Mr. Wei-Wu He: So, thank you very much. So, I think, you know, today, I will just make it relatively informal. So, for a company like OriGene come to a pathology conference and telling everybody that, okay, we're going to launch a--the most specific antibody in the pathology market, you're probably asking me, you know, "Who are you to tell me? This is an old industry."
So, and you know, is it truly our position to, you know, to claim that? And I hope I can convince you one way or another that we have some unique technologies that we can make a more specific antibody. Okay? So, that--if the take-home message is that, I think I have done my job. Okay.
So, I actually came from the genomic field. So, after I'd done my Ph.D. post-start, I joined, you know, one of the first companies started by Craig Venter called Human Genome Sciences and in 1993. And this is a time that the U.S. scientist was just starting to decode the human genome.
And I always thought, you know, by going into this field, you know, the genome is our body's--I call it the operating system of our body. So, personally, I'm a big believer that our genomic information really is eventually going to transform how we look at diseases or disease management.
And OriGene is a company kind of started with the thesis that, you know, the genome--the decoding of genome will create a huge wave of technology innovation. And these innovation can eventually be applied to every aspect of human healthcare. And today, we would like to share with you some of the technology we're developing around the antibody validation area.
So, the genomic, you know, field, you know, so, we now know since year 2001, we finally decoded our entire genome. And we know that in our whole human body now have about 20,000 protein coding genetic units. Like, say, insulin is one genetic unit or, you know, BIC1 [sp]is one genetic unit. So, we have about 20,000.
And this is really the operating system. The protein is what I call the application software of our body. And so, in--and these are the, you know, foundations of a lot of medical innovations. For instance, protein can be made into therapeutic proteins, or we can make antibody, like Avastins against our protein. And we can develop specific small-molecule drugs to target, you know, specific cancers, or we can develop diagnostic tests.
So, the--but, historically, we human being, the society always develop a product one gene at a time. You know, Canada is a wonderful place for the, you know, discovery of insulin. So, that's one--to one protein. Well, that protein was discovered first. Then eventually, the gene was cloned. But, now, with the genome is decoded, we now know there's 20-some-thousand protein coding gene.
So, how do we change the paradigm? And after the genome is decoded, how do we take the whole genome to help develop products to better manage diseases. That's really what OriGene is all about. And we're very much focused on developing the toolbox for this field.
And tools are very important. So, I gave a few examples, you know, Excrit [sp], you know, PCR machines. I think this is a peptide synthesizer, DNA sequencing machine, patch clamping. So, many of the innovation in the biology is driven by the innovation of toolbox.
So, if you look at the genome--you know, genome project, you know, the first genome cost $1 billion to decode. Right now, $2,000, so you can have your sequence genome decoded, personal genome.
So, you know, we have come a long way in genomic technology innovation. So, and so--and OriGene has continued to push the frontier of this field. So, OriGene's focus, a little bit brief introduction is we're started in 1996. We have been around for now 16 years. Time goes fast. I thought it was yesterday.
So, we have about 400 employees. Our headquarter is in Rockville, which is, you know, the head--you know, where the entire human genome was decoded.
We have actually a subsidiary in Seattle called Blue Heron, which we actually do synthetic biologies. We actually synthesized the backbone for Craig Venter. The first synthetic live organism is actually done by us. And we have a quite large organization in China. We do monoclonal antibody production in Beijing, Wuxi. And we also recently merged with a pathology company in China called ZSBio. They are actually the oldest pathology company in China. It's been there for 20 years, and we distribute a lot of the global reagents for the pathologists in China.
So, actually, OriGene's long-term goal is to develop assays for every single protein in the human proteome. So, if you think about company like Illumina of the world, they have--Affymetrix of the world, they have revolutionized how we detect the DNA and RNA.
But, today, if I challenge you and say I give you a tube of my serum and say, "Go help me detect 10,000 protein in my serum," what is the technology you're going to use. So, the technology's not quite there yet when it comes to detect the proteome.
