Combio 2006, last day roundup

Yesterday I breezed into Brisbane for the last day of the Combio 2006 meeting, to catch some talks, and make a showing to accept an award from the ASBMB.

Neal Saunders has been posting summaries of this meeting in Brisbane on his blog, so I thought I'd give my take on the last day too.

Here's are my highlights:

In David Claphams talk on transient receptor potential (TRP) ion channels, I learnt that menthol feels cold because it binds to an activates a TRP channel involved in cold sensing. Think about that next time you taste that cool minty freshness. (I woke up a 4 am to fly to Brisbane. The brain wasn't really kicking over just yet).

In the "Molecular Basis of Disease and Drug Design" session, K. Krause gave a very honest and entertaining talk on what he termed his "Night Science". ("Day Science" is the stuff that works out nicely, shows logical progression with no nasty inconsistencies or loose ends and gets talked about at plenary lectures. "Night Science" is the stuff that doesn't work out as well as we'd like .. it's confusing, there are loose ends and inconsistencies, despite carefully doing all appropriate controls. Not to be confused with "Bad Science"). Krause and his group were unlucky enough to find that a lead compound discovered through an in silico screen, which initially appeared to be a great inhibitor of alanine racemase, turned out to in fact be a potent inhibitor of another enzyme in their coupled assay. I wasn't inhibiting their target well at all (doh!).

There were actually a few examples of some somewhat disturbing results from in silico screens in this session, which I've seen similar examples of a few times before. Researchers do an in silico screen, and find some top-ranking hits, one or two of which are also good inhibitors in an assay. The co-crystal structure is solved, and reveals that the compound is not actually binding in anything like the conformation that the computational docking predicted (sometimes not even the same site). What is going on here ? Is it just the fact that in twenty random compounds one will turn out to be a weak inhibitor ? Unlikely, since then high-throughput real-world screens would have a much higher hit rate. Is it that the computational docking is half right, fitting one fragment of the compound which has high affinity well, and the other non-binding or weak binding half doesn't matter ? Probably more likely, but it still doesn't explain the cases where the compound binds in a completely unpredicted site. Food for thought: maybe many docking scoring functions for small molecules are good at selecting generally sticky molecules ...... (I don't do this kind of work directly, so I'm really an ignoramus on the issue).

I also went to the "Cancer - Emerging Drug Targets" session. Andrew Scott from the Ludwig Institute for Cancer Research presented some really encouraging results of early clinical trails for an EGFR antibody, and Michelle Haber of the Children's Cancer Institute Australia presented some results from two cell based assays, where 'high-throughput' screens have identified some inhibitors of the N-myc oncogene, and a drug efflux pump (MRP) inhibitor. I'd never really thought about it, but apparently those pesky cancer cells up regulate this efflux channel and actively pump out anti-cancer drugs, in a similar way to some parasites that become multi-drug resistant.

In the final plenary lecture, Nick Proudfoot told us about his work on transcriptional termination. It's still too early for the textbooks, but it looks like transcriptional terminators bind at the termination site and near the promoter regions in a lot of cases, turning genes into physical 'loops'. Whether this helps the RNA polymerase jump from the end of a gene straight back to the start to make the next mRNA transcript is still not proven, but it's an attractive model.

Combio is always a bit of an eclectic mix, but if you take it in the right frame of mind it can be good fun, and a nice way to broaden the scientific horizons a little. Needless to say, I slept like a log after all that.