Transforming Growth Factor-Β Limits Secretion of Lumican by Activated Stellate Cells Within Primary Pancreatic Adenocarcinoma Tumors

Published OnlineFirst April 28, 2016; DOI: 10.1158/1078-0432.CCR-15-2780

Biology of Human Tumors Clinical Cancer Research Transforming Growth Factor-b Limits Secretion of Lumican by Activated Stellate Cells within Primary Pancreatic Adenocarcinoma Tumors Ya'an Kang1, David Roife2, Yeonju Lee3, Hailong Lv4, Rei Suzuki5, Jianhua Ling6, Mayrim V. Rios Perez1, Xinqun Li1, BingBing Dai1, Michael Pratt1, Mark J. Truty7, Deyali Chatterjee8, Huamin Wang9, Ryan M. Thomas10, Yu Wang11, Eugene J. Koay3, Paul J. Chiao6, Matthew H. Katz1, and Jason B. Fleming1

Abstract

Purpose: Pancreatic ductal adenocarcinoma (PDAC) is lethal examined the effect of lumican on PSCs in a collagen-rich cancer whose primary tumor is characterized by dense composi- environment. tion of cancer cells, stromal cells, and extracellular matrix (ECM) Results: Here we identify PSCs as a significant source of composed largely of collagen. Within the PDAC tumor microen- extracellular lumican production through quantitative IHC vironment, activated pancreatic stellate cells (PSC) are the dom- analysis. We demonstrate that the cytokine, TGF-b, negatively inant stromal cell type and responsible for collagen deposition. regulates lumican gene transcription within HPSCs through its Lumican is a secreted proteoglycan that regulates collagen fibril canonical signaling pathway and binding of SMAD4 to novel assembly. We have previously identified that the presence of SBEs identified within the promoter region. In addition, we lumican in the ECM surrounding PDAC cells is associated with found that the ability of HPSCs to produce and secrete extra- improved patient outcome after multimodal therapy and surgical cellular lumican significantly enhances HPSCs adhesion and removal of localized PDAC. mobility on collagen. Experimental Design: Lumican expression in PDAC from 27 Conclusions: Our results demonstrate that activated pancreatic patients was determined by IHC and quantitatively analyzed for stellate cells within PDAC secrete lumican under the negative colocalization with PSCs. In vitro studies examined the molecular control of TGF-b; once secreted, the extracellular lumican mechanisms of lumican transcription and secretion from PSCs enhances stellate cell adhesion and mobility in a collagen-rich (HPSCs and HPaSteC), and cell adhesion and migration assays environment. Clin Cancer Res; 1–13. 2016 AACR.

Introduction Pancreatic ductal adenocarcinoma (PDAC) remains a leading 1Department of Surgical Oncology,The University of Texas MD Ander- son Cancer Center, Houston, Texas. 2Department of General Surgery, cause of cancer-related death and an unsolved health care The University of Texas Health Science Center at Houston, Houston, dilemma in the United States (1, 2). Histologically, cancer Texas. 3Division of Radiation Oncology, The University of Texas MD 4 cells comprise only a fraction of the PDAC tumor mass with Anderson Cancer Center, Houston, Texas. Department of Hepatobili- fi ary Surgery, The First Affiliated Hospital, School of Medicine, Shihezi the majority consisting of a desmoplastic brotic network of University, Xinjiang, China. 5Department of Gastroenterology and activated fibroblasts and pancreatic stellate cells (PSC), leuko- Rheumatology,The Fukushima Medical University School of Medicine, cytes, and ECM components (3–6). It is well recognized that the Fukushima, Japan. 6Department of Molecular and Cellular Biology,The University of Texas MD Anderson Cancer Center, Houston, Texas. microenvironment plays an important role in pancreatic cancer 7Department of Surgery, Mayo Clinic, Rochester, Minnesota. 8Depart- cell survival, metastatic dissemination, and resistance to ther- ment of Pathology and Immunology, Baylor College of Medicine, apy (3, 7), but the mechanisms of this influence are poorly 9 Houston, Texas. Department of Pathology, The University of Texas understood. MD Anderson Cancer Center, Houston,Texas. 10Department of General Surgery, The University of Florida College of Medicine, Gainesville, Lumican, a member of the small leucine–rich proteoglycan Florida. 11Neurodiagnostics Laboratory, The University of Texas Med- family (SLRP), is present within primary and metastatic tumors ical Branch, Galveston, Texas. derived from various human malignancies, including pancreatic Note: Supplementary data for this article are available at Clinical Cancer cancer (8–10). It is a secreted, collagen-binding ECM protein that Research Online (http://clincancerres.aacrjournals.org/). is highly expressed in connective tissue throughout the body. The Corresponding Author: Jason B. Fleming, Department of Surgical Oncology, protein moiety of lumican binds collagen fibrils and the hydro- The University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Blvd., philic glycosaminoglycans regulate interfibrillar spacing (11–14) Houston, TX, 77030-4426. Phone: 713-794-4258; Fax: 713-745-1921; E-mail: and collagen assembly (15, 16). Its complex and diverse proteo- jbfl[email protected] glycan structure suggests that lumican influences cell function doi: 10.1158/1078-0432.CCR-15-2780 through a variety of mechanisms (17, 18). We have previously 2016 American Association for Cancer Research. demonstrated that the presence of lumican in the ECM

www.aacrjournals.org OF1

Kang et al.

