Problem Set #3: Comparing Protein Structures Determined by NMR
and X-Ray Crystallography using Swiss PDB Viewer
In the past couple of weeks we have looked at structures of proteins determined by X-ray crystallography. However, some of the structures in the PDB were determined using nuclear magnetic resonance. This week we will compare structures determined using the two different techniques.
1. What fraction of the PDB is made up of protein structures determined by NMR?
Find the structure of myoglobin determined by NMR and view the structure within the PDB. What is strange about this picture?
Now look at the file, which should be 1MYF. How many myoglobin structures are found in this file?
If you were to open this file in Rasmol, only one structure would appear. As a result, we will open the file in a more complex but more powerful program called Swiss PDB Viewer. It is also available free on the web along with a tutorial and its manual at http://us.expasy.org/spdbv/. Because we will be using this program a lot this semester, I would definitely advise putting Swiss PDB Viewer onto your own computer if you have one and taking an hour or two to do the tutorial.
When you open the file, Swiss PDB Viewer will ask you how many structures to open; open all of them. You should get a very fuzzy view of your protein composed of all 12 overlayed structures. Why does this structure file contain so many similar structures?
You will also get a whole lot of menus at the top, because Swiss PDB Viewer is a pretty complicated program. However, note that there is no command-line window. Take some time to play with the program to see what it can do. In the “Wind” menu at top, select “toolbar” “control panel” “layer infos” and “seq alignment” top open all of these windows, and spread them out so that you can read them all at once. Look at the information contained in them, and see what is selected or deselected, turned on or off when you click on various things in the menus. Also try the “Select” menu at top; holding down shift while you select allows you to unselect things.
We would like some kind of measure of how similar these structures are to each other. A common statistical measure of the differences between the atoms is the Root Mean Squared Deviation between the atoms, or the RMS dev. A very small number, i.e. 1 Ā or below, means that two structures are identical; successively larger numbers indicate structures that are more different. Select all of the protein residues but not heteroatoms for one of the structures in the control panel window, and select two of the structures in the layers info window. Now go to the Fit menu at top and select “Calculate RMS deviation…” This program can calculate the deviation between two structures using just the alpha carbons or backbone atoms (this is good for two similar proteins with different side chains) or the deviation of the whole protein. It places the answer in the toolbar in red letters.
Run this program several times to calculate the RMS deviation between pairs of structures for the alpha carbons, backbones, AND side chains. What kinds of RMS deviations are you getting between structures? How many atoms are used in the calculation? Which deviations are the smallest, and which are the largest? Which parts of the structures (alpha carbons, backbones, side chains) are the most similar, and which are the most different?
You may also select particular secondary structure elements (Select-> Secondary Structure -> Helices) and only compare those parts, excluding termini and loops which can be highly variable. What kinds of RMS deviations do you get for the helices only?
This program will also calculate a Ramachandran plot for all of the residues in the protein. Select all of the residues in the “control panel “ window (or do Select->Select All) and then choose Wind -> Ramachandran plot. Sketch it out. What does this tell you? Where are most of the points? Do any amino acids have non-allowed conformations?
2. We would like to compare this structure to the structure of myoglobin obtained by X-ray crystallography to see how similar it is. Close all of the layers in Swiss PDB Viewer and open the file we looked at previously, 1DWR.pdb (you may have to go back to the PDB to find it again). Then open 1MYF.pdb, this time only opening one structure.
How well do the pictures superimpose? How well do the sequences in the Alignment window superimpose? What does this tell you?
The program is capable of matching and overlaying even vaguely related proteins, as we will see in later weeks. Go to Fit -> Magic Fit. What happens in the picture window? In the sequence window?
How well do the structures seem to match? Where do the biggest differences seem to be?
Now calculate the RMS deviation between the two structures. How different are they? How does this compare to the differences within the set of NMR structures? What does this tell you?
Based on this one example, do you believe NMR and X-ray crystallography give comparable structural information?
Please answer the following questions on another piece of paper:
3. The relative electrophoretic mobilities of a 30 kDa protein and a 92 kDa protein used as standards on an SDS-polyacrylamide gel are 0.80 and 0.41 respectively. What is the apparent mass of a protein having a mobility of 0.62 on this gel? (keep in mind that 1 Da is 1 g/mol).
4. SDS gels used to determine the lengths of proteins often contain dithiothreitol or beta-mercaptoethanol. Why?
5. Mass spectrometry cannot differentiate between leucine and isoleucine. Why not? What other method(s) could be used to decide between the two residues in an unknown peptide?
6. The octapeptide AVGWRKVS was digested with trypsin, which cuts at the carboxyl side of lysine and arginine residues. Would ion exchange or size exclusion chromatography be more appropriate for separating the products? Suppose the peptide was digested with chymotrypsin, which cuts at the carboxyl side of tyrosine, tryptophan, phenylalanine, leucine, and methionine. Which type of chromatography would now be preferred?
7. Edman degredation was used to sequence a set of peptides, but in one case the method could not be used to determine the identity of the N-terminal residue. The DNA sequence revealed that the N-terminal residue was glutamine. Why might the Edman degredation have failed? Be specific. (hint: think about the intrinsic reactivity of the terminal glutamine side chain).