Growth-hormone–induced signal transducer and activator of transcription 5 signaling causes gigantism, inflammation, and premature death but protects mice from aggressive liver cancer

Authors


  • Potential conflict of interest: Nothing to report.

  • This work and K.F., M.T., K.M., and J.W.K. were supported by grant SFB F28 from the Austrian Science Fund (FWF; to R.M. and M.M.) and the GEN-AU Program Projects “Austromouse” and “InflammoBiota” from the Austrian Federal Ministry for Science and Research (to M.M.). J.W.K. was also supported by a postdoctoral fellowship from the European Molecular Biology Organization.

Abstract

Persistently high levels of growth hormone (GH) can cause liver cancer. GH activates multiple signal-transduction pathways, among them janus kinase (JAK) 2-signal transducer and activator of transcription (STAT) 5 (signal transducer and activator of transcription 5). Both hyperactivation and deletion of STAT5 in hepatocytes have been implicated in the development of hepatocellular carcinoma (HCC); nevertheless, the role of STAT5 in the development of HCC as a result of high GH levels remains enigmatic. Thus, we crossed a mouse model of gigantism and inflammatory liver cancer caused by hyperactivated GH signaling (GHtg) to mice with hepatic deletion of STAT5 (STAT5Δhep). Unlike GHtg mice, GHtgSTAT5Δhep animals did not display gigantism. Moreover, the premature mortality, which was associated with chronic inflammation, as well as the pathologic alterations of hepatocytes observed in GHtg mice, were not observed in GHtg animals lacking STAT5. Strikingly, loss of hepatic STAT5 proteins led to enhanced HCC development in GHtg mice. Despite reduced chronic inflammation, GHtgSTAT5Δhep mice displayed earlier and more advanced HCC than GHtg animals. This may be attributed to the combination of increased peripheral lipolysis, hepatic lipid synthesis, loss of hepatoprotective mediators accompanied by aberrant activation of tumor-promoting c-JUN and STAT3 signaling cascades, and accumulation of DNA damage secondary to loss of cell-cycle control. Thus, HCC was never observed in STAT5Δhep mice. Conclusion: As a result of their hepatoprotective functions, STAT5 proteins prevent progressive fatty liver disease and the formation of aggressive HCC in the setting of hyperactivated GH signaling. At the same time, they play a key role in controlling systemic inflammation and regulating organ and body size. (Hepatology 2012)

Many growth factors and cytokines, including growth hormone (GH), are able to activate two members of the signal transducer and activator of transcription (STAT) family, namely STAT5a and STAT5b (referred to as STAT5).1, 2 GH induces multiple signaling cascades that mediate a wide range of effects, including cellular proliferation, differentiation and migration, prevention of apoptosis, and regulation of metabolic pathways.3 Through STAT5, GH also controls many facets of liver physiology and pathophysiology, including the regulation of genes associated with somatic growth, such as liver-derived insulin-like growth factor 1 (IGF-1), acid-labile subunit, and suppressor of cytokine signaling 2.4 Hepatic STAT5 also regulates the expression of sex-specific and hepatoprotective liver genes.1, 2, 5 Furthermore, STAT5b has been shown to be involved in mediating metabolic effects of GH, such as its lipolytic action on adipose tissue.6

Hepatocyte-specific loss of STAT5 in mice results in impaired liver regeneration after partial hepatectomy and metabolic defects, which manifest as hepatic steatosis.7 Nonalcoholic fatty liver disease (NAFLD), which is linked to obesity and insulin resistance, comprises a spectrum of diseases, ranging from simple steatosis to steatohepatitis, which may progress to liver cirrhosis and hepatocellular carcinoma (HCC). In addition to viral hepatitis, obesity-induced NAFLD and nonalcoholic steatohepatitis (NASH) are increasingly recognized as prime causes and promoters of liver cancer progression.8 However, the molecular mechanisms underlying this progression are still poorly understood.9

Increased STAT5b activity was found to correlate with more aggressive tumors and poor clinical outcomes in hepatitis B virus–-related HCC patients as a result of increased cell motility and tumor spread.10 Upon CCl4 challenge, mice lacking hepatic STAT5 develop liver fibrosis and, in some cases, cancer as a result of increased STAT3 activation and transforming growth factor beta (TGF-β) stabilization.11 In a mouse model of cholestatic liver disease, these mice also develop severe liver fibrosis, which was attributed to the down-regulation of hepatoprotective genes.5 Additionally, inactivation of the STAT5 locus in the murine liver has been demonstrated to result in low levels of circulating IGF-1 with concomitant high levels of circulating GH. This condition, which is clinically referred to as GH insensitivity and resistance, has been reported in patients with liver cirrhosis.12

