The MAZ transcription factor is a downstream target of the oncoprotein Cyr61/CCN1 and promotes pancreatic cancer cell invasion via CRAF–ERK signaling

Myc-associated zinc-finger protein (MAZ) is a transcription factor with dual roles in transcription initiation and termination. Deregulation of MAZ expression is associated with the progression of pancreatic ductal adenocarcinoma (PDAC). However, the mechanism of action of MAZ in PDAC progression is largely unknown. Here, we present evidence that MAZ mRNA expression and protein levels are increased in human PDAC cell lines, tissue samples, a subcutaneous tumor xenograft in a nude mouse model, and spontaneous cancer in the genetically engineered PDAC mouse model. We also found that MAZ is predominantly expressed in pancreatic cancer stem cells. Functional analysis indicated that MAZ depletion in PDAC cells inhibits invasive phenotypes such as the epithelial-to-mesenchymal transition, migration, invasion, and the sphere-forming ability of PDAC cells. Mechanistically, we detected no direct effects of MAZ on the expression of K-Ras mutants, but MAZ increased the activity of CRAF–ERK signaling, a downstream signaling target of K-Ras. The MAZ-induced activation of CRAF–ERK signaling was mediated via p21-activated protein kinase (PAK) and protein kinase B (AKT/PKB) signaling cascades and promoted PDAC cell invasiveness. Moreover, we found that the matricellular oncoprotein cysteine-rich angiogenic inducer 61 (Cyr61/CCN1) regulates MAZ expression via Notch-1–sonic hedgehog signaling in PDAC cells. We propose that Cyr61/CCN1-induced expression of MAZ promotes invasive phenotypes of PDAC cells not through direct K-Ras activation but instead through the activation of CRAF–ERK signaling. Collectively, these results highlight key molecular players in PDAC invasiveness and may help inform therapeutic strategies to improve clinical management and outcomes of PDAC.

onstrate that MAZ plays a critical role in the progression of various cancers including breast, prostate, liver, and pancreas (2,6,(12)(13)(14). In pancreatic cancer, MAZ expression is positively correlated with poor prognosis of the patients (2), and MAZ depletion delays in vitro cell proliferation, migration, and invasion of PDAC cells (2). Although the physiological changes produced by MAZ in PDAC have been characterized by a group of researchers, the molecular mechanisms through which MAZ regulates these changes and the pathway of MAZ activation in PDAC cells await detailed analysis.
PDAC development is associated with complex genetic and epigenetic changes. Several signature mutations are involved in PDAC progression. These include mutations in K-Ras and p53 tumor suppressor genes (15)(16)(17)(18)(19)(20)(21), deletion of p53, p61, SMAD4/ DPC4, and deregulation of microRNA (23,24) and chromosomal aberrations (25)(26)(27). Mutations in the K-Ras gene are prevalent in PDAC and play critical, although poorly understood, roles in initiating and empowering the progression of PDAC in human and genetically engineered mouse models in the presence of mutant p53 or the absence of other tumor suppression genes (15,16,28,29). Multiple studies describe the presence of an oncogenic K-Ras signal as insufficient for cellular transformation; additional genetic, epigenetic, dosage, or environmental factors may be required to raise the activity threshold of the K-Ras signal for tumorigenesis (20,30,31), yet these factors have not been fully elucidated. A study has shown MAZ up-regulates K-Ras transcription via binding into the G4-DNA of the K-Ras promoter in pancreatic cancer cells (7,32). However, a mechanistic link between the oncogenic role of MAZ and K-Ras activation has not yet been elucidated. Thus, our goal was to investigate whether MAZ regulation of K-Ras provides a basis for MAZ's diverse tumor biological roles such as proliferation, migration, or invasion and the stemness of PDAC cells.
CCN1/Cyr61 is a matricellular protein and a founding member of the CCN (Cyr61-CTGF-NOV) family proteins (33)(34)(35). CCN1 is a tumor-promoting factor in PDAC (28, 36 -38). Cyr61/CCN1 utilizes Notch-1-sonic hedgehog (SHh) signaling, an epithelial-mesenchymal transition (EMT) inducer (39), to transmit its tumor-promoting signals in pancreatic cancer cells (28,37,38,40,41). Previously, we have demonstrated that Cyr61/CCN1 is differentially expressed in various pancreatic cancer cells and human PDAC samples depending on their aggressive behaviors (38). We also found that interruption of CCN1 by genetical manipulation or antibody treatment reversed EMT and delayed the migration of pancreatic cancer cells in vitro and aggressive tumor growth in a xenograft model (28,38). Thus, Cyr61/CCN1 is assumed to be a novel target for inhibiting pancreatic cancer growth and differentiation. These pathobiological concepts were further supported by two independent studies indicating that the chemoresistance and metastatic potential of PDAC can be enhanced by Cyr61/CCN1 overproduction (42,43). These experimental data suggest a central role of Cyr61/CCN1 in PDAC progression; yet the mechanism by which Cyr61/CCN1 induces PDAC development is incompletely understood. Based on the functional similarities, we postulated here a possibly useful link between CCN1 and MAZ through the activation of mutant K-Ras in PDAC. Therefore, we sought to determine whether Cyr61/ CCN1 and MAZ signaling cross-talk in PDAC cells regulate oncogenic signaling.
In this report, we demonstrate that MAZ-induced invasive phenotypes (EMT, stemness, and migration) of PDAC cells are mediated by activating the downstream targets of K-Ras, CRAF-ERK (extracellular signal-regulated kinase) signaling, in PDAC cells. In contrast with previous work, we found no direct effect of MAZ on K-Ras expression. Moreover, we found that CCN1 is an upstream regulator of MAZ. The depletion of CCN1 and its downstream SHh signaling pathway dramatically blocks MAZ expression in PDAC cells, affecting the cell migration and invasion induced by MAZ. Cyr61/CCN1 knockdown significantly blocks MAZ-CRAF-ERK signaling activities in PDAC cells and thus suggesting that the oncogenic behavior of CCN1 could be mediated by MAZ-CRAF-ERK signaling pathway.

