PI3K/AKT-mediated upregulation of WDR5 promotes colorectal cancer metastasis by directly targeting ZNF407

Colorectal cancer (CRC) is the third most common cause of cancer deaths, and has a high rate of liver and lung metastasis. Unfortunately, distant metastasis is the main barrier for advanced CRC therapy and leads to a very low survival rate. In this study, we identified WDR5, a vital factor that regulates vertebrate development and cell self-renewal and reprogramming, as a novel prognostic marker and therapeutic target for CRC patients. We demonstrate that WDR5 is upregulated in CRC tissues and promotes CRC metastasis both in vitro and in vivo. In an effort to investigate the impact of WDR5 on CRC cell fate, we treated CRC cells with growth factor and inhibitor. We report that WDR5 is a novel factor in the metastasis of CRC by triggering epithelial–mesenchymal transition (EMT) process in response to the PI3K/AKT signaling pathway. Moreover, WDR5 shows a direct binding to the ZNF407 promoter on regulating cellular EMT process, leading to CRC metastasis. Hence, our findings strongly position WDR5 as a valuable marker for CRC, and inhibiting WDR5 or the associated signaling pathways may be an effective strategy for the future development of anti-CRC therapy.

Colorectal cancer (CRC) occupies the third place in the cancer death, and is common for both men and women worldwide. 1,2 The incidence rate has not significantly changed in the past decade and has even increased among adults younger than 50 during this period. 2 Although new screening strategies and improvements in treatment continue to emerge, the prognosis differs widely. [3][4][5] Moreover, distant metastasis remains the main barrier for CRC therapy. 6 Thus, it is imperative to discover molecular mechanisms and genetic alterations and to explore new biomarker and therapeutic targets for CRC. 7 Studies confirmed that the MLL and hCOMPASS complexes regulate global H3K4 methylation, 8 and one of the components of these complexes, MLL1, 9 has been reported to have oncogenic roles by modulating chromatin structures. WDR5 is the most highly conserved subunit across all of the human COMPASS-related complexes, 10 which generate H3K4me3 to influence cell development, 8,11,12 especially vertebrate skeletal system development. 13,14 Recent studies reported that WDR5 is abnormally expressed in many cancers. A cohort study of leukemia patients suggested that WDR5 may have oncogenic effects in leukemogenesis. 15 In addition, researches show that WDR5 has high expression level in prostate cancer, bladder cancer, and breast cancer, and has a positive correlationship with advanced stage and poor survival. [16][17][18] It is well established that CRC may share some similarities with these other types of cancer. In addition, WDR5 mediates somatic cell reprogramming and embryonic cell self-renewal, while epithelial-mesenchymal transition (EMT) was reported to participate in the achievement of stemness as a key step. 19,20 Therefore, these achievements suggest that WDR5 might have an oncogenic role to play in the early steps of the invasion-metastasis cascade by regulating EMT to enable cancer cells to invade deep into the stroma, and then translocate to distant parenchyma.
The PI3K/AKT pathway is frequently observed to be hyperactivated in CRC caused by genetic aberrations, such as growth factors overexpression, gene mutations and so on. Recent studies have presented that PI3K/AKT pathway activation is a key character of EMT. 21 One study in prostate cancer demonstrated a relationship between radioresistance and EMT via overactivating the PI3K/AKT/mTOR signaling pathway. 22 Another study showed that activated AKT causes EMT based on the regulation of EMT-specific markers in human squamous cell carcinoma lines. 23 Moreover, Riquelme E et al. 24 advocated a methyltranferase, EZH2, was modulated by the PI3K/AKT transduction pathway in lung cancer. As WDR5 has great influence on assembly and enzyme activity of the MLL histone methyltransferase complex, 25,26 and the mechanism by which WDR5 levels are elevated in CRC is rarely known, we investigated the regulation of WDR5 in CRC.
In the present study, we discovered that high level of WDR5 expression in primary CRC is strongly correlated with poor prognosis. Both in vitro and in vivo assays indicated that WDR5 upregulation is sufficient to promote CRC metastasis.
More importantly, molecular studies revealed that WDR5transfected cells underwent EMT and were modulated by the PI3K/AKT signaling pathway. Furthermore, gene expression microarrays were used to identify ZNF407, which was reported to affect tumor progression, 27,28 as a novel target of WDR5, which markedly increased CRC cell migration. Thus, we propose that WDR5 is a promising target in CRC prognostics and therapeutics.

