IL-6 Signaling Blockade during CD40-Mediated Immune Activation Favors Antitumor Factors by Reducing TGF-β, Collagen Type I, and PD-L1/PD-1

The tumor microenvironment is important for the development and progression of cancer, and it also plays a role in the response to treatment or lack thereof. Pancreatic cancer is a challenging disease highly resistant to conventional cancer therapeutics. The tumor microenvironment is characterized by desmoplasia with stroma including fibroblasts, stellate cells, myeloid cells, and extracellular matrix (1). Less than 10–20% of the cells are the malignant tumor cells. The activated stellate cells produce a high amount of fibronectin and collagen type I and are, hence, the major contributor to the fibrosis seen in the lesions (2). Fibrosis increases the intralesion pressure, which is believed to reduce chemotherapy exposure. Indeed, high stromal activity, as measured by upregulated α-smooth muscle actin (αSMA), is correlated with a poor prognosis (3).

The cytokine IL-6 is upregulated in pancreatic cancer, and a high serum level of IL-6 correlates with a more-advanced disease and a poorer health status (4). In pancreatic cancer, type 2 tumor-associated macrophages in the tumor stroma are the main producers of IL-6, but other cell types also contribute, including stellate cells (5, 6). Signaling through the IL-6 pathway leads to activation of STAT3, which in turn elevates TGF-β1 expression. TGF-β1 can activate production of collagen type I and plays a central role in fibrosis of the pancreas (7). TGF-β1 is also involved in immunosuppression, another characteristic of pancreatic cancer, by inhibition of Th1 cells and expansion of T regulatory cells (8, 9). IL-6 is used in some dendritic cell (DC) activation protocols, but it can also interfere with the differentiation and maturation of DCs because it promotes STAT3 activation (10, 11), and it is therefore one of the factors to promote differentiation of myeloid-derived suppressor cells (MDSCs) (6). The complete role of IL-6 in tumor immunology is, therefore, obscure. Nevertheless, the presence of macrophages in the tumor and/or an increased level of circulating MDSCs correlate to a poorer prognosis (12, 13).

We previously demonstrated that immunostimulatory gene therapy using viruses transferring CD40L (AdCD40L) is a potent stimulator of the tumor microenvironment because of its capacity to activate DCs and tilt M2 to M1 macrophages. AdCD40L showed efficacy in mouse, dog, and human (14–17). In the current paper, we explored the possibility of combining CD40L-based immune activation with IL-6 pathway blockade using virus-mediated gene therapy. We constructed LOAd713, an oncolytic adenovirus serotype 5/35 chimera that carries a gene encoding a single chain fragment (scFv) against the IL-6R in combination with a gene encoding a trimerized membrane-bound isoleucine zipper (TMZ) human CD40L. In the current paper, the effect of LOAd713 infection of pancreatic cancer cells as well as the stellate cells and myeloid immune cells that constitute most of the tumor microenvironment were evaluated.

The pancreatic cell lines BxPc3, MiaPaCa2, PaCa3, and Panc01 were a kind gift from Dr. R. Heuchel (Karolinska Institute, Stockholm, Sweden). The cell origin was confirmed by short tandem repeat analysis by the Uppsala Genome Center, Uppsala, Sweden. BxPc3 were cultured in RPMI 1640 supplemented with 10% FBS and 1% penicillin–streptomycin. MiaPaCa2, PaCa3, and Panc01 were cultured in DMEM supplemented with 10% FBS and 1% penicillin–streptomycin. All reagents were purchased from Invitrogen, Carlsbad, CA. The HEK293 and A549 cells (American Type Culture Collection, Manassas, VA) were kept in the DMEM or RPMI medium, respectively, as described above, except for A549, in which 1% sodium pyruvate was added to the medium. Pancreatic stellate cells were purchased from 3H Biomedical (Uppsala, Sweden) and kept in 2% FBS, 1% SteC Growth Supplements, and 1% penicillin–streptomycin in stellate cell medium, all from 3H Biomedical.

