MYB Promotes Desmoplasia in Pancreatic Cancer through Direct Transcriptional Up-regulation and Cooperative Action of Sonic Hedgehog and Adrenomedullin*

Extensive desmoplasia is a prominent pathological characteristic of pancreatic cancer (PC) that not only impacts tumor development, but therapeutic outcome as well. Recently, we demonstrated a novel role of MYB, an oncogenic transcription factor, in PC growth and metastasis. Here we studied its effect on pancreatic tumor histopathology and associated molecular and biological mechanisms. Tumor-xenografts derived from orthotopic-inoculation of MYB-overexpressing PC cells exhibited far-greater desmoplasia in histological analyses compared with those derived from MYB-silenced PC cells. These findings were further confirmed by immunostaining of tumor-xenograft sections with collagen-I, fibronectin (major extracellular-matrix proteins), and α-SMA (well-characterized marker of myofibroblasts or activated pancreatic stellate cells (PSCs)). Likewise, MYB-overexpressing PC cells provided significantly greater growth benefit to PSCs in a co-culture system as compared with the MYB-silenced cells. Interrogation of deep-sequencing data from MYB-overexpressing versus -silenced PC cells identified Sonic-hedgehog (SHH) and Adrenomedullin (ADM) as two differentially-expressed genes among others, which encode for secretory ligands involved in tumor-stromal cross-talk. In-silico analyses predicted putative MYB-binding sites in SHH and ADM promoters, which was later confirmed by chromatin-immunoprecipitation. A cooperative role of SHH and ADM in growth promotion of PSCs was confirmed in co-culture by using their specific-inhibitors and exogenous recombinant-proteins. Importantly, while SHH acted exclusively in a paracrine fashion on PSCs and influenced the growth of PC cells only indirectly, ADM could directly impact the growth of both PC cells and PSCs. In summary, we identified MYB as novel regulator of pancreatic tumor desmoplasia, which is suggestive of its diverse roles in PC pathobiology.

Pancreatic cancer (PC) 3 is one of the most lethal malignancies in the United States, with a dismal median survival rate of 2-8 months after diagnosis and a 5-year overall survival rate of ϳ7% (1). According to an estimate by the American Cancer Society, it is expected to overtake breast cancer to become the third leading cause of cancer-related death in the United States (2). Nearly 53,070 patients are expected to be diagnosed with PC this year and ϳ41,780 will succumb to it (1). This continued increase in mortality rates is alarming and necessitates that we generate a better understanding of PC biology by identifying key molecular determinants of its pathological phenotypes, so that novel and more effective mechanism-based therapeutic approaches can be developed.
Pancreatic tumors are highly desmoplastic in nature exhibiting the presence of dense, fibrous connective tissue surrounding the tumor cells (3,4). Desmoplasia is characterized by an increase in the presence of ␣-smooth muscle actin (␣-SMA)positive fibroblasts (myofibroblasts or activated pancreatic stellate cells) along with extensive deposition of extracellular matrix (ECM) proteins (5). This desmoplastic reaction not only gives pancreatic tumors their characteristic histoarchitecture, but is suggested to play important roles in disease progression and chemoresistance (6). Pancreatic stellate cells (PSCs) are the major cellular component of the desmoplastic stroma that synthesize and secrete several ECM components within tumor microenvironment (7)(8)(9)(10). Pancreatic tumor cell-derived signaling ligands have been shown to promote proliferation and differentiation of PSCs, and targeting of these signaling nodes is suggested to improve the efficacy of chemotherapy in preclinical models (5,6).
MYB, a cellular progenitor of v-Myb oncogene, encodes an oncogenic transcription factor that is shown to regulate a variety of cellular function by controlling expression of a wide array of genes (11). Genetic alterations of MYB in humans were first reported in acute myelogenous leukemia (12). However, in subsequent years, several studies have demonstrated genetic alterations or deregulated MYB expression in many other human cancers as well, including pancreatic cancer (13)(14)(15)(16). Recently, we demonstrated that MYB serves as a novel regulator of pan-creatic tumor growth and metastasis (17). Its expression was reported in majority of PC tissues and cell lines, while it remained largely undetectable in normal pancreatic cells. We also identified several gene targets of MYB that may mediate its impact on PC pathogenesis. However, its in-depth mechanistic involvement and diverse effects on pancreatic tumor phenotypes are not yet well understood.
