Combination therapy of DKK1 inhibition and NKG2D chimeric antigen receptor T cells for the treatment of gastric cancer

Abstract

Despite the successful application of chimeric antigen receptor (CAR)‐T cell therapy in hematological malignancies, the treatment efficacy in solid tumors remains unsatisfactory, largely due to the highly immunosuppressive tumor microenvironment and low density of specific tumor antigens. Natural killer group 2 member D (NKG2D) CAR‐T cells have shown promising treatment effects on several cancers such as lymphoma and multiple myeloma. However, the application and efficacy of NKG2D‐CAR‐T cells in gastric cancer (GC) still needs further exploration. This study identified a novel combination immunotherapy strategy with Dickkopf‐1 (DKK1) inhibition and NKG2D‐CAR‐T cells, exerting synergistic and superior antitumor effect in GC. We show that the baseline expression of NKG2D ligands (NKG2DLs) is at low levels in GC tissues from The Cancer Genome Atlas and multiple GC cell lines including NCI‐N87, MGC803, HGC27, MKN45, SGC7901, NUGC4, and AGS. In addition, DKK1 inhibition by WAY‐262611 reverses the suppressive tumor immune microenvironment (TIME) and upregulates NKG2DL expression levels in both GC cell lines and GC tissues from a xenograft NCG mouse model. DKK1 inhibition in GC cells markedly improves the immune‐activating and tumor‐killing ability of NKG2D‐CAR‐T cells as shown by cytotoxicity assays in vitro. Moreover, the combination therapy of NKG2D‐CAR‐T and WAY‐262611 triggers superior antitumor effects in vivo in a xenograft NCG mouse model. In sum, our study reveals the role of DKK1 in remodeling GC TIME and regulating the expression levels of NKG2DLs in GC. We also provide a promising treatment strategy of combining DKK1 inhibition with NKG2D‐CAR‐T cell therapy, which could bring new breakthroughs for GC immunotherapy.

Our study reveals the role of Dickkopf‐1 (DKK1) in remodeling gastric cancer (GC) tumor immune microenvironment and regulating the expression levels of natural killer group 2 member D (NKG2D) ligands in GC. We also provide a promising treatment strategy of combining DKK1 inhibition with NKG2D‐ chimeric antigen receptor‐T cell therapy, which could bring new breakthroughs for GC immunotherapy.

Abbreviations

AML

acute myeloid leukemia

CAR

chimeric antigen receptor

DKK1

Dickkopf‐1

E:T

effector : target

FCM

flow cytometry

GC

gastric cancer

HNSC

head and neck squamous cell carcinoma

IFN‐γ

γ‐interferon

IHC

immunohistochemistry

MDSC

myeloid‐derived suppressor cell

MFC

mouse forestomach carcinoma

MICA/B

MHC class I related chain A/B

NCG

NOD/ShiLtJGpt‐Prkdc^em26Cd52Il2rg^em26Cd22/Gpt

NK

natural killer

NKG2D

natural killer group 2 member D

NKG2DL

natural killer group 2 member D ligand

NS

normal saline

NTD

nontransduced T cell

ORR

objective response rate

PD‐1

programmed cell death protein

TIME

tumor immune microenvironment

TNF‐α

tumor necrosis factor‐α

ULBP1‐6

UL16‐binding protein 1–6

1. INTRODUCTION

Globally, GC has the sixth highest incidence and third highest mortality rate of all cancer types. ¹ It has long been considered as a highly heterogeneous disease with poor prognosis, and responds poorly to immunotherapy due to the suppressive TIME. ² The immunotherapy (anti‐PD‐1) in chemorefractory GC only had limited results, with an overall ORR of 12%. ³ Even with combined anti‐PD‐1 and chemotherapy in the first line, the median overall survival for patients with a PD‐1 ligand combined positive score of 5 or more is low at approximately 14.4 months. ⁴ Therefore, it is imperative to find novel immunotherapeutic approaches for GC patients.

A novel approach to cancer immunotherapy, CAR‐T cell therapy, has been shown to be very promising, which has yielded remarkable success in hematological malignancies. ⁵ The NKG2D receptor is a C‐type lectin‐like transmembrane glycoprotein and functions by binding to NKG2DLs, which in humans mainly consists of MICA/B and ULBP1‐6. ⁶ , ⁷ , ⁸ Tissues and cells are generally devoid of NKG2DLs in the physiological state, while the expression levels of NKG2DLs can be significantly upregulated under pathological conditions including infection, stress, and tumor formation. ⁹ Therefore, NKG2D has the potential to be a promising target for CAR‐T cell therapy to provide better tumor‐killing effects while reducing the occurrence of side‐effects.

