Computer-Designed Light-Switch Proteins: Controlling Biology with Photons!

Is light the original life control? In a way! So, what if we could control it better, with protein designed from the ground up? That’s what today’s paper is all about!

Light-Switch Proteins, from Scratch

Lots of proteins respond to light.

Sometimes, I forget that. Light is essential for life, from a plant’s photosynthesis to our own vision, and proteins are always ready to respond to photons coming their way.

Luckily, other scientists didn’t miss this opportunity. They’ve engineered light-responsive proteins to manipulate biology and created optogenetics and optobiochemistry, where light combines with genetics to control cell and protein activity.

But how do these proteins work?

Usually, they rely on chromophores, molecules that absorb light and cause proteins to change shape. Natural photoreceptors are powerful tools, but they come with limitations. Since they evolved for specific roles, our ability to control and reprogram them for new functions is limited.

Protein Design: Beautiful, but Tricky

In the meantime, protein design has exploded.

Now, you can go from protein binders to cages and crystals in a moment! With structures never seen in nature. There is an explosion of methods and results, especially after machine learning-based methods got popular.

Still, designing proteins that change in response to a trigger (like light) remains a challenge.

For light-responsive proteins, the go-to approach is to optimize natural proteins for new processes. But this doesn’t work all the time! Especially when it comes to entirely new applications.

Light-Responsive Proteins from the Ground Up

Enter today’s paper.

The team built de novo protein complexes that disassemble and reassemble on command, controlled by light!

It all revolves around azobenzene, a molecule capable of flipping between a cis and a trans structure based, you guessed it, on light. Light at 340 nm pushes it from planar (trans) to bent (cis). Light at 420 nm brings it back.

This transition creates steric clashes, pushing nearby atoms out of place and breaking weak interactions. Scientists have used this trick to control motor proteins, DNA origami and nanomachines! And azobenzene even has an amino acid version, in the form of phenylalanine-4’-azobenzene (AzoF).

AzoF is one of many so-called non-canonical amino acids that expand protein chemistry beyond the usual 20. A powerful way to add new functionalities to proteins! Although it requires a bit of work.

In short, the idea is to design proteins that interact and form complexes, with AzoF at the interface. When light hits, AzoF changes shape. This change interferes with the other amino acids around, making the complex fall apart!

They had 3 main challenges:

1. Designing specific protein interfaces that actually bind

2. Positioning AzoF at the interface to enable light-triggered disassembly, via induced steric clashes.

3. Designing components that, when free, maintain solubility and avoid aggregation.

It doesn’t sound easy! Let’s jump into how they did it.

How it Works: Design Strategy and Pipeline

A little problem: existing protein design pipelines don’t work with non-canonical amino acids. So, they built their own.

They started with small helical-bundle scaffolds of 3-4 helices: the building blocks for the light-responsive complexes.

Computationally, they mapped possible positions for AzoF and built a table with the best positions to incorporate AzoF, relative to each scaffold. This sped up the process and enabled a rapid assessment of sites where trans AzoF would pack well against a partner.

In the second step, they modeled how the scaffolds could assemble into complexes, optimizing the interface and the interactions of the trans AzoF. Finally, Rosetta and ProteinMPNN optimized the sequence and the chemical interactions for the most promising designs.

And from here? It’s time for experiments!

Experiments Time: Shining Light on New Designs

They designed two classes of light-responsive proteins: cyclic homo-oligomers and heterodimers.

Designing Cyclic Homo-Oligomers

Cyclic homo-oligomers are ring-shaped complexes made from multiple identical subunits.

They play key roles in biology, from catalysis to signaling and biomaterial building! So, you can imagine that controlling them with light would be powerful.

The team built lots of designs with different symmetries:

• 45 dimers

• 67 trimers

• 43 tetramers

• 45 pentamers

Of these, 22 formed the expected oligomers (not a huge success rate!). And from the 22, 13 responded to light and disassembled at 340 nm and reassembled at 420 nm. 5 of them showed a near 100% disassembly rate!

Light-Responsive Hetero-Dimers

Next, the team designed heterodimers, where a pair of proteins interacts, one with AzoF, the other one “vanilla”.

