Thanks a lot Roger Rowlett for your very detailed response. I got the
SG-Zn refinement to work exactly as you detailed.
The key was to rebuild the Cysteines in coot keeping in mind the 2.3 Å
coordination distance and not get put off by coot SS bonding things
while I tried to get the Cys-SG's in place . So after some fixing in
coot -all Zns were 2.0 to 2.5 Å away from a Cys-SG going in to refmac
and then your approach worked perfectly.
So I had to:
Rebuild and put the Cys-SG within 2.3 A of the Zinc.
Remove the SSBOND definitions from the pdb file in an editor
Run the mock refinement with refmac to have it detect the SG-ZN coordination
Merge the cif file with my ligand cif file.
Run the restrained refinement in coot
I had just one question. I am imagining that the Cys SG-Zn restraint
can be added to the standard refmac monomer dict or I should be able
to simply reuse this dummy run cif file everytime I refine a structure
with a SG-Zn coordination. Cif files are reside number agnostic is
that true.
Also since the refmac dict has the ZN defined , how come coot does not
have the cif defined?
I am assuming I can similarly make coot coordination and Zn ion aware
Regardless, thanks again for your detailed response I have the
refinement working flawlessly.
Hari
On Thu, May 13, 2010 at 8:53 PM, Roger Rowlett <[log in to unmask]> wrote:
> Hari,
>
> Before starting your dummy refmac job to write out the appropriate cif file,
> you will need to use Coot adjust the positions of the Cys and other
> Zn-ligand side chains to appropriate positions (consisatent with your
> electron density maps, of course) to coordinate to Zn. Typical Zn-S bonds
> are 2.3 A, Zn-O or Zn-N bonds are about 2.0 A. You have to do this in order
> to break the apparent S-S disulfide bond. You don't have to get the bond
> distances perfect for refmac to figure it out. You can't do a real-space
> refinement of the metal ligands in coot because it does not know about the
> metal ion you have separately added, and will instead try to insert the
> ligands into the electron density contributed by the zinc. Do the ligand
> readjustment manually to put everything in to a reasonable place for refmac
> to start from. Save this as your pdb file with your zinc ion merged in to
> start the refmac job. Before running refmac, ensure there are no disulfide
> or other undesired linkage records in your pdb file. Remove them in your
> favorite text editor as required. I can't remember at the moment if S-S
> bonds are listed as a LINKR, LINK or as some other type record, but it
> should be obvious in the PDB file. Just remove the offending S-S bond
> record. Then start your dummy refmac job, which will fail and generate a cif
> file with all the metal ligand linkage information. If you have positioned
> ligands properly, you should see in this cif file definitions for all
> Zn-ligand linkages and nothing else. Remove any blocks in the cif file that
> are not linkages you would like to incorporate into your refinement
> restraints, if necessary. Now run your refmac refinement, using the cif file
> as your input library with your metal ligand restraints. If the metal ligand
> distances are reasonable, refmac should pick them up and refine them with
> proper restraints. Refmac will write the necessary LINKR records into your
> PDB file for all future refinements of this file.
>
> We do this routinely for Zn-metalloenzymes and it works just fine with
> 2.5-2.9A data. Coot doesn't make a real link between sulfurs: it is just
> adding a bond for any atoms that are within reasonable bonding distance of
> each other. If you rotate the sulfur atoms away from each other, Coot will
> "break" the bond for you. On the other hand, if you leave a linker record in
> the pdb file for a S-S bond, then refmac will restrain this bond throughout
> refinement no matter what else is present.
>
> Good luck. You should be able to get to to work out properly.
>
> Cheers, Roger Rowlett
>
>
> On 5/13/2010 7:31 PM, hari jayaram wrote:
>
> Hello Eleanor and Roger Rowlett and everyone,
>
> I am still having a tough time with coordinated metal refinements in
> refmac ( and coot )
>
> I followed Rogers suggestion of creating a pseudo refmac run and then
> using the output cif definition for Cys -SG Zn as an input for further
> refinement. Even If I did input this cif , I see that my
> tetra-coordinated Zinc is pushed aside and my Cysteine SG - are
> disulphide
> bonded to each other creating a tetrahedral arrangement.
>
> One caveat . The Resolution of the data is no all that spectacular
> (2.5) so the whole assembly appears rather blobby with three clear Zn
> spheres like I would expect in the DELFWT maps before I model in the
> Zn.
>
> The other un-related problem is that coot seems to also want to bond
> Cysteines to each other when I do the real space refinement, but I
> think thats because SG-Zn restraints are not part of the coot
> dictionary .So the problem may very well be in the lack of a record in
> the pdb output by coot that says ..dont disulfide bond these Cys. I
> did check there were no explicit disulphide links defined or LINK or
> LINKR.
>
>
> So I am still nor refining these Cys-Zn links and have them mangled
> in my pdb, so any pointers will be greatly appreciated.
> Thanks for your help in advance
>
> Hari
>
>
>
> On Thu, May 13, 2010 at 12:50 PM, Eleanor Dodson <[log in to unmask]>
> wrote:
>
>
> Has anyone answered this?
> Eleanor
>
> hari jayaram wrote:
>
>
> Hi
>
> Am trying to build coordinated Zn+2 by CYS -SG atoms and have it
> refine properly in refmac5.5
>
> I added ZN ions in Coot at the Fo-Fc peaks. Then I defined manually
> the coordination relations in a text editor using LINK records.
> the ions were placed in the same chain that was doing the coordination.
>
> LINK ZN ZN A1001 SG CYS A 487 1555 1555
> 2.34
>
> I found that some of these ions refined well using refmac 5.5 and the
> resulting pdb had LINKR records in place indicating the coordinating
> interactions.
>
> For some of the Zn atoms however , the ions did refine , but coot
> still was covalently disulphide bonding the Cys to each other rather
> than coordinating the ZN and there were no LINKR records in the refmac
> refined pdb file. I know these interactions are legitimate because it
> is a known structure.Should I manually create LINKR records instead of
> LINK records after modeling in the ions?
>
> What is the correct procedure to model coordinating ions and have
> refmac treat them appropriately.
> Thanks in advance.
>
> Hari
>
>
>
|