Executive Case Study: Proteomics - the next quest in the post genomics era
BUSINESS IN THE POST GENOMICS ERA: CHAIRED BY: Dr Romeo Paioni, Head Scientific and External Affairs, Novartis Pharma
By on May 17, 2001BUSINESS IN THE POST GENOMICS ERA:
CHAIRED BY: Dr Romeo Paioni, Head Scientific and External Affairs, Novartis Pharma
So we can move to the second talk of this morning, Carl Foster will present Proteomics - the next quest in the post genomics era. Carl Foster has a broad education, has an MBA in Marketing and MS in Biochemistry from University of Kansas, additional education in Economics at the University of Illinois and has xxxx Administration at xxxx. After having worked for many years for Merck and xxxxx and then also for Intracell he is now the Commercial Director and Senior Vice President of Business Development for Oxford GlycoSciences.
9:00 Executive Case Study: Proteomics - the next quest in the post genomics era
Carl Foster, Commercial Director, Oxford GlycoSciences
Thank you. I'sm filling in on somewhat short notice for our CEO Michael Crander who regrets he couldn'st be here today, hopefully because we will have some good news coming out shortly on some projects we are working on but I wanted to give you a good idea of what OGS Oxford GlycoSciences is focused on and what type of company we are and a little bit about our view of the world post genomics and in particular the proteomics field.
OGS is not a newcomer to the biotech industry. We have been around since 1988, it was a spin out from Oxford University. We actually have a xxxx in Milton Park, England and in Avindon, England, both near Oxford. Currently we have around 200 employees this year; with the upcoming launch of a new product we expect to probably increase that to over 300 people. Our current market cap is $1.2 billion of that we have around 300 million in cash.
OGS is listed on both the London and Nasdaq stock exchanges. We did our offering on Nasdaq, the IPO on Nasdaq last December and raised a hundred and seventy million pounds. It was, I believe, the largest European Bio-tech financing ever and we have an extensive set of collaborations with pharmaceutical companies, government and academic partners.
To give you an idea of the type of relationships we have, on the pharmaceutical side companies like Pfizer, Bayer, Merck, Glaxo; on the technology side Cambridge Antibody, PBio Systems, Packard. We also recently signed a relationship with the Institute for Systems Biology, that's the Lee Hood, Rudy Abersolf group that came out of the university of Washington and on the drug development side, this is a piece that is relatively new to OGS. We have decided to retain ownership in some of the novel disease associated proteins that we discover and to develop those into markers and new chemical entities and immuno therapeutics on our own. The first collaboration is with MeterX where we are taking a series of breast cancer antigens that we found to be quite interesting and are developing antibodies and those will enter the MeterX T12 Development Program. We expect to begin to file the I&D for the first product later this year.
Just to kind of give you where we think we operate in this whole field of genomics and proteomics, the genome tells you what could happen . End of tape
.become quite clear that there is a lot of good information that has come out with the genome but it doesn'st actually tell you everything that is going on. We have actually done quite a lot of study at the MRNA level and have found there to be a very poor correlation and the amount of message and the actual level of protein that has been transcribed and so we choose to focus directly on the proteins to find novel disease associated proteins and then to use those as targets for the development of drugs and markers. Just kind of graphically where we participate is here in the translational post-translational modification and protein protein interactions.
The rationale for why we have chosen this field is that there are relatively few number of known protein targets with which drugs interact that the pharmaceutical industry is now more than every driving to find new protein targets and this has fuelled the investment not only in genomics but also over the past five years or so in the antibody field. We feel that proteomics offers a very direct means to find those targets and also find the disease association and to develop drugs.
So there are two main end points that we look for. One is to find a good druggable protein that can be used as a target and again we would take that protein and then perhaps create antibodies to that as we are doing with Meterex or to create new chemical entities. We would also in parallel look to find markers that would allow for early differential diagnosis of disease and/or assessment of patient response to various therapies.
What we start with is a target disease area with appropriate control indications. We then go out to a network of biosourcing universities, academic collaborators that we work with. In the case for instance of breast cancer we have worked with the Ludwig Cancer Research Institute and developed the appropriate pathology and self sorting and patient selection to find exactly the type of material that we felt would lead us to the best proteins. We then take tissue samples using differential analysis then will go into to our actual technique. We identify the proteins and try to map those on to specific pathways and ultimately come up with new drug products that would enter development. In parallel to this we find a more readily available sample like serum, cerebral spinal fluids, xxxx fluid. We take samples of that and look for the same type of markers as we are finding in the tissue and those then can be used in product development down the line.