So, we actually are pushing this in two frontier. One is we're actually developing mass spec standard for every single human protein. This is a collaboration we did with Leroy Hood at ISB in Seattle. It's funded by the government of Luxembourg. So, we just contributed almost 5,000 purified human protein to Lee. And they're generating mass spec standard on every single one of them. So, that's one direction we're going.
Another direction is to make perfect antibody for every single human protein. So, and today, since this is a pathology conference, we really want to share with you our effort in the pathology side.
So, our unique capacity is, one, is we dedicated 16 years to physically isolated a open reading frame clone for every single human protein. We are the largest collector of open reading frame clones in the whole world. Okay. So, this is actually not a easy project. NCI already spent $500 million. And so far, they only collected about 13,000 human full-length cDNA clone. So, this is a very important--this is really the operating system of our genome.
So, we have been working on it for 16 years. We're still about 2,000 loci away from the entire human genome collection. So, it will take us probably--that's one of the reason we acquired a synthetic biology company. So, because some of the genes we're trying to clone, they're so rare in human tissue. You can't clone them from the tissue anymore. We basically have to use the synthetic route to make those open reading frame clone and to make a protein.
So, we have a tremendous capacity here. We have the largest clone collection in the world. We have about 18,000 genetic loci covered. And we have acquired a company called Blue Heron, which we know how to synthesize 1 million base pairs. What's relevant here is, in order to develop assays for every single human protein, you literally have to purify every single protein. So, we have now already produced 15,000 transpected antigen. I'll, you know, elaborate a little bit more later.
But, we have already now purified 6,000 human full-length protein from mammalian cell, HEK293 cells. So, this is actually one of the reasons we can actually do high-throughput antibody production because we're already--cloning and protein production is already [unintelligible] cost for us.
So, we think by end of this year, we will--we should have 8,000 human protein fully purified from the entire human genome. And we acquire--you know, we were very prepared to get into the tissue business. We actually acquired a large tissue bank. You know, it's formally called Adeus [sp]. Later on, it was called Cytomyx. It's a tissue bank, which they spent about $70 million building. And it was maybe the tissue from Duke and Harvard and University of Chicago. And we have about 30,000 cases of cancer patient. We can even track outcome data for patients.
We actually have a very broad Luminex license from Luminex, which--so, eventually, our long-term goal is to put all these antibodies on pairs so we can eventually do multiplex pair. But, today, I will share with you our antibody production and the 10K chip.
So, the question is, how do you develop a perfect IHC antibody? So, what is the perfect IHC antibody? So, we--so, what we actually find out ever--three years ago, we--you know, we decided, you know, this is time for us to look into this field. So, we decided gear up our antibody production. We have always been watching the antibody space.
You know, so, and--you know, I was, you know, fortunate, you know, to involved in a company called Apatomix [sp]. You know, it's my colleague at HGS [sp] started a rabbit monoclonal antibody company. And we--personally, I actually invested, seeded the company early on.
And so, we have been always looking for various kind of technology to make monoclonal antibody for every single human gene. But, the cost and the quality just never reached to an area that we can truly do it. And we think now we finally may have reached to a point that we can do industrialization, like Henry Ford way of making it. So, and I'll share with you why.
So, in IHC area, in the end, quality is everything. If you don't have quality, you know, don't even bother to make an antibody because if you have a dirty antibody, you might as well not make one. Okay. So, that's our philosophy. So, if we don't have the right tool to assay the antibody, we don't even want to make it.
So, it turned out to be that, now, we have a pretty good toolbox. One of the toolbox is we can actually use our lysate [sp] as a antigen standard because, when we transpect a full-length cDNA clone into HEK293 cell, because these clone has a tag on it, we can quantitate how much antigen is in each--you know, we can literally do a titration of our antigen and immediately know which vendors p53 antibody is the most sensitive antibody because this assay vendor has the best most-sensitive p53 antibody.
But, some of the antibodies are pretty bad. You know, so, you know, this antibody has--you know, is not very sensitive, can't detect 2 nanograms of p53. But, even when they can detect it, they see multiple bands of p53.