Although there is a large body of evidence suggesting that chronic liver damage—irrespective of its etiology—carries an increased risk of HCC development,13 it has, as yet, not been determined whether elevated serum levels of GH contribute to this risk. Apart from changes in body stature (e.g., body and organ size as well as body-fat content), as well as neuroendocrine and reproductive functions, animals expressing GH from a transgene (GHtg) have been shown to have a shorter life expectancy.14 Notably, several studies have demonstrated that systemically high levels of GH in transgenic mice induce the development of hepatocellular adenomas and carcinomas.15, 16 The liver pathology reported to precede the development of HCC in GHtg mice is characterized by a sustained increase in hepatocyte turnover16, 17 and the development of chronic inflammation.18 Compared to the general population, patients treated with GH over a long period of time have been found to be at a significantly higher risk of dying from cancer.19 This is in accord with an increasing body of literature suggesting that patients with acromegaly, adults with elevated concentrations of circulating IGF-1, and individuals of tall stature are at increased risk of developing carcinomas, including HCC.20, 21

To further explore the GH-STAT5-IGF-1 axis in liver pathology, we used GHtg mice expressing ovine GH from a transgene. Strikingly, loss of STAT5 in hepatocytes reversed all pathologic alterations induced by persistently high GH levels, including adverse effects on body stature, organ size, and life expectancy resulting from kidney failure in GHtg animals. Importantly, hepatocyte morphology and chronic inflammation were also corrected to a great extent. Yet, hepatic deletion of STAT5 (GHtgSTAT5Δhep) led to a more aggressive form of HCC at earlier time points with 100% penetrance. This may be attributed to the combination of increased peripheral lipolysis, hepatic lipid synthesis, loss of hepatoprotective mediators, accumulation of mutations subsequent to loss of cell-cycle control, and aberrant activity of tumor-promoting STAT3 and c-JUN-signaling pathways.

Abbreviations

AlfpCre, Cre recombinase under albumin promoter and albumin and alpha-fetoprotein enhancers; BW, body weight; CAB, chromotrope aniline blue; CD, cluster of differentiation; DKO, double knockout; DNs, dysplastic nodules; EGFR, epithelial growth factor receptor; ELISA, enzyme-linked immunosorbent assay; ERK, extracellular-signal–regulated kinase; FFA, free fatty acid; GH, growth hormone; HCC, hepatocellular carcinoma; H&E, hematoxylin and eosin; HNF, hepatocyte nuclear factor; IGF-1, insulin-like growth factor 1; IL, interleukin; JNK, c-JUN-N-terminal kinase; LIFR, leukemia inhibitory factor receptor; MAPK, mitogen-activated protein kinase; mRNA, messenger RNA; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; NF-κB, nuclear factor kappa light-chain enhancer of activated B cells; PAS, periodic acid-Schiff base; pH2AX, phosphorylated histone H2AX; PPAR-γ, peroxisome proliferator-activated receptor gamma; PRLR, prolactin receptor; qRT-PCR, quantitative reverse-transcription polymerase chain reaction; RC, respiratory control; ROS, reactive oxygen species; Src, sarcoma; STAT, signal transducer and activator of transcription; TEM, transmission electron microscopy; TGF-β, transforming growth factor beta; TNF-α, tumor necrosis factor alpha; WAT, white adipose tissue; WT, wild type.

Materials and Methods

Transgenic Animals.

Mice with hepatic deletion of STAT5 (STAT5abfl/fl, Cre recombinase under albumin promoter and albumin and alpha-fetoprotein enhancers [AlfpCre]; referred to as STAT5Δhep, described previously1) were bred with GH transgenic animals expressing ovine GH under the control of the metallothionein promoter (GHtg, described previously16) to generate GHtgSTAT5Δhep mice. Littermates not expressing AlfpCre recombinase and the GH transgene served as controls. Male mice were used, unless indicated otherwise.

Histology and Immunohistochemistry.

Sections prepared from paraffin-embedded formalin-fixed organ specimens were stained with hematoxylin and eosin (H&E), chromotrope aniline blue (CAB), Giemsa, Prussian blue, and periodic acid-Schiff base (PAS). Immunohistochemistry was performed for cleaved caspase-3 (AF835; R&D Systems, Minneapolis, MN) and Ki67 (RM-9106; Thermo Scientific, Fremont, CA). Caspase-3-positive and Ki67-positive hepatocytes were quantified for 10 high-power fields at 200× and 400× magnification, respectively.

Additional Materials and Methods.

Animal procedures, serum biochemistry, immunoblotting, enzyme-linked immunosorbent assay (ELISA), quantitative reverse-transcription polymerase chain reaction (qRT-PCR), determination of mitochondrial function, and transmission electron microscopy (TEM) are described in the Supporting Materials and Methods.