MAZ is differentially expressed in normal and various other pancreatic cancer cell lines
To determine the status of MAZ mRNA and protein in an immortalized human pancreatic ductal epithelial (HPDE) cell line and various pancreatic cancer cell lines from less aggressive (i.e. BxPC-3) to highly aggressive cell lines (i.e. AsPC-1 and Panc-1), the mRNA and protein levels of MAZ were evaluated using Northern blotting and Western blotting, respectively (Fig. 1, A and B). MAZ mRNA and protein expression were markedly higher in the AsPC-1 and Panc-1 cell lines as compared with normal HPDE and the less aggressive cell line, BxPC-3. The MAZ protein level in HPDE cells was minimally detectable.
Cytoplasmic and nuclear subcellular fraction analysis in Panc-1 cells indicated that MAZ protein was expressed predominantly in the nucleus (Fig. 1C), which was expected, as MAZ is a transcription factor (4). The immunofluorescent analysis further validates the nuclear localization of the MAZ protein in Panc-1 and AsPC-1 cells (Fig. 1D).

Differential expression of MAZ in PDAC samples
To evaluate the status of MAZ mRNA and protein levels in the tissue samples of human PDAC and adjacent normal cells, we performed in situ hybridization and immunohistochemistry in tissue arrays containing different grades of pancreatic adenocarcinoma samples (n ϭ 28) along with adjacent normal tissue (n ϭ 6). MAZ mRNA and protein expressions were detected in ϳ80% (based on mRNA expression) and ϳ93% (based on protein detection) of the adenocarcinoma samples (Table S1). In contrast, MAZ mRNA or protein was absent or minimally detected in the adjacent normal pancreas (Table S1). In PDAC samples, MAZ expression increased gradually from low grade to high grade ( Fig. 1E and Fig. S1). MAZ was distributed mainly in the nucleus, but its expression was also detected in the cytoplasm (Fig. S2). Consistent with tissue array samples, MAZ expression was located predominantly in the nucleus of primary cultured human PDAC cells obtained with endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA, Fig. 1F).

MAZ promotes PDAC aggression
Next, the MAZ expression was evaluated in PDAC samples of genetically engineered mouse models. These include Kras LSL.G12D/ϩ , the Pdx-Cre model (referred to as the KC model (K-Ras mutant and cre)), and KC with a mutant p53 gene (designated KPC (K-Ras mutant, p53mutant, and cre)). Consistent with human PDAC samples, MAZ levels were significantly increased in the KC and KPC samples ( Fig. 1G and Table S2), and MAZ was first detected in pancreatic intraepithelial neoplasms. Thus, consistent with previous work (2), these studies suggest a strong association of MAZ overexpression and PDAC progression.
The side population (SP) of pancreatic cancer cells has mesenchymal/stemness features that are lacking in the main or non-side population (NSP). SP cells are more aggressive in the formation of subcutaneous tumors as compared with the NSP (38,40). The SP form tumors within a brief period, whereas several weeks are required for NSP to form a palpable tumor, and NSP tumors cannot metastasize (38). Thus, finally, to detect the MAZ protein levels in cancer stem cells/tumor-initiating cells, MAZ expression was evaluated in the SP and NSP of Panc-1 and AsPC-1 cells using Western blot analysis. The MAZ level was significantly higher in the SP as compared with the NSP (Fig. 2A).
To determine the pathophysiological relevance of MAZ expression in SP, we analyzed MAZ expression in a subcutaneous tumor representative Northern blotting demonstrates the differential expression of MAZ mRNA in total lysates of different pancreatic cell lines from less aggressive (BxPC-3) to highly aggressive cell lines (AsPC-1 and Panc-1) along with normal HPDE cell line. GAPDH was used as loading control. Migration of an RNA marker (Kb) is indicated on the left. B, representative Western blotting demonstrates the relative levels of MAZ protein in total lysates of different pancreatic cell lines from less aggressive (BxPC-3) to highly aggressive cell lines (AsPC-1 and Panc-1 along with normal HPDE cell line in the left panel. ␤-Actin was used as the loading control. Migration of a molecular mass marker (kDa) is indicated on the left. C, representative Western blotting demonstrates the relative levels of MAZ protein in nuclear and cytoplasmic fractions (Cyto) of Panc-1 cells. GAPDH and ␤-actin were used as for quality detection and as the loading control. The scatter plot indicates means Ϯ S.E. of three independent experiments. p values were calculated using two-tailed unpaired Student's t test. Migration of a molecular mass marker (kDa) is indicated on the left. D, representative Immunofluorescence staining of MAZ in PDAC cell lines. The nuclei were counterstained with DAPI.

