Most biologists know that the human genome, with its 100,000 or so genes, is expected to be fully sequenced by 2003, perhaps even earlier. It is less widely appreciated, however, that within the last few years, private companies have been scrambling to stake their claims to as many genes as possible before others can deposit the sequences in public databases. Although most of these data are not publicly available, it seems likely that patent applications have already been filed on the great majority of human genes.

This latter-day land grab has been made possible by a strategy pioneered by J. Craig Venter in the early 1990s, in which random cDNA clones are sequenced from a tissue of interest, generating a collection of expressed sequence tags (ESTs) that identify the genes transcribed by that tissue. Venter's strategy has been adopted by a number of biotech companies, who have been filing patent applications on a massive scale. Incyte Pharmaceuticals, for example, claims to have filed for 1.3 million ESTs, and two other companies, Human Genome Sciences (HGS) and Hyseq, also claim to have identified over 100,000 human genes, many of which they say are represented by full-length clones. The majority of these genes have no known function (although neuroscientists might note that about 50% are expressed in the brain); however, in many cases, the biochemical properties of the encoded proteins have been inferred by sequence homology, and in some recent applications, the cloned sequences have been used to express proteins and test their biological activity.

Most of these sequences, even those of unknown function, appear to be patentable, at least in the United States, where the US Patent and Trademark Office (USPTO) has taken the view that ESTs meet the criteria for patentability based on their novelty and their utility as probes. Of course, not all the sequences in the pending applications will be pursued, because patent prosecution is expensive and time-consuming. Companies will concentrate on those genes that seem most promising commercially. To some extent, these will be identified through in-house research programs; for example, HGS—according to its chairman William Haseltine—has now screened 14,000 human secreted proteins for biological activity. Yet even for companies that invest heavily in functional screening, decisions on which genes to pursue will inevitably take into account results published by other labs, who will in most cases be unaware of the applications already pending on 'their' genes. This has troubling implications.

In most countries, patent applications are automatically published 18 months after filing. In the US, however, applications remain secret until they are issued, which can take many years. This creates a loophole for the so-called 'submarine patent' strategy, in which an application is filed secretly and then kept pending until another company has invested heavily in the same technology. The patent is then brought to issue, and the investing company must either redesign its product or pay licensing fees to the patentee. Companies such as Incyte firmly deny pursuing any such strategy. Nevertheless, it remains the case that most sequence patents (particularly those for genes of unknown function) have been filed only in the US, allowing applicants to maintain secrecy as long as it is in their interests to do so.

Thus, those who are interested in commercializing gene-related technologies face an uncertain situation. Not only do they not know what other patents may be pending, it is also unclear how these patents will be interpreted. For instance, even though a patent may be awarded for a full-length gene, it might still be dominated by a prior patent claiming a partial sequence, whose assignee could then demand royalties for infringement. Similarly, a patent claiming a method for using a gene could be dominated by a patent claiming the sequence in the absence of functional data. These questions are ultimately for the courts to decide, but this could take many years; companies are unlikely to engage in expensive disputes over a patent until its value has been proven in the marketplace.

Does it matter if intellectual property rights to the human genome end up in the hands of an oligopoly of genomics companies? The USPTO has been criticized for its broad interpretation of the criteria for patentability of DNA sequences, but its director of biotechnology, John Doll, believes that the criticism is misplaced. He argues that patent law has worked well for other new technologies such as computers, which contain thousands of patented components. Doll foresees the emergence of networks of licensing agreements, in which market forces will ensure that valuable inventions are properly developed.

Not everyone is so sanguine. Both the NIH and the US National Academy of Sciences have expressed concerns that patenting genes of unknown function may inhibit the commercialization of publicly funded research. It may also affect academic research more directly. Although biotech companies have generally refrained from trying to restrict academic research, there have been exceptions. For instance, the NIH argued for several years with DuPont Corporation over the terms for using DuPont's Cre–lox technology for making tissue-specific gene knockouts. Although an agreement was eventually reached last year, there is no guarantee that future disputes will always end so amicably.

The impact may be greatest for research into diseases that are too rare to support billion-dollar blockbuster drugs. The cost of licensing and potential disputes can run to millions of dollars, and although it may be worth fighting for the rights to the biggest markets, the potential legal costs may prevent many useful applications from being brought to market. Meanwhile, one thing seems clear: the next few years should be a good time to be a patent lawyer.