Generation of non-human primate CAR Tregs using artificial antigen-presenting cells, simian tropic lentiviral vectors, and antigen-specific restimulation

Summary It is technically challenging to generate large doses of regulatory T cells (Tregs) engineered to express a chimeric antigen receptor (CAR) in non-human primates (NHP). Here, we have optimized the manufacturing of CAR Tregs by stringent sorting of Tregs, stimulation by artificial antigen-presenting cells, transduction by simian tropic lentiviral vectors, and antigen-specific expansion. The result of this method is highly suppressive CAR Tregs for use in a pre-clinical, large animal model of transplant tolerance. For complete details on the use and execution of this protocol, please refer to Ellis et al. (2022).


SUMMARY
It is technically challenging to generate large doses of regulatory T cells (Tregs) engineered to express a chimeric antigen receptor (CAR) in non-human primates (NHP). Here, we have optimized the manufacturing of CAR Tregs by stringent sorting of Tregs, stimulation by artificial antigen-presenting cells, transduction by simian tropic lentiviral vectors, and antigen-specific expansion. The result of this method is highly suppressive CAR Tregs for use in a pre-clinical, large animal model of transplant tolerance. For complete details on the use and execution of this protocol, please refer to Ellis et al. (2022).

BEFORE YOU BEGIN
The following protocol describes the expansion and transduction of CAR Tregs recognizing the NHP/human alloantigen Bw6 with CD28 and CD3z intracellular signaling domains. This protocol can be easily adapted to generate CAR Tregs with other specificities or to generate effector CAR T cells. Here, we describe the protocol for Cynomolgus macaque, but we have also generated CAR T cells from Rhesus macaque with equal success. Before starting, generate aliquots of simian tropic cHIV lentiviral vectors, irradiated K562 artificial antigen presenting cells (aAPCs), and NHP a-CD3/a-CD28 beads to have on hand.

Institutional permissions
Obtain institutional permission to perform animal studies and collect peripheral blood from NHPs under an approved Institutional Animal Care and Use Committee (IACUC) or Institutional Review Board protocol. Our protocol was approved by the University of Pennsylvania Institutional Animal Care and Use Committee (IACUC).

Manufacturing of irradiated K562 aAPCs for T cell stimulation
Timing: 2 weeks The following method will generate large quantities of irradiated K562 aAPCs for stimulating and expanding NHP T cells. The aAPCs used here were generated by transducing K562 cells with both a-CD3 CAR and human CD86 lentiviral vectors (K562.F12Q.86) or with HLA-B7 (Bw6+) and human CD86 vectors (K562.Bw6.86), performing single cell sorting of double positive clones, and characterizing the stability of transgene expression over time before selecting an optimal clone (Ellis et al., 2022). In our hands, K562-based aAPC stimulation results in larger T cell numbers versus bead-based stimulation (Ellis et al., 2022;Hippen et al., 2008Hippen et al., , 2011Maus et al., 2002;Suhoski et al., 2007;Thomas et al., 2002;Ye et al., 2011). We have used these cells successfully to stimulate Cynomolgus macaque (Ellis et al., 2022) and Rhesus macaque (Rust et al., 2020) T cells.
a. Thaw one vial of cryopreserved K562 cells rapidly in a 37 C water bath. b. Add cells to a 15 mL conical tube. c. Add room temperature or warmer R10 medium dropwise, shaking between every few drops. d. Spin down cells at 485 g for 5 min to wash.
Note: All centrifugation steps should be performed at room temperature.
e. Resuspend cells in R10 medium and count. f. Adjust cells to 200k cells/mL and incubate overnight in an appropriately sized cell culture flask in a 37 C incubator with 5% CO 2 .
Note: As K562 cells are grown in suspension, the size of the flask is dependent on number of cells, and is selected to ensure efficient gas exchange by plating at 0.2-0.5 mL/cm 2 ratio of volume to flask surface area.
2. The following day, passage cells for further propagation in a cell culture flask.
Note: Cells should be diluted with R10 medium to 50k cells/mL every other day or 25k cells/ mL every 3 days to maintain concentration under 400k cells/mL.
3. After 1 week of passaging, add 1.25 3 10 6 K562s to each well of a G-Rex 6 well plate. 4. Fill each well up to 100 mL with fresh R10 medium and return plate to incubator. 5. Grow cells undisturbed for 1 week. 6. After 1 week, remove medium in each well so that $5 mL remains. 7. Combine cells from all wells and bring up to 50 mL with R10. 8. Put cells in T150 flask and deliver 100 gy. of irradiation. 9. Wash cells with R10 medium and resuspend cells for cryopreservation at a concentration of 2.5-10 3 10 6 cells/mL in 90% FBS + 10% DMSO. 10. Pipette cells into cryovials, then put cryovials in room temperature freezing container for freezing at À80 C overnight. 11. The following day, move cryovials to liquid nitrogen for long term storage.

Generation of cHIV lentiviral vectors
Timing: 4 days NHP T cells express restriction factors that can reduce transduction efficiency by HIV based lentiviral vectors (Ellis et al., 2022;Hatziioannou et al., 2006;Richardson et al., 2014;Stremlau et al., 2004). Therefore, our simian tropic lentiviral packaging mix contains an HIV-SIV chimeric gag-pol (cHIV) (Hatziioannou et al., 2006;Uchida et al., 2009) and are pseudotyped with Cocal virus envelope glycoprotein (Trobridge et al., 2010). For best results, all DNA should be endotoxin free and CAR transgene should be placed under the EF1a promoter in the transfer plasmid (Leibman et al., 2017;Milone et al., 2009). See Figure 1 and Figure 2 for protocol outline and summary.

OPEN ACCESS
12. Aliquot lentiviral plasmid packaging mix. a. Combine 3 mg Cocal-Env plasmid, 18 mg HIV REV expression plasmid, and 18 mg HIV-SIV chimeric gag-pol plasmid. b. Fill up to 40 mL with nuclease free water.
Note: Aliquots can be stored at À20 C to À80 C for >2 years.
13. Generate transfer plasmid DNA of choice (Bw6-28z CAR is used in this protocol as described in (Ellis et al., 2022)) with endotoxin free maxiprep kit. 14. 24 h before lipofection, seed 8-10 3 10 6 HEK 293T cells in a T150 tissue culture flask in 30 mL R10 medium so that they are 70%-90% confluent the following day at the time of lipofection.
Note: HEK 293T cells detach from the plate very easily. Add/remove media to the top side of the flask.
a. Remove medium from HEK 293T cells. b. Wash cells twice gently with 10 mL PBS. c. Add 3 mL trypsin to the flask and incubate for 30 s at room temperature. d. Add 7 mL R10 to the flask to wash cells off bottom of the flask. e. Add cells to a 50 mL conical tube. f. Wash bottom of the flask with 10 mL R10 twice more and add to 50 mL conical tube. g. Centrifuge cells for 5 min at 485 g at room temperature and resuspend in 10 mL R10. h. Count cells.  Loading of T cell stimulation particles with a-CD3 and a-CD28 for Treg suppression assay Timing: 2 h 19. Pipette 100 mL of CD3-Biotin and 100 mL of CD28-Biotin into a sterile Eppendorf tube and mix well. 20. Vortex anti-Biotin MACSiBead particles and add 100 3 10 6 beads (500 mL) to the antibody-containing Eppendorf tube. 21. Add 300 mL flow buffer to bring total volume to 1 mL. The isolation of PBMCs from Cynomolgus macaque blood is not as efficient as that of human or even Rhesus macaque. We have found that Percoll gives the best results. The Cynomolgus PBMC layer will contain many more red blood cells than human, so ACK lysis is essential for efficient cell sorting.
1. Dispense anti-coagulated blood into a 50 mL conical tube and spin at 650 g for 10 min at room temperature with the deceleration and acceleration on lowest speed to avoid disruption of plasma layer atop the RBC/PBMC layer. 2. Pipette off the clear plasma (upper layer) leaving behind RBC/PBMC enriched layer. 3. Dilute remaining RBC/PBMC layer with 1 volume of PBS. 4. Place 15 mL room temperature 60% Percoll solution into a 50 mL conical, making sure to avoid getting Percoll on the sides of the tube. 5. Slowly, overlay up to 35 mL blood on top of the Percoll solution. 6. Centrifuge blood for 30 min at 650 g with the deceleration and acceleration on lowest speed. 7. Pipette off the top layer until $2.5 cm above interphase. 8. Use a pipette to collect the hazy middle layer containing PBMCs and put into a new tube. 9. Fill up new 50 mL conical tube with PBS. 10. Wash cells by spinning at 650 g for 10 min with the deceleration and acceleration returned to their highest settings. 11. Remove supernatant and lyse red blood cells.
a. Resuspend cell pellet in 1 mL of ACK lysis buffer. b. Fill up tube to 50 mL with ACK lysis buffer. c. Rock tube gently for 8 min at room temperature. d. Spin cells for 10 min at 650 g.
Note: Do not exceed 8 min of lysing time for optimal lymphocyte recovery. If cellular debris is visible following lysis, strain sample through a 70 mm filter into a new 50 mL conical tube before centrifugation.
12. Resuspend cells in 300 mL of flow cytometry buffer and move to a sterile FACS tube.

Sorting of Tregs from PBMCs
Timing: 5 h To minimize cell death, we perform a low-pressure sort using a 100 mm nozzle into 5 mL polypropylene tubes containing 1 mL of R10 medium as a cushion. Platelets will outnumber lymphocytes in the sample to be sorted. Use your flow sorter's FSC threshold to eliminate platelets from the screen. Though many platelets will remain in the Treg population following sorting, we have not seen this impact any downstream applications. Figure 3 depicts the gating strategy we use to isolate Tregs in Cynomolgus macaque.
Note: The concentration of antibodies used for flow cytometry throughout this protocol will need to be titered by each individual lab to ensure optimal performance on their flow sorter. We stain up to 100 million cells in 600 mL of flow buffer.
14. Add antibody staining master mix to cells.

Expansion and transduction of CAR Tregs
Timing: 3 weeks The goal of this step is to take stringently sorted Tregs and expand them into clinical sized doses of CAR Tregs. As few as 25k Tregs is sufficient to grow clinical sized doses of CAR Tregs. On each day of (re)stimulation, it is critical that aAPCs make physical contact with the T cells, so err on the side of choosing a plate or flask with a smaller surface area rather than a larger one. Since cells transduced with functional levels of CAR are exclusively re-stimulated by K562.Bw6.86 cells, absolute growth can be minimal between weeks 1-2 despite an increase in the percentage of CAR+ Tregs. This contrasts with antigen non-specific stimulation during week 1, where each cell is stimulated. Microscopy of stimulated Tregs can be seen in Figure 4. Figure 5 demonstrates the enrichment of CAR+ cells by antigen specific restimulation of CAR Tregs, while Figure 6 depicts the expected growth of Tregs along with expected expression of FoxP3, CAR, CTLA-4, and Helios.  32. On day 9, double media with R10 containing 600 IU/mL IL-2 and move cells to appropriate plate or flask. 33. On day 11, count cells and adjust to 500k cells/mL with R10 medium. 34. Add IL-2 to 300 IU/mL, assuming consumption. 35. On day 14 / 16 / 18, repeat the steps from days 7/ 9 / 11, but with a 1:2 ratio of aAPCs to Tregs. 36. When cells reach resting state, usually around 375-400 fL in volume as seen on a Coulter counter during cell counting, freeze up to 50 3 10 6 Tregs in 1 mL of 90% FBS + 10% DMSO solution. 37. To assess Treg product by flow cytometry, first wash 500k cells in 3 mL flow buffer and then resuspend cells in 100 mL flow buffer containing titrated amounts of the following: a. CD4-BV605 (OKT4

LAP assay
Timing: 2 h The following protocol allows the researcher to determine the extent of antigen specific activation of the CAR Treg product by measuring cell surface expression of the TGFb propeptide LAP. While the assay using aAPC targets bearing the antigen of interest is described, this assay has also been successful with plate-bound antigen. Example of expected results can be found in Figure 7. A graphical representation of the LAP assay is depicted in Figure 8.  Batches of PBMCs need to be loaded and frozen for use in Treg suppression assays. As the amount of CTV used can impact proliferation and viability of PBMCs, we have described our method below. The CTV labeled before freezing is well maintained, and doing so in larger batches increases consistency between experiments and saves time.

