Finding a Structure

The first and perhaps most important step to the modelling process is that of finding a suitable structure from which to work. It was originally thought that the best candidate for the studies may well have been a structure for the protein with no Calcium atoms bound in the EF hand sites. We found one (1cfd.pdb) but noted that this was an average of the 25 lowest energy structures as proposed by an NMR experiment (all 25 of which are contained in 1cfc.pdb) and that these structures had hugely different conformations.

You have since informed us that your studies have been performed on a mutant form of Calmodulin called VU1. Although this structure was not available on the pdb, we did find a structure, but this was also inappropriate as it was for a complex with a bound ligand. We then set about trying to find the most homologous sequence for which a structure was available. To do this we used a program called Clustal on all appropriate (40 in total) structures from the pdb. Below you can see the output from the final round of alignments, i.e. those of greatest homology.

CLUSTAL W(1.4) multiple sequence alignment


2bbm            ADQLTEEQIAEFKEAFSLFDKDGDGTITTKELGTVMRSLGQNPTEAELQDMINEVDADGN
2bbn            ADQLTEEQIAEFKEAFSLFDKDGDGTITTKELGTVMRSLGQNPTEAELQDMINEVDADGN
4cln            ADQLTEEQIAEFKEAFSLFDKDGDGTITTKELGTVMRSLGQNPTEAELQDMINEVDADGN
1cm1            ---LTEEQIAEFKEAFSLFDKDGDGTITTKELGTVMRSLGQNPTEAELQDMINEVDADGN
1cm4            ---LTEEQIAEFKEAFSLFDKDGDGTITTKELGTVMRSLGQNPTEAELQDMINEVDADGN
1cll            ---LTEEQIAEFKEAFSLFDKDGDGTITTKELGTVMRSLGQNPTEAELQDMINEVDADGN
1cfd            ADQLTEEQIAEFKEAFSLFDKDGDGTITTKELGTVMRSLGQNPTEAELQDMINEVDADGN
vu1_rs20        ADQLTDEQIAEFKEAFSLFDKDGDGTITTKELGTVMRSLGQNPTEAELQDMINEVDADGN
                   **.******************************************************

2bbm            GTIDFPEFLTMMARKMKDTDSEEEIREAFRVFDKDGNGFISAAELRHVMTNLGEKLTDEE
2bbn            GTIDFPEFLTMMARKMKDTDSEEEIREAFRVFDKDGNGFISAAELRHVMTNLGEKLTDEE
4cln            GTIDFPEFLTMMARKMKDTDSEEEIREAFRVFDKDGNGFISAAELRHVMTNLGEKLTDEE
1cm1            GTIDFPEFLTMMARKMKDTDSEEEIREAFRVFDKDGNGYISAAELRHVMTNLGEKLTDEE
1cm4            GTIDFPEFLTMMARKMKDTDSEEEIREAFRVFDKDGNGYISAAELRHVMTNLGEKLTDEE
1cll            GTIDFPEFLTMMARKMKDTDSEEEIREAFRVFDKDGNGYISAAELRHVMTNLGEKLTDEE
1cfd            GTIDFPEFLTMMARKMKDTDSEEEIREAFRVFDKDGNGYISAAELRHVMTNLGEKLTDEE
vu1_rs20        GTIDFPEFLNLMARKMKDTDSEEELKEAFRVFDKDGNGFISAAELRHVMTNLGEKLTDEE
                *********..*************..************.*********************

2bbm            VDEMIREADIDGDGQVNYEEFVTMMTSKKRRWKKNFIAVSAANRFKKISSSGAL------
2bbn            VDEMIREADIDGDGQVNYEEFVTMMTSKKRRWKKNFIAVSAANRFKKISSSGAL------
4cln            VDEMIREADIDGDGQVNYEEFVTMMTSK--------------------------------
1cm1            VDEMIREADIDGDGQVNYEEFVQMMTFNARRKLKGAILTTMLAT----------------
1cm4            VDEMIREADIDGDGQVNYEEFVQMMT----------------------------------
1cll            VDEMIREADIDGDGQVNYEEFVQMMTA---------------------------------
1cfd            VDEMIREADIDGDGQVNYEEFVQMMTAK--------------------------------
vu1_rs20        VDEMIREADVDGDGQVNYEEFVQVMMAK--------------------------------
                *********.************ .*                                   

Key

Looking generally at the sequences, a very high degree of conservation is evident but careful examination of the final line of the alignment output reveals a blank marking; a significant difference. This mutation divides the sequences into two classes, the top three have a threonine at this position and the bottom group of five have a glutamine as found in our mutant vu1. All the other non matches are mutations particular to the vu1 mutant.

The bottom group of five structures were then viewed with a modelling program to narrow the search down to the structure 1cll.pdb. This is a high resolution structure of Human Calmodulin with Calcium bound but no ligand. This was thought to be the most appropriate template to use judging from the information provided.