Our approach in creating a pH-sensitive CAR was through the usage of an pH-induced auto-proteolytic linker. We chose the linker sequence (EAAAK)5, which was shown to cleave around the tumor relevant value of pH 6.5 by Wu et al. [1]. Before testing this functionality on a CAR, we decided to design and create a simple fusion protein linked by (EAAAK)5. The fusion protein is made of a secretion signaling sequence and two times two Kappa light constant chain domains linked together with the pH-linker (Figure 1). For purification, we added a C-terminal 6x His-Tag.
After designing and ordering the whole protein sequence as a gene in pcDNA3.4, we used the ExpiCHO expression system to obtain the protein over one week. Purification was performed using an Immobilized Metal Affinity Chromatography (IMAC) followed by size-exclusion chromatography. Fractions were collected and pooled based on an SDS-Gel (Figure 2).
The calculated molecular size of our protein lies at 58 kDa. The gel bands at approximately 58 kDa are therefore likely our protein of interest. It can be seen that all fractions contain the linker protein, while fractions 33 to 36 show the thickest bands, hence highest amount of protein.
While the purification of our ph linker protein worked well, the protein showed no pH sensitive auto-cleavage properties. We tested its stability at pH values from 7.5 to 6 at 37 °C overnight (Figure 3). No cleavage products from our 58 kDa protein are visible. Given the position of the linker, we would have expected protein fragments ranging from 23 to 25 kDa.
There could be different reasons for the (EAAAK)5 linker not showing any cleavage activity. For one, the cleavage activity might not only be related to the linker itself but also parts of the connected proteins used by Wang et al. Furthermore, this auto-catalytic cleavage might be dependent on other outside factors not shown in the paper. Overall, the available characterization of the cleavage function is quite limited and has not been properly reproduced in other literature. If there are certain other conditions needed for the auto-catalytic cleavage to occur, the usability in cell culture/ in vivo conditions is likely low.
During our literature research, we did not find other pH-sensitive proteolytic peptide sequences to mimic the protease induced cleavage of the masked CAR from Han et al. [2]. There might be other ways to achieve pH-specificity, like through pH-induced conformational changes of, for example, the CAR scFv binding epitopes.
The goal of our first experiments was to determine whether we could use a synthetic bioorthogonal ligand dimer to induce a receptor system, which in turn would induce a reporter. For the ligand, we chose mCherry as well as GFP, both as homo- and heterodimer: 2xmCherry, 2xGFP and GFP-mCherry. This way, we were hoping to observe our ligands through fluorescence. Furthermore, the corresponding nanobodies(GBP [3] & LaM-2 [4]) that we used for the receptor system are well established and characterized. The transmembrane and intracellular parts of our receptor system were taken from the Modular Extracellular Sensor Architecture (MESA) from Daringer et al. [5]. For the reporter system, we chose EYFP, which Daringer et al. have used before. We later changed the reporter to miRFP680 by cloning (as described on the Engineering page). The MESA system has been used to induce expression through transcription factor release induced by Vascular Endothelial Growth Factor (VEGF) [6]. We aimed to achieve the same functionality for functionally orthogonal ligands (mCherry & GFP), before turning the system into a loop by replacing the induced reporter with the ligand itself.
As our first testing system was not designed to produce a ligand by itself, we had to purify the relevant proteins. Prof. Scheller generously provided us with plasmids for the expression and purification of mCherry-Fc, GFP-Fc and GFP-mCherry-Fc from Mossner et al. 2020 [7]. Using a protocol adapted from their publication (see Experiments), we were able to obtain all three proteins in one week using ExpiCHO and Protein A column purification. Given the reduced expression time and modified methodology compared to the publication, the purification worked surprisingly well for us (Figure 4). While some additional band impurities are visible, the main bands around 54 kDa (mCherry & GFP) and 80 kDa (GFP-mCherry) are well defined. Most of the lower bands correspond to degradation products that are also visible in Mossner et al. As the dimerization of mCherry- and GFP-Fc is based on disulfide bonds, we also checked the native redox state using N-ethylmaleimide (NEM) containing Lämmli. Given the acceptable purification, we used the three purified proteins for further experiments.
Using GFP and mCherry as homo- and heterodimeric ligands allowed us to design different receptor configurations for our MESA (Figure 5). We combined either the mCherry or GFP nanobody with the transcription factor (TF) or Tobacco Etch Virus (TEV) protease intracellular domain. Schwarz et al. [6] showed that the distance between extracellular receptor domain and transmembrane domain can affect the activation efficiency of the MESA system. Therefore, we designed all constructs with two differently sized domains linking nanobody to the transmembrane domain, called linker 1 and 2 in the following descriptions.
Each half of the built MESA system is encoded on a separate plasmid. To express our starting system in HEK293T, we transfected two corresponding MESA plasmids as well as the reporter (see Experiments for details). We then investigated the functionality of the MESA system by manually adding purified ligand 24 h after transfection. 48 h after transfection, the samples were investigated by flow cytometry. The obtained cytometry data was first gated to obtain only single cells. The used gating strategy can be seen in Figure 6.
