Francis Collins

“We’re creating a lot of patent rights early on that are going to reach far downstream into product development, and are going to create obstacles to product development,” says Rebecca Eisenberg, a University of Michigan law professor and expert on biotech patent law. The filing of patent applications for expressed sequence tags (ESTs) is one thing that worries Eisenberg. At least three million of these DNA sequences are in the patent pipeline [See table]. Most will probably never see the light of day, but a great many are likely to receive patent protection. Will products using these sequences infringe the EST patents? Is this prospect already slowing or inhibiting drug and diagnostics development in biotech and in big pharma?


Drug companies aren’t talking. Signals requested interviews with Merck & Co. Inc., Glaxo Wellcome plc, Pfizer Inc., Warner-Lambert Co., Pharmacia & Upjohn Inc., Wyeth-Ayerst (the pharmaceutical division of American Home Products Corp.), Bristol-Myers Squibb Co. and Amgen Inc. All our queries were either ignored or rejected. Obviously, no company is eager to publicly admit that it faces such a far-reaching and fundamental problem. But comments made during a panel discussion at January's Chase H&Q Healthcare Conference shine at least a thin beam of light on the vast, dark and increasingly dangerous patent landscape.

“The analogy that I would use is that of a minefield,” said Bob Levy, senior VP of science and technology for American Home Products. “We are spending an incredible amount of time now, when we find exciting targets and begin to validate them, in trying to define who has rights to what. And we’re finding, in almost every product that we look at, that someone has patented the protein, the gene, a fragment, a diagnostic test.” Levy noted that untangling patent rights, and determining which patents are dominant, are increasingly time-consuming and expensive tasks. And patent-holders must be paid. “The royalties that will be involved soon in some of the products that we are bringing to market, they’re already up into the ten, fourteen, fifteen percent [range],” said Levy. “And that may increase with time.” (See also the Signals article, "Is The Alliance Deck Becoming 'Anti-Stacked' Against Innovators?" for further discussion of the impact of stacking royalties on biotech alliances with pharmaceutical companies.)

“We’re living in this era of ‘stacking royalties,’” agreed Joffre Baker, VP of research and discovery for Genentech Inc. “There are more patents than ever out there around processes, methods, various tricks. The royalties just keep on stacking up and up.” The problem will only grow as more genes and ESTs are patented. “Everybody’s fighting over the last scrap of real estate in terms of the remaining human genes,” said Baker.


To many, it seems wrong that DNA should be patentable at all, since no one invented it. Patenting DNA, it follows, is an act of arrogance and hubris; it confers ownership to something that should not be commodified; and it lets individuals and corporations own our common genetic heritage. None of these arguments, however, have made the slightest dent in U.S. patent law, which rests on the 1980 Supreme Court decision, Diamond v. Chakrabarty, which determined that genetically modified bacteria were patentable inventions.

Other court cases have battered the “product of nature” doctrine that holds that natural products aren’t patentable, so that today the U.S. Patent and Trademark Office (PTO) and the courts view even “natural” DNA as a chemical compound that, once isolated and purified, is patentable. “I don’t think anyone is going to challenge it [in the courts],” says University of Pennsylvania bioethicist Jon Merz. “That ship has sailed.”

To date the PTO has issued about 2,000 patents on full-length genes (all species included). But that number is not important. What’s crucial is the number of DNA sequences in patent applications, since the application date, not the issue date, determines who owns what. And that number runs into the millions, thanks to ESTs. So the DNA patent land grab is almost over, even though only a few patents have issued. The frontier is closing fast.

The PTO ruled that ESTs -- short sequences of coding DNA derived, assembly-line style, from messenger RNA extracted from cells -- were patentable in the early '90s. That led to a deluge of patent applications by genomics companies. Incyte Pharmaceuticals Inc., Human Genome Sciences Inc. and (later) Hyseq Inc. filed almost as fast as they could make and sequence their cDNA clones. “Their business model was, ‘file early and often,’” remarks Steve Holtzman, chief business officer for Millennium Pharmaceuticals Inc. Millennium and other companies tossed in their own ESTs for good measure. No one knows how many ESTs are in line waiting for patents -- not even the PTO. “We quit tracking them,” says John Doll, the PTO’s director of biotechnology examination. “We had about a half a million, and that was about four years ago.”

Critics of gene patents hope public sequence databases will pre-empt many of these claims, but that’s not likely to happen. The majority of EST applications were filed before much sequence data made it into the open databases of Merck and The Institute for Genomic Research (TIGR). And sequences placed in GenBank by the Human Genome Project are almost all coming too late to nullify EST patents.


