Brian Sykes - Marius, coming to your chemokine story, I would have said exacdy the same thing when I saw IL-8 to that how could any protein not look like that, not be so beautifully symmetric. While IL-8 is a dimer, we (I and Clark Lewis) have synthesized an analog which is a monomer and fully active
in vitro. Do you think these dimers and tetramers are important?
Clore - I am glad you brought that up, it’s an interesting point. Ian synthesized an interlukin-8 in which he replaced an amide group by an N-methyl group where that N- methyl group was located directly at the 8-sheet interface. In solution that species is monomelic and if you look at the activity
in vitro, it is active. However, what he didn’t show were two aspects. First, he didn’t show what the nature of the species was when bound to its receptor, so the dimer isn’t excluded. The second thing is that although there’s absolutely no question that the form of this chemokine circulating in the blood is monomelic in so far that their association constants range from 10 -8-10 -7 Mand their concentrations in the blood are in the picomolar range, it is also quite clear that the chemokines don’t act in free solution. They act on solid support and in solid support when they are fixed to vascular endocelium, the local concentration of chemokine is actually very high and under these circumstances, they are definitely going to dimerize or in fact even oligomerize to a higher extent. In the case of hMIPlß it oligomerizes to a much higher extent. What I think is going on is that, while for just binding to the receptor in a test tube it is fine to be a monomer as far as its in vitro action is concerned, the dimeric form is key in so far as it actually has to bind to the surface of the vascular endothelium. I think that these oligomers are a key aspect in fixing the chemokines to the vascular endothelium. If they fail to do that, they just get washed away and then they have no effect whatsoever. So in vivo I think it’s important.
Carol Post - With regard to the distance restraints involving NOE’s between subunits, how were you able to distinguish intersubunit from intrasubunit interactions in the beginning? Clore - What you can do is allow the NOE’s to float. You can make them ambiguous and let them choose whether its intramonomer or intersubunit. In fact, in this particular case, you don’t need to go quite to that extent, because it is quite clear from the experiments that we did what is intersubunit and what is intrasubunit. So all we have to do is allow the computer to decide what the particular intersubunit interactions are and that’s exactly what we did to start with until we found out roughly what was going on. Once we did we could then bootstrap our way across to figure out all the remaining contacts and the assignments specifically between the subunits.
Carol Post - Marius, I am not real familiar with the p53 story. Could you explain what the oligomerization does? You said p53 interacts by activating a gene for another protein, so what about the oligomerization? Did you say the monomers were active?
Clone - The monomer has weak transcriptional activity.
Post - What about the point mutations which seem to be affecting the oligomerization?
Clore - There are four mutations in the oligomerization domain. There are thousands of mutations in the DNA binding domain. The frequency of mutations in the oligomerization domain is minute compared to that in the DNA binding domain and exactiy what role they have is unknown. The problem here is that people have sequenced p53 in plenty of tumors but nobody has sequenced p53 extensively in normal cells. I mean they have sequenced it once but not an infinite number of times. So nobody knows what the variation of the normal p53 sequence is. So I don’t even know whether those mutations actually play a role in mutagenesis or not. All I know is that they happen to be involved in the interface and it looks as if they would destabilize the tetramer. What is clear is that the efficiency of the monomer as a transcriptional activator, and for that matter just the DNA binding protein, is much reduced from that in the tetrameric form, and if you look at the DNA binding sites, the DNA binding site comprises four sites in close proximity.
Marc Adler - I have gotten used to seeing the figure in the literature that the rmsd deviation between the X-ray structure of the same protein and different crystal packing forms as 0.4 Å whereas you had mentioned that you were getting rmsd’s of 0.2 Å for your structure, so I am wondering about the discrepancy. Why do you think your structure has such a small rmsd on the backbone.
Clore - There are two things that I think you are confusing here. First, the rms between crystal structures can be much smaller than 0.4 Å. It can be down to 0.1 Å. It all depends on the quality of the structure, the type of structure, the influence of crystal packing and so on. How well the NMR structure is determined reflects simply the precision of the coordinates. The precision of the crystal structure is even higher than that. Now, why is that structure determined so well. Well, it’s determined so well because we happen to have something like 25–30 experimental restraints per residue and, I didn’t show this slide because it’s rather old, as you increase the number of restraints, the quality of the structure and the precision go up, just as it does in crystallography. If a crystal diffracts only to low resolution it means that there are relatively few experimental data points. As it diffracts to higher resolution, you have larger numbers of experimental data points, and a higher quality structure. It’s exactly the same in NMR. This doesn’t mean to say there can’t be discrepancy between this and the crystal structure of the order of 0.6 or 0.8 Å and it also doesn’t imply that the accuracy of the structure is 0.2 Å. In fact the accuracy of the structure is likely to be a factor of 2 down from that and we have shown that in calculations.