Growth hormone is permissive for neoplastic colon growth

Abstract

Growth hormone (GH) excess in acromegaly is associated with increased precancerous colon polyps and soft tissue adenomas, whereas short-stature humans harboring an inactivating GH receptor mutation do not develop cancer. We show that locally expressed colon GH is abundant in conditions predisposing to colon cancer and in colon adenocarcinoma-associated stromal fibroblasts. Administration of a GH receptor (GHR) blocker in acromegaly patients induced colon p53 and adenomatous polyposis coli (APC), reversing progrowth GH signals. p53 was also induced in skin fibroblasts derived from short-statured humans with mutant GHR. GH-deficient prophet of pituitary-specific positive transcription factor 1 (Prop1)(-/-) mice exhibited induced colon p53 levels, and cross-breeding them with Apc(min+/-) mice that normally develop intestinal and colon tumors resulted in GH-deficient double mutants with markedly decreased tumor number and size. We also demonstrate that GH suppresses p53 and reduces apoptosis in human colon cell lines as well as in induced human pluripotent stem cell-derived intestinal organoids, and confirm in vivo that GH suppresses colon mucosal p53/p21. GH excess leads to decreased colon cell phosphatase and tensin homolog deleted on chromosome 10 (PTEN), increased cell survival with down-regulated APC, nuclear β-catenin accumulation, and increased epithelial-mesenchymal transition factors and colon cell motility. We propose that GH is a molecular component of the "field change" milieu permissive for neoplastic colon growth.

Keywords: acromegaly; colon; growth hormone; growth hormone deficiency.

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Fig. 1.

Disrupted GH signaling is associated with p53 accumulation. Western blot analyses of (A) normal human skin fibroblasts and fibroblasts derived from a short-stature subject with a GHR mutation (GHRmut); (B) GHRmut fibroblasts transfected with pcDNA3.1 (control, C) or increasing amounts of pcDNA3.1hGHR DNA (total amount of transfected DNA was 1 μg in all samples) and harvested 48 h later; (C) p53 in normal human skin fibroblasts transfected with 10 pM hGHR siRNA and harvested 48 h later. Experiments were performed three times and representative results are shown. (D) Western blot analysis of p53, p21, and APC in colon mucosal biopsies derived from seven human subjects at baseline before (untreated, U) and after 8 wk treatment with pegvisomant (Peg). Western blot analyses of colon tissue derived from (E) WT and Prop1−/− mice and (F) GHR−/− mice. In each experiment three to six mice were used.

Fig. S1.

GH deficiency results in decreased proliferation and increased senescence. (A) Percent BrdU incorporation in skin fibroblasts derived from normal human (Cont) and from GHRmut subjects. The graph depicts mean ± SEM of measurements obtained from three independent experiments; *P < 0.05. (B) Real-time PCR of p53 and p21 mRNA levels in skin fibroblasts derived from normal human (Cont) and from GHRmut subjects. Normalized PCR results are expressed as fold-change vs. control taken as 1. Results are shown as mean ± SEM of triplicate measurements. (C) Western blot analysis of normal human skin fibroblasts transfected with 10 pM scramble (Scr) or hGHR siRNA and harvested 48 h later. Experiments were repeated twice. (D) Percent BrdU incorporation in normal human skin fibroblasts transfected with scramble (Scr) or siGHR RNA for 48 h. Graph depicts mean ± SEM of triplicate measurements. *P < 0.05. In A, B, and D, differences between groups were analyzed using two-tailed unpaired Student t test.

Fig. S2.

GH in pituitary and colon of Prop1−/− mice. (A) Western blot analysis of p53 in pituitary of Prop1−/− mice. (B) Real-time PCR of GH mRNA levels in colon tissue derived from WT and Prop1−/− mice. Normalized PCR results are expressed as fold-change vs. control taken as 100%. n = 7 mice per group. Results are shown as mean ± SEM of triplicate measurements.

Fig. 2.

GH deficiency attenuates tumor development by inducing p53. (A) Number of small intestine tumors per mouse and average tumor size (mm); *P < 0.01, **P < 0.001. Results were analyzed by ANOVA followed by Tukey test. Numbers above the columns indicate number of mice analyzed. (B) Small intestine of APCmin+/− and APCmin+/−Prop1−/− mice. (C) Percent of mice bearing colon tumors in 9-mo-old APCmin+/− and APCmin+/−Prop1−/− mice. (D) Western blot analysis of colon p53 in tissue derived from pretumorous 7-mo-old APCmin+/− and APCmin+/−Prop1−/− mice. (E) Immunofluorescent confocal images of GH immunoreactivity in colon tumor of APCmin+/− and APCmin+/−Prop1−/− mice. Both APCmin+/− tumor stroma and tumor epithelium are immunopositive for GH (red), whereas no GH is detected in the colon of APCmin+/−Prop1−/− mice. (Upper) (Magnification, 20×.) (Scale bar, 200 μm.) (Lower) (Magnification, 63×.) (Scale bar, 100 μm.) (F) Western blot analysis of hNCC stably infected with lentivirus expressing nonspecific scramble (Scr) or GH shRNA and HCT116 cells transfected with nonspecific scramble or hGH siRNA for 48 h. Experiments were each performed at least three times, and representative results are depicted.

