Tutorial: CATH/Gene3D

Note: We recommend you open this page's links in new tabs (hold down the Ctrl key whilst clicking the link or click the link with the right mouse button and select Open Link in New Tab) so that you don't navigate away from this page.

To help keep the tutorial short, we have provided short-cut links to complement the instructions. We recommend using Firefox or Chrome rather than Internet Explorer.

A tour through CATH and a story of survival

You are interested in a beta lactamase protein that you know confers drug resistance in a pathogenic bacterium and you want to find out more about the nature of the active site and why this pathogen is resistant to penicillin. Search Genome3D ( www.genome3d.eu) with the UniProt identifier P00811. You will see that the protein belongs to CATH superfamily 3.40.710.10 of beta-lactamases and DD-peptidases. The majority of Genome3D structure prediction methods assign the protein to equivalent superfamilies in SCOP and CATH.

Click on the Gene3D link to get more information from the CATH-Gene3D resource (or click  here if you get lost). Click on the Drugs entry of the menu on the left of the page to see information available on the drugs found to bind to the active site of this protein. There are two approved synthetic drugs and many other putative drugs for which there is experimental evidence.

Let's look at the CATH superfamily (3.40.710.10). Click on CATH superfamily (or  here). This page shows summary information for superfamily 3.40.710.10. Mouse-over the Species Diversity box to see the species distribution of the superfamily. You will see that this superfamily is largely confined to bacterial species. Now mouse-over the EC Diversity box to see the functions associated with different relatives in the superfamily. Apart from beta-lactamases, what is the largest functional group?

What about the evolution of the proteins?

Image by Mcstrother on Wikipedia

To find out when these different functions emerged in evolution, click on the  FunTree box (under Enzyme Function). You can see that FunTree provides a lot of phylogenetic and functional information for the superfamily. This is organised according to sets of similar structural clusters (SCs) on the left of the screen (below the superfamily links).

The DD-peptidases (or "D-Ala-D-Ala carboxypeptidases" in the EC diversity box, earlier) are bacterial enzymes that are also known as "penicillin binding proteins". They cross-link peptidoglycan chains to form a strong mesh-like structure in bacterial cell walls. The figure to the right describes their activity in Gram-positive bacteria.

These enzymes are serine peptidases i.e. during catalysis an acyl-enzyme intermediate is formed involving a serine residue. The use of D-amino acids by bacteria contributes to the resilience of their cell walls since proteins only contain L-amino acids and most peptidases are L-stereospecific.

About 2 billion years ago fungi evolved the ability to synthesize beta-lactam antibiotics which bind irreversibly in the active site of DD-peptidases and thus inhibit their activity. Because bacterial cell walls are constantly being broken down and remodelled by the bacteria themselves, an inability to regenerate cross-linkages rapidly results in loss of integrity of the cell wall and bacterial death. An example of a current day organism is the penicillin mould that produces penicillin and in 1928 this was the first antibiotic to be discovered by Man.

In response to this assault it is generally accepted that some bacterial DD-peptidases evolved into beta-lactamases that can break open the beta-lactam ring in an antibiotic and thus inactivate it.

How has resistance evolved? Can we see changes in the active site?

Let's look more closely at these two different relatives. Go back to the  Gene3D summary page for our protein P00811. In the Domain View section we can look at the domain family predictions for our protein, as well as regions predicted to be disordered and globular. Known experimentally-determined domain assignments for the protein from CATH are also available.

We can also look more closely at the function of our protein by finding out which functional family (FunFam) cluster it has been assigned to in CATH. FunFams are clusters of relatives that are very likely to have similar functions. Hovering the mouse over the CATH domain family prediction image, we can see that our protein matched very well to the domain 2ffyA00, leading to this domain prediction. Click on the CATH link next to this domain id (also  here). This loads the CATH webpage for this domain, which tells us that it belongs to the FunFam "Class C beta-lactamase CMY-10" (ID: 22208), and to our protein (under UniProtKB entries at the bottom of the page). We can therefore say that our original beta-lactamase, P00811 belongs to this FunFam. Click on the FunFam  link. An alternative way of searching for a FunFam of interest is to access it from a superfamily web page. Look for Functional Families in the SUPERFAMILY LINKS menu at the top left. Find FunFam 22208 (Beta-lactamase [FF: 22208] in the table (type "22208" into the filter at the top of the table) and open the link in a new browser tab.

