Custom 3D Flex Mesh - protein motion different to 3D variability results

[Kuba]

Hi all,

I’m working with a soluble, ~190kDa protein which has quite a bit of inherent motion. It’s composed of two large domains which, when activated, move in a hinge-like motion towards and away from each other. You can clearly see this motion just from the 2D class images which is nice! I’ve spent quite a bit of time trying to isolate and model the two main states: fully compressed, and fully extended. Unsurprisingly, NU-refinement shows quite a bit of continuous heterogeneity in the map, but thankfully 3DVA clustering has separated this motion into two main particle clusters, each representing the two main conformational states. The volume series of the 3DVA nicely shows the expected hinge-like motion.

However, I know that there is more motion present than what the 3DVA volumes are showing, for example, the head of the protein swivels and twisting should also be present on the main body. Therefore, I am trying to use the 3D Flex job series to extract some of this finer movement (if possible), however I’m struggling to even reproduce the overall hinge motion. After following the tutorial on 3D flex mesh preparation, I generated a custom mesh by segmenting my protein into 3 pieces.

The Fab domains in this tutorial actually displayed hinge-like motion which is not dissimilar to the motion that I expect from my two domains! So using this, I prepared my segment connections in the Flex Mesh Prep job as: 221>223,221>222. My reasoning being that the body is the most stable component, which should allow more freedom of motion for the flexible arm and head regions. Once this runs, I can see that the arm and body densities never separate. The whole movement looks more like a protein breathing motion, rather than what I saw in 3DVA. All domains still look bound to each other and the movement seems restricted, which makes me think my segment connection hierarchy is possibly wrong. I’ve also tried to set the arm region as the stable area (223>221,223>222), but this produces the same result.

I’m not the most experienced user when it comes to the 3D flex job series, so I hope I’ve not missed something extremely obvious. I can’t find something similar in the discussion forum about joined mesh density not separating, so perhaps it’s a basic parameter that needs adjusting which I’m not familiar with. Hopefully I’ve provided enough detail, but I’m happy to produce further results if necessary - thank you for any help you can give me!

Kuba

[Kuba]

Here are the additional images that I wasn’t able to upload in the post:

[Kuba]

[rwaldo]

Hi @Kuba, welcome to the forum!

This looks like an interesting, but challenging problem! I want to start out by saying that a 190 kDa will be difficult to resolve with 3D Flex, but I’ll provide some guidance below.

I also want to highlight that if your head, arm, and body each move independently (that is, the relative rotation of the arm/body is not expected to influence the relative rotation of the head/arm), 3D Flex may not perform better than performing Local Refinement of each region individually (as we do in the Yeast spliceosome case study). WIth that being said, let’s think about your 3D Flex case.

Right now, you’ve segmented your mesh based on the logical organization of your structure – one region for the body, one for the arm, and one for the head. However, this results in all three of your regions being fused together where you want them to move apart. If you look here:

#222’s mesh touches #221’s mesh at the top and bottom of the arm, so the two will not be able to move apart. You may instead want to segment your mesh like so:

With the green region being the root and the pink and yellow regions being attached to green but not each other. This way, the meshes of pink and yellow will be allowed to move apart from each other, since they are not connected at all.