Tutorial 1. Handling protein glycosylation using Coot and
In Refmac, recognition of sugar residues and covalent linkages
that they make to protein (e.g. ASN) and to each other is done
automatically. However some manual work is involved at the stage
of model building. This is not always a trivial task and the
simple example selected for t is tutorial does not reflect all
possible complications. The example structure has only one protein
molecule in the AU, the protein is small and has only one
glycosylation site, and all the sugar residues are present in a
single conformation. High resolution data provide good elec ron
density maps and make it easy to build the polysaccharide from
single sugar residues.
(1) The directory "jligand/1_glycan" contains the tutorial
data (files 4gos.mtz, model.pdb). Set up ccp4i project "1_glycan"
with this directory as the project directory.
(2) Run Refmac5 with model.pdb and 4gos.mtz to generate map
coefficients. Use default output file names.
Double click on the Refmac job in the ccp4i job list to open the
job result page. Check refinement statistics. Open output files in
Coot (Co t button in the section Output files)
(3) With output model and density from step (2) opened in Coot,
add NAG residue into the difference map near ASN A112. Save the
new model as model_refmac1_coot1.pdb in the project directory of
the project "1_glycan".
(4) Refine model_refmac1_coot1.pdb against 4gos.mtz. Use default
output file names.
(5) The complete polysaccharide in this example contains three
types of sugar residues: NAG, BMA, MAN. Reiterate steps (3,4) to
build the whole polysaccharide inferring the type of the next
added sugar residue from the features of the electron density
If Refmac recognises the new link, it automatically adds a
corresponding LINKR record to the output PDB-file.
The following line in the Refmac log-file means that geometrical
restraints associated with the covalent link are applied:
WARNING: New link was found
No chiral volume outliers in the Refmac log-file mean that the
type of link that has been applied is correct
No strong "red" or "green" density on or near the added residue
mean that the type of residue was determined correctly.
Step (3), the incorporation of the first sugar monomer, is
detailed in the next section.
Step (3) includes manual fitting of the first sugar residue into
Coot has an option of semi-automatic ligand fitting (Coot >
Calculate > Other Modelling Tools > Find Ligands) and you
can try it as well. Note, however, that the automatic procedure
does not always work, especially when the electron density
corresponds to a chain of several sugar monomers and you are
looking for only one of them. Therefore we describe manual
procedure which is guarantied to work (in our example) and which
is worthy exercising in any case.
To select the default output pdb-filename in CCP4i, place the
cursor on the input pdb-filename and press Tab. Similarly select
the default output mtz-filename.
Good bookkeeping is quite important for multi-step procedures
such as modelling polysaccharides. For this reason we pay
attention to file names in this tutorial. For the same reason we
do not run refinement directly from coot, but rather perform it
via ccp4i interface, where bookkeeping is a bit better organised
and previous work is easily traceable next time you open the
interface. However, you may try and complete the whole exercise in
a single session of Coot using Coot's own interface for Refmac.
In the example used in this tutorial, the whole polysaccharide
can be built in one go. So if you prefer, you can try and add all
missing sugar monomers during a single coot session and then
refine the complete structure at step (5). However, in more
complicated cases the step-by-step procedure is advantageous as it
helps avoid rebuilding the whole polysaccharide if something has
gone wrong with one of the monomers (incorrect monomer type has
been chosen, or incorrect link has been applied by Refmac because
of inaccuracy in the initial fitting).
At lower resolution and with longer sugar chains the chances are
high that the sugar monomers remote from protein would have
multiple conformations, but the density would not be good enough
to model them correctly. In this case, modelling the density with
water molecules (pseudo-waters) is more appropriate. Do not
over-interpret the density!
5. Details of step (1), adding NAG linked to ASN
After step (2), the model with missing glycan moiety
(model_refmac1.pdb) and corresponding 21 and 11 maps
(4gos_refmac1.mtz) are opened in Coot from Refmac result page.
Here is a detailed description on building the first sugar into
Select "Icons and text" in the toolbar at the right margin of
the main Coot window.
Go to residue A 112 ASN (coot > Draw > Go To Atom ... and
select residue from the list). Note the green density, in which we
will build the polysaccharide.
Coot > File > Get monomer...; Enter NAG in the text box,
For easier positioning of the monomer, delete H-atoms from NAG:
press "Delete" button in the toolbar, check "Hydrogens in Residue"
checkbox in the "Delete window" and click on any atom of NAG
molecule in the Coot main window.
Open "Rotate translate" from the toolbar, click any atom of NAG
molecule and position NAG so that it fits into the corresponding
density reasonably well. In the correct orientation the O1 atom of
NAG should be the closest to the ND2 of ASN A112.
าReal space refine zoneำ; move individual atoms to their correct
position using left mouse button with while keeping Ctrl button
Make sure that the O1 atom of NAG is in the correct position
(very close to ND2 of ASN A112). Delete this atom, as it is not
present in the bound NAG.
Merge NAG into the original model as a residue A 201. This
involves the following four steps.
Coot main window > Edit > Merge Molecules*; check the
box next to "NAG_from_dict" and press Merge button.
Figure out what is the chain Id and residue number of NAG (in
this case this should be B 1 for the first added monomer; chain ID
will be incremented with each added monomer if you will try and
add several monomers in a single Coot session) Then go to Coot
main window > Edit > Renumber Residues* and select Chain
ID B, residue numbers 1 to 1, Offset 200. Press Renumber button.
Coot main window > Edit > Change Chain IDs*; select
chain Id of the new NAG residue (B in our example), check "Use
residue range" checkbox and define the residue range according to
the residue number of NAG (from 201 to 201 in our example). Type
in "A" in the text box "Chain ID". Press "Apply".
Mid-mouse click on any atom in NAG to check that the procedure
worked (the merged NAG residue should become NAG A201)
* In old version of Coot, this menu item is found in
Coot main window > Calculate
Perform the last check before saving the model and exiting
Coot. Open "Display Manager" window. There are two molecules,
"model_refmac1.pdb" and "NAG_from_dict". The latter had been
initially loaded from dictionary and fitted into the density.
"Merge molecules" action has added a copy of this molecule to
Using "Display" checkbox, show and hide these molecules in turn
to make sure (a) that copy of "NAG_from_dict" is indeed present in
"model_refmac1.pdb" as residue NAG A201 and (b) that the copy is
at the same position as the original.
Save the molecule "model_refmac1.pdb". Set the filename for
output coordinates to "model_refmac1_coot1.pdb". This is a good
filename for further bookkeeping with ccp4i.
Note "CCP4i Project Directory" menu in the window "Select
Filename for Saved Coordinates". Select the project 4GOS and the
output pdb-file will be saved in the project directory.