Off-the-Shelf Prostate Stem Cell Antigen-Directed Chimeric Antigen Receptor Natural Killer Cell Therapy to Treat Pancreatic Cancer

Abstract

Background & aims: Pancreatic cancer (PC) is the third leading cause of cancer-related death with a 5-year survival rate of approximately 10%. It typically presents as a late-stage incurable cancer and chemotherapy provides modest benefit. Here, we demonstrate the feasibility, safety, and potency of a novel human natural killer (NK) cell-based immunotherapy to treat PC.

Methods: The expression of prostate stem cell antigen (PSCA) was evaluated in primary PC at messenger RNA and protein levels. The processes of retroviral transduction, expansion, activation, and cryopreservation of primary human NK cells obtained from umbilical cord blood were optimized, allowing us to develop frozen, off-the-shelf, allogeneic PSCA chimeric antigen receptor (CAR) NK cells. The safety and efficacy of PSCA CAR NK cells also expressing soluble (s) interleukin 15 (PSCA CAR_s15 NK cells) were evaluated in vitro and in vivo.

Results: PSCA was elevated in primary human PC compared with the adjacent or other normal tissues. PSCA CAR_s15 NK cells displayed significant tumor-suppressive effects against PSCA(+) PC in vitro before and after 1 cycle of freeze-thaw. The viability of frozen PSCA CAR_s15 NK cells persisted more than 90 days in vivo after their last infusion and significantly prolonged the survival of mice engrafted with human PC.

Conclusions: PSCA CAR_s15 NK cells showed therapeutic efficacy in human metastatic PC models without signs of systematic toxicity, providing a strong rationale to support clinical development.

Keywords: Immunotherapy; Innate Immunity; NK Cell; Pancreatic Cancer; Pancreatic Ductal Adenocarcinoma.

Figure 1.. Assessment of PSCA as a therapeutic target for pancreatic cancer

(A) PSCA expression in pancreatic tumors normalized to expression in normal adjacent tissues across 7 independent studies^–. The blue color and (−) sign indicate downregulation while red color and (+) sign indicate upregulation in tumors relative to the normal tissues. The number and color scale represent the ranking in percentage of the indicated gene in each analysis. P-value was calculated by median-rank analysis. (B) PSCA mRNA expression in pancreatic tumors as determined by histological grade (Grade 1 n=31; Grade 2 n=97; Grade 3 n=50). Statistical analysis was performed by one-way ANOVA. (C) The overall survival of PC patients was estimated based on the expression level of PSCA (top 33% versus bottom 33%, n=57 each). P-values were determined by the log-rank test. (D) Representative images of immunohistochemistry (IHC) of PSCA protein expression from primary pancreatic tumors or adjacent benign normal tissues. IHC images were taken under 200X and the scale bar is 100 pixels. (E) IHC of PSCA expression in normal benign tissues as indicated. (F) Schematic of the clinical grade vectors. (G) The expression of tEGFR in the engineered and control NK cells after enrichment for the CD56(+)CD3(−) population, as measured by flow cytometry.

Figure 2.. The PSCA CAR_s15 vector enhances IFN-γ secretion and tumor lytic functions of NK cells in vitro

(A) The median of fluorescence intensity (MFI) and the percentage of PSCA(+) cells in the indicated PC cell lines. (B) IFN-γ levels in the supernatants of co-cultured engineered NK cells and pancreatic tumor cells, as measured by enzyme-linked immunosorbent assay (ELISA). Statistical analysis was performed by one-way ANOVA. (C, D) Cytotoxicity of engineered NK cells against (C) PSCA(+) Capan-1 or (D) PSCA(–) PANC-1 tumor cells, at an effector: target (E:T) ratio of 1:3 measured after 48 hours of co-culture. (E, F) Real-time cell analysis (RTCA) of the cytotoxicity of engineered NK cells against (E) PSCA(+) Capan-1 or (F) PSCA(–) PANC-1 tumor cells with an E:T ratio of 1:3 and presented as the growth index of the residual cancer cells. The tumor only control group contained no NK cells. s15 = s15 NK cells; PSCA = PSCA CAR NK cells; PSCA-s15 = PSCA CAR_s15 NK cells.