Statistical Analysis.

All values are represented as means ± standard error of the mean, if not indicated otherwise. qRT-PCR quantifications and serum parameters were evaluated for significance using one-way analysis of variance, with Tukey's post-hoc test. Differences between experimental groups were considered significant at P < 0.05, P < 0.01, and P < 0.001. Kaplan-Meier plots were analyzed for significance using the log-rank test. All calculations were performed using GraphPad Prism software (GraphPad Software, Inc., La Jolla, CA).

Results

STAT5 Is Essential for Postnatal Body Growth.

GH is the major stimulator of somatic growth, cellular proliferation, and regeneration. To assess the importance of hepatic STAT5 for postnatal growth, we compared the growth curves of GHtg, GHtgSTAT5Δhep, and STAT5Δhep mice to those of wild-type (WT) littermates (Fig. 1A,B). Our data confirm that high GH levels lead to gigantism, characterized by an increase in body weight (BW) of at least 40% (Fig. 1A,B) and a 15% increase in long bone growth (Supporting Fig. 1) in male mice, compared to WT littermates. Contrary, abrogated GH signaling in the liver resulting from loss of hepatic STAT5 (STAT5Δhep) results in a reduction in body size of approximately 40% (Fig. 1A,B). GHtgSTAT5Δhep mice showed growth curves identical to those of WT littermates aged up to 9 weeks. Thereafter, the absence of STAT5 in GHtg mice resulted in growth retardation, reducing BW to approximately 80% of that in WT mice (Fig. 1A,B), whereas bone length was 5% above the WT length (Supporting Fig. 1). Compared to STAT5Δhep mice, GHtgSTAT5Δhep mice were not significantly larger. These changes were paralleled by reduced levels of IGF-1, a well-known STAT5 target and the main regulator of postnatal body growth. Both serum and messenger RNA (mRNA) expression levels (Fig. 1C,D) were found to be significantly elevated in GHtg mice, but barely detectable in GHtgSTAT5Δhep and STAT5Δhep animals. Similarly, Igf-2 expression was found to be significantly increased in GHtg mice, whereas animals lacking hepatic STAT5 showed reduced mRNA levels, compared to WT littermates (Fig. 1D). Hence, growth retardation, despite high GH levels, confirms that STAT5 is essential for GH-stimulated body growth.

Figure 1.

STAT5 is essential for postnatal body growth. (A and B) Postnatal body growth in GHtg (grey line) and GHtgSTAT5Δhep (black line) mice was compared to that in WT mice (dashed line; n ≥ 12/genotype). As already published,1 the growth curves of STAT5Δhep animals were included for reference (dashed grey line). (C) IGF-1 serum levels were assessed in 12-week-old mice by means of ELISA (n ≥ 6/genotype). (D) By means of qRT-PCR, Igf-1 and Igf-2 mRNA levels were measured in livers at 12 weeks of age (n ≥ 3/genotype). Data represent means of triplicate assays.

Reduced Life Expectancy in GHtg Mice Is Reversed by Deletion of STAT5.

High serum levels of GH were associated with premature mortality in transgenic mice (Fig. 2A). We detected severe pathologic changes in several internal organs, including the kidney, lung, and heart (Fig. 2B), which is in line with previous reports of reduced life expectancy resulting from systemically high GH levels.22 On histologic analysis, GHtg mice displayed signs of progressive inflammation in all major organs. Giemsa staining revealed very dense, partly follicular lymphocyte infiltration in the kidneys and other organs, such as the lung. Moreover, Prussian blue staining revealed the presence of activated macrophages in the kidneys, which contribute to the chronic inflammatory phenotype (Fig. 2B). The chronic inflammatory process resulted in parenchymal rarefaction, leading to cystic dilation of the renal pelvis (Fig. 2C) and, eventually, to renal failure, which was the most probable cause of death. At higher magnification, the microstructure of the kidney displayed luminal obstruction by patchy PAS-positive protein deposits in the glomeruli (i.e., glomerulonephritis). In addition, mesangial hypercellularity and thickening of the basement membrane were observed (Fig. 2B). Furthermore, chronic inflammation resulted in regenerative fibrosis within the lung and heart (CAB staining). Strikingly, GHtgSTAT5Δhep animals had a life expectancy comparable to that of WT and STAT5Δhep littermates. All pathologic changes described above were significantly improved by the deletion of hepatic STAT5. Additionally, interleukin (IL)-6 serum levels (Fig. 2D) and hepatic expression of the proinflammatory cytokines, Il-6, tumor necrosis factor alpha (Tnf-α), and Il-1β (Fig. 2E), were significantly increased in GHtg mice only, which is indicative of chronic inflammation. On the contrary, the expression of all three cytokines and IL-6 serum levels were reduced to levels comparable to WT controls in all STAT5-deficient animals (Fig. 2D,E). Nuclear factor kappa light-chain enhancer of activated B cells (NF-κB) regulates the production of inflammatory cytokines and can itself be activated by TNF-α, IL-1β, and IL-6. In GHtg livers, RelA (p65 subunit of NF-κB) mRNA expression was significantly higher, whereas p65 protein levels were slightly increased, with the expression most prominent in inflammatory cells (Supporting Fig. 2A,B). Based on the survival curve, five (six for GHtg mice) different time points were chosen for further analysis (Supporting Fig. 2C).

