DRAFT ONLY: DO NOT CITE OR CIRCULATE
Synthetic Genomics Options for Governance
Michele S. Garfinkel,* Drew Endy,‡ Gerald L. Epstein, # and Robert M. Friedman*
*The J. Craig Venter Institute, Rockville, Maryland
# Center for Strategic and International Studies, Washington, District of Columbia
‡ Massachusetts Institute of Technology, Cambridge, Massachusetts
The views and opinions expressed in this draft report are those of the authors and not necessarily those of the other study Core Group members, the participants of the workshops discussed in this draft, or of the institutions at which the authors work. The authors assume full responsibility for the report and the accuracy of its contents.
We gratefully acknowledge the Alfred P. Sloan Foundation for support of this study.
DRAFT ONLY: DO NOT CITE OR CIRCULATE
(page intentionally left blank)
27 Nov 2006 Synthetic Genomics: Options for Governance 2 DRAFT ONLY: DO NOT CITE OR CIRCULATE
TABLE OF CONTENTS
Introduction...... 5
Benefits...... 9
Risks...... 11
Framing a Policy Response...... 16
The Portfolio of Governance Options...... 18
Policy Options Presented (TABLE)...... 19
Policies for commercial synthesis firms...... 21
Policies for monitoring or controlling equipment or reagents...... 31
Policies for controlling publications and data for which open communication poses more risks than benefits...... 38
Policies for the roles of users and organizations in promoting safety and security in the conduct of synthetic genomics protocols...... 46
Choosing a Portfolio of Options...... 55
Summary of Options (FIGURE)...... 56
Summary of Portfolios (TABLE)...... 58
27 Nov 2006 Synthetic Genomics: Options for Governance 3 DRAFT ONLY: DO NOT CITE OR CIRCULATE
(page intentionally left blank)
27 Nov 2006 Synthetic Genomics: Options for Governance 4 DRAFT ONLY: DO NOT CITE OR CIRCULATE
INTRODUCTION
For decades, scientists have been searching for an efficient means to chemically synthesize genes, the building blocks of life. The first complete chemical synthesis of a gene was described in the early 1970s by Har Gobind Khorana and his colleagues. It was an arduous task, taking Khorana and 17 others years to assemble the very small (207 base-pairs) gene.1 Scientists had been “reading” the genetic code for years. Khorana and colleagues were the first to go the other way: from genetic code to a small, but functional, biological building block.
By the mid-1990s, Willem Stemmer and co-workers were able to synthesize a much larger gene and vector system (approximately 2700 base-pairs) using a variation of a standard molecular biology laboratory tool, the polymerase chain reaction.2 Stemmer’s technique has proven to be very useful in pharmaceuticals research.
In both Khorana’s and Stemmer’s experiments, as well as in those from other laboratories that were also interested in chemical synthesis of biological molecules, the goals of the researchers were both scientific and applied: to understand the natural world more completely, and to use that knowledge to make the world better.
In 2002, a team of researchers at the State University of New York led by Eckard Wimmer reported the assembly of an infectious poliovirus constructed in the laboratory directly from nucleic acids.3 Although this work was built on the prior examples of “from scratch” DNA synthesis noted above, Wimmer’s work demonstrated for the first time in a post-September 11th world the feasibility of synthesizing a complete microorganism—in this case a human pathogen—using only published DNA sequence information and mail- ordered raw materials.
The next year, a group from the Venter Institute (formerly the Institute for Biological Energy Alternatives) published a description of a similar technique applied to the construction of phiX174 (a virus that infects bacteria, called a bacteriophage).4 The advance here was not so much in length, as the viruses are of similar sizes, but in efficiency: compared to the one year or so required to synthesize and validate infectious poliovirus, fully-functional phiX174 was synthesized in approximately 2 weeks. Both
1 Agarwal KL, Buchi H, Caruthers MH, Gupta N, Khorana HG, Kleppe K, Kumar A, Ohtsuka E, Rajbhandary UL, Van de Sande JH, Sgaramella V, Weber H, Yamada T. 1970. Total synthesis of the gene for an alanine transfer ribonucleic acid from yeast. Nature 227:27-34. Khorana HG, Agarwal KL, Buchi H, Caruthers MH, Gupta NK, Kleppe K, Kumar A, Otskua E, RajBhandary UL, Van de Sande JH, Sqaramella V, Terao T, Weber H, Yamada T. 1972. Total Synthesis of the Structural Gene for an Alanine Transfer Ribonucleic Acid from Yeast. Journal of Molecular Biology 72: 209-217. 2 Stemmer WP, Crameri A, Ha KD, Brennan TM, Heyneker HL. 1995. Single-Step Assembly of a Gene and Entire Plasmid from Large Numbers of Oliogodeoxyribonucleotides. Gene 164: 49-53. 3 Cello J, Paul AV, Wimmer E. 2002. Chemical Synthesis of Poliovirus cDNA: Generation of Infectious Virus in the Absence of Natural Template. Science 297: 1016-1018. 4 Smith HO, Hutchison III CA, Pfannkoch C, Venter JC. 2003. Generating a Synthetic Genome by Whole Genome Assembly: φX174 Bacteriophage from Synthetic Oligonucleotides. Proceedings of the National Academy of Sciences USA 100: 15440-15445.
27 Nov 2006 Synthetic Genomics: Options for Governance 5 DRAFT ONLY: DO NOT CITE OR CIRCULATE
poliovirus and phiX174 are relatively small viruses, approximately 7400 and 5400 nucleotides (DNA subunits, where each subunit carries one letter of the genetic code: A, C, T, or G) respectively, but the lessons learned from these synthesis experiments are directly applicable to learning how to construct larger and more complex genomes. Further dramatic increases in the speed and accuracy of DNA synthesis would be necessary to lead eventually to the ultimate goal of the Venter Institute group: the synthesis not just of viruses but of whole bacteria. Today, a number of groups are working to design and construct from scratch novel bacterial genomes and simple eukaryotic chromosomes (e.g., yeast).
“Since the sequence is generated by chemical synthesis, there is full choice in the subsequent manipulation of the sequence information. This ability is the essence of the chemical approach to the study of biological specificity in DNA and RNA”, Khorana noted in 1979.5 Today, the rapidly-advancing technology of whole genome assembly reflects this potent observation. Synthetic genomics is a suite of techniques that permit the construction of any specified DNA sequence, enabling the chemical synthesis of genes or entire genomes. These DNA synthesis technologies applied to beneficial applications should have very positive impacts for individuals and for society.
However, the ability to reproduce very long sequences (in the tens or even hundreds of thousands of nucleotides) very rapidly and relatively inexpensively has led to concerns that bioterrorists could use these techniques to construct truly fearsome viruses such as smallpox from scratch. Synthetic genomics thus is a quintessential “dual-use” technology—a technology with broad and varied beneficial applications, but one that could also be turned to nefarious, destructive use.6 Such technologies have been around ever since the first humans picked up rocks or sharpened sticks. Nevertheless, dual-use bioscience and biotechnology, as exemplified by synthesis technology, pose special challenges, which are the subject of this report.
Synthesis technologies
Researchers have had the basic knowledge and tools to carry out de novo synthesis of gene-length DNA from nucleotide precursors for over 35 years. However, the techniques used on the first constructions were extremely difficult and constructing a gene of just over 100 nucleotides in length could take years.
Today, using machines called DNA synthesizers, the individual subunit bases adenine (A), cytosine (C), guanine (G), and thymine (T) can be assembled de novo, in any specified sequence using readily accessible reagents. Individual researchers and
5 Khorana HG. 1979. Total Synthesis of a Gene. Science 203: 614-625. 6 Atlas RM, Dando M. 2006. The Dual-Use Dilemma for Life Sciences: Perspectives, Conundrums, and Global Solutions. Biosecurity and Bioterrorism 4: 276-286. Committee on Research Standards and Practices to Prevent the Destructive Application of Biotechnology, National Research Council of the National Academies. 2004. Biotechnology Research in an Age of Bioterrorism. The National Academies Press (Washington, District of Columbia).
27 Nov 2006 Synthetic Genomics: Options for Governance 6 DRAFT ONLY: DO NOT CITE OR CIRCULATE
laboratories can buy bench-top DNA synthesizers that can string together pieces 75 to100 nucleotides long; such machines are readily available both new and used.
More commonly, however, a scientist will mail order a stretch of DNA up to 150 nucleotides in length from a commercial oligonucleotide manufacturer. Even longer gene-length (or even genome-length) stretches of DNA can be ordered from commercial gene foundries, who use proprietary state-of-the-art technology to assemble such long pieces of DNA.
In many regards, synthetic genomics is an evolutionary technology, offering a much more efficient way to achieve that which can already be done with standard recombinant DNA or other biochemical or molecular biology techniques. In another light, however, DNA synthesis offers the potential of revolutionary advances, making possible qualitatively new capabilities, broadening the number of scientists and engineers able to use biotechnology, and enabling them to consider higher-level applications without having to concentrate on the underlying molecular manipulations. Thus, this study focused on identifying new risks and benefits that could be derived from these new technologies, and on defining policy options to minimize the former without unduly hampering the latter.
The Study
The goal of this report, and the study of which it is a part, is to formulate governance options that attempt to minimize safety and security risks from the use of synthetic genomics while also allowing its development as a technology with great potential for social benefit. We focused on three societal concerns: bioterrorism, worker safety, and protection of communities and the environment outside of legitimate laboratories. We did not attempt to evaluate or assess broader issues associated with pathogenic microorganisms in particular, or biotechnology in general. These broader issues have been controversial for decades and are beyond the scope of this effort.
We follow several earlier studies that have looked at societal issues and made policy proposals or recommendations regarding synthetic genomics and synthetic biology. Among the earliest was a study examining the bioethics of synthesizing a bacterium7, following a proposal to use synthetic genomics to construct a minimal bacterial genome.8 Several National Research Council committees have considered a number of biological security issues.9 The best-known of these, commonly called the Fink Committee, issued a
7 Cho MK, Magnus D, Caplan AL, McGee D, and the Ethics of Genomics Group. 1999. Ethical Considerations in Synthesizing a Minimal Genome. Science 286: 2087-2090. 8 Hutchison III CA, Peterson SN, Gill SR, Cline RT, White O, Fraser CM, Smith HO, Venter JC. 1999. Global Transposon Mutagenesis and a Minimal Mycoplasma Genome. Science 286: 2165-2169. 9 Committee on Biological Threats to Agricultural Plants and Animals, National Research Council of the National Academies. 2003. Countering Agricultural Bioterrorism. The National Academies Press (Washington, District of Columbia). Committee on Genomics Databases for Bioterrorism Threat Agents. National Research Council of the National Academies. 2004. Seeking Security: Pathogens, Open Access, and Genome Databases. The National Academies Press (Washington, District of Columbia). Committee on Research Standards and Practices, op. cit. at 6
27 Nov 2006 Synthetic Genomics: Options for Governance 7 DRAFT ONLY: DO NOT CITE OR CIRCULATE
report that was the basis for the initiation of the National Science Advisory Board for Biosecurity 10. The Biological and Environmental Research Advisory Committee of the Department of Energy published its own report on the need for action to insure responsible and thoughtful pursuits in synthetic biology.11 Other groups and individuals have made specific proposals as well.12
Building on this base, we focused on the emerging technologies used to construct gene- and genome length stretches of DNA, and in particular, how the technology can be used to construct a microorganism (viruses, and, eventually, bacteria). The risk presented by the ability to synthesize a pathogen is the primary risk on which we focused.
We designed and held several workshops to gather and analyze information. We assembled a core group of 18 people (including ourselves); most attended every workshop and were very important in assuring that we identified, researched, and analyzed each policy challenge and option. In addition to the core group, each workshop involved other experts relevant to the workshop topic.
We could not hope to include all stakeholders (e.g., civil society groups, regulators, policymakers, research administrators, etc.) in every aspect of the study. Nevertheless, the core group included a wide variety of perspectives, including synthetic genomics researchers, sellers of synthesized DNA, policy analysts who focus on bioterrorism, and those who focus on the legal, ethical, and societal implications of biotechnology.
We held three workshops over 20 months. The first workshop in September 2005 examined Synthesis Technologies. This workshop examined currently available DNA synthesis technologies and how those technologies might change or be replaced in 5 to10 years. This workshop also identified opportunities for technical interventions to impede malicious use of the technology. The resulting overview included detailed discussions, based on commissioned papers, about the flow of reagents from raw materials to phosphoramidite precursors to finished oligonucleotides to full length genes. The workshop also explored current capabilities of computer software to screen the sequences of oligonucleotide and gene-length orders for defined sequences and malicious intent. Participants also considered explicitly how the availability of certain kinds of equipment (e.g. DNA synthesizers) and know-how affect how easy or difficult it is to construct a microorganism from raw materials.
The second workshop explored both the applications (benefits) and potential dangers or misuses (risks) of the technology. Risks and Benefits Specifically Attributable to Synthetic Genomics, held in February 2006, explored the question, “how does a world with synthetic genomics differ from one without it?” With respect to security or safety
10 Committee on Research Standards and Practices, op. cit. at 6 11 United States Department of Energy, Biological and Environmental Research Advisory Committee. 2004. Synthetic Genomes: Technologies and Impact. http://www.sc.doe.gov/ober/berac/SynBio.pdf 12 Church G. 2004. A Synthetic Biohazard Non-Proliferation Proposal. http://arep.med.harvard.edu/SBP/Church_Biohazard04c.htm The Conferees of the Second International Conference on Synthetic Biology. 2006. Declaration of the Second International Meeting on Synthetic Biology (Draft) 29 May. http://hdl.handle.net/1721.1/32982
27 Nov 2006 Synthetic Genomics: Options for Governance 8 DRAFT ONLY: DO NOT CITE OR CIRCULATE
risks, a key finding of this workshop was that today, except for a few important exceptions, there are far easier ways to obtain a pathogen, if that is one’s intent. However, within several years—and no more than a decade—not only might it be possible to synthesize any virus, but that in many cases it may become easier to synthesize a virus than to find it in nature or to obtain it from a laboratory. The workshop also explored various aspects of biosafety; a key concern was the number of new researchers coming into the field from non-microbiology backgrounds.
At the final workshop in May of 2006, Governance, we began to evaluate the various policy options that were identified during the first two workshops. We explored the current regulatory mechanisms governing synthetic genomics and evaluated new measures with potential for mitigating risk while preserving benefits.
BENEFITS
Over the course of the study, we identified several major areas where synthetic genomics could make a unique or significant contribution: as an enabling technology that is changing the nature of basic biological research and for its potential in applied biotechnology research to develop new pharmaceuticals, biological sources of transportation fuels, and manufacturing of other bio-based products.
Synthetic genomics is even today changing the nature of basic molecular biological research. As an enabling technology, it has already been shown to be a significant time saver by shortening the time needed for normally arduous recombinant DNA techniques; in the coming 5 to10 years it should become less expensive as well.13 Using synthetic genomics to rapidly change the sequence of various genes or whole genomes could be (and in some cases already is) a powerful tool for basic research in a number of disciplines. For example, various laboratories are already using synthetic genomics to understand the mechanisms of evolution at the molecular level,14 to rapidly define regulators of specific genes or gene pathways and to begin to validate, at the DNA level, the minimal requirements for life.15
This capability to quickly make subtle changes at the sequence level may lead to more efficient research and production of vaccines for human and animal health and related diagnostics. Specifically, the ability to assemble and mutate sequences rapidly could allow for the development of broadly protective vaccines against, and diagnostics for, viruses that themselves are diverse and variable, such as SARS16 and hepatitis C.17
13 Based on discussions at the first workshop, discussed above. 14 E.g., Szostak JW, Bartel DP, Luisi PL. 2001. Synthesizing Life. Nature 409: 387-390 Gibbs WW. 2005. Synthetic Life. Scientific American 290: 74-81. 15 Chan LY, Kosuri S, Endy D. 2005. Refactoring bacteriophage T7. Molecular Systems Biology 1: doi:10.1038/msb4100025 16 Baric RS, Sheahan T, Deming D, Donaldson E, Yount B, Sims AC, Roberts RS, Frieman M, Rockx B. 2006. SARS coronavirus vaccine development. Adv Exp Med Biol. 581:553-60. 17 Department of Energy op. cit. at 11.
