• Open Access

Tumor–stroma interactions reduce the efficacy of adenoviral therapy through the HGF-MET pathway

Authors


To whom correspondence should be addressed.
E-mail: kenoki@med.kyushu-u.ac.jp

Abstract

Many preclinical studies have shown the potential of adenovirus-based cancer gene therapy. However, successful translation of these promising results into the clinic has not yet been achieved. Pancreatic ductal adenocarcinoma (PDAC) is characterized by abundant desmoplastic stroma, and tumor–stromal cell interactions play a critical role in tumor progression. Therefore, we hypothesized that tumor–stroma interactions reduce the efficacy of adenoviral therapy. We investigated the effect of fibroblasts on adenovirus-based gene therapy using SUIT-2 and PANC-1 pancreatic cancer cells cultured with or without fibroblast-conditioned culture supernatant then infected with Ad-LacZ. After 48 h, the cells were stained for β-galactosidase. The results showed that the number of β-galactosidase-positive cells was significantly reduced after culture with fibroblast-conditioned supernatant (< 0.05). Because the hepatocyte growth factor (HGF)/MET pathway plays an important role in tumor–stroma interactions we next investigated the involvement of this pathway in tumor–stroma interactions leading to the decreased efficacy of adenoviral therapy. SUIT-2 cells were cultured with or without SU11274 (a MET inhibitor) and/or fibroblast-conditioned culture supernatant, then infected with Ad-GFP. After 48 h, GFP-positive cells were counted. The number of GFP-positive cells in cultures containing fibroblast-conditioned supernatant plus SU11274 was significantly greater than in cultures without SU11274. In conclusion, our results suggest that stromal cells in PDAC reduce the efficacy of adenoviral therapy through a mechanism involving the HGF/MET pathway. Control of such tumor–stroma interactions may lead to improvements in adenoviral gene therapy for PDAC. (Cancer Sci 2011; 102: 484–491)

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal human malignancies. Because of its aggressive rate of growth and metastasis, only 10–15% of patients are resectable at diagnosis.(1) Difficulties in early diagnosis and the limited efficiency of currently available systemic therapies(2–4) mean that the prognosis is extremely poor, with a median survival time of 6 months and an overall 5-year survival rate of <5%.(5,6) Therefore, new therapeutic approaches are needed.

Adenovirus-based cancer gene therapy is a promising novel approach for treating cancers that are resistant to established therapies.(7) However, although adenoviruses show promising results in vitro, they have fallen short of their expected therapeutic values in patients. The poor and heterogeneous penetration of the virus into tumors is a major cause of this failure.(8,9) In patients with head and neck cancers treated with oncolytic adenovirus, ONYX-015, the response rate was only 14%.(10)

Infiltrating PDAC is characterized by an abundant desmoplastic stroma that is unique among solid tumors. Increasing evidence suggests that tumor–stromal cell interactions within the cellular microenvironment of a variety of cancers play a critical role in tumor progression,(11) angiogenesis,(12) chemoresistance,(13) and metastasis.(14) Nevertheless, the effects of stromal components on adenoviral therapy have not been studied. We hypothesized that tumor–stroma interactions may reduce the efficacy of adenoviral therapy. To test this hypothesis, we investigated the effect of fibroblasts on the efficiency of adenoviral-based gene therapy.

Materials and Methods

Cell lines, primary fibroblast cultures, and reagents.  The human pancreatic cancer cell lines, SUIT-2 and PANC-1, were generously donated by Dr. H. Iguchi (National Shikoku Cancer Center, Matsuyama, Japan). Primary fibroblast cultures (f1, f2, and f3) were isolated from three patients with PDAC who underwent pancreatoduodenectomy in our department as previously described.(15) Each patient gave their informed consent. Cells were cultured in DMEM supplemented with streptomycin, penicillin, and 10% FBS at 37°C in 5% CO2. The MET inhibitor, SU11274, was purchased from SUGEN (South San Francisco, CA, USA). The monoclonal anti-human hepatocyte growth factor (HGF) antibody was purchased from R&D Systems (Minneapolis, MN, USA).