Although suggested by previous observations (31), the molec- Translational Relevance ular mechanisms controlling lumican secretion in human pan- The tumor microenvironment, in particular the tumor– creatic cancer have not been described. In this study, we demon- stroma interaction, is an important factor of disease progres- strate that extracellular lumican is physically colocalized with sion and outcome in patients with pancreatic ductal adeno- PSCs within the collagen dense ECM of primary PDAC tumors. carcinoma (PDAC). Lumican is an extracellular matrix protein When cultured, these PSCs secrete large amounts of lumican and whose presence with primary PDAC tumors is associated with this production is controlled at the transcription level by the TGFb improved patient outcome. Here we identify that activated –SMAD4 signaling axis. This is accomplished through SMAD4 pancreatic stellate cells isolated from primary PDAC tumors binding to novel SBEs we have identified within the promoter secrete large amounts of lumican that augments adhesion and region of the lumican gene. Once secreted, lumican in the extra- migration of activated pancreatic stellate cells in a collagen- cellular space facilitates the adhesion and migration of PSCs rich environment. We found that TGFb, a cytokine associated within a collagen-rich environment. with desmoplasia in PDAC, acts through its canonical signaling pathway to inhibit lumican transcription and secretion in Materials and Methods activated stellate cells. Together, these findings underscore the Ethics statement influential role of lumican within the stroma of localized The xenograft protocol was approved by The University of Texas PDAC and its impact on tumor biology and patient outcome. MD Anderson Cancer Center Institutional Review Board (IRB) under #LAB07-0854. Excess patient tumor was collected only after the planned surgical resection and pathologic examination were surrounding, but not within, PDAC cells is associated with complete. Patient specimens from pancreatectomies performed improved patient outcome after multimodal therapy and surgical between 2009 and 2011 at The University of Texas MD Anderson removal of localized PDAC (19). The cell source and the molec- Cancer Center (UTMDACC; Houston, TX) were selected after ular mechanisms controlling the secretion of lumican into the reviewing medical records and tissue specimens. ECM of PDAC and other solid tumors are currently unknown. Stellate cells are found in solid organs throughout the body Cell lines and reagents (16), and pancreatic stellate cells were first isolated in 1998 by Immortalized human pancreatic stellate cells (HPSCs) har- Apte and colleagues (17). Pancreatic stellate cells comprise vested from primary PDAC tumors were a gift from Dr. Rosa F. 47% of pancreatic parenchyma and normally maintain tissue Hwang (UTMDACC; Houston, TX; ref. 32). Immortalized normal architecture in health and injury states through ECM protein human pancreatic ductal epithelial (HPNE) and 293T cells were a synthesis and degradation. Pancreatic stellate cells usually exist gift from Dr. Paul J. Chiao (UTMDACC, Houston, TX). Human fi in a quiescent state, (18), but become activated in response to pancreatic broblasts (HPF) were purchased from Vitro Bio- injury. Activation of pancreatic stellate cells is marked by alpha- pharma (#SC00A5). PANC-1, BxPC-3, and MiaPaCa-2 cell lines smooth muscle actin (a-SMA) expression and secretion of were obtained from ATCC. MDA-PATC43, 50, 53, and 66 cell lines fibrillar collagen as part of an intense fibrosis until healing were derived in our laboratory (33, 34). These cells lines were fi fi occurs when the stellate cells return to the quiescent state. veri ed as human and unique by DNA ngerprinting, and all cell Within the PDAC tumor microenvironment, activated stellate lines were authenticated by short tandem repeat (STR) DNA fi cells (PSCs) are the dominant stromal cell type, and PSCs have pro ling at the Characterized Cell Line Core Facility of the MD been demonstrated as the cell source of fibrillar collagen within Anderson Cancer Center within 6 months. All cell lines were the desmoplastic ECM of PDAC tumors (20). Recent studies maintained in DMEM (Life Technologies) supplemented with b focusing on tumor–stroma interaction suggest that stroma in 10% FBS at 37 Cin5%CO2. Human recombinant TGF and PDAC acts to restrain cancer cell growth (21) with the depletion lumican (rLUM) were purchased from R&D Systems (#240-B and of a-SMA–positive myofibroblasts (PSCs) in the pancreas of 2846-LU). genetically engineered mice resulting in aggressive cancer for- mation and reduced survival (22, 23). IHC, immunofluorescence staining, and Vectra multispectral The pleiotropic cytokine, TGFb is frequently present in the analysis tumor microenvironment; it is one of the strongest inducers of Tissue microarrays (TMA) were constructed using core samples ECM production during fibrogenesis (24, 25) and desmoplasia from paraffin-embedded blocks of 27 PDAC primary tumors within pancreatic carcinoma (26). When signaling via its canon- surgically removed at MD Anderson Cancer Center (Houston, ical pathway, TGFb binds to TGFb type II receptors, which TX). These were prepared for IHC study at the Clinical Core phosphorylate TGFb type I receptors and in turn phosphorylate Laboratory of MD Anderson Cancer Center. All reactions were cytoplasmic SMAD2 and SMAD3. The phosphorylated SMAD visualized with 3,30-diaminobenzidine (DAB) as a chromogen. protein complex then binds to SMAD4 and is translocated into Isotype controls for all antibodies were negative. The TMA slides the nucleus. Once inside the nucleus, the complex binds to sites with IHC stains were scanned using the Vectra slide scanner within gene promoter regions termed SMAD-binding elements (PerkinElmer) to identify lumican expression signal (DAB stains); (SBE) to regulate transcription. Specific DNA sites identified as a representative digitized image was imported to Nuance (Perki- SBEs include CAGACA and the 8-bp palindromic sequence, nElmer) and the spectral library of hematoxylin and DAB was GTCTAGAC (27–29). Both cancer and host cells within the tumor created. Then, InForm image analysis software (PerkinElmer) microenvironment are sources of TGF-b (30), but the specific was applied to quantify LUM expression in stroma or tumor. interactions between TGF-b and different host cell types are The software was then employed to quantify DAB intensity (a.u.) incompletely understood. in specific tissue classifiers (35, 36).

