From: Yuri Pompeu
Date: 1 September 2011 03:52
I was playing around with PDBe PISA and came across the following:
For pdb entry 1OYA. The most promising interface has an area bury of around 720A^2 and DeltaG of -10.6Kcal/mol. sym_op(y,x,-z+1) and CSS of 0.039! Assembly analysis says it has no strong indications that point to stable quaternary structure.
This protein has been extensively studied and determined to be a dimer.
Entry 3RND is the same protein with one single conservative mutation deep in the active site.
They align with a RMSD of 0.3 A, 99.8% sequence identity. Superposition and inspection of the regions that contact
the adjacent monomer shows they are basically identical.
The interface here shows Area bury of 760 A^2 and DeltaG = -6.6Kcal/mol. sym_op (-y,-x,-z-1/2) CSS=0.00 !
Assembly analysis basically says no stable oligomers form. This enzyme also is dimer according to gel filtration.
Could anyone ellaborate on this please, if they feel like they have the time...
Cheers
----------
From: Karthik S
http://www.ebi.ac.uk/msd-srv/prot_int/pi_ilist_css.html
so it depends on how many 'stable assemblies' pisa can find i suppose.
more interfaces and especially if stable enough will make your
fraction go down. i would have been more surprised or worried if that
conservative mutation showed radically different CSS scores say one
close to zero and the other one or close to it. so the exclamation
marks here are really pointless (since both values are close to zero).
hence i would ignore the CSS in these two cases. CSS is a statistical
measure and does not imply biological meaning. in making me (us)
assume the latter through this one singular value leads to all
misconceptions.
--
Karthik
----------
From: Jan Dohnalek
Wasn't the original question directed to our (growing) feeling that many times PISA says No obvious oligomerization pattern but we already have evidence of dimer formation etc..
This should happen "occasionally" as the approach implied in the calculations is statistical in a sense. We should not be getting such contradictions on a regular basis.
Possible I misunderstood the original point ...
Jan
--
Jan Dohnalek, Ph.D
Institute of Macromolecular Chemistry
Academy of Sciences of the Czech Republic
Heyrovskeho nam. 2
16206 Praha 6
Czech Republic
----------
From: Ethan Merritt
I think there are at least two possibilities
1) the interface seen in the crystal is a real dimer interface,
but the PISA score fails to rate it as significant
2) the protein has crystallized as a monomer even though it
[sometimes] exists in solution as a dimer. The interface
seen in the crystal is not the "real" dimer interface and
thus the PISA score is correct.
I have no idea which, if either, of these might be the case for 1OYA.
Ethan
----------
From: Jan Dohnalek
I guess both of the mentioned possibilities occur and it is hard to judge which one it is for a particular case.
PISA is extremely useful for clear-cut cases to judge them quick. In the "borderline" ones it remains to be the task of the research teams to prove what sort of oligomerisation state is biologically relevant.
I wish we had a method that delivers a reliable answer regarding the "real state" of any protein studied...
Jan
----------
From: Yuri Pompeu
This is regarding Ethan´s point, particularly:
I see the same exact interface in a crystal of a close homologue that belongs to a different space group (hexagonal vs tetragonal system)
----------
From: Eleanor Dodson
Like Jan, I find it very useful to sort out the clear cut cases. Otherwise it is easy to get things wrong..
But isnt a buried surface area of 720 rather small for a stable interface? If there is other confirming evidence like 2 diff space groups then you feel more secure!!
----------
From: Eugene Krissinel
720 is not an impressive size for a stable interface, but it may do depending on molecule size and exact chemistry of the interface (h-bonds, salt bridges, disulphides, charges etc etc). Everything is subject to chemical environment and concentration, as usual. For these entries, PISA gives dissociation free energy of -1 kcal/mol. Given some +/- 5 kcal/mol estimated (guessed) accuracy of PISA, this may or may not be a stable thing. And yes, it has about 70-80% chances to be simply an artefact of crystal packing, according to some sort of derivations that I did in 2nd PISA paper in J.Comp.Chem. in January last year.
Having said all this, PISA is not an oracle and does not pretend to be correct in 100% of instances.
Eugene.
----------
From: Phil Evans
I get confused by these figures. As I understand it the "interface area" given in Pisa is half the loss of accessible area on forming the complex: is that right?
We're working on a complex with interface area ~500A^2, where the complex is stable enough for gel filtration, and with a measured Kd of ~2microM, Pisa estimate of DelG -2.3. Does that sound sensible?
Phil
----------
From: Jacob Keller
Well, I guess I have always been curious what is the gold standard
here--perhaps SEC, ITC, SPR, pulldowns? What if SEC shows a
polydisperse sample with weak oligomerization, or SPR a very weak
binding constant? Do we then revert to a functional assay? Or what if
the functional assay does not show anything, but the binding constant
is really strong? Or vice versa, the binding is completely
undetectable, but the functional assay shows something?
