Background Adoptive cell therapy (ACT) using genetically modified T cells has evolved into a promising treatment option for patients with cancer. However, even for the best-studied and clinically validated CD19-targeted chimeric antigen receptor (CAR) T-cell therapy, many patients face the challenge of lack of response or occurrence of relapse. There is increasing need to improve the efficacy of ACT so that durable, curative outcomes can be achieved in a broad patient population.
Methods Here, we investigated the impact of indomethacin (indo), a non-steroidal anti-inflammatory drug (NSAID), on the efficacy of ACT in multiple preclinical models. Mice with established B-cell lymphoma received various combinations of preconditioning chemotherapy, infusion of suboptimal dose of tumor-reactive T cells, and indo administration. Donor T cells used in the ACT models included CD4+ T cells expressing a tumor-specific T cell receptor (TCR) and T cells engineered to express CD19CAR. Mice were monitored for tumor growth and survival. The effects of indo on donor T cell phenotype and function were evaluated. The molecular mechanisms by which indo may influence the outcome of ACT were investigated.
Results ACT coupled with indo administration led to improved tumor growth control and prolonged mouse survival. Indo did not affect the activation status and tumor infiltration of the donor T cells. Moreover, the beneficial effect of indo in ACT did not rely on its inhibitory effect on the immunosuppressive cyclooxygenase 2 (COX2)/prostaglandin E2 (PGE2) axis. Instead, indo-induced oxidative stress boosted the expression of death receptor 5 (DR5) in tumor cells, rendering them susceptible to donor T cells expressing TNF-related apoptosis-inducing ligand (TRAIL). Furthermore, the ACT-potentiating effect of indo was diminished against DR5-deficient tumors, but was amplified by donor T cells engineered to overexpress TRAIL.
Conclusion Our results demonstrate that the pro-oxidative property of indo can be exploited to enhance death receptor signaling in cancer cells, providing rationale for combining indo with genetically modified T cells to intensify tumor cell killing through the TRAIL-DR5 axis. These findings implicate indo administration, and potentially similar use of other NSAIDs, as a readily applicable and cost-effective approach to augment the efficacy of ACT.
- immunotherapy, adoptive
- drug evaluation, preclinical
- receptors, chimeric antigen
- drug therapy, combination
- cell engineering
Data availability statement
Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as online supplemental information.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- immunotherapy, adoptive
- drug evaluation, preclinical
- receptors, chimeric antigen
- drug therapy, combination
- cell engineering
WHAT IS ALREADY KNOWN ON THIS TOPIC
The increased basal level of oxidative stress in cancer cells renders them more susceptible to redox disruption. It is well-documented that non-steroidal anti-inflammatory drugs (NSAIDs) can induce oxidative stress in cancer cells and cause growth inhibition. Emerging evidence indicates that induction of oxidative stress in cancer cells by tumor-reactive T cells contributes to the efficacy of adoptive T cell therapy (ACT). Whether the pro-oxidative activity of NSAIDs can augment the efficacy of ACT has not been explored.
WHAT THIS STUDY ADDS
We demonstrate that administration of indomethacin potentiates ACT in multiple preclinical models. Mechanistically, indomethacin-induced oxidative stress leads to enhanced expression of death receptor 5 (DR5) in cancer cells, sensitizing cancer cells to the cytotoxicity of tumor-reactive T cells through the TRAIL-DR5 axis.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Administration of indomethacin, and potentially other NSAIDs, represents a readily applicable and cost-effective approach to improve the therapeutic outcome of ACT.
Adoptive cell therapy (ACT) using T cell infusion products genetically modified to express tumor-specific TCRs or CARs has resulted in remarkable therapeutic outcomes in some patients with certain types of malignancies.1 2 However, the therapeutic potential of ACT as a treatment modality for various types of cancer in a broad patient population has yet to be fulfilled.3 4 Even for CD19-targeted CAR T cell therapy, the most advanced and clinically validated ACT, many patients failed to respond to therapy or succumbed to post-therapy relapse.5 Thus, finding ways to improve the efficacy of ACT has the potential to deliver durable, curative therapeutic benefits to more patients.
Cancer cells are often under a higher degree of basal level oxidative stress than normal cells, as reflected by increased presence of reactive oxygen species (ROS), likely driven by cell-intrinsic transformation events such as oncogene activation, tumor suppressor gene inactivation and altered metabolism, or by exogenous insults such as irradiation and chemotherapy.6 7 As a result, cancer cells tend to have increased dependence on the endogenous antioxidant defense systems to maintain redox homeostasis, making them more vulnerable than normal cells to further redox disruptions. The idea of targeting tumor redox homeostasis to achieve selective tumor killing has led to the development of numerous pro-oxidants that can incite oxidative stress in cancer cells either by promoting ROS production or by inhibiting the antioxidant systems.8–12 However, application of pro-oxidants as monotherapy for cancer has encountered challenges including inefficient drug delivery, lack of drug stability, limited tumor specificity, and tumor heterogeneity.
We recently reported that ACT can profoundly alter tumor redox metabolism, resulting in increased ROS accumulation and oxidative DNA damage in cancer cells.13 These results suggest that tilting tumor redox balance toward oxidative destruction is critical for the efficacy of ACT, raising the possibility that pro-oxidants can be employed to intensify tumor oxidative stress so as to sensitize tumor cells to ACT. Non-steroidal anti-inflammatory drugs (NSAIDs), which are widely used for pain relief and reducing inflammation, are known to induce oxidative and endoplasmic reticulum (ER) stresses that cause cancer cell apoptosis.14–22 Meanwhile, it has been well established that NSAIDs can exert antitumor effects by inhibiting the tumor-promoting COX/PGE2 axis. Inhibition of PGE2 can also mitigate immunosuppression in the tumor microenvironment (TME), leading to improved antitumor immunity.23 24 In this regard, compelling evidence from preclinical studies indicates that certain NSAIDs can augment the efficacy of anti-PD1 therapy,25 26 propelling the interest of applying NSAIDs to potentiate immune checkpoint blockade therapy. Although it has been reported that NSAIDs can sensitize tumor cells to CD19CAR T cell cytotoxicity in vitro,27 in vivo evidence supporting the use of NSAIDs in the setting of ACT is still lacking.
