REGγ Controls Hippo Signaling and Reciprocal NF-κB–YAP Regulation to Promote Colon Cancer

Colorectal cancer is one of the most commonly diagnosed cancers in both men and women. Multiple signaling pathways and predisposing factors are involved in the development of colorectal cancer, including colonic inflammation and YAP activation (1, 2). Patients with severe ulcerative colitis with hyperactive NF-κB have an increased risk of developing colorectal cancer (3). Aberrant regulation of NF-κB along with other signals is involved in colitis-associated carcinogenesis (4, 5). How inflammatory signals interact with the YAP pathway to promote colorectal cancer remains poorly understood. Our previous finding that the proteasome activator REGγ promotes experimental colitis encouraged us to explore REGγ functions further.

REGγ (also known as PA28γ), a 11s proteasome activator, promotes the degradation of multiple target proteins in an ubiquitin-independent manner in many cellular processes (6–10), including tumorigenesis and angiogenesis (11–13). In our recent study, REGγ was demonstrated to promote bowel inflammation by degrading IκBϵ and activating the NF-κB signal pathway (14). The strong correlation between YAP signaling and REGγ led us to investigate whether REGγ could be a driver for colorectal cancer development.

The Hippo signal pathway was first discovered in Drosophila fruit fly by mosaic genetic screens. The highly conserved mammalian Ste20-like kinases (Mst)1/2 and the effector, large tumor suppressor kinase (Lats)1/2, represent a core component of the Hippo pathway with the transcription coactivator Yes-associated protein (YAP) as a major downstream effector of the Hippo pathway. The Hippo–YAP signal pathway plays pivotal roles in tissue homeostasis, organ size, regeneration, and tumorigenesis (15, 16). Inactivation of YAP severely impairs dextran sulfate sodium (DSS)-induced intestinal regeneration, while hyperactivation of YAP results in polyp formation following DSS treatment (17). Deletion of Mst1/2 genes in the intestinal epithelium induces YAP hyperactivation, a phenomenon frequently found in human colonic cancer specimens (18). Despite this evidence of YAP as an oncogene (19), inactivation of YAP causes no obvious intestinal and colon defects under normal homeostasis, except alterations in colonic inflammation (17). YAP phosphorylation leads to its ubiquitination-dependent degradation (20).

In this study, we found that REGγ could activate YAP by degrading Lats1 to promote cell proliferation and colitis-associated colorectal cancer in mice. Stable silencing of REGγ in colon cancer cells retarded tumor growth, which was effectively reversed by activation of YAP. Overexpression of REGγ was found in human colon cancer samples, correlating with a poor prognosis for colon cancer patients. Prominently, reciprocal activation of the YAP and NF-κB pathways was found in human colon cancer cells with a prominent level of REGγ. Thus, REGγ plays a key role in colon cancer development via suppression of Lats1 to promote the action of YAP on NF-κB activity.

REGγ knockout mice were kindly provided by Dr. John J. Monaco at the University of Cincinnati and backcrossed for more than 10 generations in our specific pathogen-free animal center (21). Eight- to 10-week-old C57BL/6 male mice were used. Animals were maintained according to the ethical and scientific standards by the Animal Center at East China Normal University.

HCT116, HT29, HEK293T, MEF SV40, and HeLa cells were purchased from the ATCC or acquired from the Cell Culture Core at Baylor College of Medicine. The Core carries out regular examination of cell contamination or authentication. All the cells were maintained for no more than 50 passages before being thawed a new stock or acquire a low passage batch. All cells were grown in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS) and 100 μg/mL penicillin/streptomycin. HCT116 sh-N, HCT116 sh-R, HT29 sh-N, and HT29 sh-R cells were generated by retroviral shRNA vectors specific for REGγ or a control vector from OriGene. The 293-REGγ inducible cell line and MEF SV40 WT, MEF SV40 REGγ KO, HeLa sh-N, and HeLa sh-R cells were previously generated.

pCDNA3.1-Lats1, pCDNA3.1-YAP, pCDNA3.1-YAP(S127A), pCDNA3.1-HA/Flag-REGγ, and PCDH-YAPS127A were constructed in our laboratory. Lats1 siRNA (GGTTCTGAGAGTAAAATTATT) and YAP siRNA (GACAUCUUCUGGUCAGAGATT) were synthesized by Genepharma. Plasmids or siRNA were transfected to different cells and cultured for 36 or 72 hours by using transfection reagents (Cenji).