27 Nov 2006 Synthetic Genomics: Options for Governance 9 DRAFT ONLY: DO NOT CITE OR CIRCULATE
Synthetic techniques have already been applied in research on new or improved drugs. Artemisinic acid, the precursor of the antimalarial drug artemisinin, was produced in a strain of yeast specifically engineered to allow production of high titers of the product.18 Artemisinin is normally extracted from the plant Artemisia annua harvested from the environment, a process that is inefficient, expensive, can contaminate the product with other plant material, and depends on the weather and possibly even the political situation in regions where the plant is found. The production process in yeast is currently being optimized for industrial scale-up, but even this first step is a significant improvement over the drug’s current source.
Another group19 has described the total synthesis of a 32,000 base-pair polyketide synthase gene cluster. This synthesis was notable for its length (this remains one of the longest syntheses published to date) and that it yielded an active gene product. The enzyme it encodes is in a class of extremely important drugs (which includes antibiotics, transplant rejection suppression treatments, and potential anti-cancer drugs). Synthesizing many variants of these genes could provide pools of potential drugs, which can then be screened for the desired activities.20
Synthetic genomics could contribute to the search for carbon-neutral energy sources. A major application of synthetic genomics will be to overcome biological barriers to producing biofuels.21 Specifically, two hurdles to moving to liquid biofuels have been identified by a variety of individuals and groups. First, an important (or at least highly desirable) feature of a biofuel-producing organism is to have as many of the steps from input (e.g. cellulose) to output (fuel) taking place within the microorganism itself, with a minimum of post-harvest processing. Second, a large fraction of the feedstock (in particular, lignin) cannot yet be converted to liquid biofuel. No one organism, or even a few, has been discovered that can fulfill these needs. However, using synthetic genomics (and other techniques of biotechnology generally), it is imaginable that a microbe (probably a bacteria) could be engineered with many, if not all, of these features, making biofuels a viable possibility.22
Sometimes called “white biotechnology”, biobased manufacturing is becoming a reality. Plants and microbes are being engineered to produce the raw materials that can be used to manufacture products that today are petroleum-based. The hope is that biologically- based manufacturing will lead to more environmentally friendly methods of production,
18 Ro DK, Paradise EM, Ouellet M, Fisher KJ, Newman KL, Ndungu JM, Ho KA, Eachus RA, Ham TS, Kirby J, Chang MC, Witheres ST, Shiba Y, Sarpong R, Keasling JD. 2006. Production of the Antimalarial Drug Precursor Artemisinic Acid in Engineered Yeast. Nature 440: 852-853. 19 Kodumal SJ, Patel KG, Reid R, Menzella HG, Welch M, Santi DV. 2004. Total Synthesis of Long DNA Sequences: Synthesis of a contiguous 32-kb Polyketide Synthase Gene Cluster. Proceedings of the National Academy of Sciences USA 101: 15573-15578. 20 Herper M. 2006. The Biggest DNA Ever Made. Forbes.Com 13 July http://www.forbes.com/home/sciencesandmedicine/2006/07/12/dna-artificial-genes-codon- cz_mh_0713codon.html 21 Department of Energy op. cit. at 11. 22 United States Department of Energy. 2006. Breaking the Biological Barriers to Cellulosic Ethanol: A Joint Research Agenda. DOE/SC-0095 (June).
27 Nov 2006 Synthetic Genomics: Options for Governance 10 DRAFT ONLY: DO NOT CITE OR CIRCULATE
as well as environmentally preferable products themselves. For example, the environmental impacts of plastic manufacturing might be lessened through judicious use of bioengineering of metabolic pathways using synthetic genomics as one tool.23
Finally, millions of new genes are being discovered in metagenomic surveys (looking at thousands of species at the same time) of bacteria living in natural environments. Some of these could be important in engineering specific pathways into microbes as described above. Because the microorganisms they come from typically cannot be cultured in the laboratory, the genes or genomes of interest are known only by their sequence. Synthetic genomics could allow for the reconstruction of these potentially important new genes.
RISKS
Security
The biochemistry underlying the synthetic construction of genomes is in many ways similar to more widely-used techniques that have been used in laboratories since the establishment of standardized techniques for recombining pieces of DNA. In fact, the tools and techniques used in the actual construction of genomes and their subsequent use are so similar, and in some cases identical, to the recombinant DNA techniques that are now the staple of modern biological research that a simple way to compartmentalize this research for the purposes of oversight is simply not possible. Further, the databases of DNA sequence and computation-based design tools that enable this technology are rapidly growing in size and sophistication, and are available to virtually anyone in the world with a computer.
With the availability of tools and information, the only additional factors determining if any individual could construct a pathogen (for good or for bad) are access to a moderately well-equipped biology laboratory, the know-how to carry out the necessary synthetic techniques, and the biological expertise to take strands of nucleic acid and “boot” them into viable, self-replicating organisms (or transfer them into a cell so that they express their particular gene products). Although this is not yet a general laboratory technique that can be carried out by using a standard protocol manual, many workers are becoming adept at these techniques.24
As discussed above, to date, a variety of techniques all aimed at experimentally constructing ever-longer nucleic acid chains have been used to construct a partially-active 7700 nucleotide copy of poliovirus, a fully active copy of the 5400 nt bacteriophage phiX174, and a 32,000 nt polyketide synthase gene cluster, among others. Although the protocols used to construct these genes and genomes were not identical, what they had in
23 Aldor IS, Keasling JD. 2003. Process Design for Microbial Plastic Factories: Metabolic Engineering of Polyhydroxyalkanoates. Current Opinion in Biotechnology 14: 475-483. Biotechnology Industry Organization. ND (~2004/2005). New Biotech Tools for a Cleaner Environment. 24 Brent R. 2006. In the Valley of the Shadow of Death. Draft. https://dspace.mit.edu/bitstream/1721.1/34914/1/Valley2006.pdf
27 Nov 2006 Synthetic Genomics: Options for Governance 11 DRAFT ONLY: DO NOT CITE OR CIRCULATE
common was in the use of the starting material: easily commercially obtained oligonucleotides of relatively short length (under 100 nucleotides). The actual techniques for joining the oligos in the laboratory currently are relatively arduous (compared to ordering gene- or genome-length DNA via a website), but are becoming easier as researchers improve their techniques through experience.
However, it is not necessary for researchers (or anyone) to assemble the oligonucleotides in-house; firms have opened over the last several years that will fill orders for stretches of DNA that are tens of thousands (or potentially even hundreds of thousands) nucleotides long. Synthetic DNA of this length could be used directly, or with relatively few manipulations, to construct a virus. This is potentially very useful: the ability to quickly construct a virus in a laboratory could be used for making a vaccine against a new or re- emerging virus, for clarifying mechanisms of pathogenesis in looking for cures, and in basic research such as understanding evolution. At the same time, such a service could allow those with malicious intent to quickly make a virus to be used as a biological weapon.
Thus, for legitimate researchers, these gene- and genome synthesis firms can save them significant amounts of time (and, by replacing certain types of laboratory work, money as well). Mitigating the possibility that someone might take advantage of this ease of ordering to commit an act of terrorism will be covered in detail in this report.
However, genes and genomes can also be constructed without the use of commercial synthesis firms. An individual or group with patience and a moderate amount of money can easily purchase a DNA oligonucleotide synthesizer, phosphoramidites (the basic units for DNA synthesis), and a few common chemicals in order to synthesize oligos from scratch; these oligos would then be joined in the lab for a full-length gene or genome. Granted, given the emerging economies of scale in oligo and gene synthesis, this approach would likely cost more, take longer, and have a higher error rate than would use of commercial firms.
A bioterrorist, then, might be able to assemble the genomes of at least some pathogens without being detected. Whether such an individual or group could “boot” the pathogen (i.e., produce an intact and infectious virus particle starting from the intact genetic material) would be an unknown. Since some viruses can be booted just by introducing the DNA into an appropriate cell it is prudent to assume that at least some viruses could now be readily produced by anyone with an interest in doing so.
For example, many viruses, some of them pathogenic, have already been constructed from scratch in the laboratory. It has been estimated that within 2 to 5 years it will be possible to synthesize any virus.25 In the 5 to10 year time frame, some simple bacterial genomes will be synthesized with ease; once synthesized, it should be relatively
25 Venter JC. Presentation to the National Science Advisory Board for Biosecurity. 1 July 2005. Additional information from Workshops 1 and 2 of this project.
27 Nov 2006 Synthetic Genomics: Options for Governance 12 DRAFT ONLY: DO NOT CITE OR CIRCULATE
straightforward to transfer that genome to an empty bacterial cell, thus producing a living cell from the inert genome.26
But just because a potential bioterrorist could synthesize the genome of a pathogen, would they? Other routes might be much simpler and more reliable. Several options come to mind: buy it from a biological repository, steal it from a laboratory that maintains it for research, or obtain it from nature.
The easiest way, if it were possible, would be to just order it from a biological repository, such as the American Type Culture Collection (ATCC), a nonprofit organization that distributes viruses and bacteria to microbiologists around the world. However, physical access to and transfer of potentially dangerous biological samples is strictly controlled in the United States under the so-called Select Agent regulations. And while ATCC maintains many pathogens, the most dangerous ones, e.g., Ebola virus, are not available at all. Thus ordering dangerous pathogens from a U.S. repository, or one in a country with equivalent regulations, is not an effective approach.
Stealing a pathogenic virus from a laboratory that maintains it for research would certainly be possible for many pathogenic viruses, but the laboratories that maintain the most dangerous ones are, again, relatively well controlled.
Obtaining the virus from nature is yet another possibility. For some viruses, this is easy. For example, some animal viruses are quite common and require no special equipment to collect. Foot-and-mouth disease virus, which is not endemic in the United States, can be collected on a blanket from an infected barn. However, other viruses, such as Ebola, appear only sporadically, thus one would have to wait for an outbreak.
If a virus was not already readily accessible then, de novo synthesis might provide a relatively attractive or sole path for acquisition. The smaller the genome of a virus the easier it will be to synthesize, but that is not the only factor. While in principle it is possible to go from a synthetic genome to infectious virus for any viral threat, some are more difficult to produce than others.
After examining the viruses on several lists of microbial agents that pose severe threats to public health and safety, we concluded that four types of viruses would be easier to synthesize and then make infectious, than to obtain by other means. These include smallpox, filoviruses such as Ebola and Marburg, 1918 influenza virus, and perhaps the 2003 SARS coronavirus.
Smallpox (variola) virus, once a major scourge, has been eradicated and no longer exists in nature. The only known stocks of variola virus are in two high-laboratories, one in the United States (CDC in Atlanta) and the other in Russia.
26 Information from Workshops 1 and 2 of this project. See also Glass JI, Assad-Garcia N, Alperovich N, Yooseph S, Lewis MR, Maruf M, Huchison III CA, Smith HO, Venter JC. 2006. Essential Genes of a Minimal Bacterium. Proceedings of the National Academy of Sciences USA 103: 425-430.
27 Nov 2006 Synthetic Genomics: Options for Governance 13 DRAFT ONLY: DO NOT CITE OR CIRCULATE
The natural reservoirs of Ebola and Marburg filoviruses are still unknown, thus the only way to obtain them from nature is during their intermittent outbreaks. Because they are so virulent, they are kept in very few research laboratories, all of which are tightly controlled.
1918 influenza no longer exists in nature. Until recently, it did not exist in the laboratory either, but in 2006 was successfully synthesized from the DNA sequence obtained from a variety of historic samples. The virus was “resurrected” by synthesis to enable research on why this particular strain was so virulent.
The 2003 SARS coronavirus also no longer exists in nature. Again, because of its virulence, it is maintained in but a few tightly controlled laboratories. Its sequence, however, is easily obtainable from public databases and because of its size and other characteristics, like 1918 flu, is possible to synthesize SARS by chemical means.
Again, a bioterrorist might be able to synthesize any of these viruses in three different ways. The simplest would be to order large stretches of their genome (thousands to tens of thousands of bases long) directly from a gene synthesis company. Much of the hard work needed to synthesize an infectious virus would be done by the company, greatly simplifying the bioterrorist’s task. We present several options below intended to close such an obvious weakness in the system.
Once the simplest option is no longer available, a bioterrorist might then attempt to order from a commercial supplier of oligonucleotides, DNA pieces 50 or 100 bases long, and assemble these short pieces himself. If blocked from exploiting this route, a bioterrorist might turn to synthesizing oligo-length pieces of DNA, by purchasing a desktop DNA synthesizer. Each of these approaches is possible, but with each the challenges facing a would-be bioterrorist become increasingly difficult and it is less likely that an attempt would succeed.
The possibility of constructing superpathogens (novel pathogens that have never seen before in nature, and for which treatment is not effective was initially dismissed as being too difficult for even the best scientists to accomplish. However, later discussions have elicited at least a recognition that we may begin to understand the mechanisms underlying pathogenesis to the point that within 10 to15 years the idea of constructing a superpathogen would no longer be so far-fetched.
Laboratory Safety
Biosafety, the protection of workers in the laboratory, and of the communities and environment outside of the laboratory, is always a concern in any biomedical laboratory. With the addition of synthetic genomics to the toolbox that is available to researchers, new risks may be introduced, or existing risks increased.
27 Nov 2006 Synthetic Genomics: Options for Governance 14 DRAFT ONLY: DO NOT CITE OR CIRCULATE
The first and perhaps most important concern is a result of the influx of new workers to the field. Participants at our workshop expressed concern that a general degradation of biosafety practice might occur due to the relatively large numbers of new workers coming from fields such as engineering or physics, without adequate biosafety training.
The Asilomar meeting of 1975 (discussed below) is the foundation of biosafety as it is practiced in the United States today. Interestingly, the major discussion there—the safety of transmitting genes from one organism via another organism (a vector such as a virus or bacterium) into a third organism—echoes concerns expressed for synthetic genomics today: how to assess the safety of chimeric organisms that have genomes derived from a very large number of initial sources. Specifically, using standard recombinant DNA cut- and-paste techniques, it is possible to readily assemble a chimera from tens of sources. Synthetic constructions, however, could be from hundreds of sources or more. How to evaluate such constructions for biological safety concerns remains murky. While there are no data to suggest that such higher-order chimeras will be dangerous per se, this concern has nonetheless prompted some to suggest that all synthetic genomics protocols should take place under levels of biological containment used for the most dangerous human and agricultural pathogens (e.g., Biological Safety Levels [BSL] 3 or 4).27 Requiring such containment would have the effect of making such work expensive, and would thus restrict it to far fewer labs than might utilize it otherwise..
Creating a pathogen genome by combining large numbers of genetic regions not known to generate pathogenic traits is not entirely implausible: this is one mechanism for evolution (though over much longer time periods). A very small number of reports in the literature describe unexpected pathogenic qualities arising from genetic modification (usually on microorganisms that were already pathogenic; these changes for the most part had to do with extending the range of species that the pathogen could infect28). But there has been virtually no research on the biological nature of interspecies chimeras, and those carrying out risk assessment for their synthetic genomics experiments will need to carefully consider this in determining the risk level. Specifically, experiments aimed at constructing new species with unknown metabolisms or with unknown membrane compositions and thus unknown cell-cell interaction properties might be done at a higher containment level than routine.29 Higher containment might also be appropriate for experiments listed in the Fink report as “Experiments of Concern.”30 It has been suggested in our workshops and by others that BSL-2 or BSL-2+ containment is the appropriate default, unless any segments of the DNA to be constructed are known to have a higher inherent risk level.31
27 Tucker JB, Zilinskas RA. 2006. The Promise and Perils of Synthetic Biology. The New Atlantis Spring 2006: 25-45. 28 Amsellem Z, Cohen BA, Gressel J. 2002. Engineering Hypervirulence in a Mycoherbicidal Fungus for Efficient Weed Control. Nature Biotechnology 20: 1035-1039. 29 Species with unknown metabolisms would have unknown metabolic dependencies, making it difficult to design them to be auxotrophic (dependent on unique laboratory-supplied nutrients and therefore incapable of surviving in the external environment). 30 Committee on Research Standards and Practices, op. cit. at 6 31 Discussions at Workshop 2 of this project; additional information via personal communications.