MAZ promotes PDAC aggression
xenograft in nude mice generated by either the SP or NSP of Panc-1 cells. The immunohistochemical analysis (n ϭ 3) showed that the MAZ expression was markedly higher in Panc-1 SP tumor sections as compared with NSP tumor sections, where MAZ expression was less intense or almost undetected (Fig. 2B).

MAZ silencing reprogrammed MET and stemness in PDAC cells
It has been reported that MAZ promotes some surrogate in vitro biological markers for invasive phenotypes such as colonyforming ability, migration, and invasion (2). Consistent with previous reports, our current studies found that MAZ ablation in Panc-1 cells blocks the colony-forming ability, invasion, and motility of pancreatic cancer cells toward serum (Figs. S3-S5). We reasoned from these collective studies that MAZ overexpression could be a strong determinant of pancreatic cancer cell fate (i.e. invasive progression). Because EMT/stemness has been implicated in invasive progression and the metastasis of cancer cells (44,45), we thus hypothesized that EMT/stemness in pancreatic cancer cells would be dependent on MAZ over-expression to promote invasive phenotypes (Fig. 3A). To test this hypothesis, MAZ expression was knocked down by MAZspecific siRNA in Panc-1 and AsPC-1 cells (Fig. 3B), and different EMT markers were determined using Western blotting. Although siRNA-mediated silencing of MAZ was significant in both cell lines (Fig. 3B), we found that the expression of epithelial markers was significantly increased, whereas mesenchymal marker levels were markedly reduced in MAZ-depleted cells as compared with scrambled siRNA-transfected cells (Fig. 3, C-F). To validate the siRNA results, MAZ expression was knocked down by two additional MAZ-specific siRNA (siRNA2 and siRNA3) in Panc-1 and AsPC-1 cells, and two different EMT markers were determined using Western blotting (Fig. S6). We found that both siRNAs effectively altered the protein expressions of vimentin (mesenchymal marker) and E-cadherin (epithelial marker) in Panc-1 and AsPC-1 cell lines. After validation of the MAZ siRNAs, siRNA1 was used for further studies.
We finally sought to test the role of MAZ on the in vitro sphere-forming ability of pancreatic cancer cells, a functional consequence of EMT (46). We found that MAZ-depleted Panc-1 and AsPC-1 cell lines significantly lose their capacity to form spheres as compared with MAZ-positive scrambled siRNA-transfected cells (Fig. 4, A and B). Collectively, these studies suggest the degree of EMT/stemness in pancreatic cancer cells corresponds to their level of MAZ expression.