First, we wanted to test whether our MESA components are expressed and whether they are capable of ligand binding. We therefore transfected HEK293T cells with only one MESA part. Cells without any MESA part with and without added ligand were used as a control. To all MESA component cell samples ligand was added. As shown in Figure 7, Cells containing a MESA part show a significantly higher signal than those who do not. This indicates a low to no retention of ligand on the cell surface in the absence of receptors. On transfection of the receptors, the corresponding fluorescence signals increase by a factor of at least 1.5 indicating both the expression on the cell surface and ligand binding. The exception here is GFP-Linker 1-TEV MESA showing no significant binding.
Our fluorescence microscopy results visualize these findings. Figure 8 shows immunostaining images of COS7 cells that were transfected with M2TA or M2TEV. DAPI stain was used to image the cell nuclei (blue). The MESA components were stained using immunofluorescence labeling against the HA-tag (red). For both samples red cell outlines can be seen, indicating surface expression of the MESA components.
Combining the results of the cytometry data analysis and immunostaining, it can be said that the MESA components are expressed on the cell surface and that generally ligand binding takes place for both mCherry and GFP system components.
In the next step, we tested whether ligand binding induces reporter expression in our system. Schwarz et al. [6] showed that, depending on the ratio of the two transfected MESA parts, ligand induction changes, in some cases even decreasing with ligand addition. Therefore, we screened every combination of MESA parts for reporter activation with different transfection ratios of the TF and TEV component. An increase in Reporter signaling is visible in all samples compared to the controls with only the reporter transfected (Supplementary Figure 1). The obtained fluorescence signals for the miRFP reporter without and with ligand addition can be seen in Figure 9. Overall ligand addition results in, if any at all, only slight signal differences. The highest one being + 22% for the G1TA M1TEV combination at a 24x ratio of TF to TEV upon GFP-mCherry addition (Figure 7F). As previously shown [6], the ratio of transfected TF to TEV plasmid affects the reporter background signaling. Interestingly, the overall corrected APC MFI seems to be dependent on the transfected TF MESA component, with G2TA showing the lowest signaling induction. While we are not sure of the reason, we theorize that the extracellular receptor might have an influence on the intracellular domain of the MESA component, affecting TF release.
Unfortunately, none of the screened MESA combinations show a significant increase in reporter signaling on ligand addition. Schwarz et al. [6] showed reporter induction of 2x and more for their best MESA combinations. It might be that our chosen incubation times, both before adding ligand and before measuring, are not optimal. Since miRFPs have a longer maturation time compared to other fluorescence proteins [8], it is possible that our miRFP reporter does not yet show a fluorescence signal after 24 h. Additionally, we might not have screened the optimal plasmid ratios. Overrepresentation of the TF half might also be a possible reason. Both cases would lead to a high background signal that is not influenced by the ligand binding. Additionally, there are more linker lengths that might need to be considered. A broader screen with more constructs might reveal an optimal condition. It has to be added that the shown cytometry analysis results only resemble one transfection sample, so one biological replicate. While this is not enough for obtaining significant results, we were hoping to find a promising MESA component and ligand combination in this scanning experiment.
Sample | MFI APC (Median) | MFI DAPI (Median) |
---|---|---|
Reporter Only Control | 5676 | - |
M1TA , M1TEV , 6x Ratio , - | 21768 | 2951 |
M1TA , M1TEV , 6x Ratio , + | 22395 | 2905 |
M1TA , M1TEV , 12x Ratio , - | 23206 | 3132 |