Several previous preclinical studies have reported satisfactory treatment efficacy and potential application value of NKG2D‐CAR‐T cells in preclinical models of hematological malignancies and several solid tumors including glioblastoma, hepatocellular carcinoma, and breast cancer. ¹⁰ , ¹¹ , ¹² , ¹³ For example, Driouk et al. reported that NKG2D‐CAR‐T cells showed robust treatment efficacy to AML and T‐cell acute lymphoma. ¹⁰ As in solid tumors, Sun et al. reported that NKG2D‐CAR‐T cells resulted in significant tumor regression in preclinical models of hepatocellular carcinoma, ¹¹ while Yang et al. found effective killing effects of NKG2D‐CAR‐T cells on both glioma cells and tumor stem cells in a glioma model. ¹² However, current clinical trials of NKG2D‐CAR‐T cells in multiple cancer types have not yielded satisfactory results, requiring further investigation. ¹⁴ , ¹⁵ , ¹⁶

Dickkopf‐1, an antagonist of the classical WNT pathway, has been reported by previous studies to be overexpressed in a variety of tumors including GC, and is associated with poor prognosis. ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ Recent studies have found that DKK1 can create a suppressive TIME by inducing accumulation of MDSCs and inhibiting CD8⁺ T cell activation, thus promoting tumor progression in melanoma and colorectal cancer. ²¹ , ²² Notably, Malladi et al. reported in 2016 that knocking down DKK1 expression resulted in significantly increased expression of NKG2DLs, restoring NK‐mediated clearance of lung cancer cells. ²³ Therefore, we hypothesize that inhibition of DKK1 in GC cells could also lead to upregulation of NKG2DLs, which could boost the targetability of CAR‐T cells. Moreover, reversed suppressive TIME by DKK1 inhibition might also improve the expansion and persistence of CAR‐T cells, which provides the rationality of combining DKK1 inhibition and NKG2D‐CAR‐T cell therapy for enhanced immunotherapy in GC.

2. MATERIALS AND METHODS

2.1. Cell culture

Human GC cell lines NCI‐N87, MGC803, HGC27, MKN45, SGC7901, NUGC4, AGS, mouse GC cell line MFC (originally induced from the gastric tissue of 615‐line mice), HEK cell line 293 T, human embryonic lung diploid fibroblasts 2BS, and HUVECs were obtained from Cell Bank of Shanghai Institute of Biochemistry and Cell Biology. All cells were cultured in RMPI‐1640 (Corning) containing 10% FBS and 1% penicillin as well as streptomycin at 37°C under a humidified atmosphere of 5% CO₂.

2.2. Mice

Five to six‐week‐old female 615‐line mice were purchased from Institute of Hematology, Chinese Academy of Medical Sciences. Five to six‐week‐old female BALB/c nude mice and NCG mice were purchased from GemPharmatech Co. Ltd. All mice were kept in specific pathogen‐free animal facilities at Nanjing Drum Tower Hospital.

2.3. Treatment with DKK1 inhibitor in vitro

The DKK1 inhibitor WAY‐262611 was purchased from Topscience and dissolved in DMSO. NCI‐N87, MGC803, HGC27, MKN45, SGC7901, NUGC4, and AGS cells were separately seeded into 12‐well plates a day in advance. On the following day, cell culture medium was aspirated and PBS was used to wash off unadhered cells before fresh medium was replaced. WAY‐262611 at different concentrations of 0, 0.63, 1.26, 2.56, 5, and 10 μM were added and each concentration had three replicates. After 48 h of coculture, cell culture medium was removed and cells were digested with trypsin to prepare single cell suspension for FCM analysis. Human normal cell lines HUVEC, 2BS, and 293 T were treated in the same way.

2.4. Treatment with DKK1 inhibitor in vivo

To investigate the treatment efficacy of DKK1 inhibition in vivo, we established a GC syngeneic mouse model and immune‐deficient mouse model. For the GC syngeneic model, 5–6‐week‐old 615‐line mice were injected subcutaneously with MFC cells (1 × 10 ⁶ ) in 100 μL PBS. For the GC immune‐deficient model, 5–6‐week‐old nude mice were injected subcutaneously with MFC cells (1 × 10 ⁶ ) in 100 μL PBS. After tumor challenge, mice were randomized into groups treated with WAY‐262611 (10 mg/kg bodyweight) or DMSO six times in 14 days. Mice were killed by cervical dislocation and tumors as well as spleens were collected for FCM analysis.

2.5. Combination therapy experiment

NCI‐N87 cells were suspended (at a density of 5 × 10⁷/mL) in PBS. The suspension (100 μL) was injected subcutaneously into the NCG mice. After 3 days of tumor inoculation, mice were randomized into four groups (n = 5 per group): Control, WAY‐262611, KD‐025, and WAY‐262611 + KD‐025. Mice were treated with 100 μL NS or WAY‐262611 (10 mg/kg) i.p. every 2 days. After the tumor volume reached approximately 70 mm³, mice were treated with 100 μL NS or KD‐025 (1 × 10⁸/mL) by i.v. tail vein injections. Tumors were measured on alternate days and tumor volumes were calculated as follows: tumor volume (mm³) = width² × length/2. All mice were killed by cervical dislocation on day 28 post tumor inoculation. Tumors were collected for histologic analysis and the heart, liver, spleen, lung, kidney, stomach, large intestine, and small intestine were collected for H&E staining.

2.6. Cytotoxicity assay

To evaluate the antitumor ability of CAR‐T cells after DKK1 inhibition in vitro, NCI‐N87 cells were pretreated with medium containing WAY‐262611 (10 μM) for 48 h. Cells were labeled with carboxyfluorescein succinimidyl ester (Vazyme), which were defined as the target cells, and were seeded into ultralow 96‐well plates at a density of 4 × 10⁴ cells/well. The effector cells, KD‐025 and NTD, were added into the well at an E:T ratio of 1:3, 1:1, 3:1 and 6:1 in triplicate and incubated with target cells for 6 h at 37°C. All cells were stained with propidium iodide (Vazyme) after coculture to differentiate dead cells from live cells, and then qualified by FCM. The supernatant was harvested to measure the concentration of TNF‐α and IFN‐γ using the LEGENDplex HU Th1/Th2 Panel (BioLegend).

2.7. Statistical analysis

GraphPad Prism was used to undertake statistical analysis and create graphics. A minimum of three replications were carried out for all experiments. The data are expressed as the average ± SEM. Unpaired two‐tailed Student's t‐test was used to compare data between two groups when data were normally distributed, and the Mann–Whitney U‐test was used when data were not normally distributed. One‐way ANOVA was used for analysis of three or more groups. Statistical significance was defined as: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001.

Other methods including Bioinformatics analysis, mRNA RT‐quantitative PCR, Vector design, lentiviral package, CAR‐T cell preparation, FCM analysis, and IHC are provided in Appendix S1.

3. RESULTS

3.1. Inhibition of DKK1 reverses immune suppressive tumor microenvironment of syngeneic GC model

To investigate whether DKK1 inhibition showed antitumor efficacy in GC models, we established an MFC‐challenged GC syngeneic immunocompetent mouse model and a T cell‐deficient mouse model, followed by treatment with DKK1 inhibitor (WAY‐262611) or DMSO (control group for WAY‐262611) as indicated (Figures 1A and S1A). An interesting finding was that the tumor size and tumor volume decreased significantly in GC immunocompetent mice (Figure 1B,C). Meanwhile, we found a decrease of tumor volume in the GC T cell‐deficient model, however, the differences did not reach the statistical significance (Figure S1B). Considering the role of DKK1 in creating a suppressive TIME has been reported in recent studies, ²¹ , ²⁴ we hypothesized that WAY‐262611 could reverse the suppressive TIME in GC and thus achieve antitumor efficacy. Next, we undertook FCM of tumor tissues and spleens to analyze the immune status in the GC immunocompetent model treated with WAY‐262611 or DMSO. The FCM analysis showed increased tumoral and splenic infiltration of CD8⁺ T cells (Figure 1D,K) and higher proportions of tumoral IFN‐γ⁺ lymphocytes and TNF‐α⁺ lymphocytes (Figure 1E,F). Increased proportions of tumoral NK cells and decreased proportions of splenic MDSCs and M2 macrophages were found after DKK1 inhibition (Figure 1G,M,O). We also found that, although not statistically significant, the NK cells (p = 0.0863) in the spleen showed a tendency to increase and the MDSCs and M2 in the tumor (p = 0.0915, p = 0.0873) tended to decrease after inhibiting DKK1 inhibition (Figure 1H,J,L). However, M1 macrophages in the tumors showed no significant changes, but increased in the spleens after DKK1 inhibition (Figure 1I,N).

FIGURE 1.