27 Nov 2006 Synthetic Genomics: Options for Governance 15 DRAFT ONLY: DO NOT CITE OR CIRCULATE
FRAMING A POLICY RESPONSE
In the mid-1970s, influential scientists who had pioneered the emerging techniques of genetic engineering called for a moratorium on recombinant DNA research until the safety implications of that work could be more thoroughly reviewed. From February 24- 27, 1975, nearly 200 researchers convened for that purpose at the Asilomar Conference Center near Monterey, California. By the end of the conference, they concluded that the work could be done with minimal risk to the environment or to workers, provided that certain guidelines were followed. The most famous of these guidelines, appropriate containment via risk assessment, resulted in the partitioning of experiments into the risk group levels associated with containment levels now commonly referred to as Biological Safety Level (BSL) -1 through –4). In the more than thirty years since the Conference, the conclusion, that the work is or can be made safe, has held true.
There have been suggestions that synthetic genomics needs “another Asilomar.”32 But Asilomar is for the most part not the right model for the field. Asilomar was an exercise in self-governance: the community itself determined and imposed on itself those procedures needed to ensure safety. Although there was apparently some discussion over whether to consider the possibility of biological warfare, it was decided that the hazards of carelessness were enough to deal with at that meeting.33 For addressing the risks of bioterrorism, self-governance will not be sufficient. Bioterrorists, by definition, are not willing to accept the norms of the research community, and no community can control all subsequent uses to which the research results or techniques it develops might be put.
Nevertheless, the research community can take action to lessen the risk that scientific and technical advances might be misapplied, and to maintain confidence among decisionmakers and the public that continued advance of science and technology is, on balance, beneficial to society. Both questions were pressed after the attacks of September 11th, 2001, and the subsequent anthrax letter mailings, which threatened to change the relationship between the security community and the biological sciences.
With the realization that individuals or small groups with sufficient biological expertise could produce biological agents that could terrorize, if not attack, an entire country, the scientific community became concerned that law enforcement and national security officials and public opinion might demand a more restrictive and legally binding regulatory regime to lessen the risk that research materials, expertise, and facilities would be used to make weapons. In part to forestall such an outcome, but also motivated by a sense of professional responsibility, the scientific community began to address what actions it could take on its own to protect the ability of science to advance without contributing to weapons programs34 or to rogue bioterrorists.
32 Ball P. 2004. Synthetic Biology: Starting from Scratch. Nature 431: 624-626. 33 Rogers M. 1975. The Pandora’s Box Congress. Rolling Stone 19 June, p. 37 and ff. 34 Ronald M. Atlas, “Securing Life Sciences Research in an Age of Terrorism,” Issues in Science and Technology, Fall 2006.
27 Nov 2006 Synthetic Genomics: Options for Governance 16 DRAFT ONLY: DO NOT CITE OR CIRCULATE
Policy goals and selection criteria
In a following section we present 20 options for the governance of synthetic genomics. These options address three key policy goals:
• Enhancing biosecurity, by either preventing incidents of bioterrorism or by helping law enforcement identify those responsible when incidents should occur. • Fostering laboratory safety, again by either preventing accidents or by helping to respond in the event an accident does occur. • Protecting the environment, the people and natural ecosystems outside the laboratory.
For each of the 20 options, we have included our judgment about their effectiveness for achieving each of these three goals. Of course, the overall desirability of an option depends on a host of other considerations, as well. Thus we have evaluated how well each option fares on four other key criteria. These include:
• Does the option hold down costs and other burdens to both government and the affected industry? • Can the option be implemented today, or does it need additional research before it will be effective? • Does the option unduly impede biological research or progress by the biotechnology industry? • Does the option help to promote constructive applications of the technology?
Keep in mind, however, that the technologies of synthetic genomics are relatively new and rapidly advancing and evolving. There is no crystal ball with which to predict the future, nor are there policies robust enough to accommodate all plausible futures. In such rapidly changing situations, a framework of “adaptive decision making” seems most appropriate. A suite of options are put in place today that match today’s technologies, the magnitude of today’s risks and benefits, and societal priorities. The keys to success are 1) that conditions are closely monitored and 2) that decision makers are prepared and willing to modify the suite of options accordingly. Not only might tomorrow’s choice of options be different, but the options from which to choose might be drastically altered as well.
Identifying intervention points
We identified several promising points for policy intervention by considering the several ways a gene or genome can be synthesized. We identified four “factors of production” needed to construct genes or genomes: raw materials and reagents, sequence information, equipment, and know-how.
27 Nov 2006 Synthetic Genomics: Options for Governance 17 DRAFT ONLY: DO NOT CITE OR CIRCULATE
To thwart the intent of a potential bioterrorist, points for policy intervention include:
• At the point of DNA synthesis itself o Gene foundries, that can produce whole genes and genomes o Oligonucleotide manufacturers, that sell short stretches of DNA o Desktop DNA synthesizers used in individual laboratories to make short stretches of DNA o Raw materials (when linked with control of DNA synthesizers)
• Information needed for synthesis o Sequence data o Literature from which the procedures to assemble short pieces of DNA into whole genomes can be obtained
For legitimate investigators, the points for potential intervention to enhance laboratory safety and minimize risks to the environment include:
• The investigator, through such mechanisms as o Education o Training tools, such as manuals and clearinghouses • Oversight bodies, such as Institutional Biosafety Committees
The options below address specific policy options for each of these intervention points.
THE PORTFOLIO OF GOVERNANCE OPTIONS
Introduction
Below are four groups of policies that could contribute to the governance of synthetic genomics. The evaluations are presented both as text and in a summary chart. The chart is helpful for comparing the effectiveness of the various options in enhancing security and safety against other considerations (such as costs to firms). Policy options were evaluated as described earlier.
The options presented in Table 1, below, are derived from a variety of inputs. In our initial research, we recognized a general set of concerns and stakeholders that would be relevant in any discussions of security and safety. Over the course of the three workshops and discussions with the core group and other participants, we developed a deeper understanding of the needs of various actors and how these groups interact (or not) with each other. Some of the options were clearly suggested by individuals; some were developed by discussions of the larger group. In all cases, we evaluated each policy option on the criteria described in the previous section.
27 Nov 2006 Synthetic Genomics: Options for Governance 18 DRAFT ONLY: DO NOT CITE OR CIRCULATE
TABLE 1: Policy options presented in this section
Ia. Policies for commercial gene- and genome synthesis firms Ia-1. Require Firms to Use Approved Software for Screening Orders Ia-2. People Who Order Synthetic DNA Must be Certified by an Institutional Biosafety Officer or Similar “Responsible Official” Ia-3. Firms Must Use Approved Screening Software; People Who Order Must be Certified by Biosafety Officer Ia-4. Firms Must Store Information About Customers and Their Orders
Ib. Policies for commercial oligonucleotide synthesis firms Ib-1. Require Firms to Use Approved Software for Screening Orders Ib-2. People Who Order Synthetic DNA Must be Certified by an Institutional Biosafety Officer or Similar “Responsible Official” Ib-3. Firms Must Use Approved Screening Software; People Who Order Must be Certified by Biosafety Officer Ib-4. Firms Must Store Information about Customers and Their Orders
II. Policies for monitoring or controlling equipment and reagents II-1. Registration of DNA Synthesizers II-2. Licensing of DNA Synthesizers II-3. Licensing of Synthesizers, plus License Required to Buy Reagents or Services
III. Policies for controlling publications and data III-1. Self-Governance of Risky Information by Scientists and Journals III-2. Self-Governance, Assisted by a National Advisory Group III-3. Restrict Access to Information Flagged by Self-Governance Process
IV. Policies for users and organizations for promoting safety and security in the conduct of synthetic genomics research. IV-1. Education About Risks and Best Practices as Part of University Curriculum IV-2. Compilation and Use of a Manual for “Biosafety in Synthetic Biology Laboratories” IV-3. Clearinghouse for Best Practices IV-4. Broaden IBC Review Responsibilities to Consider Risky Experiments IV-5. Broader IBC Review, plus Oversight from National Advisory Group to Evaluate Risky Experiments IV-6. Broader IBC Review, plus Enhanced Enforcement of Compliance with Biosafety Guidelines
27 Nov 2006 Synthetic Genomics: Options for Governance 19 DRAFT ONLY: DO NOT CITE OR CIRCULATE
The list above does not include two additional options that, after a significant amount of discussion, we decided not to analyze in detail. The first of these was “No Changes Needed—Current Policies are Adequate.” As discussed in chapter 2, synthetic genomics provides the ability to readily synthesize certain viruses, including some known as “select agents” that currently fall under an existing set of regulations governing their access, possession, and use. These select agent regulations are designed to prevent unauthorized users from acquiring these pathogens through diversion or theft from legitimate facilities—but they lose relevance to the extent that select agents can now be synthesized directly. As a result, policymakers may seek controls on synthesis that would serve to restore some of the power that select agent controls once had.
Even if policymakers believe that the current regulatory approach can accommodate synthetic genomics, guidance is needed merely to apply existing regulation or governance mechanisms, defined in the context of one set of technologies, to a subsequent technological generation. Policymakers satisfied with the existing approach can simply choose to reject all of our proposed options.
The other rejected option was “Freeze Further Developments in Synthetic Genomics— No Other Policy Mechanism Can Adequately Mitigate Risk.” This was rejected for at least two reasons.
Pragmatically, we could not find a policy “handle” that could freeze further developments in synthetic genomics without also denying the use of synthetic DNA for a host of other uses, or that even could have high confidence in banning synthesis in the first place. There are no chokepoints specific to synthesis. Given the overlaps in materials, equipment, and products for synthesis techniques and “conventional” recombinant DNA research, any attempt to ban this work would likely result in a de facto ban domestically—and export offshore—of virtually all that we would recognize as modern biology, while, given the relative simplicity of obtaining some of the necessary equipment, not necessarily even succeeding at preventing all synthesis.
Moreover, the discussions at our workshops and our analyses failed to convince us that the risks are in fact so large that freezing the field would be more useful to society than allowing it to continue.
27 Nov 2006 Synthetic Genomics: Options for Governance 20 DRAFT ONLY: DO NOT CITE OR CIRCULATE
The Options
I. Policies for commercial synthesis firms
DESCRIPTION OF THIS INTERVENTION POINT
Though it is certainly possible to synthesize a gene- or genome-length piece of DNA from its basic building blocks using a DNA synthesizer in one’s own laboratory, most researchers today custom-order synthesized DNA from commercial firms. The genetic code of the desired piece of DNA is submitted electronically over the Internet. The desired DNA is synthesized in a specialized facility and then shipped to the researcher. By using such firms, researchers obtain more accurate DNA for their experiments, avoid the need for expensive equipment, and minimize the amount of technical expertise needed.
Similarly, the easiest path for a bioterrorist to synthesize a pathogen is to obtain custom- ordered DNA from a commercial firm. Note as well that for most pathogens, synthesizing a genome would be more difficult than either stealing it or obtaining it from nature. But for a few viral pathogens today, synthesis is a plausible alternative.
Today, two types of firms supply synthesized DNA. The first supplies shorter-length oligonucleotides (single-stranded DNA), typically up to 100 base pairs in length. The bulk of the synthetic DNA and RNA market is for such shorter-length pieces, used for a variety of purposes. To synthesize 1918 influenza, for example, a researcher (or a bioterrorist) might order several hundred oligo-length pieces of DNA that would be assembled to construct the entire 14,600 base pair genome.
A small number of firms—fewer than 100 worldwide—specialize in synthesizing gene- and genome-length pieces of double-stranded DNA (and, less routinely, RNA), sometimes incorporated into living cells for shipment. Again, using the example of 1918 influenza, the genome consists of eight segments ranging in size from about 900 to 2300 base pairs. A bioterrorist could conceivably order the eight segments and then insert them into an animal cell to form the complete virus.
For a potential bioterrorist, assembling a genome from these larger pieces is less difficult technically than starting with the shorter-length oligos, and perhaps more important, far much less time consuming. Much of the highly skilled labor needed to synthesize a genome is, in essence, readily available for hire. Thus, we believe that options that focus on “gene foundries”, i.e., firms that can synthesize gene and genome length stretches of DNA and RNA, are top priorities for preventing nefarious uses of synthetic genomics.
The difficulty of constructing a genome from commercially synthesized oligos is comparable to the difficulty of starting with oligos constructed in one’s own lab with a privately owned DNA synthesizer. However, ordering oligos from commercial firms
27 Nov 2006 Synthetic Genomics: Options for Governance 21 DRAFT ONLY: DO NOT CITE OR CIRCULATE
clearly saves time compared to synthesizing them in one’s own lab; thus, screening by oligo suppliers may be the next best intervention point for preventing potential incidents of bioterrorism using synthesized DNA.
DESCRIPTION OF OPTIONS
Clearly, commercial DNA synthesis firms have no interest in supplying potentially harmful pieces of DNA to users who are not using them for legitimate research purposes or who may be unaware of danger to themselves or others. Below we present options either to detect and thus prevent shipment of harmful genes or genomes to someone who has no legitimate need for such sequences or, at minimum, to record these shipments for surveillance or forensic purposes.
Two general approaches are possible for screening DNA orders prior to synthesis. First, computer software can be used to compare the submitted DNA sequence to those of known pathogens. First-generation software is available and already in use at several gene foundries. However, software improvements and a more refined list of potentially harmful genes and genomes would greatly enhance the effectiveness of computer-based approaches. These research needs are discussed later in this section.
The entire responsibility and burden for screening does not have to fall to commercial firms that synthesize DNA. The vast majority of their customers are employed by universities, research institutes, or private firms such as pharmaceutical companies. Most such institutions will employ a trained biosafety professional. By requiring that biosafety professionals be part of the ordering process, one can ensure that all orders are from known researchers working at known institutions, and not from rogue individuals.
Finally, there is merit to storing information about previously placed orders for forensic purposes in the event of a bioterrorist attack. The sequence of the pathogen can be compared to recent synthesis orders to identify potential matches.
We first describe each of these options below and then compare the advantages and disadvantages of each in the following section.
I-1. Require commercial firms to use approved software for screening orders
As mentioned above, computer software can be used by commercial firms to compare the DNA sequence submitted by their customers to the sequences of known pathogens. First- generation software currently exists35 and is being used by several firms today. Firms that supply synthesized DNA could be required to use “certified” software that compares the sequence of submitted DNA orders to those of known pathogens.
As mentioned previously, commercial DNA synthesis falls into two rather distinct products: 1) synthesis of short oligonucleotides, typically up to about 100 to 150 base
35 E.g., BlackWatch, developed by Rob Jones of Craic Computing, Seattle, Washington. Discussions on this and related approaches at Workshop 1 of this project based on commissioned paper from R. Jones.
27 Nov 2006 Synthetic Genomics: Options for Governance 22 DRAFT ONLY: DO NOT CITE OR CIRCULATE pairs long and 2) gene-length synthesis, producing pieces of DNA hundreds to thousands of base pairs long. Designing a screening system that is effective—both technically and administratively—for screening shorter, oligo-length pieces will be more of a challenge than designing one for gene-length pieces of DNA. Identifying the organism to which it belongs is significantly more difficult for a short piece of DNA than a full-length gene. Moreover, since oligos are used in a wide variety of different applications, the sheer volume of production of oligos far exceeds that for synthesis of genes and genomes. Again, fortunately, synthesizing a pathogen is more difficult and more time-consuming when starting with oligos than with gene-length pieces of DNA.
