Discovery and validation of the binding poses of allosteric fragment hits to PTP1b: From molecular dynamics simulations to X-ray crystallography

  1. Jack B. Greisman
  2. Lindsay Willmore
  3. Christine Y. Yeh
  4. Fabrizio Giordanetto
  5. Sahar Shahamadtar
  6. Hunter Nisonoff
  7. Paul Maragakis
  8. David E. Shaw

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Fragment-based drug discovery has led to six approved drugs, but the small size of the chemical fragments used in such methods typically results in only weak interactions between the fragment and its target molecule, which makes it challenging to experimentally determine the three-dimensional poses fragments assume in the bound state. One computational approach that could help address this difficulty is long-timescale molecular dynamics (MD) simulation, which has been used in retrospective studies to recover experimentally known binding poses of fragments. Here, we present the results of long-timescale MD simulations that we used to prospectively discover binding poses for two series of fragments in allosteric pockets on a difficult and important pharmaceutical target, protein-tyrosine phosphatase 1b (PTP1b). Our simulations reversibly sampled the fragment association and dissociation process. One of the binding pockets found in the simulations has not to our knowledge been previously observed with a bound fragment, and the other pocket adopted a very rare conformation. We subsequently obtained high-resolution crystal structures of members of each fragment series bound to PTP1b, and the experimentally observed poses confirmed the simulation results. To the best of our knowledge, our findings provide the first demonstration that MD simulations can be used prospectively to determine fragment binding poses to previously unidentified pockets.

Article activity feed

  1. PREreview

    This Zenodo record is a permanently preserved version of a PREreview. You can view the complete PREreview at https://prereview.org/reviews/7743207.

    Summary:

    Molecular dynamics simulations (MD) have emerged as an important tool in drug discovery. Their application to prospectively discover binding sites and to screen compounds interacting at those sites, remains a frontier application, even for compounds with high potency. Here, simulations are used to run a "virtual fragment screen"/"swimming experiment" for two compounds with very weak binding affinity that they identified as potential hits by SPR. In some sense, because of the weak affinity and limited interactions, these results are much more impressive and surprising than the landmark "How does a drug molecule find its target binding site" paper from the Shaw group a decade ago (https://pubmed.ncbi.nlm.nih.gov/21545110/). 

    The poses are validated with crystal structures of these fragments bound to PTP1b. The major strength of this paper is showing that MD simulations can identify weakly bound fragments at allosteric sites (even those not previously highlighted extensively in fragment screens) with binding poses that closely resemble experimentally obtained structures. The major weakness of this paper is that the authors only performed "swimming" on two fragments, which does not allow us to generalize how well MD simulations both predict other allosteric sites and how well MD simulations predict fragment poses of larger chemical spaces of varying fragment sizes. Given the large number of negative controls available from the SPR screen and the relatively large number of positive (ish? - as affinities weren't measured) controls from Keedy et al fragment screens, a larger study could be conducted - but for now, this work remains a tantalizing glimpse into the capabilities and potential of the swimming method for PTP1b.

    Major points:

    1. The authors initially performed SPR to identify two fragments for MD simulation and crystallization but this appears to be a very small number of fragments for MD. Could they discuss how feasible (or not) would it be to perform MD simulations of the whole library to identify potentially weaker fragments that may not have been detectable by SPR? Additionally, how might the "swimming" approach compare to a more conventional pipeline (e.g. identifying binding sites and fragment pose compare to mixed solvent MD with virtual docking)?

    2. The authors describe two binding sites for the fragments they identified through SPR, one that was previously identified (DES-4884) and a second site that they report as an allosteric site (DES-4779) but do not describe or provide structures of any broader structural changes to PTP1b that occur as a result of fragment binding at either site. The authors should show an alignment between apo and bound states of PTP1b to highlight allosterically induced structural changes. 

    3. The authors highlight the DES-4884 fragment inducing two phenylalanine rearrangements (Phe196 and Phe280). While Phe196 conformational change has also been reported in fragment screens, they say that the Phe280 also swings out but don't explain whether that could be of significance to future fragment screens (does Phe280 create new binding opportunities, increase/decrease fragment affinity, induce broader changes to the structure of PTP1B?) It would be helpful if the authors could contextualize the phenylalanine rearrangements with what is previously known about this allosteric site for future fragment screens. 

    Minor points:

    1. We feel that the crystallography section of the methods and materials is incomplete. It would be helpful if the authors: 1)  explained what PTP1b construct they used, 2) the crystallization conditions, 3) the structure resolutions, 4) method of obtaining fragment bound structure (fragment soaking vs co-crystallization) and 5) how they identified the fragments in their crystal structures. Further information about how crystal diffraction data was collected and processed would be helpful.

    2. Figure 3e: The authors show the DES-6016 variant of DES-4884 makes hydrogen bonds with at least two water molecules. Do the authors propose that this allosteric site requires coordination with water molecules for fragment binding? How stable are the water molecules in the MD simulation to support this?

    3. The authors mention that for each fragment there were additional minor binding sites from the MD simulations. During the crystallization experiments, were fragments detected at these binding sites as well?

    Reviewed by CJ San Felipe (UCSF) and James Fraser (UCSF)

    Competing interests

    The author declares that they have no competing interests.

  2. Version published to 10.1101/2022.11.14.516467v2 on bioRxiv
  3. Version published to 10.1101/2022.11.14.516467v1 on bioRxiv