The antibodies used were β-Actin (Sigma), anti-REGγ (Invitrogen), anti-Lats1, anti-p-Lats1(S909), anti-Lats2, anti-YAP, anti-p-YAP (S127), anti-p-YAP (S397), anti-Mst1 (Cell Signaling Technology), anti-p53 (Santa Cruz Biotechnology), anti-p21 (BD Pharmingen), anti-Flag-mouse (MBL), anti-HA-mouse (Abmart), anti-p65 (Santa Cruz Biotechnology), and anti-p-p65(S536; Abcam). Drugs verteporfin (Sigma), human TNFα, and human IL6 (Novoprotein) were used.

Cells were lysed in cold RIPA buffer (50 mmol/L Tris–HCl pH 7.4, 1 mmol/L EDTA, 150 mmol/L NaCl, 1% NonidetP-40, 0.25% sodium deoxycholate, 0.1% SDS and protease inhibitors) and run on SDS-PAGE gels. After transferring to the nitrocellulose membrane, immunoblots were analyzed using the primary antibodies at 4°C overnight. They were then incubated 1 to 2 hours with fluorescent secondary antibodies (Jackson ImmunoResearch) and detected by using Odyssey CLx (LI-COR).

Cells were harvested in lysis buffer (50 mmol/L Tris–HCl pH 7.5, 5 mmol/L EDTA, 150 mmol/L NaCl, 1% TritonX-100, 1 mmol/L Na₃VO₄, 5 mmol/L NaF) and centrifuged for 15 minutes at 10,000× g to discard the insoluble debris. Cell lysate was immunoprecipitated with anti -Flag-M2 agarose or anti-HA agarose beads for 4 to 8 hours at 4°C. Immunoprecipitates were washed three times with wash buffer (50 mmol/L Tris–HCl pH7.5, 5 mmol/L EDTA, 150 mmol/L NaCl, 1 mmol/L Na₃VO4, 5 mmol/L NaF) followed by Western blot analysis.

Cells were treated with TNFα or/and VP (verteporfin) beforehand. The detailed steps are described in ref. 22. The primer sequences are described in Supplementary Table S2.

Recombinant REGγ protein was purified in our laboratory as described previously (13). The substrate proteins were generated by in vitro translation. The protein decay experiment was conducted by incubating substrate, 20S proteasome (Boston Biochem) and purified REGγ for 1 to 3 hours in 25 μL reaction volume at 30°C with appropriate controls. The results were analyzed by Western blotting.

Cells were seeded on coverslips in 24-well plates, were washed in cold PBS three times, fixed in 4% formaldehyde, and immunostained by monoclonal anti-YAP antibodies, as well as DNA staining with 4,6-diamidino-2-penylindole (DAPI). The Alexa Fluor 546 (red) goat anti-rabbit antibody (Molecular Probes) was used for YAP. Immunofluorescence was visualized by confocal microscopy (Leica TCS SP5).

Human samples were provided by Xinhua Hospital Affiliated to Shanghai Jiaotong University. Patient organization and case access are in line with the ethical requirements of University Committee on Human Research Protection in East China Normal University.

Tissues were embedded in paraffin and then cut into different sections (4–5 μm thick). Immunohistochemical staining was scored according to the following standards: staining intensity (I) was classified as 0 (lack of staining), 1 (mild staining), 2 (moderate staining) or 3 (strong staining); staining percentage (P) was designated as 1 (<25%), 2 (25%–50%), 3 (51%–75%), or 4 (>75%). For each section, the semiquantitative score was calculated by multiplying I and P (which ranged from 0 to 12). Score 0–3 was as not significant (negative), 4–8 as weakly positive and 9–12 as strongly positive. In the analysis, low expression meant negative or weakly positive, high expression meant strongly positive. The log-rank test was performed to assess statistical significance.

Total RNA was extracted from cells or pulverized colons (liquid nitrogen treatment) using TRIzol (TakaRa). Two micrograms of total RNA was reverse transcribed in a total volume of 20 μL. Aliquots of the RT products were used for quantitative RT-PCR analysis with Mx3005P (Stratagene). Each experiment was repeated three times. Primer sequences are described in Supplementary Table S1.