Thus, screening could be required for only longer sequences (for example, greater than 500 base pairs) or for all commercially synthesized DNA, regardless of length.
I-2. People who order synthetic DNA from commercial firms must be certified as legitimate users by an Institutional Biosafety Officer or similar “Responsible Official”
Rather than placing the responsibility for screening on the firms that synthesize DNA, the responsibility for approval could be shifted to the research institutions in which scientists work. In particular, under this option, staff that place orders for synthetic DNA must be first certified as legitimate users by that institution’s biosafety officer. In order to accept an order, commercial firms would need to see that the individual researcher had been approved to place such orders by a registered institutional biosafety officer.
The institutional biosafety officer would not have to screen each shipment for hazard. Rather, the biosafety officer would merely certify that the person ordering the DNA is a legitimate user of synthetic DNA. Such approval might need to be reviewed once per year and might be linked to biosafety certification or training requirements. A list of certified researchers could be maintained and updated electronically so that individual orders could be approved with minimal time delay.
Note that this approach is somewhat similar to that used by the American Type Culture Collection (ATCC), a nonprofit organization that stores and distributes biological materials such as cell lines, bacteria, animal and plant viruses, and antisera. ATCC will only ship potentially hazardous material with the approval of a registered biosafety professional. Likewise, shipments of radioactive materials can only be received by registered biosafety professionals.
A biosafety officer, at his or her discretion, might choose to screen individual orders as well, examining the research from the perspective of laboratory safety or potential harm to the environment. Clearly, however, this would add to his or her workload and slow down the approval process considerably.
27 Nov 2006 Synthetic Genomics: Options for Governance 23 DRAFT ONLY: DO NOT CITE OR CIRCULATE
I-3. Commercial firms must use approved software for screening orders; people who place orders must be certified by a registered institutional Biosafety Officer
Yet another option is to construct a hybrid approach, combining options A and B. Under this approach, to place an order for synthesized DNA, one would have to be certified as a legitimate user by a biosafety official, but commercial DNA suppliers would also be required to screen orders for hazardous sequences.
The biosafety official would be asked to certify researchers into two categories: 1) legitimate users of synthetic DNA, and 2) researchers with ongoing experiments with pathogens or with DNA that might come from a pathogen. In the event that the submitted sequence was identified by software screening as potentially hazardous, but the individual researcher’s experiment had not been approved by the biosafety official to use potentially hazardous DNA, the biosafety official would have to be contacted before the DNA order could be filled.
In addition, shipments of certain types of hazardous genes or portions of pathogens, instead of being shipped directly to the individual researcher, might be sent to the institutional biosafety officer or to the chair of the Institutional Biosafety Committee. (IBCs were created under the NIH Guidelines for recombinant DNA research to assess the biosafety and environmental risks of proposed rDNA experiments conducted in academic and commercial settings, and to decide on the appropriate level of containment.) For example, this procedure might be required for any piece of synthesized DNA, longer than 500 base pairs length, whose sequence matches one of those on the hazardous list.
I-4. Require commercial firms to store information about customers and their orders
A far more minimal approach would be to simply require commercial firms to store information about their customers and their orders. The Toxic Substances Control Act (TSCA) already requires firms to retain records, including the identify of the customer, for many types of chemicals and other substances (including, in some cases, DNA sequences) for at least 5 years, but it is not clear whether this is an appropriate mechanism for firms making synthetic DNA.
Commercial DNA suppliers would be required to register with a designated agency such as the FBI. Information about each order would be stored at the firm for a specified period of time and made available to the FBI on demand for surveillance purposes or for forensic analysis in the event of an attack. For example, once the pathogen had been isolated and sequenced, its sequence could be compared to orders for synthesized DNA to try to find a match.
To ensure that orders would only be allowed to be delivered to known street addresses (similar to the policies of FedEx). Even then, although a matching order might be identified, there would be no assurance that the person or group that had placed the order would not have already moved.
27 Nov 2006 Synthetic Genomics: Options for Governance 24 DRAFT ONLY: DO NOT CITE OR CIRCULATE
COMPARING THE OPTIONS
Options for Gene Foundries
The primary purpose for implementing any of the first three options above is to prevent a potential bioterrorist from obtaining DNA from a commercial firm. The fourth option, rather than focusing on prevention, might help law enforcement officials respond to an incident, should it occur.
Figure 1A below summarizes our judgments about how well each of the options would enhance biosecurity, if implemented at gene foundries, that is, firms that produce gene- and genome-length stretches of DNA. The figure also includes our evaluation of each option’s effectiveness for meeting two other important goals: improving laboratory safety and protecting the environment. Finally, the figure compares the options according to a series of other important considerations, such as the costs and difficulties of implementing each option. In a later section, we discuss the effectiveness of these options when implemented by firms that produce shorter oligonucleotides.
Relative Effectiveness for Achieving Goals
For preventing bioterrorists from obtaining long stretches of a potentially pathogenic genome, we judge Option I-3, the hybrid approach, to be most effective; followed by Option I-1, screening by approved software; and finally Option I-2, requiring that customers be certified as legitimate users of synthetic DNA by their institution’s biosafety officer. Option 3 melds the strengths of the other two options. Screening software will identify potentially harmful pieces of DNA, regardless of whether the intended use is nefarious or legitimate. Certification from an institutional biosafety officer is a simple way of determining whether the customer is a legitimate user of that potentially harmful piece of DNA—the overwhelmingly likely source of such an order.
Option I-3 is likely to be the most effective option for avoiding harmful laboratory accidents or harmful releases to the environment. Under this option, a biosafety officer would be notified if a user who is not certified to use pathogenic sequences in his or her research ordered one or more such sequence, either inadvertently or deliberately. Option I-1 (screening alone) might avoid some accidental orders of harmful sequences, but would be considerably less effective.
By requiring firms to store information about orders for several years and to supply that information to the FBI in the event of a bioterrorist attack (Option 4), it might be possible to identify the individual or group responsible for an attack using a synthetic organism. This option might also be used in the event of an accidental release of a synthetic organism to the environment. Such records would provide one of very few possible leads for identifying individuals or groups “after the fact”.
27 Nov 2006 Synthetic Genomics: Options for Governance 25 DRAFT ONLY: DO NOT CITE OR CIRCULATE
n fy ti n ee r o r e n ti c e a C e S t re l m t s c p r s u s o o u r S f M e t n M s Pe I rs rd u e s fy r ie e O M ti o r ic r t d ff e s e S n c O a m C t u ir t s o ty Pl F s u F : u s fe o M r e h id M s e n a r d s W b m r e o y O r S i O G rs Bi le H F t . e . p . B . u 1 d 2 3 d 4 - r - - n - Does the Option: Ia O Ia Peo Ia a Ia Abo Enhance Biosecurity by preventing incidents? ~}z|
by helping to respond? ———} Foster Laboratory Safety — — by preventing incidents? | |
by helping to respond? ————
Protect the Environment — — by preventing incidents? | |
by helping to respond? ———} Other Considerations: Minimize costs and burdens to government and }}|z industry? Perform to potential without additional research? }z~}
Not impede research? ~~}z
Promote constructive ———— applications?
Most effective for this goal. Key to Scoring: z Most effective performance on this consideration. ~ Relatively effective. } Moderately effective. | Somewhat effective. { Minimally effective.
— Not relevant.
FIGURE 1A: Summary of Options for Gene Synthesis Firms
27 Nov 2006 Synthetic Genomics: Options for Governance 26 DRAFT ONLY: DO NOT CITE OR CIRCULATE
Relative Effectiveness on Other Criteria
While effectiveness in achieving goals is extremely important, policy choices must be made with other criteria in mind. An option whose costs exceed its benefits, or that hampers legitimate researchers more than bioterrorists, is not likely to be chosen.
The bottom half of Figure 1A displays our judgments of the effectiveness of each of the four options in meeting a series of other important criteria. For example, the first row examines the costs and burdens of each option to government and to DNA synthesis firms. Note that the costs and burdens of the first three options are inversely proportional to their effectiveness for preventing a bioterrorist attack. We judge Options I-2 and I-3 to be somewhat burdensome, and while we believe that Option I-3 would be the most effective for preventing a potential attack, it is likely to be the most the most costly and burdensome to implement.
Options I-1 and I-3, both of which rely on computerized screening of orders, will require several additional components to work effectively. First, the screening software must be tested and certified to ensure that it meets minimum standards. The testing and certification might be done by a U.S. government agency such as the CDC, or perhaps by an advisory body that is sanctioned by that agency. The same agency would also be responsible for preparing the screening list, though again, this task might be delegated to an advisory group.
Second, commercial gene foundries would be required to register with a designated agency and certify that approved software is being used. That agency might perform periodic random tests to determine whether the software was, in fact, in use.
Finally, the FBI or a similar agency must establish a “hot line” for commercial firms to notify if a suspicious sequence is detected. That agency would need to establish thresholds of concern to determine when the hot-line should be called and when an order would be denied.
The need for additional research is a second important consideration listed in Figure 1. For Options I-1 and -3 to be effective, two technical improvements are crucial: better screening software and a tailored list of risky sequences against which orders will be screened.
The software itself must be improved to identify risky orders more effectively and efficiently. Both the error rate and the amount of additional human screening required must be reduced. Moreover, the current generation of software can be “gamed” by a particularly clever individual by slightly altering the DNA sequence in ways to avoid detection but not alter its intended function.
Improvement is also needed in the list of harmful genes and genomes to which the submitted sequence is compared. The current software relies on the Select Agent list, which was constructed for an entirely different purpose. This list needs to be extended to
27 Nov 2006 Synthetic Genomics: Options for Governance 27 DRAFT ONLY: DO NOT CITE OR CIRCULATE
include not only select agent genomes but also genomes that differ in minor ways from Select Agents yet produce functionally equivalent organisms. The list also needs to include individual genes of concern, for example, toxin genes (which are currently covered by Select Agent regulations) as well as those that confer properties designated as particularly relevant to biological weapons threats, such as resistance to known treatments such as antibiotics.
The Select Agent regulations also cover transfers of synthetic DNA or RNA within the United States if the genetic material can be expressed as a select virus or toxin. Facilities sending and receiving such materials must be registered with either the CDC or the Animal and Plant Health Inspection Service (APHIS) of the U.S. Department of Agriculture (for animal and plant pathogens), and each transaction must be reported.
Yet another important consideration is the extent to which each option impedes or burdens legitimate research while seeking to prevent illegitimate uses or accidents. All of the options fare reasonably well on this criterion. Software screening will increase the cost of gene synthesis somewhat, but not by very much. Several firms already screen today and remain competitive.
Certification of legitimate users by an institutional biosafety officer would add a new approval step. Each order would not have to be certified; rather individual researchers would be certified as legitimate users perhaps once a year. Within universities or other large research institutions with biosafety officers, this extra step would add some extra administrative burden, but could be readily accommodated.
The greatest impact would be felt by researchers working for small start-up firms that do not have a biosafety officer. A mechanism would have to be established to allow such scientists to be certified by independent consultants. If such a mechanism were too burdensome, small start-up firms might be compelled to shift to in-house synthesis instead.
None of the options are effective for promoting constructive applications, though there might be some modest benefit to the added interaction between researchers and biosafety officers.
Finally, it is worth noting that all of the options lose effectiveness without the participation of other countries, giving rise to the need for internationally harmonized regulations covering gene and genome synthesis. Import rules might be able to limit the amount of DNA synthesized in other countries that is shipped to the United States. But none of the options could address the potential problem of a synthesized pathogen that is smuggled across a U.S. border.
27 Nov 2006 Synthetic Genomics: Options for Governance 28 DRAFT ONLY: DO NOT CITE OR CIRCULATE
Options for Firms that Synthesize Oligonucleotides
Figure 1B displays our judgments about the effectiveness of requiring firms that synthesize oligonucleotides to adopt one or more of these options. Again, these firms synthesize and sell pieces of DNA typically shorter than 100 bases long. In general, while implementing these options at “oligo” supply firms will certainly add another layer of protection, the risk reduction per unit of effort would be lower for these firms than for suppliers of longer gene and genome stretches of DNA.
Relative Effectiveness for Achieving Goals
As can be seen by comparing Figures 1A and 1B, implementing Options 1 and 3 at oligo supply firms would be significantly less effective for preventing incidents of bioterrorism than the same option implemented at gene foundries. This is for two reasons: First, the shorter the piece of DNA, the lower the confidence that the particular sequence that was ordered is found exclusively in a pathogenic organism and is not present in a benign organism as well. Options 1 and 3 rely on computer software to distinguish potentially harmful from benign pieces of DNA. When the results are ambiguous, the only solution is to request a review of the data by knowledgeable staff. Next-generation software might be able to clarify these cases to some extent, but a degree of ambiguity is inevitable for very short pieces of DNA.
Moreover, because oligo-length stretches of DNA have many applications other than synthetic genomics, risk reduction per unit of screening effort will be low. For oligo synthesis, not only is the needle in the haystack that one is searching for shorter, but the haystack is larger as well.
Thus, for preventing potential bioterrorists from synthesizing a harmful organism from commercial oligos, we are hard-pressed to determine whether software-based screening (Option I-1) is superior to having biosafety officers certify legitimate users (Option 2). Option 3—combining the strengths of both Options I-1 and -2—is again clearly the most effective approach. Option I-1 and Option I-3—those that rely on screening—are the most effective for fostering laboratory safety and protecting the environment.
Relative Effectiveness on Other Criteria
The pattern of relative effectiveness of these options in meeting the other important criteria listed in Figure 1B generally follows that described above for the options implemented at gene foundries, but quite often implementing these options at oligo houses would be less effective or desirable.
The costs and burdens to industry will be higher at oligo supply firms than gene foundries on a per unit or per dollar of business basis because the “false positives” that must be resolved will be more frequent with shorter sequences. Similarly, computer-based screening options require more research and development to be effective for screening shorter-length oligos. The far wider variety of uses for oligos than for genes and
27 Nov 2006 Synthetic Genomics: Options for Governance 29 DRAFT ONLY: DO NOT CITE OR CIRCULATE
fy ti n r io t e n t s e a C e u t re l m s c p r M u s o o s r S f r M e t n e s Pe I r rs rd u e u fy r t e O M i o c c rt t fi e s fa f c e S u O m C t n a ir s s y t u a r t Pl F s s e e : u r M d f o d M e o r a h ri M s s rd ig O W b O m l o y ir O n le S t O e Bi H B F . e . p . . u 1 r 2 3 d 4 - - - n - Does the Option: Ib Sc Ib Peo Ib a Ib Abo Enhance Biosecurity by preventing incidents? }}~|
by helping to respond? ———} Foster Laboratory Safety { — { — by preventing incidents?
by helping to respond? ————
Protect the Environment ———— by preventing incidents?
by helping to respond? ———| Other Considerations: Minimize costs and burdens to government and |}|z industry? Perform to potential without additional research? |z}}
Not impede research? |||z
Promote constructive ———— applications?
Most effective for this goal. Key to Scoring: z Most effective performance on this consideration. ~ Relatively effective. } Moderately effective. | Somewhat effective. { Minimally effective.
— Not relevant.
FIGURE 1B: Summary of Options for Oligonucleotide Synthesis Firms
27 Nov 2006 Synthetic Genomics: Options for Governance 30 DRAFT ONLY: DO NOT CITE OR CIRCULATE
genomes means that many more scientists will be inconvenienced by regulations applied to oligo supply houses. Finally, none of these options are effective at promoting constructive applications by researchers.