Dickkopf‐1 (DKK1) inhibition reverses suppressive tumor immune microenvironment of a syngeneic gastric cancer (GC) model. (A) Schematic of DKK1 inhibitor (WAY‐262611) treatment schedule in GC subcutaneous model. (B) Representative images of tumors and (C) tumor volumes of mouse forestomach carcinoma (MFC)‐challenged 615‐line mice. (D–K) Proportions of (D) CD8⁺/CD45⁺ cells, (E) tumor necrosis factor‐α (TNF‐α)⁺/CD45⁺ cells, (F) γ‐interferon (IFN‐γ)⁺/CD45⁺ cells, (G) natural killer (NK)1.1⁺/CD45⁺ cells, (H) Gr‐1⁺/CD11b⁺ cells, (I) CD86⁺/F4/80⁺ cells, and (J) CD163⁺/F4/80⁺ cells in tumors were detected by flow cytometry (FCM). (K–O) Proportions of (K) CD8⁺/CD45⁺ cells, (L) NK1.1⁺/CD45⁺ cells, (M) Gr‐1⁺/CD11b⁺ cells, (N) CD86⁺/F4/80⁺ cells, (O) and CD163⁺/F4/80⁺ cells in spleens were detected by FCM. Data with error bars are shown as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. D, day.

Correspondingly, we undertook FCM in the GC T cell‐deficient model. Although the tumor volume did not reach statistical significance (p = 0.088) after DKK1 treatment, the immune status was improved. The FCM analysis indicated that DKK1 inhibition induced significantly higher infiltration of NK cells and lower infiltration of M2 macrophages in the tumors (Figure S1C,D), and a decreasing trend of M2 macrophage infiltration in the spleens (p = 0.1339; Figure S1G). No significant changes were found in the splenic infiltration of NK cells (p = 0.4934; Figure S1F). Moreover, increased proportions of dendritic cells in the tumors and decreased proportions of MDSCs in the spleens were also observed in the WAY‐262611 group (Figure S1E,H). Collectively, DKK1 inhibition triggers antitumor efficacy in GC models by reversing the suppressive TIME, depending on the competent immune system.

3.2. Inhibition of DKK1 upregulates expression levels of NKG2DLs in vitro

High autocrine of DKK1 from tumor cells has been reported to decrease the NKG2DL expression on tumors, which leads to its escape from immunosurveillance. ²³ In addition, a cancer vaccine targeting MICA/B (part of NKG2DLs) showed obvious efficacy through increasing the density of MICA/B on tumor cells, while augmenting the T cell and NK cell responses. ²⁵ Therefore, we further explored the role of DKK1 in regulating NKG2DL expression in GC. First, we investigated the expression levels of NKG2DLs in the mRNA level among multiple cancer types based on the data from GEPIA (http://gepia.cancer‐pku.cn/), including GC, liver hepatocellular carcinoma, and HNSC (Figure S2A). The results showed that NKG2DLs are widely expressed among multiple cancers, especially the high expression of ULBP2, ULBP6, and MICA in HNSC, lung squamous cell carcinoma, and esophageal carcinoma. However, in GC tissues, apart from the moderate expression of MICA, other NKG2DLs were expressed at rather low levels. Second, we analyzed the mRNA expression levels of NKG2DLs among multiple GC cell lines based on the CCLE database. It is also interesting to note that most NKG2DLs were expressed at low levels among all GC cell lines, except for the moderate expression level of MICB in AGS and SNU‐601 (Figure S2B). Accordingly, we detected NKG2DL expression at the protein level in seven GC cell lines, namely NCI‐N87, MGC803, HGC27, MKN45, SGC7901, NUGC4, and AGS using FCM (Figures 2A and S3). Consistent with the results from the CCLE database, the majority of NKG2DLs were expressed at low levels in GC cell lines, except for the moderate expression of ULBP4 and ULBP2/5/6 in NCI‐N87 cells (Figure 2A). Collectively, we found that compared with other cancer types, relatively low levels of NKG2DLs were expressed in both GC tissues and GC cell lines.

FIGURE 2.

Inhibition of Dickkopf‐1 (DKK1) upregulates the expression levels of natural killer group 2 member D ligands (NKG2DLs) in vitro. (A) Expression levels of NKG2DLs at the protein level including UL16‐binding protein (ULBP)1, ULBP3, ULBP4, and ULBP2/5/6, and MHC class I related chain (MIC)A and MICB were detected by flow cytometry across seven gastric cancer cell lines: NCI‐N87, MGC803, HGC27, MKN45, SGC7901, NUGC4, and AGS. Data are shown in the bar chart. (B–E) Expression levels of ULBP1, ULBP3, ULBP4, ULBP2/5/6, MICA, and MICB in (B) NCI‐N87, (C) HGC27, (D) MKN45, and (E) MGC803 cells after treatment with different concentrations of DKK1 inhibitor (WAY‐262611) (0, 0.63, 1.26, 2.52, 5 and 10 μM) for 48 h were detected by flow cytometry. Data with error bars are shown as mean ± SEM.

