Frontier living with lipids
March 15, 2011 § Leave a comment
The completion of stage 1 of the human genome project a bit over 10 years ago marked the beginning, not the end, of an era. It didn’t mean that we can stop hunting for the molecular components involved in biological behavior, or in disease; there’s still plenty to do. But now that we believe we have a handle on a substantial fraction of what’s in the genome, it’s suddenly possible to imagine obtaining similarly comprehensive datasets for all the other components involved in biology. Lots of progress is being made in working down the hierarchy of the “central dogma”, from the information encoded in DNA to making RNA to making protein. After making comes regulating, though, and there we have only made a beginning. There are many more-or-less empty spaces on the biological map, decorated in some places with the biological equivalent of “here be dragons”.
One of those mostly empty spaces represents our current understanding of the roles of lipids in biology. We know some things. We know that some lipids, notably the phosphatidyl inositol lipids, have extremely important roles in several signaling pathways. We know that a large number of proteins have lipid-binding domains, which is a pretty good indication that lipids are important. We also know that about 5% of the genome is concerned with making lipids and moving them around; and there are an increasing number of mass spectroscopy-based efforts to catalog the identities of the lipids found in cells and start to understand how lipid profiles change in different situations. But we’re far from having a comprehensive understanding of the effects lipids have on biological behaviors.
A recent paper from a large team led by Anne-Claude Gavin makes a foray into these uncharted regions (Gallego et al. 2010. A systematic screen for protein-lipid interactions in Saccharomyces cerevisiae. Mol. Sys. Bio. 6 430). In a strategy reminiscent of the earliest steps towards protein chips, they developed a method to immobilize arrays of lipids on nitrocellulose membranes and look for direct interactions between the immobilized lipids and the proteins in a cell extract. The interactions detected appear to be mostly interactions between lipid headgroups and proteins; the 51 lipids and lipid precursors they chose (plus some controls) are ones where the headgroup would be expected to be “visible” from the cytoplasm, and the 172 proteins they selected to label with a TAP tag and test for binding all live in the cytoplasm.
One of the problems of striking out into uncharted territory is that when little is known, there is little to show that you are going in the right direction. Gallego et al. therefore take care to evaluate the quality of their data, as best they can. They show that the protein-lipid binding events they see when their proteins are expressed in yeast extracts are reproducible, and are mostly also detected when the protein is expressed in E. coli. This provides some evidence that most of the interactions they find are direct, not mediated by other proteins in the yeast extract. Several other controls establish that the method picks up the majority of direct interactions that were already known — though not those that involve large complexes, transient enzyme-substrate interactions, or binding events that involve the hydrophobic tail of the lipid. Based on their tests, Gallego et al. believe that the quality of their dataset is at least as good as that of large protein-protein interaction datasets; there are false positives, and false negatives, but there are also many interesting new clues.
So, what is novel in these data? A lot. Proteins with a known lipid-binding domain bound to an average of 3 lipids, while those not previously known to have a lipid-binding domain bound to an average of one lipid each. Most of these ~500 putative interactions are new. The authors picked 34 novel interactions involving 10 proteins for follow-up, and pursued them using secondary assays including binding to liposomes, inhibition of the production of the relevant lipid using antibiotics, live-cell imaging, and, in one extreme case, full determination of the X-ray structure of the lipid-binding domain to 2Å resolution.
For 24 of these 34 interactions from the nitrocellulose assay, the authors were able to find some level of confirmatory evidence, such as selective binding to liposomes or cell membranes containing the right lipids. Not all of the proteins in this set had known lipid-binding domains, and so the fact that the authors see lipid binding in these proteins means that our definition of lipid-binding domains needs to be broadened; in at least 2 cases, Gallego et al. have fairly confidently identified domains that must have previously unrecognized lipid-binding activity. The finding they followed up furthest involves the protein Slm1, a component of the Torc signaling pathway that binds both phosphatidyl inositol lipids and sphingolipids. They found that Slm1 has specificity for individual members of these lipid families, and that the PH domain of Slm1 binds to liposomes in a way that is dependent on both of these lipids. This caused them to suspect that this domain of Slm1 was binding cooperatively to the two lipid headgroups. They purified the PH domain and solved the crystal structure to 2Å resolution, and indeed found two nice cavities, each containing positively-charged residues, that could each bind to one of the two lipids. Mutation studies show that both sites are necessary for the function of the protein.
The authors point out that many of the interactions they discovered couldn’t have been predicted from sequence alone, and comment in their discussion that “the number of novel interactions discovered shows clearly that even major classes of metabolites, such as [the phosphatidyl inositol lipids] and sphingolipids, have been insufficiently studied”. Quite a difficult statement to argue with, if you ask me.
Gallego O, Betts MJ, Gvozdenovic-Jeremic J, Maeda K, Matetzki C, Aguilar-Gurrieri C, Beltran-Alvarez P, Bonn S, Fernández-Tornero C, Jensen LJ, Kuhn M, Trott J, Rybin V, Müller CW, Bork P, Kaksonen M, Russell RB, & Gavin AC (2010). A systematic screen for protein-lipid interactions in Saccharomyces cerevisiae. Molecular systems biology, 6 PMID: 21119626