A protocol to evaluate quantitative tumor cell killing by Jurkat cells expressing chimeric antigen receptor (CAR) targeting single tumor antigen. This protocol can be used as a screening platform for rapid optimization of CAR hinge constructs prior to confirmation in peripheral blood-derived T cells.
Chimeric antigen receptor (CAR) T cells are at the forefront of oncology. A CAR is constructed of a targeting domain (usually a single chain variable fragment, scFv), with an accompanying intra-chain linker, followed by a hinge, transmembrane, and costimulatory domain. Modification of the intra-chain linker and hinge domain can have a significant effect on CAR-mediated killing. Considering the many different options for each part of a CAR construct, there are large numbers of permutations. Making CAR-T cells is a time-consuming and expensive process, and making and testing many constructs is a heavy time and material investment. This protocol describes a platform to rapidly evaluate hinge-optimized CAR constructs in Jurkat cells (CAR-J). Jurkat cells are an immortalized T cell line with high lentivirus uptake, allowing for efficient CAR transduction. Here, we present a platform to rapidly evaluate CAR-J using a fluorescent imager, followed by confirmation of cytolysis in PBMC-derived T cells.
CAR-T cell therapy has shown great promise in hematological malignancies, evident from the 6 FDA-approved CAR-T products since 2017, as reported by the National Cancer Institute1. There are numerous CAR-T cells in clinical trials for targeting solid tumors. Engineering novel CAR targets and optimizing the CAR construct is vital to the efficacy of a CAR-T cell. Choosing the ideal CAR construct for each application is essential for accurate targeting of tumor associated antigens (TAA) while avoiding low levels of TAA expression in normal tissues2.
A CAR construct is primarily made of five compartments: (1) extracellular single-chain variable fragment (scFv) domain targeting tumor antigen; (2) hinge domain; (3) transmembrane domain; (4) intracellular cytoplasmic T cell costimulatory domain; and (5) signaling domain. Modifying each of these domains affects the precision of the CAR-T cell engaging with its target cell3. Hence, evaluating the cytotoxicity and cross-reactivity of these CAR constructs in vitro is critical to choose the right construct for progressing toward in vivo experiments. Current methods of evaluating cytolysis by T cells include 51Cr release assay, lactate dehydrogenase release assay, bioluminescent imaging assay, real-time impedance-based cell analysis, and cell-based flow cytometry assay4,5. The fluorescent imaging-based platform described here identifies the number of live vs. dead cells, which is a direct quantification of T cell cytolysis as opposed to an indirect method of evaluating the cytolysis by T cells.
Here is an easy, cost-efficient, rapid, and high throughput technique with minimal intervention to evaluate the cytotoxicity of Jurkat cells expressing epidermal growth factor receptor (EGFR) CAR against MDA-MB-231 triple-negative breast cancer (TNBC) cells and EGFR CRISPR knock out MDA-MB-231 cells. Jurkat cells are immortalized human T Lymphocyte Cells6 that have been widely used for studying T cell activation and signaling mechanisms7. Furthermore, Jurkat cells have been used for in vitro CAR testing in multiple studies8,9,10,11. Jurkat cells are easily transduced by lentivirus and have sustained proliferation, and this system was leveraged to optimize the hinge domain of various EGFR CAR constructs.
This assay can be used for screening multiple CAR constructs targeting various tumor antigens and used against multiple adherent tumor cell lines and in various effector to tumor (E:T) ratios. Additionally, multiple time points can be evaluated, and number of replicates can be modified to identify best killing among the various CAR constructs. The best constructs need to be confirmed using peripheral blood mononuclear cells (PBMCs) derived CD3 T cells. The overall goal behind developing this method is to rapidly optimize CAR hinge geometry in a high throughput manner overcoming barriers such as low transduction efficiency, followed by confirmation in PBMC derived T cells.