Out of 36 designs, 23 formed dimers, with 3 having high disassembly efficiency!

In one design, the binding affinity dropped 167x after light exposure! No wonder it disassembled.

Structural Validation: From Models to Reality

The team solved multiple crystal structures of the designed complexes, for homo-oligomers and hetero-dimers. Most of them could be superimposed on the design model with atomic accuracy!

The AzoF electron density is also clearly resolved and matches the designed placement, confirming that the computational pipeline can place AzoF precisely in the interface.

Applications: Hydrogels and Receptors

Designing switches is cool, sure.

But the team wanted to use them!

Light-Regulated Ligand-Receptor Systems

The authors converted one heterodimer (called LRD-7) into a ligand/receptor pair.

They took GEMS, a modular receptor scaffold, and fused it to the vanilla chain of LRD-7, and expressed it in cells. The other chain, with an inactive AzoF, is presented externally, and after light irradiation, it can bind to the receptor.

This activates a downstream pathway, boosting gene expression up to 18x above baseline. And the activation was reversible and time-dependent, demonstrating programmable optogenetic control!

Light-Regulated Hydrogels

My favourite part!

The authors built a hydrogel by combining a light-regulated pentamer (LRO-C5-1) and a dimeric crosslinker. The building blocks connect into a network, forming a light-sensitive gel.

Under 340 nm light, the gel softened dramatically, turning practically liquid. Its storage modulus (a measure of stiffness) dropped from 91 Pa to just 9 Pa, and switching back to 420 nm reformed the gel!

Super interesting for advanced materials!

Conclusions: Looking Ahead!

Great work! And it has the best figures I’ve seen in a while.

I’m happy to see protein design moving so fast and creating more functional designs. Scientists made incredible discoveries using light-responsive proteins, and it’s exciting to see new designs!

This approach is powerful, but still has some limitations:

• Modest initial success: Many designs fail, so an iterative design-build-test is needed. But I’m sure this will improve! For example, deep learning algorithms have decent success rates.

• Genetic code expansion: Using non-canonical amino acids requires adding machinery to incorporate them into proteins, which adds a layer of complexity.

• Wavelengths & tissue penetration: The wavelengths used have limited tissue penetration and are potentially toxic in vivo.

But all these problems can be solved! It would be cool to see more red-shifted chromophores, for example, which have higher tissue penetration and less toxicity.

So, go read the paper here! Lots of details, they even figure out predictors for design success.

If you made it this far, thank you! What do you think of computational protein design? What do you think could be the next big target? Reply and let me know!

More Room:

• CRISPR Imaging: CRISPR has many uses, and imaging is getting up there in popularity. This review highlights strategies to improve CRISPR-based live-cell imaging, focusing on enhanced signal labeling and efficient intracellular delivery using nanomaterials. It outlines current challenges( weak signals, off-target effects) and suggests that combining CRISPR with advanced nanotechnology could enable more precise, real-time molecular imaging in biology and medicine.

• Polymer Donuts: Okay okay they are toroids, I know. But “donuts” is funnier. This study reports a better method to synthesize uniform nanotoroids using a precisely designed bottlebrush copolypeptoid. The polymer’s tailored amphiphilicity and flexible architecture drive toroid formation through an intramicellar end-to-end coalescence mechanism. Control experiments and simulations confirm the mechanism, distinguishing it from other morphologies like sheets or micelles.

• Fake Cytoskeletons: We have seen many DNA nanotech ways to create synthetic cytoskeletons. But what about proteins? This study introduces a nature-inspired protein design strategy to build dynamic, tubular assemblies from two distinct protein units. These artificial structures mimic actin-like flexibility and reversibility, disassembling and reassembling in response to environmental cues like salt and temperature. Using cryo-EM, the researchers reveal diverse, helical morphologies resembling natural filaments.

• Share Plenty of Room with founders or builders

  I help biotech and deep tech companies transform complex technologies into engaging content that builds credibility with investors, partners, and potential hires. Let’s chat!

• Know someone who’d love this?
  Pass it on! Sharing is the easiest way to support the newsletter and spark new ideas in your circle.

• Got a tip, paper, or topic you want me to cover?
  I’d love to hear from you! Just reply to this email or reach out on social.