Results
WDR5 is frequently detected to be upregulated in CRC cell lines and tissues. For the purpose of addressing the potential role of WDR5 in CRC development, WDR5 expression levels were measured in human CRC cell lines and tissues. Results showed that all five CRC cell lines (HT-29, SW620, HCT-15, HCT116, and COLO205) had higher WDR5 expression level than the normal intestinal epithelial cell line (FHC) at both the mRNA level (Figure 1a; Po0.05) and the protein level ( Figure 1b). In addition, WDR5 mRNA expression was markedly higher in colorectal tumor tissues than those in paired adjacent normal mucosal tissues (qRT-PCR; n = 16; Figure 1c; Po0.05). Moreover, similar results can be counted in large samples from TCGA Colorectal Adenocarcinoma (COADREAD) that project that about twofold greater WDR5 mRNA expression levels were calculated in human CRC tissues than in normal tissues (Supplementary Figure S1B; Po0.001). Further, to examine endogenous WDR5 protein expression, we performed immunohistochemistry (IHC) on histopathologically confirmed CRC tissue sections with paraffin-embedded (n = 161; for more information, see Supplementary Table S1). Results showed a markedly higher level of WDR5 protein expression in tumor tissues than that compared to paired non-tumor mucosal tissues (Figure 1d  Elevated expression of WDR5 is associated with CRC metastasis and predicts poor prognosis. Then, the association between WDR5 expression and clinicopathological characters was analyzed. Data presented in Table 1 implied that high WDR5 expression was significantly correlated with TNM stage, lymph node status, and distant metastasis, but not with gender, age, location, size, grade, or CEA level. As shown in Figure 1f and g, patients with a high WDR5-expression signature had a shorter OS, likewise, high expression level of WDR5 was consistently associated with a reduced PFS by Kaplan-Meier analysis significantly (Po0.05). In addition, according to the data from TCGA data sets, the correlation between WDR5 expression level and the patients' survival was analyzed by defining two different follow-up times (5 years or extend to 28 January 2016). Both results showed a shorter overall survival (OS) in high WDR5 group significantly with Po0.05 (Supplementary Figure S1C). Also, the 5-year rate for high WDR5 level patients is 45.6%, compared to over 70% of low WDR5-level group at the fifth year (Supplementary Figure S1D).
Moreover, Cox proportional hazards regression analysis embodied that the WDR5 level was independently related to survival, which implicated WDR5 as an independent prognostic factor for PFS and OS (shown in Table 2). Thus, these analyses indicate that WDR5 upregulation is clinically relevant to CRC metastasis, and WDR5 is an independent prognostic factor in CRC patients.
WDR5 enhances the CRC cell migration and invasion ability in vitro. By reason of the relationship between WDR5 and lymph node status and distant metastasis, the biological function of WDR5 in CRC cell migration and invasion was explored using transwell assays (with or without Matrigel) in the context of WDR5 overexpression or deletion. WDR5 expression was evaluated, and both migration and invasion were substantially enhanced when WDR5 was overexpressed in HCT-15 cells (Figure 2a; Po0.001). In contrast, downregulation of WDR5 sharply suppressed SW620 cell migration and invasion (Figure 2b; Po0.001).
Furthermore, to simulate the liquid nature of the circulatory system that is encountered during metastasis, we examined the sphere formation ability of CRC cells after altering WDR5 expression. Interestingly, HCT-15-WDR5 cells had increased capacity for sphere formation compared with HCT-15-Ctrl cells (Figure 2c; Po0.05). In contrast, the sphere-forming ability of SW620-sh#1 and SW620-sh#2 cells was significantly decreased (Figure 2d; Po0.05), suggesting that WDR5 maintains the vitality and stemness of CRC cells. Therefore, these results prove that WDR5 promotes the metastatic behavior of CRC in vitro. In addition, we utilized the in vivo liver and lung metastasis models with the SW620-sh#2 cell line, in which WDR5 was silenced, and the SW620-Scr cell line was a negative control. Livers and lungs were collected after 8 weeks, and the metastatic nodules on the surfaces were significantly smaller and fewer in number in the SW620-sh#2 group than in the SW620-Scr group (Figure 3c,d,g and h; Po0.05). Thus, these two models indicate that WDR5 promotes CRC metastasis in vivo.

WDR5 promotes CRC cell motility by inducing EMT.
Given that EMT is a critical event involved in tumor invasionmetastasis cascade, the impact of WDR5 on EMT was  WDR5 and E-cadherin expression was further studied by IHC in the same cohort of 20 CRC samples. The results confirmed that E-cadherin expression had a significant negatively correlation with WDR5 expression (r 2 = 0.209, Po0.05; Figure 4c and d). Overall, WDR5 regulates CRC metastasis by modulating epithelial-and mesenchymal-specific proteins.