In this study, we set out to examine how administration of indomethacin (indo), a commonly used NSAID, shapes the therapeutic outcome of ACT in model systems where tumor-specific CD4+ T cells or CD19CAR T cells are used to treat murine or human B-cell lymphomas. Our results reveal oxidative stress-induced tumor-intrinsic death receptor signaling as the molecular basis by which indo potentiates the antitumor efficacy of tumor-reactive T cells, supporting repurposing NSAIDs for cancer treatment in the context of T cell adoptive immunotherapy.
Cell lines and cell culture conditions
A20 and CT26 tumor cell lines were purchased from the American Type Culture Collection. The generation of A20 cells expressing influenza hemagglutinin (A20HA) was described elsewhere.28 The death receptor 5 knockout A20 cell line (A20.DR5 KO) was generated using the Alt-R CRISPR-Cas9 System (IDT) following the manufacturer’s instruction. Limiting dilution was performed and single cell clones were analyzed by flow cytometry to select for DR5-negative clones. Nucleotide deletion in the targeted Dr5 gene loci was confirmed by DNA sequencing. A20.luciferase cells (A20.Luci) were generated by transducing A20 cells with retroviruses carrying luciferase (MSCV-luciferase). Murine CD19-expressing CT26 tumor cell line (CT26.mCD19) was created by transducing CT26 cells with a vector carrying murine CD19 (MSCV-mCD19). CD19+ tumor cells were subsequently FACS-sorted and expanded. Human B-cell lymphoma cell line Raji was provided by Dr. Huidong Shi (Augusta University). Human colorectal carcinoma cell line HT29 was provided by Dr. Gary A. Piazza (Auburn University). All tumor cell lines were cultured in RPMI 1640 (HyClone Laboratories) culture medium supplemented with 10% fetal bovine serum albumin (FBS), 1% penicillin/streptomycin (HyClone Laboratories), 1% non-essential amino acids, 1% glutamine and 0.1mM 2-mercaptoethanol at 37°C in a 5% CO2 incubator.
BALB/c mice (5–8 weeks old) were purchased from Charles River NCI at Frederick. TCR-Tg (6.5) mice on a BALB/c (Thy1.1) background expressing an αβTCR specific for amino acids 110–120 from influenza HA presented by MHC class II molecule IEd were described previously.29 BALB/c mice on the CD45.1 background and immunodeficient nonobese diabetic (NOD)–severe combined immunodeficient (Scid) IL-2RƔ-null (NSG) mice were purchased from the Jackson Laboratory. All mice were housed under specific pathogen-free conditions by Laboratory Animal Services of the Augusta University.
Mouse and human T cell viral transduction
The retroviral vector MSGV-1D3-28Z.1–3 containing mouse CD19-targeting CAR (mCD19CAR) was provided by James N. Kochenderfer at National Cancer Institute.30 The retroviral vector MSCV-luciferase was generated as described previously.31 The retroviral vector MSCV-mTnfsf10 containing mouse TRAIL was purchased from VectorBuilder. The retroviral vector SFG-humCD1928z containing human CD19-targeting CAR (hCD19CAR) was described elsewhere.32 Retroviral supernatants preparation and murine T cell retroviral transduction procedures were described previously.31 Phoenix Amphotropic cells were transfected with hCD19CAR retroviral vector using Lipofectamine 2000 to produce viral supernatants. Human T cells were isolated from the buffy coat of healthy donors using the EasySep Human T Cell Isolation Kit (STEMCELL Technologies). Purified T cells were activated by T-Activator CD3/CD28 Dynabeads (ThermoFisher) in the presence of hIL-2 (30 U/mL) for 48 hours. Activated T cells were spin-infected twice on two consecutive days with hCD19CAR retroviral supernatant using retronectin. Human T cells were further expanded for 7 days in the presence of hIL-2 (100 U/mL) before harvest for adoptive transfer.
Animal tumor models and in vivo treatments
Tumor cells were subcutaneously injected to the right flank of mice. The following numbers of tumor cells in 100 µl volume were used for each mouse: 5×106 for murine B-cell lymphoma cells (A20HA, A20 and A20.DR5KO), 3×106 for human B-cell lymphoma Raji cells, and 1×106 for murine colon cancer CT26.mCD19 cells. Tumor growth was monitored by caliper measurement every other day. Tumor areas were calculated as the product of two perpendicular diameters in square millimeters (mm2). When tumor sizes reached the desired sizes as indicated in specified experiments, mice were randomly assigned into groups to receive the specified treatments. CTX was injected intraperitoneally to mice at 150 mg/kg 1 day before T cell transfer. Indo was intraperitoneally injected to mice at 1 mg/kg. Indo administration started 4 hours before T cell transfer and continued daily for 28 days. Mice were euthanized when tumor sizes reached 300 mm2. For T cell transfer using unstimulated HA-specific CD4+ T cells, spleens collected from 6.5-Tg Thy1.1+ mice were processed into single cell suspensions. ~5 × 106 HA-specific CD4+ T cells were injected to each mouse via tail vein. For T cell transfer using pre-activated HA-specific CD4+ T cells, CD4+ T cells were isolated from the spleens of from 6.5-Tg mice using an EasySep Mouse CD4+ T Cell Isolation Kit (STEMCELL Technologies). Isolated CD4+ T cells were activated and expanded in vitro with T-Activator CD3/CD28 Dynabeads (ThermoFisher) for 3 days. 1.0×105 activated HA-specific CD4+ T cells were injected to mice via tail vein. For CAR T cell transfer,~3 x 106 mCD19CAR+ or ~0.8 x 106 hCD19CAR+ T cells were transferred to each mouse.
