression Master Mix (Applied Biosystems, Foster City, CA) according to the manufacturer's instructions. Real-time PCR was performed using an ABI Prism 7900HT (Applied Biosystems, Foster City, CA). TaqMan probes for Nox4 (Mm00479246_m1), Socs2 (Mm00850544_g1), Puma (Mm00519268_m1), Bim (Mm00437795_m1), and beta-actin (4352341E) were used (Applied Biosystems, Foster City, CA) for real-time PCR. The SYBR primers were as follows: Cdkn2b, 5′-CCCTGCCACCCTTACCAGA-3′ (forward), 5′-CAGATACCTCGCAATGTCACG-3′ (reverse); GAPDH, 5′-AACGACCCCTTCATTGAC-3′ (forward), 5′-TCCACGACATACTCAGCAC-3′ (reverse).
All statistical analyses were performed using a two-tailed, unpaired Student t test. P ≤ 0.05 was considered significant.
ChIP, chromatin immunoprecipitation; DAPI, 4′,6-diamidino-2-phenylindole; DPI, diphenylene iodonium; GH, growth hormone; HCC, hepatocellular carcinoma; IgG, immunoglobulin G; MEF, mouse embryonic fibroblast; mRNA, messenger RNA; NIDDK, National Institute of Diabetes and Digestive and Kidney Diseases; PCR, polymerase chain reaction; ROS, reactive oxygen species; STAT5, signal transducer and activator of transcription 5; TGF-β, transforming growth factor-β.
STAT5-Dependent Regulation of Nox4, Puma, and Bim in Liver Tissue.
To gain further insight into STAT5′s role as tumor suppressor and understand underlying genetic pathways, we mined microarray-based expression data from liver tissue of control and liver-specific Stat5-null mice and from Stat5+/+ and Stat5−/− mouse embryonic fibroblasts (MEFs) (for GEO accession numbers, see Materials and Methods). In addition to the reduced expression of genuine STAT5 target genes (such as Socs2) in Stat5-null liver tissue, we observed a 2.5- and 3.6-fold reduction of Nox4 and Bim mRNA levels, respectively (Supporting Table 1). Similarly, expression of Nox4 in Stat5−/− MEFs was reduced 3.3-fold (Supporting Table 2). In addition, we observed a 5.7-fold reduction of Puma mRNA in Stat5−/− MEFs. Whereas NOX4 is a reactive oxygen species (ROS)-generating enzyme, BIM and PUMA are proapoptotic proteins. Quantitative real-time PCR and western blots confirmed GH and STAT5 dependency of the Nox4, Puma, and Bim genes in liver tissue. Nox4, Puma, and Bim mRNA levels were reduced in Stat5-null livers (Fig. 1A). The Socs2 gene served as a positive control (Fig. 1A,B). NOX4, PUMA, and BIM protein concentrations were also reduced in Stat5-null livers (Fig. 1D). Actin served as a loading control, and the greatly reduced STAT5 levels verified the efficient deletion of the Stat5 locus. To establish GH-dependent expression in vivo, control and liver-specific Stat5-null mice were injected with GH followed by mRNA analyses. Whereas GH treatment of control mice induced Nox4 mRNA levels, no such increase was observed in the absence of STAT5 (Supporting Table 1, Fig. 1B).
To determine whether STAT5 directly binds to—and thereby controls—the Nox4 gene in the liver, we scanned the promoter region for GAS motifs. Chromatin immunoprecipitation (ChIP) analyses in Stat5-null livers confirmed GH-induced STAT5 binding to two GAS motifs in the Nox4 gene promoter (Fig. 1C). STAT5 binding to a GAS motif in the Socs2 gene promoter served as a positive control (Fig. 1C).
Similar to Nox4, GH-induced Puma and Bim expression in liver tissue was STAT5 dependent (Fig. 2A) and STAT5 bound to GAS motifs in the respective promoter regions as determined by ChIP analyses (Fig. 2B). Binding to the Socs2 gene promoter served as a positive control.
STAT5 Does Not Control the Antiapoptotic Genes Bcl2, Bcl2l1, or Mcl1.