II. Policies for monitoring or controlling equipment and reagents
DESCRIPTION OF THIS INTERVENTION POINT
In attempting to monitor or control the equipment, materials, expertise, or information needed to synthesize DNA, the most readily accessible intervention point would be at the level of a DNA synthesizer – a device that produces short segments of DNA (typically less than 100 bases in length) with specified sequences of the four DNA bases (abbreviated A, T, G, and C). DNA synthesizers automate the sequence of chemical reactions needed to add a specific base to an existing strand of DNA, repeating the process as many times as necessary with the appropriate reagents until the desired base sequence is complete. Commercial synthesizers range in size from that of a microwave oven to a refrigerator, cost anywhere from thousands of dollars (used) to over a hundred thousand dollars (high-end, new), and can typically produce tens to hundreds of different DNA sequences at a time. At least 15 firms in the United States and at least an additional seven worldwide sell new or refurbished DNA synthesizers. Tens of thousands of these machines have been manufactured; they are available not only from scientific supply vendors but also used on the aftermarket, including eBay.
A more recently developed synthesis approach produces tens or hundreds of thousands of distinct DNA sequences in parallel on the surface of a chip, growing each sequence one base at a time by controlling which base is added to the DNA strands in each region of the chip. Once the specified length (and thus sequence) has been attained, all of the strands are washed from the chip surface into a common solution. The technology required to produce DNA in this way is more specialized and, although commercially available, less widely available than are traditional DNA synthesizers. Customers typically order chips carrying a desired set of DNA sequences, rather than procure the equipment needed to make their own chips. Several firms worldwide make chips designed for DNA synthesis, but this is still a relatively specialized market.
In both synthesizer and chip-based production, additional processing is required to clean up the short oligonucleotides and assemble them in the proper order to form gene- or genome-length strands of double-stranded DNA.
DESCRIPTION OF OPTIONS
Methods to monitor or control DNA synthesizers, in increasing order of government control, include registration (formally notifying the government when acquiring or
27 Nov 2006 Synthetic Genomics: Options for Governance 31 DRAFT ONLY: DO NOT CITE OR CIRCULATE
possessing a DNA synthesizer) and licensing (requiring the government to grant permission before acquiring or possessing a DNA synthesizer). Registration or licensing of a synthesizer could also be made a requirement for procuring specialized raw materials (phosphoramidites) necessary for synthesis, key spare parts of synthesizers (such as the capillary tube assembly), and service contracts for synthesizers, which would make it more difficult to operate synthesizers that were not incorporated into this regime.
Any of these policies would assign to each synthesizer an official owner of record who would have responsibility for that machine, and any of the policies would help to make authorities aware of the presence of synthesizers and of the people or institutions that were conducting DNA synthesis. Discovery of a synthesizer that had not been registered or licensed would constitute prima facie grounds for suspicion.
These options would be intended to enhance security by impeding and by helping to expose illegitimate activity. They are predicated on the assumption that forcing individuals with illegitimate intent to obtain DNA synthesizers surreptitiously, to lie to governmental authorities, or to build their own synthesizers would complicate their planning, open up additional possibilities for detection, and provide unambiguous grounds for prosecution if caught. It also assumes that the use of an ostensibly legitimate synthesizer for illegitimate purposes might be detected or deterred more readily if all synthesizers were declared and accountable to specific owners of record. Finally, it assumes that the burden of registration or licensure on legitimate users of synthesizers would be acceptable.
In effect, these measures would serve as what the arms-control community calls a “confidence-building measure”—a measure that is meant to give an indication of good intent but that cannot provide reliable proof of compliance. One major difference between legitimate and illegitimate users of biology and biotechnology is that legitimate users should be willing to reveal their activities, at least qualitatively, whereas illegitimate users would seek to conceal theirs.
II-1. Registration of DNA synthesizers
Newly manufactured or imported synthesizers would be given unique identifiers, and manufacturers, importers, and distributors would collect and report to the government information about the purchasers of these machines. Criteria would also have to be developed to specify how and when custom-built synthesizers would have to be registered.
If such a regime were implemented comprehensively, it would have to include all existing DNA synthesizers, not just newly purchased ones. Provision would also need to be made for formally decommissioning machines as they were retired, and for re- registering them when they were sold or transferred. Failure to register might incur administrative or even criminal penalties, without the need to prove illegitimate intent.
27 Nov 2006 Synthetic Genomics: Options for Governance 32 DRAFT ONLY: DO NOT CITE OR CIRCULATE
Registrations could, but need not, be made a matter of public record. Providing public access to registration information would increase the transparency of activities involving DNA synthesis and give private citizens and interest groups some ability to monitor them. However, public access could also risk exposing proprietary information that a firm seeks to withhold from its competitors, such as its DNA synthesis capacity, and might invite unwarranted attention from certain pressure groups whose agendas extend well beyond whether or not regulations on DNA synthesis were being followed.
II-2. Licensing of DNA synthesizers
Similar procedures would apply as in the case of registration, with the additional element that the government could specify the criteria required of registrants and could deny licensure to those applicants who did not meet the criteria. Such a system would be similar in concept, if not detail, to the current system under which individuals must be granted permission by the U.S. government to have access to select agents.
Note that a regime that did not “grandfather” all existing DNA synthesizers raises the possibility that an individual or institution could be denied a license for equipment already possessed, making its continued possession illegal. Forcing the divestiture of property in this way could be considered a de facto government confiscation of property and might require appropriate compensation.
II-3. Licensing of Synthesizers, plus License Required to Procure Reagents or Services
Pharmaceutical companies use phosphoramidites to produce drugs such as AZT (a treatment for HIV infection) in amounts that are orders of magnitude greater than those needed for gene synthesis. One way to think about this problem is that much less than one kilogram of phosphoramidites would be sufficient to synthesize one copy of the genome of every person on earth, yet pharmaceutical manufacturers use tons of phosphoramidites per year. Such an overwhelmingly larger demand for phosphoramidites, or other controlled materials or services, would complicate any control regime: the non-synthesis users would have to be brought into the regime and required to register before getting permission to purchase the materials, and penalties would apply to those who re-transferred controlled commodities to unregistered users. However, it would be extremely difficult to enforce such a regime through material accountancy, which would require accounting for tons of material with a precision necessary to detect the diversion of the grams of material involved in DNA synthesis.
COMPARING THE OPTIONS
Relative Effectiveness for Achieving Goals
The options discussed above are intended only to enhance biosecurity. However, the security benefits are modest, since no such regime could have high confidence in preventing illegitimate synthesis. Figure 2 below summarizes the potential contributions of the various options to enhancing biosecurity.
27 Nov 2006 Synthetic Genomics: Options for Governance 33 DRAFT ONLY: DO NOT CITE OR CIRCULATE
ts s n u t l e n p g e t t, a n n e m R p e e i m m y ip ip u u Bu Equ f to o Eq Eq f f o o ed ion ir t g g u ra in in q t s s e is n n R g e e c c e e i i s R L L n . . . e -1 -2 -3 ic Does the Option: II II II L Enhance Biosecurity { by preventing incidents? |}
by helping to respond? {{{
Foster Laboratory Safety ——— by preventing incidents?
by helping to respond? ———
Protect the Environment ——— by preventing incidents?
by helping to respond? ———
Other Considerations: Minimize costs and burdens to government and z~} industry? Perform to potential without additional research? zz~
Not impede research? ~}|
Promote constructive ——— applications?
Most effective for this goal. Key to Scoring: z Most effective performance on this consideration. ~ Relatively effective. } Moderately effective. | Somewhat effective. { Minimally effective.
— Not relevant.
FIGURE 2: Summary of Options for Monitoring or Controlling Equipment or Reagents
27 Nov 2006 Synthetic Genomics: Options for Governance 34 DRAFT ONLY: DO NOT CITE OR CIRCULATE
Synthesizers are relatively small and, unless total control is imposed over the supply, easy to acquire and hide. It would be very difficult to ensure that all existing synthesizers were identified and brought into such a regime, just as it would not be physically difficult to possess, maintain, and operate an unregistered machine—at least to the extent that airtight controls could not be placed on the necessary raw materials. More significantly, synthesizers can be built from scratch, although with some reduction in throughput and efficiency compared with a purchased product. There are few externally observable indicators (other than supply of reagents) that would denote the existence or operation of an unregistered synthesizer. Therefore, requiring that materials or maintenance be provided only for synthesizers that had been or licensed would increase still further the difficulty of operating unregistered machines.
A more serious security liability than unregistered synthesizers, however, is the possibility that registered synthesizers could be used for illicit purposes, either by the registrants themselves (i.e., an apparently legitimate firm set up as a cover for illicit activity) or by individuals who have access to legitimately registered machines (e.g., employees of a firm or students at a university). It would be difficult for an owner of record or any governmental authority to monitor the usage of registered DNA synthesizers closely enough to detect such illegitimate activity. Moreover, any attempt to do so would likely mitigate much of the motivation for preferring in-house synthesis over contracting out for gene synthesis in the first place (i.e., ease of use, rapid turnaround time, and absolute confidentiality).
At the point when pathogenic viral genomes can be readily synthesized and converted into viable and infectious virus particles, access to synthesizers will be equivalent to access to pathogens from a safety and security perspective. It follows logically that synthesizers should then be subject to controls as stringent as those now applied to select agents. However, DNA synthesizers have utility in a great variety of applications that are completely unrelated to select agents, meaning that the burden on the research and commercial communities of select-agent-type controls on DNA synthesizers would be far more pervasive than the impact of the select agent rules themselves.
In establishing an owner of record for each synthesizer, these options would serve to make the operators of these machines more accountable for what is done with them. Therefore these policies may act in part to promote responsible use of the machines, helping to foster a climate in which laboratories operate more safely and accidental releases are minimized. However, any such biosafety benefits would be quite indirect.
Registering or licensing synthesizers would be of only minimal utility in responding to the accidental or deliberate release of an organism constructed with synthesized DNA. Unlike bullets, which can be associated uniquely with the gun that fired them, pieces of synthesized DNA cannot be attributed to a particular synthesizer. Authorities investigating the release of a biological agent possibly incorporating synthesized DNA might find a list of registered DNA synthesizers helpful in identifying the locations known to have the capability of synthesizing DNA. Nevertheless, such a list would reveal
27 Nov 2006 Synthetic Genomics: Options for Governance 35 DRAFT ONLY: DO NOT CITE OR CIRCULATE
nothing about possible covert synthesis capabilities that might exist in additional, unlisted locations, or whether commercially synthesized DNA might have been used.
With respect to biosecurity, a registration requirement would prohibit gene synthesis by unregistered users but would not constrain activities by registered users. Since anyone can register, this option would do very little to constrain illegitimate use. Licensing would potentially have greater biosecurity value than registration in that it would not only ban synthesis by unlicensed users but would give government the authority to limit who can be licensed. This ability could be important if the government had intelligence identifying those who might abuse synthesis. As with the Select Agent Rule, it would be possible to subject all individuals seeking routine access to a DNA synthesizer to a security vetting procedure, such as fingerprinting and checks against criminal and terrorist databases. In practice, however, it will probably be very difficult for the government to deny licenses on the basis of anything other than failure to meet published objective criteria (i.e., a criminal record, nationality in a “state sponsor of terrorism,” etc.) By way of reference, permission to have access to select biological agents has never been denied on the basis of possible terrorist affiliations, although there is no way of knowing whether anyone has been deterred from seeking access to select agents for fear of being turned down. Even so, the government may wish to have the legal authority to deny a license even if it cannot predict what circumstances would lead it to want to exercise that authority.
Relative Effectiveness on Other Criteria
The bottom half of Figure 2 is our evaluation of the effectiveness of each option with respect to other considerations. The costs and burdens to government and industry are minimized in a regime requiring only registration of synthesizers; licensing performs effectively on this consideration as well, though somewhat less so. In both cases, though, the paperwork and tracking issues are relatively straightforward compared to Option 3, requiring registration to purchase reagents. Because of the significant volume of reagents used in DNA synthesis that are also used in other ways (e.g., in pharmaceutical production), a registration requirement to purchase reagents would confer a significant burden on the agency that needed to track such registrations. To reduce this burden, however, it might be possible to issue waivers to pharmaceutical companies that use phosphoramidites exclusively for applications unrelated to DNA synthesis.
Both the registration and licensing of synthesizers would require little additional research. The identities of companies that manufacture and supply synthesizers is known. Although it might be helpful to have a list of the serial numbers of every synthesizer ever made commercially, at least initially such information would not be required. With respect to registration to purchase reagents, some research would be required to pinpoint the sources of both phosphoramidites and the chemicals used in DNA synthesis. Because many of these raw materials come from outside the United States, it might be difficult to compile a complete list.
27 Nov 2006 Synthetic Genomics: Options for Governance 36 DRAFT ONLY: DO NOT CITE OR CIRCULATE
Modest paperwork and record-keeping would be involved in transferring registration during the purchase, sale, or resale of synthesizers, and in verifying that purchasers of materials or services were legitimate users. Licensing could be considerably more complicated, depending on the information and processing time required by licensing authorities and the likelihood of a “false alarm” that incorrectly indicated an increased risk of illegitimate use. In either case, a registration or licensing requirement could impede research to some degree as these paperwork issues were dealt with.
Because the markets for synthesizers and reagents are inherently international, it is worth noting here that in its narrowest definition this scheme could be implemented solely with respect to synthesizers produced within or imported into the United States. However, a regulatory regime would become more useful (in terms of capturing synthesizer capability) the more widely it was deployed around the world, which would require that other countries enact equivalent policies. In that case, harmonized requirements among different national systems would be desirable both in terms of imposing equivalent burdens on researchers and manufacturers worldwide, and in minimizing the burden that vendors and distributors would face in tailoring export policies to specific destinations. It is not likely that the cost or burden of implementing a registration regime would be severe enough to force research in or out of certain countries, but it is conceivable that an onerous and inappropriate licensing process—particularly one that excluded significant numbers of applicants—could induce researchers to seek work in countries that do not impose such regulations.
A licensing regime may be more difficult to harmonize internationally than a registration regime, since different countries may adopt different criteria for who should or should not be licensed. Indeed, the United States bans access to select agents, without exception, to nationals of countries on the State Department’s list of “state sponsors of terrorism.” Such a list is unlikely to be acceptable to other nations. Also still to be determined is how internationally harmonized security standards for granting access to DNA synthesizers would be negotiated, e.g., within what institutional framework.
International harmonization would be more important to a regime that affected the dissemination of key reagents than it would be to a regime that regulated only the synthesizers themselves. Given that the producers of the key reagents used in DNA synthesis are located for the most part outside the United States, there would be little hope of effectively controlling access to these materials without the cooperation of the countries in which the suppliers are located. Of course, controls on reagents could be imposed on a strictly domestic basis in the United States for the purpose of providing an incentive to register U.S.-based synthesizers. If imposed unilaterally, however, such controls would not be effective in impeding those with unregistered synthesizers from acquiring the necessary reagents directly from foreign sources.
27 Nov 2006 Synthetic Genomics: Options for Governance 37 DRAFT ONLY: DO NOT CITE OR CIRCULATE
III. Policies for controlling publications and data for which open communication poses more risks than benefits
DESCRIPTION OF THIS INTERVENTION POINT
This set of policy options raises the question of whether and how controls on information related to the science and technology of synthetic genomics could or should be restricted to impede their utilization for malicious purposes.
Over the past few years, a number of groups have addressed aspects of this set of issues for biological research in general, and for genome databases in particular.
In February 2003, a group including the editors of several prestigious scientific journals issued a statement reiterating the importance of open scientific communication, but also acknowledging that “there is information that, although we cannot now capture it with lists or definitions, presents enough risk of use by terrorists that it should not be published.” 36 The group went on to conclude that “on occasion, an editor may conclude that the potential harm of publication outweighs the potential societal benefits. Under such circumstances, the paper should be modified, or not be published.”37 Drafters of this statement did not give government a role in making this determination, but rather assigned this responsibility to editors, publishers, and researchers themselves.