Obtain PBMCs from NHP blood draw as described in the method ''isolation of PBMCs from
Cynomolgus macaque whole blood''. 68. Remove 1 vial of CTV from freezer. 69. Allow vial to warm to room temperature before opening. 70. Pulse spin tube to pool reagent at bottom of tube. 71. Resuspend in 20 mL anhydrous DMSO to generate 5 mM stock solution. 72. Wash cells once with PBS and resuspend up to 25 3 10 6 PBMCs in 1 mL PBS. 73. Dilute CTV stock 1:100 to 50 mM. 74. Add 110 mL of 50 mM working stock for every 1 mL of cells in PBS for a final concentration of 5 mM. 75. Mix well and incubate for 20 min at room temperature.
Note: CellTrace proliferation dyes will undergo hydrolysis in aqueous solutions to become non-cell permeable. Dilute stock with PBS just before labeling cells.

Figure 8. LAP assay
After 24 h of co-culture between CAR Tregs and K562s, Tregs are stained for expression of LAP to determine antigen specificity of CAR Tregs.

OPEN ACCESS
76. To remove free dye from solution, add 5 volumes of R10 and incubate at room temperature for 5 min. 77. Spin cells down for 5 min at 485 g and freeze in aliquots of 2.5 3 10 6 cells/vial in 90% FBS / 10% DMSO.

Timing: 4 days
This assay quantifies the antigen non-specific suppressive activity of CAR Tregs. We suggest dedicating a large quantity of PBMCs that have been labeled with cell proliferation dye and frozen to standardize responders across batches of CAR Tregs. We find that irradiated aAPCs are too strong of a stimulus for Treg suppression assays, so the T cell Activation/Expansion Kit (Miltenyi Biotec) is used instead. See Table 1 for an example of the plate setup. Example data are depicted in Figure 9, and a graphical representation of the suppression assay can be found in Figure 10.
78. Wash CAR Tregs and Teffs with R10 medium three times to eliminate residual IL-2. 79. Resuspend cells in R10 at 750k cells/mL. 80. Thaw frozen, CTV labeled PBMCs as above. 81. Wash cells twice with R10 medium and resuspend at 750k cells/mL in R10. 82. Wash a-CD3/a-CD28 beads generated in the section ''loading of T cell stimulation particles with a-CD3 and a-CD28 for Treg suppression assay''. a. Add 20 mL of a-CD3/a-CD28 beads to 100 mL of R10 in an Eppendorf tube. b. Spin cells in a tabletop centrifuge for 5 min at 300 3 g. c. Remove media, resuspend beads in 500 mL R10 medium, and count. Adjust beads to 375k beads/mL. 83. Add 134 mL Tregs (100k cells) to well B2 in a 96 well round bottom polystyrene plate. 84. Fill up wells B3-B7 with 67 mL fresh R10. 85. Perform 1:1 serial dilution of Tregs by moving 67 mL of cells from B2 through B7. 86. Remove the last 67 mL from well B7 so that all wells finish this step containing 67 mL of CAR Tregs. 87. Add 67 mL of CTV labeled PBMCs (50k cells) to each well B2-B7. 88. Add 67 mL of beads (25k beads) to each well B2-B7. 89. Repeat steps 83-88 with Teffs instead of Tregs in wells C2-C7. 90. Create a row of control wells: a. Add 67 mL of PBMCs and 134 mL R10 to well D2 as an unstimulated control. b. Add 67 mL of PBMCs, 67 mL beads, and 67 mL R10 to well D3 as an unsuppressed control.

EXPECTED OUTCOMES
From an initial population of 25k-100k sorted Tregs, we expected between 500-1000 3 10 6 cells on day 18-21 ready for freezing ( Figure 6A). By flow, our final cell product is expected to be >80% Figure 10. Treg suppression assay Measurement of CTV dilution after co-culture between Tregs, CTV labeled PBMCs, and a-CD3/a-CD28 beads to assess Treg suppressor ability.

OPEN ACCESS
FoxP3+, >80% Helios+, and >80% CAR+. Levels of CTLA-4 were variable in this protocol ( Figures 6B  and 6C). Each batch of Tregs was subject to LAP and suppression assays, where we expected >30% LAP+ in response to target antigen (Figure 7), and 50% suppression occurring between 8:1 and 16:1 PBMC: Treg ratio (Figure 9). Depending on the strength of bead stimulus in the suppression assay and the composition of PBMCs used, the suppressor ability of the Tregs may be less or more apparent. Comparison to Teff cells as the suppressors is therefore vital to interpretation of the experiment.