Construction of LOAd adenoviral vectors LOAd(−) and LOAd700 has been described previously (17). LOAd713 contains TMZ-CD40L and an scFv against the IL-6R (scFv-aIL-6R). In short, LOAd713 was constructed by synthesis of a pUC57-Kan plasmid containing the sequence for the CMV promoter followed by scFv-aIL-6R, a T2A peptide, and then human TMZ-CD40L. The gene region was flanked by adenoviral regions containing 5′SspI and 3′ScaI sites (GenScript, Piscataway Township, NJ). After enzyme digestion of the plasmids with SspI and ScaI, the gene fragment was gel purified and inserted into LOAd(−) by homologous recombination. HEK293 cells were transfected for the first step of adenoviral production followed by expansion in A549 cells. Virus purification was made with cesium chloride centrifugation followed by dialysis using a Slide-A-Lyzer Cassette (Thermo Fisher Scientific, Waltham, MA). Viruses were suspended in a Tris buffer (Tris-HCl 10 mM [pH 7.9], with MgCl₂ 2 mM and sucrose 4%), and titer was determined as fluorescent forming units (ffu) per milliliter (18). Viral aliquots were stored at −80°C.

For detection of TMZ-CD40L and scFv-aIL-6R, HEK293 cells were transfected with 5 μg of plasmid containing only TMZ-CD40L, only scFv-aIL-6R, TMZ-CD40L in combination with scFv-aIL-6R, or with an empty (mock) plasmid. The scFv-aIL-6R used in this experiment had a c-myc tag used for detection. After 48 h, transfected cells and supernatant were harvested. The cells were incubated with 1% BSA in PBS for blocking of unspecific binding, followed by incubation with Abs for 15 min at room temperature. Abs used were anti-CD40L Brilliant Violet 421 (BV421) (clone 24-31) and mouse IgG1 κ BV421 (clone MOPC-21). After that, the cells were washed with 0.5% BSA in PBS plus 3 mM EDTA and suspended in 1% paraformaldehyde in PBS plus 3 mM EDTA before analysis in BD FACS Canto II (BD Bioscience, San Jose, CA). Data were analyzed in Flow Jo (Tree Star, Ashland, OR). The supernatants were added to wells coated with rIL-6R (Abbiotec, San Diego, CA) or CD19 as irrelevant control (Sino Biological, North Wales, PA). The myc tag was then detected by an anti–myc tag Ab (Abcam, Cambridge, U.K.) followed by a goat anti-rabbit IgG (Invitrogen) before TMB substrate (Millipore, Billerica, MA) was added. The absorbance was evaluated at 450 nm in an iMark Micro Plate Reader (Bio-Rad Laboratories, Hercules, CA).

Pancreatic cell lines were infected with of 25 ffu/cell of LOAd(−), LOAd700, or LOAd713 or left uninfected, and after 48 h, the cell phenotype was evaluated by flow cytometry as described above. Abs used were anti-CD40L BV421 (clone 24-31), anti-CD40 allophycocyanin (clone HB14), anti–IL-6R PE (clone UV4), mouse IgG1 κ BV421 (clone MOPC-21), mouse IgG1 κ allophycocyanin (clone MOPC-21), and mouse IgG1 κ PE (clone MOPC-21), all from BioLegend.

Pancreatic cell lines were infected with 100 ffu/cell of LOAd(−), LOAd700, or LOAd713 or left uninfected for 2 h in serum-free medium in a total volume of 100 μl. Postinfection, the infected cell suspensions were diluted to 100,000 cells/ml, and 100 μl in quadruplicates was added to a 96-well plate. After 48 and 72 h, the viability of the cells were measured by MTS CellTiter AQueous One Solution Cell Proliferation Assay kit (Promega, Madison, WI) according to the manufacturer’s instructions. The relative cell viability in percent was calculated as the absorbance for infected cells divided by the absorbance for uninfected cells multiplied by 100.

Panc01 cells were infected with 25 ffu/cell of LOAd(−), LOAd700, or LOAd713 or left uninfected and cultured for 48 h. Supernatants were harvested and added to fresh Panc01 cells. rIL-6 (20 ng/ml; BioLegend) was then added to the cells followed by 20 min incubation at 37°C. The level of STAT3-Y705 phosphorylation in the different groups was determined with flow cytometry according to the manufacturer’s protocol (BD Bioscience).