MAZ regulates CRAF-ERK signaling in PDAC cells
Ras family proteins (K-Ras, N-Ras, and H-Ras) are the most critical hubs in cell signaling pathways (47). The somatic mutations in Ras family protein are associated with oncogenic progression (47)(48)(49). K-Ras mutations are the most prevalent in PDAC (20,30,47). They play a critical role in PDAC development and progression (30,50). Previous studies have suggested that MAZ activates K-Ras transcription via binding into the G4-DNA of the K-Ras promoter in pancreatic cancer cells (7). However, analysis of the human pancreatic adenocarcinoma array found that MAZ ablation in Panc-1 cells minimally but not significantly decreased K-Ras mRNA expression, whereas mRNA expression of various cancer-promoting genes like MMP1, MMP-7, VEGF, IL6, Cyp2E1, and CDC42 were markedly reduced (Fig. 5, A and B). To validate the data, the impact of MAZ depletion on K-Ras protein expression was evaluated. We found that the K-Ras protein level was unchanged in scrambled siRNA and MAZ siRNA-transfected Panc-1 cells and AsPC-1 cells in which MAZ expression was reduced significantly ( Fig. 5C and Fig. S7).
The CRAF-MEK-ERK cascade is a critical mediator of Ras-dependent oncogenic progression in various cancers including pancreatic cancer (51)(52)(53). Thus, our goal was to determine the impact of MAZ on CRAF expression. We found that CRAF levels were unchanged in both MAZ siRNA and scrambled siRNAtransfected Panc-1 cells (Fig. 5D) and AsPC-1 cells (Fig. S7).
The next goal was to evaluate whether MAZ depletion impairs the activities of CRAF-ERK signaling in K-Ras mutant PDAC cells. Because CRAF phosphorylation at Ser-338 and dephosphorylation at Ser-259 are required to activate CRAF in the plasma membrane and promote RAS-RAF interaction (54,55), the effect of MAZ on CRAF phosphorylation at Ser-338
Earlier studies have shown that Ser-338 phosphorylation is regulated by p21-activated protein kinases (PAK) (56,57), whereas phosphorylation at Ser-259 of CRAF is mediated through protein kinase B (AKT/PKB) (58,59). Given the importance of these two kinase proteins in the functional regulation of CRAF, the impact of MAZ on the activities of PAK and AKT were tested. MAZ depletion in PDAC cells significantly reduced PAK activity, whereas AKT activity was expressively increased, thus indicating that MAZ-induced Ser-338 phosphorylation and Ser-259 dephosphorylation could be mediated through differential regulation of PAK and AKT in PDAC cells (Fig. 6, A and B, and Fig. S7). In support of this possibility, we observed a significant decreased in AKT phosphorylation and increased in PAK phosphorylation in MAZ-overexpressed HPDE cells generated by stable transfection of MAZ-expressing vector (Fig. 6, C-E). To validate the above results, the MAZ-transfected HPDE cells were grown in the presence or absence of a PAK inhibitor (PAKi) (FRAX486) (60), ERK inhibitor (ERKi) (U0126), or AKT inhibitor (AKTi) (wortmannin), and the phosphorylation levels of Ser-338 and Ser-259 were determined using Western blotting. The studies showed that S338 phosphorylation was significantly increased in MAZ-overexpressing HPDE cells, and this expression was blocked by PAKi. However, ERKi was unable to suppress the phosphorylation of S338 (Fig. 7A). The combined studies demonstrate that MAZ-induced S338 phosphorylation of CRAF is mediated through the activation of PAK.
Finally, we determined the activity of ERK in MAZ-overexpressed HPDE cells by assessing the ERK phosphorylation. The studies show that the ERK phosphorylation was significantly increased in MAZ-overexpressed HPDE cells as compared with vector-alone-transfected HPDE cells (Fig. 7C). MAZ does not affect total ERK protein expression (Fig. 7C). Collectively, these studies further support the notion that MAZ is a prime regulator of the CRAF-ERK signaling pathway via controlling PAK and AKT activities.

MAZ overexpression derives invasive phenotypes in normal pancreatic cancer cells via activation of CRAF signaling
Previous studies have demonstrated that CRAF-ERK promotes EMT in prostate cancer cells (61). Thus, we hypothesized that MAZ regulation of invasive phenotypes in pancreatic cancer cells is mediated by CRAF-ERK activation (Fig. 8A). To test this hypothesis, EMT and sphere formation were determined in MAZ-overexpressing HPDE cells in the presence or absence of CRAFi (ZM-336372). The expression of epithelial markers was significantly decreased, whereas the expression of mesenchymal markers was markedly elevated in MAZ-overexpressing HPDE cells as compared with vector-alone-transfected HPDE cells. We also observed that MAZ-induced EMT was significantly suppressed by the CRAF inhibitor (Fig. 8, B and C). Similarly, the studies found that the sphere-forming ability and migration toward SDF-1␣ (a chemoattractant) of HPDE cells were dramatically increased in MAZ-transfected cells and that the effect of MAZ was blocked by CRAFi (Fig. 8, D and E). Collectively, these studies suggest that CRAF signaling is required to induce invasive phenotypes by MAZ.

