From: John Philo <jphilo@mailway.com>
  To  : Rasmb <rasmb@alpha.bbri.org>
  Date: Thu, 27 Apr 2000 17:29:16 -0700

RE: Incompetence..

Normally I try to make only one comment per issue, but since (1) Jim has
definitely misquoted me, (2) I do know something about these chemical
heterogeneity issues, and (3) I'm a qualified incompetent, I'll jump back in
again.

If it is true that AUC with glycosylated proteins is a 'waste of time' then
I guess I've wasted a lot of it! I think Jim was actually referring to a
discussion we had about whether one could use sedimentation velocity on
glycoproteins to try to detect heterogeneity such as amino-terminal
variations, small protein conformational variations, etc. In my view that
would indeed be a waste of time since the carbohydrate heterogeneity will
dominate over minor differences arising from the polypeptide structure.

Sedimentation equilibrium on glycoproteins is certainly useful to
characterize the total carbohydrate content, state of association, etc., but
normally one is measuring average quantities over the population of
glycoforms. Velocity on glycoproteins can also be useful to look at
aggregates, and MAY give some indication of the overall protein shape if the
protein is very asymmetric and not too glycosylated.

With regard to whether these different protein forms can have different
stabilities toward aggregation, the answer is a definite yes! In some cases
the unglycosylated or under-glycosylated forms are quite unstable and prone
to aggregate, and at least for EPO the glycosylated forms are even
significantly more stable towards thermal or chemical denaturation. There
are also cases known where clipping at the N- or C-terminus affects
aggregation, and this is also true for more subtle chemical alterations such
as deamidation or methionine oxidation.

When one extends this to ask to what extent this sort of heterogeneity
produces a range of association constants for reversible associations, I'm
not sure there is much hard data. We did in a couple of cases report
differences in receptor-hormone binding for both glycosylated and
non-glycosylated protein hormones, and for SCF the difference was more than
an order of magnitude. Given the large effect, and the large glycosylation
heterogeneity, in this case there will clearly be a wide range of binding
affinities.

Emory is certainly right that when there is evidence for chemical
heterogeneity there is likely a range of association constants and not just
fully 'competent' or 'incompetent'. I am less certain about the "very often"
aspect of his comment that "very often, slight differences in folding can
produce slightly different lnK values".

How often do proteins really have multiple conformations that are long-lived
and distinct enough to give a range of associations? Here I am not talking
about these chemical differences or loss of a ligand like Borries'
hemeprotein, but just a range of protein conformations. Obviously in the
modern view of protein folding every protein is a time-dependent ensemble of
conformations, but I don't think you'll ever see those fluctuations in
sedimentation equilibrium. If one has other evidence for conformational
heterogeneity then yes you definitely should at least consider that in
interpreting sedimentation equilibrium data. However if the only evidence
for multiple protein conformations was a better fit of sedimentation
equilibrium data, then I would want to find other data to support that, such
as sedimentation velocity or some other directly conformation-sensitive
technique.

John Philo

-----Original Message-----
From: Jim Bloom [Jim.Bloom.B@bayer.com]">mailto:Jim.Bloom.B@bayer.com]
Sent: Thursday, April 27, 2000 1:30 PM
To: - *rasmb@alpha.bbri.org
Subject: Incompetence..


 Many thanks to Borries, Tom and Yujia Xu for their lucid explanations
of incompetence and competence.  I have had an interesting thought for the
biochemically inclined.  I do much more ESI-MS analysis than AUC and
therefore
see lots of subtle differences in proteins.  We also work with glycosylated
proteins.  Examples are an interleukin that contained four forms:  a full
length non-glycosylated, des1-2 non-glycosylated (ie. biochemists
jargon....lacking the first two amino acids on the N-terminal), and two
glycosylated full length versions.  We also have another protein with a full
length non-glycosylated population and  glycosylated at one N-linked site
and
glycosylated at two sites.   So, with this background comes a topic of
interest
to me: are these forms of the same molecule structurally "the same".   In
the
experience of the RASMBer's, has anyone seen a difference...for example
maybe
one might see aggregation with the non-glycosylated forms but not the
glycosylated?
 One other comment...in a discussion with John Philo at one time he
mentioned to me that in a sense using AUC with glycosylated proteins is a
"waste of time" ie. too much heterogeneity.  Any comments on this issue?
 Jim