Advice on Monitoring Passive Membrane Permeation via Unbiased MD

User discussions
1

GROMACS version: 2025.2
GROMACS modification: No

Hi all,

We’re currently exploring whether a particular molecule enters cells via passive diffusion across the lipid bilayer or through endocytosis. To probe this, we’re using MD simulations to assess its membrane permeability.

In our unbiased simulations, we observe repeated binding and unbinding events between the molecule and the membrane surface, but no leaflet-crossing events — the molecule never traverses from the top leaflet to the bottom and exits the membrane.

We’re wondering:

  • Are bilayer-crossing events typically rare in unbiased simulations due to high energetic barriers?

  • Is there a recommended unbiased strategy to observe these events, or would enhanced sampling (e.g., umbrella sampling) be more appropriate?

  • If we go the umbrella sampling route, are there reference systems or standards that are commonly used for benchmarking small molecule permeability?

We haven’t been able to find published examples where unbiased MD alone was sufficient to estimate passive permeation of small molecules, so any pointers (papers, parameters, collective variables, etc.) would be greatly appreciated.

Thanks in advance!

2

Hi @brjoel

Are bilayer-crossing events typically rare in unbiased simulations due to high energetic barriers?

This depends on the molecule you are investigating, especially on how big and how hydrophobic/hydrophilic it is, and on the type of lipids in the bilayer (some might be generally harder to permeate, like ceramides, than other, like POPC). In my experience, very rarely people can observe full permeation through a lipid bilayer via unbiased MD simulations. If the molecule is more hydrophilic then it won’t partition in the bilayer, while if it is more hydrophobic it might partition but then it won’t leave.

Is there a recommended unbiased strategy to observe these events, or would enhanced sampling (e.g., umbrella sampling) be more appropriate?

Again, it depends on how difficult it is for the molecule to permeate. I would say that most people use some kind of enhanced sampling technique, as even in the case of crossing you will need several events to quantify the permeability. There are, however, examples of people running unbiased MD, like this, but as you can see the permeants are small, the bilayers rather permeable, and they still require long simulations and many crossing events.

If we go the umbrella sampling route, are there reference systems or standards that are commonly used for benchmarking small molecule permeability?

I don’t know if there is a gold standard, but there are several reviews, like this, that can help in choosing an approach. In general, if I want to benchmark my set up for permeability prediction, I would check for works that use the same force field and membrane type and go for an easy permeant that they report and compare. Also, consider that other than umbrella sampling there are several other techniques to enhance the sampling, like this and this cool approaches that are really useful if you have complex systems and complex permeants.

3

Thanks!