IFN-γ surmounts PD-L1/PD1 inhibition to CAR-T cell therapy by upregulating ICAM-1 on tumor cells

Dear Editor, Chimeric antigen receptor modified T (CAR-T) cell therapy has shown potent antitumor activity against relapsed and refractory hematological malignancies. However, its efficacy in solid tumors is limited, partly because of the inhibition of PD-L1/PD-1 signaling on CAR-T cells in solid tumors. Further optimizations of CAR-T cells by disrupting PD-1 signaling have improved the anti-tumor efficacy of CAR-T cells. Here, we report an alternative approach that sensitizes tumor cells by interferon (IFN)-γ, but without modifying T cells; this strategy surmounts PD-1 inhibition and markedly enhances CAR-T anti-tumor activities in vitro and in vivo. We constructed two CAR-T cells containing CD28 or 4-1BB costimulatory receptor intracellular domains in tandem with CD3ζ that target the tumor antigens of HER2 and mesothelin (MSLN) (Supplementary Fig. S1a). Both HER2 and MSLN CAR-T cells exhibited high rates of cytotoxicity to targeted tumor cells, with robust secretion of IFN-γ and interleukin (IL)-2 (Supplementary Fig. S1b, c) and dramatically increased expression of effector activity surface molecule CD107a and intracellular granzyme B (Supplementary Fig. S1d). PD-L1 was significantly upregulated on tumor cells, and coinhibitory molecules including PD-1, TIM-3, and LAG-3 were markedly elevated on CAR-T cells (Supplementary Fig. S2a–c). Moreover, the recursive antigen encountering of HER2 CAR-T cells, mimicking the in vivo antitumor milieu where CAR-T cells serially encounter and kill tumor cells (Supplementary Fig. S3a), showed high cytotoxicity and IFN-γ release (Supplementary Fig. S3b), despite high PD-L1 expression on almost all tumor cells (Supplementary Fig. S3c). These results indicated the existence of an intrinsic mechanism of CAR-T cells that surmounts the inhibitory effects of PDL1/PD-1 during tumor cell killing. IFN-γ is an important inducer of PD-L1 expression on tumor cells. To evaluate the effects of IFN-γ-induced PD-L1 expression on CAR-T cytotoxicity, tumor cells were primed with IFN-γ for 24 h before exposure to CAR-T cells. IFN-γ concentration was optimized as <10 ng/ml to ensure induction of PD-L1 expression without inhibiting tumor cell growth (Supplementary Fig. S4a, b). Notably, however, CAR-T cells showed more potent cytolytic activity and IFN-γ secretion when exposed to IFN-γ-primed tumor cells with high expression of PD-L1 (Fig. 1a, b). Conversely, when IFN-γ was neutralized by anti-IFN-γ antibody, the cytotoxicity of CAR-T cells was significantly attenuated (Fig. 1c), which indicates that IFN-γ was critical to CAR-T activities. Notably, in these tests, the medium was replaced with fresh medium to eliminate residual IFN-γ (<10 pg/ml, data not shown), excluding the direct effect of IFN-γ on CAR-T cells. We thus speculated that IFN-γ functioned to CAR-T activity by acting on tumor cells. To examine this possibility, we first primed tumor cells with low concentrations of IFN-γ and measured the cytotoxicity of CAR-T in fresh medium containing sufficient anti-IFN-γ antibody to neutralize IFN-γ from CAR-T (Fig. 1d). CAR-T cells still exhibited similar killing activity to IFN-γprimed tumor cells when IFN-γ was neutralized (Fig. 1e). Second, we disrupted the IFN-γR2 gene (IFNGR2) in tumor cells using the CRISPR-Cas9 system to block IFN-γ signaling (Supplementary Fig. S5a–c), because IFN-γR2 has been shown to determine the extent of IFN-γ-induced signaling in given cell populations. IFN-γR2-null tumor cells grew similarly to control cells (Supplementary Fig. S5d) and did not express PD-L1 any more under IFN-γ treatment (Supplementary Fig. S5e). However, these cells were more resistant to the cytolysis of CAR-T cells (Fig. 1f). The enhancement of CAR-T cytotoxicity by IFN-γ priming was also significantly diminished as shown by two different target tumor cell lines (Fig. 1g). These results indicated that IFN-γ on tumor cells is involved in the intrinsic mechanism of CAR-T to circumvent PD-L1/PD-1. To directly characterize the function of IFN-γ in surmounting PD-L1/PD-1, we tested CAR-T activities on tumor cells that stably overexpress PD-L1. Overexpression of PD-L1 on tumor cells (Supplementary Fig. S6) did not render tumor cells resistant to CAR-T killing (Fig. 1h), and also did not impair the enhancement of CAR-T cytotoxicity to IFN-γ-primed tumor cells (Fig. 1i). Notably, however, PD-L1 overexpression in IFN-γR2-null tumor cells significantly suppressed both cytolytic activity (Fig. 1j) and secretion of IFN-γ and IL-2 (Fig. 1k). These results indicated that IFN-γ signaling on tumor cells enhances CAR-T activities by surmounting the inhibitory effects of PD-L1/PD-1. In contrast, the inhibitory function of PD-L1/PD-1 on CAR-T activity requires the absence of IFN-γ signaling in tumor cells. This is consistent with the previous study that defects of the interferon signaling pathway are associated with acquired resistance to immune checkpoint blockade therapy. We next investigated how IFN-γ surmounts PD-L1/PD-1 inhibitory effects on CAR-T. IFN-γ induces complex molecular events beyond PD-L1 overexpression in tumor cells. We first examined whether IFN-γ increased HER2 expression. However, HER2 expression was decreased in IFN-γ-primed tumor cells (Supplementary Fig. S7), which indicates that HER2 was not involved in IFN-γ-mediated effects on CAR-T. We found that IFN-γ induced intercellular adhesion molecule 1 (ICAM-1, also known as CD54) overexpression on multiple kinds of solid tumor cells (Supplementary Fig. S8a), and LFA-1 (the β2 integrin adhesion molecule) was significantly elevated on CAR-T cells after tumor cell exposure (Supplementary Fig. S8b). ICAM-1 is a cell surface glycoprotein on antigen-presenting cells (APCs) and plays a critical role during an effective immune response. The interaction of antigen-laden APCs with CD8 T cells is mediated by the interaction of ICAM-1 to its receptor LFA-1 on T cells. The ICAM-1/LFA-1 interaction promotes efficient adhesion of T cells with APCs and transmits intracellular signals to promote T-cell activation and proliferation. We then investigated the function of ICAM-1 on CAR-T activity. Knockout of ICAM-1 (Supplementary Fig. S9) in tumor cells with functional IFN-γ receptor almost completely abolished the increase of CAR-T cytotoxicity to IFN-γprimed tumor cells (Fig. 1l), which indicates the critical role of