The cells were washed with cold PBS three times 24 hours after transfection and harvested in the lysis buffer provided with the Luiferase Assay Kit (Promega). After one cycle of freezing and thawing, the cell lysates were centrifuged in 4°C at 12,000× g for 10 minutes. Then, 20 μL of supernatant was added to equal amount of luciferase assay substrate, with two repeats. Luminescence was measured as relative light units using the LUMIstar OPTIMA (BMG Labtech) illuminometer. Each individual experiment was repeated for three times.

Logarithmic phase cells were seeded in 96-well plates at 2.5 × 10³ cells per well and incubated for 24 hours. Cells were incubated with 0.5 mg/mL MTT for 4 hours and add DMSO for 15 minutes. Absorbance (490 nm) was measured and analyzed as described (13).

Twenty-one samples (healthy control, n = 11; ulcerative colitis (UC), n = 10) were chosen from the NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo) to detect the correlation of mRNA expression between the REGγ and Hippo pathways; the accession number is GSE10616. Another 53 UC data and 68 colorectal cancer data were used to investigate the correlation of REGγ, p-65, and YAP. Raw data were downloaded from GSE3629.

HCT116 sh-N and sh-R stable cells overexpressing YAP (S127A) cell lines were constructed with PCDH-YAPS127A. Eight-week-old BALB/c male nude mice were used. Cells were implanted into the dorsal flanking sites of nude mice at 2 × 10⁶ cells in 200 μL per spot. Thirty days after injection, mice were examined, and tumor volumes were measured three times. Tumor volume = 1/2(length × width²).

Colon inflammation was induced by treatment with 2% (w/v) DSS (MP Biomedicals) in drinking water for 3 to 5 days. For the colorectal cancer model, mice were given intraperitoneal injection of azomethane (Sigma) at 10 mg/kg. After 7 days, 2% DSS was given to mice in drinking water for 7 days, followed by normal water for 7 days; this process was repeated three times. Then, mice were given normal water until sacrificed. Four weeks after completing the DSS treatment, tissues were collected for the followed experiments.

Quantitative data were displayed as mean ± SEM of independent samples by using Prism software (GraphPad Software). Statistical analysis of values was performed using ANOVA. P values < 0.05 were considered significantly different.

Given that patients with ulcerative colitis have an increased risk of developing colorectal cancer, our discovery of REGγ as an activator of bowel inflammation prompted us to investigate its link with human colon cancer. Bioinformatics analyses revealed a strong correlation between REGγ expression and key components in the Hippo–YAP pathway (Fig. 1A). With the hypothesis that REGγ might be a positive regulator of YAP signaling, we generated stable REGγ knockdown (sh-REGγ), using a previously well-defined shRNA (23), in HCT116 and HT29 human colon cancer cell lines to determine whether REGγ can modulate the Hippo–YAP activities. Silencing REGγ in HCT116 and HT29 cells resulted in a significant increase in the Lats1 and p-YAP levels (Fig. 1B). Immunofluorescence with antibody against total YAP revealed an increased nuclear localization of YAP in REGγ-high HCT116 cells (sh-N), whereas a more dispersed YAP staining with less nuclear intensity was found in REGγ-deficient (sh-R) HCT116 cells (Fig. 1C). Quantitative analysis indicated a 30% decrease in the ratio of nuclear localization in REGγ silenced sh-R cells compared with the REGγ sh-N cells (Fig. 1D), suggesting a positive link between REGγ and YAP. We then measured transcriptional activities of YAP in cells with varying REGγ levels. There was a significant reduction in the expression of connective tissue growth factor (CTGF), cysteine-rich angiogenic inducer 61 (Cyr61), and amphiregulin (AREG) genes, positively regulated by YAP (24, 25), in REGγ-deficient sh-R HCT116 cells; however, genes DDIT4 and Trail, negatively regulated by YAP (26), had higher expression in the sh-R cells (Fig. 1E).

