Local Growth Hormone Facilitates Aging of the Colon Epithelial Microenvironment

Abstract

Aging is associated with the appearance of senescent cells secreting the senescence-associated secretome, facilitating a milieu favoring age-related microenvironmental changes. As we previously showed the production of local nonpituitary growth hormone (npGH) in senescent colon epithelial cells, we now elucidate mechanisms underlying npGH action in the nontumorous colon tissue microenvironment. We demonstrate autocrine npGH action in normal human colon cells (hNCC) infected with lentivirus-expressing hGH (lentiGH), as well as paracrine npGH action in hNCC cocultured with lentiGH hNCC and in intact human 3-dimensional intestinal organoids cocultured with organoids infected with lentiGH. Enriched gene ontology and pathway analysis of intact organoids exposed to paracrine npGH identified distorted extracellular matrix (ECM) and focal adhesion pathways concurrent with altered expression of ECM and cytoskeletal proteins. Significant phosphoprotein changes associated with the cytoskeleton and cell migration pathway occurred in GH-exposed hNCC. Paracrine npGH triggers these changes by activating epithelial-mesenchymal transition, as shown by suppression of E-cadherin and induction of Twist2 in cellular models, as well as in the colon of nude mice inoculated with GH-secreting xenografts. These changes are consistent with observed increased migration of hNCC overexpressing lentiGH, or in those cocultured with GH-secreting hNCC or with GH-secreting normal colon fibroblasts. Furthermore, whole exome sequencing detected increased structural variation in intact organoids cocultured with lentiGH-infected organoids, likely as a consequence of GH-mediated suppressed DNA damage repair, thereby favoring cell transformation. Our results indicate that local growth hormone facilitates aging of the colon epithelial microenvironment.

Keywords: aging; epithelial‐mesenchymal transition; extracellular matrix; growth hormone.

FIGURE 1

The npGH is expressed in senescent cells, and autocrine and paracrine GH promote β‐catenin nuclear translocation. (A‐B) Representative confocal images of npGH expression in senescent cells in (A) normal colon and (B) hyperplastic colon polyp. SA‐β‐gal, green; GH, red; colocalization, yellow. Scale bar = 100 μm. (C) Representative confocal image of hNCC expressing lentiGH. Arrow indicates GFP‐positive cell expressing GH (green) and active β‐catenin (red). Arrowhead indicates the cell cultured in proximity of a GH‐expressing cell also expressing intranuclear active β‐catenin. Scale bar = 20 μm. (D‐F) Representative cross‐sections of confocal Z‐stacks of hNCC showing localization of active β‐catenin (red) and nucleus (blue). (D) Active β‐catenin is outside the nucleus in GFP‐positive lentiV hNCC, (E) in the nucleus of GFP‐positive lentiGH hNCC, and (F) in the nucleus of GFP‐negative hNCC cocultured with lentiGH hNCC.

FIGURE 2

Autocrine and paracrine GH trigger EMT and β‐catenin nuclear translocation. (A‐E) Autocrine GH triggers EMT and β‐catenin nuclear translocation. (A) Cytoplasmic (CF) and nuclear (NF) fractions of hNCC infected with lentiV or lentiGH and analyzed 7 days after infection. ImageJ quantification of protein expression is depicted on Figure S2. (B) EMT markers in hNCC line #1 and line #2 infected with lentiV or lentiGH. (C) Assay of canonical Wnt pathway signaling through activation of a TCF/LEF luciferase reporter construct (TOPFlash) or a control reporter (FOPFlash). hNCC infected with lentiV or lentiGH were nucleofected with reporter constructs. Luciferase activity was normalized to pRL‐TK Renilla Luciferase Control Reporter Vector. Relative luciferase activity was calculated as the ratio between light units of TOPFlash divided by those of FOPFlash. (D) Migration of hNCC line#1 and line #2 infected with lentiV or lentiGH. (E) Invasion of hNCC line #1 infected with lentiV or lentiGH. Graph depicts duplicates of two or three experiments. Microscope images are depicted in Figure S2D,E. (F–J) Paracrine GH triggers EMT and β‐catenin nuclear translocation. (F) Cytoplasmic and nuclear fractions of hNCC cocultured with hNCC infected with lentiV or lentiGH and analyzed 7 days later. (G) EMT markers in hNCC cocultured with hNCC infected with lentiV or lentiGH (left) or treated with 500 ng/mL GH (right) for 24 h. (H) RT‐PCR of MMPs in hNCC cocultured with hNCC infected with lentiV or lentiGH for 1 month. (I) Migration of hNCC line#1 and line#2 cocultured with hNCC infected with lentiV or lentiGH. (J) Invasion of hNCC line#1 cocultured with hNCC infected with lentiV or lentiGH. In I, J graphs present duplicates of three experiments. In D, E, I, J results are depicted as mean ± SEM. *p < 0.05, **p < 0.01. Microscope images of migration and invasion are depicted in Figure S4E,F. ImageJ quantifications of protein expression are depicted in Figure S2A–C and Figure S4A,B.