NOTE: All cell culture work is done in a biosafety cabinet with a lab coat, gloves, and following standard aseptic techniques.
1. Generating CAR expressing Jurkats (CAR-J)
2. Plating CFSE labelled tumor cells
NOTE: MDA-MB-231 (from ATCC, HTB-26) cells were a gift from a collaborator, and EGFR KO MDA-MB-231 were created as previously described12.
3. Co-culturing CAR expressing Jurkats with CFSE labelled tumor cells
4. Preparation of plate for imaging
5. Analyzing fluorescent images
A range of E:T ratio between 1:8 and 8:1 for CAR-J1 was evaluated at 72 h which targeted EGFR on TNBC MDA-MB-231 cells. Jurkat cells were transduced with CAR lentivirus with polybrene to generate CAR-J cells as described in step 2. Cytotoxicity of CAR-J1 significantly increased with higher E:T ratio with no difference in killing at 1:8 ratio (Figure 1). More than 50% killing was observed at 4:1 E:T over 72 h. This E:T was used for subsequent experiments with duration reduced to 48 h for rapid cytotoxicity evaluation of multiple CAR constructs. EGFR targeting CAR constructs with the hinge domain modified were designed (Table 1). The 4 different hinges used are IgG long, IgG medium, IgG short and CD8a as described here13. These 3 constructs were expressed on Jurkat cells and CAR positivity determined by evaluating the percent of Flag tagged cells by flow cytometry (Figure 2A-D) as described here12. Un-transduced Jurkat cells were used as control group to determine CAR expression and the gating strategy is shown in (Figure 2E). The cytotoxicity of these 4 CAR constructs expressed on Jurkat cells was evaluated against EGFR expressing MDA-MB-231 cells and CRISPR EGFR KO MDA-MB-231 cells. Antigen specific killing was observed by all the constructs (Figure 3A) whereas there was no significant killing observed against EGFR KO cells (Figure 3B) suggesting that killing was specifically mediated through the scFv only. There was no killing by the un-transduced Jurkat cells. Representative images of tumor killing are also shown where the overlap of CFSE labeled tumor cells with PI-stained dead nucleus make them appear yellow (Figure 4). Data is representative of 3 independent experiments with 6 technical replicates per group.
This was further confirmed by expressing these CAR constructs on PBMC derived CD3 T cells. There is low expression of CAR on CD3 T cells which are then enriched by sterile flow sorting (Figure 5). However, there was no expression of CD8 hinge containing EGFR CAR. Hence the other 3 CAR constructs containing IgG long, IgG medium and IgG short were evaluated for their cytotoxic potential against MDA-MB-231 cells. There is a similar trend of killing with IgG short having the least cytotoxic potency compared to the other 2 constructs (Figure 6). To identify the best construct among IgG long and IgG medium CAR-T cells, their activation with and without exposure to TNBC tumor cells was evaluated by intracellular staining of cytokines as described previously12. MDA-MB-436 has low levels of EGFR and MDA-MB-468 has the highest EGFR protein expression based on a western blot with a panel of 13 TNBC cell lines12. IgG long CAR-T cells had the least basal activation levels (TNFa, 4-1BB) and low release of cytotoxic granules (perforin and granzyme B)14 without any exposure to tumor cells (Figure 7). Upon exposure to high EGFR expressing MDA-MB-468 cells, IgG long CAR-T cells had the highest activation based on TNFa and 4-1BB.
Figure 1: EGFR CAR-J1 cytotoxic evaluation along a range of E:T. CFSE labelled MDA-MB-231 tumor cells were plated with un-transduced Jurkat cells or EGFR targeting CAR-J1 at E:T ratio of 1:8, 1:4, 1:2, 1:1, 2:1, 4:1 and 8:1 showing increasing cytotoxicity with higher effector cells. Data are represented as mean ± SD and statistical significance was evaluated using Student's t-test. 1:8 E:T was non-significant, 1:4 E:T had ***p<0.001, and rest ****p<0.0001. Please click here to view a larger version of this figure.