This is kind of the short course on what OGS does. We start out by running many two-dimensional gel arrays. We.in some cases for instances in Alzheimer's I'sll go into a little bit more fully, we had over five hundred different samples, five hundred different protein expression maps. We stack those on top of each other using the computer image. We use embedded marker proteins, these red arrows, as a way to align all of the different gels, running to D gels is somewhat of an art form in the variability and lack of reproducibility has always been a problem. We think that we have overcome that and we use some very sophisticated work/warp in algorithms to align hundreds and hundreds of different gel samples and add those into a database. What we come up with is a proteo map of that particular disease, so in this case this might be a cerebral spinal fluid proteom of all the controls and all the disease samples that we have, those are all of the proteins mapped on to one computer gel.
The computer can then look at that gel and determine which one of those spots is differentially expressed. We can then direct a robot to go back and pull out that spot, it cuts it out, we digest the fragments and then do mass back and identify the actual protein. The gene can then be determined once we have the protein sequence.
The approach we use has proven to be reproduceful, it's sensitive, we have archiveable gels and I will talk about that a little bit more as well. We have automated this entire process, we use the very latest in mass spec interfaces and have created our own internal database informatics and data mining tools. As I said there are about two hundred people at OGS, somewhere around seventy-five of them are just in the IT area working to make sure that we can mine the data.
Our factory, that we like to call it, is about 50 000 sq ft of proteomics space, our capacity is 40 000 two-dimensional gels per year and right now we can actually do around 500 000 mass spec. We, as I said earlier, have a collaboration a technology collaboration with Applied Biosystems and we are the beta test site for their new top Tof Mac Spec instrument and we believe that that instrument will be about a hundredfold better than anything else on the market and we anticipate in the next twelve months being able to do a mass spec sequence every second.
Out of this we have been able to capture a large patent estate, each disease associated protein that we file intellectual property on is physically isolated, chemical defined and disease linked so it fits well with the criteria that the PTO's are establishing about having use of application. These are some of the areas in which we filed protein patent filings. At the end of last year we had actually around 2000 proteins filed and we expect to have twice that by the end of this year.
Just to give you an idea of the growth and scale that we have been able to accomplish. It has been almost logarithmic for the last four years and we are incorporating a new technology this year, as I mentioned we have a relationship with the Institute for Systems Biology. Rudy Abersol invented a technology called I-Cap and this would be a way to go directly from sample to mass spec without running any type of two-dimensional array and we anticipate having that up fully operational within the next year. We are running it at a bench top level right now; we hope to industrialize that process very soon.
The business model that we have for the company is threefold. We use much of the data that we find in database products. In the past we have worked with Insight Genomics and we plan to continue to build database products into the future and we retain ownership in protein and GNIP. Collaborations with our pharmaceutical partners focus on targets and markers. We hope to develop drugs and useful clinical research tools from those. In many cases we retain some rights such as diagnostics particularly if the pharmaceutical company does not have an active interest in the diagnostic field, then we retain those rights. Typical deal structure include fees, milestones and royalties. We also have our own internal research development program. As I mentioned we are going to be launching a product later this year called Bavesca; this will treat Galshay disease,. Galshay is a genetic disease that affects about 15 000 people worldwide and we have a novel therapy to treat that. We also have as I mentioned our own pipeline of new drugs coming through our collaboration with Meterex. First in the area of databases, just to give an idea of the scope, we have around 3 000 protein expression maps already in our databases and of these there are about 30 000 proteins that have been annotated.
On the collaboration side the typical way that we work with pharmaceutical company is to start with a disease area. We then develop a research plan that helps to identify the appropriate patient population and answer the specific questions that are of interest and OGS receives upfront payments, ongoing research development payments, milestones and royalties. Wherever we can we try to flow much of that data back into our database business. If you run a gel and you find 3 000 proteins, there may only be a half a dozen on there that you are really interested in; we would like to take all of the rest of that data and use it as a good background for our databases.