Human primary pancreatic stellate cells were cultured on Poly-L-Lysine–coated plates (3H Biomedical). 1 × 10⁵ cells per group were infected with 25 ffu/cell of LOAd(−), LOAd700, or LOAd713 or left uninfected for 48 h before harvest of cells to prepare total RNA (RNeasy Minikit; Qiagen, Hilden, Germany) and protein lysates. For protein extraction lysates, cells were lysed using RIPA lysis buffer (50 mM Tris-HCl [pH 7.4], 150 nM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 1 mM PMSF, and 10% Protease Inhibitor Cocktail from Thermo Fisher). The protein suspension was centrifuged to remove debris, and the lysates were then transferred to fresh tubes for storage in −20°C. Protein concentration was measured with Coomassie Plus Protein Assay Reagent (Thermo Fisher Scientific) prior to analysis using a 233-analyte custom multiplex array by service at Olink Proteomics AB, (Uppsala, Sweden). Relative values are expressed as linear normalized protein expression. Total RNA from cells was synthesized to cDNA with an iScript cDNA synthesis kit (Bio-Rad Laboratories). Gene expression was then determined by quantitative PCR using primers for GADPH (forward: 5′-GTCAAGGCTGAGAACGGGAA-3′; reverse: 5′-TCGCCCCACTTGATTTTGGA-3′) and αSMA (forward: 5′-CTGTTCCAGCCATCCTTCATC-3′; reverse: 5′-TCGCCCCACTTGATTTTGGA-3′), all purchased from Invitrogen. SYBR Green Supermix (Bio-Rad Laboratories) was used, and amplifications were done by Bio-Rad CFX96. Data analysis was performed using the CFX Manager Software (Bio-Rad Laboratories).

PBMCs (n = 2) or CD14⁺ enriched cells (n = 2; MACS beads; Miltenyi Biotec, Bergisch Gladbach, Germany) from healthy blood donors were cultured as previously described to promote MDSCs differentiation (19). In short, cells were cultured at a density of 0.5 × 10⁶ cells/ml in 10 ng/ml of IL-6 and 10 ng/ml of GM-CSF (Gentaur, Brussels, Belgium) for 7 d. Supernatants from uninfected, LOAd(−), LOAd700, or LOAd713-infected A549 cells were added to the 7-d culture at a concentration of 50% of total medium. After 7 d, the cells were harvested and analyzed by flow cytometry as described above. Abs used are as follows: anti-CD11b allophycocyanin (clone ICRF44), anti-CD11b Pe/Cy7 (clone ICRF44), anti-CD33 PE (clone WM53), anti-CD163 BV421 (clone GHI/61), mouse IgG1 κ BV421 (clone MOPC-21), and mouse IgG1 κ PE (clone MOPC-21), all form BioLegend.

CD14⁺ cells were sorted from healthy donor PBMCs (n = 5; CD14⁺ purity mean: 98.6%, ranging from 98% to 99%) and differentiated to immature CD1a⁺ DCs by culturing for 6 d with 150 ng/ml of GM-CSF and 50 ng/ml of IL-4 (Gentaur) (CD1a⁺ purity: 85.2%, ranging from 82% to 87%). The immature DCs were then infected with 50 ffu/ml of LOAd(−), LOAd700, or LOAd713 or left uninfected. The DCs were cultured for 48 h before being analyzed by flow cytometry as described above. Abs used are as follows: anti-CD1a allophycocyanin (clone HI149), anti-CD70 PE (clone 113-16), anti-CD83 PE (clone HB15e), anti-CD86 BV421 (clone IT2.2), anti-CD154 BV421 (clone 24-31), anti-HLADR PerCP (clone L243), anti–IL-6R PE (clone UV4), anti–IL-10R PE (clone 3F9), mouse IgG1 κ PE (clone MOPC-21), rat IgG2a κ PE (clone RTK2758), mouse IgG1 κ BV421 (clone MOPC-21), mouse IgG2b κ BV421 (clone MPC-11), and mouse IgG2a κ PerCP (clone MOPC-173), all from BioLegend. Supernatants were also collected and analyzed with three different techniques, namely a custom 233-analyte multiplex proteomic array by service (Olink Proteomics AB), luminex methodology (Milliplex MAP kit Human Th17 Magnetic Bead Panel HTH17MAG-14K, Millipore, Billerica, MA, USA), and for soluble IL (sIL)-6R by ELISA (Invitrogen).