MAZ expression is regulated by CCN1 in pancreatic cancer cells
Having identified the close expressional and functional proximities of CCN1 and MAZ (Refs. 2, 28, 37, and 38 and our current studies), we speculated that MAZ and CCN1 might be interrelated. To test this possibility, we first determined whether expressions of MAZ and Cyr61/CCN1 are co-localized in pancreatic cancer tissue sections by using an immunofluorescence assay. As expected, both MAZ and Cyr61/CCN1 were co-expressed in the tumor cells of a tissue section (Fig. 9A). MAZ was predominantly expressed in the nucleus, whereas Cyr61/CCN1 was expressed in the cytoplasm.
Although, both proteins are co-localized, it is unclear whether Cyr61/CCN1 regulates MAZ expression or whether MAZ, as a transcription factor, promotes Cyr61/CCN1 expression in PDAC cells. Thus, the status of MAZ protein was evaluated in CCN1 neutralizing antibody-treated Panc-1 cells and CCN1 knocked-down Panc-1 cells using Western blot analysis. A significant down-regulation of MAZ was found in both CCN1 antibody-treated and CCN1 knocked-down Panc-1 lysates (Fig. 9B). After subcellular fractionation, nuclear expression of MAZ was also reduced in CCN1 antibody-treated cells or CCN1-depleted cells (Fig. 9C). Moreover, BxPC-3 cells, which have less CCN1 and MAZ expression, when treated with human recombinant CCN1 (hrCCN1) exhibited significant upregulation of MAZ expression (Fig. 9D).
Previously, we have shown that Cyr61 regulates SHh signaling to promote invasive phenotypes (37). The present studies found

MAZ promotes PDAC aggression
that the SHh inhibitor cyclopamine significantly blocks MAZ protein expression in Panc-1 cells (Fig. 9F), suggesting that SHh signaling is also involved in MAZ regulation in PDAC cells.
The fate of CCN1 in MAZ-depleted Panc-1 cells was evaluated. Neither Western nor Northern blot analyses found changes in CCN1 expression (Fig. 9E). Altogether, these results indicate that CCN1 is an upstream regulator of MAZ and that Cyr61/CCN1 regulates MAZ via SHh signaling pathway (Fig. 9H).

Discussion
In this study, we investigated the pathobiological importance of the MAZ transcription factor in the progression of PDAC and uncovered the up-and downstream signature molecules through which MAZ controls tumor aggressiveness (summarized in Fig. 10). This schematic model elucidates all aspects of our findings, including explaining how MAZ following induc- The p values were calculated using two-tailed unpaired Student's t test. Migration of a molecular protein marker (kDa) is indicated on the right. E, representative Western blotting depicts the level of pERK, ERK, and ␤-actin in MAZ siRNA-or scrambled siRNA-transfected Panc-1 cells. ␤-Actin was used as the loading control. All data represent means Ϯ S.E. of three independent experiments. The p values were calculated using two-tailed unpaired Student's t test. Migration of a molecular protein marker (kDa) is indicated on the right.

MAZ promotes PDAC aggression
tion by Cyr61/CCN1 raises the threshold of the activity of the K-Ras signal without affecting RAS expression and promotes invasive phenotypes in PDAC cells.
The oncogenic role of MAZ in PDAC and its utility as a biomarker for poor prognosis has already been shown (2). Our studies support the previous reports and suggest that MAZ is differentially overexpressed in human PDAC samples with the increment of the grade. The differential overexpression of MAZ was also detected in various human cell lines and PDAC samples of KC and KPC mouse models (Fig. 1). MAZ overexpression was first detected in the pancreatic intraepithelial neoplasm stage, and expression gradually increased as the disease progressed. Interestingly, the side-population studies indicate that MAZ is predominantly overexpressed in the cancer stem cell population as a whole (Fig. 2), thus suggesting that MAZ may play a vital role in maintaining EMT and cancer stem cell behavior of pancreatic cancer cells as well as promoting tumor growth; these pathobiological properties are lesser or lacking in NSP cells. Indeed, our further studies have validated the hypothesis and, together with previous studies (2), specify that MAZ-expressing pancreatic cancer cells possess more aggressive behaviors (i.e. EMT, migration, and sphere-forming ability (Figs. 3, 4, and 8)) as compared with MAZ-depleted cells.
The oncogenic K-Ras mutations are one of the earliest, and most critical, of the events in the development and progression of PDAC via activation of the Raf family of serine/threonine kinase (47,50,(63)(64)(65). Previous studies suggest that MAZ binds to a GA-box with consensus sequence GGG(A/C)GG located in the K-Ras promoter (4) and activates K-Ras transcription (66). Moreover, other studies have found that sequestering the MAZ protein by a stable K-Ras promoter analog prevents MAZ from activating K-Ras transcription and delays tumor growth in mice (7). Given these studies, we hypothesized that the regulation of the phenotypic shift and aggressive behavior of PDAC cells by MAZ could be mediated through the regulation of K-Ras transcription. However, the findings presented here are based on two independent experimental approaches (Fig. 5), and the results from both are consistent in showing that MAZ does not up-regulate K-Ras mRNA or protein in PDAC cells, but rather MAZ activates CRAF-ERK, the downstream signaling molecules of K-Ras (Figs. 5 and 6).
CRAF phosphorylation at Ser-338 and dephosphorylation at Ser-259 are prerequisite events for the activation of CRAF that are needed to promote RAS-RAF interaction (54,55). Phosphorylation at Ser-338 and Ser-259 are regulated by PAK (56,57), and AKT/PKB, respectively (58,59). Our studies show that MAZ promotes Ser-338 phosphorylation by activating PAK, whereas it promotes Ser-259 dephosphorylation by suppressing AKT/PKB activity in PDAC cells (Figs. 5 and 6). Thus, these events could comprise a mechanism that contributes to CRAF activation by MAZ in PDAC cells.
The convergence of MAZ overexpression and the treatment of the CRAF inhibitor or AKT/PKB inhibitor in HPDE cells indicate that the increased phenotypic shift from epithelial to mesenchymal and stemness by MAZ in pancreatic cancer cells could be mediated by the CRAF-ERK signaling pathway via blocking phosphorylation of the Ser-259 site of RAF (Figs. 6 -8). Despite these provocative findings, it remains unclear how MAZ, being a transcription factor, regulates PAK and AKT activities via phosphorylation and dephosphorylation. Previous studies have shown that transcription factors such as FoxO, STAT3, and ATF3 activate AKT activities via AKT phosphorylation and kinase activities (67-69), which could be caused mainly by the suppression of protein phosphatase 2 (PP2A) and protein phosphatase 3 (calcineurin) (68). Further, the control of PAK signaling by transcription factor p53 has also been reported, indicating that p53 mediates Bcl-2 phosphorylation and apoptosis via activation of the Cdc42/PAK/JNK1 pathway (70). Based on these studies, we suggest that MAZ regulation of AKT and PAK activities could be mediated via a similar path-

