Protein Crystals: Programming Living Cells as Microreactors!

Protein crystals have been used to study protein structures. But they have so much potential! New materials, better drugs, or intracellular recorders.

The problem? They’re hard to make.

Can building them inside cells help solve the problem and bring protein crystals out of the structural biology lab? Today’s paper tries! It’s an exciting one.

Proteins are super versatile.

Enzymes catalyze reactions, hemoglobin transports oxygen in the blood, and actin creates a tiny skeleton inside cells. Proteins do so many things! They can even form crystals, just like salt or diamonds! Proteins self-assemble into ordered, repeating 3D lattices with pores ranging from ~2 to ~50 nm.

But why are they interesting?

Well, they have potential applications in biotech and biomedicine:

That’s a pretty exciting toolbox.

And these kinds of porous materials are getting popular. Just look at MOFs (metal-organic frameworks). Their whole thing is that they’re extremely porous, giving them a crazy surface area. They’re perfect for carbon capture, water purification, and drug delivery.

And the scientists working on them got the Nobel Prize in 2025!

So, protein crystals could be useful too. And they have nice advantages:

Okay, awesome! Why aren’t we using them already?

Well, making them is hard. Right now, in the standard route, you produce the protein in cells, purify it, and then crystallize it in vitro under specific conditions. This works, but it’s a pretty uncontrolled process, especially when it comes to crystal nucleation and growth. The result is batch-to-batch variation, which makes applications much harder!

Can we do better?

An alternative? Intracellular crystallization.

The idea is simple: make the cell itself the reactor. Instead of purifying protein first and crystallizing later, you let the cells make the crystals directly. That means no extraction, no extra handling, and much more control over the process.

This has been tried before, but today’s paper gives it a much more programmable framework!

The team built an intracellular platform for controllable and programmable protein crystallization in living mammalian cells! The result? Cells act as microreactors that confine and regulate crystal growth.

And the control goes beyond simple size. By combining genetic tagging and click chemistry, the authors can spatially pattern the crystals, load them with different materials, and even use them as drug-release vehicles!

So, how does it work?

The team started with previously designed two-component lattices.

They constructed 3 different lattices and then mutated key interface residues to generate a library of variants. When these variants were expressed in HEK293T cells, some of them formed intracellular assemblies!

Among them, they focused on FC1, which formed a high-symmetry cubic lattice with:

A perfect scaffold for guest materials! After mutational screening, they identified FC1m as the best intracellular crystallizer. Structural analysis confirmed that it closely matched the design.

FC1m is the main character of the rest of the paper!

One of the strongest results is that crystal growth inside cells is predictable. The authors modeled the cells as a microreactor that continuously supplies protein building blocks.

Under steady-state growth, the building blocks are synthesized, degraded, and incorporated into crystals at constant rates. They could even derive a growth law for the crystal edge length!

And they validate their model with microscopy. In HEK293T cells:

And when they reduced FC1m expression, the nucleation was delayed to 12 h or 17 h, depending on the expression level. So both nucleation timing and crystal growth rate can be tuned!

The coolest part of the paper.

The authors asked: Can we spatially program the crystals? To test it, they fused Halo-Tags to protein monomers and labeled the growing crystals with different colored dyes at defined times.

Since the crystal grows outward over time, the labeling creates coloured layers! The researchers show that:

So, now the porous crystal is also spatially programmable!

But the team didn’t stop at dyes.

They combined HaloTag with orthogonal click chemistry to install different reactive handles into specific crystal layers and then load guest materials using click reactions.

They used 4 ligands:

These support 4 orthogonal reactions, letting them program up to 4 distinct materials into the same crystal. They tested with:

Amazing!

But what can you do with a spatially programmed crystal?

The answer: controlled release. The team turned FC1m crystals into vessels for programmable drug release! First, they took five-layered FC1m crystals and tested their dissolution in simulated physiological environments.

They varied the pH from ~4.5 to ~6.8, and they saw that:

Great! So it does dissolve; now, what can you load on these crystals?

They settled on bFGF (basic fibroblast growth factor), a signaling protein crucial for cell proliferation, wound healing, and tissue repair. Inside cells, it triggers signalling cascades like the Akt pathway.

The team loaded bGFG into the 5-layered protein crystals and then incubated them with HeLa cells carrying an Akt biosensor to monitor the signal in real time.

When the culture is set to pH 6.8, the released bFGF causes oscillatory Akt signaling in the cells, with the pattern depending on the programmed layer order:

This is strong proof that the system can encode not only what is released, but also when!

Such cool work!

I liked how it uses cells as controllable reactors for growing protein crystals. Protein crystallization is hard. Doing it inside cells makes it more predictable and more scalable, and it could unlock many interesting applications.

And the programmability here is exciting! Imagine layering the different enzymes that make up a reaction chain, or drugs that target different resistance mechanisms…

Plus, this could unlock new materials: more sustainable conductive materials, better carbon storage options, or biomaterials that work better with electronics. There’s so much room here!

But the authors warn of some limitations:

So, still some kinks to solve, but this is a strong first step! And ehi, it’s still controllable, biocompatible, and programmable.

I liked this paper! Go here and read the whole thing. The figures are beautiful!

If you made it this far, thank you! What do you think of protein nanocrystals? Do you think they have a place in biomedicine? Do you think that producing them in cells is going to be a game changer? Reply and let me know!