Back to the future: Controlling synthetic life science trade in DNA sequences

Abstract

In recent years there has been an enormous growth of interest in synthetic genomics and synthetic biology, which we collectively refer to as the synthetic life sciences. Rapid progress in this field has enabled the synthesis of biomolecules, whole genomes, and even simple life forms, raising hopes for the development of new bioproducts capable of addressing a wide range of ecological, technological, and biomedical challenges. However, the synthetic life sciences also pose a number of biosecurity and biosafety risks. Numerous regulatory options for the control of synthetic life sciences have been advanced. In this piece, the authors discuss one of those regulatory options: control of trade in DNA sequences. After reviewing the most commonly advanced proposals for regulation of the DNA sequence trade, they consider whether a clearinghouse for centralizing the oversight of all DNA sequence ordering would provide a better means of regulating the DNA sequence trade. They conclude that though a clearinghouse could potentially provide a promising means of regulation, the technology required for an effective clearinghouse is not currently available. Current policy making should be partly concerned with ensuring development of adequate technology for regulation in the future.

For relatively new technologies, synthetic genomics and synthetic biology—collectively known as synthetic life sciences—have received a great deal of public attention lately, mostly due to the extraordinary strides in their advancement. One recent example was the design of a living cell controlled by synthetic DNA. Developed in the 21st century, the synthetic life sciences encompass a number of distinct engineering strategies drawn from the interdisciplinary convergence of molecular genetics, chemistry, nanotechnology, and engineering.

Synthetic genomics is the chemical synthesis of DNA sequences. The synthesized sequences can already exist—for example, the chemical synthesis of genes naturally found in an organism—or it can involve the synthesis of novel DNA sequences—that is, the synthesis of a gene that incorporates a specific man-made alteration. Synthetic biology, however, is both the design and construction of new biological parts, devices, and systems—and the redesign of existing natural biological systems for useful purposes. Because the synthetic life sciences may enable the synthesis of biomolecules, whole genomes, and even simple life forms, these sciences have enormous potential, as they may logically be applied to any area of science or biomedicine that utilizes or works with genetic components, including pharmaceutical development, fuel production, detoxification of chemicals, genetic therapy, and environmental control.

The synthetic life sciences are not entirely benign, however, and they pose a dual-use dilemma: While they may be used for societal good, they may also be used by malicious individuals, like terrorists, to cause harm and incite panic. In particular, critics of the synthetic life sciences worry that this technology enables the synthesis of pathogenic agents that could be used as—or in—biological weapons, allowing would-be bioweaponeers to create biological weapons agents, such as Ebola or smallpox, which would otherwise be difficult to obtain. This kind of danger has been highlighted repeatedly: In 2002, a group of researchers at the State University of New York in Stony Brook produced the first synthetic virus upon artificially synthesizing a functional poliovirus (Cello et al., 2002); in 2005, researchers at the Mount Sinai School of Medicine in New York reconstructed the 1918 Spanish flu virus (Kaiser, 2005); and in 2008 they synthesized a bat virus (a SARS-like coronavirus), which closely resembles—and is the likely progenitor of—the human SARS epidemic (Becker et al., 2008). And then in May 2010, researchers at the J. Craig Venter Institute in Maryland used synthetic DNA to create a living cell (Gibson et al., 2010). When the polio study was published in 2002, many doubted that the same technique would allow the synthesis of smallpox, because the smallpox genome is so much bigger than that of polio, thereby being much more technologically difficult to synthesize. This technology has since progressed so rapidly, however, that experts believe the artificial synthesis of smallpox is now achievable (Selgelid and Weir, 2010a , 2010b). Smallpox is one of the most feared biological weapons agents, and modeling has shown that a smallpox attack could cause devastation on the level of a nuclear attack—or a series of nuclear attacks. For example, the number of deaths and casualties from Hiroshima has been estimated at about 135,000 (Atomic Archive, undated); two separate models showed that in a city the size of Washington, D.C., 167,240 and 168,510 deaths could result from a smallpox outbreak (Chen et al., 2004). A different study on numerous U.S. cities predicted that a biological smallpox attack that put 50 percent of the population at risk could cause between 100,000 and 1 million deaths (Bicknell, 2002). Worldwide immunity to smallpox is low because immunization programs ended after the eradication of smallpox in 1980.

Another concern about synthetic biology is that researchers with ill intent could intentionally design pathogens to be used as—or in—biological weapons, for example, pathogens that are highly transmissible, virulent, and/or resistant to treatment or vaccination; and/or pathogens with characteristics (such as environmental durability) that make them especially well-suited to weaponization. This could be achieved through the alteration of just a few genes in an already-known pathogen, such as when Australian researchers created a strain of mousepox by inserting an interleukin gene into the virus genome (Jackson et al., 2001), or from synthesizing a pathogen from first principles—like what researchers at the J. Craig Venter Institute are trying to achieve. Julian Savulescu, a bioethicist at Oxford University, says that the risks associated with the creation of such artificial life are “unparalleled” and that this kind of science “could be used in the future to make the most powerful bioweapons imaginable” (Gill, 2010).