Next, we investigated whether inhibition of DKK1 in vitro could upregulate the expression levels of NKG2DLs. We found that in NCI‐N87 and HGC27 cells, the expression levels of all NKG2DLs were remarkably upregulated after WAY‐262611 coculture in a dose‐dependent manner, while ULBP4 and ULBP2/5/6 were expressed at extremely high levels (Figures 2B,C and S4A,B). Similarly, in MKN45 and MGC803 cells, the expression levels of ULBP4 and ULBP2/5/6 were also dramatically upregulated after DKK1 inhibition, with the levels of other NKG2DL expression slightly increased (Figures 2D,E and S4C,D). Taken together, we validated the low expression levels of NKG2DLs in GC cell lines at the protein level, and determined that the NKG2DLs expression in GC cell lines could be effectively upregulated through DKK1 inhibition.

3.3. Inhibition of DKK1 induces an increase in expression levels of NKG2DLs in vivo

We then explored whether DKK1 inhibition in vivo could also induce an increase in the expression levels of NKG2DLs. As NKG2DL expression showed the most substantial increase in NCI‐N87 cells after in vitro DKK1 inhibition, we established a subcutaneous NCI‐N87‐challenged GC model on NCG mice for further in vivo study, with WAY‐262611 injected i.p. at different concentrations (0, 5, and 10 mg/kg) (Figure 3A). Flow cytometry analysis of NCI‐N87 tumors showed that the expression of all NKG2D ligands were notably upregulated after WAY‐262611 treatment in a dose‐dependent manner (Figure 3B). Accordingly, ULBP4 and ULBP2/5/6 were upregulated to an extremely high expression level, consistent with the results of DKK1 inhibition in GC cells in vitro. Moreover, IHC analysis of NKG2DL expression in NCI‐N87 tumor tissues was carried out. Consistent with the FCM results, the expression levels of all NKG2DLs except MICA were obviously upregulated in tumor tissues of the WAY‐262611 group (Figure 3C). We also constructed a subcutaneous MKN45‐challenged GC model to verify our results (Figure S5A). Similarly, all NKG2DLs were markedly upregulated in the MKN45 tumors after treatment with WAY‐262611 (Figure S5B). Therefore, DKK1 inhibition can induce increased NKG2DL expression in both in vitro GC cell lines and in vivo GC tumor tissues.

FIGURE 3.

Inhibition of Dickkopf‐1 induces an increase in the expression levels of natural killer group 2 member D ligands (NKG2DLs) in vivo. (A) Schematic of WAY‐262611 treatment schedule in NCI‐N87 challenged gastric cancer subcutaneous model. (B) Representative flow cytometry (data of the expression of UL16‐binding protein [ULBP]1, ULBP3, ULBP4, ULBP2/5/6 and MHC class I related chain [MIC]A and MICB) of NCI‐N87 tumor tissues in mice. (C) Expression levels of NKG2DLs in NCI‐N87 tumor tissues were detected by immunohistochemical staining of ULBP1, ULBP2/5/6, ULBP3, ULBP4, MICA, and MICB. Scale bar, 100 μm. D, day.

3.4. Inhibition of DKK1 enhances tumor killing effects of NKG2D‐CAR‐T cells in vitro

After identifying the role of DKK1 inhibition in upregulating NKG2DL expression levels in GC, we next investigated whether WAY‐262611 could also improve the tumor killing effects of NKG2D‐CAR‐T cells (KD‐025) in vitro. Accordingly, we designed the NKG2D‐CAR lentiviral vector (NKG2D‐BBz) as shown in Figure 4A. T cells were obtained from healthy donors and confirmed to be CD3⁺ using FCM (Figure S6). CD3/28 beads were used to activate the cells for 48 h, after which the cells were transduced with NKG2D‐BBz lentivirus. The expression of NKG2D‐BBz CAR was detected on day 3 by FCM and 51.1% of the T cells were positive for NKG2D‐BBz CAR (Figure 4B).

FIGURE 4.

Inhibition of Dickkopf‐1 (DKK1) enhances the tumor killing effects of natural killer group 2 member D (NKG2D)‐chimeric antigen receptor (CAR)‐T cells in vitro. (A) Schematic representation of NKG2D‐BBz CAR. (B) NKG2D‐based CAR expression was detected by FACS with NKG2D Ab. (C) Schematic of coculture model of NCI‐N87 cells in vitro. (D, E) Levels of (D) γ‐interferon (IFN‐γ) and (E) tumor necrosis factor‐α (TNF‐α), released by KD‐025, were measured after 6 h of coculture incubation at the effector : target (E:T) ratio of 3:1. (F) Cytotoxicity of KD‐025 against NCI‐N87 cells before and after DKK1 inhibition at different E:T ratios of 1:3, 1:1, 3:1, and 6:1. (G) Representative flow cytometric plots of (F). (H) Cytotoxicity of nontransduced T cells (NTD) against NCI‐N87 cells before and after DKK1 inhibition at different E:T ratios 1:3, 1:1, 3:1, and 6:1. (I) Representative flow cytometric plots of (H). Results are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. CFSE, carboxyfluorescein succinimidyl ester; PI, propidium iodide.