Fig. 3.

GH expression in human colon tissue. Human colon tissue arrays comprising specimens from normal colon tissue, ulcerative colitis, colon adenocarcinoma, and colon adenocarcinoma metastasis. Representative images are shown. (A) GH (green) induced in colon of patients with ulcerative colitis vs. normal adjacent tissue (NAT). (Scale bar, 100 μm.) (B) GH (green) coexpressed in stromal cells with IgA (red), a marker for plasma cells. (Scale bar, 25 μm.) (C) GH (brown) expressed in benign colon adenoma stromal tissue. (Scale bar, 100 μm.) GH (green) and α-SMA (red) expressed in CAFs in (D) adenocarcinoma and in (E) metastasis. (Scale bars, 100 μm.) (F) GHR (brown) in normal adjacent tissue (NAT) and in colon adenocarcinoma. (Scale bar, 100 μm.)

Fig. 4.

GH suppresses p53, inducing ubiquitination. (A) Real-time PCR of p53 mRNA levels in hNCC treated with 500 ng/mL GH. Normalized PCR results are expressed as fold-change vs. control (100%). Results are depicted as mean ± SEM of triplicate measurements. Two independent experiments were performed, and representative results are shown. (B) Western blot analysis of hNCC and HCT116 cells treated with indicated doses of GH for 24 h. Experiments were performed at least three times and representative results are depicted. (C) Western blot analysis of p53 in hNCC untreated (C) or treated with GH for 24 h, and then treated with control (C) or 0.5 μM MG-132 for an additional 24 h. (D) p53 and ubiquitin interactions analyzed in hNCC untreated (C) or treated with GH (500 ng/mL) for 6 h. Cell lysates were immunoprecipitated (IP) with IgG or anti-ubiquitin antibodies and immunoblotted for p53. (E) Western blot analysis of Pirh2 and TRIM29 in hNCC treated with indicated doses of GH. (F) p53 and Pirh2 interactions analyzed in hNCC untreated (C) or treated with GH (500 ng/mL) for 6 h. Cell lysates were immunoprecipitated (IP) with IgG and Pirh2, antibodies and immunoblotted for p53 and Pirh2 (for input, see D). (G) Western blot analysis of p53 in hNCC stably expressing scramble (Scr) shRNA or shPirh2 RNA and treated with GH for 24 h. (H) Western blot analysis of hNCC stably expressing scramble (Scr) shRNA or TRIM29 shRNA and treated with GH (500 ng/mL) for 24 h. Experiments were each performed at least three times and representative results depicted.

Fig. S3.

Effect of GH on p53/p21 is MDM2 independent and is mediated by GHR. Western blot analysis: (A) of total and pSTAT5 in hNCC cells treated with 500 ng/mL GH and harvested at indicated time points, (B) of total and phsopho MDM2 in hNCC treated with indicated doses of GH for 24 h, (C) of p53, p21 and Pirh2 in hNCC pretreated with 20 µg/mL pegvisomant for 1 h and then treated with 500 ng/mL GH for 24 h.

Fig. S4.

GH effects on colon cells is not mediated by IGF1. Western blot analysis of (A) IGF1 in colon hNCC and HCT116 cells treated with indicated doses of GH, (B) p53 in hNCC treated with indicated doses of IGF1 for 24 h, and (C) hNCC cells stably infected with lentivirus expressing shIGF-1 RNA or scramble (Scr) shRNA and treated with 100 ng/mL GH for 24 h. Experiments were repeated twice and representative blots are shown.

Fig. 5.