WDR5 mediates the EMT process regulated by the PI3K/AKT pathway. As PIK3CA and PTEN mutations are commonly detected in CRC, and hyperactivation of the PI3K/ AKT pathway, which leads by these genetic aberrations, is frequently observed in CRC, which leads to reduced apoptosis, increased proliferation, and induced EMT process, we stimulated the PI3K/AKT pathway by treating cells with IGF-1 for 24 h and then evaluated the expression of certain proteins that are involved in the PI3K/AKT pathway and EMT. Western blotting revealed increased expression of phosphorylated AKT, Snail1, ZEB1, and WDR5 and decreased E-cadherin and ZO-1 expression in SW620-Scr and SW620sh cells with IGF-1 stimulation (Figure 5a). As expected, the mRNA levels of WDR5, Snail1, and ZEB1 were markedly increased in both SW620-Scr and SW620-sh cells after IGF-1 treatment, whereas E-cadherin and ZO-1 mRNA levels were significantly downregulated under the same conditions (Figure 5d; Po0.05). Moreover, knockdown of WDR5 reversed the impact of IGF-1 on these molecules (Figure 5a and d; Po0.05). These data demonstrate that PI3K/AKT pathway activation increases WDR5 expression and then induces the dysregulation of EMT markers.
Interestingly, when WDR5 was silenced with shRNA, the effect of LY294002 on the EMT markers was partially weakened, further strengthening the evidence that the PI3K/ AKT pathway influences EMT via WDR5 (Figure 5c and e). Moreover, the transwell assays clearly showed that LY294002 inhibited the metastatic ability of CRC cells through a WDR5-dependent PI3K/AKT axis (Figure 5f and g).
These results suggest that WDR5 is elevated by PI3K/AKT pathway activation, which eventually induces EMT.
ZNF407 is an essential downstream target of WDR5 promotion effects in CRC metastasis. Finally, to identify the downstream target genes of WDR5, gene expression profiling of HCT-15-WDR5 and HCT-15-Ctrl cells was conducted by microarray analysis. Moreover, TCGA data were downloaded from TCGA data portal to sort out genes that were positively correlated with WDR5 with Po0.01. Gene ontology analysis was performed to identify the unique genes that were functionally related to cell metastasis. Moreover, the overlapped genes among these three analyses were selected to be validated in four different CRC cell lines (Figure 6a, left). With these comprehensive analysies of microarray data, gene ontology, TCGA data, and qPCR validation, we identified and verified 29 genes for further exploration (Figure 6b and c). To be more exacting and rigorous, genes with a fold change in mRNA expression of at least 2.0 were selected (Figure 6a, middle) and intersected   To verify the prediction of binding between WDR5 and the ZNF407 promoter, Flag-WDR5 was specially detected at the ZNF407 promoter in 293T cells by ChIP analyses (Figure 6e). Moreover, WDR5 knockdown decreased the occupancy at ZNF407 promoter using anti-WDR5 or anti-H3K4me3    (Figure 6f). Moreover, luciferase assays demonstrated that WDR5 could upregulate ZNF407 promoter activity, which further confirmed that WDR5 promotes the transcription of ZNF407 by binding to its promoter region (Figure 6g). Furthermore, rescue assays evaluated by qRT-PCR showed that siRNA targeting ZNF407 reduced the expression levels of Snail1 and ZEB1, and significantly induced E-cadherin and ZO-1 mRNA expression; exogenous WDR5 expression could partially reverse these effects (Figure 6h). Ultimately, our results indicate that WDR5 potentially promotes CRC metastasis by targeting and stimulating the transcription of ZNF407.
Taken together, our study attempted to figure out that WDR5 is frequently overexpressed in CRC cell lines and tissues, and promotes CRC metastasis. Inhibiting WDR5 or the associated signaling pathways may improve the efficacy of CRC treatment as an effective strategy.

Discussion
CRC development involves multiple genetic and epigenetic alterations, which results in the dysregulation of oncogenes and tumor suppressor genes. 30 Functional loss of PTEN, activating mutation in Ras or PIK3CA, and stimulation by various growth factors in CRC represent mechanisms for activating the PI3K/AKT pathway. Herein, we described the tumorigenic function of WDR5 in CRC metastasis as a downstream target of the PI3K/AKT pathway, which leads to the altered expression of EMT markers and regulates metastasis by directly promoting the transcription of ZNF407. Several cell lines and animal models support our conclusion and proposed model (Supplementary Figure S4).
