At A Glance
Name: Gerard Manning
Position: Associate group leader, Sugen (subsidiary of Pharmacia), South San Francisco
Background: PhD in Biochemistry from Stanford, in lab of Drosophila researcher Mark Krasnow, 1997
Worked at Molecular Applications Group, Palo Alto, Calif., on bioinformatics for classification of bacterial genomes and expression analysis software (1997-1999)
Joined Sugen in late 1999, working on the role of signal transduction genes in disease
Published an article in Science last week entitled “The Protein Kinase Complement of the Human Genome”
What is your background, and how did you get into studying kinases?
I did my undergraduate work in Ireland, then graduate work in California, at Stanford University, with Mark Krasnow. It was a Drosophila genetics lab in the Biochemistry department. So my background is actually in genetics and because of that emphasis on gene function and networks of genes, I then went to work with a small bioinformatics startup called Molecular Applications Group. We did some work on microarrays, and some work on gene classification. My main project there was large-scale classification of bacterial genomes.
Then I joined Sugen, which is a biotech company dedicated to signal transduction therapies. Our main focus has been working with kinases, in part in oncology. Sugen goes back about ten years, so there’s been a gene discovery effort going on there ever since the beginning. When I joined, we had been just taken over by Pharmacia and Upjohn. We had been looking for novel kinases both by PCR cloning and by looking at the EST databases, which were growing very large at that point. When we were taken over by P and U, we had access to the Celera genomic database, and the public genomic databases were also starting to fill up. Then we moved into computational discovery of kinases with bioinformatics in the genome. And really [the publication in Science] this week is the end of a long road on that. It is our first publication on the human kinases. We published recently on some of the model organisms, yeast, fly, and worm, which were kind of warm up acts for the human kinases.
In the paper, you found a total of 518 kinases in the human genome. What are you planning to do with these kinases at Sugen?
For many years we’ve been pulling out novel kinases and characterizing them to see if they have a role in disease, then selecting some set of them to develop therapies against. This gives us a lot of new kinases. Over 100 of the 518 have either not been published or have been described in only a very sketchy manner. Also because we have every human, every fly, and every worm kinase, we have done a very comprehensive comparison of them, so we can annotate any kinase by its similarity to model system orthologs and to its evolutionary neighbors in the human genome. So that has helped us even with some well-known genes, to understand more about their function.
What about the mouse genome? In terms of kinase comparisons, that’s the one you don’t have yet...
We are working on the mouse genome, and in the paper, we mentioned that we found orthologs — exact pairwise homologs of human kinases — for almost all human kinases. That genome isn’t complete yet, and we’re hoping that we will have a complete mouse kinome maybe within a matter of months.
What do you think is the importance of the novel kinases relative to the other ones that were previously known?
Some of the kinases we found were kinases whose sequences were quite divergent. In most cases we do have homologs in model organisms that can help us. [But] in some cases they seem to be completely undescribed. Many of these are expressed in very restricted patterns. That’s why they haven’t popped up in cloning projects. That may be very useful for us, because it may be that these kinases have very selective roles. So if they have a role in disease, then inhibiting them may just inhibit the disease and not have much broader effects.
Can you also predict the phosphorylation sites of these kinases, or their substrate specificities?
We can to some extent. It’s not work that we’ve done, but a number of other groups have mapped motifs around known phosphorylation sites with a program called Scansite. One of the things that we found useful is that these programs will predict many phosphorylation sites. When we have many family members and when we have the mouse homologs for many of these, we can look cross-species and see which of those putative phosphorylation sites are conserved. So that’s helped us narrow down the highly predicted from the best reliable phosphorylation sites. We’re not working on that yet at the full kinome level. We’re being selective because the algorithms aren’t bullet-proof. But we are using that in some of our target validation and some downstream analysis on some target genes.
What are some of the kinome-scale studies that you would like to see done, either at Sugen or elsewhere?
I think it’s conceptually very interesting that we now have what we believe are all the protein kinases. We now have a handle on potentially every phosphorylation event within the cell, and that may give us a much better set of tools to go after all of the phosphorylation-based signaling pathways. One obvious way is [that] many of these genes, since they are novel, can be added to microarray platforms to look comprehensively at the expression of all kinases within any particular cell type. Another approach is looking at antibodies and phosphospecific antibodies to all of these kinases. In that regard, something that’s coming out with the paper is a poster that we’ve done in collaboration with Cell Signalling Technologies. It’s a large pullout poster with all the family tree of all of the kinases. CST and other companies have programs to develop large numbers of phosphospecific antibodies. So having the full kinome available, we hope, will spur more of those reagents being made available.
Do you see a company making an antibody array with possibly all of those phosphospecific antibodies on it?
That’s definitely a strong possibility. My understanding is that the technology is changing so rapidly that it’s hard to bet. I would wait until we have most of the antibodies. But I know there are companies doing that kind of array technology, and for us that would be wonderful if we could go into a tumor or a cell type, and for us or other people to ask ‘what is the phosphorylation state of all of these kinases?’ Because typically, kinases are activated by phosphorylation, so if we see a phosphorylated kinase we can then, to a large extent, conclude that it’s an activated kinase and what it’s signaling in that cell.
In general, what do you think the implications of this study are for proteomics?
I think it gives us a handle on all of the phosphorylation events within the cell, but given that phosphorylation is probably the most common posttranslational modification of proteins, this gives us in one sense an alphabet or a language to describe what’s going on when proteins are modified. Ideally we should be able to describe and identify every phospho-spot on a 2D gel, every band you pull down with an anti-phosphotyrosine or an anti-phosphoserine antibody. Many of these kinases we’ve discovered were completely unknown, and for several of them we’ve found that the genes in humans are unknown but in the same subfamily, but when we go to yeast and to fly and to worm, this same family is conserved. So over 600 to 800 million years this family of kinases has been present. It is obviously essential to all of these organisms. And yet we can say nothing about their function. So that’s very exciting to see something that’s been around for so long and ask the question of what they do.
Amongst the various genomes, we have found some kinases that are only found in vertebrates, and there are some that are absent in vertebrates. We also looked at some of the other domains whose function we had known something about. Where is the protein anchored, what does it interact with? Is it a transmembrane protein? All of those computational approaches can give us a much better clue as to what the protein is doing before we hit the lab.