Construction of recombinant adenovirus.  A control vector expressing the bacterial β-galactosidase gene (LacZ) was generated by homologous recombination of the pJM17 plasmid and the shuttle plasmid vector pSV2+ containing an expression cassette and the CMV early promoter/enhancer, followed by LacZ and a polyadenylation signal. Recombinant Ad-LacZ was propagated in HEK293 cells. An adenovirus expressing GFP (Ad-GFP) was generated, amplified, and titrated as previously described.(16,17)

Adenoviral infection of cells.  Cells were seeded into six-well plates and cultured in DMEM supplemented with 10% FBS either with or without concentrated fibroblast culture supernatant for 48 h. To adjust for the FBS concentration, supernatant from fibroblasts cultured without FBS was used in the experiments after concentration using an Amicon Ultra-15 10K device (Millipore, Billerica, MA, USA) at 4000g for 20 min. Cells were infected with Ad-LacZ at 10 MOI. The culture medium was replaced with fresh medium 1 h after transfection.

Assessment of transgene distribution by evaluation of β-galactosidase (β-gal) expression.  Forty-eight hours after adenovirus infection, SUIT-2 cells were rinsed with PBS and fixed. β-Galactosidase activity was detected by immersing cells in X-gal staining solution for 6 h at 37°C, as previously described.(18)

Cell proliferation assay.  Cell proliferation was evaluated by propidium iodide (PI) fluorescence intensity. Cells were plated in 24-well tissue culture plates and cultured with or without fibroblast-conditioned culture supernatant 2 days. They were then infected with Ad-LacZ at a 10 MOI. Two days later, PI (30 μM) and digitonin (600 μM) were added to each well with PI. Fluorescence intensity, corresponding to total cells, was measured with an Infinite F200 (Tecan Trading, Mannedorf, Switzerland) fitted with 535-nm excitation and 620-nm emission filters. Cell proliferation was defined as the ratio of fluorescence intensity at a given time point relative to that measured at the beginning of the experiment. All experiments were carried out in triplicate wells.

Real-time PCR and RT-PCR assays.  The Ad-LacZ DNA content of infected cells was determined by real-time PCR analysis using SYBR Premix Ex Taq 2 (Takara, Tokyo, Japan) and a Chromo4 System (Bio-Rad, Hercules, CA, USA), as previously described.(18) The primers used for the β-gal gene were: 5′-CACGGCAGATACACTTGCTG-3′ and 3′-ATCGCCATTTGACCACTACC-5′.(19) The number of viral DNA copies was calculated from a standard curve obtained for the purified adenovirus vector (CMV–β-gal) and was further adjusted according to the protein concentration of each lysate. The mRNA levels for the Coxsackie virus and adenovirus receptor (CAR), αvβ3-integrin, αvβ5-integrin, clathrin, and dynamin 2 were quantified by real-time RT-PCR assay using a QuantiTect SYBR Green RT-PCR kit (Qiagen, Tokyo, Japan), 10 ng total RNA, and primers specific for: CAR, 5′-GGCGCTCCTGCTGTGC-3′ and 5′-CTTTGGCTTTTTCAATCATCTTC-3′; αvβ3-integrin, 5′-GAGGATGACTGTGTCGTCAG-3′ and 5′-CTGGCGCGTTCTTCCTCAAA-3′; αvβ5-integrin, 5′-CCTGTCCATGAAGGATGACTTG-3′ and 5′-CTCATTGAAGCTGTCCACTCTG-3′; clathrin, 5′-CGGTTGCTCTTGTTACGG-3′ and 5′-CGGTTGCTCTTGTTACGG-3′; and dynamin 2, 5′-AGGAGTACTGGTTTGTGCTGACTG-3′ and 3′-GTGCATGATGGTCTTTGGCATGAG-5′,(19) as previously described.(18) Each sample was run in triplicate. The level of each mRNA was normalized to those of 18S rRNA amplified using the specific primers 5′-GTAACCCGTTGAACCCCATT-3′ and 5′-GCGATGATGGCTAACCTACC-3′, and expressed as a ratio compared with untreated controls.

