JLigand tutorial (link)

Biological macromolecules are polymers and, therefore, the stereochemical restraints for macromolecular refinement can be subdivided into two sets, restraints that are applied to the atoms that are all from the same monomer and restraints associated with the covalent bonds between monomers. Refmac dictionary contains three types of data blocks defining restraints: descriptions of monomers, their modifications (both for intra-monomer restraints) and links (inter-monomer restraints). Generic links for proteins, DNA, RNA and for some sugars are there in the standard Refmac library distributed with CCP4. Non-standard links can be defined in an additional user's library. Any non-standard link must also be explicitly specified in the LINK record of the PDB-file used as input for Refmac.

In general, if the restraints for compounds X and Y are already defined, there are two ways of defining restraints for the compound X-Y. Existing definitions for X and Y can be ignored and all restraints for X-Y can be put into a single description of monomer. Alternatively, necessary modifications to intra-monomer restraints for X and Y and inter-monomer restraints can be described in three additional data blocks. The second way is preferable for structural biology as it makes it possible to follow name conventions for monomers and atoms.

This tutorial uses a 1.6 Å resolution P2₁2₁2₁ crystal structure of pig cytosolic aspartate aminotransferase in complex with 2-methylaspartate (Rhee et al., 1997; PDB code 1ajs). The asymmetric unit contains a dimer formed by subunits A and B. The coenzyme pyridoxal-5'-phosphate is present as the external aldimine in subunit A, where it forms an Schiff base with the substrate analogue 2-methylaspartate. At the same time, the coenzyme in subunit B forms the internal aldimine with the side chain of Lys 258. The two aldimines will be defined as a monomer and as a link, respectively.

Useful schemes can be found in Okamoto et al., (1994): Scheme 1 presents formula and conventional atom names for 2-methylaspartate and Scheme 2 shows transaldimination from the internal to the external aldimine. Some other relevant schemes can be found in e.g. Vacca et al. (1997) and Jeffery et al. (2000). The last two are free access articles.

Before starting

DATA for this tutorial are in the directory JLigand_link_tutorial and include three files described below.

The coordinate file model.pdb and the X-ray data file data.mtz were generated from the PDB entry 1ajs. The coordinate file was modified as follows,

• Lysine-pyridoxal-5'-phosphate was replaced with lysine (residue LLP B 258),

• 2-methylaspartyl-pyridoxal-5'-phosphate was removed (residue PLA A 415),

• This incomplete model was refined using REFMAC (100 cycles).

Our goal is to produce the complete structure including ligands and refine it. This will include the following steps:

• Making description of a link between lysine and pyridoxal-5'-phosphate,

• Merging two ligands into a single ligand, 2-methylaspartyl-pyridoxal-5'-phosphate,

• Calculation of phases to generate the electron density,

• Docking ligands into the electron density,

• Structure refinement with ligand and link.

JLigand, Coot and REFMAC will be used for generation of library-file, docking and refinement, respectively. The library entries for the ligands will be created from scratch and compounds LLP and PLA defined in the standard library will be ignored.

Making description of a link

Actions (indented text) are performed in JLigand window except when the window is explicitly specified (underlined text).

• Start JLigand in the directory JLigand_link_tutorial

Terminal: cd JLigand_link_tutorial

Terminal: jligand

• Load LYS

Load Ligand: type in LYS, press enter and wait a bit

• To open help window:

Help > JLigand Mouse Help or

Help > JLigand Keypad Help

• Play with the molecule: rotate (hold the left mouse button and drag), move (hold right button and drag) and zoom (rotate mouse wheel).

• Show atom names,

Tick Atom Ids check box (the second horizontal panel from the top)

Note: Only the first letter of atom label is clickable; it is underlined when cursor is moved to it.

• Play with an atom: edit atom data (double-click on the atom, i.e. on the first letter of atom label), delete (press Del in the left panel and click on atom) and undo (press Undo button in the second line of the top panel).

• Move LYS away from the centre of display

Press right mouse button and drag

• Load PLP

• Delete O4A in PLP.

Press Del in the left panel and click on the first letter of the atom label

• Show hydrogen atoms,

Tick Hydrogens check box

• Delete HZ1, HZ2 and HZ3 in LYS. When this is done, exit the Del mode by pressing the right mouse button or any of the atom type buttons on the left panel.

• Define a bond between NZ of LYS and C4A in PLP.

