Oncolytic viruses have demonstrated in pre-clinical and clinical studies safety and a unique pleiotropic activity profile of tumor destruction. Yet, their delivery suffers from virus inactivation by blood components and sequestration to healthy tissues. Therefore, mesenchymal stromal cells (MSCs) have been applied as carrier cells for shielded virus delivery to tumors after ex vivo infection with oncolytic viruses. However, infection and particle production by MSCs have remained unsatisfying. Here, we report engineered oncolytic adenoviruses (OAds) for improved virus production and delivery by MSCs. OAds are uniquely amenable to molecular engineering, which has facilitated improved tumor cell destruction. But for MSC-mediated regimens, OAd engineering needs to achieve efficient infection and replication in both MSCs and tumor cells. We show that an Ad5/3 chimeric OAd capsid, containing the adenovirus serotype 3 cell-binding domain, strongly increases the entry into human bone marrow-derived MSCs and into established and primary pancreatic cancer cells. Further, we reveal that OAd with engineered post-entry functions-by deletion of the anti-apoptotic viral gene E1B19K or expression of the death ligand TRAIL-markedly increased virus titers released from MSCs, while MSC migration was not hampered. Finally, these virus modifications, or viral expression of FCU1 for local 5-FC prodrug activation, improved tumor cell killing implementing complementary cytotoxicity profiles in a panel of pancreatic cancer cell cultures. Together, our study establishes post-entry modification of OAd replication for improving virus delivery by carrier cells and suggests a panel of optimized OAds for future clinical development in personalized treatment of pancreatic cancer.

Pancreatic ductal adenocarcinoma (PDA) is one of the most aggressive malignancies and has poor therapeutic options. We evaluated improved oncolytic adenoviruses (OAds), in which the adenoviral gene E1B19K was deleted or a TRAIL transgene was inserted. Bone marrow mesenchymal stromal cells (MSCs) served as carriers for protected and tumor-specific virus transfers. The infection competence, tumor migration, and oncolysis were measured in cancer stem cell (CSC) models of primary and established tumor cells and in tumor xenografts. All OAds infected and lysed CSCs and prevented colony formation. MSCs migrated into PDA spheroids without impaired homing capacity. Xenotransplantation of non-infected PDA cells mixed with infected tumor cells strongly reduced the tumor volume and the expression of the proliferation marker Ki67 along with a necrotic morphology. Adenoviral capsid protein was detected in tumor xenograft tissue after intravenous injection of infected MSCs, but not in normal tissue, implying tumor-specific migration. Likewise, direct in vivo treatment correlated with a strongly reduced tumor volume, lower expression of Ki67 and CD24, and enhanced activity of caspase 3. These data demonstrate that the improved OAds induced efficient oncolysis with the OAd-TRAIL as most promising candidate for future clinical application.

Reovirus nonstructural protein 1s is implicated in cell cycle arrest at the G 2 /M boundary and induction of apoptosis. However, the contribution of 1s to these effects in an otherwise isogenic viral background has not been defined. To evaluate the role of 1s in cell cycle arrest and apoptosis, we used reverse genetics to generate a 1s-null reovirus. Following infection with wildtype virus, we observed an increase in the percentage of cells in G 2 /M, whereas the proportion of cells in G 2 /M following infection with the 1s-null mutant was unaffected. Similarly, we found that the wild-type virus induced substantially greater levels of apoptosis than the 1s-null mutant. These data indicate that 1s is required for both reovirus-induced cell cycle arrest and apoptosis. To define sequences in 1s that mediate these effects, we engineered viruses encoding C-terminal 1s truncations by introducing stop codons in the 1s open reading frame. We also generated viruses in which charged residues near the 1s amino terminus were replaced individually or as a cluster with nonpolar residues. Analysis of these mutants revealed that amino acids 1 to 59 and the amino-terminal basic cluster are required for induction of both cell cycle arrest and apoptosis. Remarkably, viruses that fail to induce cell cycle arrest and apoptosis also are attenuated in vivo. Thus, identical sequences in 1s are required for reovirus-induced cell cycle arrest, apoptosis, and pathogenesis. Collectively, these findings provide evidence that the 1s-mediated properties are genetically linked and suggest that these effects are mechanistically related.