Antibodies and reagents
Single cell suspensions were stained for flow cytometry analysis using the following fluorochrome-conjugated antibodies specific to murine CD4 (GK1.5), CD8 (53–6.7), CD90.1 (OX-7), CD45R/ B220 (RA3-6B2), CD19 (6D5), CD262 (DR5) (MD5-1), CD253 (TRAIL) (N2B2), CD44 (IM7), CD62L (MEL-14), PD1 (RMP1-14), CD11b (M1/70), CD45.1 (A20), CD45.2 (104), anti–IL-2 (JES6-5H4), anti–TNF-α (MP6-XT22), anti–IFN-γ (XMG1.2), anti–GM-CSF (MP1-22E9), anti-GzmB (GB11), and antibodies specific to human CD4 (OKT4), CD8a (HIT8a) and CD262 (DR5) (DJR2-4). Antibodies were purchased from Biolegend unless otherwise noted. Anti-mouse FOXP3 (FJK-16s), and BCL-2 (Bcl2/100) were purchased from eBioscience. MCL-1 (B-6) (sc-74436) and Anti-FLIPS/L Antibody (G-11) (sc-5276) were purchased from Santa Cruz Biotechnology. Anti-COX2 (#10010096) antibody was purchased from Cayman Chemical. Polyclonal mouse anti-rat F(ab′)2 antibody and isotype control antibody were purchased from Jackson Immuno Research. Cyclophosphamide (CTX) was purchased from Tokyo Chemical Industry (Tokyo, Japan). Indomethacin was purchased from Sigma-Aldrich. N-Acetyl-L-cysteine was purchased from Alfa Aesar. DR5 agonist antibody (MD5-1) was purchased from Bio X cell. Lipofectamine 2000 was purchased from ThermoFisher Scientific. Retronectin was purchased from Takara. D-Luciferin monosodium salt was purchased from Pierce.
Sample preparation and flow cytometry analysis
Spleen and tumor samples were processed into single-cell suspensions. For surface molecule detection, cells were stained with fluorochrome-conjugated antibodies for 15 min at room temperature in the dark. To detect intracellular molecules, including cFLIP, MCL-1, BCL2, COX2, and FOXP3, cells were stained with surface markers followed by intracellular staining using the Foxp3 Transcription Factor Staining Buffer Set (eBioscience). For cytokine intracellular staining, T cells in splenocytes or tumor cell suspensions were stimulated with the GolgiPlug Leukocyte Activation Cocktail (BD Biosciences) for 3.5 hours at 37°C. Cells were harvested and surface stained, followed by cytokine staining using the Fixation/Permeabilization Solution Kit (BD Biosciences). For detection of mCD19CAR expression in transduced murine T cells, polyclonal mouse anti-rat-F(ab′)2 antibody was used to detect 1D3 scFv with normal polyclonal mouse IgG antibody as the isotype control. For detection of hCD19CAR expression in transduced human T cells, biotinylated L protein was used to stain cells followed by PE-conjugated streptavidin (SA) as described previously.33 For intracellular ROS measurement, ROS was detected using the cell permeant CM-H2DCFDA (ThermoFisher) as previously described.13 To evaluate tumor cell death triggered by DR5 signaling, 96-well protein A-coated plates (Pierce) were used to coat MD5-1, a DR5 agonist antibody, as previously described.34 In brief, wells in a protein A-coated plate were coated with 50 µL phosphate-buffered saline (PBS) or purified MD5-1 antibody (10 mg/mL) at 4°C overnight and washed with PBS before use. 1.0×105 A20 or A20.DR5KO cells were seeded into each well in 200 µL culture media not containing 2-mercaptoethanol. DMSO or indomethacin was added to PBS or MD5-1-coated wells as indicated. After overnight culture at 37°C, cells were harvested for Annexin V staining (BioLegend) following manufacturer’s instruction. DAPI (0.5 µg/mL) was added to cells before flow cytometry analysis.
Bioluminescent imaging (BLI) was performed on a Spectral Advanced Molecular Imaging X (Ami X) system (Spectral Instruments Imaging) to detect tumor-infiltrating donor T cells as we previously described.35 Briefly, luciferase-carrying viruses were used to transduce donor T cells. Luciferase-labeled HA-specific CD4+ T cells or mCD19CAR T cells were used for adoptive transfer. For imaging, each mouse received an intraperitoneal injection of 150 mg/kg luciferin and was anesthetized by inhalation of 2% isoflurane. The photographic images were acquired and overlaid with pseudocolor luminescent images. All BLI data were analyzed using the Aura Imaging Software (Spectral Instruments Imaging). The luminescence signal in the tumor region was quantified as photon/sec as an indicator of the density of donor T cells in tumor.
RNA extraction and quantitative real-time PCR
Tumor cells were harvested after overnight treatment under the indicated conditions. RNA was extracted using TRIZOL Reagent (ThermoFisher Scientific). 1 µg of RNA was used to generate cDNA using the SuperScript III First-Strand Synthesis System (ThermoFisher Scientific). Quantitative real-time PCR was performed using SYBR Green Mix (Bio-Rad) on a BioRad iCycler equipped with an iCycler iQ Detection System. β-actin was used to normalize target gene RNA expression levels. Each sample was assayed in triplicates. All primers were purchased from Integrated DNA Technologies. The sequences of the PCR primers are shown in online supplemental table 1.