Murine melanoma B16 cells (2 × 10⁵) expressing human CD46 (kind gift from Dr. Hemmi, University of Zurich) (20) was injected in syngeneic C57BL6 mice purchased from Taconic, Denmark. At day 5, the black tumor cells were visible under the skin, and treatment was initiated. A virus expressing murine TMZ-CD40L (mLOAd700) was injected s.c. in the tumor area (1 × 10⁹ infectious particles per mouse in 50 μl) with or without coinjection of a rat anti-mouse IL-6Rα Ab (0.5 mg/mouse in 50 μl; BioXCell, West Lebanon, NH). Control mice were treated with physiological NaCl. Treatments were given twice per week, six times in total.

Statistical calculations were performed using GraphPad Prism 6 (La Jolla, CA). Kruskal–Wallis (ANOVA) with Dunn multiple comparison tests were used when more than two groups of nonparametric data were analyzed. Data sets with several groups and time points were analyzed using a two-way ANOVA with Tukey multiple comparison test. Survival was analyzed by log-rank test.

Gene constructs expressing a TMZ-CD40L (17) and/or an scFv-aIL-6R were produced and transfected into HEK293 cells. As shown by flow cytometry, the transfected cells rapidly expressed TMZ-CD40L (Fig. 1A), whereas a c-myc tag enabled detection of the scFv-aIL-6R by ELISA (Fig. 1B). Pancreatic cancer cell lines BxPc3, MiaPaCa2, PaCa3, and Panc01 were infected with serotype 5/35 adenoviruses without transgenes [LOAd(−)], with TMZ-CD40L (LOAd700), or with a combination of scFv-aIL-6R and TMZ-CD40L (LOAd713). CD40L expression is shown in Fig. 1C. The pancreatic cancer cell lines were not positive for CD40 or IL-6R (Fig. 1D), and infection with LOAd713 did not alter this expression pattern. The oncolytic capability of LOAd713 was tested in the four pancreatic cancer cell lines (Fig. 1E–H). All LOAd viruses could induce tumor cell death within 48 h, and the addition of TMZ-CD40L and scFv-aIL-6R did not alter this capacity. LOAd713 showed a significant reduction of cell viability compared with uninfected cells at both 48 and 72 h in all cell lines tested. Hence, LOAd713 can efficiently expand in pancreatic cancer cells to induce oncolysis.

The IL-6R is often soluble binding to IL-6, thereby docking into a shared signaling molecule (gp130). Hence, the presence of IL-6R was tested on the cells and in the supernatant using the sensitive Olink Proteomic assay. In Fig. 2A, it is indeed shown that expression of IL-6R can be low on cells but found in the supernatant (Fig. 2A). Binding of IL-6 to IL-6R leads to phosphorylation of STAT3 at the Y705 position. To confirm the function of the scFv-aIL-6R, Panc01 and MiaPaCa2 cells were infected with LOAd(−), LOAd700, or LOAd713 or left uninfected. After 48 h, the supernatants with and without scFv-aIL-6R were harvested and added to fresh Panc01 and MiaPaCa2 cells stimulated with rIL-6. IL-6 triggered STAT3 phosphorylation in both Panc01 and MiaPaCa2 cells (Fig. 2B, 2C), but cells cultured with supernatant from LOAd713-infected cells that contained scFv-aIL-6R showed significantly reduced STAT3 phosphorylation. Hence, LOAd713 can participate to reduce IL-6–induced STAT3 activation in tumor cells.

Pancreatic stellate cells play a major role in the tumor biology of pancreatic cancer, and because of the high content in the lesions, intratumoral delivery of LOAd713 would likely lead to infection of stellate cells. Because our virus only infects human cells and the scFv-aIL-6R does not cross-react with murine IL-6R, we sought to elucidate the role of the IL-6R blockade using human stellate cells from primary cultures. Even if replication and oncolysis will not occur in normal cells, the LOAd713 virus can infect normal cells such as stellate cells and drive expression of its transgenes because the transgene expression is under control of a promiscuous CMV promoter. Hence, primary stellate cells were infected with LOAd viruses or left uninfected. The cells were analyzed by quantitative PCR to confirm that the cells were indeed stellate cells by detection of αSMA (Fig. 3A). Further, cell lysates prepared from the cultures were analyzed using Olink multiplex proteomics. Expression of CD40 was confirmed, and CD40L was detected in the LOAd700- and LOAd713-infected cells as expected (Fig. 3B). Further, both IL-6 and IL-6R were present in the cells. In cells infected with LOAd713, a higher IL-6 level was detected, but it did not reach significance. Interestingly, molecules driving the hostile tumor microenvironment and/or cell division including LAP-TGF-β1, collagen type 1, fibroblast growth factor 5 (FGF5), hepatocyte growth factor (HGF), and TNF-like weak inducer of apoptosis (TWEAK) were all significantly decreased by LOAd713 infection (21–24). Virus infection [LOAd(−)] increased vascular endothelial growth factor (VEGF), but this was not seen with the LOAd713 virus. Interestingly, the lymphocyte chemokines CXCL10 and CCL20 were increased by LOAd713. Hence, LOAd713 infection of stellate cells reduced the pathological phenotype of these cells while enhancing their capacity to alert the immune system.