Whole genes are also adding up. Human Genome Sciences has filed applications encompassing 7,500 full-length human genes, according to chairman and CEO Bill Haseltine. Incyte has at least 6,500 genes waiting. Celera Genomics, before it stopped releasing patent numbers, accounted for another 6,500 partial and complete genes. Hyseq has disclosed filings for over 5,000 full-length genes.

Bill Haseltine

French genomics company Genset SA is also a major player. As of last October, it had generated over 90,000 sequences of 5-prime untranslated region (UTR) sequence tags of human full-length cDNA clones -- each tag belonging to a different gene. Patent applications will be filed on all of them, if they’re not already. Although Genset CEO Pascal Brandys won’t disclose the number of gene patents his company has pending, documents filed with the U.S. Securities and Exchange Commission (SEC) state that “it is the company’s intention to file patent applications covering all such full-length cDNA sequences." Even without full-length genes, as long as possession of the 5-prime UTR sequence is enough to claim, or “dominate” (in patent law parlance) the gene, then Genset already has most of the 100,000-plus human genes accounted for in its patent applications.

Counting ESTs, patent applications now blanket the entire human genome many times over. “We’ve filed patents that describe…up to 140,000 human genes,” says Haseltine. “We’re reasonably confident we were the first to find 80 to 90 percent.” For its part, Incyte claims patent filings “covering over 50,000 unique genes.” Hyseq CEO Lewis Gruber says his company now has “30,000 novel genes” in the patent pipeline.

Incredibly, the same sequence can be patented more than once. It can be in an EST, and also in the full-length gene; each is patentable. PTO guidelines allow EST applicants, if they come first, to claim “dominance” over the full gene (implicitly, through use of legal language). But the final verdict on such broad EST claims will not be in for years. “That’s only going to be solved by the courts,” says the PTO's Doll.


Since only five EST patents have issued to date, the PTO has kept companies guessing on other issues. Most important: How much about EST function must be known in order to patent it? The PTO recently tightened up requirements for documenting such EST “utility.” (It’s no longer enough just to say an EST can be used to find a gene or help map a chromosome.) But how much information will be enough remains an open question. Doll says that about one-third of EST patent applications won’t meet the current guidelines. Still, that leaves over a million patentable ESTs.

That’s alarming, says University of Michigan's Eisenberg, especially if a drug company needs a license from the owner of an EST patent in order to use the full-length gene. Making and sequencing ESTs, she notes, is a trivial (and largely automated) lab exercise, but the patent system has been treating each EST as a separate “invention.” “That’s very problematic,” she says. “Because it seems like people are making the most trivial or no contribution to ‘the art,’ and at the same time claiming broad patent rights that will allow them to capture more than their share of future research by other people.”

Rebecca Eisenberg

Eisenberg offers this scenario: Somebody gets a patent on an EST; someone else patents the full-length gene; another party patents a variant in that gene; and then someone patents a diagnostic test to test for variants predisposing for disease. “Each of these people could have their own patents,” she points out. “And somebody who wants to bring that test to market will need licenses from all of them.” Eisenberg says the proliferation of early patent rights is leading to an “anticommons,” in which companies seeking to develop useful products will be discouraged by a proliferation of patents on gene fragments.

Even worse, the diagnostic test seller might not be aware of some of the patents until too late. In the U.S., patent applications remain secret until the patent is issued. (Europe publishes applications 18 months after submission.) “You can lay low, nobody knows that your patent application is pending, and then you come out of the water with an issued patent, after somebody else has figured out a use for your sequence, and you have somebody to charge royalties of,” says Eisenberg. Such “submarine patents” can lead to heavy licensing costs, infringement penalties, or even product withdrawal, so a conceivable outcome is that certain products simply won’t be developed unless all patent-holders are identified -- something hard to do with millions of ESTs lurking underwater.

Hyseq’s Gruber applies the submarine strategy. His company selectively revises some patent applications (thus delaying PTO review), while letting others go forward, based on others’ use of Hyseq’s DNA. “If we find someone using our ESTs, we can accelerate that process and get [our patents] to issue quicker,” he says.

Unethical? Gruber points out that Hyseq (which develops some of its own drugs) is itself vulnerable to submarining. “We have the same bad effect from that that everyone else does, because our competitors have submarine patents,” he says. So, any criticism, in his view, is misplaced. “[It's] a case of people blaming the patent applicant for something that has not been corrected in the system,” Gruber says. He suggests changing PTO rules to force quick publication of patent applications: “I’d be very happy if everybody put their cards on the table.”


Genomics company executives dismiss Eisenberg’s anticommons argument and drug companies’ fears of royalty stacking. “Anytime there is a complex intellectual property portfolio that has been developed, industry generally responds by setting up beneficial cross-licensing arrangements,” says Haseltine. “The electronics business is bedeviled by that problem. Yet it manages to survive quite well.”

