In this study, we tested for the first time the possibility to specifically deliver a therapeutic gene into pancreatic tumor cells using cell surface markers.
Gene transfer with lentiviruses is a very potent approach of gene therapy since viruses can transduce cells regardless of their stage in the cell cycle [
20]. Previous studies have reported the possibility to use lentivectors for gene delivery to pancreatic cancer cells but were performed with broad tropism envelopes [
7‐
9]. This approach might represent a risk of morbidity because of possible integrations of the viruses in normal cells (off target transductions). However, viral gene therapy still holds the promise of oncotargeting gene delivery sparing the normal cells, thanks to the modified Sindbis virus glycoprotein developed by Pariente et al. [
13] and used here. This approach offers the possibility to specifically pseudotype the viral particles. We found that modified Sindbis virus glycoprotein-packaged lentiviruses could efficiently transduce the pancreatic cell lines
in vitro. Three pancreatic cell surface antigens were tested for specific lentiviral gene transfer. The use of anti-PSCA yielded very poor gene transfer, regardless of the cell line. This result was disappointing since this protein was found highly expressed in pancreatic adenocarcinoma [
21] and anti-PSCA drove efficient gene transfer in prostate tumors [
13]. Interestingly however, CLDN18 and MUC4 achieved the best results in terms of efficiency and specificity
in vitro and could be suitable for specific gene transfer in pancreatic tumor cells. To test this hypothesis, human tumor cells were grafted under the skin of immune-deficient mice. Therapeutic agent administration by intra-tumoral injections is possible in pancreatic tumors since it has been performed in clinical trials with endoscopic ultrasound injections [
22]. Moreover, intra-tumoral injection of therapeutic oncotropic lentiviruses might be safer than intra-venous delivery to limit any systemic toxicity. Anti-MUC4-pseudotyped viruses carrying the firefly luciferase reporter gene, directly injected in the tumors yielded, luminescence signals in the tumors comparable to signals obtained with viruses packaged into the non-specific envelope containing the VSV glycoprotein, in two different cell lines.
In vivo, MUC4 was a more potent antigen than CLDN18 and even than HLA antigens in the CAPAN2 cells. Remarkably, luminescence appeared confined to the tumors since no signal was detected elsewhere in the mice, even when the detection mode was used with very low stringency. This observation was made previously when the gene transfer targeting system used here was tested by others [
13‐
15]. Noteworthy, pancreatic injections of viruses in tumor-free mice led to very interesting observations. First, the broad tropism virus transduced very efficiently the pancreas and leaked in other intra-abdominal sites, and even in the testis. By contrast the MUC4 oncotropic virus showed no detectable transduction at any site when injected in the pancreas, in the absence of tumors, even at the injection site. Taken together, this set of data suggests that the oncotropic lentiviral transductions appeared safer and more specific than VSV-G-driven transductions.
The results with CLDN18 oncotropic viruses were somewhat disappointing in the
in vivo transductions, considering that the fact that similar results were obtained for CLDN18 and MUC4 oncotropic viruses
in vitro (Figure
2B). We actually noticed strong signals one week after virus injections with anti-CLD18 antibodies, but signals had partially disappeared in CAPAN2 and almost totally disappeared in MIAPACA2 cells at the time of sacrifice, after two weeks (not shown). One possible explanation could be that fixation of anti-CLDN18 might interfere with the biological function of claudin 18 in cancer cells, probably leading to cell death.
Herpes thymidine kinase (TK) in combination with the pro-drug ganciclovir remains one of the most potent systems for anticancer gene therapy approach and has given promising results in a very recent phase I clinical trial with an adenoviral system [
6]. We evaluated the transfer of the TK gene by MUC4 oncotropic lentiviruses injected in orthotopically grafted human pancreatic tumor cells. Our experimental strategy was designed to do both the follow up of tumor growth (by fluorescence) and of the virus-infected tumor cells (by bioluminescence) in live animals. Importantly and as observed before, luminescence remained confined to the tumors when viruses were injected directly in the pancreas of the recipient mice. Moreover, GCV treatment resulted in luciferase signal loss and in slowing down of the tumor growth. It would be worth now to use this strategy to examine other PDAC-specific cell surface targets, and we feel that this study presents the proof of concept of oncotargeted molecular therapy of PDAC. There are many ways to improve the system. First, several targets (cell surface markers) could be used in concert as well as several rounds of virus injections could be performed to gain in efficiency. Second, once the markers have been validated, it is now possible to use vectors pseudotyped with engineered Sindbis virus glycoprotein bearing a covalent link with the antibodies. Indeed, fusion proteins could be produced [
23] rendering the transduction very potent even in an immune-competent background. Another attractive option would be the use of the biotine/avidine combination developed more recently [
24].