So, even on detection side, there's tremendous amount of variability in the marketplace. But, only specificity side, it's even more challenging. So, because--if you think about IHC application, our interesting IHC application is the most challenging application because, if you think about a antibody pair, you know, because you're using two independent antibody, so your likelihood of getting specificity is higher because, if you made a rabbit antibody, a mouse antibody, the likelihood of those two antibody will hybridize against the same noise is low.
But, IHC, you're using one antibody. And your tissue is a --full of human protein, especially you're trying to detect cancer tissue. Every--you know, every single cancer tissue, the gene expression level is different. So, you have even more a daunting challenge. So, to me, actually, an IHC antibody specificity has to be the number one criteria of--you know, for a antibody. It's like you're going to fish to a dirty pond to try to catch a particular fish. Your fishhook better be extremely specific. Otherwise, you're going to pull out all kinds of stuff.
And then you're going to write a pathology report to the patients and tell them what to do. And that is very--to me, it's very nervous if I'm that--so, we think actually this antibody here got to be the most specific probe on Earth.
And so, how do we do that? And this is actually what we'd like to share with you, this after all these years, 16 year worth of work, we have now produced a antigen chip. Okay. So, we basically take our 16 year worth of work. We took all our open reading frame clones. We transpect them into HEK293 cells. So, this is tens and thousands of transpection. Okay?
And then we validate every single one of them by doing Western blot because all these clone has a tag on it. It has a mick [sp] and flag tag. So, we can know this gene has been expressed.
Okay. Then we can do different extraction. We can, you know, do it. And then we--in this particular case, we take the whole lysate. We don't purify the protein, okay, because a lot of membrane protein, like GPCR out of water, is very difficult to purify.
Actually, in the context of the lysate is actually--it's actually--it's pretty easy to isolate. So, then you just spot each dot is basically an over-expressed lysate antigen. Okay? So, this chip has 10,000 human antigens on it. It's about half of the whole human proteome. Okay?
So, now, with this, too, finally, when you have an antibody, you can say, well, you know, I can check entire half of the whole proteome to see whether they're specific or not. Okay. That's the message.
And eventually, we would like to have the entire human proteome on it. I think this year, this end of this year, we are planning to print a 15K chip. So, we finally now have already produced 15,000 lysate. So, we can produce a 15K chip. But, the last 2,000 or 3,000 is always going to be the most difficult. So, it's--that's the name of the game.
So, even the cDNA clone, we have 2,000 genes today. The whole world doesn't know how to clone 2,000 genes in our genome. You know, like there's a gene called titin. It's 100 kb. You know, I don't know how to clone it. We need to invent a whole new factor just for that clone alone.
So, this is actually the lysate chip, 10K chip, printed. And so, we're using this manually to assay antibody specificity. And so, this--any antibody--well, antibody can pass our 10K chip specificity, now we are calling these antibody UltraMAB, stands for ultra specific basically. Okay?
So, one of the investment we made in the last three years is to build a large-scale antibody production facility in China because antibody production, unfortunately, is extremely labor intensive. It's a numbers game. If you want to have a good high-quality antibody, the best thing to do is to have lots of hybridoma to pick from.
For instance, we are actually investing in to make a perfect ALK antibody. It turned out to be ALK antibody's pretty tough. So, we went ahead and made 200 hybridoma already. Okay. We immunized like, you know, more than a dozen animals. And we made six different antigens. We made a huge number of hybridomas. So, this way, we can have a big pile to screen for the perfect one.
And unfortunately, this whole process is extremely labor intensive. So, we actually decided to do all this in China. So, we now are actually releasing about 200 antibodies per month right now, OriGene is. Okay. So, we have only been in this game for two years. We have now released over 4,000 antibody, not all UltraMAB. So, UltraMAB is still much high--more difficult.
So, we basically using high throughput. And we do a vigorous testing. But, for UltraMAB, really, there's two criteria for UltraMAB. One is they have to pass our microarray specificity test. And we have--the pathologist we use has to like the sensitivity of our antibody. That's the two main criteria.