Western blot analysis.  Western blotting to examine MET activity was carried out as described previously.(18) SUIT-2 and PANC-1 cells were cultured with or without fibroblast-conditioned culture supernatant, SU11274/DMSO and an HGF-neutralizing antibody/PBS and lysed in lysis buffer. The proteins were fractionated by 12% SDS-PAGE and transferred to a membrane (Millipore). The membrane was then incubated with rabbit polyclonal anti-phosphorylated met (1:500; Cell Signaling Technology, Beverly, MA, USA), anti c-met (1:1000; Santa Cruz Biotechnology, Santa Cruz, CA, USA), or anti-actin antibodies, then probed with anti-rabbit IgG (1:2500; Cell Signaling Technology) or anti-mouse IgG (1:5000; Invitrogen, Carlsbad, CA, USA).

Statistical analysis.  Values are expressed as the mean ± SD. Comparisons between all groups were analyzed by one-way anova, and Student’s t-test was used for comparisons between two groups. The level of statistical significance was set at < 0.05. To confirm the induction results, experiments were repeated at least three times.

Results

Effect of fibroblasts on the morphology of pancreatic cancer cells.  To investigate the effect of fibroblasts on pancreatic cancer, we examined the morphology of pancreatic cancer cell lines cultured with or without concentrated fibroblast-conditioned culture supernatant. As shown in Figure 1, the morphology of pancreatic cancer cells treated with fibroblast-conditioned supernatant changed from round to spindle-shaped. This morphologic change was observed in both SUIT-2 and PANC-1 cells treated with the supernatant from all fibroblast strains (f1, f2, and f3), and the greatest change was seen in SUIT-2 cells treated with the supernatant from f2 and f3 fibroblasts.

Figure 1.

 Morphologic changes in pancreatic cancer cells treated with fibroblast-conditioned supernatant. (A) SUIT-2 cells and (B) PANC-1 cells were treated with or without fibroblast-conditioned supernatant for 2 days. Magnification, ×200.

Effect of fibroblasts on adenoviral gene uptake.  We next investigated the effect of fibroblasts on adenoviral gene uptake in pancreatic cancer cells. SUIT-2 and PANC-1 cells were cultured with or without fibroblast-conditioned supernatant for 48 h then infected with Ad-LacZ at 10 MOI. After 48 h the viral DNA content was quantified using real-time qPCR. As shown in Figure 2(A), the viral DNA content of cells cultured with fibroblast-conditioned supernatant was significantly lower than that of untreated cells. This suggests that tumor–stroma interactions decrease viral gene uptake by pancreatic cancer cells.

Figure 2.

 Adenoviral gene uptake and β-galactosidase (β-gal) expression in pancreatic cancer cells cultured with or without (−) fibroblast-conditioned supernatant. (A) SUIT-2 and PANC-1 cells were infected with Ad-LacZ at 10 MOI and DNA was extracted 48 h later. Viral DNA content was quantified by real-time PCR and expressed as a ratio compared with cells cultured with fibroblast-conditioned supernatant. Each value represents the mean ± SD of triplicate measurements. *< 0.05. (B) Photomicrographs of X-gal stained cells cultured with or without fibroblast-conditioned supernatant. Magnification, ×200. β-Galactosidase expression in Ad-LacZ treated cells cultured with or without fibroblast-conditioned supernatant for 48 h. Cells were infected with Ad-LacZ at 10 MOI and β-gal activity was assessed by X-gal staining 48 h after infection. (a) SUIT-2 cells; (b) PANC-1 cells. (C) Number of β-gal-positive cells. Each value represents the mean ± SD of five independent fields. *< 0.05.

Effect of fibroblasts on β-gal expression in Ad-LacZ-infected cells.  To investigate the effect of fibroblasts on the expression of transgenes delivered by the adenoviral vector, we examined the expression of β-gal in the Ad-LacZ transfected cells. SUIT-2 and PANC-1 cells were cultured with or without fibroblast-conditioned supernatant for 48 h then infected with Ad-LacZ at 10 MOI. After 48 h the cells were stained for β-gal. As shown in Figure 2(B), large numbers of cells cultured in the absence of fibroblast-conditioned supernatant showed the characteristic blue staining, whereas only a small number of cells cultured with fibroblast-conditioned supernatant were positive. Cells not infected with Ad-LacZ were all negative (Fig. 2C, < 0.05). This suggests that tumor–stroma interactions inhibit the expression of genes delivered by adenoviral vectors.

