Date: Wed, 30 May 2001 14:23:30 +0200
From: Christian Pilger
Subject: summary: titratable amino acids and counter ions

Hi all,

earlier this week I asked some questions about the preparation of protein
structures prior to MD simulations with respect to the treatment of
titrable residues and the placement of counter ions.
I would like to thank Michal Otyepka, Malcolm Gillies, Deepak Singh, Rick
Venable, Mike Ford, and Kenward Vaughan for their kind responses.

As I received some requests for a summary, I send it to the lists.

Cheers,

Christian

== response #1 ==

Date: Mon, 28 May 2001 13:52:43 +0200 (DFT)
From: Michal Otyepka
To: Christian Pilger
Subject: Re: titratable amino acids and counter ions

Dear Mr. Pilger:

> while preparing a protein structure as input for MD simulations, I
> encountered the following points, which I would like to ask your advice:
>
> - How do you usually deal with titratable residues (GLU, ASP, LYS, ARG) ?
> Are all of them simply taken in their ionized forms ? Would it be good
> to try to rationalize the ionization states of these residues by
> manually inspecting their environment ? Which methods/rules of thumb are
> available for the estimation of these ionization states ?

program GRID written by Peter Goodford (Molecular Discovery Limited) can
help you

> - In case the net charge is neutralized by counter ions: how and where
> should these be positioned ? Should one counter ion be placed directly
> at each titratable residue (i.e. number of counter ions = number of
> ioniziable residues) or should counter ions be placed (arbitrarily ?)
> around the protein structure (i.e. number counter ions = net charge of
> the whole system) ?

You can use GRID or Grasp. The counter ions are placed to the
their positions to compensate electric field of a protein. Whole systen
has to be neutralised by counter ions.

The surface titratable residues behave obviously as free aminoacids (it is
not rule!). I think that some people around Mr. McCammon and Mr. Cieplak
develop methods to compute pKa of titrable aminoacids but I am not sure.

With best regards,
Michal Otyepka

== response #2 ==

Date: Mon, 28 May 2001 14:03:38 +0200
From: Malcolm Gillies
To: Christian Pilger
Subject: Re: CCL:titratable amino acids and counter ions

Dear Christian,

> - How do you usually deal with titratable residues (GLU, ASP, LYS, ARG) ?
> Are all of them simply taken in their ionized forms ? Would it be good
> to try to rationalize the ionization states of these residues by
> manually inspecting their environment ? Which methods/rules of thumb are
> available for the estimation of these ionization states ?

I prepared a couple of slides on this topic for a talk a few years
ago. You might like to look at them:

http://wwwcmc.pharm.uu.nl/gillies/aio/

The bottom line is that there are no perfect solutions, but that
a bit of common sense, and in some cases the application of continuum
electrostatics calculations, can result in a much more reasonable
model.

cheers,

Malcolm
--
== response #3 ==

Date: Mon, 28 May 2001 08:52:41 -0700 (PDT)
From: Deepak
To: Christian Pilger
Subject: Re: CCL:titratable amino acids and counter ions

Dear Dr. Pilger

Those are very valid questions and I suspect different
people will have different answers. The best solution
is some experimental evidence pointing to the
protonation states of such residues. In the absence
of experimental evidence, I personally like to examine
the local environment of each residue. Counterions
are best used to neutralize a on-neutral protein.
Althouth the isoelectric point gives you a decent
handle on gross protein properties, it is the local
pKa that governs the protonation state of titratable
groups. In doubt, I have actually performed
calculations using both both protonation states and
examining the differences and trying to reconcile them
with any known observations. There are of cousre,
methods to try and calculate the protonation states
but I do not trust them yet.

Hope this helps

Regards

Deepak

== response #4 ==

Date: Mon, 28 May 2001 13:45:12 -0400
From: Rick Venable
To: Christian Pilger
Subject: Re: CCL:titratable amino acids and counter ions

On Mon, 28 May 2001, Christian Pilger wrote:
> while preparing a protein structure as input for MD simulations, I
> encountered the following points, which I would like to ask your advice:
>
> - How do you usually deal with titratable residues (GLU, ASP, LYS, ARG) ?
> Are all of them simply taken in their ionized forms ? Would it be good
> to try to rationalize the ionization states of these residues by
> manually inspecting their environment ? Which methods/rules of thumb are
> available for the estimation of these ionization states ?