As reported, WDR5 is a major driver of cell progression in various cancer types. For instance, WDR5 has central roles in the proliferation of androgen-dependent prostate cancer cell, and its protein expression levels are positively correlated with poor survival in breast and bladder cancer. 15,17,18 Nevertheless, none of these studies has elucidated the functional mechanism of action of WDR5 in CRC metastasis. In our study, WDR5 upregulation was frequently observed in CRC cell lines and tissues, and its overexpression level could serve as an independent predictor for survival of CRC patients. In functional studies, WDR5 significantly promoted CRC migration, invasion, and sphere formation in vitro, and also strengthened the ability to metastasize to the liver and lung in vivo. Thus, these results are in accordance with the finding that WDR5 involves in CRC metastasis, and can contribute independently as a favorable prognostic factor for OS, indicating that WDR5 can act as a potential biomarker for CRC patients.
As EMT stays crucial in tumor migration and invasion during progression and the acquisition of stemness properties, [31][32][33] we tested whether WDR5 affected cell motility by inducing   EMT. As expected, WDR5 overexpression decreased epithelial markers and increased mesenchymal markers at both mRNA and protein levels. Moreover, WDR5 and E-cadherin expression levels were related (r 2 = 0.209, Po0.05; Pearson's test). Additional studies revealed that WDR5 was upregulated by the PI3K/AKT pathway, leading to EMT in CRC cells. Interestingly, in the absence of WDR5, EMT regulation by PI3K/AKT was seriously weakened. It has been reported that PI3K/AKT can trigger cellular EMT. 34 Our data present an explanation for how WDR5 is modulated in CRC and confirm that the PI3K/AKT pathway-induced EMT entails WDR5 profoundly. More importantly, these findings provide fresh possibilities for molecular intervention in CRC as a target or as a cooperative factor to enhance the efficacy for moleculetargeting agent.
Furthermore, microarray assays were performed, and gene expression data were analyzed using various databases (i.e., gene ontology, GEO, and TCGA). Eventually, ZNF407 was determined to be directly targeted by WDR5, and CRC cell metastasis was inhibited when ZNF407 was silenced. Additional data revealed that knockdown of ZNF407 partially rescued the WDR5-induced decrease in E-cadherin mRNA expression (Figure 6h). Moreover, it was reported that missense mutations in ZNF407 affect tumor progression, as evidenced by whole-genome sequencing of gastrointestinal stromal tumors. 28 Moreover, our functional studies of ZNF407 showed that ZNF407 functioned as a promotive factor in CRC metastasis and accelerated cell proliferation slightly (Supplementary Figure S3). Together, these data suggest a novel theory that CRC metastasis is regulated by WDR5.
Although our studies led to these important discoveries, there are some limitations. First, we detected a slight, but not significant promotion of CRC cell growth in MTT and colony formation assays (data not shown). We tentatively suggest that WDR5 functions in stemness or the maintenance of cell viability, not growth, which is in conformance with the results in Figure 2c and d, showing more but not bigger spheres in the WDR5 overexpressed group than that in the control group. Notably, it is reported that tumor cells stop dividing when they undergo EMT; hence, EMT acts as a suppressor of cancer cell division, thus stagnating proliferation. 35,36 Therefore, WDR5 might be more appropriately categorized as a metastasis promoter than a tumor growth inducer. Second, CRC predominantly metastasizes to the liver and lung; 37,38 we designed two in vivo models to show that WDR5 has a positive effect on CRC metastasis. Importantly, in the multistep metastatic process, cancer cells must first survive in the circulation system and then implant at a distant organ site. On the basis of this phenomenon, in vitro sphere formation assays were conducted, and a higher sphere formation efficiency was observed in cells overexpressing WDR5 that were cultured ultralow attachment plates. In addition, Cox correlation analysis of CRC tissues indicated that WDR5 is an independent prognostic factor for PFS. Hence, our results imply that WDR5 takes pivotal part in CRC metastasis process. Third, WDR5 was reported to interact with HDAC3, resulting in a recruitment of a histone methyltransferase complex and the activation of mesenchymal gene expression under hypoxic conditions. [39][40][41] Indeed, qRT-PCR analysis of CoCl 2 -treated CRC cells showed a dose-dependent activation of HIF1-α, followed by the upregulation of WDR5 expression (data not shown). These findings suggest another potential mechanism by which WDR5 is elevated in CRC and provide reliable data to explain how WDR5 regulates EMT. Fourth, the role of EMT in metastasis is not yet completely understood, and some studies suggest that EMT is not the speed-limiting process for metastasis, instead it highlights the importance of EMTinduced chemoresistance. 35,42 Furthermore, accumulating evidence indicates that cancer stem cells are characteristically resistant to conventional cancer therapy and drive tumor progression, metastasis, and chemoresistance. 43 A previous study reported that PI3K/AKT induced chemoresistance in colon cancer cells, 44 coinciding with the regulation of cellular self-renewal and the reprogramming with a core transcription network by WDR5, which was found to be activated by PI3K/ AKT to promote CRC metastasis in our study. The role and mechanism of action of WDR5 in CRC chemoresistance represents a future research direction.