Considering that the stability of YAP is also regulated by Lats1-independent phosphorylation at S397 (27), we examined if REGγ may affect its level in colon cancer cells. As expected, expression of p-YAP (S397) in colon cancer cells and tissues was not altered when REGγ was deleted (Supplementary Fig. S1B and S1C). Unlike Lats1, Lats2 had no change in REGγ knockdown cells (Supplementary Fig. S1A). The upstream kinase of Lats1 and Lats2, Mst1, was unchanged in either sh-N or sh-R colon cancer cell lines (Supplementary Fig. S1D). These data indicate that REGγ may negatively regulate Lats1 to activate YAP in human colon cancer cells and modulate YAP downstream gene expression. Also, we observed that REGγ deletion resulted in a significant increase of Lats1 and p-YAP in HeLa, but not in REGγ^−/− MEFs (Supplementary Fig. S1E) probably due to cell specificity.

To determine a possible role for REGγ in degradation of Lats1, the degradation dynamics of Lats1 in the presence of cycloheximide, a protein synthesis inhibitor, in REGγ-deficient sh-R cells and the control cells (sh-N) was measured. Lats1 decayed much slower in the REGγ-depleted HT29 or HCT116 sh-R cells than it did in the sh-N cells (Fig. 2A and B; Supplementary Fig. S2A), indicating that REGγ affects the stability of Lats1 in these cells. Lats1-dependent p-YAP (S127) displayed an increase of stability in REGγ-knockdown cells (Fig. 2A and B; Supplementary Fig. S2A) while Lats1-independent p-YAP (S397) stability was not regulated by REGγ (Supplementary Fig. S2C and 2D). The mRNA levels of Lats1, evaluated by semiquantitative RT-PCR, had no changes in these cells regardless of REGγ abundance (Fig. 2C), suggesting a posttranscriptional regulatory mechanism.

We carried out gain-of-function experiments by using doxycycline-inducible 293 cells, a previously well-defined cellular system for conditional study of REGγ functions (6), that can overexpress either wild-type REGγ or a dominant-negative mutant REGγ unable to activate the 20S proteasome. As expected, induction of REGγ triggered degradation of Lats1 (Supplementary Fig. S2B), whereas induced expression of the REGγ-N151Y dominant-negative mutant failed to promote degradation of endogenous Lats1 as compared with the wild-type REGγ control.

Next, we analyzed if REGγ and Lats1 bind to each other intracellularly by reciprocal coimmunoprecipitation experiments. We found physical interactions between endogenous REGγ and Lats1, respectively (Fig. 2D and E). To elucidate if the effect of REGγ on Lats1 degradation is direct, we used cell-free proteolysis. Incubation of in vitro translated Lats1 with 20S proteasome or purified REGγ alone exhibited no significant degradation of Lats1, but a combination of REGγ and 20S proteasome promoted much faster turnover of Lats1 than did the 20S proteasome alone (Fig. 2F). Taken together, our results demonstrate a direct role of REGγ in degradation of Lats1 in vitro and in cells.

The finding that Lats1 is a target of REGγ-proteasome led us to verify the causal relation and specificity in the Hippo–YAP signaling. We transiently silenced Lats1 in sh-N and sh-R HCT116 cells and assessed expression of YAP downstream genes by real-time RT-PCR. While REGγ-depleted sh-R cells had reduced expression of AREG, CTGF, and Cyr61, silencing Lats1 markedly increased their expressions (Supplementary Fig. S3A). The expression of Trail, a gene negatively regulated by YAP, was downregulated with RNA interference of Lats1 in REGγ-depleted sh-R cells (Supplementary Fig. S3A and S3B). In gain-of-function experiments, exogenous Lats1 was introduced to sh-N and sh-R HCT116 cells with viral infection followed by evaluation of AREG, CTGF, Cyr61, and Trail expression. Lats1 overexpression drastically repressed transcription of AREG, CTGF, and Cyr61 in sh-N HCT116 cells (Supplementary Fig. S3C and S3D). In contrast, overexpressed Lats1 greatly enhanced Trail levels in sh-N HCT116 cells (Supplementary Fig. S3C and S3D). We also checked AREG, CTGF, and Cyr61 expressions in sh-N and sh-R HT29 cells with siLats1 or overexpression of Lats1. A similar result of REGγ-dependent YAP regulation was observed in HCT116 cells (Supplementary Fig. S3E–S3H). Therefore, we conclude that REGγ-mediated regulation of YAP activity is Lats1 dependent.

