Paul Robustelli

I have failed to teach my son anything about entropy or times arrow and hereby withdraw my application for tenure.

GitHub - paulrobustelli/Sisk_IDP_Writhe_2025: Code Accompanying "Characterizing structural and kinetic ensembles of of intrinsically disordered proteins using writhe" Code Accompanying "Characterizing structural and kinetic ensembles of of intrinsically disordered proteins using writhe" - paulrobustelli/Sisk_IDP_Writhe_2025

Check out the code

github.com/paulrobustel...

And the writhe tools python package

pypi.org/project/writ...

Our manuscript on using the knot theory descriptor writhe to build kinetic models of IDP dynamics is now out in JCTC

pubs.acs.org/doi/full/10....

We think this approach establishes a new state-of-the-art for building markov state models of IDP dynamics.

How small molecules stabilize oligomers of a phase-separating disordered protein Small molecule inhibitors of the intrinsically disordered androgen receptor activation domain have been tested in clinical trials for the treatment of castration-resistant prostate cancer. These compo...

Thanks if you made all the way through!

Heres the paper:
www.biorxiv.org/content/10.1...

and the trajectories + code:
github.com/paulrobustel...

Congrats to Jiaqi, Tommy, Borja and Stase on their awesome work!

In the future, we want to see if we can use ligand induced oligomerization simulations to design ligands that more effectively stablize AR oligomers and modulate AR condensates- which could potentially lead to more potent CPRC therapies.

We think that stabilizing R2:R2 interfaces (and other less populated R2:R3 and R3:R3 interafces) may also be how EPI-001 lowers the cloud temperature (Tc) of condensation, rigidifies AR condensates, and interferes with RNA pol II partitioning in cellls.

We think that MD + NMR show that ligands stabilize dynamic oligomerization interfaces of AR AD by bridging interactions between aromatics and forming an extended hydrophobic core that stabilizes helix folding. We see ternary complexes are definitely more rigid than apo dimers, but still dynamic.

So MD suggests these ligands effectively act as "stickers" to bring aromatic residues together from different monomers together.

Good news: so does NMR! @borjaml.bsky.social and Stase found that EPI-001 increases the magnitude of intermolecular NOEs of aromatic protons in Tau-5.

Lets zoom back out to ensemble averages. They key result is that the alkyne linker of 1aa pushes the phenyl groups further apart and gives more rotational freedom. This allows 1aa to simultaneously intercalate into both R2 monomers and form aromatic contacts that stabilize ternary complexes.

Lets take a closer look at the heterogeneity of binding poses and interactions in each kinetically metastable state in the SI.

Maybe one day we can try to design ligands to stablize specific substates?

Lets look at 1aa ternary complexes to see what we're talking about.

Ligands tend to bind at one dominant interface, engage aromatic side chains and induce helix folding, but do so with a collection of fuzzy(ish) binding modes. We feel this matches the magnitude of NMR CSPs in experiments well

We get valuable insight by looking at ensemble-averaged statistics of these states too.

We can see that in apo simulations, residues AWAAAAAQ in R2 associate in a fuzzy manner. Adding ligands extends the apo interaction interfaces and stabilizes and rigidify specific pairs of aromatic contacts.

Here are some snapshots of metastable substates identified by projecting simulation frames onto a kinetically meaningful 2D tCCA latent space.

Dig into the SI if you want to see detailed characterizations of each state.

Characterizing Structural and Kinetic Ensembles of Intrinsically Disordered Proteins Using Writhe The biological functions of intrinsically disordered proteins (IDPs) are governed by the conformational states they adopt in solution and the kinetics of transitions between these states. We apply wri...

But what do these dimer and ternary complexes look like?

Watch out! Here comes Tommy Sisk to work his IDP writhe + kinetic modeling magic.

pubs.acs.org/doi/full/10....

Tommy used tCCA to analyze fluctuations of intramolecular and intermolecular chain writhe and identify metastable complexes

We see 1000s of dimerization events. Lets go back to our NMR autocorrelation function (ACF) fitting days, and show that if we compute an ACF on time series of dimer contacts - you get a double exponential decay, showing two timescales of dissociation, with ligands slowing down both timescales.

We ran simulations of two monomers of the AR AD R2 fragment (apo), and in the presence of EPI-002 or 1aa.

We see that adding ligands substantially stabilizes the intermolecular association of R2 domains and helicity increases in ternary complexes, in perfect agreement with experiments.

So what does this look like an atomic, molecular level?

A single (or small number of) rigid binding interface(s)? Disordered binding? Fuzzy complexes?

Lets get an Anton2 grant and fire-up some unbiased 100 microsecond MD simulations of AR oligomerization in the presence and absence of ligands

Earlier this year they released a beautiful preprint with a trove of NMR, DLS and condensation assay data showing that EPI-001 stabilizes oligomers of the AR AD, predominantly by strengthening intermolecular interactions formed by the R2 helix of the AR AD.

www.biorxiv.org/content/10.1...

First the IRB Barcelona team get some important experimental results.

In 2021 Stase and co. showed the the R2 helix of the AR AD shows the largest NMR chemical shift perturbations (CSPs) with increasing AR concentration - identifying it as the dominant oligomerization interface of the AR AD.

This paper also showed that these ligands partition into AR condensates and lower their cloud temperature (Tc). In vitro and cellular results suggest their activity is potentially linked to their ability to modulate the properties of AR condensates.

But how do they this? By what mechanism?

In our 2023 NSMB paper, @xsalvatella1.bsky.social and Denes Hnisz led an effort to identify new AR inhibitors
www.nature.com/articles/s41..., and Jiaqi found that that MD correctly predicts that more potent ligands with diphenylacetylene moieties have higher affinity to AR AD monomers

In 2016, @xsalvatella1.bsky.social used #NMR to identify the binding site of EPI-001:

pubs.acs.org/doi/full/10....

In 2022, Jiaqi used MD simulations to model monomer binding modes of EPI-002 (the highest affinity stereoisomer of EPI-001):

www.nature.com/articles/s41...

Small molecules that target the disordered androgen receptor activation domain (AR AD) have shown promise for treating castration resistant prostate cancer (CPCR) in mouse models - but have shown insufficient potency in human trials.

If only we understood their mechanisms better....

Proud to share our manuscript:

"How small molecules stabilize oligomers of a phase-separating disordered protein"

www.biorxiv.org/content/10.1...

by Jiaqi Zhu and Tommy Sisk, in collaboration with @borjaml.bsky.social , Stase Bielskute-Garcia & @xsalvatella1.bsky.social

Does anyone else think saying "this paper isn't a good fit for our journal" is quite a bit less silly than saying that a paper isn't a "striking advance" / "sufficiently timeless" / etc. ?.

Who is this goofy form letter language for exactly?

Happy to have contributed a few sections on NMR to this very practical and comprehensive community review on benchmarking force fields against experimental data.

Today Writhe, our favorite knot theory descriptor for characterizing IDP dynamics, is the star of the lab pumpkin for the annual department carving contest.

If you can make sense of the design— I’ll buy you a coffee, bao bun, or wheatgrass shot at the next conference. No substitutions.

Fun day at @dartmouthchem.bsky.social
serving buldak ramen and explaining the biophysics and structural biology of heat and capsaicin activation of the ion channels to celebrate @acs.org #NationalChemistryWeek.

Far out.

Also, take note of our 15 (!) exp. refined all-atom IDP ensembles deposited in the protein ensemble database:

proteinensemble.org/entries/PED0...

If you think you have good method / force field for generating IDP ensembles, you can benchmark your agreement with exp. against these ensembles.