We next cultured NCI‐N87 cells with WAY‐262611 for 48 h, and then incubated NCI‐N87 cells with KD‐025 or NTD for 6 h at different E:T ratios of 1:3, 1:1, 3:1, and 6:1 (Figure 4C). The control group was incubated with KD‐025 or NTD in the same approach for 6 h. We found that without DKK1 inhibition, KD‐025 showed limited killing effects on NCI‐N87 cells compared to NTD (p < 0.01; Figure S7A,B). Notably, after inhibiting DKK1, the killing effects of KD‐025 on NCI‐N87 cells were significantly upregulated compared to NTD, which was also improved with the increase of E:T ratios (Figure 4F,G). In contrast, the killing effect of NTD showed no obvious difference with or without DKK1 inhibition (Figure 4H,I). Moreover, the cytokine expression in the supernatants of the cell culture was detected, and an increased release of IFN‐γ and TNF‐α by KD‐025 was observed after DKK1 inhibition, which could account for its enhanced tumor killing ability (Figure 4D,E). No significant release of cytokines was detected in NTD. We also carried out this cytotoxicity assay using MKN45 and HGC27 GC cells to substantiate the results above. The killing effects of KD‐025 were significantly upregulated and enhanced with the increasing E:T ratios (Figures S8A–D and S9A–D). In addition, we detected the increased release of IFN‐γ, TNF‐α, and interleukin‐5 by KD‐025 (Figures S8E–G and S9E–G). Together, these in vitro results indicate that inhibition of DKK1 could enhance the activation and antitumor cytotoxicity of NKG2D‐CAR‐T cells, thus improving the treatment effects of NKG2D‐CAR‐T cells in GC.

3.5. Inhibition of DKK1 in vivo enhances antitumor efficacy of NKG2D‐CAR‐T cells

Considering that the antitumor cytotoxicity of NKG2D‐CAR‐T cells could be significantly enhanced by DKK1 inhibition in vitro, we next explored whether DKK1 inhibition could augment the antitumor effect of NKG2D‐CAR‐T cells in vivo. NCI‐N87‐challenged and MKN45‐challenged xenograft GC models were established by injecting NCI‐N87 and MKN45 cells into NCG mice, as well as injecting WAY‐262611 intravenously, with KD‐025 i.v. injection on day 7 (Figures 5A and S10A). We found that DKK1 inhibition or NKG2D‐CAR‐T cell infusion alone could inhibit GC tumor growth to some extent, while the combination therapy of WAY‐262611 + KD‐025 presented a significantly better antitumor effect than each single treatment (Figures 5B–E and S10B–E). The treatment efficacy of WAY‐262611 in tumor‐bearing NCG mice was impaired compared to tumor‐bearing T cell‐deficient mice, which suggested again that the activity of WAY‐262611 is immunodependent.

FIGURE 5.

Inhibition of Dickkopf‐1 (DKK1) enhances the antitumor efficacy of natural killer group 2 member D (NKG2D)‐chimeric antigen receptor (CAR)‐T cells in vivo. (A) Schematic of WAY‐262611 and KD‐025 treatment schedule in a gastric cancer subcutaneous model. (B) Representative images of tumors, (C) tumor weight and (D, E) tumor volume in Control, WAY‐262611, KD‐025, and WAY‐262611 + KD‐025 groups. (F) Immunohistochemical analysis to detect CAR‐T cells in NCI‐N87 tumor tissues using a CD3 Ab. Scale bars, 200 μm (left), 50 μm (right). Data with error bars are shown as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. D, day.

Of note, anti‐CD3 Ab was used to detect infiltration of T cells in tumor tissues. Higher levels of CAR‐T cells were still detectable in tumor tissues at the treatment end‐point of day 28, and the infiltration of CAR‐T cells in NCI‐N87 tumors was dramatically increased after DKK1 inhibition (Figure 5F).