MAZ promotes PDAC aggression
way or other unknown pathways. To uncover these vital issues, further studies are warranted.
Our final question was how MAZ is regulated in pancreatic cancer cells. Having discovered the close expressional and functional proximities of Cyr61/CCN1 and MAZ (2,28,37), one possible explanation is that CCN1 may activate MAZ or vice versa for the pathobiological needs in pancreatic cancer cells. We conducted detailed in vitro studies to uncover the relationship between Cyr61/CCN1 and MAZ in pancreatic cancer cells. The studies showed that suppression of Cyr61/CCN1 by shRNA or Cyr61/CCN1 antibody treatment led to decreased MAZ expression and its downstream signaling molecules (CRAF-ERK) in Panc-1 cells, whereas Cyr61/CCN1 treatment increased MAZ expression in BxPC3 cells. In contrast, MAZ depletion showed no effect on Cyr61/CCN1 expression in Panc-1 cells. From this study, we propose that Cyr61/CCN1 regulates MAZ expression in pancreatic cancer cells. Further studies will be needed to provide evidence that Cyr61/CCN1-induced up-regulation of MAZ is mediated through SHh signaling (Fig. 9).
In conclusion, our study has elucidated the underlying mechanism of MAZ-induced invasive phenotypes in pancreatic cancer cells and have uncovered the relationship between Cyr61/CCN1 and MAZ in PDAC. This study suggests the idea that inhibition of MAZ could be an anti-PDAC therapeutic avenue to achieve blockage or delay the progression of the deadly disease by overcoming K-Ras addition of pancreatic cancer cells (19,71).

Ethics statement
The animal studies were approved by the Kansas City Veterans Affairs Medical Center Animal Care and Use Committee in accordance with the AAALAC animal care guidelines and the National Institutes of Health Guide for the Care and Used of Laboratory Animals. Mice were monitored daily and euthanized if they displayed excessive discomfort. The mice were fed regular commercial mouse diet (without tetracycline) with a 12-h lightdark cycle.

MAZ promotes PDAC aggression
For subcutaneous xenograft studies, male athymic nude mice were used and purchased from The Jackson Laboratory. SP and NSP Panc-1 cells (5 ϫ 10 4 ) were injected subcutaneously into the right flank of 6 -8-week-old nude mice. Tumor growth was monitored beginning the following day and continued up to 50 days. The tumor tissues were collected and saved for further experiments.

Human samples and primary culture
The pancreatic cancer tissue array, with cancer-adjacent tissue and normal tissue as the control, including tumor-node-metastasis (TNM), clinical stage, and pathology grade, were obtained from US Biolab Corp. (Rockville, MD) and the University of Kansas Medical Center core facilities. Fresh pancreatic cancer biopsy samples were obtained from endoscopic ultrasound-guided fine-needle aspiration at the Veterans Affairs Medical Center, Kansas City, MO. The samples were collected, stored, and used with the informed consent of the patients who had pancreatic adenocarcinoma and were enrolled in the study (IRB protocol No. 00868). The studies abided by the Declaration of Helsinki principles. The pancreatic adenocarcinoma samples, which were confirmed by an experienced pathologist at the Kansas City Veterans Affairs Medical Center, were collected and grown on a Petri dish. To grow tumor cells without cognate stroma, the cells were serially centrifuged to separate the epithelial cells.
For subcutaneous xenograft studies, male and female athymic nude mice were used and were obtained from The Jackson Laboratory. SP and NSP Panc-1 cells (5 ϫ 10 4 ) were injected subcutaneously into the right flank of 6 -8-week-old nude mice. Tumor growth was monitored beginning the following day and continued up to 50 days. The tumor tissues were collected and saved for further experiments.