Effect of fibroblast-conditioned culture supernatant on cancer cell proliferation.  Next, we evaluated the proliferation of pancreatic cancer cells cultured with or without fibroblast-conditioned culture supernatant and adenovirus using a PI assay. As shown in Figure 3, there was no difference in proliferation between pancreatic cancer cells cultured with fibroblast-conditioned culture supernatant and those cultured without fibroblast-conditioned culture supernatant either before adenoviral infection and 2 days after infection. These data suggest that cell death caused by viral infection under any of the conditions did not affect the results shown in Figure 2(A–C).

Figure 3.

 Cell proliferation assay for pancreatic cancer cells cultured under each condition. (A) SUIT-2 cells and (B) PANC-1 cells were plated in 24-well tissue culture plates, cultured with or without (−) fibroblast-conditioned culture supernatant for 2 days, then infected with Ad-LacZ at 10 MOI. Two days later, propidium iodide (30 μM) and digitonin (600 μM) were added to each well. Fluorescence intensity was then measured.

Effect of fibroblasts on CAR, αvβ3-integrin, αvβ5-integrin, clathrin, and dynamin 2 mRNA levels in pancreatic cancer cells.  To express the transgene, several steps are required, which include cell attachment, endocytosis, endosomal escape, cytoplasmic transport and disassembly, and DNA import. All adenovirus serotypes, except subgroup B, bind to a primary receptor, CAR. The CAR-docked particles activate αvβ3- and αvβ5-integrins and their co-receptors, triggering endocytosis.(20,21) Endocytosis of adenoviral particles is mediated by clathrin and involves the large GTPase, dynamin.(21) The present data show that tumor–stroma interactions decreased viral gene uptake and expression of transgenes delivered by the adenoviral vector. To investigate this mechanism, we quantified the levels of CAR, αvβ3- and αvβ5-integrins, clathrin, and dynamin 2 mRNA in SUIT-2 cells using real-time qRT-PCR (Fig. 4). We found that the expression of CAR, αvβ5-integrin, and clathrin mRNA was significantly lower in cells cultured with the supernatants from all fibroblasts cultures than in the control (< 0.05). The level of αvβ3-integrin mRNA expression was affected only in cells cultured with f2 and f3 conditioned supernatants. The level of dynamin 2 mRNA expression was only affected in cells cultured with f1 conditioned supernatant. These data suggest that tumor–stroma interactions prevent efficient adenovirus-based gene therapy at the point of cell attachment and endocytosis.

Figure 4.

 Effect of fibroblasts on the expression levels of Coxsackie virus and adenovirus receptor (CAR) (A), αvβ3-integrin (B), αvβ5-integrin (C), clathrin (D), and dynamin 2 (E) mRNA expression in cells treated with or without (−) fibroblast-conditioned supernatant. mRNA expression was quantified by real-time RT-PCR using total RNA from cells cultured with or without fibroblast-conditioned supernatant. Each value represents the mean ± SD of triplicate measurements. *< 0.05.

Effect of MET inhibitors on fibroblast-induced decrease in efficacy of adenoviral therapy.  Previously, we reported that the HGF/MET pathway plays a critical role in tumor–stromal cell interactions.(22–25) Therefore, to investigate the involvement of the HGF/MET pathway in tumor–stroma interactions leading to the decreased efficacy of adenoviral therapy, we used SU11274 (a MET inhibitor) and an HGF-neutralizing antibody to block this pathway.(26,27) SU11274 is a small ATP-competitive molecule that targets the ATP binding site of MET and blocks HGF-dependent Met activation.(26–28) SU11274 is selective for Met rather than other tyrosine kinases, including Flk, EGFR, PDGFßR, Tie2, c-src, cdl, and FGFR-1.(27)

The effect of SU11274 and the HGF neutralizing antibody on the expression levels of c-Met and phosphorylated Met proteins in SUIT-2 and PANC-1 cells was evaluated by Western blot analysis. As shown in Figure 5(A), fibroblast-conditioned supernatant increased the expression of the phosphorylated Met protein in SUIT-2 and PANC-1 cells, whereas both SU11274 and the HGF neutralizing antibody reversed this increase. The expression level of c-MET was unchanged. These data suggest that the fibroblast-conditioned supernatant increased the level of HGF-mediated Met phosphorylation, and that this was inhibited by both SU11274 and the HGF neutralizing antibody.