Here, we provide evidence supporting a role of MYB in pancreatic tumor desmoplasia. We show that the tumors derived from MYB-overexpressing PC cells exhibit greater desmoplasia, characterized by the expression of myofibroblast marker (␣-SMA) and extracellular-matrix proteins (collagen I and fibronectin). We also establish SHH (Sonic hedgehog) and ADM (Adrenomedullin) as direct transcriptional targets of MYB that mediate its effects on the growth of pancreatic tumor and stellate cells via autocrine and/or paracrine signaling. Together, our studies establish MYB as a key driver of PC progression and metastasis by directly impacting the tumor cells as well as by altering the tumor microenvironment.

Results
Tumors Derived from MYB-overexpressing Pancreatic Cancer Cells Exhibit High Desmoplasia-We performed histopathological analyses of tumor xenografts derived from recently developed orthotopic mouse model of pancreatic cancer (17). Since tumors were derived from MYB-overexpressing and -silenced cells, we observed expected differences in MYB expression in tumor xenograft sections as well (data not shown). When stained with H&E, tumor sections from high MYB-expressing group exhibited dense desmoplastic regions surrounding the pancreatic tumor cells, while only minimal desmoplasia was observed in that of low MYB-expressing group (Fig. 1A). This observation was further confirmed by immuno-staining the tumor sections for collagen I, fibronectin, and ␣-SMA, which are well-characterized markers of desmoplasia (5). A dramatic decrease in the expression of collagen I, fibronectin, and ␣-SMA was detected in tumors generated from MiaPaCa-shMYB cells compared with those derived from MiaPaCa-NT-Scr cells (Fig. 1, B-D). Likewise, we also observed high desmoplasia associated with elevated expression of collagen I and ␣-SMA in tumors derived from MYB-overexpressing BxPC3-MYB cells (data not shown). Together, these findings suggest that MYB may serve as a novel regulator of pancreatic tumor desmoplasia.

MYB Expression in Pancreatic Tumor Cells Is Associated with Growth Induction and Differentiation of Pancreatic Stellate
Cells in a Co-culture System-Since pancreatic stellate cells (PSCs) are the major cellular component of the stromal compartment of pancreatic tumors and responsible for desmoplastic reaction, we next investigated the functional consequences of their interaction with high and low MYB-expressing pancreatic cancer cells (PCCs) in a co-culture system. For this, we used matched cell lines, MiaPaCa-NT-Scr/MiaPaCa-shMYB and BxPC3-Neo/BxPC3-MYB that are genetically engineered to serve as control or MYB-knockdown/overexpressing cells (17). Differential MYB expression in these cell lines was confirmed by immunoblot assay prior to any functional assay (data not shown). PSCs co-cultured with BxPC3-MYB and MiaPaCa-NT-Scr cells exhibited significant enhancement in their growth (58.4 and 51.8%, respectively), while it was minimal in PSCs co-cultured with BxPC3-Neo and MiaPaCa-shMYB cells (10.4 and 9.2%, respectively) ( Fig. 2A). Enhanced growth of PSCs also correlated with increased expression of ␣-SMA (differentiation marker of PSCs) (Fig. 2B). In addition, we also monitored the growth of high and low MYB expressing PCCs when co-cul- tured with PSCs. Data demonstrate that the growth of MYBoverexpressing BxPC3-MYB and MiaPaCa-NT-Scr cells was induced (ϳ53 and 47%, respectively) when co-cultured with PSCs, whereas only little induction was observed (ϳ8%) in low MYB-expressing BxPC3-Neo and MiaPaCa-shMYB cells (Fig.  2C). Taken together, our data suggest that MYB overexpression in PC cells is of mutual significance for both pancreatic tumor and stellate cells when they co-exist and communicate freely with each other.
