From:	balbes &$at$& osiris.rti.org (Lisa M. Balbes)
Date:	Fri, 4 Oct 91 09:59:45 EDT
Subject:	Summary: Protein-small molecule interactions

WOW !  I guess I hit a good topic.  I got quite a few responses, and
several people suggested that the a discussion on the BB would be more
appropriate.  So, I am posting a brief summary of the responses I have
had so far.  Most of the solutions are computationally intensive
routines, which is what I was afraid of. If anyone else has a simple,
elegant solution, there is a large community out there waiting for it.

Summary follows.  All typos mine, all responses paraphrased.  I will send the
full text of replies to anyone who requests it.  It's 520 lines - so far.
Many Thanks to everyone who answered  - Does anyone else have something to
add?


				Lisa
NOTE:  I posted both the original question and summary to both
	the comp chem mailing list (chemistry -x- at -x- ccl.net and the
	sybyl mailing list (sybyl -AatT- quant.chem.rpi.edu).
	Replies should probably be sent to both as well.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

***** From: Carlos Faerman 

Program DOCK, by I. D. Kuntz.
This tries to fit rigid bodies against the receptor and then
scores them according to a function that tells you how good this fit of
spheres is. Once that step is done you could use CHEMPROP wich turns on the
electrostaiatic potential and possibly hydrogen bonds and analyze these
complexes.  Although it may sound rudimentary it gives a good idea
of the orientations you should discard.

References: J. Med. Chem. (1988) 31,722-729. (algorithm)
Proc. Natl. Acad. Sci. USA (1990) 87,6644-6648. (application to HIV protease)

*****  jpj - at - lotus.medicine.rochester.edu (Jeffrey P. Jones)

Use free energy difference calculations using either
a slow growth or a perturbation program. This type of calculation
is accurate to between 1 and 0.2 kcals. This method eliminates much
of the problem associated with poor sampling. See Singh et. al. JACS
1987, 109, 1607(AMBER), Beveridge, Annu.  Rev. Biophys. Biophys. Chem.
1989, 18, 431 (Great review although it is now dated) , Mitchell J. Comp.
Chem. 1991, 12, 271.  This is a lot of work but it really is great!

***** From: Yvonne Martin  MARTIN (+ at +) cmda.abbott.com

Jeff Howe at Upjohn does what you are interested in by a standard potential
energy calculation and then adding in the solvation-desolvation term
(essential) by using Scheraga's terms.

Jeff has a manuscript in press in which the computer designs molecules to fit
a binding site. For this to be reliable, he has to be able to compare and rank
the various suggestions. The above method seems to work well. For example, they
successfully build a number of ligands for which they know experimentally how
they bind. The program also suggests the analogue found experimentally.

*****  From: "Leonore A. Findsen" 

The big question here is what exactly are you looking for. The method of QSAR
combines a number of different factors to generate a number or a volume and
then you look at a range of numbers. However, you determine the factors that
QSAR will analyze.  I use molecular dynamics to look at protein-drug
interactions, but I am looking at specific non-covalent bonds. Another
methodology is to use the relative free energy of binding of two similar drugs.
Or you can do a conformational analysis if the drug and see how the collective
volume fits. I don't believe that there is just one methodology for looking
at protein-drug interactions (or at least I hope there isn't or the field I
am in might colapse).

*****  From: martin "-at-" link.sunet.se (Martin Norin)

We have just (today actually) submitted a paper on how to
calculate the enantioselectivity of enzymes, e.g. the free energy
difference between the transition state of the reaction of enantiomers
of a small substrates (esters) catalyzed by an enzyme (chymotrypsin).
We have compared the difference in potential energy between the transition
state of each enantiomer (bound to the enzyme) with experimental data.
The results agrees with experimental data semiquantitatively.We have used
molecular mechanics and dynamics in this work. We also have
used specific methods to calculate point charges of substrate atoms.


Also has had good luck with GRID. This program looks for interesting
interaction areas between molecules, especially between a
small molecule and a protein. In short the program rolls a probe, which
may be a part of a drug, over the target molecule and measures the
interaction energy between the probe and the target at every grid point.
It a unique force field which probaly is the best in the area right now
(my personal opinion).