Figure 5.

 Effect of MET inhibitors on the fibroblast-induced decrease in the efficacy of adenoviral therapy in pancreatic cancer cells. (A) Effect of SU11274 and a hepatocyte growth factor (HGF) neutralizing antibody on the expression levels of c-Met and phosphorylated Met (p-Met) proteins. SUIT-2 and PANC-1 cells were cultured with or without (−) fibroblast-conditioned culture supernatant, SU11274/DMSO or HGF neutralizing antibody/PBS, then lysed. The proteins were subjected to Western blot analysis with antibodies specific for activated p-MET or c-MET. (B) Effect of MET inhibitors on fibroblast-induced morphologic changes. SUIT-2 cells were treated with or without SU11274 and/or fibroblast-conditioned culture supernatant for 2 days. Magnification, ×200. (C) Effect of MET inhibitors on the fibroblast-induced decrease in transgene expression. SUIT-2 cells were treated with or without SU11274 and/or fibroblast culture supernatant for 2 days, then infected with Ad-GFP at 10 MOI. Two days after infection, GFP expression was observed using a fluorescent microscope. Magnification, ×200. (D) Number of GFP-positive cells. Each value represents the mean ± SD of five independent fields. *< 0.05. (E) Effect of MET inhibitors on the fibroblast-induced decrease in the levels of Coxsackie virus and adenovirus receptor (CAR), αvβ5-integrin, and clathrin mRNA expression. (a) CAR, (b) αvβ5-integrin, and (c) clathrin mRNA expression in SUIT-2 cells cultured with or without SU11274 and/or fibroblast-conditioned culture supernatant. *< 0.05.

We also investigated the effect of SU11274 on fibroblast-induced morphologic changes in pancreatic cancer cells. As shown in Figure 5(B), pancreatic cancer cells treated with both SU11274 and fibroblast-conditioned culture supernatant changed from spindle-like cells to round cells. This morphologic change was dependent upon the concentration of SU11274.

Next, we investigated the effect of SU11274 on the fibroblast-induced decrease in the expression of transgenes delivered by the adenoviral vector. SUIT-2 cells were cultured with or without SU11274 and/or fibroblast-conditioned supernatant for 48 h, then infected with Ad-GFP at 10 MOI. After 48 h the GFP-positive cells were counted using a fluorescent microscope. Figure 5(C) shows that a large number of SUIT-2 cells cultured with SU11274 (2.5 μM) and fibroblast-conditioned supernatant were GFP-positive (similar to control cells cultured without SU11274 and fibroblast-conditioned supernatant), whereas only a small number of cells cultured with fibroblast-conditioned supernatant were GFP-positive (Fig. 5D, < 0.05). This suggests that inhibition of the HGF/MET pathway improves the efficiency of gene delivery by adenoviral vectors, which was prevented by stromal cells.

Effect of MET inhibitors on fibroblast-induced changes in expression levels of CAR and αvβ5-integrin mRNA.  In this study, we found that the CAR, αvβ5-integrin, and clathrin mRNA expression was downregulated in pancreatic cancer cells cultured with fibroblast-conditioned supernatant. We next investigated the effect of SU11274 on the expression levels of CAR, αvβ5-integrin, and clathrin mRNA expression. Figure 5(E) shows that the expression of CAR, αvβ5-integrin, and clathrin mRNA was significantly higher in cells cultured with SU11274 and the supernatant from all fibroblast cultures than in cells cultured without SU11274 (< 0.05).