Detection of prostaglandin E2 (PGE2) and indomethacin
To quantify the levels of PGE2 and indomethacin in tumor tissues, mice were sacrificed, and tumors were excised, weighed and cryopreserved in −80°C. Tumor samples and controls were homogenized 1:10 with 5 mM ammonium acetate buffer using an Omni THQ homogenizer. Due to the endogenous amount of PGE2 in the tumor, calibration standards, blanks and QCs were prepared by spiking 5 mM ammonium acetate (50 µL) with the appropriate amount of PGE2 and indomethacin to achieve concentrations ranging from 0.5 to 10,000 ng/mL. Blank and QC samples were also prepared in tumor homogenate for each analyte. Standards, blanks, QCs and samples were spiked with internal standard (10 µL of 1000 ng/mL PGE2 and indomethacin in acetonitrile). Acetonitrile (500 µL) was added to an equal sample volume of tumor homogenate to precipitate the proteins and vortexed well. After centrifugation for 5 min at 21 000 × g, the supernatant was transferred to an autosampler vials and analyzed by liquid chromatography–mass spectrometry (LC/MS/MS). The LC/MS/MS system consisted of Shimadzu system (Columbia MD) equipped with LC20-AD dual HLPC pumps, an SIL20-AC HT autosampler, and a DGU-20A2 in-line degasser. Detection was performed using an Applied BioSystems 5500 QTRAP (Applied Biosystems, Foster City, CA) triple quadrupole mass spectrometer operated in the positive ion mode. Mass calibration, data acquisition and quantitation were performed using Applied Biosystem Analyst V.1.7.2 software (Applied Biosystems, Foster City, California, USA).
Data were analyzed using Prism V.8.0 (GraphPad Software). Experimental values are expressed as mean±SE of mean (SEM). Survival was compared using a Log-rank (Mantel-Cox) test. For comparison between two groups, statistical analysis was performed using unpaired Student’s t-test. One-way analysis of variance followed by Tukey’s multiple testing correction was used for comparison of more than two groups. P values less than 0.05 were considered statistically significant.
Indo enhances the efficacy of ACT in a mouse B-cell lymphoma model
Initial experiments were conducted to determine if indo administration could enhance the efficacy of ACT in a well-characterized murine B-cell lymphoma model system.28 29 31 In this system, mice with established A20 tumors expressing a model tumor antigen HA (A20HA) received pre-conditioning chemotherapy with cyclophosphamide (CTX), followed by infusion of splenocytes containing HA-specific CD4+ T cells (HA-CD4sp) derived from 6.5 TCR-Tg mice (figure 1A). Indo was administered daily for 28 days by intraperitoneal injection at a dose of 1 mg/kg. Indo alone did not have any discernable impact on tumor growth compared with untreated mice (figure 1A,B, indo vs noTx). CTX alone transiently delayed tumor growth and moderately improved survival, but all mice succumbed to relapse. Indo administration following CTX (CTX+indo) did not improve tumor growth control or mouse survival compared with the CTX only group. Tumor growth inhibition was observed when CTX-conditioned mice were infused with HA-CD4+ T cell-containing splenocytes (CTX+HA-CD4sp), in which 20% of mice in this treatment group achieved complete tumor regression. Notably, indo administration in addition to splenocyte infusion (CTX+HA-CD4sp+indo) further improved tumor growth inhibition and led to complete tumor regression in 60% of mice. Lowering the indo dose to 0.3 mg/kg resulted in loss of the beneficial effect, and increasing the dose to 4 mg/kg resulted in toxicity in mice as reflected by severe weight loss (data not shown). Thereafter we used indo at 1 mg/kg dose for all in vivo studies.
Clinical application of ACT often involves reiterated stimulation of patient-derived T lymphocytes to expand and/or genetically modify them to generate the final infusion products. To model this scenario in mice, CD4+ T cells isolated from the splenocytes of TCR-Tg mice were expanded ex vivo with CD3/CD28 T-Activator Dynabeads. We performed a cell number titration experiment and found that adoptive transfer of 0.1 million activated HA-CD4+ T cells (HA-CD4act) to CTX-conditioned tumor-bearing mice resulted in significant tumor growth delay, but only occasional complete tumor regression. We used this suboptimal therapeutic condition in subsequent experiments to examine the impact of indo administration on ACT efficacy. Consistent with the results of ACT using TCR-Tg splenocytes, indo administration following adoptive transfer of activated HA-specific CD4+ T cells (CTX+HA-CD4act+indo) exhibited augmented antitumor activity, as reflected by significantly improved tumor growth inhibition and prolonged survival (figure 1C). It is noteworthy that indo administration at 1 mg/kg for 4 weeks did not cause weight loss in mice (figure 1D), indicating that treatment was well-tolerated at this dose. Moreover, the mice that had achieved complete tumor regression in the CTX+HA-CD4act+indo group were resistant to tumor rechallenge (figure 1E), suggesting the formation of immune memory.
Indo enhances the efficacy of CD19CAR T cell therapy in murine and human B-cell lymphoma models
We next determined if the beneficial effect of indo could be replicated in more clinically relevant CD19CAR T cell therapy models. Mouse total T cells were transduced to express CAR targeting murine CD19 (mCD19CART) (figure 2A). In vitro killing assays showed that indo pretreatment rendered A20 B-cell lymphoma cells more sensitive to the cytotoxicity of mCD19CAR T cells (figure 2B,C). To test if indo could augment the efficacy of mCD19CART in vivo, syngeneic BALB/c mice with established A20 tumors were conditioned with CTX followed by infusion of mCD19CAR T cells in the absence or presence of indo administration (figure 2D). As shown in figure 2D,E, indo by itself had no effect on tumor growth kinetics (indo vs noTx), nor did it improve the transient antitumor effect of CTX (CTX +indo vs CTX). However, the addition of indo significantly enhanced the efficacy of mCD19CART therapy (CTX +mCD19CART+indo vs CTX +mCD19 CART).