We evaluated the safety of this combination therapy to avoid “on‐target, off‐tumor” toxicity. We first analyzed the correlation between DKK1 and NKG2DLs in gastric normal tissues from The Cancer Genome Atlas stomach adenocarcinoma data collection. We found that DKK1 expression is positively correlated with ULBP2–6 expression, and is not significantly correlated with ULBP1, MICA, or MICB expression (Figure S11), which indicates that the downregulation of DKK1 might not cause the NKG2DL upregulation in normal human tissues. Next, we treated human normal cell lines HUVEC, 2BS, and 293 T with different concentrations of WAY‐262611 and there was no significant increased expression of NKG2DLs in HUVEC, 2BS, or 293 T cells (Figure S12). In addition, no significant differences of the mouse NKG2DLs were found in the mouse heart, liver, spleen, lung, or kidney after WAY‐262611 treatment (Figure S13). Moreover, no apparent CAR‐T‐related toxicity or obvious weight loss was observed in either the single or combination treatment groups (Figures S14 and S15). These results indicated that the infiltrating abilities and treatment efficacy of NKG2D‐CAR‐T cells in GC were both enhanced after DKK1 inhibition, thus providing robust preclinical evidence for the combination therapy of DKK1 inhibition and NKG2D‐CAR‐T cells.

4. DISCUSSION

Development of CAR‐T cell therapy for solid tumors is slow and remains challenging, with CAR‐T cell treatment for GC still lacking ideal targets. A recent phase I trial of claudin 18.2‐specific CAR‐T cells in refractory advanced GC showed promising results with an ORR of 57.1%, ²⁶ suggesting that CAR‐T cell therapy could be beneficial to GC patients and more suitable targets as well as combination therapy and deserves further exploration. Although prior studies have reported the success of NKG2D‐CAR‐T cell therapy among multiple cancer types preclinically and elucidated the modulating role of DKK1 in antitumor immunity, little is known about the therapeutic efficacy of combination therapy of NKG2D‐CAR‐T cells and DKK1 inhibition in GC. Based on this research, we first clarified the antitumor effects and immunoregulating role of DKK1 inhibition in GC, which reverses the suppressive TIME of GC and augments the CD8⁺ T cell and NK cell responses. Next, we found that the NKG2DL expression was comparatively low in GC tumors and cell lines, but they could be noticeably increased both in vitro and in vivo after DKK1 inhibition. Additionally, DKK1 inhibition also improves the activation and tumor‐killing abilities of NKG2D‐CAR‐T cells in in vitro coculture models. Moreover, DKK1 inhibition enhances the tumor targetability of NKG2D‐CAR‐T cells, leading to remarkably improved treatment effects in the in vivo xenograft GC model, and was verified of no obvious toxicity. Therefore, this study provides a novel and potential combination strategy for enhanced immunotherapy in GC.

The NKG2DLs are widely expressed among multiple cancers including breast cancer, ovarian cancer, glioma, and AML. ²⁷ , ²⁸ , ²⁹ Evidence has been emerging that high NKG2DLs expression levels are correlated with improved clinical outcomes in various solid tumors like breast cancer and colon cancer. ³⁰ , ³¹ , ³² However, we found that the NKG2DL expression in GC tumor tissues and cell lines are generally low and to a great extent heterogenous. It has been reported that NKG2D‐CAR‐T cells are capable of targeting tumor cells that express NKG2DLs effectively and specifically, ³³ and the tumor killing effects of NKG2D‐CAR‐T cells are correlated with the number of tumor cells that express the ligands. ³⁴ Moreover, the density of NKG2DLs in tumor cells plays a role in the expansion of NKG2D‐CAR‐T cells, which decides its long‐lasting therapeutic efficacy. Therefore, cotreatment with NKG2D‐CAR‐T cell therapy with effective approaches that increase the density of NKG2DLs in tumor cells could amplify its antitumor response.

In the current research, a series of NKG2DLs have been reported including MICA/B, ULBP1‐6 in human, and H60a/b/c, RAET1 α, β, γ, δ, ε and MULT1 in mouse. ³⁵ Given that NKG2DLs are not homologous in human and mouse, in this study we designed CAR‐T targeting human NKG2DLs for better clinical translation. Previous studies have reported that drugs or irradiation that cause DNA‐damaging reactions can consistently increase the expression of ULBP1‐3, MICA, and MICB. ³⁶ Correspondingly, Peng et al. found that gefitinib (epidermal growth factor receptor tyrosine kinase inhibitor) promoted the expression of NKG2D in NK cells and upregulated the expression of ULBP1, ULBP2, and MICA in tumor cells. ³⁷ Malladi et al. also verified that knockdown of DKK1 led to a significant increase in the expression of NKG2DLs including ULBP2, ULBP5, and MICA on the surface of the tumor cells, which in turn restored NK cell‐mediated clearance. ²³ In our study, we found that expression levels of NKG2DLs in GC tumor and cell lines can be significantly upregulated by DKK1 inhibition, which brings improved treatment efficacy of NKG2D‐CAR‐T cells in the xenograft GC model. Therefore, we provide a novel therapeutic approach through enhancing the NKG2DL expression to reinstate the NKG2D axis‐mediated immunosurveillance, thus exerting synergistic tumor killing effects in combination therapy of NKG2D‐CAR‐T cells. Moreover, further research is required to pinpoint the mechanism behind DKK1 inhibitor‐induced upregulation of NKG2DLs in GC.