Figure 2.

High serum levels of GH reduce life expectancy. (A) Kaplan-Meier plot of male and female mice of all four genotypes over 65 weeks. GH transgenic mice were found to have a reduced life span. By contrast, GHtgSTAT5Δhep animals had a life expectancy comparable to that of WT and STAT5Δhep littermates (n ≥ 37/genotype; n ≥ 8 for GHtgSTAT5Δhep mice). (B) Histologic analysis of the kidneys, lung, and heart of 28-week-old WT, GHtg, and GHtgSTAT5Δhep animals stained with PAS, Giemsa, Prussian blue, and CAB revealed progressive pathologic changes, including glomerular changes, follicular infiltrates in the kidneys and lung, activated macrophages, and fibrosis of the lung and heart (indicated with arrows). (C) Both macroscopically and in microscopic overviews (25× magnification), the renal pelves of 28-week-old GHtg animals often appeared cystic and dilated. The renal parenchyma was very thin and had an atrophic appearance. (D) IL-6 serum levels were assessed in 40-week-old mice by means of ELISA (n = 7/genotype). (E) Hepatic expression of the proinflammatory cytokines, Il-6, Tnf-α, and Il-1β, was assessed in 40-week-old animals by means of qRT-PCR (n ≥ 3/genotype). All assays were performed in triplicate.

Pathologic Changes in GHtg Hepatocytes Are Reversed by Deletion of STAT5.

In the liver, persistently high expression of GH caused substantial cellular alterations. Hepatocytes of GHtg mice were abnormally large and had larger nuclei of polymorphic shape (Supporting Figs. 2D and 3A). Additionally, irregular intranuclear inclusions and an increased number of mitochondria were detected (Supporting Fig. 3A). Assessment of mitochondrial function by means of respiratory control (RC) ratios, which characterize mitochondrial activity regarding adenosine triphosphate synthesis, revealed that the mean values were diminished in GHtg mice at 40 weeks of age, whereas STAT5Δhep mice displayed significantly increased RC ratios for complex I and II substrate-dependent respiration (Supporting Fig. 3B). Moreover, corticosterone levels in GHtg mice were markedly increased (Supporting Fig. 2E). High corticosterone levels, a consistent finding in GHtg mice, are likely linked to oxidative-stress–induced aging of these mice.14 At the same time, hepatic mRNA levels of the anti-inflammatory factors, Il-10 and Tgf-β, which are able to protect the liver from severe injury,23 were increased in GHtg animals (Supporting Fig. 2F). Strikingly, upon deletion of STAT5 in GHtg mice, the increase in hepatocyte size, turnover, sinusoidal cellularity, irregular nuclear morphology, as well as the corticosterone, Il-10, Tgf-β, and RC values were reversed and reached levels comparable to those of WT controls (Supporting Figs. 2 and 3).

Increased Fat Deposition and Higher Tumor Burden in GHtgSTAT5Δhep Livers.

Increased postnatal body growth in GHtg mice is accompanied by an alteration of body proportions, resulting in an acromegaly-like phenotype, enlargement of internal organs such as hepatomegaly, and reduced peripheral body fat.24 Compared to WT littermates, all mice expressing the GH transgene were found to display increased liver/BW ratios at all time points analyzed. However, liver weight in GHtg mice never exceeded 12% of BW, whereas in GHtgSTAT5Δhep mice, ratios of more than 15% were observed at 28 and 40 weeks of age (Fig. 3A). Because of hepatosteatosis, the liver/BW ratios in STAT5Δhep mice, compared to WT littermates, were elevated as well. Yet, the ratios remained in the same range as those of GHtg controls over the whole period of analysis.

Figure 3.