Figure 2: EGFR CAR Jurkat production for in vitro cytotoxic evaluation. (A-D) Nearly 90%-100% CAR expression of 4 different constructs CAR IgG long, CAR IgG medium, CAR IgG short, CAR CD8a. (E) Gating strategy for CAR positivity determination: Debris excluded in SSC-A vs FSC-A plot; Single Cells selected in FSC-H vs FSC-A plot; and Live cells selected by viability dye DAPI negative gating. Please click here to view a larger version of this figure.
Figure 3: Cytotoxicity evaluation in co-culture of CAR-J and CFSE labelled tumor cells. (A) CFSE labelled MDA MB 231 cells were plated with CAR-J at 4:1 E:T ratio for 48 h showing varying efficacy in tumor cell killing. (B) CFSE labelled CRISPR EGFR KO MDA MB 231 cells did not experience any killing by any of the CAR-J cells. Data are represented as mean ± SD and statistical significance was evaluated using one-way ANOVA. **p<0.01; ****p<0.0001; ns: not significant. Please click here to view a larger version of this figure.
Figure 4: Visual representation of images after 48 h of co-culture. (A) Tumor only group with CFSE (green) labelled tumor cells. (B) Addition of un-transduced Jurkat cells with red cells being the dead Jurkat cells and (C) overlay of green and red indicating dead tumor cells (yellow) as shown by arrows. The number of green cells that are not yellow are quantified. Please click here to view a larger version of this figure.
Figure 5: EGFR CD3 CAR-T cell enrichment. (A-B) CAR expression evaluated by determining percent of Flag positive and IgG positive cells before enrichment and (C) post enrichment by sterile flow sort. Please click here to view a larger version of this figure.
Figure 6: Cytotoxicity evaluation in co-culture of CD3 CAR-T and CFSE labelled tumor cells. CFSE labelled MDA MB 231 cells were plated with CAR-J at 4:1 E:T ratio for 48 h showing varying efficacy in tumor cell killing. Data are represented as mean ± SD and statistical significance was evaluated using one-way ANOVA. **p<0.01; ****p<0.0001; ns: not significant. Please click here to view a larger version of this figure.
Figure 7: EGFR CD3 CAR-T cell Activation. CAR-T cells were exposed to tumor cells and intracellular cytokine levels for(A) TNFa, (B) 4-1BB, (C) perforin, and (D) granzyme B were evaluated. Data are represented as mean ± SD and statistical significance was evaluated using one-way ANOVA *p<0.05; **p<0.01; **p<0.001; ***p<0.0001; ns: not significant. Please click here to view a larger version of this figure.
No. of PA EGFR806 scFv | Vh-Vl Linker | Hinge/Spacer | TM | Cytoplasmic |
CAR IgG Long | Whitlow (18 aa) | IgG4 EQ CH2 CH3 (229 aa) | CD4 | 4-1BB / CD3z |
CAR IgG Medium | Whitlow | IgG4 CH3 (129 aa) | CD28 | 4-1BB / CD3z |
CAR IgG Short | Whitlow | IgG short (12 aa) | CD4 | 4-1BB / CD3z |
CAR CD8a | Whitlow | CD8 (45 aa) | CD8a | 4-1BB / CD3z |
Table 1: CAR construct designs. Abbreviations: aa = amino acids; TM = Transmembrane.
1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | |
A | Triton-X | MDA MB 231 Only | Triton-X | MDA MB 231 EGFR KO Only | ||||||||
B | MB231 + Mock 4:1 | MB231 EGFR KO + Mock 4:1 | ||||||||||
C | MB231 + EGFR IgG Long 4:1 | MB231 EGFR KO + EGFR IgG Long 4:1 | ||||||||||
D | MB231 + EGFR IgG Medium 4:1 | MB231 EGFR KO + EGFR IgG Medium 4:1 | ||||||||||
E | MB231 + EGFR IgG Short 4:1 | MB231 EGFR KO + EGFR IgG Short 4:1 | ||||||||||
F | MB231 + EGFR CD8a 4:1 | MB231 EGFR KO + EGFR CD8a 4:1 | ||||||||||
G | ||||||||||||
H |
Table 2: Sample plating strategy.
