FAP-retargeted Ad5 enables in vivo gene delivery to stromal cells in the tumor microenvironment

Abstract

Fibroblast activation protein (FAP) is a cell surface serine protease that is highly expressed on reactive stromal fibroblasts, such as cancer-associated fibroblasts (CAFs), and generally absent in healthy adult tissues. FAP expression in the tumor stroma has been detected in more than 90% of all carcinomas, rendering CAFs excellent target cells for a tumor site-specific adenoviral delivery of cancer therapeutics. Here, we present a tropism-modified human adenovirus 5 (Ad5) vector that targets FAP through trivalent, designed ankyrin repeat protein-based retargeting adapters. We describe the development and validation of these adapters via cell-based screening assays and demonstrate adapter-mediated Ad5 retargeting to FAP⁺ fibroblasts in vitro and in vivo. We further show efficient in vivo delivery and in situ production of a therapeutic payload by CAFs in the tumor microenvironment (TME), resulting in attenuated tumor growth. We thus propose using our FAP-Ad5 vector to convert CAFs into a "biofactory," secreting encoded cancer therapeutics into the TME to enable a safe and effective cancer treatment.

Keywords: adenovirus vectors; cancer—gene therapy; cancer—targeted therapy; cell delivery; tumor microenvironment.

Graphical abstract

Figure 1

Generation and validation of DARPin-based hFAP-specific adenoviral retargeting adapters (A) Schematic representation of a bispecific trimeric DARPin adapter for adenoviral retargeting. The retargeting DARPin (orange) with specificity for a selected cell surface molecule (e.g., FAP) is fused via a long flexible linker to the knob-binding DARPin 1D3 (green) that in turn is fused via a short linker to the trimerizing protein SHP from lambdoid phage (yellow). The bispecific trimeric DARPin adapter forms a highly stable clamp around the fiber knob (red) to block natural cellular interactions and redirect adenoviral tropism to selected cells (e.g., FAP⁺ cells). (B) Flow cytometry analysis of hFAP expression of the parental HT1080 and HT1080hFAP cell line upon hFAP antibody staining. (C) Cell-based adapter binding assay on target and non-target cells. The purified “Top 9” adapters constructed with the selected hFAP-specific DARPins were titrated on hFAP⁺ HT1080hFAP and hFAP^– HT1080 cells in the concentration range of 0.1–100 nM. Binding was detected via flow cytometry by staining of the His-tagged adapter, and specific binding signals were determined as ΔMFI = MFI (HT1080hFAP cells) − MFI (HT1080 cells). The non-binding control adapter E3_5 was applied as a negative binding control. Bars represent specific binding signals of single point measurements. Representative data of two independent experiments are shown. MFI, mean fluorescent intensity. (D) Transduction of target and non-target cells by hFAP adapter-retargeted Ad5. Recombinant Ad5 encoding iRFP670 was pre-incubated with the “Top 9” hFAP adapters (colored filled bars) or the E3_5 blocking adapter (black empty bar) and tested for transduction of hFAP⁺ HT1080hFAP and hFAP^– HT1080 cells in comparison with the untargeted Ad5 (black filled bar) at a multiplicity of infection (MOI) of 20 plaque-forming units (PFU)/cell. Transduction was assessed from cellular expression of iRFP670 detected by flow cytometry. Dashed lines indicate cut-off levels above which functional and hFAP-specific adapters were identified. Bars represent mean transduction level of two biological replicates ± standard deviation (SD). Statistics: unpaired t test; ∗p < 0.05, ∗∗p < 0.005; p values are indicated for each sample with respect to the untargeted Ad5 for HT1080hFAP cells or with respect to the E3_5-Ad5 for HT1080 cells. Representative data of three independent experiments are shown. (E) Flow cytometry analysis for CAR expression of the HT1080 and HT1080hFAP cell line in comparison with the positive control HeLa cell line upon CAR antibody staining.

Figure 2

Selected hFAP adapters mediate adenoviral transduction of hFAP-expressing human fibroblasts Flow cytometry analysis for (A) hFAP expression and (B) CAR expression of the human fibroblast D551 cell line upon antibody staining. The HeLa cell line served as a positive control in the CAR expression analysis. (C) Transduction of human fibroblasts by hFAP adapter-retargeted Ad5. Recombinant Ad5 encoding iRFP670 was pre-incubated with the selected “Top 4” hFAP adapters or the E3_5 blocking adapter and analyzed at two different MOIs (PFU/cell) for transduction of hFAP⁺ D551 cells in comparison with the untargeted Ad5. Transduction levels were determined via cellular expression of iRFP670 detected by flow cytometry. Bars represent mean transduction level of two biological replicates ± SD. Statistics: unpaired t test; ∗p < 0.05, ∗∗p < 0.005; p values are indicated for each sample with respect to the untargeted Ad5. Representative data of two independent experiments are shown. (D) Incubation time-dependent transduction of human fibroblasts by hFAP adapter-retargeted and untargeted Ad5. Recombinant Ad5 encoding TdTomato was pre-incubated with the hFAP adapter no. 6 and analyzed for transduction of hFAP⁺ D551 cells in comparison with the untargeted Ad5 after 4, 24, and 48 h incubation time of the (retargeted) adenoviral vector with the cells at an MOI of 10 (PFU/cell). Transduction was measured via cellular expression of TdTomato detected by flow cytometry. Representative data of two biological replicates of one experiment are shown.