Changes in body composition and metabolic parameters secondary to altered GH-STAT5 signaling. (A) Compared to WT mice, all animals expressing the GH transgene displayed increased liver to BW ratios at all time points analyzed (P < 0.05). Compared to GHtg controls, there was a significant increase (P < 0.001) in the liver/BW ratios of GHtgSTAT5Δhep mice at 28 and 40 weeks (n ≥ 8/genotype). (B) Macroscopic appearance of representative livers (upper panel) and visceral fat pads (lower panel) of 28-week-old WT, GHtg, STAT5Δhep, and GHtgSTAT5Δhep mice. An increased tumor burden was seen in GHtgSTAT5Δhep mice, which displayed large steatotic nodules (>9 mm in diameter). (C) The WAT/BW ratios, analyzed at four time points, showed a dramatic decrease in peripheral fat in GHtg and GHtgSTAT5Δhep mice after 12 weeks of age. (D) At 12 weeks of age, FFA serum levels, which were determined using an enzymatic test, were elevated in all mice lacking STAT5 (n ≥ 6/genotype). (E) Serum triglyceride, cholesterol, and glucose levels were measured at 12 weeks of age (n ≥ 4/genotype). Serum insulin levels were determined by ELISA at 12 weeks of age (n ≥ 6/genotype). (F) Liver damage was quantified by measuring the levels of the liver transaminases, AST and ALT, at all time points indicated (n ≥ 6/genotype).

The dramatic increase in liver weight observed in GHtgSTAT5Δhep mice can be explained by two factors: on the one hand, GHtgSTAT5Δhep mice showed a higher tumor burden at 28 and 40 weeks of age, compared to GHtg animals (Fig. 3B; Supporting Fig. 4B,C), and on the other hand, there was increased fat deposition in the liver (Figs. 3B and 4A). High GH levels, resulting from transgene expression and the deletion of hepatic STAT5 in GHtgSTAT5Δhep animals, led to massive lysis of peripheral fat. This is indicated by the reduced white adipose tissue (WAT)/BW ratio (Figure 3C) and the increased release of free fatty acids (FFAs) into the serum (Fig. 3D), which subsequently accumulated in their livers to a greater extent than in STAT5Δhep mice (Supporting Fig. 4). This process, which was observed at a very young age already, compromised the metabolic competence of the liver. At 12 weeks of age, serum triglyceride and cholesterol levels were found to be significantly elevated in GHtgSTAT5Δhep mice, whereas their glucose levels were only slightly increased (Fig. 3E). At the same time, high serum insulin levels were measured in GHtgSTAT5Δhep mice (Fig. 3E), which is indicative of insulin resistance. Moreover, increased lipid accumulation caused chronic damage to the hepatocytes, as could be observed from the elevated transaminase levels (Fig. 3F). As compared to WT littermates, they were increased in GHtg, STAT5Δhep, and, particularly, in GHtgSTAT5Δhep mice.

Figure 4.

Loss of hepatic STAT5 reverses GH-mediated changes in hepatocytes, but leads to increased fat deposition. (A) On histologic analysis, liver morphology in 12-week-old GHtg mice revealed increased sinusoidal cellularity and large dysplastic hepatocytes with large polymorphic nuclei. By contrast, the livers of GHtgSTAT5Δhep mice were comparable to those of WT mice in terms of cell size, inflammation, and nuclear shape. However, GHtgSTAT5Δhep mice showed massive lipid accumulation in the liver (H&E, 200×; inlay: 400× magnification). (B and C) As revealed by cleaved caspase-3 and Ki67 stainings, there was increased hepatocellular apoptosis and mitosis in 12-week-old GHtg mice (n ≥ 4/genotype).

Loss of Hepatic STAT5 Is Associated With Increased Lipid Synthesis in the Liver and Decreased Expression of Hepatoprotective Factors.

The extensive hepatic steatosis observed in GHtgSTAT5Δhep mice (Figs. 3B and 4A) results not only from lysis of peripheral fat depots, but also from enhanced expression of lipogenic regulators upon loss of hepatic STAT5,7, 25-27 particularly peroxisome proliferator-activated receptor gamma (Ppar-γ) and its target gene, cluster of differentiaiton (Cd)36 (Supporting Fig. 5A). This suggests that the steatotic phenotype is likely to be caused by a combination of lipolysis, increased lipid transport to the liver, and the induction of de novo lipid synthesis in the liver. Additionally, deregulated expression of genes that have been described to protect the liver from injury creates an environment facilitating damage-induced hepatocarcinogenesis. This makes it more likely for potentially harmful events, such as inflammation and accumulation of lipids within hepatocytes, to result in mutations. In fact, the mRNA levels of epithelial growth factor receptor (Egfr), prolactin receptor (Prlr), hepatocyte nuclear factor (Hnf6), and leukemia inhibitory factor receptor (Lifr), which are all considered to have hepatoprotective functions,5 were reduced to barely detectable levels in mice lacking STAT5 independent of transgenic GH expression (Supporting Fig. 5B). Conversely, expression of all four genes was significantly increased in livers of GHtg mice.