MDSCs are one of the major suppressive immune cells present in the tumor microenvironment of pancreatic cancer (25). In vitro, MDSCs can be differentiated from human PBMCs, or CD14⁺ cells (e.g., monocytes), by addition of IL-6 and GM-CSF (6). We show in this study that MDSCs defined as CD11b^highCD33⁺ myeloid cells are indeed increased in cultures of PBMC (Fig. 4A) or monocytes (Fig. 4B) when IL-6 and GM-CSF are added to the cultures obtained from two different blood donors. To investigate if the addition of the scFv-aIL-6R in LOAd713 can block MDSC differentiation, the PBMCs and CD14⁺ cells stimulated with IL-6/GM-CSF were also subjected to supernatants from LOAd virus–infected and uninfected A549 lung cancer cells. These cells produce higher amounts of scFv-aIL-6R because they are more resistant than pancreatic cancer cell lines to immediate oncolysis. When supernatants from LOAd-infected A549 cells were added to the cultures, the percentage of CD11b^highCD33⁺ cells was reduced, regardless of the virus used, whereas supernatant from uninfected cells only modestly differed from IL-6/GM-CSF stimulation alone. The CD14⁺ cells seemed even less prone to differentiate into CD11b^highCD33⁺ cells when supernatant from LOAd700-infected cells was added. The CD14⁺ cells were also evaluated for the differentiation into macrophages defined as CD11b⁺CD163⁺. IL-6 and GM-CSF prevented differentiation into CD163⁺ macrophages compared with monocytes cultured in medium only. The addition of supernatants from LOAd700-infected cells completely disrupted differentiation of CD163 macrophages, but this was not seen when supernatants from uninfected, LOAd(−), or LOAd713-infected cells were added, suggesting that the tumor supernatant contains other molecules that counteracted the addition of IL-6/GM-CSF by promoting M2 macrophage differentiation (Fig. 4C). However, supernatant from LOAd700-infected tumor cells reduced M2 differentiation, which we have noted using CD40L gene therapy in animal models previously (26). These experiments suggest that LOAd viruses can affect the tumor cell microenvironment to reduce the differentiation of MDSCs, which could lead to a reduced immunosuppression in the tumor. However, simultaneous CD40 stimuli and blockade of IL-6 signaling seemed to allow differentiation of CD163 macrophages that also have been connected to immunosuppression.

The microenvironment of pancreatic cancer is known for its content of myeloid cells and lack of antitumor immunity. CD40L is a potent activator of myeloid cells, especially DCs that drive formation of tumor immunity. Because IL-6 blockade may prevent or accelerate DC maturation, we sought to investigate the ability of LOAd713 to activate immature DCs. Monocyte-derived immature DCs were infected with LOAd713, LOAd700, or LOAd(−) or left uninfected. Infection with LOAd(−) did not upregulate activation markers on DCs compared with uninfected DCs; fold change was ∼1 in all five donors. LOAd713 and LOAd700 infection led to a significant upregulation of the maturation markers CD83 and CD86, whereas CD70 showed a significant fold change in the LOAd713 group only (Fig. 5A). There was no difference in the expression of HLA-DR, IL-6R, and IL-10R between infected DCs regardless of the virus used. DC supernatants were analyzed for released cytokines (Fig. 5B). LOAd713- and LOAd700-infected DCs secreted significantly higher levels of immunostimulatory cytokines IL-6, IL-12(p70), IFN-γ, and TNF-α compared with uninfected DCs. The immunosuppressive cytokine IL-10 was only significantly upregulated in supernatants from DCs infected with LOAd(−). These data demonstrate that LOAd713 is a potent activator of myeloid cells, such as DCs, which is an important step to activate antitumor immunity. Note that even if the CD14⁺ monocyte purification was performed prior to DC maturation (mean: 98.6%, ranging from 98% to 99%), it cannot be excluded that other cell types can still be present in primary cell cultures, such as lymphocytes, that can participate to produce cytokines in these cultures.