Figure 3

Retargeting of Ad5 to murine fibroblasts using a human/mouse cross-reactive FAP DARPin (A) Analysis of hFAP DARPins for binding to mFAP via ELISA. Purified DARPins, previously selected to be hFAP specific, were analyzed via ELISA for cross-reactivity to mFAP. The unselected, non-binding DARPin E3_5 was applied as a negative binding control. The maltose-binding protein (MBP)-specific DARPin off7 and recombinant MBP were applied as a technical positive binding control. The dashed line indicates a cut-off signal set on the negative binding control to select mFAP-binding DARPins. Bars represent mean transduction level of two biological replicates ± SD. Statistics: unpaired t test; ∗p < 0.05, ∗∗p < 0.005; p values are indicated for each sample with respect to the E3_5 control DARPin. (B) Flow cytometry analysis of mFAP expression of the NIH3T3 and NIH3T3mFAP cell line with mFAP antibody staining. (C) Cell-based DARPin binding assay on target and non-target cells. Purified DARPins selected as mFAP binders by ELISA were analyzed for binding on mFAP⁺ NIH3T3mFAP and mFAP⁻ NIH3T3 cells. Binding was detected by flow cytometry upon FLAG tag antibody staining of the FLAG-tagged DARPin. The unselected, non-binding DARPin E3_5 was applied as a negative binding control. Bars represent specific binding signal of single point measurements. Representative data at 1 μM DARPin concentration of a titration experiment are shown. MFI, mean fluorescent intensity. (D) Transduction of target and non-target cells by mFAP adapter-retargeted Ad5. Recombinant Ad5 encoding iRFP670 was pre-incubated with the mFAP adapter no. 6 or the E3_5 blocking adapter and tested for transduction of mFAP⁺ NIH3T3mFAP and mFAP^– NIH3T3 cells in comparison with the untargeted Ad5 at an MOI of 2 (PFU/cell). Transduction levels were determined via cellular expression of iRFP670 detected by flow cytometry. Bars represent mean transduction level of two biological replicates ± SD. Representative data of three independent experiments are shown. (E) Flow cytometry analysis for CAR expression of the NIH3T3 and NIH3T3mFAP cell line in comparison with the positive control A549 cell line upon CAR antibody staining.

Figure 4

Successful retargeting of Ad5 to FAP-expressing fibroblasts in vivo (A) HER2-overexpressing NCI-N87 tumor cells and GFP-labeled, mFAP-expressing NIH3T3mFAP fibroblast cells were co-injected subcutaneously into the flank of SCID/beige mice. After tumor establishment (200 mm³ tumor volume), mice were treated intratumorally with 3 × 10⁹ PFU FAP-retargeted or untargeted Ad5 encoding TdTomato. Three days post injection, tumors were harvested and analyzed by flow cytometry. Transduced cells were detected via TdTomato expression and further characterized by cell surface marker staining or GFP expression. Each data point represents a single mouse. Bars represent mean ± SD of five mice per group. Statistics: unpaired t test; ∗p < 0.05. Representative data of two independent experiments are shown. (B) Quantification of (A), indicating mean values of transduced cells. (C) Immunohistochemical analysis of (A) to investigate the cell specificity of FAP-retargeted Ad5. Representative immunofluorescence images of tumor tissues stained for HER2 (cyan) and counter-stained with DAPI (blue) for nuclei staining. FAP⁺ cells were detected via GFP expression (green), and cells transduced with Ad5 were detected via TdTomato expression (magenta). Scale bars 1 mm.

Figure 5

Efficient in vivo delivery of anti-cancer therapeutics using FAP-Ad5 (A) Growth analysis of HER2⁺ tumor xenografts upon Ad5-mediated treatment with trastuzumab (TZB) or with TZB (Herceptin) as a protein. HER2-overexpressing NCI-N87 tumor cells and NIH3T3mFAP cells were co-injected subcutaneously into the flank of SCID/beige mice for tumor establishment. At a tumor volume of 50 mm³, mice were treated intratumorally with 9 × 10⁸ PFU FAP-retargeted Ad5 encoding TZB (FAP-Ad5-TZB; n = 5), or one single dose of 200 μg of TZB (Herceptin) as a protein (n = 3), or three doses of 200 μg of TZB (Herceptin) as a protein (n = 3), or PBS (n = 5). Arrows indicate time points of injection for the corresponding treatment. Data points represent mean ± SD. Statistics: unpaired t test; ∗p < 0.05, ∗∗p < 0.005, ∗∗∗p < 0.0005; p values are indicated in black color for FAP-Ad5-TZB with respect to PBS and in blue color for FAP-Ad5-TZB with respect to TZB (Herceptin) 1×; results for TZB (Herceptin) 1× and TZB (Herceptin) 3× with respect to PBS were statistically non-significant (p > 0.05). (B) Tumor weights of harvested tumors from (A) 19 days post injection. Each data point represents a single mouse. Statistics: unpaired t test; ∗p < 0.05, ∗∗p < 0.005. (C) Detection of TZB within harvested tumors from (A) 19 days post injection. Representative immunofluorescence images of tumor tissues stained for TZB (cyan) and counter-stained with DAPI (blue) for nuclei staining. Scale bars 1 mm.