Resource Sharing in Biomedical Research (1996)

The last of the six case studies examined is neither a repository of scientific community property, like the American Type Culture Collection, nor a center for visiting scientists, like the Cornell synchrotron. The Human Genome Center at Lawrence Livermore National Laboratory in fact is not technically a shared resource at all, but a federally owned, contractor-operated research and development laboratory that has become, by default, a supplier of valuable materials and information to the international scientific community without specific funding to do so. The resources shared by Livermore fall into several categories described below.

Lawrence Livermore National Laboratory (LLNL) is one of several Department of Energy (DOE) national laboratories. Located about 50 miles southeast of San Francisco in the Livermore Valley, it is operated for DOE by the University of California. Founded in 1952 by E.O. Lawrence as a second nuclear weapons laboratory (joining Los Alamos), LLNL now applies its skills in high-performance computing, advanced engineering, and management of large research and development projects, to a broad range of the nation's technological challenges. The laboratory has 12 scientific or engineering directorates; the Biology and Biotechnology Research Program, which houses the Human Genome Center, is one of those 12 directorates.

When it began in 1963 as the Biology and Environmental Research Program, its initial task was to study the dose to man of isotopes in the

environment as a result of fallout from weapons testing, and the immunologic and genetic consequences of radiation. By 1992, the program was doing little radiation work but was heavily involved in studying the mechanisms of genetic damage from toxic chemicals and disease. A reorganization yielded the Biology and Biotechnology Research Directorate, with an increased emphasis on biotechnology and structural biology, while environmental studies joined similar efforts in the Environmental Programs Directorate.

The Biology and Biotechnology Research Program has an annual funding level of approximately $30 million, primarily from peer-reviewed research grants. About half of that funding is through Department of Energy grants; the other half is from various sources, including the National Institutes of Health (NIH) and other federal agencies, as well as industry. Its activities fall under four programs:

1. Health Effects assesses exposure to toxic agents, carcinogens, and mutagens for a wide variety of sources; also studies DNA repair, the genetics of cancer susceptibility, and biodosimetry.
2. Health Care applies Livermore-developed technology to disease detection and treatment.
3. Structure-Function Analysis investigates proteins and other molecules responsible for maintaining the integrity of the human genome.
4. Genomics develops recombinant DNA clones, DNA mapping and sequencing techniques, and instrumentation and informatics tools to characterize the genes of microorganisms, animals, and humans. The LLNL Human Genome Center is the focus of this activity.

DOE has established genome centers at three sites, Los Alamos National Laboratory (LANL), Lawrence Berkeley National Laboratory (LBNL), and Lawrence Livermore National Laboratory (LLNL). Together with 18 multidisciplinary Genome Science and Technology Centers supported by the NIH National Center of Human Genome Research, they form the backbone of the Human Genome Project. The Human Genome Center at LLNL was established in 1990 as an outgrowth of ongoing work on DNA repair genes, specifically on chromosome 19. A multidisciplinary team of chemists, biologists, physicists, mathematicians, engineers, and computer scientists, the center is organized into four broad areas: Resources, Physical Mapping, DNA Sequencing, and Enabling Technologies. Each area consists of multiple projects led by a principal investigator. Together they labor at three

tasks—creating biological resources useful for genomic research, developing instrumentation and informatics for genome research, and locating genes.

The Resources Group provides cloned DNAs, synthetic DNA oligomers, DNA library production, DNA fingerprinting, restriction mapping, and fluorescence in situ hybridization (FISH) for the center, as well as maintaining all relevant data in an easily accessible database.

The Enabling Technologies Group provides center biologists with support in genome informatics, covering data acquisition, storage, integration, and display; with statistical and mathematical expertise critical to mapping and sequencing; and with instrumentation in such areas as fluorescence, electrophoresis, and high- throughput sample handling.

The Physical Mapping Group localizes biologically interesting features (genetic markers, genes, and regions conserved between species) of the genome, with special emphasis on human chromosome 19.

The Sequencing Group works in three areas: (1) highly accurate "finished" sequencing of genomic DNA containing genes of interest; (2) support sequencing for other researchers, including partial and full-length cDNA sequencing and generation of sequence tagged sites (STS); and (3) sequencing of large sections of chromosome 19.

