CYR61/CCN1 Regulates dCK and CTGF and Causes Gemcitabine-resistant Phenotype in Pancreatic Ductal Adenocarcinoma

Pancreatic ductal adenocarcinoma (PDAC) is a common form of pancreatic cancer and now the third leading cause of cancer-related death in the United States, and an anticipated second leading cause of cancer-related death by 2030 (1–3). The prognosis of PDAC remains depressing and a difficult malignancy to treat (4). Although the location of the pancreas leads to late diagnosis and one of the vital weaknesses in treatment and unresectability, a major contributor to the poor clinical outcome is chemoresistance.

The treatment of choice for PDAC in the early stage is surgery followed by adjuvant therapy. Regrettably, the surgery is not always a choice as most patients present with locally advanced unresectable disease or their disease has already metastasized to the distant organs, leaving them with a poor prognosis (5, 6). Thus, gemcitabine, a nucleoside analogue, has been widely used as a first-line chemotherapeutic drug for PDAC with advanced stages. It can be used alone or in combination with other agents (7). It is reported that PDAC cells are comparatively more sensitive to gemcitabine than other anticancer drugs (5). However, the majority of the patients with PDAC eventually develop resistance to gemcitabine with poorly known mechanisms (5). Thus, understanding the mechanism of resistance of gemcitabine by PDAC cells is urgently needed.

Gemcitabine resistance in pancreatic cancer cells is a complex and multifactorial process (8, 9). Gemcitabine activation in cancer cells needs a rate-limiting transition from the inactive prodrug to an active compound through a series of phosphorylations by a rate-limiting enzyme deoxycytidine kinase (dCK; refs. 10, 11). The epigenetic silencing or inactivating mutations of dCK play a crucial role in gemcitabine resistance (12, 13). However, how dCK is epigenetically inactivated in PDAC cells is not well understood.

In addition, desmoplasia, a dynamic process regulated by the multiple signaling cross-talk of tumor cells and surrounding stroma, plays a critical role in chemoresistance (5, 14–19). Desmoplasia leads to a significant increase in the production of dense and fibrous connective tissues, extracellular matrix proteins, stroma with high interstitial pressure around the tumors, and proliferation of myofibroblast cells/active pancreatic stellate cells (PSC) to maintain the tumor growth, metastasis, and a barrier to chemotherapy penetration (16). Multiple studies have shown that CTGF, a cysteine-rich, matricellular protein, is involved in gemcitabine resistance in many cancer cells including PDAC (20). The PDAC cells' secreted CTGF activates stellate cells to induce the desmoplastic reaction. However, no molecule has yet been found that could regulate CTGF levels in PDAC cells.

We and others have previously shown that CYR61/CCN1, which belongs to the CCN family of growth factors, acts as an oncoprotein in PDAC (21–25). CYR61/CCN1 promotes growth and invasive phenotypes of PDAC cells via integrin-dependent noncanonical pathways. Moreover, our studies also showed that CYR61/CCN1 reprograms the epithelial–mesenchymal transition (EMT) and maintains cancer stemness/tumor-initiating properties that are associated with metastasis and drug resistance (18, 21, 22, 26–30). This provokes the hypothesis that PDAC cell–secreted CYR61/CCN1 signaling contributes to gemcitabine resistance. In this study, we validated the premise and demonstrated that CYR61/CCN1–depleted PDAC cells are sensitive to gemcitabine. Furthermore, we show that CYR61/CCN1 can regulate dCK and CTGF in PDAC cells and stellate cells, and concurrently activate stellate cells. On the basis of these previously unrecognized mechanisms we suggest that CYR61/CCN1 promotes gemcitabine resistance through the regulation of dCK and CTGF, and thus, CYR61/CCN1 signaling may represent a new target for sensitization of gemcitabine in PDAC.

Animal protocols were approved by the Kansas City VA Medical Center Animal Care and Use Committee, in accordance with the Association for Assessment and Accreditation of Laboratory Animal Care International animal care guidelines, and NIH Guide for the Care and Use of Laboratory Animals. Male and female athymic nude mice (nu/nu genotype), 6–8 weeks old, were purchased from Jackson Laboratories and acclimatized in our facility for 1 week.