OF2 Clin Cancer Res; 2016 Clinical Cancer Research

Pancreatic Stellate Cells Regulate Lumican Secretion

For double color immunofluorescence (IF), HPSCs were seeded 8880, Bio-Rad). Relative expression levels were determined by into 8-well Lab-Tek II chamber slides (#154534, Thermo Fisher normalizing the expression level of each target to GAPDH, and DDC Scientific), fixed and permeabilized for 10 minutes the next day. relative mRNA fold changes were determined using the 2( t) The chamber slides were blocked with 1% BSA for 30 minutes, methods. and the mixture of anti-lumican and anti-SMAD4 antibodies (5 ng/mL) was added and incubated overnight at 4 C. TMA slides SMAD4 and LUM knockdown assays in HPSCs were deparaffinized and rehydrated, and then the paired mixtures Human retroviral short hairpin RNAi against human SMAD4 of lumican with a-SMA primary antibodies (5 ng/mL) were plasmid (shSMAD4) was purchased from Addgene (plasmid incubated overnight at 4 C. Secondary antibodies either conju- #15724). Three human lumican knockdown shLUM plasmids gated with Alexa Fluor 488 or with Alexa Fluor 594 of different (shLUM/KDs) were purchased from Sigma-Aldrich (TRC# species, depending on primary antibodies, which were purchased TRCN0000156246 labeled as shLUM/KD-1, TRCN0000154276 from Life Technologies (#A11008, A11012, and A11005). 2 ng/ml labeled as shLUM/KD-2, TRCN0000153890 as shLUM/KD-3). of these antibodies were incubated with the slides for 30 minutes Detailed products and iRNA sequences are available on the Sigma- at room temperature, and the slides were counterstained with Aldrich website. Its control noncoding shRNA plasmid was pur- DAPI (#D8417, Sigma-Aldrich) for 10 minutes, and mounted chased from MD Anderson Core. Procedures for transfection and with DAKO Fluorescent mounting medium (#S3023, Dako). As a infection into HPSCs were conducted as described previously negative control, HPSCs or TMA slides were applied with IgG. (38). The effectiveness of SMAD4 knockdown was validated by Images were captured with an Olympus FV1000 Laser Scanning Western blotting analysis. HPSC/shLUM/KD-1, -2, -3, and /CTL Confocal microscope and analyzed with FlowView software at the were transiently infected in HPSCs for 48 hours, and then collec- Flow Cytometry and Cellular Imaging Facility of MD Anderson tions of cell lysate and supernatant for the Western blot analysis Cancer Center. All images were captured with the same exposure and ELISA were pursued separately. Simultaneously, adhesion time on all samples. These profiles were then used to discriminate and migration assays were performed. The effectiveness of lumi- the individual colors on multi-stained slides (37). Three random can knockdown was validated by both Western blotting and images were chosen to estimate Pearson coefficient values (20 ELISA assays. magnification). Pearson correlation coefficient calculated the linear correlation between two variables X and Y. Electrophoretic mobility shift assay and LUM promoter luciferase reporter and mutagenesis assays Western blot analysis, ELISA, and PCR The LUM promoter fragment was examined for recognition by Cells were harvested and solubilized in RIPA buffer as described nuclear proteins by electrophoretic mobility shift assay (EMSA) previously (38). Whole-cell lysates (20 mg) were separated by according to a protocol described previously (38). Three oligo electrophoresis on 8%–12% SDS polyacrylamide gels, transferred probes containing SBE consensus–binding sites were designed to PVDF membranes (#10600023, GE Healthcare), and probed and are described in Supplementary Table S1. The super-shift with different dilutions of antibodies of interest. The antibodies experiments were performed with anti-SMAD4 antibody (#SC- used in this study, lumican (#ab98067, ab168348) and a-SMA 7966, Santa Cruz Biotechnology). The three SBEs and mutant (#ab5694) were purchased from Abcam. Phospho-Smad2 oligos were purchased from Sigma-Aldrich. (Ser465/467, #3101), phospho-Smad3 (Ser423/425, #9520), LUM promoter dual-luciferase activity was performed with a phospho-AKT (Ser473, #4060), phospho-MEK1/2 (Ser217/221, pGL2-LUM promoter construct, including part of the LUM first #9154), SMAD4 (#9515), Smad2/3 (#5678), and MEK1/2 exon and potential SBE sites (SBE1) as described previously (38). (#4694) were purchased from Cell Signaling Technology. Vincu- The pGL2-LUM plasmid was cotransfected into HPSCs and 293T lin (#ab18058) and TBP (#ab818) for protein loading control cells with an internal control, TK-Renilla luciferase. The activities were from Abcam. Reactive bands were visualized with enhanced of Firefly and Renilla luciferase were determined using a dual chemiluminescent reagents (GE Healthcare). For TGF-b stimula- luciferase reporter assay system (#E4550, Promega) after 48 hours tion and phosphorylation studies in HPSCs, fresh 5% FBS medi- of transfection. Firefly luciferase activity was normalized to the um was added and the cells were incubated with or without TGFb Renilla luciferase activity of the internal control. For site-directed (5 ng/mL) for 2, 8, or 24 hours. HPSC nuclear and cytosol lysates mutagenesis assays, one base pair in the first SBE sequence in the were separated by buffer A and buffer C. LUM promoter region from 811 bp to 817 bp was changed The ELISA was performed in fresh supernatant derived from using a mutagenesis kit (#200522, Agilent Technologies, Inc.). serum-free media of HPSCs, HPSC/shSMAD4, HPSC/LUM/KDs, and PDAC cells of 80% confluence in 10-cm dishes, washing Cell matrix adhesion and migration assays dishes with PBS twice, and 3-mL serum-free media added per dish Cell matrix adhesion and migration assays were performed on with or without treatments. Supernatant was collected at 8 and 24 collagen I–coated plates as described previously (39) (#A11428, hours, and ELISA was performed in 5 to 10X dilutions of the 96-well plate; #A11428, 6-well plate; Life Technologies). Briefly, supernatant (#MBS701117, MyBioSource). The optical density of HPSCs, HPSC/shSMAD4 cells, and other PDAC cells were seeded each well was read at 450 nm using a FluoStar Omega reader (5 104 cells) with serum-free medium in 96 wells where these (BMG Labtech, Inc.). cells were incubated with or without TGFb (10 ng/mL), rLUM (0.2 Total RNA was extracted using TRIzol reagent, and comple- mg/mL), supernatant collected from HPSCs, or serum-free media mentary DNA was prepared with the iScript Reverse Transcription (as control). HPSC/CTL and HPSC/shLUM/KDs were incubated Supermix Kit (#170-8841, Bio-Rad). The lumican primers for- in serum-free media after 48-hour infection. The cells were incu- 0 0 0 ward: 5 -CAGACTGCCTTCTGGTCTCC-3 , and reverse: 5 -AGCT- bated at 37 C for 24 hours to determine adhesion to collagen I. 0 CAACCAGGGATGACAC-3 . The expression level of human LUM Each plate was shaken at 1,000 rpm for 15 seconds and washed mRNA was quantified using iQ SYBR Green Supermix (#170- twice with PBS carefully to remove nonadherent cells. Afterward,

www.aacrjournals.org Clin Cancer Res; 2016 OF3

Kang et al.

fresh 10% FBS media and 20-mL MTT (5 mg/mL; #M5655, Sigma- tary Fig. S1B). Similar methods were applied to test colocalization Aldrich) were added into each well subsequently. The plates were of lumican and a-SMA, a described marker of PSC marker, in the incubated for another 3 hours, and the relative number of same collection of primary PDAC tumor specimens. Lumican attached cells was determined by the MTT method. (green) and a-SMA (red) expression levels were positively corre- Time-lapse migration analysis was performed on Olympus IX- lated (Pearson coefﬁcient, r ¼ 0.78) indicating that lumican 81 DSU Spinning Disk Confocal microscope (Olympus) system expression in stroma is positively associated with PSCs (P > equipped with the WeatherStation incubator (PrecisionControl) 0.0001; Fig. 1C). Using the Kaplan–Meier method, the median using a 10 objective. Migration was assessed using two-well survival of the stromal lumican-high group (DAB intensity culture inserts (#160255, Ibidi USA Inc.) placed into 6-well mean) was 16.05 months compared with 10.2 months for the collagen I–coated plates. Brieﬂy, these cells were seeded separately stromal lumican-low patient group (

OF4 Clin Cancer Res; 2016 Clinical Cancer Research

Pancreatic Stellate Cells Regulate Lumican Secretion

Figure 1. Lumican expression in 27 PDAC specimens in TMA. A, represented IHC images of primary tumors, 10 (top), 40 (bottom) magniﬁcation. B, VECTRA analysis calculates DAB intensity (a.u.) among tumor and stromal cells each case, mean values were plotted by GraphPad Prism. C, represented double immunoﬂuorescence staining of lumican (green) and a-SMA (red; left), and separate signals per case were recorded and analyzed by FluoView, values were plotted by GraphPad Prism (right). Two TMA cores images (10) per specimen were taken for analysis. Results are the mean SEM. D, using the Kaplan–Meier method, overall survival was compared between the patients with primary PDAC tumors with stromal lumican expression levels greater (stromal lumican high; n ¼ 12) or less than (stromal lumican low; n ¼ 11) the mean signal value.

www.aacrjournals.org Clin Cancer Res; 2016 OF5

Kang et al.