To further characterize the DC profile, a 233-analyte multiplex proteomics array was used (Fig. 6). DCs infected with LOAd713 demonstrated an increased expression of CD40L and CD40. Infection with either LOAd713 or LOAd700 led to upregulation of many cytokines, costimulator factors, and chemokines, including IL-12, TNF, IFN-γ, 4-1BB, CXCL9, CXCL10, CXCL11, and CCL4 involved in antitumor immunity. As expected, activation of DCs also upregulated programed death-1 (PD-1) and programed death ligand-1 (PD-L1) as seen in the LOAd700-infected DCs. PD-L1 functions by binding to PD-1 expressed on activated T cells, which induces an anergic state with an inhibition of antitumor responses (27). Interestingly, IL-6R blockade by LOAd713 reduced both PD-1 and PD-L1 on the DCs. Ultimately, this may reduce regulatory signaling upon contact with T cells.

Animal models are notoriously difficult using the LOAd family of viruses because adenoviruses do not replicate in mouse cells. Hence, no oncolysis will occur. Further, the serotype 35 fiber on the LOAd virus enables infection of human cells only because it targets human CD46 with no cross-reactivity to animal CD46. Finally, neither the TMZ-CD40L transgene nor the scFv-aIL-6R cross-reacts with the murine counter receptors. Nevertheless, attempting to explore LOAd viruses in an in vivo model, we cloned a LOAd virus expressing a murine TMZ-CD40L: mLOAd700. The murine B16 melanoma cell line has been engineered to express human CD46 (20) and was used in syngeneic C57BL6 mice. Because of a lack of any commercially available hybridoma expressing antimurine IL-6R Ab for cloning of a murine scFv, the IL-6 blockade in vivo was attempted by administering a full-length anti–IL-6R Ab. The B16-hCD46 tumor cells grew rapidly in mice (Fig. 7A), but repeated (n = 6) s.c. injections of mLOAd700 at the tumor site significantly prolonged survival compared with control (p = 0.0145). However, combination with full-length Ab targeting murine IL-6R hampered the effect noted by mLOAd700 alone, which can be due to Ab-dependent cellular cytotoxicity of IL-6R–positive cells. The day after the third treatment, five mice per group were sacrificed, and tumors were analyzed for immune cell infiltration (Fig. 7B). mLOAd700-treated tumors had an increase of T cells. They were active as demonstrated by increased PD-1. Interestingly, addition of IL-6R blockade tended to reduce PD-1 in two mice. Further, mLOAd700 increased APCs (CD11b⁺MHCII⁺) in the tumor. These cells had increased PD-L1, but the addition of IL-6R blockade showed reduced PD-L1 on these cells. Finally, mLOAd700-treated tumors showed decreased suppressive myeloid cells (CD11b⁺MHCII⁻) compared with mice treated with IL-6R blockade alone.

Pancreatic cancer is a devastating disease with rapid progression and high mortality. Treatment options are few and have a minor impact on survival. Hence, new therapeutic strategies are warranted. The microenvironment of pancreatic cancer is characterized by its desmoplastic stroma, which contributes to tumor progression, metastasis, and treatment resistance (28). One major contributor to the desmoplastic microenvironment is IL-6. Upon binding to the IL-6R, STAT3 is phosphorylated and translocated into the nucleus where it binds to STAT-responsive elements that control genes involved in proliferation, survival, angiogenesis, and immunosuppression (29–32). For example, STAT3 induces the expression of the immunosuppressive factors TGF-β1 and IL-10 at the same time it reduces the production of proinflammatory cytokines and chemokines (29). The immunosuppressive nature of the tumor microenvironment inhibits the infiltration of effector T cells to the cancer cells, classifying pancreatic cancer as a “cold” tumor (33, 34). The few infiltrating T cells may explain the lack of effect of checkpoint blockade Abs. Such Abs aim to block the negative signals that inhibit effector T cells, and thus far they include Abs targeting CTLA-4, PD-1, and PD-L1 (27). In a phase IIa clinical trial evaluating αCTLA-4 (ipilimumab), no effect on survival could be seen in pancreatic cancer, and in another trial using αPD-L1 (BMS-936559), no response was noted in the 14 pancreatic cancer patients treated (35, 36). In contrast, addition of the immune activator αCD40 to αPD-L1 treatment in a syngeneic orthotropic mouse model of pancreatic cancer increased the overall survival and the antitumor immunity (37). Hence, stimulation of CD40 may support the conversion of a cold tumor to an immunologically “hot” tumor, which enables checkpoint Ab therapy.