Cell lines
The human epithelial ovarian cancer cell line SK-OV-3, lung cancer cell line A549 and lentivirus packaging cell line HEK 293TD were obtained from American Type Culture Collection (ATCC;Manassas, VA, USA). All the cell lines were cultured in Dulbecco's modified Eagle's medium (thermo fisher scientific) supplemented with 10% heat-inactivated FBS and 1% penicillin and streptomycin. All cell lines were maintained in a humidified incubator with 95% air and 5% CO 2 at 37°C.

DNA constructs and lentivirus production
The extracellular fragment was composed of HER2-specific scFv from HER2, linked to human CD8α hinge region, transmembrane domain of the CD8α. Intracellular region included 4-1BB cytoplasmic domain and CD3ζ molecule, simply called HER2BBZ (HER2-CAR) was cloned into lentivirus-vector pWPXLd. Another CAR targeting to mesothelin with CD28 cytoplasmic and CD3ζmolecule was termed as MSLN28Z(MSLN-CAR), all components were cloned into lentivirus-vector pWPXLd. The subsequent lentivirus production, concentration and quantification were completed in our lab. Tumor cells were engineered to express enhanced firefly luciferase (ffluc) by transduction with lentivirus carrying the ffluc gene under the control of the CMV promoter. To construct a recombinant eukaryotic vector bearing the gene of human ICAM-1(CD54), the vector pWPXLd was used to construct pWPXLd -ICAM-1 using the restriction sites BamHI and EcoRI. The vector pCDH expressing the human PD-L1 gene was cloning by EcoRI and XbaI restriction site. Knockout cell lines were generated using the CRISPR-Cas9 system. To generate knockout IFNGR2，ICAM-1 cells, cells were transduced with pLentiCRISPRV2 vector (Addgene) encoding the annealed double-stranded single-guide RNAs(sgRNAs) targeting IFNGR2 and ICAM-1 by BsmB I (NEB). The sgRNA sequence TCGCCTGTACAACGCAGAGC(sgRNA IFNGR2#1) and GGACCTGCTCTGCGT TGTAC(sgRNA IFNGR2#2) were used for generating IFNGR2 KO cells. The sgRNA sequences GCGGCTGACGTGTGCAGTAA(sgRNA ICAM-1#1) and CAACTTGTC AGCCCCCGGGT(sgRNA ICAM-1#2) were used to generate bulk populations of ICAM-1 KO cells. Lentivirus was generated as previously described. [1] Briefly, 293T cells were transfected with packaging plasmid psPAX2, PMD2.0G (Invitrogen)and CAR lentiviral backbone plasmid using a modified calcium phosphate method. Viral supernatants were collected after transfection 48 and 72 hours and concentrated through ultracentrifugation at 70,000g for 120 minutes. Lentiviral pellets were resuspended in phosphate-buffered saline (PBS)-lactose solution (4 g lactose per 100 mL PBS), aliquoted and stored at −80°C. Lentiviral titers were quantified using 293T cells based on CAR expression.