In conclusion, WDR5 is vital for cell development, and its dysregulation is essential in CRC particularly. We demonstrate that WDR5 is upregulated in CRC and promotes CRC metastasis by clarifying the underlying mechanism. This study show that WDR5 is a valuable marker for CRC and has a pivotal role in CRC metastasis.

Materials and Methods
Patient tissue specimens and follow-up. All formalin-fixed, paraffinembedded colorectal tissues used in this study were archived from Sun Yat-sen University Cancer Center (SYSUCC) with written informed consent following the Ethics Committee approval, and associated clinical data and pathology report review were obtained from raw case reports.
Each tissue sample was collected from patients before surgery and neoadjuvant therapy from 1 September 1999 to 31 December 2007, containing tumor tissues (percentage of tumor cells 470%) and non-tumor tissue (45 cm lateral from the cancerous edge). All patients were contacted for at least 5 years, and the last censor date was 1 March 2014.
The normalized level_1 clinical data and mRNAseq data are from TCGA database (data version 2016_01_28) for the Colorectal Adenocarcinoma (COADREAD) project.
Cell lines, cell culture, and stimulation. The HEK-293T (human embryonic kidney cell line), FHC (normal colon epithelium cell line), and the human CRC cell lines, including HCT116, HT-29, HCT-15, SW620, and COLO205, were obtained from American Type Culture Collection, and cultured under standard conditions. GP293 cells were from our own laboratory and grown in DMEM. 45 All cells were regularly authenticated by STR profiling at the China Center for Type Culture Collection. Sensitive Mycoplasma detection was conducted regularly by a PCR-based method.
For blockade or activation of the PI3K/AKT pathway, LY294002 (Cell Signaling Technology, Boston, MA, USA) or IGF-1 (Cell Signaling Technology) was used. Cells were plated in six-well plates, and 24 h after seeding, cells were treated for another 24 h with LY294002 (10 μM), IGF-1 (100 ng/ml), or DMSO/PBS (control) in a serumfree medium. Cells were then washed in PBS, RNA was isolated for qRT-PCR, and cell lysis was used in western blot analysis as described below.
Quantitative reverse transcription PCR. Total RNA was extracted from cultured cells using TRIzol (Invitrogen, Carlsbad, CA, USA), and was reversetranscribed using M-MLV reverse transcriptase (Promega, Madison, WI, USA). cDNA was used as template for quantitative reverse transcription PCR (qRT-PCR) using Stratagene MX3000P Sequence Detection System with SYBR Green qPCR mixture (Invitrogen). Each qRT-PCR experiment was performed in triplicate. GAPDH was used as loading control. The primer sequences of the indicated genes are provided in Supplementary Table S3. WDR5 promotes colorectal cancer metastasis X Tan et al Immunohistochemical staining. IHC staining was performed according to a previous study. 45 Briefly, IHC for WDR5 was performed in 161 paraffin-embedded primary CRC tissues and adjacent noncancerous tissues. The following primary antibodies were used: rabbit anti-E-cadherin monoclonal antibody (1:200, Cell Signaling Technology) and goat anti-WDR5 polyclonal antibody (1:100, R&D System Inc, Minneapolis, USA). Antibody dilution solution was used as negative control. IHC staining was evaluated using the H-score and dichotomized according to OS with an ROC curve.
Stable cell line construction. WDR5 full-length sequence and control sequence were inserted into the pLNCX2 viral vector (Clontech, Mountain View, CA, USA). WDR5 shRNA was synthesized at Invitrogen and the short hairpin target sequences are provided in Supplementary Table S2. Stable cell lines HCT-15 for overexpressing WDR5 and SW620 for silencing WDR5 were constructed by lentivirus packaging and transduction according to a previous method with slice modifications. 45 WDR5 expression in these stable cell lines was validated by western blotting and qRT-PCR evaluation.
Migration and invasion assays. Tumor cells (1 × 10 5 each) were serumstarved and seeded into Boyden chambers with or without Matrigel-coated inserts (8-μm pore, BD Falcon, Franlin Lakes, NJ, USA) for invasion or migration assays, respectively. The chambers were then placed in 24-well plates containing medium and 10% FBS. After 24 h, cells attached to the undersurface of the membrane were fixed in ethanol, stained with crystal violet, and counted using a microscope.