Case study: chitinase-allosamidin complexes

A few important things before we start:

You can find this file (and the lectures) online at http://davapc1.bioch.dundee.ac.uk/shanghai04 (but please don't click on this link now, instead continue to use the version you are currently viewing)

Aims

The aims of this practical are to make you familiar with:

This will be done through a case study on family 18 chitinases and their interaction with a natural product inhibitor, allosamidin.

Introduction

Family 18 chitinases are enzymes that hydrolyse chitin, a polymer of N-acetylglucosamine and an essential component of the fungal cell wall, nematode egg shells and insect/crustacea exoskeletons. These enzymes are believed to be good drug targets against pathogenic chitin-containing organisms. Several chitinase inhibitors are known, such as allosamidin, and the peptide inhibitors CI-4 and Argadin/Argifin. Allosamidin is a nanomolar inhibitor of most family 18 chitinases, and is made of two unusual sugars (N-acetylallosamines) and a pseudo-sugar moiety termed allosamizoline (see Figure below).

We will use the crystallographically determined chitinase-allosamidin complexes to demonstrate some practical aspects of the material covered in the lectures. Most of the experimental data used in this practical is covered by this publication.

Choosing the right PDB file

We will first try to apply rational criteria for selecting a PDB file from a range of possible ones.

You should now get at least 7 complexes, 1 of which is an allosamidin derivative. Now we would like to decide which of these is the "best" structure to use for analysis of chitinase-allosamidin interactions. We do that through the criteria discussed during the lecture, that is the resolution of the diffraction data, the crystallographic R-factor, the difference between average B-factors for the protein and the ligand and the ligand electron density. This gives you a table showing resolution and R-factor. Basically from these criteria all structures look reasonable, with as "best" structure 1LLO. Now use the "Create a tabular report" function again, but look at "Crystallisation description". The pH optimum of chitinases is around pH 6.5 - is 1LLO still such as good option? As a further check let's compare some electron densities. As you can see, even at 2.5 Å, a respectable resolution for a protein-ligand complex, there is still a lot of uncertainty, and the density looks more like a "shapeless envelope" rather than closely following the contours of the allosamidin molecule. Now use the Electron Density Server again to investigate the human chitinase-allosamidin complex (PDB code 1HKK). When you have opened the viewer, use the "Select" function and center on residue 138 - you will see an allosamidin molecule sitting close to it. Does the electron density for allosamidin in this structure look better?

This is a good time to experiment a little bit - if you are interested in a particular protein-ligand complex as part of your own work, why not have a quick look at the electron density now!

Now download a slightly cleaned up version of this 1HKK PDB file by clicking here and save it as humanallo.pdb . To further investigate the degree of order of the ligand in the complex, we will now compare the B-factors of the protein versus the ligand. Start rasmol by typing rasmol humanallo.pdb.

As you can see, the mobility of the ligand (fat sticks) is very similar to that of the surrounding residues(thin sticks), giving us further confidence that allosamidin really was occupying all active sites in the crystal in an ordered fashion.

PRODRG

As explained during the lectures, PRODRG is a useful tool for generating ligand coordinates/topologies and other small molecule-related goodies. The webserver is at http://davapc1.bioch.dundee.ac.uk/prodrg, a FAQ is here, a recent PRODRG paper here and a related review here. Please note that although a license is required to run PRODRG locally (sorry, University of Dundee rules), we're happy to waive license fees for academic users.

Even if we didn't have any allosamidin coordinates, we could generate these with the PRODRG server. Just to make this point, let's try to generate coordinates for allosamidin's unusual allosamizoline moiety. Go to the PRODRG server and try to put in: 

O              
|          C
c---c--O   |
|   |  |   |
|   |  C---N
|   |  "   |
C-C-C--N   |
| |        C
O O
Try the same again with this MDL Molfile as written out by ChemDraw: 
thing2.mol
  ChemDraw09130400252D

 14 15  0  0  0  0  0  0  0  0999 V2000
   -0.2062    0.4125    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
   -0.2062   -0.4125    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
    0.5784   -0.6674    0.0000 N   0  0  0  0  0  0  0  0  0  0  0  0
    1.0633   -0.0000    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
    0.5784    0.6674    0.0000 O   0  0  0  0  0  0  0  0  0  0  0  0
   -0.9909    0.6674    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
   -1.4758   -0.0000    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
   -0.9909   -0.6674    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
   -1.2044   -1.4643    0.0000 O   0  0  0  0  0  0  0  0  0  0  0  0
   -2.3008   -0.0000    0.0000 O   0  0  0  0  0  0  0  0  0  0  0  0
   -1.2044    1.4643    0.0000 O   0  0  0  0  0  0  0  0  0  0  0  0
    1.8883   -0.0000    0.0000 N   0  0  0  0  0  0  0  0  0  0  0  0
    2.3008    0.7145    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
    2.3008   -0.7145    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
  1  2  1  0      
  3  2  1  1      
  3  4  2  0      
  4  5  1  0      
  5  1  1  1      
  1  6  1  0      
  6  7  1  0      
  7  8  1  0      
  8  2  1  0      
  8  9  1  1      
 10  7  1  1      
  6 11  1  1      
  4 12  1  0      
 12 13  1  0      
 12 14  1  0      
M  END
You can see how in the context of an SD file (a database flat file based on the MDL Molfile format) PRODRG would be useful for automatic generation of coordinates for a large library of molecules!

If *you* have a favourite ligand - try it out and see if it works.

Protein-ligand complexes with GROMACS

OK - now back to the main topic - the chitinase-allosamidin complexes. Download this tar file and unpack it with gzip -d -c complexes.tgz | tar xvf - . A dir/ls command will show that you have generated a number of PDB files and GROMACS mdp files, the function of which will become clear later. We will now use GROMACS to energy minimize the chitinase-allosamidin complex and perform a short MD run. Do the following:

Now we will do a few simple analyses (always keeping in the back of our mind that this is an unrealistically short simulation). Do: Now that you are an experienced GROMACS protein-ligand simulator, use everything written above to address the following question. If you look at Figure 1 and Table I you will see four different allosamidin derivatives have been studied in complex with the human chitinase. Their PDB codes and inhibition IC50's are also given. Can you energy mimimize these complexes and see if a correlation between protein-inhibitor interaction potential energy (~ deltaH) and IC50 exists? The necessary PDB files are already in your directory - this time use energy minimization in vacuum (i.e. do not create a box with water, but keep the crystal waters) using the emvac.mdp file also already in your directory. If you find this too difficult or you are running out of time, skip this and move on to the next section.

Building ligand derivatives into a receptor with LigTor

Now for a demonstration of the PRODRG/LigTor building option. We start from a "stripped down" version of allosamidin, which lacks the 6-OH and the 2-N-acetyl groups and part of the allosamizoline moiety - we will be re-attaching these missing groups (with PRODRG) and optimize them in context of a receptor structure (with LigTor). Download the scaffold bare allosamidin here and call it hackallo.pdb and inspect it with rasmol, and compare it with the full allosamidin molecule which should already be in your directory as 1hkk-ligand.pdb. Could we now rebuild the full allosamidin and automatically choose the right conformation of the 6-OH and 2-N-acetyl side chains in context of the receptor structure? To do this, do: