Cells are the building blocks of life. The way cells recognize other cells and external signals can lead to a variety of biological consequences, including cell growth, death, and motility. Researchers want to understand cell-cell communication, reverse engineer it, and ultimately sculpt cell interactions that surpass natural capabilities. Although cell therapies already exist, the future of such cell therapies will likely involve profound modification of patient cells to treat a spectrum of diseases and repair tissues.
In a previous article, we reviewed a study that modularly replaces the extracellular region of proteins to recognize different ligands; These “reassembled” proteins transduce the same signaling pathway as long as the transmembrane and intracellular compartments remain intact. Here, we discuss a paper that instead emphasizes the intracellular compartment of the cell. Researchers at the University of California, San Francisco theoretically reconstructed the signaling domains of CAR T cells and explored possible effects on cell-cell communication.
Construction of a chimeric antigen receptor
cchimeric Aantigen AndThe receptor (CAR) requires genetic modification to express new, synthetic components. Figure 1 depicts the three main regions of a CAR T cell: the antigen-binding domain, the transmembrane domain, and the signaling domain. Scientists often fixate on the binding domain and tailor it to a specific therapeutic target (eg, a protein found in cancer cells). Here the researchers, however, focus on the structure of the signaling domain and its effect on CAR T cell function.
A CAR T cell signaling domain typically contains a CD3ζ T cell receptor (TCR) molecule and any combination of costimulatory molecules. Costimulatory molecules contain multiple signal motifs or small peptides that bind to specific downstream signaling molecules. These molecules affect T cell signal transduction in a variety of ways. Two examples include 4-1BB, which can enhance T cell memory and persistence, and CD28, which is associated with effective T cell killing but reduces T cell persistence.
Expanding the possibilities through machine learning
Researchers in the Wendell Lim lab sought to uncover the unexplained rules that govern costimulatory signaling and thereby optimize the properties of cART cells. They used a synthetic signal motif library, machine learning, and a unique conceptual approach to discover combinations beyond what occurs naturally.
From words, to sentences, to language
Researchers look at natural signaling proteins, pull signal motifs from them, and synthetically assemble combinations of signaling motifs to create unique signaling programs. This approach can be conceptualized as a sentence-building exploration.
Figure 2 illustrates this rearrangement between different “words”—signal motifs—individual “sentences” or signal programs. To understand and predict the “language” of these combinations, the team then used machine learning algorithms called neural networks to identify the underlying “grammar” of the dataset. This reveals the importance of word order, word meaning, and word combinations in the final product—otherwise reorganized as signaling motifs affecting T cell phenotype identity, function, and arrangement.
The team generated a library of anti-CD19 CAR T cells with different costimulatory domains. Each cell contains one, two or three signal motifs derived from natural signaling proteins (see Figure 2). The team inserted 12 native signal motifs alongside a spacer motif at random positions i, j And k to generate a total of 2,379 distinct motif configurations, as seen in Figure 3.
Next, the researchers screened random subsets from the library to classify T cell cytotoxicity and ability to proliferate (stemness). This process produces unique and unusual combinations, comparable to costimulatory molecules 4-1BB (ie: M10-M1-M1-ζ).
Decoding “Language” Using Predictive Neural Networks
Signaling motif sequences have different levels of cytotoxicity and stemness according to experimental analysis. The team then used this data to understand the invisible rules surrounding costimulatory molecule design.
An artificial neural network proved crucial for this investigation. As seen in Figure 4, the data were partitioned to train or test algorithms for predicting the cytotoxicity or stemness of a chimeric antigen receptor. This mechanism has explained several associations, such as the ability of costimulatory domains such as 4-1BB to enhance cytotoxicity and stemness.
Successful prediction with M1 costimulatory molecules
Can neural networks accurately predict the fate of a T cell with a specific costimulatory combination? The team tested the waters by adding the costimulatory molecule M1 to the 4-1BB-like versus CD28-like signaling domain. The neural network predicted that adding the M1 motif would exhibit enhanced cytotoxicity and stemness in the 4-1BB-like domain while having no effect on the CD28-like counterpart.
In an in vitro model, CAR T cells with a 4-1BB-like domain and M1 motif effectively killed tumor cells and maintained T cell stemness; On the other hand, the addition of the M1 motif did not cause any change for the CD28-like derivative. This accurate prediction also translated into results in mouse models. 4-1BB/M1 CAR T cells delayed tumor cell growth for two weeks longer than 4-1BB alone CAR T cells. These observations show how a neural network can be used to accurately predict T cell properties depending on the synthetic signaling motifs involved.
Prospects for CAR T therapy
It can be difficult to predict how a synthetic receptor component will affect the resulting cell properties. This study unravels some of this mystery with signal motif libraries and machine learning. By analyzing CAR T cell costimulatory domain combinations, the team developed a neural network that successfully predicts T cell phenotype based on the costimulatory molecules present. This, in turn, revealed the rules of CAR T costimulatory signaling that could be used to design better synthetic signaling domains. Similar libraries and subsequent analyzes can be applied to improve other modular regions of a CAR T cell.