T cell isolation, lentiviral transduction, and ex vivo expansion
Human peripheral blood mononuclear cells (PBMCs) were isolated from consented research participants (healthy donors) under protocols approved by Ethics Committee of the State Key Laboratory of Biotherapy. On the day of leukapheresis, peripheral blood mononuclear cells (PBMC) were isolated by density gradient centrifugation over Ficoll-Paque (GE Healthcare) followed by multiple washes in DPBS (thermo fisher scientific). Freshly isolated PBMC were then cultured in complete X-VIVO 15 medium (Lonza) containing 100 U/mL recombinant human IL-2 (peprotech) and 5% human AB serum (Sigma-Aldrich) and stimulated with anti-CD3 and anti-CD28 magnetic beads for 24-36 hours (Life Technologies). For CAR lentiviral transduction, desired lentivirus at a multiplicity of infection of 5-10 were added to each retronectin-coated well plates and centrifuged at 1000g for 2 hours at 32 . After centrifugation, activated T cells were added to each well and incubated at 37 with 5% CO 2 . Cells were then cultured in and replenished with fresh complete X-VIVO containing cytokines every 2-3 days. After 7 days, beads were magnetically removed, and cells were further expanded in complete X-VIVO containing cytokines to achieve desired cell yield.

T7EN1 cleavage assay and sequencing
Cells were harvested and digested with 100 µg/ml Proteinase K in lysis buffer (10 µM Tris-HCl, 0.4 M NaCl, 2 µM EDTA and 1% SDS). Genomic DNA was extracted by phenol-chloroform and alcohol precipitation. The T7EN1 cleavage assay was performed as follows: briefly, targeted regions of ICAM-1 and IFNGR2 were PCR-amplified from genomic DNA using 2 × Taq Master Mix (Vazyme) and the products were purified with a PCR cleanup kit (Axygen). Purified PCR product was denatured and re-annealed in NEBuffer 2 (NEB) using a ProFlex 3 x 32-Well PCR System (Applied Biosystems). Hybridized PCR products were digested with T7EN1(NEB) for 30 min and separated by 2% agarose gel. Primers for PCR are listed in Supplementary Table1.

Western blot analysis
Cells were harvested and lysed with RIPA lysis buffer containing a protease inhibitor cocktail (Selleck Chemicals). The lysates were centrifuged for 10 min at maximum speed at 4 °C and the supernatants were mixed with sample loading buffer (5×) and boiled for 5 min. For immunoblotting, proteins from whole-cell lysates were resolved by 10 or 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to PVDF membranes. The membrane was blocked for 1 h at 25 °C using 5% non-fat milk and incubated in the presence of anti-IFNGR2 overnight at 4 °C with gentle rocking. The membrane was then washed three times with TBS-T buffer and incubated with horseradish peroxidase-conjugated secondary anti-mouse IgG antibody in 5% non-fat milk TBS-T buffer for 1 h at 25 °C. The membrane was washed three times with TBS-T buffer. PVDF membranes were then exposed for an enhanced chemiluminescence assay using the Gel imaging instrument (BIO-RAD ChemiDoc MP). α-Tublin was used as a loading control. The gray value of all the bands is measured by ImageJ.

Quantitative PCR
Total mRNA from cultured T cells was isolated by RNeasy Mini Kit (Qiagen Inc). cDNA was then synthesized via PrimeScript 1st Strand cDNA Synthesis Kit (Takara Bio Inc). Quantitative PCR was performed using SYBR Green PCR Master mix (Applied Biosystems) in a reverse transcription PCR (RT-PCR) system (Thermo Fisher Scientific). The comparative Ct values of genes of interest were normalized to the Ct value of β-actin. Then, the 2-Δct method was used to determine the relative expression of the genes, while the 2-ΔΔct method was used to calculate the fold changes of gene expression over control.