Select the  Alignment tab to view the multiple sequence alignment of this functional family. Highly conserved positions have been identified by the Scorecons algorithm ("Scoring residue conservation" Valdar WSJ (2002), Proteins: Structure, Function, and Genetics. 43(2): 227-241) and are shown highlighted in red. Can you spot the SXXK motif (where X can be any amino acid) associated with the catalytic dyad used by this family? It should be around alignment positions 80-83.

Right-click on the Back to all FunFams button to open a new browser tab and to switch back to the browser tab containing the table of FunFams, open FunFam 22165 (D-alanyl-D-alanine carboxypeptidase [FF: 22165]). Click on the page's  Alignment tab. FunFam 22165 contains DD-peptidase relatives. You will see a similar motif as both enzymes possess the SXXK motif catalytic dyad (additional catalytic residues are often suggested but these are also shared by both classes of enzyme). This should be around alignment positions 69-72. So the difference in their catalytic activity is likely to be due to substrate binding residues that perhaps subtly alter the position of the substrate, alter the energetic stability of binding, stabilisation of the transition state intermediate and/or subsequent hydrolysis of the acyl-enzyme intermediate.

Can you see any highly conserved residues which lie close to the SXXK motifs but which are different in the beta-lactamase and DD-peptidase alignments? For example, the following positions involve quite significant changes: 85 (Phe) in 22208 versus 74 (Met) in 22165, and 88 (Val) in 22208 versus 77 (Tyr) in 22165.

How close are these residue mutations?

CATH provides two structural comparison resources. The  CATHEDRAL web-server allows users to submit a structure to be scanned against a library of representative CATH domain structures. The  SSAP web-server allows a quicker structural alignment and superposition of two structures.

To save time in this tutorial, we have performed these searches for you. If you click  here, you'll see a search of the 2ffyA00 beta-lactamase structure against CATH. The top hit (4netB00) is for another beta-lactamase within the same superfamily. Now let's compare the structure of the beta-lactamase with a structure from a DD-peptidase family. Look at this superposition (which will be prepared using the SSAP web-server mentioned above) of protein structure 2ffyA00 for the beta lactamase P00811 against the structure 3itbD01 for the DD-peptidase (UniProtKB id: P08506). How similar are the structures? The residues involved in the SXXK motif in the active site are highlighted in green. How similar are the active site regions of these functionally diverse domains?

Here are some residue positions that are highly conserved within each FunFam but different between the two FunFams:

22208 (2ffyA00) : ...SXXKT...V...
----------------: ...|||||...|...
22165 (3itbD01) : ...SXXKI...Y...

You can verify that each of these is highly conserved in the two FunFam alignment browser tabs you now have open. Let's select these residues in PyMOL: look at the two sequences at the top of the PyMOL window, find the SXXK motif (highlighted in green) and then, for both sequences, select the residues two and five residues after the K. (Note: the two sequences aren't aligned in PyMOL so these residues are in different positions.) Type show sticks, sele and then look closely at the side chains of these residues. Would you say that these mutated residues are likely to have an impact on the binding of the peptide substrate?

Some thoughts to take away

Novel beta-lactamases rapidly evolve and confer resistance against new Man-made derivatives of beta-lactam antibiotics (typically within two years) as is the case for all other existing classes of medicinal antibiotics, resulting in what may be an imminent crisis for modern medicine. People may start dying from scratches again as they often did in the pre-antibiotic days of previous centuries. Pharmaceutical companies are reluctant to invest in the development of new classes of antibiotics since they would need to be kept in reserve for emergencies and thus be infrequently used and result in little revenue.