Sendai virus is robust and consistent in delivering genes into human pancreatic cancer cells

Abstract

Background: Pancreatic ductal adenocarcinoma (PDAC) is a highly intratumorally heterogeneous disease that includes several subtypes and is highly plastic. Effective gene delivery to all PDAC cells is essential for modulating gene expression and identifying potential gene-based therapeutic targets in PDAC. Most current gene delivery systems for pancreatic cells are optimized for islet or acinar cells. Lentiviral vectors are the current main gene delivery vectors for PDAC, but their transduction efficiencies vary depending on pancreatic cell type, and are especially poor for the classical subtype of PDAC cells from both primary tumors and cell lines.

Methods: We systemically compare transduction efficiencies of glycoprotein G of vesicular stomatitis virus (VSV-G)-pseudotyped lentiviral and Sendai viral vectors in human normal pancreatic ductal and PDAC cells.

Results: We find that the Sendai viral vector gives the most robust gene delivery efficiency regardless of PDAC cell type. Therefore, we propose using Sendai viral vectors to transduce ectopic genes into PDAC cells.

Keywords: Gene delivery; Pancreatic cancer; Sendai virus (SeV).

Fig. 1

The heterogeneous susceptibility to LV transduction among human pancreatic cancer epithelial cells. (A) The schematic diagram for the transduction of primary human PDAC with LV-GFP. (B–C) GFP expression in epithelial cells derived from freshly isolated human PDAC cells. MUC1 is a marker for PDAC. While most larger epithelial cells express GFP (B), smaller ductal epithelial cells barely express GFP (C).

Fig. 2

Transduction of human PDAC cells with LV-GFP. (A–B) Effect of LV-GFP transduction on the cell viability of PDAC cell lines (MIA PaCa-2, BxPC3, Panc-1, SW1990 (basal-like subtypes, circles) and CFPAC-1, Capan-1, and HPAF-II (classical subtypes, squares)) (A) and PDX-derived primary PDAC cells (ST-7599, ST-7270, and ST-14490) (B). Human normal fibroblasts (BJ6) and HPDE cells (H6c7) were used as controls. The percentage of live cells was calculated using D Horizon™ Fixable Viability Stain 660 by flow cytometry. Statistical significance was computed through linear regression model coefficients (p > 0.05, Table S2). The asterisks in the figure refer to the P-value of linear regression model coefficients. * - P ≤ 0.05; ** - P ≤ 0.01; *** - P ≤ 0.001; **** - P ≤ 0.0001. (C–D) Transduction efficiency of LV-GFP in PDAC cell lines (MIA PaCa-2, BxPC3, Panc-1, SW1990 (basal-like subtypes, circles) and CFPAC-1, Capan-1, and HPAF-II (classical subtypes, squares)) (C) and PDX-derived primary PDAC cells (ST-7599, ST-7270, and ST-14490) (D). BJ6 and H6c7 cells were used as controls. Functional infectious units per cell (IFU/cell) were calculated based on LV GFP titer in human fibrosarcoma HT1080 cells (Table S1). The dependence of the fraction of GFP-expressing cells to the number of viral IFU/cell were fitted with a second-degree polynomial regression model. Statistical significance was computed through ANOVA to test whether there were any significant interactions of categorical variables (cell types) in the regression models estimating the relationships of the percentage of GFP-expressing cells to the number of viral IFU/cell used (Table S3 for p-value). The asterisks in the figure refer to the P-value of the ANOVA test. * - P ≤ 0.05; ** - P ≤ 0.01; *** - P ≤ 0.001; **** - P ≤ 0.0001. Each experiment was repeated at least three times independently (n ≥ 3). All data are represented as mean ± SD unless specified. SD bars can be smaller than the size of the symbols.

Fig. 3

Transduction of human PDAC cells with SeV-GFP. (A–B) Effect on SeV-GFP transduction on the cell viability of PDAC cell lines (MIA PaCa-2, BxPC3, Panc-1, CFPAC-1, SW1990, Capan-1, and HPAF-II) (A) and PDX-derived primary PDAC cells (ST-7599, ST-7270, and ST-14490) (B). BJ6 fibroblast and H6c7 HPDE cells were used as normal controls. The percentage of live cells was calculated using D Horizon™ Fixable Viability Stain 660 by flow cytometry. Statistical significance was computed through linear regression model coefficients (Table S2 for p-values). (C–D) Transduction efficiency of SeV-GFP in PDAC cell lines (MIA Paca-2, BxPC3, Panc-1, CFPAC-1, SW1990, Capan-1, and HPAF-II) (C) and PDX-derived primary PDAC cells (ST-7599, ST-7270, and ST-14490) (D). As controls, BJ6 and H6c7 cells were used. Functional infectious units per cell (IFU/cell) were calculated based on SeV-GFP titer in LLC-MK2 cells (Table S1). The dependence of the fraction of GFP-expressing cells to the number of viral IFU/cell was fitted with a second-degree polynomial regression model. Statistical significance was computed through ANOVA to test whether there are any significant interactions of categorical variables (cell types) in the regression models estimating the relationships of the percentage of GFP-expressing cells to the number of viral IFU/cell used (Table S3 for p-value). The asterisks in the figure refer to the P-value of the ANOVA test. * - P ≤ 0.05; ** - P ≤ 0.01; *** - P ≤ 0.001; **** - P ≤ 0.0001. Each experiment was repeated at least three times independently (n ≥ 3). All data are represented as mean ± SD unless specified. SD bar can be smaller than the symbol's size.

Fig. 4

Comparison of transduction efficiencies between LV-GFP and SeV-GFP. Transduction efficiencies are compared in control cell lines (HT1080, BJ6, and H6c7) (A), classical subtype PDAC lines (CFPAC-1, Capan-1, and HPAF-II) (B), basal-like subtype PDAC lines (MIA Paca-2, BxPC3, Panc-1, and SW1990) (C), and PDX-derived primary PDAC cells (ST-7599, ST-7270, and ST-14490) (D). The IFU/cell is calculated based on LV-GFP and SeV-GFP titer in corresponding cells. Each experiment was repeated at least three times independently (n ≥ 3). A second-degree polynomial regression model fitted the dependence of the fraction of GFP-expressing cells on the number of viral IFU/cell used. Statistical significance was computed through ANOVA to test whether there are any significant interactions of categorical variables (viral vector type) in the regression models estimating the relationships of the percentage of GFP expressing cells to the number of viral IFU/cell used (Table S4 for p-value). The asterisks in the figure refer to the P-value of the ANOVA test. * - P ≤ 0.05; ** - P ≤ 0.01; *** - P ≤ 0.001; **** - P ≤ 0.0001. All data are represented as mean ± SD unless specified. An SD bar can be smaller than the symbol's size.

Fig. 5

Relative transduction efficiencies of LV-GFP and SeV-GFP in PDAC cell lines and PDX-derived primary PDAC cells. The relative transduction efficiencies of LV and SeV vectors in PDAC cells (cell lines or primary cells) were calculated by normalizing titers obtained from PDAC cells to the titers obtained on the control H6c7 HPDE cells (n = 3). Statistical significance was computed with the nonparametric Mann-Whitney U test. The relative transduction efficiency of SeV-GFP was significantly higher than that of LV-GFP across all tested PDAC cells (Mann Whitney U test, p<0.05). The relative transduction efficiency of LV-GFP was significantly lower in the classical subtype than in the basal-like subtype of PDAC (Mann Whitney U test, p = 0.04). In contrast, there was no difference in the relative transduction efficiencies of SeV-GFP between classical and basal-like subtypes of PDAC cells (Mann Whitney U test, p = 1). Basal-like subtype PDAC lines (MIA Paca-2, BxPC3, Panc-1, and SW1990); classical subtype PDAC lines (CFPAC-1; Capan-1; HPAF-II; Primary PDAC cell (ST-7599; ST-7270; ST-14490); foreskin fibroblast BJ6; HPDE cell (H6c7). Since the presented data are not the result of direct measurements and do not satisfy all propagation of error requirements [81], the standard deviations are not shown.