<HTML>
<HEAD>
<TITLE>Re: [RASMB] fringe deviation</TITLE>
</HEAD>
<BODY>
<BLOCKQUOTE>Dear Barbara<BR>
<BR>
Both sedimentation velocity and sedimentation equilibrium can work well when it comes to determining Ka/Kd values, but there is a real problem with currently used software when it comes to estimating weak interactions - and I take it from the concentration range in c that you are using that it is a truly weak interaction that you are looking at. The contributions from the thermodynamic non-ideality (BM) term can be of the same order of magnitude to those arising from Ka, and since (in the approximated forms of the equation often used, as in WinNONLIN, they are additive, plainly one can never unscramble these two effects. <BR>
<BR>
As regards SE vs SV, I use both. I have to disagree with John Philo when he asserts that "non-ideality effects are much stronger in velocity than in equilibrium". This is incorrect. For ideal spheres, the limiting concentration dependence coefficient for velocity is 4.0 ml/g (the kinetic ks - see my chapter 21 in "Analytical Ultracentrifugation in Biochemistry and Polymer Science" 1992) whilst the corresponding coefficient for (1/M) in equilibrium is 8.0 ml/g (value for 2BM - see any good physical chemistry text). The shape-dependence of these coefficients is not at all identical, so for a non-globular protein it is probably true that one can find 2BM to be bigger than ks. But - either way - you have to unscramble Ka away from non-ideality. And as Allen Minton has correctly said, you will not find WinNONLIN of any use here.<BR>
<BR>
So - how to do it? Let us look at the two possibilities:<BR>
<BR>
<B>(1) SV<BR>
</B><BR>
There are problems with trying to unscramble c-effects out of single runs. Among other things, the effects of pressure on the solvent and on the system are <U>not</U> negligible (see [1]). But this does not matter. Within a single rotor you can do 3 experiments over whatever range of c you can get. Strictly control all the parameters (e.g. scans used) to be equivalent between your 3 channels. Find the s(w) for each - various methods, but Peter Schuck's c(s) gives you the s(w) by simple integration . Providing you know the s value for the monomer (no problem with a weakly interacting system - do another run on one or more low c samples) you can fit s(w) vs c with stability using Ka as a floated parameter and get an excellent estimate for Ka - since the ks factor is readily incorporated into the fitted equation. I am happy supply the equations for this approach - they can be used in any decent fitting program.<BR>
<BR>
This method works <U>at even 2+ logs away from Kd</U> - so no need to use impossibly high c values. The precision with which s values can be measured relative to each other in the same rotor is amazing (<0.1%) so that is why 3 channels are plenty. I like this method - it is very 'visual', you can see whether your fit makes sense or not. And sorry about another contradiction (John, Borries) - but you do <U>not</U> need a vast range in c. I have a litter of data around in which I have successfully used e.g. 0.5, 1.0 and 2.0 mg/ml (for systems with a significant c-dependence of s).<BR>
<BR>
<B>(2) SE<BR>
</B><BR>
A big advantage of SE is that just one channel with a minimal volume of solution suffices - at least provided you have sorted out that you system <U>is</U> an equilibrium, not a mixture. I am currently working on souped-up software which segregates Ka away from 2BM. I will be releasing same very soon. But what is publicly available right now is of limited (or even no) use for this purpose. Perhaps Allen can fill us in on the 'correct' treatments which he says are available? <BR>
<BR>
To sum up: in your case, I would use SV. The suggestion from John Philo that you use absorption optics off-peak is a good one - but bear in mind that the numerical aperture of the XL-I/A optics is very limited, so in early stages of a higher-c run you are liable to get 'schlieren' effects, in which you see a peak where you expect to see just an integral boundary! Using 3 mm centrepieces limits this problem. And you can always discard a few early scans, <U>just so long as you stick to the same regime for all channels</U>.<BR>
<BR>
All best with your researches. Weak interactions may be a pain - methodologically - but from the point of view of biology, signalling and all that, they are important.<BR>
<BR>
Kind regards<BR>
<BR>
Arthur Rowe<BR>
<BR>
[1] N Errington, P. Mistry & A J Rowe (2002) "Protein hydration varies with protein crowding and with applied pressure: a sedimentation velocity study" Progr Colloid Polym Sci <B>119</B> 58-63<BR>
<BR>
<BR>
</BLOCKQUOTE>--<BR>
*******************************************************<BR>
Arthur J Rowe<BR>
Professor of Biomolecular Technology<BR>
NCMH Business Centre<BR>
University of Nottingham<BR>
School of Biosciences<BR>
Sutton Bonington<BR>
Leicestershire LE12 5RD UK<BR>
<BR>
Tel: +44 (0)115 951 6156<BR>
+44 (0)116 271 4502<BR>
Fax: +44 (0)115 951 6157<BR>
email: arthur.rowe@nottingham.ac.uk<BR>
arthur.rowe@connectfree.co.uk (home)<BR>
Web: www.nottingham.ac.uk/ncmh/business<BR>
*******************************************************<BR>
<BR>
<TT><BR>
</TT><BLOCKQUOTE><TT><BR>
Hello!<BR>
<BR>
I am trying to determine the association constant for a monomer-dimer equilibrium. I know the association occurs because I've seen it by velocity experiments.<BR>
<BR>
I run equilibrium at 3 concentrations and 3 speeds, my conc. are high and therefore I used interference optics. I have a fringe displacement at equilibrium of up to 50 fringes. I fit my data with WinNonlin and at the best I can get a SQUARE ROOT OF VARIANCE=9.3317E-02. Since I have such a big fringe displacement when I plot deviation vs. indipendent variable, in the worst case I have a deviation of -0.25 to 0.375 fringes, is it reasonable?<BR>
<BR>
thanks <BR>
barbara<BR>
<BR>
---<BR>
Barbara Lelj Garolla Di Bard<BR>
Dr. Mauk's Lab<BR>
Dept. of Biochemistry and Molecular Biology<BR>
University of British Columbia<BR>
2146 Health Sciences Mall<BR>
Vancouver, B.C. V6T 1Z3<BR>
Phone: (604) 822-2526<BR>
<BR>
<BR>
_______________________________________________<BR>
RASMB mailing list<BR>
RASMB@rasmb-email.bbri.org<BR>
http://rasmb-email.bbri.org/mailman/listinfo/rasmb<BR>
</TT></BLOCKQUOTE><TT><BR>
</TT>
</BODY>
</HTML>