Figure 3.. The PSCA CAR_s15 vector increases NK cell functional markers without decreasing the native activation receptors

(A) Flow cytometry analysis of NKp30, NKp44, NKp46, NKG2D, and CD16 on the CD 56(+) s15 or PSCA CAR_s15 NK cells stimulated with Capan-1 cells with an E:T cell ratio of 1:3 measured after 24 hours of co-culture. (B) Flow cytometry analysis of intracellular TNF-α, IFNγ, and CD107a on engineered CD56(+) NK cells without or with stimulation by Capan-1 cells. (C) Quantification of the results shown in panel B (n=3). P-values were determined by the t-test.

Figure 4.. PSCA-directed CAR NK cells retain functional properties after a freeze-thaw cycle

(A) Cell counts of previously frozen engineered NK cells (n=3/group) over time assessed by the MUSE cell analyzer post thaw. Cells were kept in the freezing buffer at room temperature once they were thawed. (B) Expression of the engineered constructs (assessed by measuring surface expression of tEGFR) before and after a freeze-thaw cycle (n=4). The cells shown in (A) and (B) were cryopreserved for two weeks prior to the assessment. (C) Cytotoxicity of fresh or NK cells derived from a same donor and frozen for 6-month, at an E:T ratio of 1:3 using the RTCA assay. (D) Comparison of NK cell distribution in mice after i.p. or i.v. delivery by flow cytometry. One dose of 8×10⁶ frozen s15 NK cells was thawed then delivered to the immunocompromised mice and evaluated after 1 week. Cells positive for human CD45 and human CD56 were considered NK cells.

Figure 5.. PSCA CAR_s15 NK cells suppress PC in a metastatic mouse model

(A) A schematic of PSCA-directed treatment with PSCA CAR_s15 NK cells in a human metastatic PC model created by injecting Capan-1_luc cells into NSG mice. The figure was generated at BioRender (https://biorender.com/). (B) The efficiency of NK cell transduction used for in vivo experiments as measured by surface density expression of tEGFR. NT=non-transduced NK. (C) Time-lapse luciferase imaging of the metastatic PC mouse model after the indicated treatments (WK=week). (D) Quantification of the bioluminescence images from panel C up to day 32. Statistical analysis was performed using linear mixed models. (E) Survival analysis of mice in the indicated groups from panel C as analyzed by the log-rank test. (F) Representative images of pancreas (T=tumor tissue; P=pancreatic tissue) and liver in the PC mouse model with the indicated treatments 50 days post initial treatment. (G, H) Flow cytometric analyses of (G) tumor cells (ZsGreen, detected in the GFP channel) and (H) NK cells (human CD45(+)/CD56(+)) from pancreas and liver. Both tumor and NK cells were gated on the singlets and living (SYTOX-) lymphocyte gate.

Figure 6.. Absence of tumor cells after PSCA-directed CAR NK cell treatment.

Representative images from (A) anti-human PSCA immunohistochemistry (IHC) (B) hematoxylin and eosin (H&E) staining, and (C) anti-human MUC1 IHC images of the indicated tissues from the metastatic PC mouse model after the indicated treatments under the indicated magnification. Scale bar is 100 pixels.

Figure 7.. EGFR functions as a safety switch for PSCA CAR NK cells in vivo.

(A) Schematic of the two vectors used to transduce NK cells. (B) The purity of PSCA CAR_s15 NK cells co-expressing luciferase_ZsGreen, as determined by flow cytometry for tEGFR and ZsGreen. (C) Time-lapse luciferase imaging of PSCA CAR_s15 NK cells after cetuximab or vehicle (saline) treatment (n=3/group). (D) Quantification of the bioluminescence shown in panel C. Statistical analysis was performed using linear mixed models. (E) Persistence of infused NK cells in the indicated organs after the indicated treatments, as shown by flow cytometry for ZsGreen and human CD56. (F) Representative images of IHC for human NKp46 in the indicated organs after the indicated treatments. Infused NK cells are indicated by arrows.