JPK
--
*******************************************
Jacob Pearson Keller
Northwestern University
Medical Scientist Training Program
*******************************************
----------
From: Bosch, Juergen
----------
From: Jacob Keller
mea culpa! How about FRET?
JPK
----------
From: Andreas Förster
AUC !
Andreas
--
Andreas Förster, Research Associate
Paul Freemont & Xiaodong Zhang Labs
Department of Biochemistry, Imperial College London
http://www.msf.bio.ic.ac.uk
----------
From: Jacob Keller
NMR...take that!
JPK
--
----------
From: Bosch, Juergen
----------
From: Huang, Jing
Jacob,
You may also try DARTS (drug affinity responsive target stability). In this method, small molecule (or protein) binding leads to specific protection of its target protein from protease digestion. It is completely label free, and appears to be able to detect very low affinity interactions (high microM for small molecules). We'd be happy to send detailed protocols if you are interested.
http://www.pnas.org/content/106/51/21984.full
Best,
Jing
--
Jing Huang, Ph.D.
Associate Professor
UCLA
Department of Molecular & Medical Pharmacology
jinghuang at mednet dot ucla dot edu
http://labs.pharmacology.ucla.edu/huanglab/
----------
From: Jacob Keller
I did a similar assay years ago, but since the results were negative,
never published anything--it was seeing whether nucleotides bound to
my protein of interest by time courses of proteolysis +/- nucleotide.
One tricky part of the assay, however, is to be sure that the compound
of interest doesn't inhibit the protease--did you address that? I
guess you would have to have some control proteins for that...
Jacob
--
----------
From: Huang, Jing
Excellent point indeed. We always include (at least one, preferably multiple) control proteins that are proteolysed equally across to make sure that the observed "target stabilization" is not due to fortuitous protease inhibition.
What protease did you use for your nucleotide binding case? The protease sometimes matters. For example, thermolysin mainly only digests proteins that are unfolded, whereas pronase, which is a mixture of various proteases, can digest both folded and unfolded proteins.
Best,
Jing
From: Jacob Keller
Date: 1 September 2011 03:52
I was playing around with PDBe PISA and came across the following:
For pdb entry 1OYA. The most promising interface has an area bury of around 720A^2 and DeltaG of -10.6Kcal/mol. sym_op(y,x,-z+1) and CSS of 0.039! Assembly analysis says it has no strong indications that point to stable quaternary structure.
This protein has been extensively studied and determined to be a dimer.
Entry 3RND is the same protein with one single conservative mutation deep in the active site.
They align with a RMSD of 0.3 A, 99.8% sequence identity. Superposition and inspection of the regions that contact
the adjacent monomer shows they are basically identical.
The interface here shows Area bury of 760 A^2 and DeltaG = -6.6Kcal/mol. sym_op (-y,-x,-z-1/2) CSS=0.00 !
Assembly analysis basically says no stable oligomers form. This enzyme also is dimer according to gel filtration.
Could anyone ellaborate on this please, if they feel like they have the time...
Cheers
----------
From: Karthik S
http://www.ebi.ac.uk/msd-srv/prot_int/pi_ilist_css.html
so it depends on how many 'stable assemblies' pisa can find i suppose.
more interfaces and especially if stable enough will make your
fraction go down. i would have been more surprised or worried if that
conservative mutation showed radically different CSS scores say one
close to zero and the other one or close to it. so the exclamation
marks here are really pointless (since both values are close to zero).
hence i would ignore the CSS in these two cases. CSS is a statistical
measure and does not imply biological meaning. in making me (us)
assume the latter through this one singular value leads to all
misconceptions.
--
Karthik
----------
From: Jan Dohnalek
Wasn't the original question directed to our (growing) feeling that many times PISA says No obvious oligomerization pattern but we already have evidence of dimer formation etc..
This should happen "occasionally" as the approach implied in the calculations is statistical in a sense. We should not be getting such contradictions on a regular basis.
Possible I misunderstood the original point ...
Jan
Jan Dohnalek, Ph.D
Institute of Macromolecular Chemistry
Academy of Sciences of the Czech Republic
Heyrovskeho nam. 2
16206 Praha 6
Czech Republic
----------
From: Ethan Merritt
I think there are at least two possibilities
1) the interface seen in the crystal is a real dimer interface,
but the PISA score fails to rate it as significant
2) the protein has crystallized as a monomer even though it
[sometimes] exists in solution as a dimer. The interface
seen in the crystal is not the "real" dimer interface and
thus the PISA score is correct.
I have no idea which, if either, of these might be the case for 1OYA.
Ethan
----------
From: Jan Dohnalek
I guess both of the mentioned possibilities occur and it is hard to judge which one it is for a particular case.
PISA is extremely useful for clear-cut cases to judge them quick. In the "borderline" ones it remains to be the task of the research teams to prove what sort of oligomerisation state is biologically relevant.
