[RASMB] Re: RASMB digest, Vol 1 #319 - 5 msgs
John Philo
jphilo at mailway.com
Thu Jul 15 14:18:15 PDT 2004
Chi-Yuan,
It would have been quite helpful to know the weight-average sedimentation
coefficient for each of those distributions. The lower signal/noise for the
low concentration sample translates to less resolution of peaks (which may
not be real anyway), which complicates trying to compare by eye. But anyway
it looks to me like there is considerably more material in the ~7-9 S range
at the higher protein concentration, and thus that the weight-average is
higher as the concentration increases. If that is true then you must have an
associating system. I think the enormous change in best-fit f/f0 ratio
between these data sets also is a red flag that things are changing pretty
drastically with concentration. That might indicate that some
oligomers/aggregates that are quite hydrodynamically extended or partially
unfolded are formed at the higher concentration, but it also might be a hint
that the data are being strongly influenced by dynamic effects
(interconversion of species during the run).
To me the broad nature of the distribution and very wide range of
sedimentation coefficients that are present suggests this is a pretty
non-specific association rather than formation of some discrete, preferred
quaternary structure. While as I said by eye it looks to me like things are
changing somewhat with concentration, this is really a remarkably weak
concentration dependence given the ~7-fold change in concentration and given
that some of these species appear to be pretty high oligomers. Thus I would
also suspect that you have both reversible and irreversible oligomers
present. So overall I would tend to call this aggregation.
Since there does seem to be concentration dependence, I agree with Jack
Correia 100% that it would be inappropriate to try to integrate individual
peaks unless you can establish that those represent true molecular species.
Thus the original issue of what is the right degree of maximum-entropy
smoothing is basically irrelevant---probably the only valid result is the
weight-average value for each sample.
John Philo
Alliance Protein Laboratories
-----Original Message-----
From: rasmb-admin at server1.bbri.org [mailto:rasmb-admin at server1.bbri.org] On
Behalf Of medakachou
Sent: Wednesday, July 14, 2004 8:16 PM
To: rasmb at server1.bbri.org; John Correia
Subject: Re: [RASMB] Re: RASMB digest, Vol 1 #319 - 5 msgs
Dear RASMB,
I attached a pdf file (distribution.pdf) containing c(s) distribution plot
at high and low protein conecentration. The parameters are included, also
the rmsd, f/f0, residual bitmap. p = 0.68 and the resolution is 250 (smin to
smax is 0.1 to 25). Thanks everyone's comments and suggestion.
Chi-Yuan Chou
PhD student, the Institutes of Life sciences, National Defense Medical
Center, Taipei, Taiwan
e-mail: r6243023 at yahoo.com.tw
Lab homepage: http://www.enzkin.org/
----- Original Message -----
From: "John Correia" <jcorreia at biochem.umsmed.edu>
To: <ls890067 at ndmctsgh.edu.tw>; <rasmb at server1.bbri.org>
Sent: Thursday, July 15, 2004 6:45 AM
Subject: Re: [RASMB] Re: RASMB digest, Vol 1 #319 - 5 msgs
> Isolating the main peak, reconcentrating the sample, and generating
> the same distribution suggests reversibility, the classic test,
> although its not clear what a 10 peak pattern means? Discrete peaks
> suggest irreversible aggregates. Is there reducing agent in the
> buffer? Cys in the protein?
>
> What optical system and how good are the fits, rms values? c(s) can
> get more peaky with very low noise levels & apparently low p values -
> I personally only do .95. I always check c(s) distributions with
> g(s), although for a broad distribution g(s) may be hard to apply.
> Alternatively what do the Ls-g(s) distributions look like? Ultimately
> I plot c(s) and g(s) as a function of concentration to develop
> hypotheses about the data. The shape of the conc dependence of the
> boundaries is often informative, although I would never fit a c(s) or
> Ls-g(s) distribution shape to extract molecular information. Maybe a
> g(s) since if properly done its the derivative of the boundary.
>
> Then I go to direct boundary fitting, individually & globally - in my
> case Sedanal but sedfit/sedphat can work depending upon the model.
> Years ago (the 70's) there was a lively, but non email (didn't exist),
> discussion about fitting raw data vs smoothed data or extracted
> moments
> - many methods give similar answers, but if possible always fit the raw
> data directly.