Gemcitabine (Eli Lilly, and Company) was purchased through the VA pharmacy. Crystal Violet stain was purchased from Sigma (Millipore-Sigma). Human polyclonal anti-rabbit CYR61, CTGF, and dCK antibodies and polyclonal goat anti-rabbit IgG-HRP and monoclonal goat anti-mouse IgG-HRP were purchased from Santa Cruz Biotechnology and Thermo Fisher Scientific, respectively. CYR61 human recombinant protein (hrCYR61) was purchased from PeproTech. All other chemicals were purchased from either Sigma or Thermo Fisher Scientific. Authentication of the antibodies, specifically, CTGF and CYR61, which are closely related and belong to the same family, were confirmed, and found no cross-reactivity of the antibodies (Supplementary Fig. S1).

Pancreatic cancer cell lines BxPC-3, Capan-1, AsPC-1, and Panc-1 were purchased from ATCC. The cell lines were maintained as described previously (22). Human pancreatic stellate cells (HPaSteC) was purchased from ScienCell (ScienCell Research Laboratory) and maintained in stellate cell media and FBS. FEF3 cells, a fetal human esophageal fibroblast, were a gift and maintained in high-glucose DMEM with FBS. Authentication of the cell lines was performed by determining short tandem repeat (STR) profiles using the Promega PowerPlex 16 system. This was performed once and compared with external STR profiles of the cell lines (when available). Cell lines were also figured out to be Mycoplasma-free prior to use.

The Connectivity Map represents the network of functional connection of the drug-resistant genes with CYR61/CCN1 in pancreatic cancer patient samples. The network was designed in IPA (QIAGEN Inc., https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis), using its knowledge base of known, experimentally validated, molecular interactions. The network reflects the potential pathways by which CYR61 interacts with drug-resistant genes.

The CYR61-knockout stable Panc-1cell line was generated using our previous method (21, 22).

Human CYR61 gene knockout AsPC-1 cells were generated using CYR61/CCN1 Human Gene Knockout Kit (CRISPR; OriGene). Three separate transfections were carried out in AsPC-1 cells. Donor DNA was cotransfected with either one of two guide RNA (gRNA) vectors or with scrambled control using EndoFectin Max Transfection Reagent (GeneCopoeia). Equilibrate gRNA vectors (1 and 2 each 5 μg) and donor DNA (5 μg) and endofectin were diluted in 250 μL of Opti-Mem I (Invitrogen) and incubated for 20 minutes at room temperature to form DNA-Transfection reagent complexes and added to the cells. After selection by puromycin (0.5 μg/mL), the status of CYR61 was determined using Western blotting, and one of the clones (clone 1), which exhibits complete ablation of CYR61 (Supplementary Fig. S2), was used for further studies.

The side population (SP)/stem cells from Panc-1 cell line were isolated according to the methods described previously (21). Briefly, cells were suspended in the culture media at concentration of 1 × 10⁶ cells/100 μL. Vybrant-Violet solution (10 μmol/L) and Verapamil (50 μmol/L) solution were added into the sample and incubated at 37°C for 90 minutes. The cells were incubated with propidium iodide (2 μg/mL) immediately to exclude dead cells before flow cytometry analysis. SP cells were identified, sorted, and analyzed on a BD FACSAria SORP flow cytometer (BD Biosciences) using approximately 405 nm excitation and 440 nm emissions. Sorted cells (i.e., SP and non-SP) were washed in serum-free medium and then cultured in DMEM with 10% FCS for several days in 5% CO₂ at 37°C.

To create stable gemcitabine-resistant Panc-1^CYR61(+) and Panc-1^CYR61(−) cell line, cells were regularly exposed to increasing doses of gemcitabine. Cells were first treated with 50 nmol/L of gemcitabine for 6 days continuously and triggered cell death more than 60%. After 8 weeks of continuous exposure of increasing concentrations of gemcitabine (100 nmol/L–250 nmol/L), the cells were considered as stable to gemcitabine. Then, the cells were cultured after withdrawal of drug for 7–8 passages and resistant stability was checked after exposing the cells to gemcitabine (250 nmol/L) for another 7 days. Gemcitabine treatment dose was applied at the time of medium change at 2–3 days interval.

Vectors containing dCK gene (plasmid pDONR223-DCK) or vector alone (Addgene) was transiently transfected in Panc-1 cells using EndoFectin Max Transfection Reagent (GeneCopoeia). Briefly, 5 μg of DNA and EndoFectin were diluted in 250 μL of serum-free Opti-Mem I (Invitrogen) and incubated for 30 minutes at room temperature to form DNA-Transfection reagent complexes and added to the 70%–90% confluent Panc-1 cells. After 48 hours, transfection efficiency was checked by Western blot analysis.