With the findings that REGγ drives the tumor promoter YAP (15) and hyperexpression of REGγ exists in human colorectal cancer, we investigated the role of the REGγ–YAP pathway in colon cancer cell growth and experimental tumor formation. In both HCT116 and HT29 cells, cell viability assays were performed to demonstrate that depletion of REGγ markedly attenuated the growth of these human colon cancer cells 4 days after seeding (Supplementary Fig. S4A and S4B). Silencing YAP had a similar effect to REGγ depletion on cell growth (Supplementary Fig. S4A and 4B). However, Lats1 knockdown increased cell viability in HCT116 and HT29 cells (Supplementary Fig. S4C), indicating Lats1-dependent effects on cell growth. We established a murine colitis-associated cancer model by injecting a single dose of DNA-methylating agent azoxymethane followed by three rounds of 2% dextran sodium sulfate (DSS) treatment (1 week each round) at intervals of 2 weeks. Colon tumors developed in the REGγ wild-type mice were significantly more severe than those in the REGγ^−/− mice (Fig. 3A). We found that REGγ^+/+ mice developed 13.5 tumors per mouse while REGγ^−/− mice had only 6.8 (Fig. 3B). The average tumor size in WT mice was 6.5 mm in diameter, while it was 2.1 mm in REGγ^−/− mice (Fig. 3C). Expression of proliferating cell nuclear antigen (PCNA) at the protein and mRNA levels was much higher in tumor tissues from REGγ^+/+ mice than from REGγ^−/− mice (Supplementary Fig. S4D and S4E). In line with molecular analysis, immunohistochemical staining with anti-PCNA demonstrated significantly higher numbers of PCNA-positive cells in colon tumors from WT mice than in REGγ^−/− mice (Supplementary Fig. S4F). These findings indicate that lack of REGγ inhibits the proliferation of human colon cancer cells and the progression of colitis-associated tumor formation.

We compared the variation of YAP signal–related molecules in colon tumors from REGγ^+/+and REGγ^−/− mice. While Lats1 and p-YAP protein levels in REGγ^−/− tumors were considerably higher than those in REGγ^+/+ tumors, YAP levels were significantly lower in REGγ^−/− tumors by Western blot analysis (Fig. 3D), likely due to sustained increase of REGγ in tumor areas. Immunohistochemical studies supported the observation of elevated Lats1 and p-YAP in REGγ-high tumors versus reduced total YAP in REGγ-defective tumors (Fig. 3E). By RT-PCR, expression of genes (AREG, CTGF, and Cyr61) positively regulated by YAP was significantly reduced in REGγ^−/− tumors; however, transcription of genes (DDTI4 and Trail) regulated by YAP as a suppressor was greatly reinforced in REGγ-deficient tumors (Fig. 3F), exhibiting a positive correlation between REGγ and YAP activity in the development of experimental colon tumors.

REGγ is known to regulate multiple tumor suppressors, and lack of REGγ is associated with tumor suppression (6, 7, 13, 14). To determine if activation of YAP signaling in sh-R HCT116 cells may override tumor suppressive functions with REGγ deficiency, we generated a constitutively active YAP (S127A) construct (28) and stable cell lines expressing activated YAP were created in sh-N and sh-R HCT116 cells. Cells expressing active YAP or control vectors were injected into dorsal sides of nude mice as previously described to generate xenograft tumors (23). Mice with REGγ-defective sh-R HCT116 cells had smaller tumor loads than mice with normal REGγ (Supplementary Fig. S5A), substantiating the proliferative action of REGγ on colon cancer cells. In contrast, expressing constitutively active YAP in sh-R HCT116 cells restored tumor size like that derived from sh-N HCT116 cells (Supplementary Fig. S5A–S5C). These results suggest that active YAP, acting downstream of REGγ, can fully overcome the tumor suppressive effect by REGγ depletion in colon cancer cells. Therefore, we conclude that the REGγ–YAP pathway promotes colon cancer cell proliferation and tumor growth.

An interesting observation was that the differences of Lats1 and YAP levels in nontumor colon tissues between REGγ^+/+ and REGγ^−/− mice were boosted in tumor samples (Fig. 3D). These data raise a possibility that Hippo–YAP activity was further enhanced by cross-talk with other signals regulated by augmented REGγ during tumorigenesis. The fact that REGγ promotes both NF-κB (14) and YAP signaling led us to investigate whether REGγ promotes cross-talks between the NF-κB and YAP pathways.

