In this example, we will see how to use Swiss-PdbViewer to do standalone modelling. This will involve manipulation of amino-acid sidechains by mutating them. Keep in mind that once you have modelled a protein, you need to perform a few steps of energy minimisation to improve the model by removing bad contacts. No direct energy minimisation is provided in Swiss-PdbViewer, and you will need an external force field such as AMBER or GROMOS.
Before doing the modelling, we will compare an insulin structure with an insulin-like structure so that you can see the differences. Mature insulin is composed of two protein chains (B and A), while Insulin-like has only one chain. In fact, proinsulin is synthesised as a single protein chain, but a peptide fragment (domain C) is removed by proteolytic cleavage which leads to two separate peptidic chains, linked by disulfide bridges.
Open the file 1HIQ.pdb (insulin)
Open another copy the file 1HIQ.pdb (insulin)
Select chain B (simply click on a B in the control Panel).
Switch to the second copy of the insulin and select chain A
Now Create a merged layer from selection (from the edit menu).
Close the two copies of 1HIQ only a "_merge_" layer should remain.
Remove All Hydrogens (from the build menu)
Save it as "1HIQswap".pdb
Close the layer.
Open the file 3GF1mdl1.pdb (insulin-like)
Remove All Hydrogens (from the build menu)
Open the file 1HIQswap.pdb (insulin)
Select Group Kind: Cys (hold the shift key while invoking this command to act on both proteins).
Color by Selection (hold the shift key while invoking this command to act on both proteins).
Fit Molecules (auto)
Improve Fit
Switch the view between the two proteins (with Control Tab) and note how the SS-bond pattern looks similar. Also note the stretch of residues present only in 3GF1mdl1. This is the C domain absent in the insulin (cleaved during the maturation). This part was just to convince you that there is indeed an homology between insulins and insulins-like proteins.
Now we will model a C. elegans insulin-like. But before, we will demonstrate that proteins might adopt several conformations. This is particularly the case in insulin-like, whose structure is maintained by several disulphide bridges. The PDB file 3gf1 contains ten NMR models. We will compare model #1 and model #4.
Close all layers
Open the file 3GF1mdl1.pdb (insulin-like)
Open the file 3GF1mdl4.pdb (insulin-like)
Select Group Kind: Cys (hold the shift key while invoking this command to act on both proteins).
Color by Selection (hold the shift key while invoking this command to act on both proteins).
Observe the differences: Disulphide bridges are at the same place, at the core of the protein, while "surface" residues can adopt various conformations. This is a fact that has to be kept in mind during modelling. Also, in this case, we can test our modelling hypothesis on various NMR models to test the validity, and then choose the best starting point to do the actual modelling.
Load Raw Sequence to model Q21506.aa (from the Swiss-Model menu).
Select Group Kind: Cys (hold the shift key while invoking this command to act on both proteins).
Color by Selection (hold the shift key while invoking this command to act on both proteins).
Manually align the sequence onto the template (try to match the Cys and observe what happens you have two possibilities to align the Cys (the spdbv project file can be obtained here)
Note how the first possibility makes only one SS-bond. In pink, is the location of the gap, and you can see it will be easy to bridge it by removing the C-like domain.
Note how the second possibility automatically makes a SS-bond between Cys 6 and Cys 36, and how Cys 24 is almost able to make a SS-bond with Cys 50). In pink, is the location of the gap, and you can see it will be easy to bridge it by removing the C-like domain. This alignment sounds better. Let's check that this alignment is also valid if one uses the NMR model #4 as template:
In fact, Cys 24 and Cys 50 are even closer, and it will be very easy form them to make a SS-bond by slightly rotating any of the Cys around its Carbon alpha. We can also check that it is indeed possible to bridge the gap using the "Build Loop" feature of the Build menu. Select Asn 26 and Pro 30 as anchor residues:
Then you need to check the sidechain packing and do an energy minimisation on the structure in order to release the bad contacts. The best approach here is to use Swiss-Model once you have verified that your alignment was correct. Once you get back a model, check that it corresponds to what you want, tweak some parts, and do an energy minimisation with a program such as AMBER or GROMOS.
Last modification: Please note that these pages have not been updated since 1999 and are provided on a "as is basis". We are currently working on an updated version of this course.