ADM and SHH Are Novel Direct Transcriptional Targets of MYB in Pancreatic Cancer Cells-To understand the molecular basis of MYB-potentiated phenotypic changes, we interrogated our differential gene expression data from deep sequencing analysis of MYB-expressing and -silenced pancreatic tumor cells (GEO accession number GSE61290). 4 Among several differentially expressed genes, we found a decreased expression of ADM (adrenomedullin) and SHH (sonic hedgehog) in MYBsilenced cells, which have previously been implicated in the development of desmoplasia in pancreatic tumors (5,18). To ascertain the effect of MYB silencing on ADM and SHH, we analyzed their expression in MYB-expressing (MiaPaCa-NT-Scr) and -silenced (MiaPaCa-shMYB) cells as well as MYB-null BxPC3 and ectopically MYB-expressing BxPC3-MYB cells at both mRNA (by quantitative RT-PCR) and protein (by immunoblot) levels. We observed enhanced expression of ADM and SHH in high MYB-expressing cells as compared with low MYB or MYB-null pancreatic cell lines both at the transcript (Fig. 3, A and B) and protein (Fig. 3C) levels. Since both ADM and SHH are secreted proteins, we also measured their levels in culture supernatants of MYB-overexpressing/silenced PC cells by ELISA. As expected, significantly enhanced levels of ADM (3.6fold; Fig. 3D) as well as SHH (5.3-fold; Fig. 3E) were recorded in the medium secretions of BxPC3-MYB in comparison to the BxPC3-Neo cells, while these were decreased considerably (ADM, 5.4-fold; SHH, 6.8-fold) in MYB-silenced MiaPaCa cells (Fig. 3, D and E). Tumor sections derived from mice of high and low MYB-expressing groups also exhibited similar association in IHC analyses (Fig. 3F).
To examine if ADM and SHH were direct transcriptional control of MYB, we performed in silico analyses of ϳ1-kb DNA region 5Ј upstream of their coding DNA sequence (CDS) (GenBank TM accession numbers D43639 and AY422195.1, respectively) using web-based applications (ALGGEN-PROMO and TFsearch). Several putative MYB binding sites were detected within this region of both ADM and SHH (Fig. 4A). We next conducted ChIP analyses to confirm the direct binding of MYB to the ADM and SHH promoters. This was followed by PCR amplification of ChIPed DNA using primers in the flanking regions of putative MYB binding sites (Fig. 4A). It was revealed that MYB was able to bind to the ADM and SHH promoters at least in the regions that are in close proximity to transcriptional initiation sites (Fig. 4, B and C). Furthermore, MYB-silenced MiaPaCa-shMYB cells exhibited significantly reduced PCR amplification signal for both ADM and SHH upstream regions supporting the specificity of MYB-dependent chromatin pulldown (Fig. 4, B and C, respectively). Together, these findings establish MYB as a novel transcriptional regulator of ADM and SHH in pancreatic tumor cells.
To ascertain if the impact of ADM and SHH was solely paracrine or involve an autocrine loop as well, we treated high MYB- 4 Transcript profiling: GEO accession number GSE61290. FIGURE 2. MYB promotes growth of pancreatic stellate cells. A, pancreatic stellate cells (PSCs; 2 ϫ 10 5 /well) were seeded in the 6-well plates, and pancreatic cancer cells (PCCs; 1 ϫ 10 5 /well) were seeded into insert chamber. After 24 of culturing, insert having PCCs was placed over the well containing PSCs and allowed to grow for 144 h with medium replacement after every 48 h. Induction of growth in PSCs relative to respective monocultures (PSCs/ PSCs) was examined by counting the number of viable cells. B, PCCs (1 ϫ 10 5 /well) were seeded in the 6-well plates, and PSCs (2 ϫ 10 5 /well) were seeded into insert chamber and growth of PCCs was examined by counting the number of viable cells after 144 h as described above. For control, both well and insert chambers were seeded with PCCs. Data presented as mean Ϯ S.D., n ϭ 3; *, (p Ͻ 0.05). C, total protein from PSCs co-cultured with high (BxPC3-MYB and MiaPaCa-NT-Scr) and low (BxPC3-Neo and MiaPaCa-shMYB) MYB-expressing cells was isolated and the expression of ␣-SMA was examined by immunoblot assay. ␤-actin was used as internal control.