First, we inspected whether the inflammatory microenvironment, such as activation of NF-κB, may promote YAP activity in human colon cancer cells. HCT116 and HT29 human colon cancer cells treated with TNFα or IL6 induced marked increase of YAP transcription (Fig. 4A; Supplementary Fig. S6A). Chromatin immunoprecipitation (ChIP) assays showed recruitment of p65 to the YAP1 promoter following TNFα incubation (Fig. 4B), suggesting a direct role of p65 in YAP transcription. In addition, TNFα stimulation enhanced YAP1 binding to the promoter of its downstream gene, Cyr61 (Supplementary Fig. S6B). Activation of NF-κB by TNFα or IL6 significantly induced expression of AREG, CTGF, and Cyr61, but attenuated transcription of Trail and DDIT4 in HCT116 or HT29 cell lines (Fig. 4C and D; Supplementary Fig. S6C and S6D), establishing a positive regulation by the NF-κB signal on the YAP pathway. To determine whether inflammation influenced YAP activation, mice were treated with DSS for 5 days and analyzed for the expression of YAP target genes in colon epithelium. RT-PCR results demonstrated that an inflammatory microenvironment, as evidenced by elevation of NF-κB target genes (Supplementary Fig. S6F), strikingly correlated with an increase in the expression of YAP (Fig. 4E) and its target genes ARGE, CTGF, and Cyr61 (Fig. 4F; Supplementary Fig. S6E), an indication of YAP activation.

Next, we analyzed if YAP activation may regulate NF-κB signaling. To our surprise, transcription of p65 was significantly enhanced by expression of a constitutively active YAP protein (YAP-S127A) but attenuated by a YAP inhibitor (Fig. 5A; Supplementary Fig. S7A). ChIP analysis with an anti-YAP1 antibody revealed YAP1 enrichment at the promoter region of p65 (Fig. 5B), suggesting YAP1 can directly promote the transcription of p65 in human colon cancer cells. To validate our observation, an NF-κB luciferase reporter assay was performed in the presence or absence of active YAP. Constitutively active YAP (S127A) strongly enhanced NF-κB activity in HCT116 or HT29 cells (Fig. 5C; Supplementary Fig. S7B). Moreover, we applied a YAP inhibitor to TNFα- or IL6-treated cells, followed by assessing transcription of NF-κB target genes. Consistently, TNFα or IL6 treatment led to a robust increase in transcription of IL8 and IL6, downstream genes of NF-κB, whereas IL8 and IL6 expression were blocked by YAP inhibitor verteporfin (VP) in HCT116 or HT29 cells (Fig. 5D and E; Supplementary Fig. S7C and S7D). To rule out the potential off-target effect by the YAP inhibitor, we used siRNA against YAP to elucidate its function in cross-talk with NF-κB. Antagonizing YAP function markedly attenuated TNFα- or IL6-induced expression of IL8 and IL6 (Fig. 5F and G; Supplementary Fig. S7E and S7F) in two different colon cancer cell lines. Consistently, ChIP assays indicated that a YAP inhibitor effectively blocked TNFα-induced p65 binding to IL8 promoter (Fig. 5H). Thus, the YAP signal pathway could intensify the activity of NF-κB in human colon cancer cells.

Clearly, REGγ-deficient cells had impaired responses to TNFα or IL6 treatments (Figs. 4C–D and 5D–G). Moreover, the reciprocal effects of NF-κB and YAP signals were simultaneously suppressed in cells lacking REGγ (Figs. 4C and D and 5D–G), indicating that REGγ modulates the interaction between the NF-κB and YAP pathways. Collectively, our study identifies a previously unknown mechanism by which the REGγ proteasome, a new regulator of the Hippo–YAP pathway, promotes a positive feedback regulation between YAP and NF-κB to coordinate the growth of inflammation-associated colon tumor.

To delineate the clinical implication of REGγ–YAP–NF-κB signaling in human colon cancer, a cohort of 68 human colorectal cancer samples and 53 UC samples obtained from the dataset GSE3629 was analyzed for correlation of gene expression. Strong positive correlations for REGγ–YAP1, REGγ–RELA (p65), and YAP1–RELA were observed in colorectal cancer tissues (Fig. 6A) as well as in UC tissues (Supplementary Fig. S8A).