We usually take joint intellectual property rites, we don'st do any type of Fee for service business and we try to maintain value capture wherever we can, as I said some companies are not interested in for instance diagnostics, other companies that may be xxxxx in the immuno therapeutic field may not have any small molecule capabilities and so we would retain NCE rights as an example.
The types of collaborations that we have with Pfizer - we'sre looking at targets and markers for Alzheimer's and Athrosclerosis; with Merck we just finished a study looking at mechanisms and type 2 diabetes; also in the agricultural area is the previous speaker had interest there; we are working with Piner Hybrid the corn company to understand gene traits of corn. Probably one of the most exciting new opportunities we are pursuing this year is this area of drug safety and compound profiling and the idea is that if you take a sample from an animal and look at their proteomic map and then treated that same animal with some particular therapeutic regimen, maybe it is a pre-clinical drug candidate that you have and then take another sample and note the proteomic changes. The proteins that change may be able to be mapped to certain toxicology pathways and we have done a study with the FDA that shows in one particular case with Fiberate that we could have predicted that that drug would have had cardiotoxic effects and so by determining which proteins and which pathways are going to be affected, we hope to be able to screen pre-clinical drug candidates more effectively to show which ones may be of interest.
Another way that you could use this is with drug rescue and for instance we had a company come to us recently that had a product in phase 2 ready to go into phase 3 and they became concerned about a rare adverse event that may or may not show up in the small population of their phase 3 studies. They asked us to run the proteomic analysis of that to determine whether we though it was predictive of this particular adverse event and then if it was to compare it to two pre-clinical back-up compounds they had; the idea would be that if those looked better they would take those forward instead of an expensive phase 3 study with a slightly risky compound. With Bayer we are working in the area of targets, markers to respiratory disease.
This just gives you kind of an overview of one of our collaborations with Pfizer. In this one we filed joint IP on a number of bio markers that changed in Alzheimer's disease. We hope that those can be used in clinical design to improve patient selection, predict a value, monitor drug responses, they take some of their Alzheimer's products through to market and we have started a secondary program in Arthro-sclerosis.
From this work we were able to build a cerebral spinal fluid proteom. As I mentioned there were hundreds and hundreds of samples from controls from various types of disease patients and we have put all of those into one master proteom of cerebral spinal fluid; there are over 2000 individual proteins in there, almost all I believe all now have been identified and quantified and we think that this is going to be a very significant resource for companies moving forward looking for new targets and disease markers. This will be a database product that we plan to launch some time this summer.
Just give you an idea of the type of samples we put in there; the Alzheimer's patients we had age and sex matched controls, we had many longitudinal samples, patients over time, I believe it is about every six months for several years. We wherever possible try to take samples from family members that do not have the disease to give us as broad a spectrum as we can.
This is actually one of our computer generated 2D array. This actually doesn'st exist but it is the cubed computer reading of what one of the gels would look like and by taking the XY co-ordinance of the actual dark fuzzy spots making a cleaner computerized picture, the computer can then sort the data much more efficiently and direct robots to go back and retrieve actual proteins more efficiently.
Just to show you how you zoom in. One of the things that we are not used to reading 2D gels that you can see from a picture like this are actually the post translational modifications, that is kind of these chains, some are phosphorylation chains, some are glycosolytion chains and it is often times the post translational modification that was actually important. If you looked just at the MRNA or at the genomic level you would not see the many different changes that happen after the protein was actually made. We have one case where the important isoform of a protein was the 48th dot in the post translational modification chain and that would have never been picked up unless you were looking specifically at a 2D array.
To give you an idea of the computer power that is required to sort through something like this we had in the Alzheimer's database over 500 protein expression maps. Many of those are run in duplicate, there were 700 000 gel features that features a protein spot. We map those on to an XY co-ordinance system and then we sort those out using their molecular weights and isolectic points and then do the mass spec. In this case out of the 2000 features we found about 115 that were of some interest as either a target or marker.
Again the type of diseases that we are working in, Alzheimer's, schizophrenia all the major C&S diseases, we try to get a broad capture of various tissue types. I will say that bio-sourcing and tissue collection is really a critical thing and it was good to hear that in Utah they don'st have that much of a problem but it is a critical thing to get good material and in the case of for instance Alzheimer's the material you most often get, the primary brain tissue is from deceased patients and it would be nice to be able to get, in some fashion, brain tissue from living patients but that as you can understand is not something people volunteer for.