Effect of HGF neutralizing antibody on fibroblast-induced decrease in efficacy of adenoviral therapy.  To investigate whether removal of HGF from the fibroblast supernatants ameliorated the inhibitory effect of the supernatants on the efficiency of adenoviral gene delivery, we used an HGF neutralizing antibody. First, we investigated the effect of the neutralizing antibody on fibroblast-induced morphologic changes in pancreatic cancer cells. As shown in Figure 6(A), pancreatic cancer cells treated with both the HGF neutralizing antibody and fibroblast-conditioned culture supernatant changed from spindle-like cells to round cells when compared with cells treated with fibroblast-conditioned culture supernatant alone. Next, we investigated the effect of the HGF neutralizing antibody on the fibroblast-induced decrease in the expression of transgenes delivered by the adenoviral vector. SUIT-2 cells were cultured with or without the HGF neutralizing antibody (0.1 or 1.0 ng/mL) and/or fibroblast-conditioned supernatant for 48 h, then infected with Ad-LacZ at 10 MOI. The viral DNA content was quantified 48 h after infection. As shown in Figure 6(B), the viral DNA content of cells cultured with both fibroblast-conditioned supernatant and the HGF neutralizing antibody was significantly higher than that of cells cultured with fibroblast-conditioned supernatant alone. X-gal-positive cells were then counted. Figure 6(C) shows that a large number of SUIT-2 cells cultured with the HGF neutralizing antibody and fibroblast-conditioned supernatant were X-gal-positive (although fewer than in cells cultured without fibroblast-conditioned supernatant), whereas only a small number of cells cultured with fibroblast-conditioned supernatant were positive (Fig. 6D, < 0.05). These data suggest that HGF in the fibroblast supernatant influences gene delivery by adenoviral vectors.

Figure 6.

 Effect of hepatocyte growth factor (HGF) neutralizing antibody on the fibroblast-induced decrease in the efficacy of adenoviral therapy. (A) Effect of the HGF neutralizing antibody on fibroblast-induced morphologic changes. SUIT-2 pancreatic cancer cells were treated with or without (−) HGF neutralizing antibody and/or fibroblast-conditioned culture supernatant for 2 days. Magnification, ×200. (B) SUIT-2 cells were treated with or without the HGF neutralizing antibody and/or fibroblast-conditioned culture supernatant for 2 days, then infected with Ad-LacZ at 10 MOI. DNA was extracted 48 h later. Viral DNA content was quantified by real-time PCR and expressed as a ratio compared with cells cultured with fibroblast-conditioned supernatant. *< 0.05. (C) Photomicrographs of X-gal stained cells cultured with or without fibroblast-conditioned supernatant and/or HGF neutralizing antibody. Magnification, ×200. β-Galactosidase (β-gal) expression in Ad-LacZ treated cells cultured with or without fibroblast-conditioned supernatant and/or HGF neutralizing antibody for 48 h. Cells were infected with Ad-LacZ at 10 MOI and β-gal activity was assessed by X-gal staining 48 h after infection. (D) Number of β-gal-positive cells. Each value represents the mean ± SD of one or five independent fields. *< 0.05.

Discussion

In the present study, we found that: (i) viral DNA uptake and expression of genes delivered by adenoviral vectors into cells cultured with fibroblast-conditioned supernatant was significantly lower than that in untreated cells; (ii) the levels of CAR, αvβ5-integrin, and clathrin mRNA expression were significantly lower in cells cultured with fibroblast-conditioned supernatant than in controls; and (iii) SU11274, a specific inhibitor of MET, and an HGF neutralizing antibody improved adenoviral infection rates and increased the expression of transgenes delivered by the adenovirus.

Many in vitro and in vivo studies have shown the potential of adenovirus-based cancer gene therapy. However, in many clinical trials, the efficacy of adenoviral therapy was poor.(29,30) One of the reasons for the low efficacy of adenoviral therapy is inefficient spreading capacity, which limits the propagation of infection.(8) The causes of inefficient spreading capacity are overproduction of fibers, which mask adenoviral receptors,(31) the narrow spacing between tumor cells,(9) and fibrillar collagen within the extracellular matrix.(32) The present data suggest that tumor–stroma interactions are a major cause of this inefficient spreading capacity through indirect interactions. The direct physical interaction between stromal and cancer cells is also important. McKee et al.(32) reported that fibrillar collagen, which is produced by fibroblasts,(33) can regulate the initial distribution of viral vectors in certain fibrous tumors, and that pretreatment with proteases improved the efficiency of viral infection.(32,34) However, Lopez et al.(35) reported that stromal cells improved the oncolytic efficacy of conditionally replicative adenoviruses, which is inconsistent with our data. The condition of fibroblasts is very different under in vitro culture conditions and depends on several factors, including the passage number, the cell density, and the cultural condition. In this study, to avoid the effects of the concentration of FBS on the cancer cells and the efficiency of adenoviral therapy, we cultured fibroblasts in DMEM without FBS and used concentrated supernatant to adjust the FBS concentration. This difference in culture conditions may account for this discrepancy.