To determine whether STAT5 also controls expression of antiapoptotic genes, we analyzed mRNA levels of the Bcl2, Bcl2l1, and Mcl1 genes in control and Stat5-null livers. The respective mRNA levels did not change significantly in the absence of STAT5, suggesting that these genes are not under STAT5 control (Supporting Fig. 1A). Moreover, Bcl2, Bcl2l1, and Mcl1 mRNA levels did not change upon acute GH treatment of mice (Supporting Fig. 1B). We also explored direct STAT5 binding to the respective genomic loci in MEFs through ChIP-sequencing analyses. Although GAS motifs were identified in the Bcl2, Bcl2l1, and Mcl1 gene promoters, no significant STAT5 binding was observed (Supporting Fig. 1C). In addition, no binding was observed in the miR15/16 locus. Binding to the promoter-bound GAS motif in the Socs2 gene served as a positive control.
Expression of Nox4 in MEFs Is Under STAT5 Control.
To gain mechanistic insight into the STAT5 control of Nox4, Puma, and Bim and their interrelationship, we resorted to Stat5−/− MEFs and Stat5−/− MEFs ectopically expressing STAT5A (Stat5−/−/ Stat5A) using a retroviral expression vector. This system also permitted us to study links between STAT5- and NOX4-promoted ROS production. Overexpression of STAT5A in Stat5−/− MEFs led to a further increase of Nox4 and Socs2 expression (Supporting Fig. 2A), and GH-induced expression of these genes was restored (Supporting Fig. 2B). STAT5-mediated induction of NOX4 was also observed at the protein level (Supporting Fig. 2E). To address whether the Nox4 gene is under direct GH/STAT5 control, Stat5+/+ and Stat5−/− MEFs were stimulated with GH. Whereas Nox4 expression was induced 1.9-fold in Stat5+/+ MEFs, no induction was observed in Stat5−/− MEFs (Supporting Fig. 3A). Similarly, Socs2 gene expression was not stimulated by GH in Stat5−/− MEFs (Supporting Fig. 3A). ChIP assays confirmed that STAT5 binds to the conserved proximal GAS motifs in the Nox4 gene promoter (Supporting Fig. 2C). STAT5 binding to the Socs2 gene promoter served as a positive control. Western blot analyses confirmed the reduction of NOX4 in Stat5−/− MEFs (Supporting Fig. 2D). NOX4 and BIM levels were increased in Stat5−/−/Stat5A MEFs compared with parental Stat5−/− MEFs, further supporting that STAT5 directly controls expression of these genes (Supporting Fig. 2E).
STAT5 Controlled Expression of Puma and Bim.
Expression of Puma and Bim was STAT5-dependent and under GH control in MEFs (Supporting Fig. 3A). Western blot analyses confirmed the reduction of PUMA and BIM in Stat5−/− MEFs (Supporting Fig. 2D). Overexpression of STAT5A in Stat5−/− MEFs further increased Puma and Bim mRNA levels (Supporting Fig. 4A), and GH-dependent induction of Puma and Bim expression was observed in Stat5−/−/Stat5A MEFs but not in Stat5−/− MEFs carrying an empty control retrovirus (Supporting Fig. 4B). Tyrosine phospho-STAT5 was detected in GH-stimulated Stat5+/+ MEFs (Supporting Fig. 3C), and elevated levels were observed in Stat5−/−/Stat5A MEFs (Supporting Fig. 3D). Levels of phospho-p53 were also increased in Stat5−/−/Stat5A MEFs compared with parental Stat5−/− MEFs (Supporting Fig. 2E). Puma as a p53 target gene might be regulated by STAT5/p53 signaling.
One GAS motif was identified at position −605 in the Puma gene, and two conserved GAS motifs were identified at positions −3684 and −540 in the Bim gene (Supporting Fig. 4C). ChIP analyses in Stat5+/+ MEFs confirmed GH-induced STAT5 binding to these GAS motifs (Supporting Fig. 4C). Binding to the Socs2 gene promoter served as a positive control.
To explore the mechanistic links between phospho-p53 and expression of a subset of p53 target genes, we analyzed Stat5−/− and Stat5−/−/Stat5A MEFs. Expression of Bax, Fas, Noxa, and Ataf was increased in Stat5−/−/Stat5A MEFs compared with Stat5−/− MEFs carrying an empty control retrovirus (Supporting Fig. 5). Expression of the p53 gene was not changed in Stat5−/−/Stat5A MEFs compared with Stat5−/− MEFs.