expressing PC cells with ADM-(22-52) and 5E1 alone or in combination. Similarly, we also treated low MYB-expressing cells with ADM-(1-52) and/or r-SHH. We observed that the growth of high MYB-expressing PC cells was inhibited (22.3%, BxPC3-MYB; 23.9%, MiaPaCa-NT-Scr) when treated with ADM-(22-52); however, limited effect was observed upon Subsequently, medium was replaced with 2% FBS-containing medium and allowed to grow for next 48 h. Thereafter, culture supernatants were collected and centrifuged to clear the cellular debris and levels of (D) ADM and (E) SHH were measured using respective ELISA kits. Expression levels of ADM and SHH were normalized with cell number at end point, and data shown as mean Ϯ S.D., n ϭ 3; *, (p Ͻ 0.05). F, tissue sections of orthotopically developed tumors from MiaPaCa-NT-Scr/MiaPaCa-shMYB PC cells were incubated overnight with ADM or SHH specific antibodies at 4°C. Thereafter, sections were incubated at room temperature with respective polymer and probe, and immunoreactivity was visualized by using DAB Chromogen followed by hematoxylin counterstain. treatment with 5E1 (Fig. 6A). Interestingly, when we treated low MYB-expressing cells with either ADM-(1-52) or r-SHH or combination, we did not observe much induction in their growth (7.9, 6.5, 10.6% in BxPC3-Neo and 8.6, 6.4, 13.1% in MiaPaCa-shMYB, respectively) (Fig. 6B). Treatment of PSCs with ADM-(1-52), r-SHH, ADM-(22-52) and 5E1 alone or in combination confirmed their growth response to both ADM and SHH (Fig. 6C). Taken together, data suggest that ADM and SHH cooperatively mediate the effect of MYB on the growth of pancreatic stellate as well as tumor cells via paracrine and/or autocrine signaling.

Discussion
Although some doubts have been raised regarding the functional significance of extensive desmoplasia in PC pathobiology, the fact remains that it is a unique feature of PC and coevolves with the malignant disease during the course of its progression (6,9,10,19,20). Therefore, it remains of utmost importance to understand the molecular basis of its evolution as well as its biological importance in PC development. Our novel findings provide compelling data to suggest a role of MYB in pancreatic tumor desmoplasia, and thus further adding to its pathobiological significance in PC.
We recently reported a novel role of MYB in pancreatic tumor growth and metastasis (17). These effects were suggested to involve a positive impact of MYB on cell-cycle progression, apoptosis-resistance, as well as malignant features of pancreatic tumor cells via targeting of functionally relevant gene targets. The data presented herein demonstrate that MYB-overexpressing pancreatic tumors also differ in their histopathology. Tumors derived from MYB-expressing PC cells were highly fibrotic, as demonstrated by the immunostaining of ECM proteins, collagen I, and fibronectin, while only minimal reactivity was reported in tumors derived from MYB-silenced PC cells.
Both collagen I and fibronectin are not only major components of pancreatic tumor-associated stroma, but they have also been shown to promote cancer development by inducing EMT and invasiveness of pancreatic tumor cells (21)(22)(23). A major source of these ECM proteins is myofibroblasts, which originate from activated pancreatic stellate cells (PSCs) (7, 8). Accordingly, our  immunohistochemical data demonstrated high levels of smooth muscle ␣-actin (␣-SMA), a well-established marker of myofibroblasts, within desmoplastic stroma of MYB-overexpressing pancreatic tumors. This also suggested that MYBoverexpressing PC cells interact differentially with PSCs to promote desmoplastic reaction during tumor development.
Positive interaction of MYB-overexpressing pancreatic tumor cells with PSCs was also confirmed in our in vitro co-culture studies. MYB-overexpressing PC cells not only promoted the growth of PSCs in co-culture, but also induced the expression of ␣-SMA suggesting their activation and differentiation into myofibroblasts. These effects were less pronounced in MYB-silenced or MYB-null cells thus supporting a role of MYB in this tumor-stromal cross-talk. However, stimuli for PSCs should come from the changes in its microenvironment (24 -26), and specifically in this case from factors secreted by MYBoverexpressing cells. These factors were characterized to be SHH and ADM that exhibited greater expression in MYB-overexpressing cells at RNA, protein and secretome levels. More importantly, our data suggested mostly a paracrine action of SHH, while the other factor ADM, influenced the growth of both PSCs and PC cells and thus likely acted in both paracrine and autocrine manner. These findings were consistent with published data that has suggested mostly a paracrine signaling for pancreatic tumor cell-derived SHH (27). Inhibition of SHH was shown to decrease desmoplasia through direct activation of hedgehog signaling in PSCs (5), while it indirectly affected PC metastasis and lymphangiogenesis due to altered tumor microenvironment (28). In other reports, ADM has also been shown to be up-regulated in PC patients (29). It has been suggested that ADM acts directly on the tumor cells, but also influences tumor progression through its effects on endothelial cells and PSCs (18,30). Moreover, despite some conflicting data suggesting a suppressive effect of PSCs on PC progression and survival (9, 10), our data indicate mutual benefit of pancreatic tumor and stellate cells cross-talk at the molecular level.
