Ascl2 activation by YAP1/KLF5 ensures the self-renewability of colon cancer progenitor cells

Achaete scute-like 2 (Ascl2) is the Wnt signaling target, its regulation by other signaling is undefined. Now we demonstrated that CD133+/CD44+ cell population from HT-29 or Caco-2 cells exhibited cancer stem cell (CSC) properties with highly expressed Ascl2, which is related to the Hippo signaling pathway. YAP1 interference in CD133+/CD44+ HT-29 or Caco-2 cells reduced their proliferation, colony-forming ability and tumorsphere formation in vitro and inhibited the ‘stemness’-associated genes and Ascl2 expression. Enforcing YAP1 expression in HT-29 or Caco-2 cells triggered the opposite changes. Ascl2 interference reversed the phenotype of YAP1-enforced expressed HT-29 or Caco-2 cells. Krüppel-like factor 5 (KLF5) protein, not KLF5 mRNA levels, were increased due to YAP1 overexpression which is reported to prevent KLF5 degradation. Co-immunoprecipitation (Co-IP) assays demonstrated that YAP1 bound with KLF5 in HT-29 and Caco-2 cells. Luciferase and chromatin immunoprecipitation (ChIP) assays indicated that both YAP1 and KLF5 bound to the first two loci with GC-boxes in Ascl2 promoter and induced Ascl2 transcription. The decreased Ascl2 transcription by YAP1 interference required an intact KLF5 binding site (GC-box) within Ascl2 promoter, KLF5 knockdown reduced YAP1 binding and Ascl2 luciferase reporter activity upon YAP1 overexpression. Positive correlation among YAP1 and Ascl2 mRNA levels was observed in colorectal cancer (CRC) samples. Thus, our study demonstrated that Ascl2, a fate decider of CRC progenitor cells can be activated by the Hippo signaling pathway in CRC progenitor cells, and ensured their self-renewability.


INTRODUCTION
Colorectal cancer (CRC) is a leading cause of morbidity and mortality in developed countries [1]. Cancer stem cells (CSCs), which exhibit stem-like features, sustain tumor formation, cause metastasis, and are resistant to therapy, have been proposed to explain the functional heterogeneity and carcinogenesis [2][3][4][5]. Protein-coding genes and their products participate in the 'stemness' maintenance and tumorigenicity of CRC progenitor cells [6][7][8]. Thus, it is pivotal to identify the signaling pathways involved in CRC progenitor cells to develop novel reagents that target the refractory CRC progenitor cell population [9].
In 2010, crosstalk study between Wnt and Hippo signaling has been suggested, and evidence of cytoplasmic TAZ inhibiting canonical Wnt signaling has been provided, highlighting that the Hippo pathway can control other signaling cascades [33]. The β-catenin-YAP1-TBX5 (transcription factor) complex is essential to the transformation and survival of β-catenin-driven cancers [34]. These two pathways are known to be important for epithelial development and homeostasis, and there is accumulating evidence that the Hippo cascade engages in crosstalk with Wnt signaling in epithelial tissues [35][36][37]. Essential crosstalk between Hippo and Wnt occurs during epithelial transcriptional control. YAP1 reprograms Lgr5 + cryptic stem cells by inhibiting the Wnt homeostatic program while inducing a regenerative program that includes the activation of EGFR signaling, which drives cancer initiation [38]. However, understanding the molecular mechanisms connecting these signals will require further investigation.
Here, we provided a study that YAP1 combined with KLF5, further bound to the Ascl2 promoter, and then transcriptionally activated Ascl2 expression. This mechanism had an impact on the self-renewability of colon cancer progenitor cells. A correlation between YAP1 and Ascl2 was present not only in the CD133 + CD44 + CRC cell population but also in colon cancer tissue samples. Ascl2 regulation by YAP1 provided evidence that this Wnt target gene can also be transcriptionally activated by YAP1, an effector of Hippo signaling in CRC progenitor cells, and YAP1-activated Ascl2 expression in CRC progenitor cells controlled their cell state stability and conferred their ability to self-renewability.