It might be worth considering (or calculating) burial, e.g. accessible
surface area, for GLU, ASP, LYS, ARG; the rule of thumb for these 4 is
that they are typically charged. HIS residues can be a bigger problem,
esp. for enzymes, because of the tautomerism in addition to the charge
state. People have used computational methods to try to estimate the
protonation state, esp. Poisson-Boltzman methods. I know there are some
freely available progs for this, but I don't have refs handy.

> - In my case the protein has an experimentally determined isoelectric
> point of roughly pH=4.7. Does this mean, that the protein will have a
> negative net charge at pH=7 ? Then, is there a method to correlate the
> isoelectric point with this net charge ? And, are MD calculations
> reasonable, when the net charge of the system differs from zero ?

Yes, the net charge will be negative. There can be problems with a
non-zero net charge for particle-mesh Ewald methods.

> - In case the net charge is neutralized by counter ions: how and where
> should these be positioned ? Should one counter ion be placed directly
> at each titratable residue (i.e. number of counter ions = number of
> ioniziable residues) or should counter ions be placed (arbitrarily ?)
> around the protein structure (i.e. number counter ions = net charge of
> the whole system) ?

I don't think there's complete agreement about the best wat to get a net
zero charged system, or how to place the ions. I use a method for
randomly replacing water molecules more than 5 or 6 A away from any
non-water molecule. Two common neutralizing approaches are: (1) just
enough counterions for net zero charge, or (2) e.g. 0.1 M salt added to
(1) to better model ionic strength.


== response #5 ==

Date: Mon, 28 May 2001 13:57:53 -0400
From: Michael Ford
To: Christian Pilger
Subject: Placing counter ions

We have found that it may be best to simply place the ions near their
corresponding charged group then run 500-100ps and take the avg coordinates
for the ion.

The CION program, part of AMBER, is really designed for DNA and my give you a
"salty" system with a bunch of Na and Cl ions.

Mike Ford

== response #6 ==

Date: Mon, 28 May 2001 14:28:19 -0700
From: Kenward Vaughan
To: Christian Pilger
Subject: Re: titratable amino acids and counter ions

On Mon, May 28, 2001 at 01:49:36PM +0200, Christian Pilger wrote:
> Dear protein modelers,
>
> while preparing a protein structure as input for MD simulations, I
> encountered the following points, which I would like to ask your advice:
>
> - How do you usually deal with titratable residues (GLU, ASP, LYS, ARG) ?
> Are all of them simply taken in their ionized forms ? Would it be good
> to try to rationalize the ionization states of these residues by
> manually inspecting their environment ? Which methods/rules of thumb are
> available for the estimation of these ionization states ?

I have no experience with this in the field of MD,etc, but from a chemical
perspective the side chains of residues (not the amino acids themselves)
will behave like any other carboxylic acid or amine. With the vast bulk of
those having pKa's and pKb's in the range of 4-5 (the amines being related
to pOH), it is totally reasonable to assume they are >99% ionized at neutral
pH (pOH).

If surrounded by others of similar behavior (bases-bases and acids-acids)
the expected effect of multiple ionizations becoming less and less likely
makes sense, but as the strength of electrostatic attractions follow the
inverse-square law I cannot guess how important this effect is. Unless
there are several within say.. 5 or 6 angstroms, it may not make a lot of
difference on the % ionization. But this is speculation on my part.


> - In my case the protein has an experimentally determined isoelectric
> point of roughly pH=4.7. Does this mean, that the protein will have a
> negative net charge at pH=7 ? Then, is there a method to correlate the
> isoelectric point with this net charge ? And, are MD calculations
> reasonable, when the net charge of the system differs from zero ?

Yes to the negative at pH 7 question. Don't know about the 2nd one. And an
"I would certainly hope so!" to the 3rd. :-)


> - In case the net charge is neutralized by counter ions: how and where
> should these be positioned ? Should one counter ion be placed directly
> at each titratable residue (i.e. number of counter ions = number of
> ioniziable residues) or should counter ions be placed (arbitrarily ?)
> around the protein structure (i.e. number counter ions = net charge of
> the whole system) ?

Beyond my experience, sorry. I would guess place them arbitrarily and let
them settle into place.


> Any comments are highly appreciated.
...

Hope this helps!


Kenward Vaughan