We analyzed 172 cases of Chinese human colon tumors with matched nontumor controls by immunohistochemistry. REGγ was highly expressed in over 60% human colon cancer samples in all stages compared with paired nontumor tissues (Fig. 6B; Supplementary Fig. S8D). Approximately 53% of samples had high expression of YAP and 76% cases had positive p-p65 staining (Fig. 6C; Supplementary Fig. S8B). While Lats1 and p-YAP levels were lower in tumor samples compared with nontumor controls, total YAP exhibited a high correlation to REGγ levels (Supplementary Fig. S8C). REGγ in tumor specimens displayed a strong positive correlation with YAP and a negative association with Lats1 and p-YAP (Fig. 6B; Supplementary Fig. S8C), consistent with our observation in vitro and in animal studies.

Among the 172 colon cancer cases with complete clinical records and postoperative follow-up for more than 2 years, Kaplan–Meier analysis indicated a significantly lower survival rate in REGγ-high patient groups than in REGγ-low patients (Fig. 6C, top). Importantly, overexpression of REGγ was not correlated with age, sex, or tumor size (Supplementary Fig. S8D). Meanwhile, high expression of YAP was also associated with a marked reduction in survival rate (Fig. 6C, middle). Correlation of a lower survival rate with p-p65 overexpression appeared to have a trend, although it was not statistically significant (Supplementary Fig. S8B). In patients with high expression of REGγ/YAP and positive p-p65, the survival rate was further reduced compared with the other cases (Fig. 6C, bottom). Taken together, we conclude that REGγ overexpression is likely a marker for poor prognosis of human colon cancers.

It is thought that YAP is constitutively inhibited by the Hippo kinase cascade, which, in epithelia, is activated by physiological cell–cell contact during normal homeostasis, but intestinal damage increases YAP abundance and nuclear residence dependent on Hippo (19). However, it is still unclear how Hippo signal molecules may be regulated in response to inflammation. In this study, we have shown that Hippo kinase signal transduction is inhibited by REGγ proteasome–dependent degradation of Lats1 to enhance oncogenic YAP in colon cancer cells (Supplementary Fig. S9). As a result, REGγ promotes colon cell proliferation in vitro and in vivo in a YAP-dependent manner. Furthermore, we have discovered a REGγ-dependent positive feedback loop for reciprocal regulation between the YAP and NF-κB signaling, suggesting that cancer cells hijack proteasome degradation, inflammation, and YAP pathways (Supplementary Fig. S9). Corresponding with these findings, overexpression of REGγ is found in over 60% of human colon cancer samples, correlated with elevated levels of YAP/NF-κB in those samples, and associated with a poor prognosis, suggesting that REGγ is a master regulator for the development of human colon cancer.

REGγ-mediated degradation of Lats1 but not Lats2 or Mst1/2, in epithelial cells but not in MEFs, reflects a specific mechanism, suggesting additional layers of inhibitory regulation by REGγ in the Hippo kinase cascade, activated by physiological cell–cell contact during normal homeostasis in epithelia, but disturbed upon intestinal damage (19). In fact, REGγ is transcriptionally activated by NF-κB upon colon epithelial damage during experimental colitis (14). Elevated REGγ in colon epithelial cells not only exacerbates inflammation but also bridges activation of the YAP pathway by proteasome-dependent degradation of Lats1, explaining in part how disrupted cell–cell contact by the inflammatory microenvironment may lead to loss of Lats1 and increased residence of nuclear YAP.