To determine the effect of different concentrations of gemcitabine (0, 0.10, 0.25, 0.5, 1.0, and 5 mmol/L) for 72 hours on the viability of different pancreatic cancer cell lines, we performed crystal violet–based cell viability (Cell Biolabs) in a 96-well format as described earlier by us (22, 31, 32). The cell viability was measured by means of a Spectra Max 340 microplate reader (Molecular Devices) at a wavelength of 600 nm. The data on cell viability were either reported as an optical density (OD) value or percent viable cells of four different experiments ± SD. The percent cell viability was calculated according the following equation:

where AbsT is the absorbance of gemcitabine-treated cells and AbsC is the absorbance of untreated cells.

Apoptotic cell death was determined using cell death detection ELISA kits (Roche Diagnostics) as described previously by Maity and colleagues (33). Briefly, untreated and treated cells were lysed with lysis buffer and then cytoplasmic supernatants were collected, and total protein of each sample was measured. Cell lysate was added in the streptavidin-coated microplate and allowed to react with a buffer mixture containing anti-histone–biotin and anti-DNA–peroxidase. Microplates were washed with incubation buffer. The ABTS (2,2¢-azino-di-(3-ethylbenzthiazoline-6-sulfonic acid)) chromogen substrate was added to get a color reaction that was measured in the ELISA reader at 405 nm.

The IHC and immunofluorescence staining were performed according to our method described previously (21).

Expression of CD133 and CXCR4 in Panc-1 cells were sorted by flow cytometry using a BD FACSAria III (BD Biosciences) equipped with blue (488 nm) and red (633 nm) lasers. Briefly, Panc-1 cells were harvested using Hank-based enzyme-free cell dissociation buffer (Gibco) and 10⁶ cells were stained with either preconjugated PE anti-human CD184 (CXCR4) or APC anti-human CD133 antibody (BioLegend) for 30 minutes at 2°C–8°C or on ice. Cells were washed several times in PBS containing 0.1% BSA or 1% FCS at 1,200 rpm. Cell pellet was resuspended in 200 μL wash buffer and subjected to flow cytometry analysis.

The PDAC cells were plated in triplicate at a density of 5,000 cells per well of 6-well plate. Twenty-four hours after plating, the cells were treated with different concentrations of gemcitabine (0, 0.1 μmol/L, and 0.5 μmol/L) every alternative day for 7 days. The cells were then fixed and stained with crystal violet and solubilized with 10% acetic acid. The absorbance was determined at 600 nm using Microplate Reader (Spectra Max 340, Molecular Devices).

Cells were seeded in soft agar at a concentration of 2,000 in a 96-well plate per the manufacturer's (Cell Biolabs Inc.) protocol. All conditions were seeded in triplicates. The cells were treated with gemcitabine (0.1 μmol/L and 0.5 μmol/L) or vehicle every alternative day for 10 days. The cells were lysed with lysis buffer, and the plates were read in a fluorimeter using a 485/520 nm filter set.

Panc-1^CYR61(+) and Panc-1^CYR61(−) cells were used to form the spheres according to our earlier method (33). Briefly, the cells were cultured in ultra-low attachment 6-well plates (Corning Inc.) at a density of 3,000 cells/well. Because the half-life of gemcitabine is very short and rapidly converted into inactive metabolite (34, 35), every two days after seeding the cells for 24 hours, the medium was 3/4th replaced very carefully with fresh medium containing different doses of gemcitabine. The procedure was followed by 4 days and 6 days depending on the experimental setups. Spheres were counted and area measured when the diameter reaches more than 60 μm.

For protein analysis, cells were harvested and subjected to Western blotting as described previously (22).

To investigate the interaction of Panc-1^CYR61(+) and Panc-1^CYR61(-) with HPaSteC, we have developed a unique coculture in vitro experimental model. Briefly, 1% agarose gel was allowed to solidify in 12-well plate. After solidification, a round hole/ring-like structure was made in the center and periphery. Tumor cells (5,000 cells) were seeded in the center of the agarose ring and HPaSteC (30,000 cells) were seeded in the outer part of agarose ring of each well and were allowed to attach for overnight. Solid agarose ring acts as barrier to avoid mixing of tumor cells and HPaSteC at the time of seeding. After attachment of cells, agarose barrier ring was taken out carefully without disturbing the cells and coculture were allowed for 5 days. Cells were then fixed in methanol and immunofluorescence of α-SMA was performed with anti-mouse α-SMA, and counterstained with DAPI. Photograph was captured under Leica microscope.