Original Reference:  J. Med. Chem (1985) _28_, 849-857.
Also Reynolds, C. A., Wade, R. C., Goodford, P. J. "Indentifiying
targets for bioredictive agents: using GRID to predict selective
bimding regions of proteins." J. Mol. Graphics. 7: 103-108, 1989.

GRID is a computational procedure for detecting energetically favourable
binding sites on molecules of known structure.  GRID can distinguish between
selective binding sites for different Probes.  The Target and Probe can be
studied in a condensed phase such as water, so that results from GRID are
particularly relevant to biological systems in which the "in vacuuo"
approximation can be misleading."

***** From: bio320 \\at// cvx12.inet.dkfz-heidelberg.de (Friedrich Rippmann)

A recent state-of-the-art paper on interaction energies of
ligands to proteins is e.g.  DM Ferguson, RJ Radmer, PA Kollman
Determination of Relative Free Binding Energies of Peptide Inhibitors
yo the HIV-Protease J. Med. Chem.  34, 1991, p. 2654-2659

I think it is clear that you have to apply some kind of a molecular
dynamics protocol to get something like an interaction energy. Probably
it is not realy necessary to calculate deltaG (or rather deltadeltaG)
as in the approach detailed in the above paper (there are many such
approaches now using perturbation theory/thermodynamic cycle to get
delta G). All these approaches are rather computationally expensive,
and there are usually many problems (charges seem to be very important
sometimes) before you get reasonable results.  To overcome the comp.
time problems I devoloped a relatively simple MD protocol.
Basically I cycle through
1) minimization
2) clculation of (non-bonded) interaction energy
3) dynamics
and again 1) and so on.
I then average the energies in 2) and get an interaction value with
a standard deviation (I usually perform ca. 20 cycles). A plot of
these energies (for a number of compounds) against experimentally
determined delta Gs is nicely linear, so from a progmatic point of view
it works (should read pragmatic, of course). I have finished a paper
on that, but not submitted it yet.
Certainly many people would be interested in an easy way to get these
energies, but nobody has got a simple solution yet.

***** From: EA_MOORE%VAX.ACS.OPEN.AC.UK : at : OHSTVMA.ACS.OHIO-STATE.EDU

Dr.W.G.Richards at Oxford university, U.K. is an expert on drug/receptor
interaction and may be able to help you. He has written a book
Quantum Pharmocology Butterworths,London (2nd Edn. 1983).
I don't know of any FTP able molecular mechanics but the Oxford Molecular
package can be obtained through a site licence not exceeding 600 pounds
sterling per annum. Apply to CHEST -x- at -x- BATH
           CHEST
           Bath University Computing Service
           Claverton Down
           BATH
            BA2 7AY
            England.

*****  From: Rick Loncharich 

There is a paper by Pettitt and Karplus  Top Mol. Pharmacol. V3, 1986, 75-113
entitled "interaction energies: their role in drug design".
Application of the technique is in Lau and Pettitt J. Med. Chem 1989, 32,
2542-2547. Be careful there are minor points to this procedure that
are not explicitly written down.



%%%%%%%%%%%%%%%%%%%%% balbes { *at * } osiris.rti.org %%%% standard disclaimer
%%%%%%%%%%
 Lisa M. Balbes, Ph.D.
Research Triangle Institute            Life would be boring without problems.
   P. O. Box 12194			     919-541-6563
Research Triangle Park, NC 27709-2194        919-541-6767 xt 6563 (Msgs 24 hr)



Similar Messages
06/28/1993:  interaction function
06/28/1995:  Re:POSTED RESPONSES: Quantitative assessment of novel ligands
08/01/1996:  Re: CCL:M:Heat of formation calculation using MOPAC.
10/01/1993:  torsion of conjugated systems -- summary
03/05/1996:  host-guest complexes
06/08/1993:  undergrad computational chem
10/10/1994:  Summary Interaction
11/27/1996:  RE: CCL:pKa calculation
08/01/1995:  Spin contamination, effect on energy and structure.
03/05/1996:  CONFORMERS


Raw Message Text