M1TA , M1TEV , 12x Ratio , + | 22502 | 3420 |
M1TA , M1TEV , 24x Ratio , - | 21210 | 3332 |
M1TA , M1TEV , 24x Ratio , + | 22502 | 3266 |
M2TA , M2TEV , 6x Ratio , - | 19576 | 2350 |
M2TA , M2TEV , 6x Ratio , + | 19347 | 2350 |
M2TA , M2TEV , 12x Ratio , - | 16691 | 2433 |
M2TA , M2TEV , 12x Ratio , + | 15967 | 2466 |
M2TA , M2TEV , 24x Ratio , - | 14483 | 2588 |
M2TA , M2TEV , 24x Ratio , + | 14618 | 2604 |
G1TA , G1TEV , 6x Ratio , - | 21820 | 4024 |
G1TA , G1TEV , 6x Ratio , + | 21512 | 4091 |
G1TA , G1TEV , 12x Ratio , - | 20862 | 4955 |
G1TA , G1TEV , 12x Ratio , + | 19854 | 4692 |
G1TA , G1TEV , 24x Ratio , - | 19211 | 4575 |
G1TA , G1TEV , 24x Ratio , + | 20280 | 4934 |
G2TA , G2TEV , 6x Ratio , - | 15455 | 5556 |
G2TA , G2TEV , 6x Ratio , + | 13296 | 5235 |
G2TA , G2TEV , 12x Ratio , - | 11693 | 5125 |
G2TA , G2TEV , 12x Ratio , + | 12698 | 5050 |
G2TA , G2TEV , 24x Ratio , - | 11377 | 5580 |
G2TA , G2TEV , 24x Ratio , + | 9685 | 5114 |
M1TA , G1TEV , 6x Ratio , - | 22502 | 3359 |
M1TA , G1TEV , 6x Ratio , + | 23371 | 3016 |
M1TA , G1TEV , 12x Ratio , - | 22555 | 3332 |
M1TA , G1TEV , 12x Ratio , + | 19438 | 2826 |
M1TA , G1TEV , 24x Ratio , - | 19901 | 3539 |
M1TA , G1TEV , 24x Ratio , + | 21310 | 3332 |
G1TA , M1TEV , 6x Ratio , - | 20424 | 3754 |
G1TA , M1TEV , 6x Ratio , + | 19529 | 3831 |
G1TA , M1TEV , 12x Ratio , - | 19576 | 4407 |
G1TA , M1TEV , 12x Ratio , + | 19948 | 4308 |
G1TA , M1TEV , 24x Ratio , - | 17904 | 4537 |
G1TA , M1TEV , 24x Ratio , + | 21310 | 4453 |
M2TA , G2TEV , 6x Ratio , - | 17286 | 2821 |
M2TA , G2TEV , 6x Ratio , + | 18502 | 2568 |
M2TA , G2TEV , 12x Ratio , - | 15672 | 2665 |
M2TA , G2TEV , 12x Ratio , + | 15563 | 2718 |
M2TA , G2TEV , 24x Ratio , - | 13669 | 2723 |
M2TA , G2TEV , 24x Ratio , + | 14720 | 2476 |
G2TA , M2TEV , 6x Ratio , - | 15419 | 4761 |
G2TA , M2TEV , 6x Ratio , + | 14961 | 4791 |
G2TA , M2TEV , 12x Ratio , - | 12757 | 5393 |
G2TA , M2TEV , 12x Ratio , + | 13481 | 5180 |
G2TA , M2TEV , 24x Ratio , - | 12524 | 5486 |
G2TA , M2TEV , 24x Ratio , + | 12381 | 5849 |
[1] Wu, Y. J., Fan, C. Y., & Li, Y. K. (2009). Protein purification involving a unique
auto-cleavage
feature of a repeated EAAAK peptide. Journal of Chromatography B, 877(31), 4015–4021.
https://doi.org/10.1016/J.JCHROMB.2009.10.009
[2] Han, X., Bryson, P. D., Zhao, Y., Cinay, G. E., Li, S., Guo, Y., Siriwon, N., & Wang, P.
(2017).
Masked Chimeric Antigen Receptor for Tumor-Specific Activation. Molecular Therapy, 25(1),
274–284.
https://doi.org/10.1016/J.YMTHE.2016.10.011
[3] Rothbauer, U., Zolghadr, K., Muyldermans, S., Schepers, A., Cardoso, M. C., & Leonhardt,
H.
(2008). A
versatile nanotrap for biochemical and functional studies with fluorescent fusion proteins.
Molecular
and Cellular Proteomics, 7(2), 282–289.
https://doi.org/10.1074/mcp.M700342-MCP200
[4] Fridy, P. C., Li, Y., Keegan, S., Thompson, M. K., Nudelman, I., Scheid, J. F., Oeffinger,
M.,
Nussenzweig, M. C., Fenyö, D., Chait, B. T., & Rout, M. P. (2014). A robust pipeline for rapid
production of versatile nanobody repertoires. Nature Methods 2014 11:12, 11(12),
1253–1260.
https://doi.org/10.1038/nmeth.3170
[5] Daringer, N. M., Dudek, R. M., Schwarz, K. A., & Leonard, J. N. (2014). Modular
Extracellular
Sensor
Architecture for Engineering Mammalian Cell-based Devices.
https://doi.org/10.1021/sb400128g
[6] Schwarz, K. A., Daringer, N. M., Dolberg, T. B., & Leonard, J. N. (2017). Rewiring human
cellular
input-output using modular extracellular sensors. Nature Chemical Biology, 13(2), 202–209.
https://doi.org/10.1038/nchembio.2253
[7] Mossner, S., Phan, H. T., Triller, S., Moll, J. M., Conrad, U., & Scheller, J. (2020).
Multimerization strategies for efficient production and purification of highly active synthetic
cytokine
receptor ligands. PLOS ONE, 15(4), e0230804.
https://doi.org/10.1371/JOURNAL.PONE.0230804
[8] Shcherbakova, D. M., Baloban, M., Emelyanov, A. V., Brenowitz, M., Guo, P., & Verkhusha,
V. V. (2016). Bright monomeric near-infrared fluorescent proteins as tags and biosensors for
multiscale imaging. Nature communications, 7(1), 1-12.
https://doi.org/10.1038/ncomms12405