Cell lines and cell culture
Pancreatic cancer cell lines (BxPC-3, AsPC-1, and Panc-1) were purchased from American Type Culture Collection (ATCC, Manassas, VA). The cell lines were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), penicillin (100 units/ml), and streptomycin (100 units/ml) at 37°C incubator in the presence of 5% CO 2 . The HPDE cell line was a gift from the University Health Network (Toronto, Canada) and grown in keratinocyte serum-free media supplemented with EGF and bovine pituitary extract (Life Technologies Inc.). Approximately 60% confluent cells and not more than six passages were used for each experiment. The authentication of these cells was verified using short tandem repeat profiling (STR analysis of DNA), and mycoplasma contamination was checked every 60 days.

Reagents and antibodies
DMEM was obtained from Sigma, and FBS was purchased from Hyclone (Logan, UT). Human polyclonal anti-rabbit MAZ antibody was purchased from Bethyl Laboratories (Montgomery, TX). Mouse monoclonal antibodies against ␤-actin and CCN1 were purchased from Sigma. The antibodies against lamin 1B, glyceraldehyde-3-phosphate dehydrogenase   (GAPDH), cytokeratin 19, E-cadherin, ␤-catenin, vimentin, CD44, Twist, K-Ras, pCRAF, CRAF, pERK, and ERK were purchased from Cell Signaling Technology (Danvers, MA). NE-PER nuclear and cytoplasmic extraction reagents were purchased from Thermo Scientific. All other chemicals were obtained either from Sigma or Fisher Scientific. The authentication certificates for all chemicals, drugs, and antibodies were obtained from these companies. The fresh working solutions of the chemicals and drugs were prepared once a month to guarantee effectivity.

Isolation of NSP and SP cells by flow cytometry
The SP and NSP of Panc-1 cells were isolated by a BD FACSAria SORP flow cytometer (BD Biosciences) using ϳ405-nm excitation and 440-nm emission as described previously by Haque et al. (38). The sorted cells (i.e. SP and NSP) were cultured briefly in DMEM with 10% FCS in 5% CO 2 at 37°C, and cell lysates were extracted for further evaluation.

In situ hybridization
In situ hybridization was carried out essentially as described by the manufacturer (Innogenex, San Ramon, CA) with minor modifications (72,73).

Subcellular fractionation
Semiconfluent (ϳ60%) cells were scraped into cold PBS and collected by centrifugation at 500 ϫ g for 5 min. The cells were washed once with PBS and centrifuged, and the cell pellet was suspended into ice-cold cytoplasmic extraction reagent 1 (CER1) and incubated on ice. After 10 min, ice-cold CER II was added to the tube. Samples were spun at 16,000 ϫ g for 5 min to generate the cytosolic fraction (supernatant). The insoluble (pellet) fraction was suspended in ice-cold nuclear extraction reagent. The pellet was homogenized by vortexing and centrifuged at 16,000 ϫ g for 10 min to generate the nuclear fraction (supernatant).

Transfection
Mismatched siRNA, MAZ-siRNA (OriGene, catalog no. SR302819) or plasmids containing the MAZ gene (OriGene, catalog no. RC219635) were transfected in Panc-1 cells or HPDE cells by electroporation using the Neon transfection system (Life Technologies) according to our previous methods (74).

Western blotting
The Western blot analysis was the same as described previously (36). Briefly, total protein lysates were extracted from cells using radioimmune precipitation assay lysis buffer. ϳ50 g of total protein lysates was run on SDS-PAGE; the fractionated proteins were transferred to nitrocellulose membrane (Trans-Blot transfer medium, Bio-Rad) and subsequently probed with primary antibodies (1:1000 dilutions) overnight at 4°C. Following a washing of the membranes with TBS-T, the membranes were incubated with secondary antibodies conjugated with horseradish peroxidase for 30 min at room temperature. Immunoreactions (antigen-antibody) were detected by Enhanced Chemiluminescence (ECL, Pierce) using a Kodak imaging system (Carestream Health, Rochester, NY).