This idea was carried forward in a study by a National Academies’ panel of experts, called the Fink Committee after its chairman, MIT biologist Gerald Fink. Reporting out in October 2003, the Fink Committee recommended “relying on self-governance by scientists and scientific journals to review publications for their potential national security risks.”38 This recommendation endorsed the statement from the editors and publisher’s group, but did not provide any guidance for what to do with information that was judged should not be published. The Fink Committee rejected the creation of a category of “sensitive but unclassified” information in the life sciences, stating that the risks “of a chilling effect on biodefense research vital to U.S. national security as the result of inevitably general and vague categories is at present significantly greater than the risks posed by inadvertent publication of potentially dangerous results.”
One of the other recommendations of the Fink Committee was that the United States government should establish a national advisory board to facilitate dialogue between the scientific community and the national security community, to advise the U.S. government on any regulatory matters involved in addressing security concerns posed by fundamental
36 “Statement on Scientific Publication and National Security,” Journal Editors and Authors Group, Science 299: 1149. 37 Ibid. 38 Committee on Research Standards and Practices, op. cit. at 6. p. 7
27 Nov 2006 Synthetic Genomics: Options for Governance 38 DRAFT ONLY: DO NOT CITE OR CIRCULATE
biological research, and to serve as a resource for the self-governance aspects of such a regulatory regime.
The United States government implemented this recommendation in March 2004 with the creation of the National Science Advisory Board on Biosecurity (NSABB), which in turn formed a subgroup on communications to address issues such as the review of security- sensitive publications in the life sciences. The communications subgroup recognized that the communication of scientific research involves many means and stages other than formal publication of final results. Starting with a reiteration of the importance of open and unfettered sharing of information and technologies for validating and advancing scientific research and contributing to beneficial applications, the subgroup went on to develop points to consider for identifying and assessing the risks and benefits of communicating research information. It also formulated options for the content, timing, distribution, and/or context of research information that posed security concerns.39
The NSABB subgroup’s guidelines are addressed not to government but to researchers, research administrators, educators, and the scientific publishing community; they do not envision or assign a government role in reviewing or restricting publication. Among the dissemination options that are explicitly mentioned are limiting access “to selected individuals on a ‘need to know’ basis,” along with the identification of categories of individuals who should have access and under what circumstances; or recommending that “the product not be published or otherwise made accessible to the public.” Note that limiting access of the information to selected individuals constitutes the creation of a category of “restricted” information, which, if the government had sufficient authority and legal rights to the information, could include placing it under security classification.40
Also in October 2003, as the Fink Committee’s report was being released, a workshop was held at the National Academies to address concerns posed by the potential for misuse of genome sequence data, and to examine policies governing access to such databases. The report summarizing and elaborating on this workshop argued against any kind of monitoring of or restrictions to access, concluding that “rapid, unrestricted public access to primary genome sequence data, annotations of genome data, genome databases, and Internet-based tools for genome analysis should be encouraged.”41 This conclusion was motivated as much by the practical difficulties in limiting access to genome data as by the judgment that such limitations would be undesirable.
39 See “Tools for the Responsible Communication of Research with Dual Use Potential,” section 2 (pp. 7- 15 of NSABB Draft Guidance Documents, prepared for (and, with minor modifications, approved at) the July 2006 NSABB meeting; available at http://www.biosecurityboard.gov/pdf/NSABB%20Draft%20Guidance%20Documents.pdf 40 With the exception of certain nuclear weapons-related information and patent secrecy orders, the U.S. Federal Government cannot classify information that it does not have ownership rights over through means such as its having been created by federal employees or with federal resources, or acquired under agreement with the federal government. 41 Committee on Genomics Databases for Bioterrorism Threat Agents, op. cit. at 9. p. 7
27 Nov 2006 Synthetic Genomics: Options for Governance 39 DRAFT ONLY: DO NOT CITE OR CIRCULATE
DESCRIPTION OF OPTIONS
Any option for restricting the dissemination of information would start with a risk/benefit assessment such as was laid out by the NSABB Subgroup on Communications. These guidelines would lead reviewers to consider the reasonably anticipated risks of disseminating the information in question, the benefits to science and to public health of disseminating the information, and, after weighing the two against each other, the options regarding communication.
The challenge will come in applying this process to any specific case, where the ability to predict either risks or benefits may be inversely proportional to the true significance or power of the innovation. Moreover, those research findings that offer the greatest potential benefit may also be those with the greatest potential for abuse. At any rate, although it may be very important to have the ability to restrict dissemination in cases that clearly merit such an outcome, such cases are likely to be extremely rare. In 2003, the American Society for Microbiology implemented a process to screen manuscripts submitted to its journals for biosecurity considerations. In over two years, only three of more than 16,000 manuscripts received additional scrutiny on security grounds, and modifications were made to only one. None were withheld from publication.42
To the extent that results from some line of research can be anticipated to raise questions about future dissemination, it would be far preferable to address those questions at the proposal stage, going through the biosafety and biosecurity review procedures similar to those discussed in other options presented here. With a working system of proposal review in place, any controls on communication would need to be introduced only upon the discovery of unanticipated results that raised serious security concerns.
In the longer run, questions can be raised about the efficacy of any policies that are overlaid on a model of scientific communication that relies on “gatekeepers” such as peer reviewers, editors, and publishers. It is already acknowledged that scientific communication happens throughout the research process, including stages before a paper is ever submitted for publication such as writing proposals, conducting collaborative research, informally circulating results via email and the web, and presentations at conferences. In the future, it is possible that scientific communication will increasingly evolve from one based on pre-publication review to a “Wikipedia” type mechanism where results are circulated worldwide immediately and are reviewed and vetted afterwards in the form of subsequent comments and postings.43 Such a trend would place sole responsibility for what to communicate on the individual scientist and would negate the role of subsequent filters.
42 2006. National Science Advisory Board for Biosecurity. March 30th. http://www.biosecurityboard.gov/meetings/NSABB%20March%202006%20meeting%20minutes%20- %20Final.pdf , p. 21. 43 http://openwetware.org
27 Nov 2006 Synthetic Genomics: Options for Governance 40 DRAFT ONLY: DO NOT CITE OR CIRCULATE
Options for less-than-open dissemination of research information and technology differ on what type of review and advice is provided, and what kind of dissemination would result for information that was withheld from public distribution.
III-1. Self-Governance of Risky Information by Scientists and Journals
Investigators in the past have chosen to discontinue lines of work, destroy specimens, and forego publication of research results that appeared to pose dangers that were not sufficiently compensated by beneficial applications.44 Such a decision would require a judgment that dissemination of the results would have a greater chance of facilitating serious harm through accidental or intentional misuse than they would have of, for example, contributing importantly to curing natural diseases. Deciding to withhold formal dissemination would also require knowing with some confidence the full extent to which individuals already have access to the information, and therefore would have to agree to participate in any decision to forego future use.
This individual self-governance framework broadens to include scientific community institutions such as publications when the Journal Editors and Authors Group statement and the NSABB subgroup guidelines are taken into account. Such community self- governance also includes the possibility that a researcher may disagree with a journal’s decision to seek modifications, or restricted dissemination, of a paper. Since no binding authority or adjudicatory mechanism exists to resolve disputes or enforce decisions, the effectiveness of any such decision depends on those arguing for restriction being able to make a clear and compelling case.
III-2. Self-Governance, Assisted by a National Advisory Group
Option 1 above could be augmented by the designation of a national-level group of experts that could be convened on a confidential basis to advise on the publication of research that raises particular security concerns. The ASM experience suggests that such cases would not arise very often, and that the workload on such a board would not be overwhelming despite the tremendous number of scientific publications in biological science each year.
The NSABB has served in precisely this role in reviewing publication of papers on the reconstruction of the 1918 flu virus, and its interdisciplinary mix of scientific, security, and other areas of expertise brings precisely the right sets of expertise to bear on this question. The group envisioned in this option is likely to be the same group envisioned by option 6 in the section “Users and Organizations.” This might be the current NSABB, or a successor body convened for that purpose.
Such a board in general would have no binding authority to enforce its decisions, which would raise Constitutional questions among others. However, if convened as a federal
44 There have been some reports of “accidental” or “incidental” human cloning in non-federally funded embryonic stem cell research in which investigators have reported destroying the product of the experiment.
27 Nov 2006 Synthetic Genomics: Options for Governance 41 DRAFT ONLY: DO NOT CITE OR CIRCULATE
body, it could recommend actions the federal government could enforce with regard to publications of federal employees. It could, however, develop an informal advisory mechanism to convey its advice to a set of journals to which the paper in question might be submitted to help ensure that none of them inadvertently undercut the decisions of another not to publish.
III-3. Restrict Access to Information Flagged by Self-Governance Process
Under this option, information that was determined by the self-governance process described above as posing unacceptable security risks if openly communicated, would remain available for legitimate scientific research into deliberate and natural biological threats, but in a restricted environment. Note that this option was specifically rejected by the Fink Committee, which argued against the creation of a category of “sensitive but unclassified” information that would have limited the distribution of certain scientific results. They argued that no system for disseminating this information now exists, and that creating one would be a daunting task. Indeed, creating such a system would implicitly redefine what science is, since science is predicated on the ability to review and validate research conducted by others.
Were implementation of such a system to be attempted, it would require, in addition to determining what type of information should be so controlled, the following:
a. Identification of a process for creating and granting access to the restricted database. Where would this data reside? Who would have control over it? What would be the procedure for providing access to it, given that once the restricted information was released to the public (whether accidentally or intentionally), it could never be taken back?
b. Identification of procedures to be following in making use of restricted database. What kind of conclusions/research/underlying analysis based on such restricted information could be published? How would scientific results drawing on this information be validated? How could restrictions on further dissemination be maintained once the data had been released, even on a restricted basis?
c. Evolution of restrictions. In the category of this option where this information is retained—restricted dissemination—it would be important to specify how long such restrictions would have to be maintained. Items in this database would need to be reviewed periodically to determine when the rationale for restricting their dissemination ceased to exist. (This type of review is supposed to take place with classified information, including the assignment of a “declassify by” date at the time of its classification. An equivalent review process would need to be applied to any “sensitive but unclassified” designation.)
27 Nov 2006 Synthetic Genomics: Options for Governance 42 DRAFT ONLY: DO NOT CITE OR CIRCULATE
Placing any scientific information in a restricted “sensitive but unclassified” database would be both rare and newsworthy. Therefore, unless such an action could be carried out very quietly, it would be counterproductive to do it at all, lest the attempt to do so shine a spotlight on precisely that information that should not be widely disseminated. Moreover, placing restrictions on scientific information would make sense only if it had not already been widely disclosed through informal means.
COMPARISON OF THE OPTIONS
Relative Effectiveness for Achieving Goals
These options only apply to the goal of enhancing biosecurity, and specifically only to preventing an incident. All of these options will be difficult to implement in an era where scientific results are communicated in many more modes, and much more quickly, than when formal research publication was the primary means of communication. The options differ in how effectively they might be implemented, what uses might be made of the information that is not disseminated widely, and what the possible adverse consequences of the option might be. The options are summarized in Figure 4 below.
All of the options assume that information to be withheld from the public domain is flagged by a self-governance process. In the latter two, this self- governance process is advised by others as well. Presumably, the same papers will trigger review and possible restriction in each of these cases, meaning that the options do not differ with respect which information would be withheld from the public domain. It is assumed, based on the ASM experience, that any such decisions would be very rare. Therefore, all of these options are judged to be only “somewhat effective” in preventing security incidents (i.e. the misuse of the withheld information), since very little information in the end is likely to be withheld. The third option, in which restricted research would proceed in “sensitive” channels, provides the possibility that the restricted information could nevertheless be utilized in biodefense applications, and are hence judged “minimally effective” at enabling a response to security threats. In the other cases, where no provision is made for continuing the research in a closed environment, no credit is given to the ability to respond.
Relative Effectiveness on Other Criteria
The options for controlling sensitive information would impose varying burdens on the government, and on research communities, including the firms supplying them. The self- governance process common to all of them presumes that scientists, editors, and publishers have arrived at an awareness of the possibility that research information might contribute to misuse, and that some modest effort has gone into institutionalizing reviews for such information within the publication process. Given the loss to science of any
27 Nov 2006 Synthetic Genomics: Options for Governance 43 DRAFT ONLY: DO NOT CITE OR CIRCULATE
y n b o ti d a te m d is e n s for c a s n n f s I a o t A p n s , u o r e ti e o t e c n c r s v n e n G s o a i a e n c n y c G r S r r c f- e e o l v y v is A e b t o o v c S n d i y G G tr f- io f- A b t s l s el a l el a e d s S a S n R e s m n o g e . r r . i . g c 1 o u 2 t 3 o I- f o I- a I- la r Does the Option: II In J II N II F P Enhance Biosecurity by preventing incidents? |||
by helping to respond? ——{
Foster Laboratory Safety ——— by preventing incidents?
by helping to respond? ———
Protect the Environment ——— by preventing incidents?
by helping to respond? ———
Other Considerations: Minimize costs and burdens to government and z~| industry? Perform to potential without { additional research? ~z
Not impede research? ~~}
Promote constructive ——— applications?
Most effective for this goal. Key to Scoring: z Most effective performance on this consideration. ~ Relatively effective. } Moderately effective. | Somewhat effective. { Minimally effective.
— Not relevant.
FIGURE 3: Summary of Options for Controlling the Distribution of Information for Which Open Communication Poses More Risks than Benefits
27 Nov 2006 Synthetic Genomics: Options for Governance 44 DRAFT ONLY: DO NOT CITE OR CIRCULATE
information that is not fully communicated, and the potential chilling effect that the existence of a review procedure might have for certain areas of legitimate science anticipated to raise questions, all of these options are assessed at imposing at least a small cost on future scientific research (and therefore score no better than “relatively effective” at “not impeding research.” A “sensitive but unclassified” approach (option III.3) would be sufficiently complicated, that we judge it to be the most likely to impede research.
The self-governance alone approach (option III.1) imposes no additional costs to government and little to industry, with the possible exception of the scientific publishing industry. Establishing a national advisory group (option III.2)—which may be the NSABB, or may be the national-level group referred to in option IV.6—would impose some additional cost to government, notching down the “minimize costs and burdens to government and industry” one step.
The sensitive-but-unclassified approach (options III.3) will result both in more burden for government (which must oversee this type of research) and for industry (which would have to adhere to new rules). New mechanisms for making information available on a restricted basis would have to be established, thus this option is judged only “somewhat effective” at minimizing costs.
Since no “sensitive but unclassified” mechanism now exists across the international community, creating such an option would involve the greatest amount of additional research. Protocols and mechanisms for conducting classified research already exist; but managing “risky” information in the “sensitive but unclassified” environment” would require further discussion between the scientific and engineering communities and those who would be responsible for enforcing the information controls.
The globalization of science poses additional questions for option III-3. One of the key attributes of a restricted “sensitive but unclassified” database for the scientific community would be that access would be granted on the basis of scientific expertise and standing, not citizenship. No such internationally accepted system exists today, and any such system would be considerably harder to construct than a system based on nationality. Moreover, each national authority could make such determinations in its own way, hampering international harmonization.
Implementing a self-governance mechanism among scientists, editors, and publishers raises significant international concerns, since the relevant communities in different countries do not yet view deliberate biological threats in the same way. Coming to a better understanding of these issues internationally, if not harmonizing national approaches, is currently one of the tasks before the NSABB.
27 Nov 2006 Synthetic Genomics: Options for Governance 45 DRAFT ONLY: DO NOT CITE OR CIRCULATE
IV. Policies for the roles of users and organizations in promoting safety and security in the conduct of synthetic genomics protocols
DESCRIPTION OF THIS INTERVENTION POINT
The focus of this intervention point is the user of DNA synthesis technology and the institutions, organizations, and extra-institutional communities that support and oversee such work. Unlike the options identified for other intervention points, these options apply directly only to legitimate users of DNA synthesis technology; indirectly, these options may have positive implications for biosecurity. Collectively, the options address how users are trained, how the safety of their work is judged, and how standards of practice can be informally or formally enforced.