A novel therapy would benefit from simultaneously targeting the IL-6 signaling pathway and stimulating the tumor microenvironment to promote antitumor immune responses via CD40. Lau et al. (38) reported that a combination of an immune stimulator with IL-6 blockade increases apoptosis and decreases metastasis in a lung cancer model, and Caetano et al. (39) have shown that IL-6 blockade inhibits progression of lung cancer and can change the tumor microenvironment toward an antitumor phenotype. In this article, we report the design of a novel oncolytic adenovirus, LOAd713, for the treatment of pancreatic cancer. Oncolytic viruses have unique features for the treatment of cancer with their capability to infect and selectively replicate in cancer cells, leading to immunogenic cell death and activation of the immune system (40). Adenoviruses can also be armed with therapeutic genes to further promote antitumor responses. We have previously shown that CD40L gene therapy with adenoviruses can convert cold tumors to hot tumors with a high infiltration of T cells (15, 17, 41, 42). To further improve therapeutic efficacy, the LOAd713 virus carries two transgenes, an Ab scFv binding the IL-6R and TMZ-CD40L. Because of the facts that LOAd viruses only infect human CD46⁺ cells, that human CD40L does not cross-react with murine CD40, and that our scFv-aIL-6R cannot bind murine IL-6R, most experiments in this study needed investigation in human in vitro systems (17, 43).

Infection of pancreatic cancer cell lines with LOAd713 led to a reduction of cell viability due to oncolysis. The oncolytic capability of LOAd713 was similar to that of the control virus without any transgenes, showing that addition of the two transgenes did not affect the function of the oncolytic virus. Addition of scFv-aIL-6R present in the supernatant of LOAd713-infected cells to IL-6–stimulated Panc01 or MiaPaCa2 cells reduced STAT3 phosphorylation, showing that the scFv-aIL-6R led to an IL-6/IL-6R pathway blockade.

A major player in the pathogenesis of pancreatic cancer is the stellate cell. These myofibroblast-like cells produce factors associated to proliferation, inflammation, extracellular matrix remodeling, cell motility, and invasion (44). In our analysis, the stellate cells expressed CD40 as well as both IL-6 and IL-6R. Stellate cells can be infected with LOAd viruses and express the transgenes, but because of a lack of hyperphosphorylated retinoblastoma in normal cells, they will not enable virus replication. Indeed, stellate cells infected with LOAd713 were viable and showed a significant reduction of the immunosuppressive factor TGF-β. TGF-β can suppress T lymphocytes directly by inhibiting activation and proliferation and indirectly through differentiation of T regulatory cells (8, 9). TGF-β also plays a role in the fibrosis of the pancreas by regulating the production of collagenase and metalloproteinases in stellate cells (7). In line with reduced TGF-β in our experiment, collagenase type I was also significantly reduced. Further, LOAd713-infected stellate cells produced significantly reduced levels of FGF5, HGF, and TWEAK, factors that are connected to tumor progression and suppression of antitumor responses (23, 24, 45–48). Nevertheless, the overall response to LOAd713 infection of stellate cells seems beneficial because reduced production of tumor-promoting molecules and desmoplasia may be important factors to enhance therapeutic efficacy. Further, LOAd713-infected stellate cells increased expression of the chemokines CXCL10 and CCL20 that support T lymphocyte infiltration.

Pancreatic cancer is characterized by the infiltration of immune-suppressive cells, especially myeloid suppressors (25). IL-6 production by stellate cells is connected to the differentiation of MDSCs through the activation of STAT3 (6). When PBMCs were cultured with IL-6/GM-CSF, they differentiated toward a MDSC-like phenotype. To determine if virus infection of tumor cells could affect myeloid differentiation, supernatants from LOAd virus–infected, or uninfected, tumor cells were added to the PBMC cultures. The differentiation of MDSCs was blocked in all LOAd-virus groups regardless of the presence of transgenes. Hence, the differentiation blockade was connected to the overall inflammatory response to the adenoviral backbone and not to the transgenes. If CD14⁺ cells were used instead of crude PBMCs, a similar effect was seen. However, addition of supernatant from LOAd700-infected cells had the greatest capacity to block MDSC differentiation. Hence, CD40 stimulation is an important factor to reduce MDSC differentiation, which is in line with our previous published results in mice (26). Simultaneous IL-6/IL-6R blockade may revert some of that effect.