We also tested whether indo could improve the efficacy of human CD19CAR T cell therapy in a xenograft tumor model. Total T cells from the PBMC of healthy donors were transduced with a retroviral vector carrying human CD19-targeting CAR (hCD19CART) as previously reported.32 The transduction efficiency of hCD19CAR T cells was 50%–60% (online supplemental figure 1A), and the reactivity of transduced T cells to CD19-expressing human B-cell lymphoma Raji cells was verified in vitro (online supplemental figure 1B). For in vivo studies, Raji cells were subcutaneously implanted in the right flank of immunodeficient NSG mice. As shown in figure 2F, mice with established Raji tumors were divided into four groups to receive the indicated treatments: no treatment (noTx), indo, hCD19CART, or hCD19CART+indo. Indo alone did not alter tumor growth, and infusion of hCD19CART at the dose we used (0.7–0.9×106 CAR+ cells) had only modest antitumor effects; however, the combination of indo and hCD19CART markedly delayed tumor growth and improved mouse survival (figure 2F,G).
Although CD19CAR T cell therapy has become an effective treatment option for patients with B-cell malignancies, so far CAR T therapy has had limited success in treating solid tumors. We, therefore, tested if indo could potentiate CAR T therapy in a typical solid tumor model. For this purpose, mouse colorectal cancer CT26 cells were engineered to express murine CD19 (CT26.mCD19) (online supplemental figure 2A). CT26.mCD19 cells were implanted subcutaneously to the flank of BALB/c mice. Mice with established tumors received CTX-conditioning followed by infusion of a suboptimal dose of mCD19CART cells, in the absence or presence of indo administration (online supplemental figure 2B). The combination of indo and mCD19CART markedly improved mouse survival compared with mCD19CART alone (online supplemental figure 2B). Taken together, our results suggest that the beneficial effect of indo in ACT is not limited to the CD4+ T cell-based ACT model system, but may have broad implications for ACT including the use of genetically modified CAR T cells.
The beneficial effect of indo on ACT is independent of its COX2/PGE2 inhibitory activity
We sought to identify the molecular mechanism underlying the beneficial effect of indo on ACT. Indo is a nonselective COX inhibitor capable of blocking the production of PGE2. PGE2 in the TME can act to suppress antitumor immunity, and promote tumor progression, metastasis and chemoresistance.23 36 To test whether the beneficial effect of indo in ACT was due to its inhibition of the COX/PGE2 axis, we examined the intracellular level of COX2 in tumor cells by flow-cytometry analysis. Figure 3A shows that COX2 was barely detectable in A20 cells but highly expressed in CT26 cells. Correspondingly, PGE2 was marginally detected by ELISA in the supernatants from cultured A20 cells whereas it was abundant in CT26 cell culture supernatants (figure 3B). Our data showing that CT26 cells contain high levels of COX2 and PGE2 was in line with the findings of published studies,25 37 validating the robustness of our assays. Considering that PGE2 can be derived from tumor cells as well as stromal cells, we designed an experiment to measure the levels of PGE2 in the whole tumor mass. Using the HA-CD4act ACT model system as shown in figure 1, tumors were collected from mice under the specified treatment conditions. PGE2 in tumor tissues was quantified by LC-MS/MS. Figure 3C shows that PGE2 levels were marginally detected (5–40 ng/g) in all A20HA tumor tissues regardless of the treatment. In contrast, CT26 tumor tissues contained high levels of PGE2 (17933±9152 ng/g). Moreover, indo administration markedly reduced PGE2 in CT26 tumor tissues (1216±494 ng/g), reflecting the biological activity of indo in vivo. The presence of indo in tumors from indo-treated mice was confirmed and quantified by LC-MS/MS (figure 3D).
The low levels of PGE2 in A20 tumors as measured by LC-MS/MS analysis was in conflict with our initial expectation that indo-mediated PGE2 reduction led to enhance antitumor immunity in the lymphoma ACT models. To examine the impact of indo on T cells, we transferred T cells to CTX-conditioned tumor-bearing mice with or without indo administration and collected tumor tissues for FACS analysis at the specified time point (figure 3E). In the HA-CD4act ACT model, the frequencies of tumor-infiltrating donor T cells, and the percent distribution of naïve (CD44-CD62Lhigh), effector memory (CD44+CD62Llow) and central memory (CD44+CD62L+) subsets within the donor population were similar between the two groups (figure 3F). No differences were observed in the levels of T cell exhaustion markers PD-1 and Tim3 (figure 3G). In addition, donor T cells under the two conditions were comparable in producing functional molecules including IFNg, TNFa, IL2, GMCSF and GZMB (figure 3H). Furthermore, the frequencies of immunosuppressive cells, including Foxp3+ Tregs (in both donor and host CD4+ T cells) and CD11b+Gr1+ MDSCs, were also similar between the two conditions (figure 3I). Similar results were observed for mCD19CAR T cells used in the ACT model, with the exception that mCD19CAR T cells produced more IL2 and GMCSF when indo was given after CAR-T infusion, likely reflecting some inherent differences between CAR-transduced T cells and TCR-Tg T cells (online supplemental figure 3A). Consistent with the FACS analysis data, BLI-based T cell tracking experiments indicate that indo administration did not alter, that is, neither enhanced nor reduced, donor T cell tumor infiltration in this ACT model (figure 3J,K). Similar donor T cell tumor infiltration patterns were observed in the ACT model in which mCD19CART cells were used for adoptive transfer (online supplemental figure 3B,C).