The PTO’s John Doll is confident that biotech will thrive, not in spite of layers of DNA patents, but because of them. “A company wants to make money on [its] patents,” he says. “Licensing them, and selling them, for a reasonable royalty is the way they make money and the way they fund further research.” Doll points out that every computer maker uses hundreds of parts that require licenses, yet PC prices have dropped dramatically and sales are booming. “When we look back at… emerging technologies and breakthroughs in that technology, the licensing and cross-licensing has always happened in a rational manner,” he says.

John Doll

It’s true that genomics companies aren’t planning to hoard most genes. Instead, they’re promising to nonexclusively license DNA databases for reasonable royalties. “Our view is, get them into peoples’ hands broadly,” said Incyte CEO Randy Scott at the Chase H&Q conference. Once Human Genome Sciences’ exclusive partnership with a five-company consortium expires next year, “we plan a very aggressive licensing program to make sure that others use our intellectual property to create products,” says HGS’ Bill Haseltine. Celera Genomics has also promised nonexclusive licenses, except for a few hundred genes to be patented and exclusively licensed to pharmaceutical companies. “Nobody, it seems, especially us, wants to see broad ‘blocking’ patenting,” said Celera’s Craig Venter at a recent talk at the University of Michigan.

But offering broad database access (for drug discovery purposes) doesn’t address the royalty stacking problem for actual products. Incyte’s Randy Scott says he understands the drug companies’ complaints. "If the genomics industry doesn’t hear that and listen to it, and back off, to keep the royalties at a [reasonable] rate, then we’re just taking money out of our own pockets,” he said, since drug companies will simply avoid developing products based on inaccessible patents.

Rebecca Eisenberg doesn’t expect the biotech industry to come together amicably to widely cross-license patents and otherwise ease product development. “I haven’t seen any signs of that,” she says. “I mean, every biotechnology product that has come to market has been accompanied by litigation -- to the death.”

Biotech litigators have certainly thrived. Genentech sued Amgen over use of vectors for expressing cloned genes. Amgen sued Transkaryotic Therapeutics Inc. for infringing its erythropoietin patents. Hyseq and Affymetrix Inc. sued each other over DNA chip technology, and Affymetrix also sued Incyte and Synteni Inc. (which Incyte acquired in Jan. 1998). Human growth hormone has spawned several lawsuits, involving Genentech, Novo Nordisk and others. Insulin and t-PA have also been heavily litigated.

Hyseq's Gruber argues that the lawsuits will diminish as the industry matures. “There are over 1,500 biotech companies, and basically a lot of their value is based upon their intellectual property -- they don’t have product,” he says. “Any exacerbation of litigation here I think [is] coming from the fact that it’s still a relatively young industry.”

The optimists also point to PCR, the OncoMouse and cre-lox transgenic mice as examples of vital research tools that are widely available at reasonable rates. (See the Signals articles on the OncoMouse and cre-lox controversies for more details.) But the cre-lox issue “was [only] resolved after huge, high-level resources were brought to bear on it,” Eisenberg points out. “We’re more at risk not for the highly salient controversies that get written up in Science and get the attention of the director of the NIH, but the more mundane transfers that get bogged down in disputes over the terms of agreements…I haven’t seen any signs that the problem is abating.”


One possible future flash point is DNA chips. Lewis Gruber says chip companies should license Hyseq’s EST library for use as probes. “That way they’re protected when one of our patents issues,” he says. “[So] they don’t have to worry that they’re going to have a product on the market, only to suddenly infringe someone’s patent.” Some chip systems now include tens of thousands of probes, so Gruber’s scenario could be a nightmare for chip companies, since they’d have to track down the patent-holders (or future patent-holders) of at least some sequences and then pay royalties. Gruber says that certain important sequences might not even be available later if chip companies don’t act now.

That’s a red herring, contends Affymetrix general counsel Vern Noviel. Affymetrix expression chip probes are only 20 bases long, he points out, so it should be easy to work around patented ESTs. Hyseq “might have a…200-base snippet of EST out of a thousand-base gene,” says Noviel. “We can live with that. We just pick probes out of the other part.” Or, if necessary, take the gene off the chip. “We have 40,000 genes and ESTs on our chips right now, and it’s almost more than people can handle,” says Noviel. “And putting that next Hyseq gene on there -- give me a break. Incrementally, [it's] not much more valuable.” But no one knows for sure how EST patents will affect chip makers. It’s just one more uncertainty that companies must consider as they move forward with product development.

..	Incyte Pharmaceuticals	Human Genome Sciences	Hyseq	Celera Genomics	Genset	Millennium Pharmaceuticals
ESTs Filed: Issued:	1.2M 1	Undisclosed1 None	0.9M None	None None	90,0002 None	Undisclosed 1
Full genes Filed: Issued:	6,500+ 490	7,500+ 112	5,000+ None	6,500+3 None	2,500+4 50+	Undisclosed 50 (approx.)