>
> So if your system is a broad but reversible distribution what are the
> options? Indefinite? You are describing big shifts? You need to
> establish endpoints, the s value of the monomer and the endpoint of
> the association, if there is one? Integrate the entire distribution
> and plot weight average S vs concentration. Remember the behavior of
> a simple titration experiment, you need at least two orders of
> magnitude to go from 10% to 90% saturation. Plot the data vs log
> conc. Does it look like a binding curve and does it saturate? To go
> higher in c try the 0.3 mm path centerpieces, you can gain a factor of
> 5 in concentration. Try 230 nm or interference to go lower, or can
> you estiamte s1 from the other isoform's data?
>
> Without seeing the data or the actual distributions it is hard to
> know!
>
>
>
> -------------------------------------------------------------------
> 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/
> -------------------------------------------------------------------
>
>
>
> >>> "medakachou" <ls890067 at ndmctsgh.edu.tw> 07/13/04 11:38 PM >>>
> ----------------------------------------------------------------------
> ----
--------
> 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
> --------------------------------------------------------------------------
--------
>
> > Date: Tue, 13 Jul 2004 10:59:00 -0500
> > From: "John Correia" <jcorreia at biochem.umsmed.edu><~!B*+R^&>> To:
> <jphilo at mailway.com>, <r6243023 at ms48.hinet.net>,<~!B*+R^&>>
> <arthur.rowe at nottingham.ac.uk>, <rasmb at server1.bbri.org><~!B*+R^&>>
> Subject: Re: [RASMB] difference of p = 0.95, 0.68 and 0.55, the
> > confidencelevel in the sedfit c(s) distributi
> >
> > This is a MIME message. If you are reading this text, you may want
> > to consider changing to a mail reader or gateway that understands
> > how to properly handle MIME multipart messages.
> >
> > --=__Part0E2FB7D4.3__=
> > Content-Type: text/plain; charset=US-ASCII
> > Content-Transfer-Encoding: 7bit
> >
> > One problem I have with these discussions is they are method of
> analysis
> > focused and until Arthur's illustion to "know your system" not
> > focused on the molecules, the mechanism, the interactions. Were we
> > just
> talking
> > about one run and one concentration or a series of conectrations?
> > Why do you want to integrate? Are the "peak" positions constant or
> > do
> they
> > change with concentration? Is this in fact an impure system of
> > nointeracting species or aggregates, or are these aggregates of a
> single
> > component? Is there any reversible interaction going on?
> >
> > The goal is to describe your system in molecular and mechanistic
> > terms and then fit the data individually & globally to that model to
> > prove
> the
> > hypothesized mechanism. Statistics, assumptions, simulations are
> > all important. Now what is going on in your system?
>
> Dear John,
>
> Yes, I should give more information about my case. I expressed and
> purified a 34 kDa protein and it's N-terminal or C-terminal truncated
> fragments by E. coli expression system. The protein purity is > 99% by
> SDS-PAGE. This protein has two isoforms. I study them in three
> different
> concentration: 0.15, 0.50 and 1.00 mg/ml in PBS (pH7.3). By using
> sedimentation velocity and c(s) distribution analysis, I found one
> isoform's N-terminal truncated fragments showed a 10 "peaks" pattern
> whose s is from 3
> to 23 at 1.00 mg/ml. While at 0.15 mg/ml, only 5 peaks were found at s =
> 3
> to 12. It means it should be a single component aggregation and is
> concentration-dependent. The other isoform didnot show this
> characteristics.
> I want to give my paper some quantitative data about this difference, so
> I
> chose "Origin peak fitting module" and analyzed the pattern of gaussian
> peaks. The reviewer thought it is overinterpretation (about the
> fused-"peaks") and suggested me lowering the cinfidence level to p =
> 0.7.
> I've tried and found the resolution is better (every "peaks" is still
> existed). These two days I've tried Jack Lebowitz's comment and gained
> some
> quantitative data. I'm going to compare them and hope it can make my
> paper
> more quantitative sound.
> By the way, while I isolated the major "peak" species by using
> gel-filtration chromatography (S-300 column) and concentrate them (I
> need
> higher concentration), It just change back to the same "multi-peaks"
> situation (by sedimentation velocity). I think it's not a
> non-interacting
> but a associating system, right? Thanks your help.
>
> Chi-Yuan Chou
> PhD student, the Institutes of Life sciences, National Defense Medical
> Center, Taipei, Taiwan
> e-mail: r6243023 at yahoo.com.tw
>
>
>
> _______________________________________________
> RASMB mailing list
> RASMB at rasmb-email.bbri.org
> http://rasmb-email.bbri.org/mailman/listinfo/rasmb
>
More information about the RASMB
mailing list