And I'll share with you a few example why UltraMAB is so important. So, this is actually a case we did with a antibody called ERCC1. And I think in this audience, everybody probably know what ERCC1 is. It's basically--it's one of the DNA repair enzyme. So, if ERCC1--the thesis is, if ERCC1 is overexpressed, when cisplatin is used and creates the damage to the DNA, the tumor cell knows how to repair very quickly. So, therefore, the tumor cell is resistant to cisplatin treatment.
So, if the patient has high level of ERCC1, they are not candidate, not good candidate for cisplatin treatment. Okay. That's what I understand. Okay.
So, this is the New England Journal of Medicine paper looking at ERCC1-negative tumor with drug treatment, without treatment. There's definitely a difference. But, if the tumor is positive, it's less of a difference I think. That's the thesis.
You know, I heard--some people don't believe this, but, you know, this is at least the New England Journal of Medicine paper. So, this is not my job to convince you whether ERCC1 is a marker today. But, I'll convince you that the ERCC1 used to publish New England Journal of Medicine is not a good antibody.
So, there's actually a group in University of Pennsylvania basically first published a paper saying that, you know, that antibody is not specific. So, how do you know the data is good? So, in their cancer research paper in 2009, so this is--is this better? Oh, here we go. Maybe this is much bigger, more powerful, yeah.
So, this actually--this is a saw line. They actually knocked out ERCC1. So, if they knocked out ERCC1, technically, you shouldn't see anything at all from ERCC1 antibody. But, they still see a band. It's about the same size from this clone called 8F1.
Okay. So, what is this band? They obviously don't know. Actually, as a matter of fact, if you read the criticism of that paper by the French group is that they should do mass spec on this band to find out if they are. But, you know, doing mass spec is not that easy. Okay?
So, what we did is we immediate took this antibody and hybridized with our 10K chip. And of course--unfortunately, this is hard to read. But, of course, this antibody binds to ERCC1. But, it binds to another gene called a PCYT1A. I don't even know what that gene is, so obviously--so, it's very strong signal. Okay?
So, if you actually run a Western and it's clearly this hybridoma, ERCC1 not only binds to ERCC1, but it'll also bind to this protein called a PCYT1A. So, without the protein chip, it will take forever to find out what it binds to. But, took--this takes us one day. We can actually find out what it binds to. And it--because we already--usually, by the time you see a signal, we already have the lysate, right? So, we just run a Western blot to validate it.
Yeah, so, this antibody, this hybridoma, people--pathologists are using indeed, at least on our chip, binds to two antigens, okay, ERCC1 and a gene called PCYT1A. Okay?
So, and another paper even recommended a rabbit polyclonal antibody called FL-297. On our chip, this antibody is even worse. Okay? It binds to at least three different proteins. It binds to a protein called PDPK1. It binds to another protein called FERMT3. And of course, it binds to ERCC1.
So, the question is, when you have these dirty probes, how much confusing--confusion is created in the field? Okay. So, this is actually just to show you this is a rabbit polyclonal antibody. It cross-reacts with this protein called FERMT1.
So, it turned out to be this cross-reacting element is actually a nuclear membrane protein. And ERCC1 is a nuclear protein. So, we think it should interfere with the signal. Okay? So, it--so, you know, so, this is actually an antibody against PCTY1. And you can see it's a--it has similar pattern. My understand--I'm not a pathologist. Okay. So, if I'm making false statement, correct me. Okay?
This is PCYT1A. And this is 8F1. And apparently, they have the same nuclear localization. So, pathologists always like to say, okay, "If they--if p53 is supposed to be in the nucleus, and I see nucleus staining, you know, the antibody may be good." Okay?
So, to me, that's not good because, at least when we do Western blot, we separate the protein pretty far apart. But, when you look at the cellular compartment, every compartment still have thousands of protein in it, okay, so doesn't really--you know, being in the right location is really no mean--by no mean it's the right antigen. Okay?