Apart from the low expression levels of NKG2DLs, the suppressive TIME, which is generally responsible for CAR‐T exhaustion, is also an important factor limiting the treatment efficacy of NKG2D‐CAR‐T cell therapy. ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ In the last 5 years, DKK1 has also been reported to negatively modulate the TIME and impair the effects of immunotherapies in various cancers. ²¹ , ²² , ²⁴ , ⁴² D'Amico et al. reported that DKK1 promotes tumor growth by targeting the immature myeloid population in melanoma and lung cancer models. ²¹ Additionally, Haas et al. clarified that DKK1 has a negative correlation with tumor infiltrating CD8⁺ T cells and a positive correlation with MDSCs in melanoma and breast cancer. ⁴² Furthermore, inhibition of DKK1 has been clarified to induce a “hot” tumor microenvironment and functions well in immune‐modulatory therapy. ²⁴ Accordingly, we found DKK1 inhibition also shows antitumor effects in an immunocompetent GC model and remodeled the suppressive GC TIME. Inhibition of DKK1 could potentially enhance the expansion and persistence of NKG2D‐CAR‐T cells by reducing other negative regulatory factors in the TIME of GC. Moreover, an exciting result of an ORR of 90% in patients with high DKK1 expression has been reported in the DisTinGuish trial (NCT04363801), combining DKN‐01 (anti‐DKK1) and tislelizumab (anti‐PD‐1) plus chemotherapy treatment in advanced gastroesophageal adenocarcinoma patients. ⁴³ Therefore, inhibition of DKK1 in combined immunotherapy shows promising potential value in clinical application, which provides compelling corroboration for future applications of combined therapy of DKK1 inhibition and NKG2D‐CAR‐T therapy in GC. Collectively, reversing the low expression levels of NKG2DLs as well as inducing a “hot” TIME in GC by inhibiting DKK1 could allow the successful application of NKG2D‐CAR‐T cell therapy to GC patients.

Apart from its therapeutic efficacy, the safety of this novel combination therapy also matters. In physical status, to avoid the attack of the autoimmune system, NKG2DL expression is tightly regulated and NKG2DLs are absent or poorly expressed in normal tissues. ⁹ Some ULBP molecules are detected to be expressed at the mRNA level by normal cells, however, their expression at the protein level is relatively low, controlled by the posttranscriptional regulation. ⁴⁴ Although MICA was reported to be exclusively expressed by gastrointestinal epithelial cells of normal humans, recent studies have revealed the main expression of MICA is restricted in the cell nucleus. ¹² Clinical trials of NKG2D‐CAR‐T cells in AML and colorectal cancer have confirmed treatment safety without obvious treatment‐related issues. ¹⁴ , ¹⁵ In our results, no apparent syndromes such as weight loss were observed in the mouse models. In addition, H&E staining of mouse organs suggested no treatment‐related toxicity in the single or combination treatment group, indicating the safety of such combination therapy in the NCG mouse model. However, due to the low genetic homology of NKG2DLs in humans and mice, the safety evaluation of the NCG mouse model cannot fully assess the risk of off‐target effects of the combination therapy in humans. ⁴⁴ Therefore, more clinical studies are needed to confirm the safety of combination therapy in the future.

In summary, this study reveals that inhibition of DKK1 could significantly remodel the suppressive TIME in GC and upregulate NKG2DL expression in both GC tumor tissues and cell lines. Of note, the combination therapy of DKK1 inhibition and NKG2D‐CAR‐T therapy exerts a superior antitumor effect compared with single treatment both in vitro and in vivo, which could bring new insight for the immunotherapy of GC.

FUNDING INFORMATION

This work was funded by grants from the National Natural Science Foundation of China (82073382) and the Fundamental Research Funds for the Central Universities (0214–14,380,506). This study was designed, collected, analyzed, interpreted, and written entirely independently from funding sources.

ETHICS STATEMENTS

Approval of the research protocol by an institutional review board: The study protocol was approved by the Institutional Review Board of Nanjing University.

Informed consent: All healthy donors provided written informed consent.

Registry and registration no. of the study/trial: N/A.

Animal studies: All animal experiments were approved by the Drum Tower Hospital's Institutional Animal Care and Use Committee (approval number: 2020AE01064).

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest for this article.

Supporting information

Appendix S1

Zhang Y, Liang K, Zhou X, et al. Combination therapy of DKK1 inhibition and NKG2D chimeric antigen receptor T cells for the treatment of gastric cancer. Cancer Sci. 2023;114:2798‐2809. doi: 10.1111/cas.15828