Legitimate researchers carry out their work with the assumption that they are pursuing constructive lines of inquiry and that their research will seek to benefit individuals and society as a whole. In order to carry out their work, scientists and engineers (especially, although not exclusively, at universities) have at their disposal a number of support mechanisms that provide guidance or enforce rules. With appropriate training and practice, investigators know the resources available to them and which rules they need to follow. By utilizing the potentially close relationships between researchers and the bodies that guide them, it might be possible to develop safer laboratories and reduce the risk that synthetic pathogens could be released accidentally from a laboratory. Although some elements of these options apply to all microbiology laboratories, they also address the unique features of synthetic genomics.
Specifically, we have identified six options that address the needs of researchers in fulfilling their roles as responsible scientists or engineers. This set includes self- governance options by the scientific community, such as the education of trainees by senior researchers, but it also contains options that rely on outside involvement in governance. Some of the options include some type of penalty for non-compliance, but many of them provide guidance that legitimate researchers would be expected to follow during their professional research activities.
The first three options (education, a safety manual for synthetic biology, and a clearinghouse mechanism for best practices) are predicated on the involvement of institutions and/or individual experts outside the immediate community of synthetic biologists (e.g., university administrators, the CDC or the NIH, and the as-yet undetermined group that would run the clearinghouse).
27 Nov 2006 Synthetic Genomics: Options for Governance 46 DRAFT ONLY: DO NOT CITE OR CIRCULATE
DESCRIPTION OF OPTIONS
IV-1. Education about Risks and Best Practices as Part of University Curriculum
A critical component of any scheme that requires the knowledge and use of guidelines (or adherence to rules or regulations) is to educate scientists about these measures and how they should be used. Further training in research ethics, the societal implications of science, and even related aspects of law could be helpful. Although the precise mechanism could vary, the general approach would be to educate students about biosafety and biosecurity issues at the same time that they are being introduced to experimental concepts in synthetic genomics. Training in laboratory best practices could be provided for undergraduates, graduate students, and even faculty who have been working in other fields and now wish to conduct research in synthetic genomics.
One issue that would have to be taken up in any synthetic genomics curriculum is that of “dual-use knowledge” as discussed in the Fink Report. If students are not made explicitly aware that contributions to the open scientific literature could, in whatever distant and unlikely way, facilitate the nefarious plans of a bioterrorist, then they will not understand the need for biosecurity measures such as the screening of DNA sequence orders or keeping an eye out for suspicious activities in the laboratory.
Continued improvements in DNA synthesis technology will lead to dramatic increases in the amount of DNA being synthesized and a rapid increase in the diversity of users of the technology. Today most users of DNA synthesis technology are research professionals who work at well-funded commercial organizations (e.g., biotechnology and pharmaceutical companies). Many of these professionals lack access to continuing education programs that could inform them about the social implications of synthetic genomics and its governance.
As a second example, the student synthetic biology competition called iGEM (International Genetically Engineered Machines)45 has been expanding at a rate of about 300 % per year. The 2006 event, the third one held in three years, drew about 380 students from around the world. When surveyed, only about 1 % of the participating students said that they were aware of the 1975 Asilomar Conference on recombinant DNA research, the reasons for the conference, or the research oversight framework that resulted. The fact that iGEM students are young and, if already in college, tend to major in physics, history, graphic design, mathematics, or engineering—and only occasionally in biology—helps to explain this lack of knowledge.
Finally, it is worth emphasizing that the level of technical expertise required to use DNA synthesis technology is extraordinarily low and getting lower. For example, any laptop computer can be used to access public DNA sequence databases on the Internet, download and use free software for editing DNA (including sophisticated protein design software), place an order for DNA synthesis on a website, and arrange for rapid delivery by overnight mail. Thus, it is already naïve to expect that all well-intentioned users of
45 http://www.igem2006.com
27 Nov 2006 Synthetic Genomics: Options for Governance 47 DRAFT ONLY: DO NOT CITE OR CIRCULATE
DNA synthesis will have completed a degree program in biology, biological engineering, or a related field whose curriculum mandates some form of biosafety and biosecurity training. Moreover, for individuals whose education does provide such training, there is little opportunity to update this knowledge as technology and best practices change.
IV-2. Compilation and Use of a Manual for “Biosafety in Synthetic Biology Laboratories”
Several laboratory manuals and guides already exist for use both by researchers and the Institutional Biosafety Committees responsible for the safe conduct of research at their institutions. For example, the NIH issues guidelines for the safe handling of recombinant DNA,46 and the CDC publishes a handbook, Biosafety in Microbiological and Biomedical Laboratories (BMBL)47 that covers general laboratory safety. Several clinical laboratory guides are available. Recently, the World Health Organization (WHO) published a guidance document on laboratory biosecurity to complement its existing manual on laboratory biosafety.48
All of these documents will continue to be useful for several aspects of work in synthetic genomics. However, there are a few defining aspects of that type of work that the existing documents do not address. One major concern is with multi-source chimeras that are assembled from the DNA of hundreds of different organisms (in contrast to the tens of sources used by existing genetic engineering methods) as well as entirely novel synthetic constructs. In both cases, it is not known whether and to what degree chimeras containing DNA from many different sources could be pathogenic, regardless of the lack of pathogenic starting material. The data on this topic are mostly anecdotal: experiments using recombinant DNA have been conducted for upwards of 30 years and to date there is no evidence of emergent pathogenicity. At the same time, there has been little study of the emergence of pathogencity as a result of the recombination of pieces of nucleic acid. With respect to the design and construction of totally novel viral genomes, there is virtually no data indicating how one could make (or avoid making) a pathogen.
For this reason, it might be desirable to develop a new set of biosafety guidelines for researchers working with a large number of synthetic genes, or entire genomes. Certainly, the existing biosafety guidance could be modified to cover synthetic genomics and synthetic biology. However, given that synthetic genomics differs in several respects from current genetic engineering techniques, it would seem to be worthwhile to prepare a new biosafety manual, even if it incorporates large verbatim sections of the BMBL or another existing set of guidelines.
46 National Institutes of Health, 2002. NIH Guidelines for Research Involving Recombinant DNA Molecules. 47 Centers for Disease Control and Prevention. 1999. Biosafety in Microbiological and Biomedical Laboratories. 4th Edition. 48 World Health Organization, “Biorisk Management: Laboratory Biosecurity Guidance,” WHO/CDS/EPR/2006.6 (Geneva, Switzerland: WHO, September 2006).
27 Nov 2006 Synthetic Genomics: Options for Governance 48 DRAFT ONLY: DO NOT CITE OR CIRCULATE
By default, a new biosafety manual for synthetic biology might be drafted at the CDC and NIH, which are responsible for the current BMBL. Other agencies might also want to be involved in drafting and updating such a document. Certainly, a non-governmental organization such as the American Biological Safety Association could make an important contribution to such a document, either independently or in collaboration with or under contract to the CDC. Irrespective of which agency leads such an effort, a critical component would be the need for continual review and updating. As outlined in earlier sections of this report, the science of synthetic genomics is changing rapidly, and the social context of the research is changing rapidly as well. The current BMBL is seven years old; for a new manual titled “Biosafety in Synthetic Biology Laboratories” (BSBL) to be effective, it would have to be revised in response to new data in a timely manner.
The contents of a BSBL might also serve as the minimum knowledge required for certification of researchers by their local biosafety officers for the purpose of being able to order synthetic genes and genomes from commercial firms. (See the section above on screening of synthetic DNA ordered from commercial suppliers.)
A BSBL could also be a critical component for the institutional expert review of experimental protocols. A well-written biosafety manual (and the training that would accompany it) could address in one place safety problems that are generic to molecular biology and those specific to synthetic genomics and synthetic biology. Because the new manual would be aimed at Institutional Biosafety Committees and biological safety officers, in addition to laboratory heads and individual researchers, it is likely that the guidelines would actually be followed.
A laboratory biosafety manual for synthetic genomics would focus exclusively on minimizing the physical hazards associated with such experiments and would not address the issue of dangerous knowledge.
IV-3. Clearinghouse for best practices
This option would provide a central source for information on laboratory best practices for synthetic genomics. In an emergency it could provide useful information (see below) but it would not serve as a reporting hotline or as part of an emergency response.
Several clearinghouses for best practices already exist and might serve as models for this option. For example, the National Fire Protection Association maintains a web site containing a comprehensive library of information and has a toll-free number for advice on technical questions.49 The University of Chicago has a clearinghouse for scientists trying to find the answers to questions about regulatory compliance.50
Who would run such an information clearinghouse would depend on a number of factors, such as whether or not a professional society of synthetic biologists is eventually
49 National Fire Protection Association. 2006. http://www.nfpa.org (checking with them on whether they have any ethics/caller protection issues) 50 University of Chicago. 2005. University Research Administration: Research Compliance.
27 Nov 2006 Synthetic Genomics: Options for Governance 49 DRAFT ONLY: DO NOT CITE OR CIRCULATE
established and whether the government decides that a clearinghouse would be a worthwhile addition to other regulations governing synthetic genomics.
IV-4. Broaden IBC Review Responsibilities to Consider Risky Experiments
The report of the Fink Committee51 included a list of seven experiments of concern. The text included the idea that IBCs could become more attentive not just to the biosafety implications of certain areas of research with DNA (“dangerous research”) but of the biosecurity implications of such research (“dangerous knowledge”).
There are at least three different biosafety concerns with respect to the process and products of synthetic genomics in the laboratory. First, similar to the situation with traditional recombinant DNA technologies, there are concerns about working with specific, identifiable pathogens. Next, and somewhat related, are concerns about working with chimeras that combine genes from different organisms (specifically when large portions of the engineered product are derived from a pathogen). Finally, concern has been expressed about the possible emergence of pathogenicity from splicing pieces of DNA from dozens of different source organisms. If many pieces of non-pathogenic DNA are combined in ways that have never occurred in nature, could this process possibly give rise to something dangerous?
Although the initial product of synthetic genomics projects is a strand of DNA that is chemically indistinguishable from any other natural or recombinant DNA, synthetic genomics has raised new concerns with respect to laboratory and environmental safety. These concerns relate both to the process (is there anything about working with synthetic DNA that is inherently different than working with natural DNA?) and the product (are products made from synthetic DNA likely to be more dangerous than products made with genetic engineering?).
The seven experiments of concern described in the Fink report include demonstrating how to render a vaccine ineffective, conferring resistance to therapeutically useful drugs, enhancing the virulence of a pathogen or rendering a nonpathogen virulent, increasing transmissibility, altering host range, enabling the evasion of a diagnostic or other detection, and enabling weaponization. While such experiments may strike investigators as extreme, many of them happen all the time in laboratories under different guises. For example, many gene therapy experiments seek to develop viral vectors that can evade the human immune system. This characteristic is clearly related to the concern of enhancing the virulence of a pathogen, yet many investigators (and IBCs) may not think of it that way.
As synthetic genomics technologies become more widespread, it will be vital for IBCs to identify and review experiments of security concern. Minimally, this could be the seven specific types of experiments mentioned in the Fink Report; local committees could decide to add other criteria.
51 Committee on Research Standards and Practices, op. cit. at 6.
27 Nov 2006 Synthetic Genomics: Options for Governance 50 DRAFT ONLY: DO NOT CITE OR CIRCULATE
IV-5. Broader IBC Review, plus Oversight from National Advisory Group to Evaluate Risky Experiments
For a variety of reasons it is preferable to conduct reviews of experiments and enforce the biosafety rules at the local or institutional level. Occasionally, however, an experiment will be so novel as to elude local expertise with respect to risk assessment, or it may be controversial for other reasons. Over the past three decades of recombinant DNA research, the NIH’s Recombinant DNA Advisory Committee (RAC), among its other duties, has provided oversight of experiments that for any reason could not be approved by local IBCs. The largest single group of such experiments has involved various gene therapy protocols.
A similar national oversight body might be called upon to review synthetic genomics experiments, particularly those involving the construction of chimeric microorganisms using DNA from many different organisms, an area where there is little precedent and hence a lack of local expertise for review. This new review and oversight body might also be asked for advice in cases where an experimental protocol might contribute directly to bioterrorism. Such a national body could be housed in a number of agencies. It could be the RAC itself, or the National Science Advisory Board for Biosecurity (NSABB) which, like the RAC, is run out of the NIH Office of Biotechnology Activities. Alternatively, the national oversight committee could be placed in a different agency within the Department of Health and Human Services such as the CDC, or it could be taken out of HHS entirely. Some experts would like to see a national oversight body for synthetic genomics within the Department of Homeland Security. There is no reason, however, that this body could not be outside of government. For example, it might be administered by a consortium of universities, with the voluntary participation of commercial biotechnology and pharmaceutical firms.
To make such an approach more or less equivalent to the current RAC system, there should be an additional level of review above the new national-level oversight body. In the case of the RAC, for example, the NIH director has final say on the approval of recombinant DNA research protocols.
IV-6. Broader IBC Review, plus Enhanced Enforcement of Compliance with Biosafety Guidelines
In addition to any institutional penalties against principal investigators who fail to comply with IBC rules, penalties can be levied against their institutions (usually universities) as well. These institutional penalties typically involve the revocation of NIH grants. According to a study by the Sunshine Project, however, the NIH Office of Biotechnology Activities has never revoked a grant to punish violations of the NIH Guidelines, so this potential sanction may have lost at least some of its credibility.52 Criminal penalties can be invoked as well, usually in cases where another individual is harmed. Beyond the more or less voluntary nature of compliance with the NIH Guidelines, investigators and
52 Sunshine Project. 2006. Survey of Institutional Biosafety Committees. http://www.sunshine- project.org/ibc/
27 Nov 2006 Synthetic Genomics: Options for Governance 51 DRAFT ONLY: DO NOT CITE OR CIRCULATE institutions are subject to certain legally binding regulations, including the Toxic Substances Control Act and the rules of the Occupational Safety and Health Administration, as well as tort liability.
Thus, institutions have an incentive to ensure that their researchers are well-trained, but after any initial training there is little follow up to insure that the rules as taught (either formally by an Environmental Health and Safety Office, or informally within the laboratory) are being followed. For the most part, this is because of a well-earned trust of researchers. Synthetic genomics, however, could produce new questions in biosafety that cannot be answered by researchers just by using their tacit knowledge and may require additional training beyond that initially offered on starting in a biology laboratory. This option proposes that biosafety rules and guidance relevant to synthetic genomics, both those that already exist and new ones that may be developed, ought to be strictly enforced.
There are few useful models to follow when it comes to the enforcement of biosafety guidelines. In general, the research community, including those responsible for oversight, has not yet learned how to punish “rogue” scientists who choose to operate outside of community norms.
COMPARING THE OPTIONS
All of the options discussed above are aimed at legitimate researchers. Specifically, they address biosafety (the safety of laboratory workers and the surrounding communities, and protection of the environment) and mechanisms for achieving it. In some cases, however, there may be indirect benefits for biosecurity. A summary table is found below in Figure 4.
Relative Effectiveness for Achieving Goals
For fostering laboratory safety (specifically, the safety of scientific workers) and protecting the surrounding communities and the environment, educating laboratory workers is of great importance. This option involves teaching workers how to avoid incidents and what to do in case one occurs. This could be both formal classroom teaching and training in the laboratory with an experienced researcher. The latter type particularly could be very effective in that much laboratory training occurs this way (one- on-one) and is valued by students in adopting the best laboratory technique possible. In the event of an environmental release, prior education is likely to be less effective, as it will most likely only cover generalities.
The education option could also have a small positive impact on biosecurity. How laboratories are maintained and secured is important for biosafety, but it can also contribute to preventing potential bioterrorists from obtaining dangerous biological materials by stealing them or using facilities to which they should not have access.
