Tutorial 1: PDB structure of the M2 receptor
This tutorial will show you how to read a PDB file and transform it into a structure network. We load a structure of the M2 receptor, a GPCR protein, and perform a simple analysis of the connecting residues between an allosteric and orthosteric ligand.
The SenseNet plugin can load network data from several sources. As input source, we use the pdb structure 4mqt. Download and save the structure.
Load the network
Start Cytoscape install SenseNet using the App Manager located in 'Apps' submenu at the top window border. Once SenseNet is installed and loaded, go to the Control Panel on the left side of the screen. This panel contains several tabs, starting with Network, Style and Select. Click the tab named SenseNet. Select the 'Import network' button. A pop-up dialog will apear, where we configure our import settings. In the top menu box, select PDB structure contacts as import mode. You will see the panel expand to give you the specific input options. Click the button next to PDB file and select the previously downloaded 4qmt.pdb file. Leave the rest of the options on their default values. Finally, click the OK button at the bottom of the dialog.
Depending on your current layout settings, the nodes might be clustered together very closely. We recommend the following layout settings: In the top menu, select 'Layout - Settings...'. In the 'Preferred Layout' tab, set 'Prefuse Force Directed Layout'. In the 'Layout Settings tab', select 'Prefuse Force Directed Layout'. A new set of settings will appear. Of these, set 'Number of iterations' to 10000. and 'Default Spring Coefficient' to 1E-5. The resulting network will be spread thinner. You may also try different settings until you find an appropriate solution.
In this structure, the M2 receptor was co-crystallized with two ligands: The agonist iperoxo (IXO) and an allosteric modulator which is called 2CU in the pdb file. Kruse et al. reported that the presence of this modulator increased the affinity for iperoxo. We want to check if we can see this allosteric relation in our network.
You should see the residue interaction network in the main window. The nodes correspond to the amino acids of the structure, while the edges show the number of atom-atom contacts between two residues. The edge width increases with the number of interactions. You can see the exact number of contacts within an edge by hovering over it with your mouse. The large number of nodes and edges can seem overwhelming at first. However, the ligands should provide a good starting point for analysis. In order to find them in the network, we will use node coloring. Select 'Style' from the tabs on the left. At the bottom of the panel, you will see new subtabs for Node, Edge and Network options. Select the 'Node' subtab. Click on the Triangle next to 'Fill Color' to expand the options. For the 'Column' option, select 'con/residue index' and for the 'Mapping Type' option, select 'Discrete Mapping'. Scroll down to the bottom of the table until you find the entries 501 and 502. Right-click on the empty cell right next 501 and select 'Edit - Edit Selected Discrete Mapping Values'. Set a red color for residue 501. Repeat this for residue 502 with green color. You should see both colors applied to the respective nodes.
The first thing to notice is that the ligands are placed fairly centrally in the network. They also seem to be seperated by only a few amino acids. We want to see which amino acids are located between them using the shortest paths analysis tool. First, select the IXO node in the main window. Then, hold the shift key and select the 2CU node. Both nodes should light up in yellow to show that they are selected. Now, select the 'SenseNet' tab on the top left part of the Control panel. In the middle of the panel, find the 'Analysis - Paths' section and click 'Shortest paths'. A result panel should appear on the right side, showing you a table containing the paths found between the selected two nodes. If you have followed all the steps, there should be exactly one path. Click on the path in the table to highlight the nodes and associated edges. As you can see, the closest contact between IXO and 2CU goes through TYR-426, which establishes 6 contacts to IXO and 4 to 2CU. TYR-426 is part of Transmembrane Helix (TM) 7, one of the substructures undergoing conformational changes during the switch from the inactive to the active state.
Analyse the extended path network
What are other contacts in the interface between the ligands? We can use another path analysis to find out. Make sure to select only the IXO and 2CU nodes and then click 'Suboptimal paths'. A dialog will open asking for the minimum and maximum path lengths we want to find. As we know from our previous analysis, the shortest path between those nodes is of length 2. Thus, a minimum length of 0 and a maximum length of 5 should find more interesting paths. This time, the table on the right side will show a variety of pathways. You can click on each of them and see where they go through. Select all paths in the table by holding the shift key while selecting. You should see quite a number of nodes and edges selected now. As we are not interested in the other nodes at this time, we will adjust the presentation to focus on the nodes we have selected. The easiest way is to press CTRL + N (or alternatively: Click 'File/New/Network/From selected nodes,all edges'). You will end up with a new network containing only your selected nodes. Press F5 to recalculate the layout in your new network. You can also pull some nodes manually until you are content with the layout. The subnetwork is centered around the orthosteric and allosteric ligand as "hub nodes". The residues located between these nodes are potential candidates for transmission of allosteric signals.
Looking more closely at the subnetwork you created, it should be apparent that some nodes contribute to more paths than others. We can visualize this by calculating the 'betweenness centrality' of those nodes. In the SenseNet tab, click the 'Analysis - Network interactions - Centrality' button. In the appearing dialog, select 'Uniform' for the multiple edges weights option in order to calculate the centralities of an unweighted network. An overview of the results will appear on the right result panel. On the top of the results panel, click the Node auto style button. In the dialog, find the Style property option and select Size from the selection box and 'con/centrality' as column. Finally, click OK to generate a new network style. Nodes which are part of more shortest paths in the network will appear larger. You can see that TYR-426, TYR-430, TYR-403, TRP-422 and TYR-80 have particularly high betweenness centralities. It could be hypothesized that these residues play a role in mediating allosteric information between the binding sites.
This tutorial showed an example workflow for investigating contact pathways between an allosteric and an orthosteric ligand in a PDB structure. There are many other options which could be considered, which were not mentioned in order to keep the tutorial short. For example, one could be interested in treating polar and hydrophobic contacts seperately, include hydrogen bonds and secondary structure, or link the network to a 3D structure viewer.