Loss of Hepatic STAT5 Promotes the Development of HCC in GHtg Mice.

Although the characteristic alterations observed in GHtg mice were reversed by the deletion of hepatic STAT5, earlier tumor formation was observed in GHtgSTAT5Δhep mice (Fig. 3B). To gain more insight into the cellular mechanisms contributing to hepatocyte damage and earlier tumor formation in GHtgSTAT5Δhep mice, the livers of all genotypes were subjected to histologic analysis. In hepatocytes, high long-term expression of GH resulted in greatly increased rates of hepatocellular apoptosis (caspase-3 staining; Fig. 4B), which, in turn, led to enhanced regenerative proliferation (Ki67 staining; Fig. 4C) and attracted immune cells to the sites of injury (Fig. 4A; Supporting Fig. 2D). These processes, which, in GHtg mice, were observed in the first weeks after birth already, precede the onset of hepatic inflammation (Supporting Table 1), a key feature of liver pathology in these animals.18 As a result of increased hepatocyte turnover and chronic inflammation, GHtg mice first developed neoplastic nodular lesions at 40 weeks and carcinomas at 52-60 weeks of age (Fig 5A). GHtgSTAT5Δhep mice, by comparison, displayed dysplastic nodules (DNs) at 28 weeks already (Fig. 5B). At this time, their tumors (nodules >9 mm in diameter; Fig. 3B, right panel) already exceeded those observed in GHtg mice in size at all time points analyzed (nodules 1-2 mm in diameter; data not shown). At 40 weeks of age, all GHtgSTAT5Δhep mice had large steatotic and solid tumors, which hepatopathologists classified as full-blown HCC. Histologically, all tumors—HCC as well as early lesions—in GHtgSTAT5Δhep mice were similar to those observed in GHtg animals, but aggressive tumors were observed 3 months earlier (Fig. 5C).

Figure 5.

Loss of hepatic STAT5 promotes HCC progression in GHtg mice. (A) H&E staining of GHtg livers at (a) 28 and (b and c) 40 weeks. The first DNs were observed at 40 weeks. GHtg mice did not develop carcinomas before the age of 52 weeks (d-f: livers at 62 weeks; tumors in different mice, magnification of 100× and 200×, respectively). (B) H&E staining of GHtgSTAT5Δhep livers revealed increased accumulation of lipids within hepatocytes (a) and DNs at 28 weeks of age already (b-c). At 40 weeks of age, all GHtgSTAT5Δhep mice had large steatotic and solid HCCs (d-f). (C) In both genotypes, early neoplastic nodular lesions—histologically classified as DNs—were microscopically characterized by loss of normal lobular architecture and an irregular growth pattern. They appeared sharply demarcated from the surrounding liver parenchyma, which was often compressed (black asterisk). Steatotic hepatocytes of varying size and staining pattern (yellow asterisk), but no portal triads, were observed. Moreover, focal areas of cellular atypia, characterized by slight pleomorphic nuclei, coarsely clumped chromatin, large nucleoli, and cytoplasmic basophilia, were found (blue asterisk) (a-b). Full-blown HCC was characterized by the loss of lobular plates, more pronounced cellular pleomorphism (blue asterisk), and an increased mitotic index. Tumors displayed a solid or trabecular growth pattern, with more abundant steatotic hepatocytes (yellow asterisk) (c-d).

Increased Compensatory STAT3 Activity in the Livers of GHtgSTAT5Δhep Mice Leads to Elevated c-JUN Levels.

To gain molecular insight into the processes promoting the malignant transformation of hepatocytes in GHtgSTAT5Δhep animals, we evaluated the activation of other signaling proteins and pathways known to regulate the transcription of GH-responsive genes. In the absence of STAT5 proteins, these include pY-STAT1 and pY-STAT3 and their downstream target genes,11 activation of sarcoma (Src) kinase, the mitogen-activated protein kinase (MAPK) pathway, which has a role in cellular growth and differentiation, and the phosphoinositide 3-kinase/AKT/mammalian target of rapamycin pathway, which mediates survival and growth signals to cells.3 Enhanced AKT and extracellular-signal–regulated kinase (ERK) activation, caused by high concentrations of circulating insulin and IGF-1, has been suggested to promote tumor formation in overweight and obese individuals.8 However, our findings indicate that neither AKT nor ERK nor Src activation causes enhanced HCC development in GHtgSTAT5Δhep mice (Supporting Fig. 5C).

Upon loss of hepatic STAT5, pY-JAK2 levels were slightly decreased at 40 weeks of age independent of GH expression, whereas JAK2 protein levels were slightly increased (Fig. 6A). Furthermore, the protein levels of the potential STAT5 targets, BCL-2, BCL-XL, and cyclin D1 were slightly reduced (Supporting Fig. 5C). Most notably, loss of STAT5 in GH-transgenic mice led to enhanced STAT3 and STAT1 activity (Fig. 6A), which was not observed in any other genotype. In addition, increased protein levels (Fig. 6B; Supporting Fig. 5D) and elevated phosphorylation of the proto-oncogene, c-JUN, a direct transcriptional downstream target of STAT3, were observed in GHtgSTAT5Δhep animals (Fig. 6B). Moreover, activation of major stress-dependent MAPK-signaling pathways, which are known to be triggered by chronic liver damage and to be involved in the pathogenesis of HCC, was found to be enhanced in GHtgSTAT5Δhep mice. Compared to all other genotypes, tumor-bearing GHtgSTAT5Δhep mice exhibited increased c-JUN-N-terminal kinase (JNK) 1 and p38 activation (Fig. 6B), which, in turn, can activate c-JUN.28, 29 Furthermore, their p53 activity—despite quite heterogeneous p53 levels (Fig. 6C)—was found to be significantly lower than in GHtg animals, as shown by the expression levels of its known targets Ataf, Bax, Noxa, Puma, and Fas (Fig. 6D). Hence, loss of p53 function and a significant increase in double-strand breaks (phosphorylated histone H2AX [pH2AX] staining; Fig. 6E) might contribute to enhanced development of more aggressive HCC. In summary, the development of HCCs in GHtgSTAT5Δhep animals appears to be enhanced by the activation of tumor-promoting c-JUN and STAT3-signaling cascades and accumulation of mutations after loss of a cell-cycle checkpoint (Supporting Fig. 6).

Figure 6.

Aberrant signaling accelerates HCC development in GHtgSTAT5Δhep mice. (A) Western blotting analysis of livers from 40-week-old animals showed compensatory activation of other STAT proteins in GHtgSTAT5Δhep mice only. Results are representative of three independent experiments. (B) In 40-week-old GHtgSTAT5Δhep animals, increased expression and phosphorylation of c-JUN, as well as of JNK1 and p38, was found by means of Western blotting analysis. Results are representative of three independent experiments. (C) Representative Western blotting showing p53 protein levels in livers of 40-week-old animals. (D) qRT-PCR measurements of p53 and its downstream targets Ataf, Bax, Noxa, Puma, and Fas revealed increased mRNA levels in GHtg mice and reduced levels in GHtgSTAT5Δhep animals (n ≥ 5/genotype). (E) Representative liver sections of 40-week-old mice stained with antibodies against pH2AX to detect DNA damage. Quantification of positive hepatocytes by image analysis revealed increased DNA damage in GHtgSTAT5Δhep mice (n ≥ 5/genotype).

Discussion

In patients, liver cirrhosis and hepatic dysfunction have been shown to be associated with severe impairment of the GH-IGF-1-insulin receptor axis, resulting in low IGF-1 levels and reduced response to exogenous GH (i.e., GH insensitivity/resistance).12 Mice with hepatic STAT5 deletion had depleted bioactive IGF-1, leading to dwarfism.1 Interestingly, Igf-2 expression, although largely independent of GH-signaling,21 showed a similar pattern as Igf-1 expression. IGF-2, which has an important role in embryonic development, has also been implicated in tumor development, growth, and metastasis.21

The lack of negative feedback via IGF-1 resulted in reduced body growth and high endogenous GH levels (data not shown). In GHtgSTAT5Δhep mice, high GH levels resulting from both transgene expression and deletion of hepatic STAT5 caused increased lipolysis of peripheral fat depots and release of free fatty acids, which were deposited in the liver. This is supported by the finding that disruption of GH signaling by hepatocyte-specific deletion of JAK2 results in GH resistance and the development of hepatosteatosis, which was suggested to be related to increased expression of Cd36.30 Interestingly, GHtg animals showed elevated Cd36 levels, but lysed peripheral fat was not accumulating in their livers, which remains mechanistically unexplained. STAT5 deficiency additionally resulted in uncontrolled lipid synthesis in the liver and impaired glucose and insulin tolerance.7, 25 It can be assumed that these complex mechanisms are chiefly responsible for the steatotic phenotype observed in GHtgSTAT5Δhep mice. Over time, increased hepatic lipid load increasingly compromises the metabolic competence of the liver, chronically damages hepatocytes, and aggravates the GHtg cancer phenotype. This is supported by growing evidence that obesity, which is associated with hepatosteatosis and elevated transaminase levels, promotes neoplastic transformation of hepatocytes.8 Moreover, we have recently shown that combined deficiency in hepatic GH-STAT5 and glucocorticoid-glucocorticoid receptor signaling (double-knockout [DKO] mice) increases hepatic lipid load and HCC formation. In these mice, increased stress parameters, combined with high reactive oxygen species (ROS) levels and concomitant DNA damage, contributed to HCC development on top of hepatic steatosis.25

Hepatocarcinogenesis in humans is known to be a multistep process involving the accumulation of different genetic alterations induced by chronic inflammation and oxidative DNA damage that finally leads to malignant transformation of hepatocytes. These pathologic mechanisms were also observed in GHtg mice, which developed tumors histologically closely resembling human HCC at 52-60 weeks of age.

Interestingly, histologic analysis of mouse livers showed that the hepatocellular size and the mitochondrial number and function observed in GHtg mice are dependent on STAT5. The hepatic GH-STAT5 axis was also responsible for complex disease phenotypes and premature mortality in GHtg mice. Possibly, enhanced synthesis of growth factors, such as IGF-1 and IGF-2, within the liver contributes to premature mortality. Surprisingly, HCC development in mice was found to be induced by high GH levels independent of STAT5. Compared to GHtg controls, GHtgSTAT5Δhep mice had even more advanced lesions. Most strikingly, they developed HCC 4 months earlier than GHtg controls, even though STAT5 deficiency abrogated hepatocellular nuclear polymorphism, normalized hepatocyte turnover, reduced sinusoidal cellularity (i.e., inflammation), and increased life expectancy as the pathologic alterations in the kidneys, lung, and heart disappeared.

In addition to lipid accumulation in the liver, loss of the hepatoprotective factors, EGFR, PRLR, LIFR, and HNF6, in STAT5-deficient animals promoted chronic hepatocyte damage. At the same time, loss of STAT5 resulted in compensatory GH-induced STAT3 activity in GHtgSTAT5Δhep mice. STAT3 activation was not induced by IL-6 because IL-6 serum levels were significantly lower in GHtgSTAT5Δhep mice. It is well documented that, in the absence of STAT5, STAT1/3 proteins are aberrantly recruited and activated by the GH receptor. Increased STAT3 activity has been reported in HCC.31, 32 Its oncogenic signals stimulate proliferation of aberrant hepatocytes,32 which has been linked to accelerated tumor development.11 Loss of STAT5 has also been shown to cause increased hepatocyte proliferation secondary to diminished p15INK4B expression.33 Moreover, in the absence of STAT5, GH-activated STAT3 promotes the transcription of proto-oncogenes, such as cell-cycle progression and survival genes. Accordingly, there was prominent expression of c-JUN in GHtgSTAT5Δhep animals. In the absence of STAT5, STAT3 might thus contribute to tumor induction and progression in GHtgSTAT5Δhep animals.

Chronic liver damage caused by metabolic changes in these animals additionally induces JNK1 activity, which, in turn, activates c-JUN. Moreover, increased p38 stress kinase activation might contribute to activation of c-JUN.29 Following c-JUN activation34 or other oncogenic mechanisms potentially involving STAT3,35 p53 activity was inhibited, thus impeding clearance of cells harboring double-strand breaks. In contrast to DKO animals, GHtgSTAT5Δhep mice did not show significant changes in ROS levels (data not shown) or chronic stress parameters. However, hepatic steatosis, inflammatory parameters, and aberrant STAT3 and c-JUN activation were more pronounced than in DKO mice.25

Our data suggest that the HCC development observed in GHtg mice in the absence of STAT5 was caused by two distinct molecular processes. First, loss of hepatic STAT5 results in deregulation of the STAT5 target genes involved in the protection of hepatocyte integrity and in lipid metabolism. Second, rerouting of GH signaling to other STAT proteins, which may partially compensate for the loss of STAT5, activates signaling cascades that are distinct from STAT5 and promotes cancer cell proliferation. Nevertheless, it is most likely the combination of increased lipid synthesis, lipodystrophy, deregulated expression of hepatoprotective factors, accumulation of mutations facilitated by a loss of cell-cycle control, and elevated activity of tumor-promoting c-JUN and STAT3 that causes HCC.

Acknowledgements

The authors thank J. Marjanovic (LBI-CR, Vienna, Austria) and T. Behling (LBI for Traumatology, Vienna, Austria) for their excellent technical assistance. The authors are indebted to I. and M. Friedbichler of Innsbruck Medical University (Innsbruck, Austria) for providing editorial assistance in the preparation of the manuscript for this article. The authors also thank M. Trauner for helpful discussions and critical revision of the manuscript.

Ancillary