Last but not least, we show how soluble components could be included in a model such as this by increasing the complexity. This creates defined matrix, intermembrane and cytoplasmic areas

The initial simulations provided us the chance to have a quick look at properties we can measure from simulations such as this

After this, it was time to put it all together and simulate! The entire model was ~4 million Martini particles with 651 protein chains!

The curvature of the inner mitochondrial membrane meant the assembly took a bit more care, but using Blender we created a map to place our proteins and lipids on in the right locations

Then came time to assemble both membranes found at a crista! The outer membrane was more straightforward to build, with a flat bilayer being a good approximation

All of the protein models were simulated on a smaller scale, to measure stability, membrane curvature and lipid interaction. The final poses with surrounding lipids could also be fed into the larger models

But there weren't always experimentally resolved structures to work off! The large subunit of MICOS was modelled with AlphaFold and verified using available biochemical data before including further in our models

First off, to get the structures of human mitochondrial membrane proteins, we used AlphaFold to complete experimental structures when chains or regions were missing

What better way to use my first proper post than to share my first big piece of post-doctoral work with @cg-martini.bsky.social!
Here, we used integrative modelling to build and simulate a mitochondrial cristae.
Find the paper here (rdcu.be/eujAC) or see below for a quick overview 👇