Figure 2. Lumican expression and secretion in pancreatic normal and tumor cell lines, and its secretion level associated with cell adhesion and migration in HPSC and PDAC cells. A, Western blot analysis shows lumican protein in pancreatic cell lines. B, RT-PCR results show lumican mRNA level in pancreatic cell lines. C, ELISA results present secreted lumican in HPSC, PANC-1, MDA-PATC53, and MiaPaCa-2 cells at 8 and 24 hours (absorbance readings were normalized by 1106 cells/mL). Lumican standard was performed accordingly. Absorbance reading values were plotted by GraphPad Prism. Y, normalized absorbance readings. Adhesion (D)and(E) migration assays were performed onto collagen I–coated plates in HPSC, PANC-1, MDA-PATC53, and MiaPaCa2 cells. Results are the mean SEM of three independent experiments. , P < 0.05; , P < 0.001; , P < 0.0001.

TGFb inhibits lumican expression and secretion through hours compared with non-TGFb–exposed cells (P < 0.05 and canonical signaling pathway in PSCs 0.0001, respectively; Fig. 3B). Lumican (green) and SMAD4 (red) TGFb is a potent regulator of cell–matrix and cell–cell adhesion IF staining after TGFb exposure confirmed SMAD4 accumulation (41), and TGFb cytokines originate from various cell types within in the nucleus after 2 hours, and a reduction of cytoplasmic PDAC (42). We therefore sought to understand the relationship lumican after 8 hours (Fig. 3c). Similar TGFb stimulation assays between TGFb and lumican expression in HPSCs. We first con- were performed in HPaSteCs (Supplementary Materials and firmed that HPSCs secrete all three isoforms of TGFb (TGFb1, Methods); lumican secretion was inhibited significantly when TGFb2, and TGFb3), and that TGFb1 is dominant (Supplemen- HPaSteCs were exposed to TGFb for 8 and 24 hours (P > 0.05 or tary Fig. S1C and Materials and Methods). When HPSCs were P > 0.0001 comparing with no TGFb exposure, separately; Sup- exposed to exogenous TGFb (5 ng/mL) in vitro, the TGFb pathway plementary Fig. S2A), and Western blotting confirmed lumican was initiated and lumican levels within the cytosol were pro- expression was reduced and correlated with TGFb signaling path- foundly reduced after 24 hours, as demonstrated by immuno- way activation after exposing to TGFb for 24 hours (Supplemen- blotting (Fig. 3A). Secreted lumican was also reduced at 8 and 24 tary Fig. S2B).

OF6 Clin Cancer Res; 2016 Clinical Cancer Research

Pancreatic Stellate Cells Regulate Lumican Secretion

Figure 3. Lumican expression and secretion reduction while TGFb pathway was activated in HPSC cells. A, Western blot analysis demonstrated expression levels of lumican, SMAD4, phosphor-Smad2, phosphor-Smad3 in nucleus and cytosol of HPSC cells after exposing with or without TGFb (5 ng/mL) at 2, 8, and 24 hours. Smad2/3 as phosphor-Smad2, or -Smad3 controls. TBP and vinculin were used as the nucleus and cytosol loading control, separately. B, secreted lumican level was measured by ELISA after TGFb treatment (5 ng/mL) or none in 8 and 24 hours in HPSC cells supernatants, lumican readings at 450 nm were normalized by cell number (1 106 cells), bar graph was plotted by GraphPad Prism; , P < 0.05; , P < 0.0001. C, double immunoﬂuorescence (IF) staining of lumican (green) and SMAD4 (red) in HPSC cells after exposing to TGFb at 0, 2, 8, and 24 hours. Nuclei were counterstained with blue DAPI. Images were captured and merged by using an Olympus FV1000 Laser Scanning Confocal microscope and analyzed with FlowView software.

To verify that secreted lumican reduction was a consequence of compared with the combination of TGFb and SB431542 (P < TGFb pathway signaling, HPSCs were exposed to TGFb with or 0.0001; Fig. 4C). After SMAD4 was silenced in HPSCs (HPSC/ without the TGFb inhibitor, SB431542 (#ab120163, Abcam) for 2 shSMAD4; Fig. 4D), secreted lumican increased significantly at 8 hours (43). In HPSCs exposed to TGFb, the canonical pathway and 24 hours when compared with control HPSC (P < 0.0001; Fig. was activated, and cytoplasmic lumican expression reduced. 4E). Together, these results suggest that canonical TGF-b–SMAD4 However, in the presence of the TGFb inhibitor, this pathway is a crucial signaling pathway to negatively regulate lumican was inactive and cytoplasmic lumican remained at basal levels. expression and secretion within PSCs. TGFb noncanonical pathway signaling proteins, such as phospho- AKT and phospho-p44/42, were also not initiated in HPSCs (Fig. SMAD4 binds to SBEs located within the promoter region of 4A). Lumican (green) and SMAD4 (red) IF staining confirmed lumican gene (LUM) these findings (Fig. 4B). ELISA was performed with the same On the basis of our preliminary observations (Figs. 3 and 4), we treatment groups for 24 hours, and demonstrated that secreted identified three candidate SBE sequences located within the LUM lumican was significantly reduced after TGFb exposure (P < 0.05) promoter (-811, -3526, and -4303 bp). We hypothesized that

www.aacrjournals.org Clin Cancer Res; 2016 OF7

Kang et al.