Although the synthesis of pathogenic organisms is technically feasible, it is only possible in a relatively small number of research laboratories worldwide because of the inextricable high costs and required expertise, so in some cases, such as with Ebola, it might still be easier to obtain natural isolates of pathogens than to synthesize them. However, as costs deflate and technological capacity improves, it is likely that pathogen synthesis technology will become more accessible in the near future. The biosecurity concerns attached to the synthetic life sciences create challenges for regulators and policy experts, both nationally and internationally, because any legislative and regulatory system must find a balance between the need to provide security and the need to facilitate research. Several regulatory options have been identified, including (Bernauer et al., 2008; Garfinkel et al., 2007; Maurer et al., 2006; NSABB, 2006): the regulation of research; the regulation of the publication and dissemination of research results; the implementation of new codes of conduct; the establishment of a code of ethics for scientists; the registration and licensing of the technology; and the control of the DNA sequence trade (Samuel et al., 2009).

Many reports, discussion papers, and white papers have elaborated different approaches to regulate and govern synthetic biology (Bernauer et al., 2008; Church, 2004; Garfinkel et al., 2007; Maurer et al., 2006; SB2 Conferees, 2006), and numerous stakeholders, including scientists, research funding bodies, and governmental advisory groups, have supported policy proposals and recommendations for the control of the DNA sequence trade. Although there are several regulatory options worthy of consideration, the control of the DNA sequence trade has been a major focus, prompting at least the hope that a workable policy framework might be possible. Amid the discussions on the governance of the DNA sequence trade, the idea of a centralized clearinghouse has occasionally been raised, but it has not been discussed seriously. While the establishment of a clearinghouse is theoretically a better way to regulate the DNA sequence trade, the question from a technological standpoint is whether such an ambition is feasible and how this regulatory ambition should drive the development of future technology.

The commercial synthesis of DNA sequences

Short oligonucleotide sequences (DNA sequences up to 50–100 base pairs in length) have been used widely by numerous researchers for many years because they can be synthesized easily using a personal DNA synthesizer, or they can be purchased from a commercial supplier. Such DNA sequences are generally not used in the synthetic life sciences because synthetic DNA synthesis often requires much larger gene-length fragments of DNA (normally 1–3 kilobase pairs i.e., 1,000–3,000 base pairs in length). The synthesis of gene-length DNA fragments is much more difficult and has only become possible within the past decade or so, costs having steadily declined over this time (Tucker and Zilinskas, 2006). For example, in 2000, custom oligonucleotides were priced at $10 per DNA base pair; by early 2005, this rate had dropped to as little as $2, or in some cases $1.60, per base pair. Only about 50 companies worldwide—nearly half of which are in the United States—have the capabilities to perform this type of DNA synthesis.

In principle, either (short) oligonucleotide or (long) gene-length DNA synthesis technology could be the entry point of regulatory intervention (Garfinkel et al., 2007). The short length of oligonucleotides and their widespread use for many purposes other than the synthetic life sciences means that regulatory intervention is likely to be inefficient, expensive, and ineffective (Garfinkel et al., 2007). In contrast, because it is easier to determine the nature of longer DNA sequences (i.e., what genes they contain and which organism, pathogenic or not, they come from), and because the technology required to produce them is still relatively limited worldwide—and often used in the synthetic life sciences—it is likely that regulating gene-length DNA synthesis would better address at least some of the biosafety and biosecurity concerns raised by these sciences. Regulating gene-length DNA synthesis does, however, have limitations because a number of pathogenic organisms, such as the poliovirus, have shorter genomes and can be produced only by using short oligonucleotide synthesis (Wimmer, 2006). In addition, short DNA sequences coding for production of protein toxins could be synthesized and then placed into replicating organisms, which means that otherwise harmless organisms could be turned pathogenic by the addition of one gene, or a few genes. In spite of these limitations, gene-length DNA synthesis still offers an effective regulatory intervention point for reasons mentioned above. Most, though not all (Garfinkel et al., 2007), policy recommendations have therefore focused on the regulation of longer, gene-length DNA sequences (Bernauer et al., 2008; Maurer et al., 2006; NSABB, 2006).

DNA sequence trade responses and technical solutions

Most proposals discussing possible regulatory control of the gene-length DNA sequence trade have focused on a number of key intervention points, including the screening of customer orders for potentially dangerous DNA sequences; limiting the sale of DNA sequences to registered personnel in companies, universities, and other establishments; and retaining records of orders to be accessed in the event of a bioterrorist attack (Garfinkel et al., 2007). The licensing and regulation of bench-top oligonucleotide synthesizers (those that can fit on a bench measuring less than 50 centimeters by 50 centimeters) have also been considered (Garfinkel et al., 2007). Currently, the most popular regulatory options, however, are screening customer orders and limiting the sale of DNA sequences.