YAP1-enforced expression in HT-29 and Caco-2 cells increased Ascl2, KLF5 and 'stemness'associated genes expression which were reversed by Ascl2 knockdown
To confirm whether the YAP1-enhanced selfrenewability of colon cancer progenitor cells was related to a KLF5-dependent Ascl2 increase, the relative 'stemness'-associated genes expression (mRNA) levels and protein levels in lv-YAP1/HT-29 and lv-YAP1/Caco-2 cells were determined, and they were found to be significantly increased compared to their respective control cells ( Figure 5). The KLF5 mRNA levels were unaltered, but its protein levels were increased, and it is reported that KLF5 degradation could be prevented by increased YAP1 expression [39][40]. Ascl2 mRNA and protein expression  Figure 5C). The results indicated that YAP1 overexpression in HT-29 and Caco-2 cells increased Ascl2 and 'stemness'-associated genes expression and KLF5 protein level.

YAP1 and KLF5 combined and bound to Ascl2 promoter
There were four loci in the Ascl2 promoter that had a GC-box (GGGCGG), which are potential binding sites for KLF5 [41]. YAP1 is a transcriptional co-activator and (C and D) Relative Ascl2, KLF5 and 'stemness'-associated genes expression levels in both mRNA (C) and protein (D) in YAP1 enforced expressed Caco-2, and further Ascl2 interfered lv-YAP1/ Caco-2 cells. YAP1 nuclear accumulation was based on the blotting using the extracted nuclear proteins in relative cells. β-actin was used as a loading control, Lamin B1 was used as an internal control for the nuclear fraction.
has been reported to bind with KLF5 in breast cells [40]. We performed the co-immunoprecipitation assay using anti-KLF5 or anti-YAP1 antibodies, the immunoprecipitants of anti-KLF5 or anti-YAP1 antibodies in HT-29 and Caco-2 cells can be detected by both anti-KLF5 and anti-YAP1 antibodies (Figures 6A-6D). Four loci within the Ascl2 promoter that had a GC-box (GGGCGG) were selected for chromatin immunoprecipitation (ChIP) assay. Chromatin isolated from YAP1-interfered CD133 + CD44 + HT-29 or Caco-2 cell population and their control cells was subjected to immunoprecipitation using IgG and a rabbit polyclonal IgG against KLF5 and IgG and a rabbit polyclonal IgG against YAP1. Binding at the first two loci that contained a GC-box in YAP1-interfered CD133 + CD44 + HT-29 or Caco-2 cell population was significantly reduced compared with their respective control cells (Figures 6F-6I). In order to confirm whether KLF5 depletion reduces YAP1 binding and Ascl2 luciferase reporter activity upon YAP1 overexpression, we firstly observed the interference efficiency of three different KLF5 siRNAs in Lv-YAP1/ HT-29 or Lv-YAP1/Caco-2 cells and confirmed si-KLF5-2 was the most efficient sequence to inhibit KLF5 expression ( Figure 6J). ChIP assays using lv-YAP1/HT-29, lv-YAP1/ Caco-2 cells and their respective control cells indicated that both KLF5 and YAP1 binding at the first two loci that had a GC-box was significantly increased compared with control cells (Figures 6J-6M). ChIPs 3 and 4 in both YAP1interfered CD133 + CD44 + CRC cell population and YAP1enforced expressed CRC cells gave negative results, which indicated that both KLF5 and YAP1 bound to the first two loci that had a GC-box, but did not bind the other two loci. KLF5 knockdown in Lv-YAP1/HT-29 or Lv-YAP1/Caco-2 cells reversed the increased YAP1 or KLF5 binding to Ascl2 promoter upon YAP1 overexpression ( Figures 6K-6N).