Female and male BALB/c nude mice (n = 5) were inoculated subcutaneously with Panc-1^CYR61(+) cells or Panc-1^CYR61(−) cells (10⁶/mouse) as per our previous method (21). Tumor size and body weight were measured two to three times per week. Once the tumor became palpable, mice were randomized according to tumor size and gemcitabine treatment was started. The mice were injected with gemcitabine (50 mg/kg in PBS) or vehicle (PBS) intraperitoneally twice per week and the tumor growth and body weight were measured for 21 days (36, 37). Tumor growth and relative tumor volume (RTV) were measured as per our established method (33) using studylog^R measurement tools and software (California) three times a week. Antitumor efficacy was measured as a function of tumor growth inhibition (TGI) calculated by the equation:

where T and C is the mean size of tumors in the treated (T) and control (C) groups, respectively, and T₀ represents the tumor size at randomization. The treatment response was evaluated using RECIST criteria.

The statistical analysis was performed using the GraphPad Prism 6 and PASS¹⁵softwares. Results are shown as mean ± SD. Means between the groups were calculated and compared among or within variants using an unpaired, two-sided Student t test. P value of <0.05 was considered statistically significant. The entire studies were performed blindly by two or more investigators.

The CYR61/CCN1 expression has been shown to increase during the progression of PDAC, and it is needed for the proliferation, differentiation, and invasive phenotypes of pancreatic cancer cells (19). In addition, CYR61/CCN1 promotes the Sonic Hedgehog (SHh)-signaling pathway, which confers drug resistance in various cancers (38), suggesting that CYR61/CCN1 could be a key factor of drug resistance in PDAC cells (19). The current IHC analysis is consistent with our earlier findings and shows that the number of CYR61/CCN1-expressing cells was significantly elevated in PDAC samples as compared with the sections of adjacent normal cells and initial stages of the disease (PanIN; Fig. 1A and B). Moreover, gene interaction network profiles in the Connectivity Map reference database showed links among several chemoresistant gene signature pathways and CYR61/CCN1 overexpression (Fig. 1C). Collectively, this material supports the premise of the drug resistance features of CYR61/CCN1.

The drug resistance of pancreatic cancer cells is associated with the subpopulation of CD133⁺ or CD133⁺/CXCR4⁺-cancer stem cell–like phenotypes (39). In this study, FACS analysis found that CD133⁺ and CXCR4⁺ cells were markedly less in CYR61-ablated Panc-1 cells as compared with parental Panc-1 cells (Fig. 1D). Moreover, immunofluorescence and immune Western blot studies also showed that CYR61/CCN1 blockade significantly reduced the production of CD133 and CXCR4 in pancreatic cancer cell lines (Fig. 1E and F). This further reinforces the concept of the drug-resistant ability of CYR61/CCN1 in pancreatic cancer cells.

On the basis of the preceding data, CYR61/CCN1 can be considered a driver of chemoresistance. Thus, in this study, we evaluated how and to what extent CYR61/CCN1-positive pancreatic cells adapt to survive with gemcitabine. To do so, we knockdown CYR61/CCN1 in Panc-1 and AsPC-1 cells using siRNA or CRISPR/Cas-9, respectively (Supplementary Figs. S2 and S3). Cells were then treated with different doses of gemcitabine for 72 hours and an analysis of cell viability and apoptosis was performed. We found that lower concentrations of gemcitabine (0.1 or 0.25 μmol/L) significantly impaired the viability of Panc-1^CYR61(−), while no or minimal effect was detected in Panc-1^CYR61(+) cells (Fig. 2A). In contrast, higher concentrations of gemcitabine (0, 0.5, 1.0, or 5.0 μmol/L) significantly blocked cell growth in both cell types with a higher effect in Panc-1^CYR61(-) (Fig. 2B). These findings were further corroborated in AsPC-1 cells in which the CYR61/CCN1 gene was ablated by CRISPR/Cas-9 (Fig. 2C). The growth inhibition of gemcitabine in both CYR61-positive and CYR61-negative Panc-1 cells presided via apoptosis (Fig. 2D). Collectively, the studies show that CYR61/CCN1 makes pancreatic cancer cells resistant to gemcitabine.

Previously, we have shown that two subpopulations can be obtained with the Panc-1 cell line. These include SP/tumor-initiating cells (TIC)/cancer stem cells (CSC) and non-SP (NSP)/neoplastic cells (21). SP cells expressed elevated levels of CYR61/CCN1, while CYR61/CCN1 expression was undetected in NSP cells (ref. 21; Supplementary Fig. S4). To corroborate the drug-resistant role of CYR61/CCN1 in pancreatic cancer cells, SP and NSP cells were exposed to different doses of gemcitabine in the presence or absence of hrCYR61 (250 ng/mL) for 72 hours, and cell viability was measured. The study showed that gemcitabine was significantly more effective in NSP cells as compared with SP cells (Fig. 2E). Moreover, the effect of gemcitabine on NSP cells can be reduced by treating them with hrCYR61 as compared with SP (Fig. 2E), thus supporting the premise that CYR61/CCN1 overexpression protects pancreatic cancer cells from gemcitabine-induced cell death.