Select Link in the left panel, click on N of NZ label and then on C of C4A label

Note: linked monomers are not editable except for the link's bond order; undo link to edit.

• Change the single bond NZ-C4A to double

Double-click on the bond

Edit Bond Type: Bond Type: select double > Submit

• Regularise the linked ligands

Ligand > Regularise > LYS-PLP

• Save the description of the link LYS-PLP in a new file refmac.lib, which will be used as an additional library for refinement. Also, save a template of the LINK record in linkrecord.pdb. The last file will be edited and added to the pdb-file.

File > Save as Link > LYS-PLP

Save as CIF-library: File: type in refmac.lib and press Save

Info: OK

Save Link Record: Yes

Save PDB Link Record: Save

Info: OK

NOTE: We will not need coordinates of the whole compound for docking because the coordinates of its constituents will be available from the standard library.

Merging two ligands into a single ligand

• Remove LYS-PLP

Ligand > Delete > LYS-PLP

Delete Ligand: Yes

• Load ASP

• Remove the atoms HA, H2 (HN1) and H3 (HN2). Names in brackets are for the old version of monomer library.

• Add a carbon atom bonded to CA. Hydrogen atoms bonded to the new carbon will be added automatically later during regularisation

Select C in the left-margin panel, click on CA to make it active (the active atom is underlined), press shift and click on canvas to place the new atom.

• Following the convention in Okamoto et al., (1994) rename CB to CB1

Double click on C of the CB label

Edit Atom Details: Atom Id: type in CB1 and press Submit

• Also rename C1 (the added carbon atom) to CB2, O to OT1 and OXT to OT2. A good reason for renaming the OXT atom is that the atoms with such a name are difficult to handle in Coot.

• Move modified ASP away from the centre of display

• Load PLP

• Remove O4A atom

• Link N of modified ASP and C4A of modified PLP

• Change the single bond N-C4A to double

• Regularise the linked ligands

• Change the ligand's Id from ASP-PLP to e.g. MSP (three-letter Id makes life easier)

Double click on the ASP-PLP label under the picture of the ligand (you may need to move up the ligand to see the label)

Edit Ligand Id: Ligand Id: type in MSP and press Submit

• What are behind the screen are two linked monomers. Currently the conversion to a single monomer is performed when saving into a file

File > Append as Monomer > MSP

Append to CIF-library: Select refmac.lib and press Append

Info: OK

• View the file refmac.lib

View > View File ...

View File: select refmac.lib and press View

NB. The ligand MSP is defined in a single data block, data_comp_MSP. The previously created restraints for LYS-PLP are defined in three data blocks. Two data block, data_mod_LYSmod1 and data_mod_PLPmod1, define modifications to be applied to monomers LYS and PLP from the standard library, and one data block, data_link_LYS-PLP, defines extra restraints associated with the covalent bond between modified LYS and PLP.

Currently Coot reads coordinates of a ligand from a PDB-file, not from a CIF-library file. Therefore we need to read the monomer MSP and write its coordinates in a separate PDB-file.

• Remove the link MSP

• Read the monomer MSP from the file refmac.lib

File > Open > CIF File

Import Ligand from CIF-library: select refmac.lib, press Open and wait a few seconds

• Save coordinates of MSP excluding hydrogen atoms

File > Save Coordinates > MSP

Save Coordinates: select Exclude Hydrogens and press Submit

Save Coordinates in PDB-file: keep the default selection File: MSP.pdb and press Save Coordinates

Info: OK

• The files refmac.lib and MSP.pdb are all we need from JLigand, so it can be exited and we proceed to the calculation of phases.

JLigand > Quit

Select an Option: Yes

Calculation of phases

At this point we need to generation an MTZ-file with phases. This file will be used to draw the electron density maps in Coot.

• Run REFMAC with zero cycles of refinement and model.pdb and data.mtz as input. This can be done using CCP4I, or you can copy and paste the following four lines into the command line,

refmac5 xyzin model.pdb hklin data.mtz xyzout refmac1.pdb hklout refmac1.mtz

make hydr no

ncyc 0

end

Docking ligands into the electron density

• Start Coot and open files refmac1.pdb and refmac1.mtz,

Terminal: coot

Coot: File > Open Coordinates...

Select Coordinates File: select refmac1.pdb; OK

Coot: File > Auto Open Mtz...

Select Dataset File: select refmac1.mtz; OK

• Examine the electron density. The green density corresponding to the missing ligands can be easily found.