STAT5/NOX4-Dependent Regulation of ROS Levels.
To determine whether ROS generation is under direct STAT5/NOX4 control, Stat5+/+ and Stat5−/− MEFs were cultured and assayed for ROS using DCF-DA and lucigenin. DCF fluorescence, an indicator of ROS, was stronger in Stat5+/+ MEFs than in Stat5−/− MEFs (Supporting Fig. 6A). Treatment with H2O2 further increased the production of ROS in Stat5+/+ MEFs compared with Stat5−/− MEFs (Supporting Figs. 6A and 7A). The lucigenin chemiluminescent assays established that STAT5 deficiency led to a reduced level of intracellular ROS in MEFs (Supporting Fig. 6B). Treatment of Stat5+/+ MEFs with diphenylene iodonium (DPI), a NOX inhibitor, reduced ROS levels (Supporting Figs. 6A and 7B). Although DPI inhibits several NOX members, NOX4 is the only one expressed at appreciable levels in liver tissue. This suggests that ROS in MEFs originates from NOX4.
Puma and Bim Are Regulated by NOX4.
To explore mechanistic links between STAT5A and NOX4 and expression of the Puma and Bim genes, we analyzed Stat5−/− MEFs in the absence and presence of retrovirally introduced STAT5 (Stat5−/−/Stat5A). Upon treatment of MEFs with DPI, expression of Puma and Bim was reduced only in MEFs expressing STAT5A (Supporting Fig. 6C). These data provide evidence that the Puma and Bim genes are regulated by STAT5 through NOX4 signaling. STAT5A-induced expression of the Cdkn2b gene, encoding a cell cycle inhibitor p15INK4B, was partially suppressed in the presence of DPI (Supporting Fig. 8A,B) suggesting the STAT5 target Cdkn2b is also under NOX4 control.
Treatment of MEFs with H2O2 further induced Puma mRNA levels in the presence of STAT5A but not in the absence of STAT5 (Supporting Fig. 6D). Simultaneous treatment with DPI led to a suppression of Puma expression (Supporting Fig. 6D). Cell survival in the presence of H2O2 was less affected in the absence of STAT5 (Supporting Fig. 6E). Simultaneous treatment with DPI led to a rebound of cell survival in the presence of STAT5A and to a lesser extent in the absence of STAT5 (Supporting Fig. 6E). These data suggest that STAT5/NOX4 signaling in MEFs controlled PUMA-induced apoptosis and p15INK4B-regulated cell cycle inhibition.
Puma and Bim Are Regulated by NOX4 in Hepatocytes.
To explore a possible relationship between STAT5/NOX4 and the Puma and Bim genes in hepatocytes, the cell line AML12 was treated with the NOX inhibitor DPI. This resulted in reduced levels of Puma and Bim mRNA (Fig. 2C). DPI treatment also resulted in decreased Cdkn2b expression; however, it did not change expression of the STAT5 target gene Socs2. Although DPI inhibits several NOX members, NOX4 is the only family member expressed at appreciable levels in hepatocytes.24 These data imply that the direct STAT5 target gene Cdkn2b is also regulated by STAT5/NOX4 signaling.
As shown above, STAT5 did not bind to the Bcl2, Bcl2l1, and Mcl1 gene loci, and expression was not controlled by STAT5 (Supporting Fig. 1A-C). To test whether these antiapoptotic genes were regulated by NOX4, AML12 hepatocytes were treated with the NOX inhibitor DPI. Expression of Bcl2, Bcl2l1, and Mcl1 was similar in treated and untreated cells (Supporting Fig. 1D), suggesting that these genes are not under STAT5/NOX4 control.
Immunohistochemistry was used as an independent means to corroborate the importance of STAT5 on the accumulation of NOX4, PUMA, and BIM. NOX4, PUMA, and BIM were observed in liver tissue of control mice (Fig. 3B-D, left panels) and at lower levels in liver-specific Stat5-null mice (Fig. 3B-D, right panels). GH-induced nuclear phospho-STAT5 staining was observed in control mice, but not in the absence of STAT5 (Fig. 3A).
STAT5-Dependent Regulation of Hepatoprotective Proteins.