Our next goal was to investigate whether hrCYR61 treatment (250 ng/mL for 72 hours) enhances the gemcitabine-resistant phenotype in human pancreatic ductal epithelial (HPDE) cells, which lack CYR61/CCN1 expression. We found that CYR61 is unable to resist the effect of gemcitabine in HPDE cells, suggesting a unique physiologic environment is needed for CYR61 to act as a drug-resistant molecule. This environment is possibly lacking in HPDE cells (Fig. 2F). However, to determine this, further studies are warranted.

To better understand the role of CYR61 in gemcitabine resistance in pancreatic cancer cells, we tested the effect of gemcitabine on anchorage-dependent growth (ADG) and anchorage-independent growth (AIG) of Panc-1^CYR61(-) and Panc-1^CYR61(+) cells. We saw that a low (0.1 μmol/L) and high (0.5 μmol/L) doses of gemcitabine significantly impaired ADG in Panc-1^CYR61(−) as compared with Panc1^CYR61(+) (Fig. 3A and B). Like ADG, gemcitabine blocked AIG in both populations. However, the CYR61-ablated cells were roughly three times less likely to undergo AIG, suggesting that gemcitabine in a low or high dose reduces the transformation ability of Panc-1^CYR61(−) cells (Fig. 3C).

Our previous studies revealed that CYR61 promotes EMT followed by cancer stemness in pancreatic cancer cells (21, 26). Several studies have shown that cancer stem-like cells are able to proliferate and expand unlimitedly in vitro as tumor sphere formations, and thus, we wanted to investigate whether gemcitabine, in the absence of CYR61, was able to significantly destroy the sphere-forming ability of pancreatic cancer cells. We found that Panc-1^CYR61(+) and Panc-1^CYR61(−) cells could form the spheres in presence or absence of gemcitabine (Supplementary Fig. S5). However, the number and the size of spheres are significantly reduced in gemcitabine-exposed Panc-1^CYR61(−) cells as compared with gemcitabine-exposed Panc-1^CYR61(+) cells in both concentrations (0.1 and 0.5 μmol/L; Supplementary Fig. S5; Fig. 3D), indicating CYR61 ablation makes pancreatic cancer stem cells sensitive to gemcitabine.

To produce the cytotoxic effect of gemcitabine, it needs to be activated by complex intracellular phosphorylations to yield gemcitabine diphosphate (dFdCDP) and triphosphate (dFdCTP). dCK plays a vital role in initiating this sequential process (Fig. 4A). Lack of expression of this enzyme, which is a common event in PDAC, diminishes the functional efficiency of gemcitabine. We found that the expression level of CYR61/CCN1 is inversely proportional to the expression level of dCK. The PDAC cell line BxPC-3, which is not very gemcitabine-resistant as compared with other cell lines (ref. 40; Supplementary Fig. S6), overexpressed dCK and had almost no CYR61/CCN1 expression. In contrast, the dCK expression is downregulated in CYR61/CCN1-overexpressing and gemcitabine-resistant Panc-1 cells (Fig. 4B and C). Moreover, we found that the expression level of dCK can be restored in parental and gemcitabine-induced over resistant Panc-1 cells by knocking down CYR61/CCN1 (Fig. 4D) or by reducing dCK levels by treating BxPC-3 with hrCYR61 (Fig. 4E). Collectively, this study suggests that CYR61/CCN1-induced chemoresistance could be mediated through the downregulation of dCK. To corroborate the notion, we transfected Panc-1 cells with the expression vectors containing dCK gene or vector alone in Panc-1 cells. The dCK-overexpressing and underexpressing Panc-1 cells (Fig. 5A) were then treated with gemcitabine (5 μmol/L) for 72 hours and cell viability was measured. We found that the cell viability was significantly reduced in dCK-overexpressing Panc-1 cells as compared with vector alone–transfected cells (Fig. 5B).