Here we have proposed a rapid method to efficiently evaluate the target-specific cytolytic activity induced by CAR expression in Jurkat cells. All CAR constructs have the same scFv but different hinge and transmembrane domains which have been shown to affect CAR-T cells potency13. Further evaluation of non-specific killing by these CAR-J was done by culturing them with antigen knock out (KO) cells. This demonstrates that the killing is tumor antigen specific and not due to basal activation by the CAR-J. Cytotoxic potential of a range of E:T ratios can be easily determined identifying the least number of effector cells required for eliminating tumor cells at specific time points. Multiple cell lines from same or different cancer type and normal cells can be also used to evaluate the specificity of these constructs. The most potent candidates identified by CAR-J platform can then be further characterized by expressing CAR on PBMC derived CD3 T cells and performing similar killing assays and evaluating exhaustion and activation upon exposure to antigen positive and negative tumor cells.
It must be noted that Jurkat cells are CD4+ cells with an altered T cell receptor (TCR) signaling pathway15. Hence, intracellular signaling domain optimization of CAR constructs should ideally be done with CD3 T cells16 and may not work best with Jurkat cells. Interestingly, CD4 T cells showcase cytotoxicity as demonstrated in a previous publication where EGFR targeting CD4 CAR-T cells with IgG long hinge are able to completely eradicate TNBC tumors in an intracranial tumor model12. Additionally, CD4 CAR T cells show long term potency and better killing as compared to mixture of CD4 and CD8 T cells or CD8 T cells alone in GBM tumor model making them clinically important T cells subset for effective CAR-T therapy17.
Jurkat cells produce IL2 and upregulate CD69 upon activation, though they do not secrete all the cytokines of primary T cell activation8,18. However, evaluating IL2 and CD69 levels informs regarding the Jurkat cells getting activated and not about the direct cytolysis of target cells which are being quantified in absolute tumor cell numbers by this approach. However, evaluation of cytotoxicity and activation is important to fully understand the effects of novel CAR molecules.
The duration of co-culture in this experiment can be modified based on the constructs being used. CFSE labeling on tumor cells was found to be detectable until 4 days after plating. Hence, to attain a high signal to noise ratio, experiments were performed within 72 h. Since CFSE intensity decreases on division of tumor cells, the amount of CFSE used and the duration may need to change as per the tumor cells used. Since this is a high throughput experiment to analyze multiple constructs at a time, the seeding density of target cells in a 96 well plate also needs to be optimized. Target cells must be maintained so that they do not grow overconfluent or have competitive growth restriction by the end of the assay without the need to change media to minimize any intervention. Larger well plates may be used to accommodate more cells and increase the longevity of the experiment. However, the images need to be stitched to get a full picture of the well and each well has to be worked on individually, which may not make it quick and high throughput.
Another approach of long-term fluorescent labeling tumor cells is by lentiviral transduction with green fluorescent protein (GFP) or red fluorescent protein (RFP). Proper clonal selection must be done to select highly enriched (nearly 100%) brightest colonies for the best signal to noise ratio. The viability dye needs to be appropriately selected based on the fluorescent detectors in the imaging instrument.
Background fluorescence depends on the media being used and hence must be checked to avoid loss of signal while acquiring images. Regular media contain components that emit significant fluorescence when excited. It is recommend to use a low background media19. PBS may be used but it may affect the cells during image acquiring as it takes roughly 20 min to acquire images depending on the settings set for capturing images.