Northern blot analysis and quantitative real-time PCR (qPCR) analysis
For Northern blot analysis, total RNA concentrations in cell extracts were measured using nonradioactive Northern blot analysis as described previously (75). Briefly, total RNA was separated by formaldehyde/agarose gel electrophoresis and transferred to a nylon membrane. The membranes were hybridized with a digoxigenin-labeled gene-specific probe. Relative expressions of mRNA were calculated by densitometry using one-dimensional image analysis software (Kodak Image Station). GAPDH was used as a loading control. For qPCR, RNA extraction and cDNA preparation were performed as described previously (76). The qPCR was performed on a StepOne real-time PCR system (Applied Biosystems). SYBR Green master mix (Applied Biosystems) was used with the MAZ and GAPDH primers. The relative amount of mRNA was calculated by the comparative ⌬Ct method with GAPDH as an internal control.

Colony formation assay
Anchorage-dependent growth of Panc-1 cells was assessed by a colony formation assay using a Cyto-Select 96-well cell transformation assay kit (Cell Biolabs, Inc., San Diego, CA) following the manufacturer's instructions. Briefly, Panc-1 cells were transfected with control siRNA or MAZ siRNA and then seeded on 96-well plates. The cells were incubated at 37°C in a humidified incubator with 5% CO 2 for 8 days. The cell numbers were counted using SpectraMax 340 (Molecular Devices).

In vitro invasion assay
In vitro invasion assays were performed as described previously (36). Briefly, control siRNA and MAZ siRNA-transfected Panc-1 cells (10,000 cells/well) were seeded onto a Matrigelcoated (invasion) top chamber insert containing serum-free DMEM. The complete DMEM (supplemented with 10% FBS) was added to the bottom chamber as a chemoattractant. After 48 h, the migrated cells that were attached to the outer surface of the insert were stained with 0.1% crystal violet solution. Crystal violet-stained cells were solubilized with 10% acetic acid, and optical density was quantitated in a microplate reader at 600 nm. Three wells were examined for each condition, and the experiments were repeated three times.

Wound-healing assay
The wound-healing assay was performed per our previous method (37). Briefly, mismatched and MAZ siRNA-transfected Panc-1 cells were plated in 6-well plates. The monolayer confluent cultures were scratched with a sterile tip, washed with serum-free medium, and cultured in DMEM containing 2% FBS. Cells were photographed at different times. Wound closure was evaluated using TScratch software (77).

Sphere formation assay
A sphere-forming assay was performed as described previously (22). Briefly, cells were seeded (1.0 cell/l) into ultralow MAZ promotes PDAC aggression attachment Petri dish containing serum-free MammoCult medium supplemented with 4 g/ml heparin and 0.48 g/ml hydrocortisone (Stem Cell Technologies, Vancouver, Canada). The sphere cultures were supplemented with fresh medium on alternative days and could grow for days. The numbers and volumes of the spheres were determined by Leica microscopy using NIS-Elements BR imaging software (Leica Microsystems, Wentzler, Germany). In this study, spheres less than 2000 m 2 were not considered ideal.

Human pancreatic adenocarcinoma mRNA array
Different mRNA profiles associated with pancreatic cancer progression and metastasis were analyzed in parental and MAZ-depleted Panc-1 cells using a TaqMan human pancreatic adenocarcinoma mRNA array (Applied Biosystems, catalog no. 4414177) as per the manufacturer's protocol. Briefly, total RNAs were extracted from scrambled siRNA or MAZ siRNA-transfected Panc1 cells using TRIzol reagent (Thermo Fisher Scientific) and reverse-transcribed using a High-Capacity cDNA reverse transcription kit. 10 l of TaqMan gene expression master mix (Applied Biosystems) and 10 l of template cDNA were added to each well of the TaqMan array plates. Quantitative real-time PCR was performed using Applied Biosystems' real-time PCR 7500 systems and SDS software with the following thermal cycle: 2 min at 50°C and 10 min at 95°C followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. Three endogenous genes (GAPDH, HPRT1, and GUSB) were used as internal controls. The results were analyzed using DataAssist TM software (Applied Biosystems). The relative quantitative expression of genes was examined by comparing Ct value.

Immunohistochemistry
Immunohistochemical analyses were performed as per our previous method (22). Briefly, tissue sections were deparaffinized in xylene, rehydrated in different grades of alcohol, washed with PBS, blocked with tissue blocker (Life Technologies), and immunostained by polyclonal human MAZ antibody (1:1000) overnight. Following counterstaining with hematoxylin (blue), images were captured under a Leica photomicroscope.

Statistical analysis
All statistical analyses were performed using the GraphPad Prism 6.05 software package (GraphPad Software). All data are expressed as means Ϯ S.E. Statistically significant differences between groups were determined by ANOVA and the paired Student's two-tailed t test. A value of p Ͻ 0.05 was considered statistically significant.