Because loss of STAT5 is correlated with the development of liver disease, it is possible that STAT5 promotes the expression of hepatoprotective genes. We therefore analyzed whether the hepatoprotective genes Hnf6, Lifr, Egfr, and Prlr were under GH/STAT5 control. While GH treatment of control mice induced Hnf6, Lifr, Egfr, and Prlr mRNA levels, no such increase was observed in the absence of STAT5 (Supporting Fig. 9). Expression of Hnf6, Lifr, Egfr, and Prlr mRNA was slightly, yet not significantly, reduced in liver-specific Stat5-null mice (Supporting Fig. 9). Thus, reduced levels of hepatoprotective proteins may contribute to the development of liver disease in liver-specific Stat5-null mice.
Loss of STAT5 Promotes the Development of HCC in Liver-Specific Stat5-Null Mice.
We have shown that loss of STAT5 from liver tissue resulted in hepatosteatosis and HCC upon CCl4 exposure in 3-month-old mice.3.25 To investigate whether loss of STAT5 can lead to the development of HCC without chemical injury, we analyzed control and liver-specific Stat5-null mice at 17 months of age. Severe hepatic steatosis and HCC were observed in all four experimental mice analyzed, but not in age-matched controls (Figs. 4, 5), and nodules were observed in two of the four mice. To investigate molecular consequences associated with the development of HCC, we analyzed phoshpho-STAT5 and phospho-STAT3 levels in control and liver-specific Stat5-null mice at 17 months of age. phospho-STAT3 levels were greatly elevated in liver-specific Stat5-null mice at 17 months of age (Fig. 4C) but not at 2 months. To determine whether loss of STAT5 correlated with increased cell proliferation, tissue sections were stained for phospho-histone H3 as a measure of cell proliferation (Fig. 5D). The number of phospho-histone H3–positive nuclei in liver-specific Stat5-null mice at 17 months was higher than in age-matched controls.
As expected, levels of Nox4, Puma, Bim, and Socs2 mRNA were reduced in 17-month-old liver-specific Stat5-null mice compared with age-matched controls (Supporting Fig. 10A). In contrast, and as expected, Bcl2l1 and Mcl1 mRNA levels were not altered (Supporting Fig. 10B). Unexpectedly, Bcl2 mRNA levels were increased in experimental mice (Supporting Fig. 10B).
CCl4 Treatment Results in STAT5-Dependent Increase of Puma and Bim.
To further investigate whether CCl4 treatment contributes to the deregulation of Nox4, Puma, and Bim, we analyzed control and liver-specific Stat5-null mice at 3 months of age. CCl4 treatment induced Puma and Bim mRNA levels in control mice, but not in liver-specific Stat5-null mice (Supporting Fig. 11). In contrast, no change of Nox4 expression was observed. Using immunohistochemistry, NOX4, PUMA, and BIM were detected in liver tissue of control mice both in the absence and presence of CCl4 (Fig. 6A-C). In contrast, reduced NOX4, PUMA, and BIM staining was observed in liver-specific Stat5-null mice in the absence and presence of CCl4 (Fig. 6A-C).
To establish whether loss of STAT5 and reduced levels of PUMA and BIM correlated with increased cell proliferation, we stained tissue sections for Ki-67 as a measure of cell proliferation (Fig. 7A). The number of Ki-67–positive cells increased in liver tissue of liver-specific Stat5-null mice that had been treated with CCl4 (Fig. 7A). In addition, activation of the apoptotic marker cleaved caspase-3 was decreased in liver tissue of Stat5-null mice treated with CCl4 compared with treated control mice (Fig. 7B). Levels of the proapoptotic protein BAX were decreased in liver tissue from Stat5-null mice compared with control mice (Supporting Fig. 12A,B). Concentrations of proliferating cell nuclear antigen, an indicator of cell proliferation, were elevated in liver-specific Stat5-null mice treated with CCl4 (Supporting Fig. 13A,B).
STAT5/NOX4-Dependent Regulation of Apoptosis Signaling in the Liver.
To establish GH or TGF-β–dependent apoptosis signaling in vivo, control mice were injected with GH or TGF-β followed by protein and mRNA analyses. Whereas GH treatment of control mice induced caspase-3 activation and expression of Nox4, Puma, and Bim, no such increase was observed in the absence of GH (Supporting Fig. 14A). TGF-β treatment of control mice, but not experimental mice, induced caspase-3 activation and expression of Nox4, Puma, and Bim mRNA levels (Supporting Fig. 14B). This finding suggests that caspase-3 activation and expression of Puma and Bim by GH or TGF-β treatment induced apoptosis by STAT5/NOX4.