So, here's an example. So, if you do qPCR, we want to find out, okay, so this--let's say this protein will be detected by 8F1, a commonly antibody by pathologists. Would that create problem for cancer? So, we just look at cancer patient. It turned out to be some of the cancer patient expressed a lot more PCYT1 than actually ERCC1.
See, look at this patient. The ERCC1 level is very low. But, the PCYT1A is extremely high. So, for this particular patient, whatever the ERCC1 positive pathology is writing, I think it's mostly probably coming from the signal from PCYT1, not from ERCC1.
So, we think this is probably one of the biggest challenge of IHC antibody because, if you don't know how clean that antibody is and because cancer is so heterogeneous, how do you know that particular patient didn't overexpress another gene in this case that actually is the contaminant you are detecting? Okay?
So, that's why we think the UltraMAB is a--you know, we think it's important enough. That's why we decided to dedicate tremendous amount of energy in this area.
So, we decided to, you know, try our strategy to make the most specific antibody for ERCC1. So, and this just--we went ahead, immunized huge number of animals. We eventually got 12 IF positive, nine IHC positive antibody. And we got a 15 flow positive antibody. But, out of all these antibody, only two antibody actually can pass the test of being specific on our 10K chip and then the pathologist likes the reading. Okay?
And it turned out to be there's only one antibody. It's called a 4F9, and another antibody is--so, this is actually the 4F9 protein chip. It's very specific. We couldn't see any--because each dot, you can digitalize it. So, you can actually have a real reading for each dot. So, you can definitely--you can't miss it when you see a contaminant showing up on our 10K chip. Okay?
So, and this is our 4F9. We call this one would be an UltraMAB. And it clearly don't cross-react with this gene PCYT1A like 8F1. And this is the two hybridoma we have. One is called 4F9; one is called 2E12, and passed our UltraMAB. And luckily, this one now has already been licensed, one of the large anti--you know, pathology company.
So, these are actually now slowly coming to the market through different channels. We actually--since we have, you know, our ZSBio channel in China, so the Chinese pathologists are already using these antibody. So, I always challenge the U.S. colleagues and say, "If the Chinese can actually use better quality antibody, I think you need to do the same for the American patients, too, or Canadian patients, too, because you--we should never use dirty antibodies for our patients." Okay?
So, ERBB-2, we actually have done quite a few antibody. We actually picked out a few to showcase to you that what we actually realized, I'm making a very provocative statement. We actually find about 50 percent of IHC antibody are not specific, 50, more than 50 actually.
So, here's--you know, we--actually, we show these data not to make other people mad because we really just want to share with the field that, with this new innovative technology, we finally have a test for specificity. Okay?
So, here is the commonly used FDA-approved 4B5 monoclonal antibody for HER2. Okay? And when you look at our chip, it binds to three genes. It clearly binds to HER2. It binds to HER4. And it binds to another gene called ZSCAN18. And this is FDA approved. Okay?
And it can be easily validated by Western blot. And this clone 4B5 binds to HER2, HER4, and ZSCA18. We actually thing that this is why--because HER4 is highly expressed in some of the gastric cancer patients. So, if you have an antibody binds to both HER2 and HER4, if one patient has extremely high HER4, you're going to give them a HER2-3-plus reading at IHC. Okay?
So, that's why some of IHC data don't correlate with fish data because, at the protein level, you are really seeing these three proteins, right? But, at the DNA level, you're not seeing all these three gene amplified. So, this is actually just showing you that, you know, specificity is such a important, you know, criteria in IHC area.
And this is another--this is actually a polyclonal antibody because there's actually a group of people make--you know, this polyclonal, again, binds to multiple targets, at least three in this case. This is already considered to be a very, very high quality polyclonal. It's affinity purified and everything. And it still binds to three targets in the whole human proteome.
I'm actually hoping that, by the time we made another 10K chip, you know, we don't find some--even our UltraMAB are seeing some strange targets. I hope that's not the case. Okay? So, but, 10K chip is the best we have right now.