Figure 4. Lumican expression and secretion modification after TGF-b pathway blocked or SMAD4 knocked down in HPSC cells. A, Western blot analysis showed lumican, SMAD4, phosphor-Smad2, phosphor-Smad3, phosphor-AKT, and phosphor-p44/42 in nucleus and cytosol of HPSC cells after exposing with or without TGFb (5 ng/mL) and SB431542 (10 mmol/L) for 2-hour treatment. Smad2/3 as phosphor-Smad2, or –Smad3 controls, AKT and p44/42 as phosphor-AKT and –p44/42 controls. TBP and vinculin were used as the nucleus and cytosol loading control, separately. B, double immunofluorescence (IF) staining of lumican (green) and SMAD4 (red) in HPSC cells after treatment with TGFb, SB431542, or combination at 2-hour exposure. Nuclei were counterstained with blue DAPI. Images were captured and merged by an Olympus FV1000 Laser Scanning Confocal microscope and analyzed with FlowView. C, secreted lumican level was measured by ELISA after exposing to TGFb (5 ng/mL), SB431542 (10 mmol/L), or combination or none in 24 hours in HPSC cells supernatants. D, efficacy of SMAD4 knockdown in HPSC cells was confirmed by Western blot analysis. E, comparison of secreted lumican level was measured by ELISA in HPSC and HPSC/shSMAD4 cells supernatants at 8 and 24 hours. Bar graphs were plotted by GraphPad Prism. , P < 0.05; , P < 0.0001.

OF8 Clin Cancer Res; 2016 Clinical Cancer Research

Pancreatic Stellate Cells Regulate Lumican Secretion

Figure 5. Map of multiple SBEs in LUM promoter. A, three SBEs with a CAGACA sequence (blue circles 1, 2, and 3), correspondently to SBEs. B, EMSA showed that SBE1 oligos had strong DNA and nuclear protein interaction bands (lane 1), binding was quenched by wild-type (WT) oligos (lane 2) and not by mutant (M) oligos (lane 3), and anti-SMAD4 antibody (Ab) was supershifted binding activity (lane 4). C, EMSA shows SBE1 binding activities were regulated by exposing to TGFb at 2, 8, and 24 hours, but not on SBE3. OCT-1 DNA–binding activities were determined as loading controls. D, the SBE1 sequence in the LUM promoter (top) and the Mut1 mutant sequence (G!C; bottom). E, LUM pGL2-SBE1 promoter dual-luciferase activity construct of potential SBE1 site (double arrow on 1a) was stimulated by TGFb (5 ng/mL), the ratios of pGL2-LUM activities were presented at 0, 2, 8, and 24 hours in HPSC and 293T cells. F, LUM pGL2-MUT1 luciferase activity ratios of Mut1 with or without TGFb (5ng/mL) at 0, 2, 8, and 24 hours in HPSC and 293T cells. Results are the mean SEM of three independent experiments. , P < 0.01; , P < 0.0001. these sites cooperate to regulate LUM transcription (Fig. 5A). strated that SBE1 oligos have speciﬁc binding to HPSC nuclear Binding activities of SMAD4 to these sites was determined by protein (Lane 1). This binding was only blocked by wild-type EMSA and LUM dual-luciferase reporter assays. EMSA demon- oligos (Lane 2) and not by mutant oligos (Lane 3), and this was

www.aacrjournals.org Clin Cancer Res; 2016 OF9

Kang et al.

super-shifted by an anti-SMAD4 antibody (Lane 4, arrow showed tary Fig. S3B. These data demonstrate that lumican augments shift band; Fig. 5B). When HPSCs were exposed to TGFb, SBE1 HPSC adhesion and migration in a collagen I–rich environment. binding was temporarily decreased at 2 hours, then recovered Intriguingly, supernatant from HPSCs can also increase the binding at 8 and 24 hours, although not with other SBEs adherence of low-secreting lumican PDAC cell lines, such as (Fig. 5C). So, a construct including SBE1 was then cloned into PANC-1 and MiaPaCa2 cells, to collagen I (P < 0.0001; Supple- pGL2-Basic vectors (Promega; pGL2-SBE1) to validate the mentary Fig. S4A). Conversely, HPSC supernatant depleted of functional activity of specific binding within the LUM promoter lumican by anti-lumican antibody suppresses PANC-1 adhesion (Fig. 5A). HPSC and 293T cells were then transiently transfected (P < 0.01; Supplementary Fig. S4B), whereas supernatant from with pGL2-SBE1 and exposed to TGFb; luciferase activity HPSC/shSMAD4 enhances PANC-1 adhesion (P < 0.01; Supple- was increased 2 hours after TGFb stimulation suggesting that mentary Fig. S4C). Panc-1 cell adhesion was reduced (P < 0.01) SBE1 cooperates to enhance LUM activations in both HPSC and after exposure to supernatant obtained from HPSCs after exposure 293T cells. Luciferase activity was maintained at both 8 and 24 to TGFb and reduction in lumican production (Supplementary hours in HPSC (P < 0.01 and P < 0.0001, respectively), and Fig. S4D). We cocultured HPSCs and PANC-1 (where HPSCs were declined significantly after 8 hours in 293T cells (Fig. 5D). To labeled with GFP and PANC-1 labeled with RFP), and serial prove the importance of the SBE1 on construct pGL2-SBE1, we images were analyzed at several time points. The results showed specifically mutated the construct at -811 from CAGACA to that HPSCs migrated faster and circulated near PANC-1 cells (red CACACA (Fig. 5E). After this mutagenesis, no luciferase activity arrows; Supplementary Fig. S4E). Together, these results suggest in response to TGFb demonstrated that binding at SBE1 is that HPSC-secreted lumican influences the adhesion and motility necessary for LUM transcriptional control (Fig. 5F). Together, of both tumor and stromal cells in a collagen I–rich environment. these assays confirmed that exposure to TGFb induces SMAD4 fi translocation and binding to a novel, speci c, and functionally Discussion active SBE within the LUM promoter. In this study, we found that activated PSCs derived from human Lumican regulation affects HPSC adhesion and primary PDAC tumors secrete large amounts of lumican; a finding migration in vitro which correlates with the colocalization of extracellular lumican As previously experiments demonstrated that higher lumican- with a-SMA–expressing PSC, and fibrillar collagen, in primary secreting cells, HPSCs and MDA-PATC53, achieved increased tumors. We demonstrated that TGFb uses its canonical pathway to adhesion and enhanced migration, versus the lower lumican- negatively regulate lumican production in HPSCs; this occurs secreting cells, PANC-1 and MiaPaCa2, and correlation between through by SMAD4 translocation and binding to a SBE lumican and collagen I was also confirmed in these patients' (CAGACA) within the promoter region of the lumican gene. specimen lumican/collagen I double IF staining which showed Interestingly, we found that secreted lumican in the extracellular Pearson r ¼ 0.47 (P > 0.05; Supplementary Fig. S1B). More assays environment preferentially enhances cell adhesion and migration were performed to further confirm how lumican regulation was of HPSCs when on collagen. Together, these data provide a clearer specifically involved in these adhesion and migration phenomena understanding of the control mechanisms and consequences of in vitro. An anti-lumican antibody was used to deplete lumican in extracellular lumican within PDAC tumors. HPSC supernatant for 24 hours, which resulted in less secreted Lumican is usually localized to areas of pathologic fibrosis and lumican on ELISA compared with HPSCs treated with an IgG1 has been demonstrated in the ECM within various human malig- control antibody (P < 0.001; Fig. 6A). Correspondingly, HPSC nancies including pancreatic cancer (8–10, 44, 45). In pancreatic adhesion was decreased in the antibody-treated group compared cancer, lumican mRNA is highly expressed in acinar cells, islet with serum-free or IgG1-treated groups (P < 0.05; Fig. 6B). In cells, and proliferating fibroblasts within the desmoplastic stroma HPSCs with silenced SMAD4 (HPSC/shSMAD4), adhesion to surrounding cancer cells and activated stellate cells have been collagen I was increased (Fig. 6C). Migration assays demonstrated implicated as the cell source (9). Similar studies have demon- increased motility in HPSC exposed to rLUM (P < 0.0001) as well strated that PSC are the cell source of fibrillar collagen within the as in HPSC/shSMAD4 (P < 0.05), but not in TGFb treatment desmoplastic ECM of PDAC tumors (20). The colocalization of group, which indicates TGFb likely impacts HPSC motility. These lumican with desmoplasia is logical given our understanding results clearly demonstrate that external or secreted lumican that lumican binds collagen fibrils and regulates interfibrillar promote HPSC motility in vitro (Fig. 6D). More specific assays spacing during collagen assembly (17, 18, 46). TGFb is one of to validate the correlation of lumican expression or secretion with the strongest inducers of ECM production during fibrogenesis HPSC's motility were performed. While silencing lumican in (24, 25), and largely drives the desmoplastic stroma observed in HPSC by three lumican shRNA plasmids, HPSC transiently was pancreatic carcinoma (26). TGFb mediates type I collagen gene infected by the three plasmids (shLUM/KD-1, shLUM/KD-2, and expression through a synergistic cooperation of the transcription shLUM/KD-3). Lumican expression was effectively reduced in the factors Smad2/Smad3 and Sp1 acting on the promoter (24, 47). knockdown clones compared to control stellate cells as measured To date, however, no studies have described the mechanism by by Western blotting (Fig. 6E) and ELISA (Fig. 6F); adhesion (P < which TGFb could control lumican production by stellate cells 0.01 or P < 0.001; Fig. 6G) and migration (P < 0.001 or P < 0.01; activated within PDAC tumors. Fig. 6H) of the knockdown clones was significantly reduced In the described experiments, we demonstrate that TGFb acts compared to control cells. Representative images of migration in through its canonical signaling pathway to negatively control HPSCs, HPSCs treated with TGFb, HPSCs treated with rLUM, and lumican transcription in PSCs through novel SMAD4-SBE bind- HPSC/shSMAD4 at 6 and 12 hours are shown in Supplementary ing. In addition, we demonstrate how extracellular lumican Fig. S3A, and representative images in HPSC/CTL and three influences the function of PSCs within a collagen-rich environ- HPSC/shLUM/KDs at 6 and 12 hours are shown in Supplemen- ment. ELISA of media conditioned by PSCs (HPSCs and HPaSteC)