GH suppresses APC and increases β-catenin nuclear translocation. Western blot analysis of ERK/APC pathway in (A) hNCC treated with indicated doses of GH for 24 h and (B) hNCC stably expressing scramble (Scr) shRNA or p53shRNA. Experiments were repeated twice and representative blots shown. (C) Western blot analysis of hNCC stably infected with lentivirus expressing shp53 RNA or scramble (Scr) shRNA and untreated (C) or treated with GH (500 ng/mL). (D) Western blot analysis of cytoplasmic and nuclear fractions of hNCC untreated (C) or treated with GH (500 ng/mL) for 24 h. Experiments were each performed twice and representative results are shown. (E) Assay of canonical Wnt pathway signaling through activation of a TCF/LEF luciferase reporter construct (TOPFlash) or a control reporter (FOPFlash). hNCC were nucleofected with reporter constructs and treated with GH (500 ng/mL) for 24 h. Fold-induction was calculated as normalized relative light units of TOPFlash divided by those of FOPFlash. Results are mean ± SEM of triplicate measurements; *P < 0.05. The experiment was repeated four times with similar results and representative results shown. (F) Real-time PCR assessment of β-catenin target genes MMP2 and MMP9 expression in hNCC treated with GH (500 ng/mL) for 24 h. Normalized PCR results are expressed as fold-change vs. control taken as 1. Results are depicted as mean ± SEM of triplicates measurements. Two independent experiments were performed, and representative results depicted; *P < 0.05. (G) Western blot analysis of c-myc in hNCC treated with GH (500 ng/mL) for 24 h. In E and G, the differences between groups were analyzed using two-tailed unpaired Student t test.

Fig. S5.

GH induces nuclear β-catenin accumulation. Intensities of protein bands from Western blot analysis of cytoplasmic and nuclear fractions of hNCC untreated (control) or treated with GH (500 ng/mL) for 24 h quantified, normalized to β-actin, and depicted as percent from control, taken as 100%. The experiment was performed twice and mean results are shown.

Fig. S6.

GH induced nuclear β-catenin accumulation depends on p53. Western blot analysis of β-catenin in cytoplasmic and nuclear fractions of hNCC stably infected with lentivirus expressing shp53 RNA or scrambled shRNA (shScr) and untreated (C) or treated with GH (500 ng/mL). Experiments were repeated twice with similar results and representative blots are shown.

Fig. 6.

GH induces EMT, suppresses apoptosis, and increases motility. Western blot analysis of PTEN and EMT factors in (A) hNCC and (B) HCT116 cells treated with indicated doses of GH for 24 h. (C) Western blot analysis of cleaved caspase 3 in cells treated with GH (500 ng/mL) for 24 h. Experiments were performed at least three times, and representative results shown. (D) Migration of hNCC and HCT116 cells treated with GH (500 ng/mL) and harvested 48 h after plating. (E) Migration of HCT116 cocultured for 48 h with hCF infected with lentivector or lentiGH. In D and E, for quantification, the number of migrated cells per 1,000 cells in five randomly chosen fields in each duplicate transwell were counted and means calculated. Results are presented as mean ± SEM of three independent experiments; *P < 0.05. (F) Number of colonies and arbitrary colony size formed in soft agar by HCT116 cells cocultured with hCF infected with lentivector or lentiGH, and tested for anchorage independent growth. Colony size was determined using ImageJ software. Results are presented as mean ± SEM of duplicates from two independent experiments; **P < 0.01 vs. control. In D–F, the differences between groups were analyzed using two-tailed unpaired Student t test.

Fig. S7.

GH enhances cell motility. Migration of hNCC and HCT116 cells treated with GH (500 ng/mL) and harvested 48 h after plating. Representative images are shown. (Magnification, 10×.)

Fig. 7.

GH suppresses p53 in human intestinal organoids and in murine colon. (A) Western blot analysis of intestinal organoids treated with GH (500 ng/mL) for 48 h. (B) Athymic nude mice were injected subcutaneously with 500,000 HCT116 cells stably infected with lentimGH or lentiV and killed 5 wk after injection. (Upper) Western blot analysis of representative colon tissues derived from nude mice bearing mGH-secreting xenograft tumors. Four control and five experimental mice are shown here, and the rest of the animals are shown in Fig. S9A. (Lower) Intensities of protein bands from Western blot quantified, normalized to β-actin, and depicted as fold-change relative to control animals taken as 1. n = 9 mice/group, *P < 0.05. Differences between groups were analyzed using two-tailed unpaired Student t test. (C) A proposed model for colon p53-GH interaction. Local GH is induced as a result of DNA damage or inflammation. Circulating GH is induced as a result of acromegaly. High GH (circulating or local) suppresses p53, APC, PTEN, and apoptosis, and stimulates EMT, enabling a progrowth mucosal field change.

Fig. S8.

Three-dimensional intestinal organoids. (A) iPSC intestinal organoids are polarized and form crypt-like structures. (B) Organoids immunostained for phalloidin (red), a marker for actin fibers. (Magnification, 4×.)

Fig. S9.

Effects of high circulating GH in mice bearing mGH-secreting tumor xenografts. Athymic nude mice were injected subcutaneously with 500,000 HCT116 cells stably infected with lentimGH or lentiV and killed 5 wk after injection. n = 9 mice/treatment group. (A) Western blot analysis of p53 and p21 in representative colon tissues not shown in Fig. 7. (B) Serum GH and IGF1. *P < 0.05, **P < 0.01. (C) Western blot analysis of IGF1 and p53 in representative liver and colon (from Fig. 7) tissues.