Transcriptional regulation of Ascl2 by YAP1
We produced a series of pGL3-Basic firefly luciferase reporters containing different truncated versions of Ascl2 promoter ( Figure 7A). A reduction in Ascl2 promoter activity was evident from the luciferase reporter assay involving Ascl2 promoter (-2588/+620) in YAP1-interfered CD133 + CD44 + HT-29 and Caco-2 cells compared with their respective controls ( Figures 7B and  7C). Then, we compared the abilities of the truncated promoters to drive Ascl2 expression in YAP1-interfered CD133 + CD44 + HT-29 and Caco-2 cells, and significantly lower luciferase activity was observed in YAP1-interfered CD133 + CD44 + HT-29 and Caco-2 cells when using the pGL3-Ascl2 promoter (-628/+620). However, no difference in luciferase activity using the pGL3-Ascl2 promoter (-82/+620) ( Figures 7B and 7C) was observed. The results indicated that the first three GC-boxes might be responsible for the transcriptional regulation of the Ascl2 gene, and only the region containing the first two GC-boxes can be bound by KLF5 and YAP1 based on ChIP assays. Thus, the first two GC-boxes in the Ascl2 promoter region were used for the transcriptional regulation of Ascl2 by KLF5 and YAP1.

Correlation between Ascl2 and YAP1 expression levels in CRC samples
To verify whether Ascl2 expression levels in human CRC samples were related to YAP1 expression levels, quantitative real-time PCR was used to quantify Ascl2 mRNA as well as YAP1 mRNA levels in 52 CRC samples and their normal mucosa. Ascl2 and YAP1 mRNA levels in cancerous tissues were significantly higher than those in normal mucosa ( Figures 8A and 8B). The Ascl2 mRNA expression was correlated with YAP1 expression ( Figure  8C) in human CRC samples.

DISCUSSION
We demonstrated for the first time that the Hippo coactivator YAP1 was a molecular conveyor of selfrenewal properties in CRC progenitor cells and Ascl2 can be activated by the Hippo signaling pathway in CRC progenitor cells, which ensures the self-renewability of CRC progenitor cells. YAP1 directly activated Ascl2 transcription in concert with its sequence-specific DNA binding partner KLF5.
The Hippo signaling pathway is gaining recognition as an important player in both organ size control and carcinogenesis because the disruption of the core kinase cascade initiated by Hippo and the phosphorylation of YAP/TAZ (Mst1/2, Sav1 and Lats1/2 and YAP1) can lead to carcinogenesis [42][43][44]. YAP1, an effector of the Hippo signaling pathway, has been reported as an oncogene in several tumor types [45][46][47]. TAZ (a transducer of the Hippo pathway) is required for breast CSCs to sustain self-renewal and tumor-initiation capacities [48]. The reciprocal regulation of tumor-initiating stem-like cells by TLR4, and TGF-β requires YAP1 and IGF2BP3 in hepatocellular carcinoma (HCC) [49]. Although YAP binding with β-catenin and TBX5 is thought as being central to the mechanism for YAP1-mediated regulation of Wnt/β-catenin signaling [34,36], our observations in this study identified a novel mechanism that YAP1 could confer CRC progenitor cell properties via combination www.impactjournals.com/oncotarget with KLF5 and bound to the KLF5 binding site (GC-box) in the Ascl2 promoter induced its transcriptional activation ( Figure 8D).
Ascl2 plays a critical role in intestinal stem cells, CRC progenitor cells and human colorectal cancer [10][11][12][13][14] and is also overexpressed in gastric cancer and lung squamous cell carcinoma. It can accelerate cell growth, increase tumor resistance to 5-FU, and be regarded as an independent prognostic indicator [50,51]. We have now provided evidence that YAP1-activated Ascl2 confered the acquisition of YAP1-induced CRC progenitor cell self-renewal properties. YAP1 and Ascl2 may facilitate transformation by the co-regulation of downstream targets or by the activation of distinct targets that synergize to promote the transformed phenotype.
As a coactivator, YAP1 has been reported to regulate a series of genes, such as EGFR, ITGB2, CTGF, SOX9, ALPP, PIK3C2B, AREG, and BIRC5 [48,53,54]. The similarity of these studies is that YAP1-regulated transcription depends on transcription factor partners, such as TEAD, p73, ErbB4, and RUNX2 [55]. In this study, our initial observation is that YAP1 combined with KLF5 in HT-29 and Caco-2 cells, bound to the proximal Ascl2 promoter and induced its transcriptional activation. YAP1enforced expression in HT-29 and Caco-2 cells did not increase KLF5 mRNA levels but significantly increased KLF5 protein levels, this result indicated that YAP1 did not transcriptionally induce KLF5 expression but possibly prevented KLF5 degradation, which is supported by a series of studies [39,40]. KLF5 is an important transcription factor that is involved in regulating a larger number of genes, and KLF5 potentially functions as an oncogene [56]. Multicellular tumor spheroid (MCTS) formation in HT-29 colon carcinoma cells required high Quantitative real-time PCR was performed in CRC tissues (CA) and their peri-cancerous mucosa (N) to assess Ascl2 and YAP1 KLF5 mRNA levels. Ascl2 (A) and YAP1 (B) mRNA levels were higher in CRC samples when compared with those in the peri-cancerous mucosa. Ascl2 expression in CRC tissues was positively correlated with YAP1 expression (P=0.001) (C). A working model detailing YAP1 activated Ascl2 and ensured the self-renewability of colon cancer progenitor cells. YAP1 was either modified by phosphorylation or not (Phosphorylated YAP1 was represented via the red route, whereas YAP1 without phosphorylation was represented via the green route) (D).
KLF5 expression, and tumor formation can be inhibited by KLF5 knockdown [57]. We analysed the Ascl2 promoter (-2588~+620) and found a potential TEAD (TEA domain family members (TEAD1-TEAD4)) binding sequence residing at +222 of Ascl2 promoter ( Figure 6E). This potential TEAD sequence is not involved with the activation of Ascl2 gene upon YAP1 overexpression, because the Ascl2 luciferase reporter activity ( Figure 7A-7C) was unaltered although YAP1 overexpression, The accordant increase of YAP1 and Ascl2 in human colon cancer samples was consistent with the observations in vitro. This indirect in vivo evidence, together with the in vitro results, strengthens our hypothesis for a role of YAP1 in the regulation of Ascl2.
Taken together, we suggest that YAP1/KLF5activated Ascl2 expression in colon cancer progenitor cells confers their self-renewability. This novel mechanism provides a possible target for colon cancer progenitor cells.