Collectively, our data indicate that the beneficial effect of indo in potentiating ACT against B-cell lymphoma appears to be independent of its inhibitory effects on COX/PGE2, and that indo administration does not directly impact the expansion, migration, activation and function of the donor T cells.
Indo-induced oxidative stress upregulates DR5 expression in tumor cells in vitro
Many NSAIDs, including indo, can induce oxidative stress in cancer cells, leading to apoptosis.17–22 We confirmed that indo can induce oxidative stress, as indicated by increased ROS levels, in cultured mouse A20 lymphoma cells in a dose-dependent manner (figure 4A). Similar results were obtained for the human lymphoma cell line Raji (figure 4B), as well as mouse (CT26) and human (HT29) colon cancer cell lines (online supplemental figure 4). As expected, adding the antioxidant N-acetylcysteine (NAC) to the cultures neutralized indo’s ability to induce ROS (figure 4C). It has been reported that indo-induced oxidative stress in cancer cells can activate the cellular stress sensor C/EBP-homologous protein (CHOP), which upregulates DR5, sensitizing tumor cells to TRAIL-mediated cell death.20 We also found that A20 cells treated with indo had increased levels of molecules associated with oxidative/ER stress, including phosphorylated eIF2a (P-eIF2a), ATF4 and CHOP (online supplemental figure 5A). DR5 was constitutively expressed in A20 cells (online supplemental figure 5B). Indo treatment increased Dr5 transcripts, along with genes associated with oxidative/ER stress, including Nrf2, Ho-1, Atf4 and Chop (online supplemental figure 5C). Correspondingly, DR5 protein was found to be upregulated in A20 cells by indo in a dose-dependent manner (figure 4D). Enhanced expression of DR5 by indo was also observed in Raji cells (figure 4E). Indo-induced DR5 upregulation in A20 cells was abolished by the addition of NAC (figure 4F), implying the critical role of oxidative stress in enhancing DR5 expression. In addition, we found that indo treatment reduced the levels of cFLIP, MCL-1 and BCL2 in A20 cells (online supplemental figure 6A–C), which is consistent with published studies reporting that NSAIDs promote cancer cell apoptosis by downregulating prosurvival molecules.16
We asked if the increased expression of DR5, driven by indo, would augment apoptosis in cancer cells when triggered by its ligand. Published studies have reported that plate-coated MD5-1, a DR5 agonist antibody, can induce DR5 signaling-dependent apoptosis in cultured cells.34 38 Using this approach, A20 cells were cultured alone, or in the presence of indo and MD5-1 Ab, either alone or in combination. Figure 4G shows that the presence of MD5-1 or indo individually resulted in slight elevation of cell death in A20 cells, whereas the co-presence of MD5-1 and indo led to a more pronounced increase in cell death. To further define the role of DR5 in indo sensitization of A20 cells to MD5-1, we used CRISPR/Cas9 technology to create DR5-deficient A20 cells (A20.DR5KO) (figure 4H). A20.DR5KO cells were unresponsive to MD5-1 (figure 4I, DMSO +MD5-1 vs DMSO), confirming the loss of DR5 signaling in this mutant cell line. Importantly, indo could no longer sensitize A20.DR5KO cells to MD5-1-induced cell death (figure 4I, indo +MD5-1 vs DMSO +MD5-1 or indo). To confirm that the tumor sensitizing effect of indo was not restricted to A20 cells, we examined the cytotoxicity of indo and activated T cells, either alone or together, on luciferase-expressing 4T1 cells (4T1.Luci) by BLI. We found that the combination of indo and activated T cells exhibited enhanced cytotoxicity to 4T1 cells compared with either treatment alone (online supplemental figure 7). Since the T cells used in the culture were polyclonal, tumor cell killing by T cells was not antigen-driven but likely mediated via ligation of death receptors on tumor cells by their ligands on T cells. An enhanced engagement of DR5 and its ligand TRAIL between 4T1 cells and activated T cells may account for the improved cytotoxicity when indo and activated T cells were used together. Taken together, our results suggest that the increased expression of DR5, driven by the pro-oxidant activity of indo, can render cancer cells more susceptible to DR5 ligation-induced cell death.
In vivo evidence of indo-boosted DR5 expression in tumor cells and upregulation of TRAIL in donor T cells in the ACT setting
Based on our in vitro data, we posited that indo administration may boost DR5 expression in tumor cells, thereby sensitizing them to T cell cytotoxicity in the ACT setting. We first examined whether indo administration led to DR5 upregulation in tumor cells in vivo. For this purpose, A20 cells were implanted subcutaneously to CD45.1 mice so that A20 lymphoma cells (CD19+CD45.2+) could be unequivocally distinguished from the endogenous B cells (CD19+CD45.2-) at the time of FACS analysis. Tumor-bearing mice were subjected to four different treatment conditions and tumors were resected for FACS analysis 3 days after treatment (figure 5A). Figure 5B shows that A20 tumor cells recovered from CTX-conditioned, indo-administered mice had significantly increased DR5 expression compared with tumor cells from untreated or indo-treated mice, while DR5 expression in tumor cells from CTX-conditioned mice varied. The data show that indo administration following CTX-preconditioning consistently upregulates DR5 expression in tumor cells. Next, we examined whether the donor T cells were a source of TRAIL. To this end, HA-CD4act T cells were transferred to CTX-conditioned A20HA tumor-bearing mice. Five days later, spleens and tumors were collected to examine the levels of TRAIL in the donor CD4+ T cells as well as the endogenous CD4+ T cells by FACS. Figure 5C shows that TRAIL expression was significantly higher in tumor-specific donor CD4+ T cells than in bystander host CD4+ T cells. Similar results were obtained in the mCD19CART model in which mCD19CART cells exhibited higher TRAIL expression than endogenous T cells (figure 5D). These in vivo data support the possibility that indo administration enables enhanced TRAIL-DR5 engagement between donor T cells and tumor cells in the ACT setting.