Desmoplasia in PDAC plays a critical role in the intrinsic and extrinsic resistance to gemcitabine (15). A desmoplastic reaction in PDAC is marked by an intense increase in the proliferation of α-smooth muscle actin (α-SMA)-positive fibroblasts (41). Stromal fibroblasts in PDAC are recognized as activated PSCs (42). In a normal pancreas, PSCs are quiescent. But in response to a pancreatic injury or neoplasm, PSCs transform from a quiescent to “activated” state, which is also called a “myofibroblastic state” (42). Activated PSCs/myofibroblasts are characterized by α-SMA expression. Consistent with earlier work, we found that a PSC (HPaSteCs) cell line is always in an active state under culture conditions and achieves a myofibroblast-like phenotype with high expression of α-SMA (Fig. 6A). PSCs do not or minimally express CYR61 under a regular culture setup (Fig. 6A). Given the importance of PSCs/myofibroblast cells in desmoplasia during PDAC development, we first investigated the effect of CYR61/CCN1 on α-SMA in stellate cells. We found the level of α-SMA significantly increased in CYR61 recombinant protein (250 ng/mL)-treated stellate cells as compared with untreated cells (Supplementary Fig. S7). Next, we examined the effect of pancreatic cancer cell–secreted CYR61 on fibroblast cells' activity and growth. To do so, we cocultured Panc-1^CYR61(+) or Panc-1^CYR61(−) and HPaSteCs for 7 days in our unique experimental settings (Fig. 6B, left). We found that α-SMA⁺ HPaSteCs were significantly reduced when they interact with CYR61/CCN1-depleted pancreatic tumor cells as compared with Panc-^1CYR61(+) where abundant α-SMA⁺ PSCs were detected (Fig. 6B, middle and bottom). Finally, we investigated whether CYR61 has any role in the viability of stellate cells. To do so, HPaSteC cells were grown in conditioned media (CM) of Panc-1^CYR61(+) or Panc-1^CYR61(-) for 72 hours, and cell viability was measured. The results show that the viability of HPaSteCs was significantly decreased when they were grown in CM of Panc-1^CYR61(-) cells as compared with the CM of parental CYR61 positive Panc-1 cells (Fig. 6B, bottom right bar graph). Collectively, these studies suggest that CYR61 plays a vital role in survival and activation of stellate cells that needed for desmoplasia.

CTGF/CCN2, a profibrotic secretory growth factor, plays a vital role in tumor–stromal interaction and desmoplasia in PDAC (20, 43–45). CTGF/CCN2 is overexpressed in PDAC samples, various PDAC cells and reactive fibroblast/stellate cells (43, 44). We found CTGF was highly expressed in AsPC-1 and Panc-1 cell lines as compared with BxPC-3 and Capan-1 cell lines where CTGF/CCN2 expression was undetected or minimally detected (Fig. 6C). We then analyzed CTGF expression levels during the blockade of CYR61/CCN1 expression by shRNA or function by specific antibody treatment in PDAC cells. We found that CTGF production was significantly impaired in CYR61/CCN1–ablated Panc-1 and AsPC-1 cells (Fig. 6D and E).

Finally, we determined the CTGF status in human fibroblasts (FEF3; ref. 46), and HPaSteCs. We observed CTGF expression was undetected in HPaSteCs (Fig. 6F). However, treatment of hrCYR61/CCN1 protein (250 ng/mL for 48 hours) enhanced the production of CTGF in HPaSteCs (Fig. 6G). Collectively, these results suggest an autocrine–paracrine pathway is involved in CYR61/CCN1-induced CTGF/CCN2 expression in pancreatic cancer cells and pancreatic stellate cells (Fig. 6H).

Finally, we tested the role of CYR61/CCN1 in a pancreatic cancer cell subcutaneous transplant xenograft model. Weekly measurements of tumor volumes revealed that tumors in the Panc-1^CYR61(+) group (n = 5) continued to increase significantly throughout a 21-day period in the presence of gemcitabine (50 mg/kg/twice a week by injection) as compared with Panc-1^CYR61(−) tumors without having significant difference in body weight (Fig. 7A), supporting the concept that CYR61/CCN1 is a prime driving force toward PDAC cells' resistance to gemcitabine.

Because we found in this study that CYR61/CCN1 controls stellate cells' activity and proliferation (Fig. 6), we next sought to address whether CYR61/CCN1 is linked with desmoplastic reaction in vivo. The studies showed that the desmoplastic reaction was markedly higher in Panc-1^CYR61(+) tumors as compared with Panc-1^CYR61(-) tumors where desmoplasia was minimal or undetected (Fig. 7B). The nonsignificant effect of gemcitabine on desmoplastic reaction was observed in Panc-1^CYR61(+) tumors.

