From: Gumbart, JC (
Date: Sun Apr 04 2021 - 13:39:18 CDT

It’s not that we thought constant pH wouldn’t work, but just that we are not especially familiar with the method. Here’s an example of a study on a related enzyme in which they used it successfully:;!!DZ3fjg!qiAD3f7mt7lc30MuT_0aXrx3MwRRPaAlonE0FSFEs9I85WynxYQ-mzOPINsueH5f3g$

I’m also not sure that constant pH would be shorter than the individual simulations we ran, given that I think (but could be wrong) that it’s slower, plus we would have had to run longer than 20 ns to get anything converged. Not to mention the time it would have taken us to become familiar with the approach and how to use it well.

You have a range of methods to choose from with varying levels of complexity, including just guessing based on the local environment (what we often do!), semi-empirical methods (Schrodinger’s protein preparation wizard may be able to handle nucleic acids? but it’s not free), running multiple simulations to test to different combinations of assignments (you don’t have to be exhaustive - many can be disregarded at the outset), and then constant pH. None of them are necessarily right or wrong and selecting one just depends on the time/accuracy trade-off you want to make.


On Apr 4, 2021, at 9:53 AM, Francesco Pietra <<>> wrote:

I never used constant pH MD, although I was aware of NAMD implementation of it.
The reason to investigate HIS protonation was twofold: the overwhelmingly presence of nucleosides and the two recent papers mentioned on previous posts in this thread, where constant HIS protonation was used and the various possible states explored by MD. Why they did not take the shorter route of constant pH MD? The authority of the authors let me think that constant pH MD is no panacea.


On Sun, Apr 4, 2021 at 2:28 AM Miro Astore <<>> wrote:
You might consider constant ph simulations? They're supported in namd.

On Sat., 3 Apr. 2021, 6:43 pm Francesco Pietra, <<>> wrote:
Hi Josh and James
I had already seen another checking the MD stability for the various protonation states of HIS:

Histidine protonation and the activation of viral fusion proteins, Daniela S. Mueller, Thorsten Kampmann, Ragothaman Yennamalli, Paul R. Young, Bostjan Kobe
and Alan E. Mark, Biochemical Society Transactions (2008) Volume 36, part 1

With so many HISs as in my case, the combinations scale tremendously up and one fears to miss the most important overall state.
There is a need of new software to assign the protonation states of HIS. Also, if there is a fine tuning of HSE vs HSD, changeable during MD in response to the interaction with the strongly negative electrostatics of the nucleosides, things become complicated indeed.

But I'll read James' paper and I'll try the suggested Propka and DelphiPKa, if not else to see the output.

What about the famous ROSETTA? Perhaps, however, it just uses Propka or Delphi PKa.


On Sat, Apr 3, 2021 at 1:35 AM Vermaas, Josh <<>> wrote:
Admittedly, it has been a while since I looked at propka output in detail, but when I run regular proteins through pdb2pqr (;!!DZ3fjg!qiAD3f7mt7lc30MuT_0aXrx3MwRRPaAlonE0FSFEs9I85WynxYQ-mzOPINuHafiJkQ$ <;!!DZ3fjg!tsqIFHVKijNxX_B1fbFz9jygI2pMsbnGZC8el-7pPOZ9fT3WcvW5y8s__expakzqDA$>), it claims to use PropKa to assign charges, and the pqr files that come out definitely seem to have a made an attempt to choose between HSD and HSE, so I assume there is something in the PropKa output that it is using to determine histidine protonation. For the applications I was using it for, this level of detail has been fine, but JC is right, you can also just try all protonation state combinations and see what makes sense.


From: "Gumbart, JC" <<>>
Date: Friday, April 2, 2021 at 4:43 PM
To: "Vermaas, Josh" <<>>
Cc: Francesco Pietra <<>>, VMD Mailing List <<>>, "<>" <<>>, "<>" <<>>
Subject: Re: vmd-l: histidine protonation

I’m not sure about DelphiPka, but Propka doesn’t tell you which nitrogen to put the proton on. We found running simulations of different states and checking their stability to be the most reliable (albeit more time-consuming) way to assign protons to histidines:;!!DZ3fjg!qiAD3f7mt7lc30MuT_0aXrx3MwRRPaAlonE0FSFEs9I85WynxYQ-mzOPINuaSFpX1g$ <;!!HXCxUKc!gvFYR3OpPr1fjuH1jJQ1NtrWmWfgzesHlw-bJ5W_WnlWlHNPuvfRXD9_XBJvm84$>

We ran for 20 ns x 3, but you can often see a disruption sooner.


On Apr 2, 2021, at 2:26 PM, Vermaas, Josh <<>> wrote:

Hi Francesco,

Propka (;!!DZ3fjg!qiAD3f7mt7lc30MuT_0aXrx3MwRRPaAlonE0FSFEs9I85WynxYQ-mzOPINvQQCkSzw$ <;!!DZ3fjg!qLqvG5aUIsZadcvLdv2KKVS99Ta3HEX53KL7IdL8pM3CNQJ6bPm9CzlpqFDRl_oZJA$>) or DelphiPka (;!!DZ3fjg!qiAD3f7mt7lc30MuT_0aXrx3MwRRPaAlonE0FSFEs9I85WynxYQ-mzOPINt9pKSfWA$ <;!!DZ3fjg!qLqvG5aUIsZadcvLdv2KKVS99Ta3HEX53KL7IdL8pM3CNQJ6bPm9CzlpqFCAQcr_zA$>) in principle should be able to handle non-protein ionizable groups.


From: <<>> on behalf of Francesco Pietra <<>>
Date: Friday, April 2, 2021 at 1:36 PM
To: VMD Mailing List <<>>, "<>" <<>>, "<>" <<>>
Subject: vmd-l: histidine protonation

I am faced with the problem of establishing the protonation status of many histidine residues present in a nucleic acid-protein ensemble, both inside and on the periphery. All that in the framework of CHARMM36 for NAMD MD.

Servers for direct assignment, like H++, are useless because of the presence of nucleic acids.

I came across two interesting old posts by Zhu and Henin to this concern and wonder whether there is anything additionally useful for the above ensemble
From: Fangqiang Zhu (<,HSE,HSP)>)
Date: Sat Nov 29 2003 - 00:22:03 CST
From: Jérôme Hénin (<,HSE,HSP)>)
Date: Sat Nov 29 2003 - 04:37:30 CST
Thanks for advice
francesco pietra