We plan to continue our expertise in the C&S area, continuing to build our databases and xxxx two other key diseases. We plan to start these studies this year. One of the strategies that has worked well for us is that when we have a collaboration like we did with Pfizer in the C&S area, we then build on that expertise with other collaborators and even our own internal research. Once you have worked out the biology and the pathology of the tissue, running the gels, understanding the proteins in that area becomes much easier. In addition to C&S we are doing this in cancer expanding beyond just breast cancer into the other major solid tumors.
Underlying all of the protein and gene discoveries that we have made we believe that the core to keeping ahead of thenow I saw recently there were about 400 companies that have morphed into proteomic's companies in the past six months or a year or so. We believe that in order to stay ahead of these companies we need to continue to be a leader in the proteomic's technology itself. We had a patent issue last April, it is entitled Computer assisted methods and apparatus for identification and characterization of bio molecules in a biological sample. We believe that this technology, which I will describe in more detail, gives us a very strong position in doing 2 dimensional type of work and helps to broaden what we call our land grab strategy of trying to capture and identify as many disease associated proteins as possible.
This particular patent covers the imaging of the 2-dimensional array, in our case it was a gel, to generate computer readable output, using that output and sorting it according to previously defined criteria. In this case it would be something like me Tell me all of the spots that show up in the disease samples and not in the control samples,'s and then generating machine readable instructions to direct a robotic device. Now each of these things has been around for quite a while. People have been running 2-dimensional gels for thirty or forty years, people have been imaging those gels, people have been using robots to cut spots. What this patent covers is the computerization of those steps. When you use a computer to make the decisions for you that is what is covered by our patent. It does cover the use of polychromide or other gels, it does cover the covalent bonding of the gel to a solid sport system. In our case we have gels where after you run the gel you can remove one plate, the gel remains stuck to the other side. It allows you to very easily archive these, we can go back for years at a time and pull out gels and re-stain them and the proteins can still be found where they should be. We develop our own fluorescent dyes, we build our own imagers and robot devices. Most of the platform that we have is custom built in-house and proprietary.
Just to give you an idea of what that patent means. These are two images of candidalbucans, this is an image of a non-virulent strain and this is a virulent strain. These are the proteomic 2-dimensional arrays behind those. Now if you had to eyeball and figure out without using the computer which of these thousands of spots, many of which are very fuzzy, some of which may overlap, which of the thousands of spots are different between these two gels, it would be very difficult. Using the computer to do that is what is covered in our patent.
So as I said we think this is going to give us a very powerful marketing advantage. We expect that our collaborators will be very pleased to work with us since we do have this leadership position. It does raise the barrier to entry to other competitors that may try to pursue this route or to big pharma companies that try to develop their own in-house technology. Beginning this year we will also pursue a commercialization strategy that may involve licensing this technology to various users or instrumentation companies.
Just to wrap up OGS, we believe that proteomics is going to be a core part of the future drug discovery and development effort and that we are a leader in the proteomics field. We have a proven and industrialized process, we have demonstrated this in a broad number of diseases in various applications and have collaborations with many leading companies in the industry. We continue our land grab strategy with many protein patent filings and we'sll launch our first product with Veska later this year in the US through our own sales and marketing force and next year in Europe also through our own sales and marketing organization.
Thank you.
ROMEO: Thank you very much Carl for this very comprehensive illustration of all the activities, interesting aspects. Unfortunately Carl will not be with us for the panel so we may take some questions related to this topic. May I ask who would like to start? Yes, please.
QUESTION: Andreas Weith, Boehringer Ingelheim. A question that comes up repeatedly on the citation actually of those who market and who do proteomics is that the correlation between messenger R&A abundance and protein abundance in the cell is very poor and whenever you ask these guys how poor is this correlation you get very mixed answers. Can you comment on this?
CARL: We have done a lot of studies in-house and I don'st know that there is any statistic I can tell you as to what the correlation is, it varies a lot. We find some proteins where there is very high expression level, we find other proteins that are very abundant where there is no detectable expression level and so it is really all over the range. All I can say is the one slide I have seen from our scientist is that you would statistically say there is no correlation or very little correlation between the levels.