The beneficial effect of indo is dependent on tumor cell DR5 expression and can be amplified by TRAIL overexpression in donor T cells
We postulated that if the beneficial effect of indo in ACT is due to enhanced TRAIL-DR5 engagement, then DR5-deficiency in tumor cells would nullify this effect. To test this, mice inoculated with A20 or A20.DR5KO tumor cells were treated with CTX followed by mCD19CAR T cell transfer, in the absence or presence of indo administration (figure 6A schema). Confirming our prior observation (figure 2D,E), indo administration significantly improved the effectiveness of mCD19CART in controlling the growth of A20 tumors (figure 6A, left graph). In contrast, the beneficial effect of indo on mCD19CART was abolished in mice bearing A20.DR5KO tumors (figure 6A, right graph). It is worth noting that A20 and A20.DR5KO tumors exhibited similar growth kinetics in response to mCD19CART in the absence of indo, indicating that the basal level of DR5 in A20 tumor cells, even if ligated by TRAIL from the CAR T cells, is insufficient to cause tumor cell death to an extent that confers discernable therapeutic benefit. In addition, A20.DR5KO and A20 tumors grew at similar rates in untreated mice (data not shown), suggesting that DR5 deficiency per se does not alter tumor growth kinetics in mice. Together, these data indicate that the beneficial effect of indo in the context of ACT relies on tumor cell DR5 expression.
We reasoned that enhancing TRAIL expression in donor T cells may lead to more effective engagement with indo-boosted DR5 in tumor cells to intensify tumor cell apoptosis, resulting in further improvement in therapeutic outcomes. To test this, HA-CD4act T cells were virally transduced to overexpress TRAIL (TRAIL.HA-CD4act) at the time of transfer (figure 6B). Notably, indo administration following TRAIL.HA-CD4act T cell transfer exhibited markedly improved survival benefit, with over 50% of mice achieving complete tumor regression compared with less than 30% complete tumor regression in mice that received HA-CD4act and indo (figure 6C). Interestingly, TRAIL.HA-CD4act T cell transfer in the absence of indo did not result in improvement of mouse survival compared with HA-CD4act T cell transfer, indicating that the enhanced expression of DR5, boosted by indo, is essential for sensitizing cancer cells to TRAIL-expressing T cells. Mice showing complete tumor regression after receiving the TRAIL.HA-CD4act and indo combination treatment were rechallenged with A20HA tumor cells. Seven out of nine mice were completely protected from tumor growth, and the remaining two mice had significantly lower tumor burden compared with tumor-challenged naïve mice, indicating the development of immune memory (figure 6D).
Here, we demonstrate that indo administration can augment the efficacy of ACT against mouse and human B-cell lymphomas using T cells expressing a tumor-specific TCR or CD19CAR. Mechanistically, our data reveal that the ability of indo to potentiate ACT relies on its pro-oxidative activity that leads to enhanced engagement of the TRAIL-DR5 axis between tumor-reactive T cells and tumor cells. To our knowledge, our study provides the first in vivo evidence that the use of NSAID such as indo is compatible with and can benefit T cell-based adoptive immunotherapy.
Multiple lines of evidence have documented that the antitumor activity of NSAIDs can be multifaceted, likely involving inhibition of COX and blockage of PGE2 synthesis, as well as other COX-independent mechanisms, such as suppression of Wnt/β-catenin signaling, downregulation of survival molecules, induction of oxidative/ER stress, and fostering an immunogenic milieu in the TME. As an inflammatory mediator, PGE2 is known to promote tumor progression, metastasis, angiogenesis and chemoresistance.23 36 PGE2 can induce Treg cells and MDSCs, thereby exerting potent immunosuppression on tumor-reactive T cells and NK cells.23 24 PGE2 can be produced by tumor cells as well as stromal cells, such as myeloid cells and fibroblasts.39 40 It has been shown in various mouse tumor models that inhibition of COX2 or PGE2 can mitigate immunosuppression and promote antitumor immunity.23 24 Suppression of COX/PGE2 is well-established to be the major mechanism by which NSAIDs, such as celecoxib and aspirin, enhance the efficacy of anti-PD-1 therapy in multiple mouse tumor models.25 26
Here, we show that A20 tumor cells barely express COX2 and produce a negligible amount of PGE2. Moreover, PGE2 was not detected in the tumor masses resected from mice implanted with A20HA tumors. Our in vivo studies show that indo administration alone marginally altered tumor growth and mouse survival compared with untreated mice in the A20HA (figure 1B,C) and A20 (figure 2E) models. We also did not detect PGE2 in human B-cell lymphoma Raji cell culture (data not shown), consistent with the results of a previous report showing that human B-cell lymphoma cells do not produce PGE2.39 In accordance with what we observed in the murine lymphoma model, indo administration alone did not delay Raji growth in NSG mice, nor did it improve mouse survival (figure 2F,G). These results indicate that the COX/PGE2 axis does not contribute to tumor progression in the B-cell lymphoma models used in our study. Therefore, the beneficial effect of indo in ACT is unlikely attributable to COX inhibition and PGE2 reduction. The COX-independent ACT-potentiating effect of indo may have important implications for repurposing and expanding the use of NSAIDs in cancer immunotherapy. A variety of other NSAIDs besides indo can be tested to identify those that are most effective and least toxic in the setting of ACT. In addition, novel non-COX inhibitory derivatives of NSAIDs can be developed to suit the needs of ACT while reducing the potential adverse effects associated with COX inhibition, such as gastrointestinal, renal, and cardiovascular toxicities.41
Although our study underscores the importance of the TRAIL-DR5 axis in indo-mediated enhancement of ACT efficacy, our results do not exclude other mechanisms by which indo may contribute to an improved therapeutic outcome. It is conceivable that for tumors with high levels of COX/PGE2, indo administration can inhibit this pro-tumorigenic signaling pathway and thus exert direct antitumor effects. Meanwhile, indo may mitigate PGE2-driven immunosuppression in the TME, thereby conditioning a favorable immune milieu for ACT. Moreover, indo-mediated downregulation of prosurvival molecules, including c-FLIP, MCL-1, BCL2 and survivin may further promote tumor eradication. These distinct but mutually reinforcing mechanisms may synergize with ACT to achieve substantially improved therapeutic outcomes, even for hard-to-treat solid tumors, which often express high levels of COX/PGE2.