Livermore chose chromosome 19 as its initial target and has completed an integrated metric physical map of human chromosome 19 that spans 95 percent of the euchromatin. The center has also developed and applied new biochemical and mathematical approaches for constructing ordered clone maps and has sequenced several DNA repair genes in humans and rodent species. Impressive as the center's own research has been, it is the unintended consequences of such success that are most relevant to this report. The center's success has produced a demand for materials and information from other scientists that is far beyond what center staff and budget can meet. The following sections describe the center's experiences in attempting to share locally developed instrumentation and technology, information, and biological materials with the wider scientific community.

Instrumentation and technology is the area in which LLNL and center policies and experiences have been most conventional and most "successful." Instrumentation with commercial potential is basically kept proprietary until an invention disclosure or patent application is filed. Recent examples include a miniature diode laser-based miniflow flow system and a battery-operated portable PCR (polymerase chain reaction) device. In September 1995, a patent was awarded to the University of California for a technology developed by center scientists called "chromosome painting," which uses FISH to stain

specific locations on chromosomes. The center hopes to make this technology widely available through a nonexclusive sublicensing program.

The center also develops noncommercial instruments, for example, devices that do not appear to have commercial potential because they are either modifications of something that already exists or simply because the center did not think potential demand justified a proprietary approach. Access to various instruments is given to external users on an as-available basis, even though the center is not technically a user facility, on the condition that users provide their own people and their own reagents. Center projects are always the first priority however, and outside laboratories often ask to buy a replica of the instrument instead. When this happens, such instruments are usually provided at cost or requestors are given the designs and allowed to build their own. One example is the LLNL high-speed flow sorter. One of the other national laboratories provided the funds for the center to build a replica of its high-speed flow sorter. At the same time, the center was approached by a commercial firm hoping to market the instrument and, in fact, agreed to a limited exclusive license for that instrument.

This approach is not limited to instruments of modest cost or to U.S. laboratories. A private foundation in the United Kingdom has asked LLNL for help in building a smaller version of a very large, very expensive biological accelerator mass spectrometer currently in use at Livermore.

The center maintains an extensive database for chromosome 19 information and transfers data monthly to the Genome Data Base at Johns Hopkins. Many of those data are available to anyone with access to the Internet. The center home page includes a map of chromosome 19 that is updated every six months. This map does not reflect all information on hand, even at the moment of updating, partly because of lack of staff and partly because of obligations to collaborators. It is important to note that the chromosome database and map include not just data gathered by center scientists at LLNL, but data from scientists working on chromosome 19 genes all over the world. It is a big advantage, probably a necessity, for all involved to have their own data put into context in a single database and map. This is the primary incentive for these scientists to forward their data and/or clones at an early stage. On the other hand, they are not always anxious for potential competitors to have the fruits of their labor too early in the process. The Human Genome Project is dependent upon, and therefore insists upon, sharing however, and the LLNL center's general rules call for holding unconfirmed data private until verified, but for no more than six months. Some exceptions

may be negotiated for proprietary or otherwise sensitive data, but in no case will the data be kept private after publication. The center will occasionally act as a matchmaker when two laboratories with data in the six-month privacy window would obviously benefit from collaboration. Some laboratories still refuse to contribute to the database under these policies, but in general the chromosome 19 community seems to have accepted the policies as necessary for the health of the field. Further details of the LLNL Human Genome Center policies can be found below in the section on ownership and access issues.

Because of the labor-intensive nature of producing them, sharing genome-related materials with all who want them has proved far more difficult for LLNL, despite advances in automation. The center's attempts to cope with requests of this sort in fact constitute the major contribution of the LLNL case study to this report.

Livermore's initial involvement in the Human Genome Project was a joint effort with Los Alamos National Laboratory called the National Laboratory Gene Library Project, a task now essentially complete. All the libraries were made. They consisted of human chromosome-specific lambda libraries, both small insert and large insert, as well as human chromosome-specific cosmid libraries, and these were distributed to the scientific community after initial quality control.