• Load the description of MSP from CIF-library file and coordinates from PDB-file:

Coot: File > Import CIF dictionary

Select File: select refmac.lib; OK

Coot: File > Open Coordinates

Select Coordinates File: select MSP.pdb; OK

• Fit MSP into its density,

Coot: Calculate > Other Modelling Tools

Other Modelling tools: Find Ligands; Close

Find Ligands: Enlarge this window; in Select Ligands to Search for: tick .../MSP.pdb and Flexible? and then press Find'em! button

New Ligands: OK

Fitted Ligands: press Fitted Ligand #0 and then OK

• Refine fitting,

Coot: Calculate > Model/Fit/Refine

Model/Fit/Refine: Select Map

Select Map for Fitting: OK

Model/Fit/Refine: Real Space Refine Zone

Coot: Click twice on the ligand.

Coot: Keep left mouse button pressed and drag atoms into their density.

Coot: To handle an individual atom, move cursor to it, keep ctrl-key and left mouse button pressed and drag the atom into its density.

Accept refinement: Accept

Model/Fit/Refine: Close

• Append fitted ligand to the protein molecule (it makes sense to do this right now in order to mask the density for the next fitting)

Coot: Calculate > Merge Molecules...

Merge Molecules: Enlarge this window; in Append/Insert Molecule(s): tick Fitted ligand # 0, and then press Merge

• Clean up

Coot: Display Manager

Display Manager: Enlarge this window to see all the molecules and Delete Molecule buttons.

Display Manager: Unselect Display check box for Fitted ligand # 0. If the copy of fitted ligand is there in the graphics window then previous steps have been done correctly and you can proceed and delete used objects.

Display Manager: In section Molecules: press Delete Molecule next to MSP.pdb, Fitted ligand # 0 and Fitted ligand # 1. In section Maps: press Delete Molecule next to Masked (by protein).

Display Manager: Close

• Load description and coordinates of PLP from the standard library,

Coot: File > Get Monomer...

Get monomer molecule from libcheck: 3 Letter Code: enter PLP and press OK

• Fit PLP into its density

• Refine fitting

• Delete O4A atom from PLP

Coot: find O4A and centre on it; you may need to zoom in, as this atom can be very close to NZ of LYS

Model/Fit/Refine: Delete

Delete item: tick Atom

Coot: click on O4A

• Append fitted ligand to the protein molecule

• Clean up

• Save the merged molecule:

Coot: File > Save Coordinates...

Save Coordinates Molecule Selector: keep default selection .../refmac1.pdb and press Select Filename

Select Filename for Saved Coordinates: keep default selection refmac1-coot-0.pdb and press OK

• The molecule with appended ligand is saved into the file refmac1-coot-0.pdb and Coot can be exited:

Coot: File > Exit

Structure refinement with ligand and link

• Edit the file linkrecord.pdb. This file contains one line, which is an instruction for Refmac to apply the link LYS-PLP to specified residues.

If you have followed instructions precisely, refmac1-coot-0.pdb will contain PLP as residue No 1 in chain D ("PLP D 1"). Check if this is correct using a text editor or coot.

Edit chain IDs and residue numbers in linkrecord.pdb. These should be: "LYS B 258" and "PLP D   1". Keep format of the record as it was, i.e. four characters for the residue number etc.

• Insert the LINK-record from linkrecord.pdb into refmac1-coot-0.pdb, just before CRYST1 record.

• Run Refmac with the completed model. Use CCP4I, or copy and paste the four lines shown below into the command line. Note that the additional library refmac.lib is specified as libin in command line. If CCP4I is used, refmac.lib must be specified in the corresponding text-box.

refmac5 xyzin refmac1-coot-0.pdb hklin data.mtz libin refmac.lib xyzout refmac2.pdb hklout refmac2.mtz

make hydr no

ncyc 5

end

• Check, if the output file refmac2.pdb contains LINKR record (a few lines before CRYST1). If it does, then everything was done correctly.

• Notice the decrease in R-free compared to the previous Refmac run. (The values of R-factors can be found in the headers of PDB-files refmac1.pdb and refmac2.pdb.)

• Use Coot to examine the electron density of the ligands. (Start Coot; open refmac2.pdb and refmac2.mtz; navigate to the residues C1 and D1 using Draw > Go To Atom menu.)

NB. If you want Coot to display link, substitute " " for "R" in the LINKR record in refmac2.pdb. If this does not work, replace the LINKR record with the LINK record from linkrecord.pdb.