Footnotes:
1. Total ESTs generated: 2.5M (approx.)
2. 5' UTR ESTs (approx.). Other ESTs: Undisclosed
3. As of 10/20/99. Described as "new gene fragments. Included…are receptors, ion channels and secreted proteins."
4. Full-length cDNA

The market for SNP genotyping chips, for example, could explode if and when individual genetic profiling catches on. But SNPs, like ESTs, are patentable. (The PTO says that about twenty patents on SNPs have been issued to date.) Will SNP chip makers have to worry about licensing their SNP probes? Not to worry, says Noviel. “Let’s assume that one or two of them one day would end up being patented,” he says. “The amount of effort required to take those off [the chip] is less than…it is to figure it out.” Nevertheless, Affymetrix is taking the precaution of filing its own SNP patents -- between ten and twenty thousand so far, says Noviel. “We send them in [to the PTO] in boxes,” he says. “Literally we’re trying to figure out how to get these patents filed on CD-ROMs, because they’re so big.”

The prospect of SNP patents led ten major pharmaceutical companies to create the SNP Consortium in April, 1999. The Consortium’s goal is to place 300,000 SNPs, evenly spaced throughout the genome, in the public domain. (To date, about 10,000 have been released.) That’s to facilitate whole-genome disease gene association studies, considered the key to unlocking the genetic roots of complex diseases like diabetes, heart disease and schizophrenia.

But SNPs have other, more directly lucrative uses, including disease diagnosis and genetic profiling for pharmacogenomics. “Annotated” SNPs -- those known to be markers for disease, drug response, or important traits -- are likely to be far more valuable than the random SNPs used in association studies. “It’s probably more important to have detailed data on 50,000 SNPs, rather than generate half a million,” says Genset's Brandys.

It was Genset’s SNP discovery partnership with Abbott Laboratories that first alarmed other drug companies and led to the formation of the SNP consortium. Brandys won’t say how many SNPs Genset has filed patent applications on, but the eventual total is likely to be large. “For every major patent issued on genes, we expect to have several SNPs involved,” he says. As for random “generic” SNPs, “we’re still trying to devise the most appropriate strategy on “generic” SNPs,” Brandys says. “It’s something that’s not completely settled in-house, to tell you the truth.”

Some genomics companies aren’t bothering to patent SNPs at all. “Patenting SNPs is like patenting grains of sand on the beach,” says Haseltine. “There are so many potentially useful SNPs…I would guess in excess of several hundred million, that any one collection is likely to be gotten around by any other collection.”


Simple economics should limit the number of SNP and EST patent applications that companies actually proceed with. “From soup to nuts, to get a U.S. patent issued [takes] ten to fifteen thousand dollars,” says Millennium’s Holtzman. That amount grows with the addition of other countries. According to Holtzman, a single patent, issued in the 10 major market countries, would cost over $200,000. In 1996 the PTO limited the number of ESTs per review to 10, and Doll now says that his examiners will -- in most cases -- only look at one at a time. That means a separate $1,000 application fee for each one. So companies will have to be selective, and only proceed with the most valuable applications.

Because of money and because the PTO is tightening up the utility requirement, argues Millennium's chief patent counsel Mark Boshar, genomics companies that have focused on quality patents as opposed to quantity will emerge as the ultimate winners in the paper chase. “We never did put all our eggs in the gene discovery basket of ‘turning the crank, let’s discover genes for discovery’s sake,’” says Boshar. Millennium has focused on specific disease pathways, patenting genes along the way, including detailed functional information on their protein products and, when possible, their specific therapeutic use. “Those who have just pursued a gene discovery approach, with the help of bioinformatics and computers, may find themselves with their IP strategy somewhat at risk,” says Boshar. “We’ve got a patent strategy, particularly on the use side, that’s ironclad.”

Other genomics companies don’t concede that their patent applications are inferior. HGS' Haseltine, for example, boasts of testing his gene products in 130 cell types, measuring 200 different parameters, and thus generating detailed functional information. But Millennium’s approach -- focus on a disease and identify its genetic causes -- seems more likely to yield detailed, nuanced biological knowledge than Human Genome Sciences’ automated screening approach.

Still, it’s very possible that quantity will ultimately be rewarded at the expense of quality. On Feb. 15, 2000, the PTO issued a patent to Human Genome Sciences for the CCR5 gene, which was found in 1996 to be a co-receptor for HIV entry into T cells. Human Genome Sciences did not know this when it filed in 1995, but the company can now demand royalties on AIDS drugs that target this receptor.