see below
On Wed, Mar 13, 2024 at 3:33 PM Sasha Buzko <Sasha.Buzko.immunitybio.com>
wrote:
> Thanks, Carlos.
>
> I guess, at this point I’m primarily trying to figure what I can do (if
> anything) to improve behavior of the solvent-exposed non-peptide linker.
> From the experimental data that we do have, the peptide shouldn’t be
> binding to the carrier protein and be available for binding elsewhere. Is
> it possible that the GAFF force field used for the linker was parameterized
> for binding pockets and locally hydrophobic environments? In this case, it
> may not work well in bulk solvent.
>
Junmei Wang might be more qualified to answer this, but I don't think this
was done. If anything, a lot of the testing is done on hydration free
energies. It might be possible that the link itself is becoming more
compact, perhaps due to inaccurate internal energies.
> Or can there be another reason?
>
>
>
It also could be related to the chemistry of the linker and if it's well
modeled by GAFF.
Alexey Onufriev did some tests on IDPs with ff19SB+OPC, DOI:
10.1021/acs.jctc.8b01123.
others on this list may have better suggestions.
> Can you think of anything I could at least try?
>
>
>
possibly try a water model or force field that has been validated for IDPs?
generally I don't think training for IDPs should be needed, but our models
all have weaknesses so sometimes you need to choose something tested on
your sort of application.
the DESRES group has done quite a bit of work on water for this kind of
problem - take a look at doi:10.1021/jp508971m if you haven't already. OPC
may be similar but we don't have anything to recommend yet.
carlos
> Thanks again
>
>
>
> Sasha
>
>
>
> ---
>
> Oleksandr “Sasha” Buzko
>
> Computational Biology
>
> ImmunityBio
>
> 9922 Jefferson Blvd
>
> Culver City, CA 90230
>
>
>
> *From:* Carlos Simmerling <carlos.simmerling.gmail.com>
> *Sent:* Wednesday, March 13, 2024 12:46 PM
> *To:* Sasha Buzko <Sasha.Buzko.ImmunityBio.com>; AMBER Mailing List <
> amber.ambermd.org>
> *Subject:* Re: [AMBER] Peptide linked to a protein: force field
> limitations?
>
>
>
> *This email originated from outside of NANT.*
> Do not click links or open attachments unless you recognize the sender and
> are expecting the content.
> If you believe this email is suspicious, report using “*Report Phish*”
> button.
> ------------------------------
>
>
>
>
>
> there could be lots of things going on with the force field and water
> model, but I would start by thinking about what you mean when you say "
> tighter association than experimentally observed". The experiment would
> most likely involve some measure of equilibrium constant, and to compare to
> that you would need to see many events where the peptide reversible
> contacts the protein and then dissociates, as with a small molecule
> equilibrium process. That would be very challenging to simulate with any
> statistical reliability. Certainly seeing it move to the protein surface
> could be consistent with experiments if there is any binding affinity at
> all - but also could be inconsistent and the association might be too
> strong. You really can't say much from your current data.
>
>
>
> but again, it could well be an issue with the computational models. there
> are definitely articles in the literature describing over-stabilization of
> compact structures, particularly notable for simulations of IDPs. I think
> it's an area of active research since (1) experimental data have been
> scarce and (2) generating a reliable distribution of these unfolded
> structures is typically too costly to achieve.
>
>
>
>
>
> On Wed, Mar 13, 2024 at 10:56 AM Sasha Buzko via AMBER <amber.ambermd.org>
> wrote:
>
> Dear All,
>
> I'm running explicit solvent simulations of a globular protein with a long
> linker connected to a helical peptide. The linker is a long CYX-bound
> non-peptide with a PEG-amide type of structure. I use ff19SB with OPC water
> model, add ions to ~0.15 mM level and to neutralize the system.
>
> The equilibration runs fine but then I see an overall drift of the
> connected peptide to the carrier protein. It's basically trying to fold
> into the protein.
>
>
>
> As a control, I ran 1qm9 (solution NMR structure), which has two loosely
> associated subdomains connected with a peptide linker. After 200 ns, the
> subdomains came much closer to each other and the linker remained solvated.
> In other words, there is tighter association than experimentally observed.
>
> Not sure how much the non-peptide nature of my linker is impacting this.
>
>
>
> I'm wondering how much of it is real and how much is due to underestimated
> solvation. Does anyone have experience modeling these types of
> solvent-exposed structures? Do you think I've hit the limits of the force
> field or am I missing something?
>
>
>
>
>
> Thank you in advance!
>
>
>
> Sasha
>
>
>
>
> ---
> Oleksandr "Sasha" Buzko
> Computational Biology
> ImmunityBio
> 9922 Jefferson Blvd
> Culver City, CA 90230
>
> CONFIDENTIALITY NOTICE
> This e-mail message and any attachments are only for the use of the
> intended recipient and may contain information that is privileged,
> confidential or exempt from disclosure under applicable law. If you are not
> the intended recipient, any disclosure, distribution or other use of this
> e-mail message or attachments is prohibited. If you have received this
> e-mail message in error, please delete and notify the sender immediately.
> Thank you.
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
> <https://urldefense.com/v3/__http:/lists.ambermd.org/mailman/listinfo/amber__;!!LM3lv1w8qtQ!Sor9wUqbaFOVCVyW2mg4LUkH_uUrWrbUM_8pWFeFBSQ_zvwJxO1-cbSI3u54h6aNNXYfZPnfuT3eqQ7ITMKguzM2x6GQ-BqFmg$>
>
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Mar 13 2024 - 14:00:02 PDT