REGγ-triggered negative regulation of Lats1 and positive regulation of YAP are shown here via loss-of-function and gain-of-function experiments. We verified that Lats1 is indeed a direct target of the REGγ proteasome in a cell-free in vitro degradation system, where no ATP or ubiquitin is required. REGγ depletion results in differential regulation of total YAP and phosphorylated YAP, likely due to quick degradation of phosphorylated YAP by ubiquitin proteasome pathway in the cytosol. The physiological functions of the noncanonical REGγ proteasome in Lats1 degradation appear to maintain the homeostasis of Lats1 for rapid proliferative needs of colon epithelial cells, whereas the traditional ubiquitin–proteasome system may execute transient and robust clearance of Lats1 upon oncogenic signals. Changes in oncogenic p53 mutation, loss of TGFβ components, or increased NF-κB activities can dramatically promote REGγ levels (14, 22), leading to pathological activation of YAP. Expressing constitutively active YAP completely abrogates inhibition of tumor growth in REGγ-depleted colon cancer cells in mouse models. However, no striking increase in p53 expression was found after REGγ deletion in vitro or in vivo (Supplementary Fig. S10A and S10B), indicating that p53 might play a minor role in this context.

Accumulating evidence indicate that YAP is overexpressed in human colon cancers and most colon cancer–derived cell lines (2, 18, 19). Hyperactivation of YAP results in widespread early-onset polyp formation following DSS treatment (3). Co-overexpression of YAP and transcriptional coactivator with PDZ-binding motif (TAZ) is proposed as an independent predictor of prognosis for patients with colorectal cancer (2). In this study, we have also found significant elevation of YAP in both mouse colitis–associated colon tumor and human colorectal cancer tissues, which is significantly associated with a poor prognosis in human colorectal cancer patients. There is a strong positive correlation between overexpression of REGγ and total YAP but not phosphorylated YAP, with concomitant reduced Last1 in tumor tissues. Perhaps phosphorylated YAP is quickly degraded by the ubiquitin proteasome pathway in the cytosol.

Although YAP functions as an oncogenic factor, a second signal pathway provided by tissue damage or inflammation is required for growth-promoting function of YAP in hepatocytes (29). Many studies have revealed the connection between colon carcinogenesis and inflammation, in which NF-κB is a central player (30, 31). In this study, we have found that NF-κB is able to enhance YAP transcription and its activity in human colon cancer cells and mouse-colitis model, providing a direct link between inflammatory signals and activation of YAP. Meanwhile, YAP can bind the p65 promoter to enhance NF-κB signaling, reinforcing a positive feedback between NF-κB and YAP signals during colon tumor formation. Our findings are supported by a previous high-throughput ChIP-seq study suggesting recruitment of NF-kB to YAP1 promoter in head and neck squamous carcinoma cells (32).

As a critical transcriptional mediator of inflammatory responses, p65 is likely to play a key role in the initiation of cancer formation. With the notion that NF-κB enhances REGγ transcription during colonic inflammation (14), there is a cross-talk network among REGγ, NF-κB, and YAP signals, in which REGγ appears to be the hub. We believe the REGγ-regulated network is much more complicated than we have mentioned here and illustrated in Supplementary Fig. S9. For instance, we have previously found REGγ-dependent regulation of Wnt signaling (10), which is important for the development of colon cancer (33). Wnt/β-catenin signaling has been reported to regulate transcription of YAP in colorectal carcinoma cells (34) and nuclear accumulation of YAP correlates with β-catenin activation (18). Future studies will integrate all these signal pathways to elucidate REGγ functions in colorectal cancer carcinogenesis.

No potential conflicts of interest were disclosed.

Conception and design: Q. Wang, X. Gao, W. Wang, J. Fu, J. Xiao, Y. Dang, X. Li

Development of methodology: Q. Wang, X. Gao, F. Chen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Q. Wang, X. Gao, T. Yu, W. Wang, W. Zhou, Q. Huang, L. Cui

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Q. Wang, X. Gao, T. Yu, L. Yuan, J. Dai, G. Chen, J. Yang, X. Li

Writing, review, and/or revision of the manuscript: Q. Wang, X. Gao, R.E. Moses, J. Fu, L. Li, Y. Dang, X. Li

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Q. Wang, X. Gao, C. Jiao, P. Zhang, Y. Dang

Study supervision: Y. Dang, X. Li

This work was supported by the National Basic Research Program of China (2015CB910402 and 2016YFC0902102), the Science and Technology Commission of Shanghai Municipality (17ZR1407900, 14430712100, and 14ZR1411400), the National Natural Science Foundation of China (91629103, 81471066, 31401012, 31200878, 81672883, and 81401837), Shanghai Rising-Star Program (16QA1401500), and the Applied Basic Research Program of Science and Technology Department of Sichuan Province (2015JY0038).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.