The general policy followed in this distribution is that there is no bar to further distribution by the recipients of those libraries. Commercial use of individual clones is permitted, but not of the libraries. The center wanted to preclude the possibility that someone would tie up the libraries in some commercial agreement that prevents their use by the rest of the community, but people can take individual clones and commercialize them. The only quid pro quo requested of the users is acknowledgment in publications or in presentations that they received those libraries or those clones through the National Laboratory Gene Library Project, a request the center director estimates has been honored at least half the time.

The lambda libraries are probably the easiest to handle. Quality control of these libraries was performed in-house, although because of the large amount of quality control required, selected external laboratories were involved as well. In fact the center was dissatisfied with many of these laboratories and in the end had to recheck most of the work. Those clones were provided to the American Type Culture Collection (ATCC) for distribution. ATCC had received some initial funding from DOE to subsidize the cost of that distribution, which seems to be working quite well. Several thousand of those

libraries have been distributed worldwide. LLNL (and Los Alamos) receive no monetary remuneration for the clones or the libraries that go to ATCC.

Cosmid libraries are a little more difficult; so Livermore and Los Alamos basically split the 24 chromosomes present in the human genome. In some cases, multiple libraries were made. Chromosome 22, for example, had three libraries made of it, and the X chromosome had two libraries made of it. In all cases, these libraries were very extensive. The plan was to make at least a fivefold-deep library—in many cases they are 10- or 15-fold deep—and eventually develop an arrayed library in microtiter dishes.

This work took a lot of effort, and the center could not make libraries fast enough for the users who wanted them. Livermore distributed about 85 or 90 of these arrayed libraries and Los Alamos about the equivalent number, peaking in 1993 when the center had two full-time staff who did nothing but worry about this distribution and the rearraying. At that point in 1994, the center realized it just could not afford this work anymore and basically told the world that the services was being discontinued. The notion was that since there were now at least 82 copies from Livermore and an equivalent number from Los Alamos at other laboratories, it was time for those laboratories to assume the burden of further sharing.

A problem arose immediately. Many of the laboratories did not want to distribute the libraries, despite the fact that the center gave them permission, because they did not have the resources (staff, money, time, etc.). The center then tried a second approach, which was to tell laboratories requesting libraries, "Okay, you send people and supplies to our laboratory; our people will show yours how to rearray the libraries, and you can take them back with you." Unfortunately, it took only four months to discover that center staff were spending just as much time doing this work as if they had done it themselves, leaving them in their present quandry about what to do next.

One of the things that the staff have already done is provide plate pools to ATCC for distribution. These are nonarrayed sets of cosmid libraries. They have also reduced the amount of work that has to be done on rearraying by changing from 96-well microtiter trays to 384-well microtiter trays, which itself was costly in terms of staff time (two people doing this for almost a year to get the libraries rearrayed and to have at least a fivefold-deep, five-replicate set of each library).

The cost of arraying and shipping 150-some-odd libraries produced by the two national laboratories, at about $5,000 each, comes to about $750,000, which does not include the cost of having to rearray or set them up again. Both LLNL and Los Alamos have decided that it has become too costly for either laboratory to consider such distribution without some additional funding. Also it is not clearly in the best interest of a research and development

laboratory to do this work out of a budget that provides no funds for such service to other researchers.

Solicitation for external vendors has met with mixed success. The two center directors, with the encouragement of DOE, went to Commerce Business Daily and asked whether any industrial groups out there were willing to screen these libraries and distribute individual clones to end users. They included two qualifiers: (1) the group would not transfer those libraries to a third party for distribution, and (2) the libraries should not be rearrayed so as to create new clone libraries. The clones could be distributed, but the libraries could not be further distributed beyond them, in part because of ongoing collaborations on some of these libraries and some of the clones in them. The second qualifier was that any proposal for commercialization come back to the original national laboratory for permission to continue. The United Kingdom Resource Center and the German Resource Center both expressed great interest in being distributors and had no problem with the restrictions. They will be recipients of the libraries. However, only two responses were received from for-profit companies in this country, and neither was willing to meet the requirements of the qualifiers.