In our previous work, TMZ-CD40L potently activated DCs, and, in turn, TMZ-CD40L–activated DCs promoted expansion of Ag-specific T cells (17). IL-6, as mentioned before, can play a role in immunosuppression, but the cytokine is also linked to enhanced maturation of DCs and had been used in mixtures for the ex vivo maturation of DCs (49). In this article, we show that the addition of the scFv-aIL-6R to the virus and therefore the blocking of IL-6 signaling did not hamper the capacity of TMZ-CD40L to promote DC activation. Activated DCs upregulate not only costimulatory molecules but also both PD-L1 and PD-1. Interestingly, LOAd713-infected DCs reduced the levels of both PD-L1 and PD-1. Spary et al. (50) have shown that DCs can upregulate PD-L1 in an IL-6– and STAT3-dependent manner. Hence, IL-6/IL-6R blockade by the scFv-IL-6R in LOAd713-infected DCs may explain this finding. LOAd713-infected DCs also had lower levels of sIL-6R when measured by ELISA compared with LOAd700-infected cells. It is not clear if this is due to a biological effect of the scFv-aIL-6R or to a steric interference to the detection Ab. In this study, we have used DCs differentiated from monocytes obtained from five blood donors. There is some variation of the LOAd virus responses among the different blood donors, which does not seem to correlate to infectivity, but may be due to different base levels of biomarkers on DCs and also to the genetic/proteomic makeup variations among individuals. However, the pattern of which biomarkers are upregulated on DCs seems similar even if the magnitude can be different. It may be of great interest to investigate patient DC responsiveness to LOAd virus in correlation to treatment responses in later clinical trials.

There are no available animal models that can be used to fully evaluate the LOAd713 virus in vivo. Adenoviruses do not replicate in mouse cells but can show some oncolysis in Syrian hamster. Nevertheless, the LOAd family of viruses has a serotype 35 fiber that targets the virus infection to human cells expressing CD46. This receptor is not similar across species, but the human CD46 receptor can be cloned into murine cells such as B16 melanoma. Hence, we used B16-hCD46 cells in syngeneic mice and treated them with a LOAd virus expressing a murine version of TMZ-CD40L. mLOAd700 increased T cell infiltration in the treated tumors and reduced growth rate, thereby increasing the survival rates of the mice compared with controls. Unfortunately, there are no commercially available hybridomas that can be used to clone an anti-mouse IL-6R scFv, but we attempted to block IL-6 signaling by injecting a full-length Ab targeting the murine IL-6R. However, combining mLOAd700 with this full-length Ab showed worse survival compared with mLOAd700 alone. This can be due to Ab-dependent cellular cytotoxicity causing death of cells expressing IL-6R because full-length Abs can be used for depleting cells as well as for signaling blockade. IL-6 in the tumor microenvironment is driving tumor growth and causing fibrosis, but systemically it may have a beneficial effect on the immune system. s.c. injections will unfortunately also spread systemically because the tumors are too small to contain the injection volume. Our model can only show convincing benefits of TMZ-CD40L gene transfer via LOAd virus, and it cannot reveal if local IL-6R blockade is beneficial or not.

In conclusion, LOAd713 is a genetically engineered oncolytic adenovirus that can induce tumor cell oncolysis. Further, LOAd713 can target the tumor microenvironment by activating immune cells and reducing desmoplastic-inducing proteins in pancreatic stellate cells.

We thank Dr. Silvio Hemmi at the University of Zurich for kindly sharing B16-hCD46 cells.

A.L. is the chief executive officer and a board member of Lokon Pharma AB and holds a royalty agreement and a contract research agreement with Lokon Pharma AB at her adjunct position at Uppsala University. R.A. has a contract research agreement with Lokon Pharma AB. The other authors have no conflicts of interest to disclose. A patent has been filed in regard to the LOAd viruses presented in this article (WO2015155174).