tag:blogger.com,1999:blog-1136153439451224584.post1154882832849689770..comments2024-03-27T06:45:59.174-07:00Comments on Practical Fragments: Fragment linking on RPA: another protein-protein interaction inhibitorDr. Teddy Zhttp://www.blogger.com/profile/07288045760981372367noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-1136153439451224584.post-85178543585851369912013-12-20T05:27:02.053-08:002013-12-20T05:27:02.053-08:00I am one of the coauthors on this paper. I regular...I am one of the coauthors on this paper. I regularly read this blog, but do not regularly comment. Thanks for the writeup - you have correctly identified some of the challenges in this target and in the chemical matter that we have reported on in this paper. Lewis brings up an interesting point about the LLE above. Although I honestly did not really consider the possibility that the multiple binding modes for many of the fragments may be a indicator of desolvation energy playing a large role in the binding, it is quite clear to us that the hydrophobicity of the linked compounds is a major player in the binding affinity. We based most of this optimization work (and all of the fragment evaluations) on LE, but have incorporated LLE concepts into later optimizations (the LLE of this series is all pretty terrible, rarely going above 0).<br /><br />Regarding the fragment library, ours is built around the Rule of 3, but is not completely Rule of three compliant. Because the stated goal of our lab is to apply fragment screening to difficult PPI targets, we elected to relax some of the Ro3 requirements for certain substructures that have been common among PPI inhibitors in the past. I will point out that the compound 4 is not a fragment from our library, but was from a fragment-inspired purchasing campaign to expand the SAR. We did not limit to Ro3 compliant molecules in this purchasing and it thus led to some real grease balls. We were confident that we could build this hydrophobicity out of them, but this simply did not work in most cases.<br /><br />As Lewis mentioned above, this is a pretty difficult target. We are competing with an amphipathic helix that relies on considerable hydrophobic contacts for binding. We have worked on multiple series for this and the optimizations (even with LLE used as a metric and starting with fragments that are better than this from a clogp/LLE standpoint) keep coming back to highly hydrophobic molecules with acids. Balancing the physical properties with the binding affinity has been a significant challenge.<br /><br />This work continues - we should have at least a couple more papers coming out on this work in the near future. Alex Watersonnoreply@blogger.comtag:blogger.com,1999:blog-1136153439451224584.post-30231633049178758222013-12-18T06:26:54.181-08:002013-12-18T06:26:54.181-08:00Lewis raises an excellent point about LLE. I'v...Lewis raises an excellent point about LLE. I've never quite figured out how to apply LLE to fragments, but I'd have to agree that LLE values < 0 sound a bit dubious.<br /><br />On a related note, the <a href="http://practicalfragments.blogspot.com/2011/06/ligand-lipophilicity-efficiency-at.html" rel="nofollow">LLEAT</a> values are also very low, with all molecules having scores < 0.10.<br /><br />Looking at the two fragments more closely reveals that they are both very lipophilic, with cLogP > 3. Although the experimentals state that the fragment library was chosen according to "criteria related to the commonly used 'rule of 3'", they go on to state (presumably erroneously) that "compounds were removed from consideration if they possessed... a cLogP <i>less</i> than 3.0" (italics mine). A telling slip?Dan Erlansonhttps://www.blogger.com/profile/07927082337051189270noreply@blogger.comtag:blogger.com,1999:blog-1136153439451224584.post-65512246481821577592013-12-17T00:11:45.307-08:002013-12-17T00:11:45.307-08:00A property that I think is of relevance here is LL...A property that I think is of relevance here is LLE. For a few of the compounds here it would look like LLE is < 0 suggesting that much of the binding is occurring through the fact that the compound does not want to be in solution. Identifying both compounds binding at different sites when the crystallography of the compounds separately places them in the same site is consistent with this. Not to say that this is not good work, this target is likely extremely challenging. For the ABT BCL family inhibitors the LogP is indeed high, but for some of them they are extremely potent (<10 pM) thus achieve a positive LLE. Personally, a cutoff of LLE < 0 is often a good metric for identifying promiscuous compounds (as always not to be used as a hard rule!).Anonymoushttps://www.blogger.com/profile/13838291462523671842noreply@blogger.com