The Human Genome Center at LLNL has a very high resolution metric map of cosmids that spans nearly the entire chromosome 19. The correct order of the cosmids is known, as are the location and size of the gaps, and individual genes have been located both relative to each other and by restriction fragment. Other researchers hoping to find additional genes thus have a powerful incentive to contribute their candidate clones to LLNL, where they can be located within 50 kilobases of another marker. The resulting burden on center staff is substantial, however, and in recent years the center has taken a different tack. Robotics are used to make high-density filters on which the equivalent of 38,400 clones spanning the whole of chromosome 19 can be placed systematically. The resulting library is then sent to the would-be collaborator, who does the hybridization and notes which location "lights up." LLNL then pulls the corresponding clone from the freezer, verifies the match, and sends it to the collaborator for further research. More than 2,000 such clones had been sent to collaborators through 1995.

The center aggressively pursues collaborations, with subsequent involvement in publications, in part because laboratory policy requires that user fees, if charged, must offset the entire cost. Because its staff believes this would constitute a powerful disincentive for academic scientists, the center has chosen to distribute libraries to collaborators without charge.

The Human Genome Center at LLNL is also a central player in the Integrated Molecular Analysis of Genome Expression (IMAGE) Consortium, along with Washington University and Merck and Company. This group, whose founders included a scientist from the staff of Livermore, is dedicated to the widespread disbursement of cDNA clones. LLNL is basically the consortium's receiving center, accepting and arraying libraries from universities and industry, checking against a master array to minimize redundancy, and updating a tracking database. The clones are then sent to Washington University, and the information is transmitted to the dbEST Database at NIH. LLNL initially took on the task of distributing clones to requestors, but found that, as with chromosome 19, demand soon outstripped the center's capability, even with one staff member committed to the task full-time. Solicitation through Commerce Business Daily was more successful in this case however, and there are now several distribution centers, including ATCC, from which interested laboratories can obtain clones.

The Human Genome Center has responded to the exponential growth in requests for information and material resources by developing explicit guidelines and agreement forms, the essence of which has been conveyed above, but they will be summarized here for convenience.

Information and materials developed by or relinquished to LLNL and published in the open literature will be made freely available to the scientific community, to the extent that the center has the resources to comply with the request. In rare cases, requests for clones or other materials provided to LLNL by third-party collaborators may be denied if those collaborators have requested an exception to the LLNL release policy. The center requests that recipients maintain the clone or probe names assigned by LLNL and that subsequent reports and publications acknowledge contribution of such data or materials by LLNL. Libraries, clones, and other reagents provided may not be distributed beyond the recipient's lab, nor may they be used for commercial purposes without explicit permission of LLNL.

Unpublished information and materials are generally available either through collaboration or after the lapse of a suitable time period, generally six months from the date the material or data are entered into the LLNL database. Collaboration with one or more members of the center staff is the preferred mode of access. Collaborators are requested to provide LLNL with something of value to it—such as clones, probes, information, joint publications. Those

providing probes or primers will in turn be sent cosmids or YACs (yeast artificial chromosomes) that are positive, along with information on the contig status of these clones.

In the event of a request for collaboration within an area in which LLNL has an existing collaboration with a third party, the existing and potential collaborators will be notified and agreement to expand the existing collaboration sought. If no agreement can be reached, the original collaboration will be honored, and the six-month hold rule will govern release of data to the scientific community.

The issues and problems experienced by Livermore and its long and successful history of resource sharing are informative. The biggest issue is inadequate funding to support what has become exponential growth of demands for the products of successful programs. As the programs have outgrown the capacity of the initiating institutions to support them and distribute their products, there has generally been difficulty identifying other institutions (commercial or nonprofit) willing to assume responsibility for the expanding project and to make readily available, without constraints, the materials and information, even when charging fees.

The unique funding aspects of LLNL require that it either charge a fully allocated cost for materials and information or that the same be provided free of charge, except for shipping costs. The former would be prohibitively expensive and, according to the center director, would stymie resource sharing.

Another not infrequent problem is that multiple investigators utilizing the resources of the Human Genome Center, (i.e., sharing those resources) are not willing to share with each other, even when another (competitive) investigator might be in the immediate geographic proximity.

It seems that the pattern at Livermore is to initiate a project but as it becomes successful, the laboratory is not able to keep up with the demands of sharing. The center then decides to cease distribution other than providing it on a one time basis to subdistribution centers, if and when such can be found. LLNL then will recreate the same phenomenon with a newer technology and again exceed its capacity for sharing—perhaps an unavoidable price of success, but a price that may not be in the best interests of science.