Cell culture
The HT-29 and Caco-2 human colonic adenocarcinoma cell lines were obtained from the American Type Culture Collection (ATCC) and routinely cultured in McCoy's 5A medium (Sigma, USA) supplemented with 10% fetal bovine serum (FBS; HyClone, USA) at 37°C in a 5% CO 2 humidified incubator.

Flow cytometry cell sorting and flow cytometry analysis
The single-cell suspensions of HT-29 and Caco-2 cells were incubated with APC-conjugated anti-human CD133 antibody (Miltenyi Biotec) and FITC-conjugated anti-human CD44 antibody (Miltenyi Biotec) in PBS for 10 min at 4°C. The APC-conjugated mouse IgG2b isotype control antibody (Miltenyi Biotec) and the FITCconjugated mouse IgG1 isotype control antibody (Miltenyi Biotec) served as negative controls. The CD133 + /CD44 + and CD133 -/CD44cell population was sorted using a fluorescence-activated cell sorter (BD Aria, USA).

Cell proliferation
Isolated cells were seeded at 10 000 cells/well in 24-well plates. Then, 20 μl of CCK-8 solution (Beyotime Biotechnology, China) was added to each well and cultured at 37°C for 4 h. On the 1st, 2nd, 3rd and 4th days, the absorbance was measured using a microplate reader (Thermo, USA) at 450 nm. The experiments were repeated in triplicate.

Colony formation assay
An equal number of viable cells were seeded at a density of 1 000 cells/well in 6-well plates. After incubating for 14 days at 37°C, the colonies that adhered to the bottoms of the plates were counted using an inverted microscope (Olympus, Japan).

Tumorsphere formation
To evaluate the tumorsphere-formation potential, when single cells reached a density of 1 000 cells, they were plated in 48-well ultra-low attachment plates (Corning Inc, USA) with DMEM/F12 (HyClone, USA) serum-free media supplemented with B-27 (Gibco, USA), epidermal growth factor (EGF, Peprotech, USA) and basic fibroblast growth factor (bFGF, Peprotech, USA). Every 3 days, fresh medium was added. The experiment was terminated after 7 days, and spheres were quantified using an inverted microscope (Olympus, Japan).

