[RASMB] Questions on NONLIN

Tue Mar 11 18:56:00 PST 2003

The "rules" for using nonlin and designing experiments are system
dependent.  If you form a 12-mer then at sigma = 1 for the monomer it
equals 12 for the polymer - plot the concentration of 12-mer vs radial
position and you'll find the majority of it near/in the pellet.  This is
why you cannot nail down the size of the largest species, the program
has very little data to work with.  In general more curvature is
helpful, but for systems like this sed equil is a challenge.  My guess
is the slowest speed (sigma = 0.5 or less) will be best.

Try nailing down the size and fit with a fixed Nmer and a fixed monomer
sigma.  Have you tried sed vel as a function of concentration.  No
problem seeing monomer & oligomer there.  You could fit by weight
average to get K's and by extrapolation (if its weak) of the peak
position (1/s vs 1/c) get S for the N-mer.  This will help to limit the
N-mer size, especially if you have a pdb of your protein and could build
a polymer of various dimensions.   Negative stain EM or cryo-EM could
also help here.

In the "old" days a graphical analysis with a program like BIOSPIN and
then a two-species plot (search Roark & Yphantis) could help determine
the oligomer size, although the uncertainty will probably still be
large.  Only Walt Safford is old enough to remember & still do this.  I
suspect you need to have significant data at saturation for this to
work.  But we are trying to revive Biospin in a windows setting.

As a rule I never automatically try multiple speeds until I can tell I
need them, and even then I worry about stability.  Nine concentrations
(3 six channel centerpieces in an XLA) at a single speed will typically
do the job.  In your case any higher speeds doesn't help.

I prefer a global fit to determine the offsets, especially if its a
good fit.  Comparisons with an overspeed can help.

A Molar K of 5E-15 M for 12-mer is not tight - imaging each interface
contributes equally and parse the product of K's over that value - its
only -19.5 kcal for the complex and only -1.6 kcal/mol on a monomer
scale.  Adding  cooperativity makes it close, but its still not tight.

Try velocity to also exclude aggregation - any heterogeneity and your
equil system is intractable.

-------------------------------------------------------------------
 Dr. John J. "Jack" Correia
 Department of Biochemistry
 University of Mississippi Medical Center
 2500 North State Street
 Jackson, MS  39216
 (601) 984-1522                                 
 fax (601) 984-1501                             
 email address: jcorreia at biochem.umsmed.edu     
 homepage location: http://biochemistry.umc.edu/correia.html  
 dept homepage location:    http://biochemistry.umc.edu
-------------------------------------------------------------------

>>> <lelj at interchange.ubc.ca> 03/11/03 03:52PM >>>
----------------------------------------------------------------------------------
The older archived RASMB emails can be found at:
http://rasmb-email.bbri.org/rasmb_archives 
and current archives at
http://rasmb-email.bbri.org/pipermail/rasmb/ 
Search All the Archives at:
http://rasmb-email.bbri.org/rasmb_search.html 
----------------------------------------------------------------------------------

Dear RASMB 'ers

First of all I would like to thanks all the people that replied to my
email about the A to B association system. But yet I have many more
questions!!! I realized that I need to better characterize my
self-associating system before I can go on doing experiments of the
complex.
Here are a list of questions that I would like to ask you (I'm using
NONLIN), some are more general about experimental conditions other are
more specific. 
I would like to thank you all in advance for the patient that you'll
have in reading and hopefully answering:

1. Is it true that for value of sigma lower than 1 it's harder to fit
your data (I do have WAY more problems when sigma=0.7-0.01 than
sigma>1)?
If this is the case should I just run experiments at higher speed (20K,
30K) and forget about the lower ones (8K-12K)?

2.Is it better to run experiments in the 6hole cuvette or in the 2
holes? Why should I choose one over the other?

3.If at the end of my experiment I over speed (56K) to measure the
absorbance at the meniscus is it wise to fix the values of delta=abs
meniscus and don't let the value to float at all? Is this always the
case?

4. I generally design an experiment as follow:
3 concentrations (in the 6 hole cuvette) 3 different speeds and over
speed at the end. Then I try to fit these 9 data set. Can I try to fit
together data set that comes from different experiments? (i.e. runs in
different days with different concentrations? (With the same
buffer!!!))?

5.I find that at my third speed and sometimes at my second speed, some
of the protein is pelletted at the bottom of the cell..... Can I live
with it and still try a global fit?

6.I work with a self-associating system, which is NOT monomer-dimer or
tetramer, but I believe is monomer-12mer. 
I managed to fit 3 data sets at 20K and low concentration (measuring
abs= 230nm).
First of all:  I can only globally fit these 3 data sets and data sets
at lower speed don’t fit with each other and single data fit give
me values different from the one at 20K. 
Moreover, it looks like the binding is very tight lnK2=9 (with K2=12
and sigma= monomer).  I calculate Kdissociation in Molarity to be=5E-15
M using the following equation: Ka(M-1)=[K2(Abs-11)x(e monomer x
1.2cm)E11]/12 and considering Kd=1/Ka.  The square root of variance is
7E-3 and I have nicely dispersed deviation. 
Does it make any sense having such tight interaction and be able to
detect it with AUC? 
I also tried to fit using sigma=12-mer but I have no fit and also
sigma=dimer or tetramer or 6-mer in equilibrium with other species but I
have no fit.

7. Sometimes if I set sigma=monomer and allow N2 to float I have values
that are not integer number or are bigger than what I expect (N2=15.5).
If then I fix the value of N2 to a little smaller system (let’s
say 12 instead of 15.5) I have the same fit with the same deviation and
square root. (12 is the smaller number I can use to have a good fit!)
Should I always try to fit for N2 as small as possible?
I also tried to fit monomer- # -12mer where #=dimer/tetramer examer/
free to float etc. and I have no results. The free float give me #=0.2

8. With such a big associating system, is it enough if I can manage to
global fit 9 data set or do I need more? (so far I haven't been able to
fit more than 3 conc. at one speed.....)
Can I stay away from high concentrations?

Thank you all for your immense patience... 
Best regards,
Barbara

--
Barbara Lelj Garolla Di Bard
Dr. Mauk's Lab
Dept. of Biochemistry and Molecular Biology
University of British Columbia
2146 Health Sciences Mall
Vancouver, B.C. V6T 1Z3
Phone: (604) 822-2526

_______________________________________________
RASMB mailing list
RASMB at rasmb-email.bbri.org 
http://rasmb-email.bbri.org/mailman/listinfo/rasmb