QUESTION: Michael Rhodes, Columbia University. It is easy to understand why you focus on cerebral spinal fluid for.because you can, you can'st get a lot of tissue out of xxxx subjects but what kind of information does your company have about the relation of the proteomic expression and CSF compared to xxxx neural tissue itself?
CARL: Well that is what we hope to discover and why we pursue parallel paths with each patient that we do get tissue from, we also try to get cerebral spinal fluid in that case so that you can look for.. if you find a very significant protein that you would like to develop as a drug target from the tissue, you would like to be able also match that up with a similar protein finding at the CSF level so that you could monitor and diagnose much easier and then ultimately, you know, in many diseases if you could do that in serum it would even be better than CSF. So that is the ultimate goal, it doesn'st always happen, some proteins don'st show up in the CSF that showed up in tissue or they are not as predictive.
QUESTION: It came as an utter surprise to me that you said your technology are two detail electro xxxx is both reproducible and sensitive. This actually is the first statement that I ever heard from somebody producing 2-D details/gels and I need you to comment on it because my impression so far and I think I am not far off track is that there is a huge number of proteins that don'st enter the gel, that don'st show up in the gel because the dynamic range of staining in the gel is not sufficient unless you heavily over-stain the gels in order to see the low abundance proteins.
CARL: Yeah, that has been a problem in the past and I think we have been able to overcome that to a large degree. As I said running 2-D gels is somewhat as an art form as you know yourself. We do a couple of things. We spend a lot of time creating chemical ways to screen out high abundant proteins first of all so that the images that I showed you of the 2-D gels had already had things like albumen removed. If you don'st remove the albumen it is just a big black smear. So we spend a great deal of time removing the key high abundant proteins so that the low abundant proteins will show through much better. When you remove all the high abundant proteins you have effectively been able to concentrate more of the low abundant proteins into a sample and therefore you get better resolution of those low abundant proteins. We also, I mentioned develop our own proprietary fluorescent dyes. These dyes are not commercially available and we believe that they are logarithmically more sensitive than the commase blue and the silver staining and other things that you could buy on the market.
QUESTION: (Can'st hear, sorry)
CARL: I'sm sorry, the what that we use?
QUESTION: The logarithms, the computer programs.that you align the gels, are they xxxx in the US xxxx
CARL: No, no. Theit's not. Yeah, the patent is very broad, it says that if you use computers to generate, you know, picking instructions for a robot, so it could be any program essentially.
QUESTION: (Can'st hear, sorry)
CARL: Yeah, well the 400 figure was one I saw in some trade journal recently. They estimated there were about 400 and you are right, many of them were DNA companies, genomics companies that moved over, you know, in the past twelve months the large companies like Solera and some of the Insight have seen their stocks go from the $250 level down to $20 level and so many of the companies especially the smaller ones changed their spin and said they were really going to be more of a proteomics company. But there are a number of very exciting new companies that are coming along, companies that are looking at post-translational modifications and protein protein interactions, things like that.
ROMEO: Okay, maybe a last question. Yes please.
QUESTION: Rudy Abersol, Technology the iCad is a very exciting achievement I think regarding the part that xxx will be the only really robust technology in order to assess protein or composition and abundant. In that regard there will only be xxxx containing proteins that will be labeled so again, you will not be able to look at the proteom as it is. What is the percentage of xxxx containing proteins as opposed to those that don'st?
CARL: I don'st know. You would have to take a guess at that, you may have a better guess than I do. We think that the iCad technology is going to work alongside 2D arrays. One of the downfalls of iCad is that you cannot look at the post translational modifications like we can on the 2D arrays and so we think that you will use both in parallel to do very high throughput to find interesting things with iCad but then to go back and use 2D arrays to check other aspects that may not show up.
ROMEO: Okay, I think we can now close here the first part of the session of this morning. Your representation and also the question on patenting was an excellent bridge to the next presentation fully dedicated to this topic, Patenting in the human genome. The program now foresees allows a long break of about 50 minutes for coffee and a visit of the exhibition. Since we are a little bit delayed I would suggest that we cut it to forty minutes and we resume at 10.30 sharp. Thank you again Carl Foster for your presentation.