[RASMB] downward sloping baselines or "i no longer have much hair

[Borries Demeler]

> I have checked the RASMB archives and there does not seem to be anything 
> on this particular subject.
> The problem is not the XL-I; the problem is an overly complex sample and 
> reference.
> 
> The sample contains a protein that absorbs at 290 nm as well as a ligand 
> that absorbs at 290 nm.
> The reference contains just the  ligand. Both are in matching buffers.
> 
> The first scan: The baseline absorbance starts below zero and becomes 
> more negative at higher radius.
> Scan 100: The sample has almost sedimented completely giving the 
> baseline absorbance what appears to be a "bow".
> Scan 200: The sample has sedimented completely and the absorbance is 
> well below zero. It is more negative at high radius than low radius.
> 
> The problem is clearly the sample. I have not run the "control" in which 
> the sample compartment contains no protein but does contain ligand; 
> reference compartment contains no ligand.
> 
> This problem must have come up in the past. Needless to say, all 
> analyses are feasible but they are not likely to be correct. I'm at 
> wits' end.
> 
> thanks for any help you can give. i am attaching an xlgraph of the three 
> scans.
> jack kornblatt

Hi Jack,

I would try to run the sample + ligand against water or nonabsorbing
buffer in the reference cell. Then you can see both ligand, protein and
ligand-protein complex sediment together and you can analyze the mixture.

It looks like your free ligand in the reference is sedimenting slowly and
is building up an equilibrium gradient which absorbs significantly towards
the bottom of the cell, and in the sample channel it has bound mostly
to the protein, and therefore doesn't build up a shallow equilibrium
gradient. When the reference is subtracted from the sample channel,
it absorbs enough to cause a negative absorbance. Since the ligand is 
mostly bound and sequestered in complex, it sediments much faster 
and the XLA is subtracting apples from oranges.

If you measure against water, you won't have this problem and should
get a flat baseline. There may still be some free ligand building up
an equilibrium gradient, but that can be easily modeled with finite element
fitting. Feel free to send me the data that is measured against water and
I'll give it a try. It's much easier to model if you just have water in
the reference.

-Borries