OF10 Clin Cancer Res; 2016 Clinical Cancer Research

Pancreatic Stellate Cells Regulate Lumican Secretion

Figure 6. Lumican secretion associated with cell matrix adhesion and migration in collagen I–coated plates. A, ELISA shows secreted lumican was reduced after HPSC cells supernatant were treated with speciﬁc lumican antibody. B, secreted lumican reduction correlated with decreased HPSC adhesion. C, cell matrix adhesion in HPSC and HPSC/shSMAD4 cells. D, time lapse of migration assays were performed on according of HPSC, HPSC plus TGFb, HPSC plus rLUM, and HPSC/shSMAD4 conditions in collagen I–coated plates, separately. E, Western blotting showed effectiveness of lumican knockdown in HPSC/CTL and HPSC/shLUM/KDs. F, ELISA showed secreted lumican reduction in three HPSC/shLUM/KDs' clones comparing with HPSC/CTL (P < 0.01 and P < 0.001). G, adhesion assays showed reduction in these HPSC/shLUM/KDs clones compared with HPSC/CTL (P < 0.01 and P < 0.001). H, time lapse of migration assays showed migration delayed in these HPSC/shLUM/KDs clones compared with HPSC/CTL (P < 0.001 and P < 0.01). Bar graphs were plotted by GraphPad Prism. Results are the mean SEM. , P < 0.05; , P < 0.01; , P < 0.0001.

www.aacrjournals.org Clin Cancer Res; 2016 OF11

Kang et al.