In vitro tumor killing and T cell functional assays
For tumor killing assays, CAR T cells and tumor targets were co-cultured at effector:tumor (E:T) ratios of 1:5 or 1:10 in X-VIVO medium in the absence of exogenous cytokines in 48 or 96-well in a final volume of 400ul or 200 µl per well plates for 24-72 h and analyzed by flow cytometry as described above. All samples were set in triplicate. Tumor cells were plated overnight in the absence or presence IFN-γ prior to addition of T cells. Next, 50 µl of supernatant per well was collected to measure LDH release using a cytotoxicity LDH Assay Kit (Promega) according to the manufacturer's instructions. The cell lysis percentage was calculated as follows: Cytotoxicity (%) = (Experimental -Effector Spontaneous-Target Spontaneous)/(Target Maximum-Target Spontaneous) × 100.
To test for degranulation and cytotoxicity, CAR T cells were incubated with tumor cells for 5 hours in the presence of CD107a antibody and Golgistop protein transport inhibitor (BD Biosciences). After the coculture, cells were harvested, fixed, permeabilized, and stained for intracellular staining. Degranulation (CD107a staining) and intracellular marker staining (Granzyme B) were examined by flow cytometry as described above.
For T cell activation and inhibition assays, CAR-T cells and tumor targets were co-cultured in X-VIVO medium in the absence of exogenous cytokines in 48-well plates for the indicated time points and analyzed by flow cytometry for specific markers of T cell activation.

Stress test
To mimic recursive antigen encounters, we started a coculture of adherent target cells and freshly nonadherent effector cells at E:T=1:10. After 24 (round I) and 48(round II) hours, nonadherent effector cells were moved to a new set of adherent target cells at an indicated E:T ratio.

Neutralization assays
HER2-positive SK-OV-3 tumor cells were seeded at wells of a 96-well U-bottom plate at 1 × 10 4 cell/200 µl and cultured overnight with or without IFN-γ primed. The next day the media was removed and cells were washed once with PBS. HER2 CAR T cell with fresh medium in the presence of IFN-γ neutralizing antibodies or isotype were then added to the tumor cells, the percentage of cytotoxicity of CAR-T cell was calculated as described above. The supernatants were harvested after 24h cultured and performed the downstream experiment.

Elisa cytokine assays
Supernatants from tumor killing assays were collected at indicated times and frozen at -20°C for further use. The cell-free supernatant was diluted and analyzed for IFN-γ and IL-2 production using a Human IFN-γ ELISA Ready-SET-Go Kit (eBioscience) in accordance with the manufacturer's protocol. Plates were read at 450 nm using a Microplate Reader of Multiskan Ascent (Thermo Scientific) and ascent software 2.06. Values represent the mean of triplicate wells.

Xenograft model of ovarian cancer
B-NSG (NOD-Prkdcscid Il2rgtm1/Bcgen) mice were obtained from Beijing Biocytogen Co.,Ltd, 5-8 week-old NSG mice were maintained under pathogen-free conditions in-house of Sichuan University. In vivo experiments were performed in accordance with the Institutional Animal Care and Use Committee of the State Key Laboratory of Biotherapy, Sichuan University. For in vivo tumor studies, SK-OV-3-Luc(WT), SK-OV-3(IFNGR2 KO) (0.2 million/mouse) were prepared in a final volume of 200 µl medium and engrafted in 6-8 weeks old female B-NSG mice by intraperitoneal (i.p.) injection and allowed to established tumor for 6 days. Tumor engraftment was monitored by bioluminescence imaging and mice were assigned to treatment groups so that each group possessed equal average tumor luminescence, then received i.p. adoptive transfer of 2 million HER2 CAR-T cells per mouse and IFN-γ(20ug×2) every other day prior to CAR-T cell. Tumor growth was monitored at least once a week via IVIS Bioluminescence imaging (PerkinElmer, Inc., Waltham, MA, USA) and flux signals were analyzed with Living Image software (PerkinElmer). For imaging, mice were i.p. injected with 200 µL D-luciferin potassium salt (Perkin Elmer) suspended in PBS at 3 mg/mouse. Mice were also monitored for survival and sacrificed when the total luminescence of luciferin-treated subject exceeded 1 × 10 11 photons/s.

Statistics
Data were presented as mean ± SD. For comparisons between two groups, Student's unpaired two tailed t test were employed. p < 0.05 was regarded statistically significant.