Real-time PCR
To quantitatively determine the expression levels of each gene, real-time PCR was performed using firststrand cDNA, forward and reverse primers and the SYBR premix Ex TaqTM Green II (TaKaRa, Japan). The primer sequences are summarized in Table 1, and β-actin was used as an endogenous control. Expression levels were analyzed using the comparative Ct method (2 -∆∆Ct ). The real-time PCR reactions were independently performed in triplicate.

Western blotting
The cytoplasmic proteins and nuclear proteins were extracted using NE-PER nuclear and cytoplasmic extraction reagents (Thermo, USA). Separated protein bands were transferred onto PVDF membrane, and the membrane were probed overnight at 4°C with a primary antibody ( Table 2). Protein bands were visualized using the chemiluminescent Pierce ECL kit (Thermo Scientific), and quantification of the Western blots was performed using ChemiDox TM XRS with Image Lab TM software. Detailed Western blotting procedures have been described previously [58].

YAP1 overexpression
Lentivirus particles expressing YAP1 were produced by GenePharma Co. Ltd. China. HT-29, and Caco-2 cells were transfected with the lentivirus particles using an LV5 vector with a YAP1 insert. Stably transfected cells with GFP were sorted using flow cytometry (BD FACS Aria II; BD Biosciences, Franklin Lakes, NJ, USA) and isolated using puromycin selection (Solarbio, Beijing, China).

Co-immunoprecipitation
Cell lysates were immunoprecipitated with anti-KLF5 (Santa Cruz, SC22797) or anti-YAP1 (Cell Signaling, 14074) polyclonal antibodies according to the instructions of the BeaverBeads TM Protein A/G Immunoprecipitation Kit (22202-20), separated by SDS/ PAGE and analyzed by Western blotting.

ChIP assays
ChIP assays were performed using a ChIP assay kit (Millipore, 17408) according to the manufacturer's instructions. Soluble chromatin was prepared from siRNA-YAP1-1/HT-29 CD133 + /CD44 + or siRNA-YAP1-1/Caco-2 CD133 + /CD44 + cell population and their control cells as well as lv-YAP1/HT-29 or lv-YAP1/ Caco-2 and their control cells. The chromatin was immunoprecipitated with anti-KLF5 (Santa Cruz, SC22797, USA) or anti-YAP1 (Cell Signaling, 14074, USA). The precipitated DNA was subjected to real-time PCR using specific primers. The primer sequences and the lengths of the amplified PCR products are shown in Table 3.

Transfection and luciferase assays
The Ascl2 5′-flanking sequence (-2588/+620 bp region), amplified using PCR from intestinal tissue DNA, was inserted into the luciferase reporter vector www.impactjournals.com/oncotarget pGL3-Basic (Promega, Madison, WI, USA). The primer pairs used to produce each promoter fragment are listed in Table 3. All transfections were performed using the FuGENE HD transfection reagent (Promega, USA). Luciferase activity was measured 48 h post-transfection using the Dual-Luciferase Reporter Assay System (Promega) as described in our previous report [59]. The pGL3-Basic vector containing the Ascl2 promoter (-628/+620) and one GC-Box site (GGGCGG) served as the wild-type construct used for the generation of the Ascl2-Luc mutant, which harbors a mutation in the GC-BOX site (AAGTAG) created by PCR-based sitedirected mutagenesis. The transfection of the luciferase reporters containing the various Ascl2 promoters and the luciferase assays were performed as described in our previous report [17,59].

Tissue samples
A total of 52 patients with colorectal cancer who were scheduled for colonoscopy or for surgical resection were enrolled in this study. Cancerous samples and their peri-cancerous mucosa were collected by biopsy or from the resection samples. The fresh samples were immediately stored in liquid nitrogen for further quantitation of Ascl2, YAP1 and KLF5 mRNA levels using quantitative real-time PCR analysis. This study was approved by the local clinical research ethics committee. All the subjects provided informed consent prior to their colonoscopy or resection surgery.

Statistical analysis
GraphPad Prism 6.0 software was used for all statistical analyses. The differences were deemed significant when P<0.05 and very significant when P<0.01.

Author contributions
The study was conceived and designed by XW and RW. The experiments were performed by XW, JY, YS and HC. The data were analyzed by XW, SL, and LL. The manuscript was written by XW and RW.