Several mechanisms have been reported for tumor-stromal interactions culminating into ECM remodeling (31,32). In fact, not only the tumor cells, but also the hypoxic and inflammatory tumor microenvironment could also influence stromal composition of pancreatic tumors (23,24,26). Studies on bi-directional tumor-stromal cross-talk in PC have reported signaling pathways that could influence the production of SHH and ADM (27,33). Moreover, hypoxic tumor microenvironment has also been shown to promote synthesis and secretion of SHH by pancreatic tumor cells to promote desmoplasia and thus engage in vicious hypoxia-desmoplasia loop (34). Similarly, ADM has also been reported to be up-regulated by hypoxia in pancreatic tumor cells (35). In this regard, our findings establishing MYB as a novel transcriptional regulator of SHH and ADM are of great significance and indicative of important roles of MYB as a mediator in active cross-talk between tumor cells and a variety of factors within the tumor microenvironment. It is likely that MYB, either mediate the tumor-tumor microenvironment signaling or act in cooperation to promote positive tumor-stromal interaction, and thus facilitate pancreatic tumor development.
In summary, we have identified important role of MYB in the growth promotion and differentiation of pancreatic stellate cells through directly regulating SHH and ADM. We have also demonstrated that ADM and SHH cooperatively mediate the effect of MYB on the growth of pancreatic tumor and stellate cells via paracrine and/or autocrine signaling. These findings establishing MYB as a novel regulator of pancreatic tumor desmoplasia are of great significance from the molecular pathogenesis standpoint and provide us novel insight into PC pathobiology.

Experimental Procedures
Cell Lines and Culture Conditions-All the pancreatic cancer cell lines used in this study were procured and maintained as previously described (17). Pancreatic stellate cells (generously gifted by Dr. P. K. Singh, Eppley Cancer Institute, Omaha, NE) were maintained in DMEM supplemented with 20% FBS and 100 M each of penicillin and streptomycin in a humidified atmosphere of 5% CO 2 at 37°C. All the cells were tested and determined to be free of mycoplasma every month and prior to beginning of any functional assay.
Reagents, Gene Constructs, and Antibodies-The following reagents were used: Western Blotting Analysis-Western blotting was performed using standard procedures as described earlier (36,37). Briefly, cell lysates were resolved on 10% polyacrylamide gels and transferred to PVDF membranes. Blots were subjected to a standard immunodetection procedure using specific antibodies against various proteins and visualized using SuperSignal West Femto Maximum sensitivity substrate kit with a LAS-3000 image analyzer (Fuji Photo Film Co., Tokyo, Japan).
Immunohistochemical and Histological Analyses-Immunohistochemical analyses were performed as described earlier (38). All the antibodies were used at 1:100 dilutions except antibody against fibronectin (1:50). Tumor tissue sections were stained with H&E and visualized under microscope (100ϫ and 400ϫ), and photographed.
Enzyme-linked Immunosorbent Assay (ELISA)-Cells (1 ϫ 10 6 /well) were seeded in 6-well plates in regular growth medium. After 24 h, medium was replaced with low (2%) serum-containing medium and cells were further allowed to grow for next 48 h. Thereafter, culture supernatants were collected and centrifuged to clear the cellular debris. ADM and SHH levels were then detected using respective ELISA kit as per manufacturer's instructions.
Co-culture Assay-Pancreatic stellate cells (PSCs; 2 ϫ 10 5 / well) were seeded in the 6-well plates, and pancreatic cancer cells (PCCs; 1 ϫ 10 5 /well) were seeded into insert chamber (1-M pore size). Insert having PCCs was placed over the wells containing PSCs and PSCs/PCCs co-culture was allowed to establish for next 24 h. Thereafter, medium (control or treatment) was replaced after every 48 h and number of viable cells were counted at 144 h by trypan blue dye exclusion assay. In control experiments, both well and insert chambers were seeded with either PSCs or PCCs.
WST-1 Assays-Cells (3 ϫ 10 3 ) were seeded in 96-well plates, and medium was replaced next day with control or treatment medium. Cells were allowed to grow for the next 6 days, and medium replenished every 48 h. Subsequently, WST-1 assay was performed as per the manufacturer's instructions.
Statistical Analysis-All experiments were performed at least three times and numerical data expressed as mean Ϯ S.D. Wher-ever appropriate, the data were also subjected to unpaired twotailed Student's t test, and p Ͻ 0.05 was considered as significant.