and other cell types within PDAC allowed us to demonstrate that Disclosure of Potential Conflicts of Interest PSCs, activated stellate cells harvested from primary PDAC, E.J. Koay reports receiving a commercial research grant and speakers bureau secrete large amounts of lumican into the extracellular space. honoraria from Philips. No potential conflicts of interest were disclosed by the Manipulation of the TGFb canonical signaling pathway allowed other authors. us to determine that negative regulation at the transcriptional level is the primary means by which the cytokine TGFb exerts Authors' Contributions control over lumican secretion from PSCs. These observations Conception and design: J.B. Fleming, Y. Kang, J. Ling, X. Li, H. Wang Development of methodology: J.B. Fleming, Y. Kang, J. Ling, M.J. Truty, expand and provide needed detail to previous reports (9). E.J. Koay Several studies have demonstrated that SLRP proteins, specif- Acquisition of data (provided animals, acquired and managed patients, ically decorin, can modulate cellular behavior, including cell provided facilities, etc.): J.B. Fleming, D. Roife, Y. Lee, R. Suzuki, M.V.R. Perez, migration and proliferation during embryonic development, M. Pratt, M.J. Truty, H. Wang, R.M. Thomas tissue repair, and tumor growth, in addition to their extracellular Analysis and interpretation of data (e.g., statistical analysis, biostatistics, matrix function as regulators of tissue hydration and collagen computational analysis): J.B. Fleming, Y. Kang, D. Roife, Y. Lee, H. Lv, fi – R. Suzuki, M.V.R. Perez, M.J. Truty, D. Chatterjee, H. Wang, Y. Wang, M.H. Katz brillogenesis (48 50). We previously found that the presence of Writing, review, and/or revision of the manuscript: J.B. Fleming, Y. Kang, lumican in the ECM surrounding was associated with improved D. Roife, Y. Lee, R. Suzuki, M. Pratt, M.J. Truty, H. Wang, R.M. Thomas, E.J. Koay, patient outcome after multimodal therapy and surgical removal M.H. Katz of localized PDAC (39) suggesting that lumican influences cells Administrative, technical, or material support (i.e., reporting or organizing within the tumor microenvironment. Our laboratory and others data, constructing databases): J.B. Fleming, Y. Kang, R. Suzukim, B. Dai, have demonstrated that extracellular lumican impacts the activity P.J. Chiao Study supervision: J.B. Fleming, Y. Kang and survival of pancreatic cancer cells (39, 40), but the effects of extracellular lumican on host cells is poorly characterized. Our Acknowledgments observations show that high levels of lumican secretion augment The authors thank Skip Viragh Family Foundation and Various Donors adhesion of HPSC and PDAC cancer cells to collagen I, and this in Pancreatic Cancer Research. effect is abrogated by anti-lumican antibodies and genetic silencing of lumican in HPSC. This suggests that extracellular lumican, Grant Support regardless of the cell source, effects cell adhesion of multiple cell The study was supported by Cancer Center Support Grant (CCSG) core types in a collagen-rich environment. Interestingly, lumican also resources of Flow Cytometry and Cellular Imaging Facility under NIH/NCI increased cell migration when on collagen I, but the migration of award (P30CA016672). Sequencing and Microarray Facility is funded by NCI #CA016672. CCSG for the STR DNA fingerprinting and the Characterized Cell HPSC was increased to greater degree than cancer cells. The Line cores were funded by NCI grant (CA016672). D. Roife and M.V.R. Perez molecular mechanism of these phenotypic differences remains were supported in part by NIH grant T32CA009599. This work was also to be elucidated, but the observations serve as an example of the supported by grants from the Skip Viragh Family Foundation and Various differential influence of lumican on the cellular compartment. Donors in Pancreatic Cancer Research (to J.B. Fleming). These findings offer a deeper understanding of the biologic The costs of publication of this article were defrayed in part by the payment of advertisement function of lumican within the PDAC tumor microenvironment, page charges. This article must therefore be hereby marked in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. and provide a fundamental platform upon which to discover other predictive markers or to design novel antitumor strategies in Received November 13, 2015; revised March 24, 2016; accepted April 7, 2016; PDAC patients. published OnlineFirst April 28, 2016.

References 1. Howlader NNA, Krapcho M, Garshell J, Miller D, Altekruse SF, Kosary CL, 10.MatsudaY,YamamotoT,KudoM,KawaharaK,KawamotoM, et al. SEER Cancer Statistics Review, 1975–2011, National Cancer Institute. Nakajima Y, et al. Expression and roles of lumican in lung ade- Bethesda, MD;2014. nocarcinoma and squamous cell carcinoma. Int J Oncol 2008;33: 2. Siegel R, Ma J, Zou Z, Jemal A. Cancer statistics, 2014. CA Cancer J Clin 1177–85. 2014;64:9–29. 11. Greiling H.Structure and biological functions of keratan sulfate proteogly- 3. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell cans. EXS 1994;70:101–22. 2011;144:646–74. 12. Cornuet PK, Blochberger TC, Hassell JR. Molecular polymorphism of 4. Bhowmick NA, Neilson EG, Moses HL. Stromal fibroblasts in cancer lumican during corneal development. Invest Ophthalmol Vis Sci 1994; initiation and progression. Nature 2004;432:332–7. 35:870–7. 5. Lu P, Weaver VM, Werb Z. The extracellular matrix: a dynamic niche in 13. Stamov DR, Muller A, Wegrowski Y, Brezillon S, Franz CM. Quantitative cancer progression. J Cell Biol 2012;196:395–406. analysis of type I collagen fibril regulation by lumican and decorin using 6. Nikitovic D, Papoutsidakis A, Karamanos NK, Tzanakakis GN. Lumican AFM. J Struct Biol 2013;183:394–403. affects tumor cell functions, tumor-ECM interactions, angiogenesis and 14. Qi J, Zeng J, Zhao F, Lin SH, Raja B, Strych U, et al. Label-free, in situ SERS inflammatory response. Matrix Biol 2014;35:206–14. monitoring of individual DNA hybridization in microfluidics. Nanoscale 7. Feig C, Gopinathan A, Neesse A, Chan DS, Cook N, Tuveson DA. 2014;6:8521–6. The pancreas cancer microenvironment. Clin Cancer Res 2012;18: 15. Svensson L, Narlid I, Oldberg A. Fibromodulin and lumican bind to the 4266–76. same region on collagen type I fibrils. FEBS Lett 2000;470:178–82. 8. Naito Z. Role of the small leucine-rich proteoglycan (SLRP) family in 16. Kalamajski S, Oldberg A. Homologous sequence in lumican and fibro- pathological lesions and cancer cell growth. J Nippon Med Sch 2005;72: modulin leucine-rich repeat 5–7 competes for collagen binding. J Biol 137–45. Chem 2009;284:534–9. 9. Ping Lu Y, Ishiwata T, Asano G. Lumican expression in alpha cells 17. Grover J, Chen XN, Korenberg JR, Roughley PJ. The human lumican gene. of islets in pancreas and pancreatic cancer cells. J Pathol 2002;196: Organization, chromosomal location, and expression in articular cartilage. 324–30. J Biol Chem 1995;270:21942–9.