27 Nov 2006 Synthetic Genomics: Options for Governance 52 DRAFT ONLY: DO NOT CITE OR CIRCULATE
c t ti s s e e t e c h th ti g B n c i d y a rs n r e s S a P v lu in t s s O p k y p s t e w u i e B e w; o w, R f i e r e t a r v i i n t s o v G v u " f e e io s y o e e Re R r Re m b B i s o e " r u C s A o C i C rc r t o IB s B v B n fo a h I d I fo o r g n ie r r ti l o it A n a in e il e l e E a u b r d d d c a a a ib a a a d u s n L s n d e a le ro ro ro e y n tio c E ic M g C B o B B n . t . . . p . a . a c lo N -1 a -2 -3 -4 s -5 -6 h r io e y n Does the Option: IV P IV B IV IV R IV b IV E Enhance Biosecurity — — by preventing incidents? | ||}
by helping to respond? | ————— Foster Laboratory Safety by preventing incidents? ~ z }}~ ~
by helping to respond? ~ ——| —— Protect the Environment by preventing incidents? ~ z } }~~
by helping to respond? |~~| —— Other Considerations: Minimize costs and burdens to government and ~|}}}| industry? Perform to potential without additional research? ~||}}~
Not impede research? z~z}~|
Promote constructive { —— applications? z }|
Most effective for this goal. Key to Scoring: z Most effective performance on this consideration. ~ Relatively effective. } Moderately effective. | Somewhat effective. { Minimally effective.
— Not relevant.
FIGURE 4: Summary of Policies for Users and Organizations
27 Nov 2006 Synthetic Genomics: Options for Governance 53 DRAFT ONLY: DO NOT CITE OR CIRCULATE
The development and use of a new biosafety manual for synthetic genomics, or BSBL, scores very high both for preventing biosafety incidents and helping to respond to an environmental release. In the latter case, the manual would contain step-by-step instructions for dealing with an accident. Equally effective in the event of an environmental release would be telephone access to a good clearinghouse that would provide explicit instructions.
Broadening IBC review responsibilities to include the “Experiments of Concern” as defined by the Fink committee report will have a small impact on biosecurity. Combining the broadening of IBC review responsibilities with oversight by a National Advisory Group results in a somewhat higher score for preventing a biosecurity incidents. None of these will help in responding to incidents.
Broadening IBC review scores reasonably well (moderately effective) in preventing incidents that could harm either laboratory workers or those in the surrounding environment. We judge this option initially to be somewhat less effective in responding to incidents than education or the use of a good manual. However, when combined with oversight by a National Advisory Group or with enhanced enforcement, the combinations are relatively effective at preventing biosafety incidents. None of the options IV-4, -5,- or -6, in any combination, is particularly effective for responding to incidents.
Relative Effectiveness on Other Criteria
Although none of the options discussed above (except perhaps enhanced enforcement) could be implemented today without additional research, for many of the options, the additional information needed should not be too arduous to come by. For example, for educating new workers to the field, although there are no standard curricula, there are some examples of training programs for both the scientific and professional ethics aspects of research. For the use of a new manual BSBL does not yet exist, but other biosafety manuals with significant relevant information already do.
Once these issues have been clarified, it is possible that the options discussed above could make a significant contribution to enhancing the biosafety of research in synthetic genomics, but with varying impacts in other areas. These impacts could also depend on how the “additional research” questions are ultimately resolved. For example, depending on who is chosen to run the clearinghouse, it could either minimize costs to government and industry (and universities) or increase costs. Establishing education programs and preparing a BSBL would require some initial financial investment on the part of government, academia, and perhaps industry. If implemented effectively, however, these biosafety measures should minimize overall operating costs in the long run. Moreover, if a professional society of synthetic biologists were to be established and assumed primary responsibility for establishing these programs, they would entail essentially no costs to government or industry. Having local IBCs conduct reviews of proposed experiments of concern involving synthetic genomics would have a moderate effect on minimizing costs and burdens, as it could contribute toward preventing laboratory accidents.
27 Nov 2006 Synthetic Genomics: Options for Governance 54 DRAFT ONLY: DO NOT CITE OR CIRCULATE
The impact on promoting the constructive applications of synthetic genomics does not correlate directly with achieving the overall objectives of biosafety and environmental protection. Specifically, IBC reviews of experiments of concern would have a good payoff in terms of promoting constructive applications, but they would not be the best option for fostering laboratory safety or protecting the environment. On the other hand, the preparation and use of a BSBL would have little if any impact on fostering constructive applications, but it would have a positive influence on safety both inside and outside the laboratory.
From the standpoint of the user, virtually all of the review and oversight options would impede research to some extent. However, the availability of tools such as a BSBL or a clearinghouse would not impede the advance of synthetic genomics and might even accelerate progress by suggesting better ways to carry out research protocols; a National Advisory Group might clear the way for research as well by offering “gold standard” advice that might be difficult to come by on a local level.
Choosing a Portfolio of Options
Figure 5 includes our evaluations of all 20 options proposed in the previous sections. The challenge that faces decisionmakers is to choose a portfolio of options that will achieve the multiple goals desired.
Briefly reviewing, the top half of the table includes our judgment of how effective each option is for achieving three key goals: enhancing biosecurity, fostering laboratory safety, and protecting the environment. Increasing the number of options adopted will likely enhance the Nation’s ability to achieve these goals; however, no option is without downsides.
The bottom half of the figure includes rankings of how well these options perform on four additional important considerations: What costs and other burdens do they impose on government and industry? Can they perform to potential today or do they require additional research? Will it unduly impede progress in synthetic biology and other related research? Finally, does the option also help to promote constructive applications, rather than just prevent undesirable ones?
While we have provided our best judgments about the broad benefits and costs of each of these options, our ability to do so is severely limited for some of these goals and other considerations. For example, while we have pointed out that synthetic genomics would rarely be the preferred method used by a bioterrorist, we really have no way of judging the overall likelihood of such an event. Thus, we can only judge the relative effectiveness of the options for enhancing biosecurity, that is, how each option compares to others. Quantitative estimates of the added security provided by each option are just not possible.
27 Nov 2006 Synthetic Genomics: Options for Governance 55 DRAFT ONLY: DO NOT CITE OR CIRCULATE
Gene Foundries Oligo Manufacturers DNA Synthesizers Publications and Data Users and Organizations
n y y n if n if n ts s y e t t s l b io e r io r io n a t d r- p r e n t t e n t lu e d " u c t rn a n e s C e a s C e a n p g te a s v o u S t e le u t e le , a u s rm n ie t r l t r rm r rm e t t o i e s i r O p s c p M s c p n n J s o c k s ; G , s u o o u o o m Re f n ty o e u rs S f s rs S f p e e , s n s t w w y w M t e n r M t e n i y f s A I a e a B e r M e P I e P I u m m o t p n Ri f r r i ie o ie t s d s re s d s p p u s , u o r a o v s n s r r u y e r r u y e q i i B e ti e o t t s fo e v i v f r u f r E u u c c r u b e v e e ie O M ti o O M ti o o n s ve o io a R d r ice r t ct ice r t f q q t n e n G s o L e R R m d f e s a f e s o E E a i a e b s A e f c S f f c S d c y G "B y u C C l C c n m Ce t u m Ce t n f f n n r - A r r u O la r s O la r s o o e r S r o cc lf g o IB s B a B i t n s i t io ir e e n o o h I n I fo o ty P F s u a r ty P F s u t g g v y v is A e o s f l n ie r r F s s u b t S ti e l o g t io n e o : u M r M e e o : u M r ra in in q o o v c c a i n e li e t e e f d e d f d e t e n d i y a i ri d i d d E a h ri M s o r a h ri M s s s s G G tr c t u B b n s rd s rd i n n R f- io f- A b c n a a i a Na a d e W b O m ig O W b O m g e e l t l l s u a ic e s o s o o e io y ir O l io y ir O e e e a e d d r a t l e r n r y r c G s e S n e S e ic ic a e e c r B l H F t O e B l H F t s S S n R E P M C ti B o B b B n . e . p . B . u . . p . B . u R L L n . rm . o . g . t . th . . p . t . a e . . . e ti g 1 2 3 c 4 s 5 h 6 1 d 2 o 3 d 4 o 1 r 2 o 3 d 4 o 1 2 3 1 o 2 3 - s - n - a - - - h - r - e - n - b - c - e - n - b - - - ic I- f I- a I- a e y r e ig n Does the Option: Ia O Ia P Ia a Ia A Ib S Ib P Ib a Ib A II II II L II In II N II Fl IV B IV S IV P IV R IV s IV E Enhance Biosecurity { — — by preventing incidents? ~}z|}}~|~}z|}}~| |}|||| ||}
by helping to respond? ———} ———} {{{ ——{ | ————— Foster Laboratory Safety — — { — { — —————— by preventing incidents? | | ~ z }}~ ~
by helping to respond? ——————————————~ ——| —— Protect the Environment — — —————————— by preventing incidents? | | ~ z}}~ ~
by helping to respond? ———} ———| ——————|~~| —— Other Considerations: Minimize costs and burdens to government and }}|z|}|zz~}z~|~|}}}| industry? Perform to potential without { additional research? }z~}|z}}zz~~z ~||}}~
Not impede research? ~~}z|||z~}|~~}z~z}~|
Promote constructive —————————————— { —— applications? z }|
Most effective for this goal. Key to Scoring: z Most effective performance on this consideration. ~ Relatively effective. } Moderately effective. | Somewhat effective. { Minimally effective. — Not relevant. Figure 5: Summary Chart of All Options Discussed in the Report
27 Nov 2006 Synthetic Genomics: Options for Governance 56 DRAFT ONLY: DO NOT CITE OR CIRCULATE
Similarly, we can estimate the relative effect each option may have on slowing the progress of the field of synthetic genomics, but we are not able to takes this much further. While we believe the potential for the technology is high, we have no crystal ball that can tell us the future of the field with and without each option.
Decisionmakers will bring their own values, priorities, prior beliefs, and degrees of risk aversion regarding safety and security threats to the table. The relative emphasis among the goals and other considerations will vary, leading to different assessments of the desirability of each of the policy options. To help each decisionmaker choose a preferred set of options, we have constructed several portfolios that range from a modest set of controls on the new technology, to one that is quite aggressive.
Table 2 presents the mix of options within each of three illustrative scenarios. The options are again arranged by “intervention point”, that is, whether they apply to gene foundries, manufacturer of oligonucleotides, laboratory-scale DNA synthesizers, to the publications and data that enable synthesis, or the users of the technology themselves and the organizations in which they work. Note that we can construct many groupings that would use slightly different options, with slightly different outcomes. These three are presented as examples only.
The first portfolio is aimed at plugging the biggest holes in the current system of governance of synthetic genomics. The options included are those that we judge to provide the greatest benefits at the lowest cost and burden. The second and subsequent portfolios add options that will enhance biosecurity and biosafety, but the relative “bang for the buck”—the added benefit compared to undesired impacts—of these added options will be lower than those in the preceding portfolios. Each successive portfolio strikes a different balance between concern for the potential harm that might be posed by synthetic genomics versus concern about foregoing synthesis’ benefits or about imposing other costs on society.
Individual decisionmakers will prefer a different balance. That balance, of course, may shift through time as more is learned about the nature of both the risks and benefits. Thus the flexibility of the overall portfolio is another important consideration. Decisionmakers should expect that the program they adopt today will need to be reconsidered in several years time.
Again, the first portfolio includes those options that provide the greatest benefit at the lowest cost and burden—many of which are being done voluntarily today. For example, the first option listed in Table 2, Gene foundries must screen orders, is already being done by the majority of firms voluntarily. This option is aimed simply at the relatively small fraction (perhaps 25%) of U.S. firms that do not. The next two options, requiring both Gene foundries and oligo manufactures to store information, we believe is also being done today by many U.S. firms for business and regulatory reasons. Again the goal of these options is to ensure that all firms store their orders and that the FBI would be able to obtain such records in the event that an incident should occur.
27 Nov 2006 Synthetic Genomics: Options for Governance 57
DRAFT ONLY: DO NOT CITE OR CIRCULATE
Portfolio: Intervention Point Option 123 Gene Foundries Ia-1. Require Gene Foundries to Use Approved Software for z Screening Orders Ia-2. To Order From Gene Foundries, People Must Be Certified by an Institutional Biosafety Officer or Similar “Responsible Official” Ia-3. Gene Foundries Must Use Approved Screening Software; zz People Who Order Must be Certified by Biosafety Officer Ia-4. Gene Foundries Must Store Information about Customers and zzz their Orders Oligo Manufacturers Ib-1. Require Oligo Manufacturers to Use Approved Software for Screening Orders Ib-2. To Order From Oligo Manufacturers, People Must Be Certified z by an Institutional Biosafety Officer or Similar "Responsible Official" Ib-3. Oligo Manufacturers Must Use Approved Screening Software; z People Who Order Must be Certified by Biosafety Officer Ib-4. Oligo Manufacturers Must Store Information about Customers zzz and their Orders DNA Synthesizers II-1. Registration of DNA Synthesizers z
II-2. Licensing of DNA Synthesizers z
II-3. Licensing of Synthesizers, plus Licensing to Buy Reagents and Services Publications and Data III-1. Self-Governance of Risky Information by Scientists and zzz Journals III-2. Self-Governance Assisted by National Advisory Group zz
III-3. Restrict Access to Information Flagged by Self-Governance Process Users and IV-1. Education About Risks and Best Practices as Part of University zzz Organizations Curriculum IV-2. Compilation and Use of a BSBL (“Biosafety in Synthetic zzz Biology Laboratories”) Manual IV-3. Clearinghouse(s) for Best Practices z
IV-4. Broaden IBC Review Responsibilities to Consider Risky zz Experiments IV-5. Broader IBC Review, plus Assist from National Advisory zz Group to Evaluate Risky Experiments IV-6. Broader IBC Review, plus Enhanced Enforcement of IBCs z
TABLE 2: Summary of Portfolios
27 Nov 2006 Synthetic Genomics: Options for Governance 58
DRAFT ONLY: DO NOT CITE OR CIRCULATE
Education about the risks and best practices, is already occurring in some University curricula, but not many; this option is directed at the majority of students and researchers new to the field who may not have had rigorous biosafety training or have not had the opportunity to think through the potential societal impacts of the work they will be carrying out. Similarly, self-governance of risky information already occurs in the research community, but perhaps somewhat more tacitly than would be ideal. Finally, the development and use of a biosafety manual developed explicitly for synthetic biology laboratories (a “BSBL”) would make such information easily accessible to this new community of scientists and engineers.
The second portfolio adds several addition options to the mix. Again, the added benefit from these options compared to undesired impacts, will be lower than those in the first portfolio. Some decisionmakers will judge these to be useful additions; others may choose to forgo them.
For example, the second portfolio adds an additional option for both gene foundries and oligo manufacturers: orders can only be placed by legitimate researchers, as certified by a registered biosafety professional. As in the first portfolio, gene foundries must still screen their orders, but because of the lower effectiveness and increased burden of screening short pieces of DNA as compared to genes, oligo manufacturers are not required to do so.
The second portfolio also includes registration of DNA synthesizers. Though synthesizing a pathogen with only a laboratory synthesizer and the necessary reagents requires additional time and skills, it is nonetheless possible.
Organizations would shoulder a new burden and would be asked to broaden local institutional review of proposed research involving DNA synthesis to include implications for bioterrorism. National level reviews for bioterrorism and biosafety (both pre-approval of experiments and publication of results) could be introduced here to deal with issues that are not covered in a BSBL or where scientists or publishers may want additional guidance.
The options in the third portfolio begin to address concerns about biosecurity and biosafety that might never be encountered by most legitimate users, or that may be considered to be unduly burdensome. This portfolio requires licensing of synthesizers, rather than just registration. A requirement for oligo houses to screen their orders (under the hybrid option) is introduced here, as its technical feasibility remains unclear. A clearinghouse would be added to augment many of the topics included in a BSBL. Finally, enhanced enforcement of biosafety guidelines is included to increase the effectiveness of either current or expanded IBC reviews.
27 Nov 2006 Synthetic Genomics: Options for Governance 59
DRAFT ONLY: DO NOT CITE OR CIRCULATE
(End of report, page intentionally left blank)
27 Nov 2006 Synthetic Genomics: Options for Governance 60