FIGURE 3

Paracrine GH induces EMT markers in human colon fibroblast, organoids, and murine colon. (A) Twist2 expression in intact hNCC cocultured with hCF infected with lentiV or lentiGH. (B) EMT markers in organoids cocultured for 1 month with organoids expressing lentiV or lentiGH. (C) Migration and invasion of hNCC cocultured with hCF infected with lentiV or lentiGH. Graph presents duplicates of two experiments. Microscope images of migration and invasion are depicted in Figure S5. (D) EMT markers in the colon tissue of mice carrying GH‐secreting or vector‐secreting xenografts. ImageJ quantification of protein expression for B and D is depicted Figure S6.

FIGURE 4

(A–D) Paracrine GH alters ECM organization, cell adhesion and cytoskeleton proteins. Three lines of organoids were infected with lentiV or lentiGH, both expressing GFP, and cultured for 5 weeks then sorted for GFP‐positive (expressing either lentiGH or lentiV) and GFP‐negative intact neighboring cells. GFP‐negative cells cocultured with GFP‐positive lentiV or lentiGH organoid cells were compared. (A) Reactome analysis of RNA‐seq results showed ECM organization pathway was most affected by paracrine GH. (B) ECM organization genes (Reactome). (C) RT‐PCR validation of ECM organization pathway genes in three organoid lines. (D) RT‐PCR validation of genes involved in ECM adhesion in three organoid lines. (E‐F) KEGG analysis of RNA‐seq results (see Figure S8B). (E) Representative color‐coded heat map reflecting fold‐change of cell adhesion gene expression between organoid cells cultured in the presence of lentiV versus cells cultured in the presence of lentiGH. (F) Validation of adhesion gene expression by RT‐PCR. Normalized averaged PCR results from three organoid lines are expressed as fold‐change versus control. In C, D, F results are depicted as mean ± SEM. *p < 0.05, **p < 0.01. Expression of CLDN3 and THBS1 is depicted in C and D. (G) Cytoskeleton proteins in organoid cells cultured in the presence of lentiV or lentiGH. (H) Cytoskeleton proteins in the colon tissue of WT and GHRKO mice. ImageJ quantification of protein expression is depicted in Figure S9A,B.

FIGURE 5

Paracrine GH alters phosphorylation of cytoskeleton and cell migration proteins and suppresses myosin expression. hNCC were treated with 500 ng/mL GH then analyzed for phosphoproteomic changes. Depicted results are from three independent biological replicates. (A) Over‐representation analysis of human phenotype ontology database depicting pathway changes after 6 h of treatment. (B, C) Volcano plots after 24 h of treatment depicting log2 ratios of peptides that belong to (B) cytoskeleton rearrangement and (C) cell migration pathways, plotted against the negative log10 of the p value of their fold change. Peptides above −log10 = 1.35 (shown in red) corresponding to p < 0.05 are considered statistically significant. (D) Myosin expression 24 h after treatment with GH. ImageJ quantification of protein is depicted in Figure S9C.

FIGURE 6

Paracrine GH induces chromosomal instability increasing structural variants. The experiment is described in Figure 4. (A) Bar charts of SV type across GH‐affected (GH, n = 3) and vector‐affected (n = 3) samples after whole‐genome sequencing. BND, breakend; DEL, deletion; DUP, duplication. (B) Distribution of protein‐level effect types for SV. (C) Distribution of protein‐level effect types for SNP/Indels. (D) Selected GH‐unique high‐impact variants subtypes with pathway enrichment analysis of 679 genes. (E) Mucin genes from “Defective GALNT12 causes CRCS1” pathway with genomic position of SNP/Indel, dbSNP id where given, associated gene, and Phred score.