LIMITATIONS
The ability to generate large batches of Tregs is dependent on the ability to get high levels of transduction during the first week of growth. Low transduction abilities stifles growth during week 2, affecting overall cell numbers. In rare occasions, Tregs were found to have low suppressor ability in vitro, which correlated with lower percentage of FoxP3+ cells, and were thus discarded. We believe that these batches were due to non-stringency during the sorting step. Suppressive batches were later able to be generated from the same animals by using more stringent CD25+ gating. For animal health, IACUC limits both the volume and frequency of blood draws.

Potential solution
In the case where transduction percentage is low, increasing the amount of virus may help increase the transduction percentage. Alternatively, the batch of virus may not have been optimally generated. In this case, first ensure that HEK 293T cells are healthy and grow to 70%-90% confluency every 2 days. If not, replace unhealthy HEK 293T cells with another frozen batch. Next, check the fidelity of the transfer plasmid, which should have been isolated with an endotoxin free maxiprep kit. When run on an agarose gel, DNA should be supercoiled and free of genomic DNA contamination. Liberating the CAR insert with restriction enzyme digestion should generate DNA fragments of the correct size. Re-sequence the DNA transgene by Sanger sequencing, as viral transfer plasmids are prone to recombination due to homology in long terminal repeats.

Problem 2
Treg expansion is poor (expansion and transduction of CAR Tregs).

Potential solution
Check that the aAPCs still express both CD86 and aCD3 by flow cytometry by staining with His-Tagged Cynomolgus CD3ε protein, followed by CD86-BV421 and a-His Tag-AF647. As stimulation requires T cell-aAPC contact, plating in cell culture plates with too large of a surface area may prevent activation. Err on using a plate with a smaller surface area. Ensure that IL-2 was thawed no more than 2 weeks prior. Poor initial stimulation could also be due to cell death during the sorting process. Ensure that a low-pressure sort is used and that the cells are sorted into polypropylene tubes cushioned with cell culture media.

Problem 3
Tregs are not suppressive and/or do not express FoxP3 (step 21).

Potential solution
Stringent sorting of Tregs is essential to maintaining FoxP3 and suppressor ability in expanded cells. If cells are not phenotypically or functionally Tregs at the end of culture, re-assess sorting gates by moving CD25 and or CD45RA gate towards higher expression. Sorted Tregs should be in the top 1%-2% of CD25+ cells among CD4+ CD8-cells.
Problem 4 PBMC yield is low following isolation from peripheral blood (step 8).

Potential solution
As the isolation of PBMCs from Cynomolgus macaque blood is more difficult than from human blood, yield may be low. Typically, we recover between 50 3 10 6 -100 3 10 6 PBMCs from 15-20 mL of blood. If recovery is low, try pipetting more of the hazy middle layer in step 8, including some of the red blood cell layer, to ensure that all PBMCs are removed from the tube. These excess red blood cells will be efficiently lysed in step 11. Additionally, perform a complete blood count on your NHP to check for lymphopenia.

Potential solution
If the Tregs have high expression of FoxP3, CD25, and Helios and stimulation-induced LAP activation but are not suppressive in the suppression assay, you may need to optimize the assay. Excess activation of PBMCs can render the Tregs unable to provide any noticeable suppression. Ensure that all IL-2 has been washed out of the assay by washing Tregs three times before use, and that the amount of a-CD3/a-CD28 beads in each well is accurate by carefully counting the beads. If there is no/low proliferation in any wells including the unsuppressed condition, try lowering the amount of CTV loaded into the PBMCs, increasing number of a-CD3/a-CD28 beads, or increasing the duration of the experiment. Additionally, check the viability of the CTV labeled PBMCs after thawing, as warming too slowly or freezing too quickly can have drastic effects.

RESOURCE AVAILABILITY
Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, James Riley (rileyj@upenn.edu).
Materials availability aAPCs expressing pan-primate aCD3 and human CD86 are available from the lead contact with a completed Materials Transfer Agreement.

Data and code availability
This protocol did not develop any code.

ACKNOWLEDGMENTS
The authors would like to thank One Lambda for their aBw6 scFv sequence. We would also like to thank University Laboratory Animal Resources animal care technicians, veterinarians, and veterinary nurses, as well as the Penn Cytomics and Cell Sorting Resource Laboratory. These studies have been generously funded by the The Leona M. and Harry B. Helmsley Charitable Trust and National Institute of Diabetes and Digestive and Kidney Diseases (DK122644 and DK132725).