The initial finding that ligation of death receptors (DR4 and DR5) by their ligand TRAIL leads to selective death in cancer cells incentivized the development of recombinant human TRAIL (rhTRAIL) and death receptor agonist antibodies as cancer therapeutics. Despite the promising data from numerous preclinical studies, results from clinical studies showed limited efficacy of these reagents as monotherapy.42–47 The TRAIL or death receptor-based therapies have limitations related to the short half-life of rhTRAIL, heterogeneous expression of death receptors in cancer, weak agonistic potential of anti-DR4 or anti-DR5 antibodies, and acquisition of resistance by cancer cells. There is emerging evidence that engagement of the TRAIL-DR5 axis can be attained through cellular therapy to drive a robust antitumor effect. It has been shown that mice reconstituted with TRAIL-transduced bone marrow cells were less prone to tumor growth following engraftment.48 Using a murine model of allogeneic hematopoietic stem cell transplantation, Ghosh et al reported that infusion of TRAIL-overexpressing T cells can achieve a curative result in tumor-bearing mice by increasing graft-versus-tumor response while suppressing graft-versus-host disease.49 With direct relevance to adoptive T cell therapy using CAR T cells, DeSelm et al reported that TRAIL produced by CAR T cells on recognizing the targeted antigen on tumor cells can eliminate antigen-negative tumor cells previously exposed to low-dose radiation.50 Moreover, two recent CRISPR-based screening studies demonstrated that tumor-intrinsic death receptor signaling critically regulates CD19CAR T cell cytotoxicity and functional exhaustion.51 52 Based on these published studies and our current data, we put forth the hypothesis that simultaneous upregulation of TRAIL in donor T cells and DR5 in cancer cells can further magnify cell death in cancer cells. Our data suggest that ectopic expression of TRAIL on the surface of donor T cells has the advantage of providing a reliable source of TRAIL to effectively crosslink DR5 in a localized manner. Overexpression of TRAIL in tumor-specific T cells, especially those engineered to express CARs or TCRs, can be readily achieved during the T cell genetic modification process.
One potential concern for enhancing TRAIL-DR5 engagement in the ACT setting is whether fratricide will occur among donor T cells. Although others have shown that activated T cells, including CAR T cells, are prone to DR5-mediated cell death,53 54 we did not observe any obvious detrimental effects on donor T cells when mice received indo treatment or when donor T cells were genetically modified to overexpress TRAIL. The exact mechanism of the differential response of tumor cells and donor T cells to indomethacin is not clear. It is possible that cancer cells are more sensitive than T cells to indomethacin at a certain dose range because cancer cells are known to have higher basal levels of endogenous oxidative stress as compared with normal cells.6 From a practical perspective, if TRAIL-induced apoptosis in T cells does become a concern, this issue can be addressed by engineering T cells to overexpress a dominant negative form of DR5 to disrupt DR5 signaling, or an antioxidant gene such as catalase to mitigate oxidative stress. In this regard, it has been reported that T cells engineered to coexpress CAR and catalase exhibit potent antitumor activity.55
In summary, our study demonstrates that the pro-oxidative property of indo can be exploited to enhance death receptor signaling in cancer cells, providing rationale for combining indo with genetically modified T cells to intensify tumor cell killing through the TRAIL-DR5 axis. These findings implicate indo administration, and potentially similar use of other NSAIDs, as a readily applicable and cost-effective approach to augment the efficacy of ACT.
Data availability statement
Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as online supplemental information.
Patient consent for publication
We thank Drs Yukai He, Tohru Fukai, Rafal Pacholczyk and Tsadik Habtetsion for helpful discussion and advice. We thank Jeanene Pihkala at the Georgia Cancer Center Flow Cytometry Core for assisting data collection and analysis. We thank the Georgia Cancer Center Biorepository for providing human peripheral blood mononuclear cells. We appreciate the technical assistance from Drs. Huabin Zhu and Nagendra Singh. This work was supported by National Institutes of Health grant CA215523, CA238514 and CA264983 to GZ.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Contributors NSA contributed to research design, performed research, analyzed data and wrote the manuscript; CB, OO, MYG, Z-CD and GG performed experiments; RB and MLD provided critical reagents; HX, LB, DM and GP analyzed data and edited the manuscript. GZ conceived the study, designed research, analyzed data, wrote the manuscript and acts as guarantor.
Funding This work was supported by National Institutes of Health grant CA215523, CA238514 and CA264983 to GZ.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Author note Study approval: All animal experiments and procedures were performed in accordance with the institutional protocol and were approved by the Institutional Animal Care and Use Committee (IACUC) of Augusta University.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.