PD-L1 CAR effector cells induce self-amplifying cytotoxic effects against target cells

Background Immune checkpoint inhibitors and chimeric antigen receptor (CAR)-based therapies have transformed cancer treatment. Recently, combining these approaches into a strategy of PD-L1-targeted CAR has been proposed to target PD-L1high tumors. Our study provides new information on the efficacy of such an approach against PD-L1low targets. Methods New atezolizumab-based PD-L1-targeted CAR was generated and introduced into T, NK, or NK-92 cells. Breast cancer MDA-MB-231 and MCF-7 cell lines or non-malignant cells (HEK293T, HMEC, MCF-10A, or BM-MSC) were used as targets to assess the reactivity or cytotoxic activity of the PD-L1–CAR-bearing immune effector cells. Stimulation with IFNγ or with supernatants from activated CAR T cells were used to induce upregulation of PD-L1 molecule expression on the target cells. HER2–CAR T cells were used for combination with PD-L1–CAR T cells against MCF-7 cells. Results PD-L1–CAR effector cells responded vigorously with degranulation and cytokine production to PD-L1high MDA-MB-231 cells, but not to PD-L1low MCF-7 cells. However, in long-term killing assays, both MDA-MB-231 and MCF-7 cells were eliminated by the PD-L1–CAR cells, although with a delay in the case of PD-L1low MCF-7 cells. Notably, the coculture of MCF-7 cells with activated PD-L1–CAR cells led to bystander induction of PD-L1 expression on MCF-7 cells and to the unique self-amplifying effect of the PD-L1–CAR cells. Accordingly, PD-L1–CAR T cells were active not only against MDA-MD-231 and MCF-7-PD-L1 but also against MCF-7-pLVX cells in tumor xenograft models. Importantly, we have also observed potent cytotoxic effects of PD-L1–CAR cells against non-malignant MCF-10A, HMEC, and BM-MSC cells, but not against HEK293T cells that initially did not express PD-L1 and were unresponsive to the stimulation. Finally, we have observed that HER-2–CAR T cells stimulate PD-L1 expression on MCF-7 cells and therefore accelerate the functionality of PD-L1–CAR T cells when used in combination. Conclusions In summary, our studies show that CAR-effector cells trigger the expression of PD-L1 on target cells, which in case of PD-L1–CAR results in the unique self-amplification phenomenon. This self-amplifying effect could be responsible for the enhanced cytotoxicity of PD-L1–CAR T cells against both malignant and non-malignant cells and implies extensive caution in introducing PD-L1–CAR strategy into clinical studies.

PD-L1 and CD206 expression in macrophages was assessed after 4 days of differentiation. T cells or PD-L1-CAR T cells conditioned supernatants from the 24-hour co-incubation with MDA-MB-231 cell line were placed instead of the medium for 48 h to assess the impact of these supernatants on induced PD-L1 expression. The cells were fixed with 4% paraformaldehyde (15 min, at 25°C). After washing, the cells were incubated with blocking buffer consisting of PBS with 2.5% normal donkey serum and 1% bovine serum albumin for 30 min (at 25°C), followed by overnight incubation (at 4 o C) with primary goat anti-CD206 antibody ( The images were analyzed using Gen5 3.04 software (BioTek, RRID:SCR_017317). The mean fluorescence intensity within secondary masks was measured. The secondary mask was designed around each DAPI stained nucleus (which constituted a primary mask).

RNA isolation and quantitative real time PCR
Total RNA was isolated using RNeasy kit (Qiagen). RNA concentration was measured with NanoDrop 2000 spectrophotometer (Thermo Fischer Scientific, Waltham, MA, USA) and 500 ng of RNA was BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)

Fluorescence Activated Cell Sorting
NK-92 cells from 80% confluent culture flasks were centrifuged and resuspended in PBS in a final concentration of 5 × 10 6 cells/ml. Then, the cells were transferred through a cell strainer to ensure separation of the cells. The sorting was performed according to surface CAR expression with FACS Aria III cell sorter (BD Biosciences, La Jolla, CA, USA).

CAR construction
PD-L1-targeting CAR synthesis was performed by Creative Biolabs, and the design was as follows:

Lentiviral and retroviral T cells modifications
PBMC isolated from buffy coats were seeded onto the 6-well plate at a cell density 1 × 10 6 cells per 1 ml in 3 ml RPMI full medium and stimulated using 2.4 ug/ml of PHA-L for 3 days. To produce control or CAR viral particles, HEK-293T cells were seeded at 10 cm plates -4 plates per virus and transfected using a polyethyleneimine (PEI) transfection protocol to deliver a transfer plasmid with gene of interest simultaneously, VSV-G envelope expressing plasmid pMD2.G (a gift from Didier Trono; RRID:Addgene_12259) and lentiviral packaging plasmid psPAX2 (a gift from Didier Trono; RRID:Addgene_12260). After 48 h, the lentivirus-containing supernatant was harvested, filtered through a 0.45 μm pore size filter, and concentrated by overnight centrifugation at 2 500 × g at 4°C. The culture medium from the stimulated PMBC was replaced with concentrated lentiviral supernatant, and 4 µg/ml of polybrene (Sigma-Aldrich) was added. After 1 h of spinoculation round (2 000 × g at 32°C), the PMBC were kept in a humidified atmosphere containing 5% carbon dioxide (CO2).
The next day, the viral supernatant was replaced with the fresh portion of complete culture RPMI-1640 medium supplemented with 200 U/ml IL-2 (PeproTech) and Dynabeads Human T-Activator CD3/CD28 (Thermo Fisher Scientific) at a bead-to-cell ratio of 1:5). In the case of retroviral transduction, 2 x 10 cm plates of Phoenix Ampho cells were seeded per virus, and a gene of interest was co-transfected together with pHIT60 and pCOLT-GALV plasmids [4], a kind gift of Tuna Mutis (VUmc the Netherlands).
Retroviral supernatant was collected, spun down at 500 x g for 5 min, and concentrated using Retro-X-Concentrator (Takara). Stimulated PBMC's were transferred on the 50 µg/ml Retronectin (Takara) coated plates, and concentrated retroviral supernatant was applied. After 1 h of spinoculation round (2 000 × g at 25°C), the PMBC were kept in a humidified atmosphere containing 5% carbon dioxide (CO2). The spinoculation with the second batch of viral supernatant was repeated on the next day.
Similar to lentiviral transduction, on the following day, the viral supernatant was replaced with the fresh portion of complete culture RPMI-1640 medium supplemented with 200 U/ml IL-2 (PeproTech) and Dynabeads Human T-Activator CD3/CD28 (Thermo Fisher Scientific) at a bead-to-cell ratio of 1:5.
The CAR expression on the surface of the T cells was evaluated by flow cytometry 48-72 h after transduction.

In vitro mRNA synthesis and mRNA electroporation
For CAR mRNA synthesis in vitro, the PD-L1-CAR construct was subcloned into pCIpA102 plasmid (a kind gift of Stein Saebøe-Larssen from OUS). Linearized plasmid was cleaned up using QIAquick PCR Purification Kit (Qiagen) and was utilized as a template in mRNA synthesis reaction (Promega RiboMax Large Scale). ARCA (Anti Reverse Cap Analog) was used in the mRNA synthesis reaction to ensure proper cap orientation and by that 100% mRNA activity. Produced mRNA in concentration 1 µg/µl was were mixed with 20 or 40 µl (1 µg/µl) of previously synthesized mRNA or water (for MOCK control).
After the electroporation (500 V, 2 ms, squarewave), cells were transferred onto a 6-well plate containing 2 ml of pre-warmed complete medium supplemented with 20 ng/ml of IL-15 for NK cells.
After 18 h, the PD-L1-CAR expression was evaluated by flow cytometry.

CAR surface staining
To assess CAR surface expression, the effector cells (T, NK, or NK-92 cells) were washed 3 times with EasySep buffer (PBS supplemented with 2% FBS and 1 mM EDTA) and stained using goat anti-human

Generation target cell lines for luciferase based killing assays
HEK293T cells were modified with plasmid pLenti7.3/V5 TOPO-RedLuc encoding the red luciferase gene and green fluorescent protein, as described previously [5].

Preparation of T cells/NK-92 parental cells or PD-L1-CAR-T/NK-92 cell supernatants
MDA-MB-231 parental or sgPD-L1 breast cancer cells were seeded onto a 6-well plate at a cell density 1 × 10 6 mln per well in 2 ml full RPMI-1640 medium and left overnight to attach. Next, the medium was removed and replaced with either unmodified or PD-L1-CAR T or NK-92 parental or PD-L1-CAR NK-92 cells. The effector cells were added at a 1:1 E:T ratio and co-incubated for the next 24 h. Afterward,

RTCA-based killing assay
For RTCA assay, adherent target cells were seeded onto 16-well E-Plates (ACEA Biosciences) at a cell Results obtained from individual mice or mesurement means according to the treatment method are presented. The blinding was not applied, however, the experiments were supervised by two experimenters, independently. The distribution of mice to the experimental groups was random and no animals were excluded during the experiment. The sample size was determined based on the assumed increase in tumor diameter. The experimental group size was calculated by power analysis or resource equation approach as described in [7]. The number of Caspase-3/7 positive cells was calculated using Gene 5 3.04 software (BioTek, Agilent).

Cytokine array
The cytokine arrays were co-incubated with the supernatants collected from the 24-h co-cultures of

Statistical analysis
Statistical analysis was performed with GraphPad Prism 8 (GraphPad Prism, RRID:SCR_002798). Results are shown as the mean value ± standard deviation (SD) unless otherwise stated. Normality was checked using the Shapiro-Wilk test. The comparison between two groups for paired samples was analyzed using Student t-test or Wilcoxon test and for unpaired samples using unpaired t-test or Mann-Whitney test depending on data distribution. For the in vivo experiment, a two-way ANOVA test was applied. The P values of less than 0.05 were considered significant.

Suppl. Figure 4. The efficacy of PD-L1-CAR NK-92 and primary NK cells against breast cancer cells and generation of the PD-L1-knockout derivative of MDA-MB-231 cell line and the PD-L1-overexpressing derivative of MCF-7 cell line and their susceptibility to PD-L1-CAR NK-92 cells cytotoxic activity.
A. Functional

Suppl. Figure 6. The efficacy of HER-2-CAR T cells against MCF-7 breast cancer cells.
A. The

Suppl. Figure 7. The killing potential of CAR-PD-L1-bearing NK-92 cells towards the bone marrowderived mesenchymal stem cells and generation of the PD-L1 overexpressing HEK-293T cell line.
A