Abstract

Our Ambrosia T is divided into two parts, killing elements and control elements. For killing elements, we engineered an anti-Dipeptidyl Peptidase 4 (DPP4) chimeric antigen receptor (CAR), containing anti-DPP4 single chain fragment variable (scFv), CD8 transmembrane domain, 4-1BB co-stimulator domain, and CD3-ζ domain. And for control elements, we incorporated a caffeine-operated synthetic module (COSMO) and three IL-6-based negative feedback loops to control the activation and inactivation of CAR-T cells. In the same way, our experiment consisted of two parts, verification of the lethality of CAR-T cells and verification of the effect of switches, which jointly prove the feasibility of our design. CAR structures and sequences we provided demonstrated superior killing efficiency in the experiments. Moreover, we expressed switches separately with anti-CD19 CAR in Jurkat cells and found that cells equipped with IL-6R-PD-1 showed reduced of cytotoxicity, which proves that our negative feedback element regulation is effective.

Engineered CAR-T cells target DPP4 to clear senescent cells

CARs are unique receptors that are designed to target a specific antigen to functionally reprogram T lymphocytes. When the CARs recognize to their targets, it will activate and promote the proliferation of T cells and secretion of granzymes and perforins. Then, the target cells will be killed. Since the killing properties of T cells target all target cells, targeted killing of senescent cells by CAR-T cells is possible. DPP4 was a potential senescent marker which selectively expressed on the surface of senescent, but not proliferating, human diploid fibroblasts. We engineered an anti-DPP4 CAR, containing anti-DPP4 scFv, CD8 transmembrane domain, 4-1BB co-stimulator domain, and CD3-ζ domain. When the engineered CAR-T cells recognize the DPP4 highly expressed senescent cells, the downstream 4-1BB and CD3-ζ will be activated to promote the proliferation of T cells and secretion of granzymes and perforins to achieve the goal of killing the senescent cells (FIG. 1).


Figure 1. The overview of anti-DPP4 CAR-T cells clearing senescent cells.


Figure 2. The results of anti-DPP4 CAR-T cells clearing senescent cells.


Figure 3. The results of repeated experiments of anti-DPP4 CAR-T cells clearing senescent cells.

Inducible Caffeine switches regulate the activation of CAR-T cells

The COSMO was inserted between 4-1BB co-stimulator domain and CD3-ζ domain. In the absence of caffeine in the environment, when engineered CAR-T cells bind to their respective targets, the CD3- domain cannot be activated due to the blockage of COSMO, and the CAR T cells cannot fully activate, preventing them from carrying out their normal function of killing senescent target cells. However, if the caffeine is presented in the environment, the two engineered CARs can dimerize under the action of COSMO, resulting in the activation of downstream CD3 signaling pathway. Thus, CAR-T cells could be fully activated and perform normal functions (FIG.2). This is a human-controlled CAR-T cell turn-on via COSMO.


Figure 4. The structure of caffeine inducible CAR and its mechanism to control the activation of CAR-T cells.


Figure 5. The results of Inducible Caffeine switches regulate the activation of CAR-T cells.

As the result, The CAR-COSMO-M killing system showed good lethality after the addition of caffeine, but the CAR-COSMO modulation system did not seem to completely inhibit the killing of Raji cell lines by CAR-M, which we believed is in error with the real result. After analysis, we thought that two biases in the experiment lead to erroneous experimental results. The first reason is that we used an enzymatic marker to measure the data but did not consider the effect of the color of the coffee itself on the data. In addition, we selected macrophages as chassis cells, which are not in full contact with the target cells, suspension cells of the Raji cell line. This may result in poor killing efficacy. Anyway, the result showed that the CAR-COSMO-M killing system exhibited good killing in the presence of caffeine, verifying the lethality of our product and the feasibility of the COSMO switch.

IL-6 based negative feedback loop to shut down the CAR-T cells

Firstly, we designed an anti-IL-6 synthetic Notch receptor (SynNotch). The anti-IL-6 scFv was connected to the Notch regulatory domain which followed by the recombinant transcriptional factor, Gal4-KRAB. When the concentration of IL-6 in the environment is higher than a certain threshold, a large amount of IL6 will bind to anti-IL-6 scFv, leading to the conformational change of SynNotch receptor, which is cleaved by intracellular enzymes. Then the downstream Gal4-KRAB transcription factor is released and transferred to the nucleus, where it binds to the UAS-pSV40 promoter in the nucleus. Consequently, the downstream CAR gene expression was inhibited. The second design was quite similar to the first one, except that the anti-IL-6 scFv was replaced by the low affinity IL-6 receptor binding domain. Moreover, we designed a recombinant receptor which combined the IL-6 receptor extracellular binding domain and PD-1 intracellular domain. When the extracellular IL-6 concentration is high enough, it will bind to the IL-6R extracellular domain and activate the intracellular inhibitory receptors signaling modules. Therefore, the downstream signal of CD3-ζ will be inhibited and inhibit the CAR-T cells activity.


Figure 6. The structure of IL-6 based negative feedback loop.


Figure 7. Results of IL-6 based negative feedback loop to shut down the CAR-T cells (blue, Raji; orange, Raji cocultured with untransfected Jurkat; gray, Raji cocultured with transfected Jurkat). The unit of IL6 concentration is pg/ml.

Here we transfected Jurkat cells with our constructed plasmids and confirmed expression of the synthetic receptor and CAR by flow cytometry. These CAR-T cells were co-cultured with the leukemia cell line Raji and examined cytotoxicity over time. We co-cultured CAR-T cells with designed NFL or NFL in media containing different concentrations of IL-6(0, 1, 10, 100, 1000 pg/ml). We found that for CAR-T cells transfected with IL-6R-PD1, the activity of CAR was successfully inhibited at high level of IL-6,which is consistent with what we expected. However, the results did not show that high concentration of IL-6 could inhibit the cytotoxicity of IL-6-scFv-Notch-Gal4krab or IL-6R-Notch-Gal4KRAB modified CAR-T cells.

The explanations we made for not achieving the desired result is that, first, for synthetic Notch receptors, the receptor cannot be activated when the ligand is soluble, rather than presented on the cell membrane. This suggests the possibility that synthetic Notch is not suited to respond to soluble IL-6. Second, transfection efficiency in our experiments was not high, and the use of a dual-promoter system may have affected the expression of the latter protein, CAR. In summary, our IL-6R-PD-1 achieved the desired effect, demonstrating that the negative feedback loop based on IL-6 can act as a switch to regulate the killing effect of CAR-T cells.

In summary, we transduced into Jurkat or RAW 264.7 cell lines with composite Parts rather than individual Parts, and both the killing and regulatory modules showed good killing properties in the corresponding experimental contexts.