OF12 Clin Cancer Res; 2016 Clinical Cancer Research

Pancreatic Stellate Cells Regulate Lumican Secretion

18. Chakravarti S, Stallings RL, SundarRaj N, Cornuet PK, Hassell JR. Primary 34. Kim MP, Evans DB, Wang H, Abbruzzese JL, Fleming JB, Gallick GE. structure of human lumican (keratan sulfate proteoglycan) and localiza- Generation of orthotopic and heterotopic human pancreatic cancer xeno- tion of the gene (LUM) to chromosome 12q21.3-q22. Genomics grafts in immunodeficient mice. Nat Protoc 2009;4:1670–80. 1995;27:481–8. 35. Mansfi eld JR, Hoyt C, Levenson RM. Visualization of microscopy-based 19. Li X, Truty MA, Kang Y, Chopin-Laly X, Zhang R, Roife D, et al. spectral imaging data from multi-label tissue sections. Curr Protoc Mol Biol Extracellular lumican inhibits pancreatic cancer cell growth and is 2008;14:9. associated with prolonged survival after surgery. Clin Cancer Res 2014; 36. Levenson RM, Fornari A, Loda M. Multispectral imaging and pathology: 20:6529–40. seeing and doing more. Expert Opin Med Diagn 2008;2:1067–81. 20. Apte MV, Park S, Phillips PA, Santucci N, Goldstein D, Kumar RK, et al. 37. Webb JR, Milne K, Watson P, Deleeuw RJ, Nelson BH. Tumor-infiltrating Desmoplastic reaction in pancreatic cancer: role of pancreatic stellate cells. lymphocytes expressing the tissue resident memory marker CD103 are Pancreas 2004;29:179–87. associated with increased survival in high-grade serous ovarian cancer. Clin 21. Rhim AD, Oberstein PE, Thomas DH, Mirek ET, Palermo CF, Sastra SA, Cancer Res 2014;20:434–44. et al. Stromal elements act to restrain, rather than support, pancreatic ductal 38. Kang Y, Ling J, Suzuki R, Roife D, Chopin-Laly X, Truty MJ, et al. SMAD4 adenocarcinoma. Cancer Cell 2014;25:735–47. regulates cell motility through transcription of N-cadherin in human 22. Ozdemir BC, Pentcheva-Hoang T, Carstens JL, Zheng X, Wu CC, Simpson pancreatic ductal epithelium. PLoS ONE 2014;9:e107948. TR, et al. Depletion of carcinoma-associated fibroblasts and fibrosis 39. Sawhney RS, Cookson MM, Omar Y, Hauser J, Brattain MG. Integrin induces immunosuppression and accelerates pancreas cancer with reduced alpha2-mediated ERK and calpain activation play a critical role in cell survival. Cancer Cell 2014;25:719–34. adhesion and motility via focal adhesion kinase signaling: identification of 23. Gore J, Korc M. Pancreatic cancer stroma: friend or foe? Cancer Cell 2014; a novel signaling pathway. J Biol Chem 2006;281:8497–510. 25:711–2. 40. Alberts B, Johnson A, Lewis J, Raff M, Roberts K, Walter P. In: Cell 24. Leask A, Abraham DJ. TGF-beta signaling and the fibrotic response. FASEB J junctions, cell adhesion, and the extracellular matrix. Molecular biology 2004;18:816–27. of the cell. 4th edn. New York, NY: Garland Science; 2002, 44. (hardback). 25. Nakerakanti S, Trojanowska M. The Role of TGF-beta Receptors in Fibrosis. 1616 pp. PMCID: PMC4244961. doi: 10.1093/aob/mcg023. Open Rheumatol J 2012;6:156–62. 41. Wang H, Radjendirane V, Wary KK, Chakrabarty S. Transforming growth 26. Lohr M, Schmidt C, Ringel J, Kluth M, Muller P, Nizze H, et al. factor beta regulates cell-cell adhesion through extracellular matrix remo- Transforming growth factor-beta1 induces desmoplasia in an experi- deling and activation of focal adhesion kinase in human colon carcinoma mental model of human pancreatic carcinoma. Cancer Res 2001; Moser cells. Oncogene 2004;23:5558–61. 61:550–5. 42. Truty MJ, Urrutia R. Basics of TGF-beta and pancreatic cancer. Pancreatol- 27. Jonk LJ, Itoh S, Heldin CH, ten Dijke P, Kruijer W. Identification and ogy 2007;7:423–35. functional characterization of a Smad binding element (SBE) in the JunB 43. Inman GJ, Nicolas FJ, Callahan JF, Harling JD, Gaster LM, Reith AD, et al. promoter that acts as a transforming growth factor-beta, activin, and SB-431542 is a potent and specific inhibitor of transforming growth factor- bone morphogenetic protein-inducible enhancer. J Biol Chem 1998;273: beta superfamily type I activin receptor-like kinase (ALK) receptors ALK4, 21145–52. ALK5, and ALK7. Mol Pharmacol 2002;62:65–74. 28. Shi Y, Wang YF, Jayaraman L, Yang H, Massague J, Pavletich NP. Crystal 44. Qin H, Ishiwata T, Asano G. Effects of the extracellular matrix on lumican structure of a Smad MH1 domain bound to DNA: insights on DNA binding expression in rat aortic smooth muscle cells invitro. J Pathol 2001;195:604–8. in TGF-beta signaling. Cell 1998;94:585–94. 45.NikitovicD,BerdiakiA,Zafiropoulos A, Katonis P, Tsatsakis A, Kar- 29. Massague J, Wotton D. Transcriptional control by the TGF-beta/Smad amanos NK, et al. Lumican expression is positively correlated with the signaling system. EMBO J 2000;19:1745–54. differentiation and negatively with the growth of human osteosarcoma 30.YangL,HuangJ,RenX,GorskaAE,ChytilA,AakreM,etal. cells. FEBS J 2008;275:350–61. Abrogation of TGF beta signaling in mammary carcinomas recruits 46. Iozzo RV.Matrix proteoglycans: from molecular design to cellular function. Gr-1þCD11bþ myeloid cells that promote metastasis. Cancer Cell Annu Rev Biochem 1998;67:609–52. 2008;13:23–35. 47. Cutroneo KR.How is Type I procollagen synthesis regulated at the gene 31. Koninger J, Giese T, di Mola FF, Wente MN, Esposito I, Bachem MG, et al. level during tissue fibrosis. J Cell Biochem 2003;90:1–5. Pancreatic tumor cells influence the composition of the extracellular 48. Iozzo RV.The family of the small leucine-rich proteoglycans: key regulators matrix. Biochem Biophys Res Commun 2004;322:943–9. of matrix assembly and cellular growth. Crit Rev Biochem Mol Biol 32. Hwang RF, Moore T, Arumugam T, Ramachandran V, Amos KD, Rivera A, 1997;32:141–74. et al. Cancer-associated stromal fibroblasts promote pancreatic tumor 49. Fullwood NJ, Davies Y, Nieduszynski IA, Marcyniuk B, Ridgway AE, progression. Cancer Res 2008;68:918–26. Quantock AJ. Cell surface-associated keratan sulfate on normal and migrat- 33. Kang Y, Zhang R, Suzuki R, Li SQ, Roife D, Truty MJ, et al. Two-dimensional ing corneal endothelium. Invest Ophthalmol Vis Sci 1996;37:1256–70. culture of human pancreatic adenocarcinoma cells results in an irreversible 50.WightTN,KinsellaMG,QwarnstromEE.Theroleofproteoglycans transition from epithelial to mesenchymal phenotype. Lab Invest 2015; in cell adhesion, migration and proliferation. Curr Opin Cell Biol 95:207–22. 1992;4:793–801.

www.aacrjournals.org Clin Cancer Res; 2016 OF13

Transforming Growth Factor-β Limits Secretion of Lumican by Activated Stellate Cells within Primary Pancreatic Adenocarcinoma Tumors

Ya'an Kang, David Roife, Yeonju Lee, et al.

Clin Cancer Res Published OnlineFirst April 28, 2016.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-15-2780

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2016/05/12/1078-0432.CCR-15-2780.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2016/08/12/1078-0432.CCR-15-2780. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.