And so, we actually now have a HER2 candidate which is very specific. But, unfortunately, we don't even know if the pathologists like this antibody yet. It's being tested right now. We're hoping this antibody will be loved by the pathologist. Okay. So, this will be our HER2 test, okay, so the UltraMAB.
So, I think, in summary, we basically looked at the antibody on the market. And significant number of IHC antibodies are not specific when you are testing them against 10,000 antigens.
Okay. We basically--we have developed a tool, a toolbox, which is basically a 10K--half of the whole human proteome antigen chip. Now, we can use this chip to kind have--at least have act--you know, to have the first check how specific the antibody is.
And so, we are now committed to launch a whole new set of antibody for the IHC market. Basically, they have to pass the 10,000 antigen standard. And it has to meet the pathology's sensitivity test. And right now, we already released about 10 UltraMAB. And this year, our goal is to release about 100 of them because it's actually a lot of work to get one UltraMAB right because we actually--even some of our monoclonal antibodies, we thought it was pretty good. But, the moment you look at 10K chip, they bind to four, five antigens.
So, monoclonal antibody by definition, people think monoclonal antibodies are monospecific. It turned out to be 80 percent of monoclonal antibody are not monospecific. Okay? So, you need to do a lot of selection to find that 10 percent of antibody which truly are monospecific.
Okay. So, we actually used full-length protein to do immunization. So, we have a pretty wide spectrum to pick the right epitope. If you just use the peptide--I actually think one of the problem of these HER2 antibodies is they're made by a peptide antibody. So, because they have a very--you force the animal to only react to that epitope. Okay. So, you don't have much choice.
When you use full-length protein, you at least have a much bigger library to pick from. That's our theory is. Okay. So, this is basically--and we are actually, you know, initiating a monoclonal antibody co-development program. So, if anybody's interested in a monoclonal antibody, UltraMAB, for instance, ALK, we are actually developing it already. And we will be developing different kinds--you know, my theory in the genomic field is, eventually, we will have 1,000 drugs like Zarcory [sp], Erisa [sp].
And you know, because to me, the cure of cancer, the best cure is prevention, like HPV vaccine or HBV vaccine. The second best cure is early screening. You know, we're actually using this chip to do autoantibody screening working with Dr. Bass at M.D. Anderson developing a screening test for ovarian cancer.
But, the last result is this targeted therapeutics. This is what I call laser-targeted missiles. But, if you want to use laser-targeted missile, you better have a good GPS system because, if you don't have a good GPS system, your missile is going to shoot all over the place.
So, the IHC antibody or the fish testing are--in my opinion is just a GPS system. And your GPS system, the more position your GPS system is, the better care we can provide to the patient. Okay? So, if your antibody's dirty, a HER4 positive gastric cancer, you tell them that you're HER2 positive, that's not good because you're basic giving people a very expensive drug with no effect.
So, we're very much interested in working with everybody to eventually at least upper-grade all the primary antibody to these very, very specific antibody. We are very committed to launch a 15,000-antigen chip by end of this year. So, most of the antibody will be checked against two-third of human proteome.
You know, we think, if after we can check two-third of human proteome, if that antibody doesn't bind anything beyond two-third of the proteome, that antibody might be pretty specific.
Now, actually, let me, you know, share with you, actually the chip we developed, most of the drug target was preferentially selected. So, we always work on so called the drugable genome first, GPCR, IN channel, tyrosine kinase receptor.
So, that 10K chip has most of the drugable targets in there. So, in the sense this chip is already pretty good when it comes to develop a pharmaceutical, it's not necessarily the most comprehensive human proteome tool. But, for drug target, it's already pretty good because we always clone the drug target first, GPCR, IN channel, other word.
So, this is, you know, our effort to--you know, to contribute to this field. And we really are looking forward to work with everyone. So, if we show some of your antibody, some of the, you know, antibody is not specific, it's not personal. We're not trying to make anybody look bad. We just really want to say, you know, maybe the cancer patient do deserve a better HER2 antibody.
Okay. So, and we hope that we can all work together to provide the patient with the most specific antibody on Earth. Thank you very much.