Finally, we tested whether a CYR61/CCN1 deficiency along with gemcitabine can alter the expression level of α-SMA, CTGF, and dCK. We found that α-SMA and CTGF are markedly downregulated while the levels of dCK were significantly upregulated in the cell lysates of CYR61-deficient Panc-1 cells and along with gemcitabine treatment using immuno-Western blotting (Fig. 7C).

While gemcitabine is still considered as a first-line therapy and is given alone or combined with other agents to attack locally advanced or metastatic pancreatic cancers, the success is very limited and disappointing (5, 6, 8, 47, 48). The failure of gemcitabine to attack aggressive pancreatic cancers cells is due to intrinsic (metabolic malfunction) or extrinsic resistance (drug delivery/desmoplasia) or both to gemcitabine (9, 15, 47). Thus, multiple studies have been carried out to understand the mechanism, and they gave new insights into the gemcitabine resistance of aggressive cancer cells having cancer stem cell–like properties (47, 49–51). These studies, collectively, suggested that simultaneously targeting intrinsic- and extrinsic-resistant events may increase the toxic effects of gemcitabine and increase patient survival. Our studies show that depleting CYR61/CCN1 impairs cancer stemness behaviors, while concomitantly suppressing intrinsic and extrinsic resistant phenotypes of gemcitabine in aggressive PDAC cells via the regulation of dCK and CTGF. Therefore, targeting CYR61/CCN1 could be an ideal strategy to promote the effects of gemcitabine in killing pancreatic cancer cells.

In PDAC, CYR61/CCN1 needs to maintain cancer stem cell/tumor-initiating behaviors that are critical for invasive progression and drug resistance (19). In this study, we gave three evidences that broaden our understanding of the involvement of CYR61/CCN1 in chemoresistance to PDAC cells. First, we show that compared with normal, and early stages of the disease such as PanIN1, CYR61/CCN1 was found to be overexpressed in invasive pancreatic cancers (Fig. 1A and B). Second, we showed through the interaction network profile analysis that CYR61/CCN1 signaling is necessary to make cancer cells chemoresistant (Fig. 1C). Third and finally, we showed that CYR61/CCN1 depletion in pancreatic cancer cells significantly reduces the number of CD133- and CXCR4-producing cells (Fig. 1D and E), which have the capacity to resist chemotherapy (51). Taken together, our current studies validate the drug-resistant function of CYR61/CCN1 in pancreatic cancer cells.

Given the importance of the role of CYR61/CCN1 in chemoresistance, we investigated the definite role of CYR61/CCN1 in resistance to gemcitabine. We showed that the depletion of CYR61/CCN1 in PDAC cell lines significantly enhances the growth inhibition efficacy of gemcitabine via enhancing apoptotic cell death (Fig. 2). This finding was further supported by an analysis of side-populations (SP and NSP) of pancreatic cancer cells (Fig. 2E). The studies revealed that SP cells in which CYR61/CCN1 is overexpressed are more resistant to gemcitabine as compared with NSP cells having minimal or no expression of CYR61/CCN1. We found that NSP cells were less likely to survive than SP cells after exposing to them to gemcitabine. Interestingly, the growth-inhibitory effect of gemcitabine could be reduced significantly when the NSP cells were exposed conjointly with hrCYR61. Collectively, these studies suggest that CYR61/CCN1 overexpression in aggressive pancreatic cancer cells is a driving force for gemcitabine-resistant phenotypes. Ironically, the above concept was not pertinent in the nontransformed human pancreas cell line (HPDE) lacking CYR61/CCN1 expression (Fig. 2F), suggesting a unique physiologic environment is needed for CYR61 to act as a drug-resistant molecule, which is possibly lacking in HPDE cells.

Gemcitabine-resistant pancreatic cancers cells have increased anchorage-dependent and independent colony formation, migration, invasion, and sphere-forming ability (52). These aggressive features were weakly affected by gemcitabine treatment. The current findings reveal that CYR61/CCN1 depletion makes aggressive pancreatic cancer cells sensitive to gemcitabine even at a low concentration and significantly reduces the transforming ability (Fig. 3A–C), and sphere-forming skill of PDAC cells, implicating that CYR61/CCN1 ablation reverses the cancer stem cell properties and they could be mediated by reprogramming the mesenchymal–epithelial transition (21, 53).