One of the limitations of this assay is not to be able to take images at multiple time points of the same 96 well plate after adding a viability dye. The effect of addition of PI over long time was not evaluated for this publication. However, it would be ideal to be able to detect live and dead/dying target cells over a period using the same plate. It may be difficult to segregate individual cells if cells become over confluent. The area of cell monolayer and fluorescence intensity may be used to compare percent of cells living/dead as described20.
Alternate methods used to detect cytolysis by CAR-T cells do not directly measure the absolute number of dead and live cells, rather evaluate potency by indirect methods which are described in detail elsewhere4,5. So, this method gives an absolute account of effector cells and may be used in co-culture of 3 cell types as well depending on the capacity of the imaging instrument. There are very few cells required (5000 per well) which may change along with the freedom to choose a time point which are huge advantages of using this assay. Target cells in suspension and in 3-dimensional spheroid culture can also be used in culture with effector cells and cytolysis determined as described elsewhere21. Furthermore, this platform can be used to screen large libraries of small molecules and compounds to delineate the most effective molecules and multiple dosing can be done at multiple time points.
Besides the speed and the flexibility of the system, the final benefit is that of cost. The fluorescent imaging machine, plates, and reagents are all common and affordable to purchase, especially when compared to devices that are more sophisticated and may even use precious metals in their single use plates.
The authors have nothing to disclose.
MDA-MB-231 were a kind gift from Dr. Shane Stecklein. The authors acknowledge funding from the University of Kansas Cancer Center to conduct this research.
15 mL Conical Tube (Sterile) | Midwest Scientific | #C15B | Any similar will work |
50 mL Conical Tube (Sterile) | Thermo Scientific | 339652 | Any similar will work |
Black/Clear 96 well plate | Falcon | 353219 | Celligo has a list of compatible plates |
Celigo 4 Channel Imaging Cytomenter | Nexcelcom Bioscience | 200-BFFL-5C | Any similar will work |
Celigo Software | Nexcelcom Bioscience | Version 5.3.0.0 | Any similar will work |
Cell Culture Incubator | Thermo Scientific | HeraCell 160i | Any similar will work |
Cell Culture Treated Flasks (T75, various sizes, Sterile) | TPP | 90076 | Any similar will work |
CFSE | Tonbo | 13-0850-U500 | Any similar will work |
Cytek Muse Cell Analyzer | Cytek | 0500-3115 | Any similar will work |
DMEM | Gibco | 11995-040 | Any similar will work |
FBS | Gemini bio-products | 900-108 | Any similar will work |
Flow Cytometer | Cytek, BD, etc | Aurora, LSR II, etc | Any similar will work |
FlowJo Sortware | Becton Dickinson & Company | Version 10.7.1 | Any similar will work |
Fluorobrite DMEM | Gibco | A18967-01 | Any similar will work |
GraphPad Software | GraphPad | Version 9.3.1 (471) | Any similar will work |
Multichanel Pipette | Thermo Scientific | Finnpipette F2 | Any similar will work |
PBS | Gibco | 10010-031 | Any similar will work |
PenStrep | Gibco | 15070-063 | Any similar will work |
Pipette tips (Sterile, filtered, 1 mL, Various sizes) | Pr1ma | PR-1250RK-FL, etc | Any similar will work |
Pipettors | Thermo Scientific | Finnpipette F2 | Any similar will work |
Propidium Iodide | Invitrogen | P1304MP | Any similar will work |
RPMI | Corning | 10-041-cv | Any similar will work |
Serological Pipette Aid | Drummond Scientific | 4-000-105 | Any similar will work |
Serological Pipettes (Sterile, various sizes) | Pr1ma | PR-SERO-25, etc | Any similar will work |
Sodium Pyruvate | Corning | 25-000-CI | Any similar will work |
Sterile Reservoirs | Midwest Scientific | RESE-2000 | Any similar will work |
Table top centrifuge | Eppendorf | 5810R | Any similar will work |