While in many cell types the transcription factor STAT5 provides proliferative and survival cues by activating respective genetic programs, it serves as a bona fide tumor suppressor in liver tissue.3, 25 Loss of STAT5 from liver tissue leads to hepatosteatosis and the development of HCC upon CCl4 treatment. STAT5′s function as tumor suppressor can be attributed in part to its ability to regulate the cell cycle control genes Cdkn2b and Cdkn1a.25 In addition, the presence of STAT5 also suppresses inappropriate cytokine-induced activation of STAT3, an oncoprotein in its own right.
We now provide evidence for additional venues used by STAT5 to control cell death and thus suppress the development of HCC. Whereas CCl4 exposure is required to induce HCC in 3-month-old liver-specific Stat5-null mice, 17-month-old mice develop HCC in the absence of this chemical insult. Thus, loss of STAT5 by itself is sufficient to fundamentally alter cellular metabolism conducive to disease development. In this study, we have identified and investigated additional STAT5 target genes whose deregulation likely contribute to the development of HCC in the absence of STAT5. Notably, STAT5 controls ROS production through the activation of the Nox4 gene and it activates the genes encoding the proapoptotic and tumor suppressive proteins PUMA and BIM. We therefore propose that STAT5 protects hepatocytes through several pathways, including the activation of cell death programs executed by NOX4, PUMA, and BIM.
Studies on mice from which the genes encoding NOX4, PUMA, and BIM had been deleted, as well as tissue culture cells expressing reduced levels of these proteins, provided sound evidence for these proteins in cell death programs. In hepatocytes, NOX4 is required for TGF-β–induced apoptosis19 and loss of NOX4 from lung epithelium is protective from TGF-β–induced apoptosis.26 In heart tissue, NOX4 protected cells from pressure overload–induced apoptosis.20 In addition to NOX4, NOX1 and NOX2 have also been linked to cell death in hepatocytes, as CCl4-dependent hepatic fibrosis and ROS generation were attenuated in the absence of the latter two isoforms.24, 27, 28 In addition, BIM was also required for tumor cell apoptosis induced by a vascular endothelial growth factor A antagonist.29 Roles for BIM and PUMA in suppressing oncogenesis have been described for B cell leukemias30 and intestinal cells,31, 32 respectively. In those cases, BIM and PUMA exerted a strong apoptotic effect, and their loss led to enhanced tumorigenesis.
Although STAT5 directly controls the expression of p15INK4B,25 PUMA, and BIM (Fig. 8), it can also exert its function through activating another direct downstream target gene Nox4, which encodes NOX4, a key regulator of ROS.18, 20 We further provide evidence for a direct link between NOX4 and PUMA and BIM. Inhibiting NOX4 activity led to decreased expression of PUMA and BIM and p15INK4B. The mechanism of this regulatory venue is still elusive.
A picture is evolving that distinct signaling pathways emerging from STAT5 contribute to the protection of hepatocytes (Fig. 8). Hyperactive GH signaling imposed by a GH transgene promoted inflammatory liver cancer in mice, and loss of STAT5 in these mice resulted in accelerated HCC.33 This study linked STAT5 to hepatoprotective genes and the aberrant activation of c-Jun in the absence of STAT5. Moreover, Mueller et al.34 reported that the combined loss of STAT5 and the glucocorticoid receptor resulted in the development of frank HCC. In that study, development of HCC was associated with GH and insulin resistance and high ROS levels. Because NOX4, the enzyme generating ROS, is under STAT5 control, the source of ROS in STAT5 glucocorticoid receptor double knockout mice needs to be identified.
Although loss of STAT5 is sufficient to induce hepatic steatosis and HCC, the extent to which the loss of individual STAT5 executors (NOX4, PUMA, BIM, p15INK4B) would sensitize hepatocytes to injury and lead to pathological changes is unclear. Lastly, the molecular basis of STAT5′s cell specificity, promoting proliferation in the hematopoietic system and apoptosis in liver, remains an enigma. Although STAT5 can activate genes controlling cell proliferation, survival, and death, it is fair to propose that the relative activity of these pathways will determine whether STAT5 is an oncoprotein or a tumor suppressor.