The acquisition of chemoresistance against gemcitabine is complex in its chemistry and pathobiology. Classically, gemcitabine is activated inside the cells after sequential phosphorylations, and the first step of phosphorylation is carried out by an enzyme dCK (Fig. 4A; refs. 13, 54). In gemcitabine-resistant pancreatic cancer cells, the expression of dCK is remarkably less, and upregulation of dCK makes these cells sensitive to gemcitabine. Thus, the inactivation of dCK is considered as a critical intrinsic pathway to make cancer cells resistant to gemcitabine (13, 54). Our studies find that dCK expression in PDAC cells can be suppressed by CYR61/CCN1 overexpression or treatment. Furthermore, induced expression of dCK in CYR61-overexpressing PDAC cells promotes sensitive of gemcitabine in gemcitabine-resistant cells (Fig. 5). Therefore, we suggest that CYR61/CCN1 overexpression in pancreatic cancer cells coopt the resistant phenotype of these cells via suppressing dCK (Figs. 4 and 5). A recent study has shown that pancreatic stromal cell–secreted CYR61, which is induced by the TGFβ–ALK5–Smad pathway, has been involved in gemcitabine resistance (50). Although we did not find CYR61/CCN1 expression in the stroma of human PDAC or mouse model, we can now speculate that the CYR61/CCN1 from both sources could be involved in regulation of gemcitabine resistance to PDAC cells.

Despite some controversies (55), many patients with PDAC are resistant to gemcitabine because of desmoplasia, which is about 90% of the tumor volume and help in promoting intrinsic and extrinsic resistance to gemcitabine (47, 49). One of the primary cellular compartments of desmoplasia is cancer cell–associated fibroblasts (CAF). During the progression of the PDACs, CAFs are generated from quiescent fibroblasts to active myofibroblast-like phenotypes/stellate cells (PSCs) characterized by the expression of α-SMA (41, 42, 47). However, the mechanism to kick-start this transformation process is largely unknown. The current studies with a unique in vitro model show that pancreatic cancer cell–secreted CYR61/CCN1 plays a vital role in the transformation of CAFs from quiescent fibroblasts and their proliferation (Fig. 6A and B). CTGF/CCN2 plays a vital role in tumor–stromal interaction and desmoplasia in PDAC (20, 43–45). Recent studies have shown that tumor cell–derived CTGF is a prerequisite signal for fibroblast activation (56). Because our current studies found that CTGF expression in pancreatic cancer cells and fibroblast cells is regulated by CYR61/CCN5 (Fig. 6C–H), we suggest that CTGF could be an intermediate signaling molecule necessary for fibroblast activation. However, further studies are warranted.

Finally, to validate the in vitro data, we tested the impact of gemcitabine in a preclinical mouse model. We observed that gemcitabine significantly suppressed the growth of subcutaneous tumors that were developed by CYR61/CCN1-depleted Panc-1 cells (Fig. 7A). Moreover, in CYR61/CCN1–depleted tumors, CAF transformation and CTGF production were significantly decreased and at the same time dCK production was markedly elevated (Fig. 7B). These studies collectively suggest that pancreatic cancer patients' intrinsic and extrinsic resistance of gemcitabine are CYR61/CCN1-dependent, and therefore, blocking CYR61/CCN1 activity would be an ideal strategy to kill cancer cells as well as destroy desmoplastic growth in PDAC by gemcitabine.

D.D. Von Hoff reports receiving other commercial research support and is a consultant/advisory board for Eli Lilly. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Ghosh, A. Das, S.K. Banerjee, S. Banerjee

Development of methodology: G. Maity, A. Ghosh, V.G. Gupta, I. Haque, A. Das, S.K. Banerjee, S. Banerjee

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Maity, A. Ghosh, V.G. Gupta, I. Haque, A. Das, K. Dhar, S.K. Banerjee

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Maity, A. Ghosh, V.G. Gupta, I. Haque, K. Dhar, S. Gunewardena, S.K. Banerjee, S. Banerjee

Writing, review, and/or revision of the manuscript: G. Maity, A. Ghosh, V.G. Gupta, I. Haque, S. Sarkar, S. Bhavanasi, D.D. Von Hoff, S. Mallik, S. Kambhampati, S.K. Banerjee, S. Banerjee

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): G. Maity, S. Sarkar, S.K. Banerjee

Study supervision: A. Ghosh, S.K. Banerjee, S. Banerjee

We thank Archana De for technical help, VA Research office and Midwest Biomedical Research Foundation for administrative and secreterial supports. The work is supported by Merit review grant from Department of Veterans Affairs (5I01BX001989-04, to S.K. Banerjee and I01BX001002-05, to S. Banerjee), KUMC Lied Basic Science Grant Program (to S.K. Banerjee), and Grace Hortense Greenley Trust, directed by The Research Foundation in memory of Eva Lee Caldwell (to S. Banerjee and SK).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.