If synthetic biology is trying to make cells more like computers, one thing that has been sorely lacking is input-output. It’s hard to get arbitrary information about the outside environment into cells. We do have some sensors that we can modify to sense different inputs, like chimeric antigen receptors (CARs) for T-cells, but they only have one output: T-cell activation.
Now, in a pair of Cell papers from Wendell Lim’s lab, Leonardo Morsut and Kole Roybal remedy this situation with synNotch receptors, essentially amounting to standardized USB cables for eukaryotic cells. Then, to demonstrate its potential, they make dual-input immunotherapy T-cells that can precisely attack tumors without collateral damage.
Notch is a rather unique family of cell-surface receptors that only detect Delta ligands on other cells, in a handshake form of cell-cell communication. If Notch finds another cell that knows the secret Delta handshake, it releases an intracellular transcription factor (TF) that goes into its nucleus and activates response genes.
Previous work had replaced either the intracellular TF domain or the extracellular ligand-binding domain, so this group at UCSF wondered if they could use the core of Notch as a plug-and-play sensor, taking a general input and turning it into a general output. Yes indeed:
- When they tested a combinatorial library of 4 input domains and 4 output TFs, all worked to turn extracellular signals into gene expression.
- Extending the Notch core lowered leaky background expression, probably because the core works as a force sensor. Some particularly bulgy domains may have fit a bit tight, so they just had to loosen the receptor’s belt a Notch.
- Multiple synNotch receptors were orthogonal to themselves and the native pathway, meaning no crossed signals.
- Combining synNotch sensor cells with planted signaling cells, the team could induce patterned cell differentiation.
Combinatorial Sensing For Immunotherapy
One promising approach for tackling cancer is to reprogram immune cells to sense and destroy cancer. This can be hard though, because cancer cells don’t usually have a single defining feature that separates them from normal cells. You couldn’t recognize Donald Trump just by looking for a toupée, you have to consider multiple features. In the same way, immunotherapy would have a lot less collateral damage if reprogrammed T-cells could integrate multiple signals. Using synNotch, the UCSF team could do just that:
- synNotch was used to “arm” a T-cell in response to one input, causing expression of a CAR.
- Only after being armed by synNotch, the CAR activated the T-cell in response to a second signal.
- In mice with tumors expressing either one or both of the signal antigens, immunotherapy T-cells specifically targeted and shrunk only the dual-antigen tumor.
- T-cells didn’t attack single-antigen tumors, even in mice that had arming-antigen tumors in other locations, showing there is little risk of armed T-cells escaping a tumor and attacking normal tissue. This might be because CAR expression only lasts about as long as it takes T-cells to crawl around to another tissue.
Of course, there are always a few caveats:
- synNotch senses force, so it doesn’t work for soluble ligands.
- The system didn’t work well with outputs other than strong TFs, since there is no signal amplification.
Nevertheless, the synNotch system could be hugely useful, especially for cancer immunotherapy or patterned differentiation of stem cells.