Two steps, attachment and internalization, are required for an adenovirus to enter a host cell.(21,36) During the first step, the viral fiber protein attaches to the CAR on the host cell surface. During the second step, adenoviral particles are rapidly internalized through the interaction between the penton base capsid protein and αvβ3- and αvβ5-integrins on the cell surface. Adenovirus internalization is clathrin mediated and involves the large GTPase dynamin.(21) Therefore, deficiency of CAR, αvβ3-integrin, αvβ5-integrin, dynamin, and/or clathrin will limit the gene transfer capacity of adenoviruses. In this study, CAR, αvβ5-integrin, and clathrin mRNA expression levels were significantly lower in cells cultured with fibroblast-conditioned supernatant than in cells cultured without fibroblast-conditioned supernatant. A decrease in the expression of these genes may be the reason for the lower gene transfer capacity of adenoviruses in pancreatic cancer cells cultured with fibroblast-conditioned supernatant. We previously reported that adenovirus-mediated gene expression in radio-resistant pancreatic cancer cells, in which αvβ3-integrin was highly upregulated (66–120-fold), was lower than that in radiosensitive cells.(37) Integrins play a critical role in gene transfer by adenovirus vectors.(20,38) Both αvβ3- and αvβ5-integrin promote adenovirus internalization into cells, but only αvβ5-integrin promotes penetration of the endosomal membrane and endosomal escape.(39,40) It is possible that extreme overexpression of αvβ3-integrin competitively blocks the formation of αvβ5 complexes, resulting in failed endosomal escape and decreased expression of transgenes delivered by the adenoviral vector in radio-resistant cells. Taken together, these data suggest that suitable levels of αvβ3-integrin expression are required for efficient gene transfer by adenovirus vectors.

A few reports show a relationship between adenoviral gene therapy and the tumor microenvironment. Bruning and Runnebaum(41) showed that transforming growth factor-β reduces CAR expression levels and adenoviral gene transfer into ovarian cancer cells. In the present study, we used fibroblast-conditioned culture supernatant that contained not only transforming growth factor-β, but also many other growth factors such as HGF, vascular endothelial growth factor, fibroblast growth factor, and cytokines.(42–44)

Hepatocyte growth factor is a 90-kDa multidomain glycoprotein first identified as a potent mitogen for adult rat hepatocytes in primary culture.(45) It is secreted by mesenchymal cells and has high affinity for the c-Met receptor, which is frequently overexpressed in PDAC. The interaction between HGF and the c-Met receptor increases the rate of proliferation, invasion, migration, and angiogenesis of cancer cells.(46–48) Previously, we also reported that the HGF/MET pathway plays a critical role in tumor–stroma interactions.(22–25) Therefore, we chose HGF from the many factors associated with tumor–stroma interactions and investigated the role of the HGF/MET pathway in our study. We found that inhibition of the HGF/MET pathway improved the efficiency of gene delivery and increased the levels of CAR, αvβ5-integrin, and clathrin mRNA expression, which were suppressed by stromal cells. However, it is possible that other growth factors, such as fibroblast growth factor, vascular endothelial growth factor, and stromal cell-derived factor-1 (SDF-1), also affected the efficiency of adenoviral gene therapy.

In summary, we showed that fibroblast-conditioned supernatant inhibited the efficiency of viral DNA uptake and the expression of genes delivered by adenoviral vectors. We also showed that inhibition of the HGF/MET pathway (and, therefore, inhibiting tumor–stroma interactions) improved the efficiency of gene delivery by adenoviral vectors. This suggests that the stromal cells in pancreatic cancer have a detrimental effect on adenoviral therapy. We conclude that control of tumor–stroma interactions may lead to improvements in adenoviral gene therapy for PDAC.

Acknowledgment

This study was supported in part by a Grant-in-Aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Disclosure Statement

The authors have no conflict of interest.

Ancillary