I wish we had a method that delivers a reliable answer regarding the "real state" of any protein studied...
Jan
----------
From: Yuri Pompeu
This is regarding Ethan´s point, particularly:
I see the same exact interface in a crystal of a close homologue that belongs to a different space group (hexagonal vs tetragonal system)
----------
From: Eleanor Dodson
Like Jan, I find it very useful to sort out the clear cut cases. Otherwise it is easy to get things wrong..
But isnt a buried surface area of 720 rather small for a stable interface? If there is other confirming evidence like 2 diff space groups then you feel more secure!!
----------
From: Eugene Krissinel
720 is not an impressive size for a stable interface, but it may do depending on molecule size and exact chemistry of the interface (h-bonds, salt bridges, disulphides, charges etc etc). Everything is subject to chemical environment and concentration, as usual. For these entries, PISA gives dissociation free energy of -1 kcal/mol. Given some +/- 5 kcal/mol estimated (guessed) accuracy of PISA, this may or may not be a stable thing. And yes, it has about 70-80% chances to be simply an artefact of crystal packing, according to some sort of derivations that I did in 2nd PISA paper in J.Comp.Chem. in January last year.
Having said all this, PISA is not an oracle and does not pretend to be correct in 100% of instances.
Eugene.
----------
From: Phil Evans
I get confused by these figures. As I understand it the "interface area" given in Pisa is half the loss of accessible area on forming the complex: is that right?
We're working on a complex with interface area ~500A^2, where the complex is stable enough for gel filtration, and with a measured Kd of ~2microM, Pisa estimate of DelG -2.3. Does that sound sensible?
Phil
----------
From: Jacob Keller
Well, I guess I have always been curious what is the gold standard
here--perhaps SEC, ITC, SPR, pulldowns? What if SEC shows a
polydisperse sample with weak oligomerization, or SPR a very weak
binding constant? Do we then revert to a functional assay? Or what if
the functional assay does not show anything, but the binding constant
is really strong? Or vice versa, the binding is completely
undetectable, but the functional assay shows something?
JPK
*******************************************
Jacob Pearson Keller
Northwestern University
Medical Scientist Training Program
*******************************************
----------
From: Bosch, Juergen
Hi Jacob,
you forgot cross-linking to stabilize a weak complex and verify that it exists.
Jürgen
......................
Jürgen Bosch
Johns Hopkins University
Bloomberg School of Public Health
Department of Biochemistry & Molecular Biology
Johns Hopkins Malaria Research Institute
615 North Wolfe Street, W8708
Baltimore, MD 21205
----------
From: Jacob Keller
mea culpa! How about FRET?
JPK
----------
From: Andreas Förster
AUC !
Andreas
Andreas Förster, Research Associate
Paul Freemont & Xiaodong Zhang Labs
Department of Biochemistry, Imperial College London
http://www.msf.bio.ic.ac.uk
----------
From: Jacob Keller
NMR...take that!
JPK
----------
From: Bosch, Juergen
Free flow electrophoresis would be another option, by the way anybody on the East Coast who has one of those instruments ? I'd be interested to get an email directly.
Thanks,
Jürgen
----------
From: Huang, Jing
Jacob,
You may also try DARTS (drug affinity responsive target stability). In this method, small molecule (or protein) binding leads to specific protection of its target protein from protease digestion. It is completely label free, and appears to be able to detect very low affinity interactions (high microM for small molecules). We'd be happy to send detailed protocols if you are interested.
http://www.pnas.org/content/106/51/21984.full
Best,
Jing
--
Jing Huang, Ph.D.
Associate Professor
UCLA
Department of Molecular & Medical Pharmacology
jinghuang at mednet dot ucla dot edu
http://labs.pharmacology.ucla.edu/huanglab/
----------
From: Jacob Keller
I did a similar assay years ago, but since the results were negative,
never published anything--it was seeing whether nucleotides bound to
my protein of interest by time courses of proteolysis +/- nucleotide.
One tricky part of the assay, however, is to be sure that the compound
of interest doesn't inhibit the protease--did you address that? I
guess you would have to have some control proteins for that...
Jacob
----------
From: Huang, Jing
Excellent point indeed. We always include (at least one, preferably multiple) control proteins that are proteolysed equally across to make sure that the observed "target stabilization" is not due to fortuitous protease inhibition.
What protease did you use for your nucleotide binding case? The protease sometimes matters. For example, thermolysin mainly only digests proteins that are unfolded, whereas pronase, which is a mixture of various proteases, can digest both folded and unfolded proteins.
Best,
Jing
From: Jacob Keller
----------
From: Jacob Keller
Good ol' trypsin--any reason why not?
JPK
----------
From: Huang, Jing
Ideally, the more nonspecific the protease, the better. (That would be a separate protein engineering project, or chemical catalyst project..) For now, we found that Pronase works well enough. Cheers, Jing
Sent: Monday, September 05, 2011 3:44 PM
No comments:
Post a Comment