Customer registration and approval

In an effort to reduce the biosafety and biosecurity risks associated with DNA synthesis, stakeholders have proposed the development of harmonized international regulations that would only permit verified institutional customers to place orders with companies that synthesize gene-length DNA (Garfinkel et al., 2007). This type of government regulation already exists in Germany and in the United States, where firms are required to limit the synthesis and delivery of specific DNA sequences (i.e., those DNA sequences that code for virulent factors or toxins) to those researchers and institutions authorized to receive them (Graf and Wagner, 2007). In addition, the U.S. Department of Health and Human Services (HHS, 2009), the International Association Synthetic Biology (IASB, 2009), and the International Gene Synthesis Consortium (IGSC, undated) have recently produced a set of guidelines recommending the background screening of new customers. Many gene synthesis companies voluntarily perform background screening of new customers (Bernauer et al., 2008); however, customer screening is not well standardized, and government regulations in different countries are vague about customer screening requirements. As a result, there are no clear protocols for reporting any concerns raised while vetting new customers, and no background investigations have yet been flagged for further action.

Screening of gene-length DNA sequences

Many of the policy proposals to regulate the synthetic life sciences have strongly supported commercial firms, which synthesize gene-length DNA, to develop a standardized screening process to safeguard against malicious individuals attempting to synthesize dangerous pathogens (Church, 2004; Maurer et al., 2006; NSABB, 2006; SB2 Conferees, 2006). First-generation screening software, which searches through customer orders to detect any suspect sequences from pathogenic organisms, is now available (Bernauer et al., 2008). This software mostly uses a basic local alignment search tool (BLAST)—an algorithm for comparing DNA and protein sequence information. A BLAST search compares a query sequence with those already in the database to identify matches above a certain threshold (i.e., above a certain degree of sequence similarity). A large number of gene synthesis providers in Europe and the United States have a voluntary screening system in place; this means that when a customer’s order reveals suspect sequences, a more intensive investigation of the customer (or order) ensues (Bernauer et al., 2008). In general, however, current screening practices are non-uniform and disorganized. Many companies do not screen orders (Maurer et al., 2006), and, of those that do, many use their own idiosyncratic procedures.

Such lackadaisical measures have provoked calls for stronger national and international systems of governance and for a more standardized screening process (Maurer et al., 2006; NSABB, 2006, 2007; SB2 Conferees, 2006). The protocols recently published by the HHS (HHS, 2009), the IASB (IASB, 2009), and the IGSC (IGSC, undated) all include guidelines directing DNA synthesis companies to screen customers’ orders against the information held by Genbank, which is the National Institutes of Health’s genetic sequence database—an annotated collection of all publicly available DNA sequences. Further, the guidelines also require firms to screen customer orders against a list of select agents (pathogens or biological toxins that pose a threat to public health and safety). The biotechnology industry has also taken steps toward self-regulation. In 2008, the IASB stated that its members would carry a seal of approval on their websites confirming that they screen DNA sequences, thereby encouraging researchers to order sequences only from the participating companies and putting pressure on the minority of firms that cut costs by not screening (Joshi, 2008). This was reiterated in November 2009 in the IASB’s code of conduct for best practices in gene synthesis (IASB, 2009). If such screening of DNA sequences can become the “moral norm,” then those companies complying with screening regulations would in all likelihood enjoy a competitive advantage.

Although there has been some progress toward the institutionalization of comprehensive and coordinated screening of DNA sequence orders, existing screening procedures are far from optimal, and it remains unclear how systematically and effectively orders can be screened. A number of concerns are apparent.

Inefficient screening software

Because of the similarities between DNA sequences in both pathogenic and nonpathogenic organisms (known as gene conservation), it can be difficult to correctly associate stretches of DNA to particular organisms—even with the best available screening software. This leads to current screening practices generating too many (false) positives, where sequences from nonpathogenic organisms are then flagged for manual review. This is especially true for genes that are highly conserved, for example those genes necessary for survival, such as genes involved in metabolism and replication.

Foggy follow-on measures

It remains unclear what should be done following the identification of a “suspect” gene sequence in a customer order. The 2007 Alfred Sloan Foundation report made specific reference to FBI involvement in the event of suspect sequences being detected and argued for an agency to oversee procedure and regulation (Garfinkel et al., 2007). In the United States, companies are currently collaborating with the FBI in a pilot project to establish points of contact and procedures for responses to positive hits (Bernauer et al., 2008). However, there is currently no analogous central competent authority in the European Union (apart from the German Export Authorities) (Bernauer et al., 2008), and there has been little or no discussion regarding the possibility of an international center for biosecurity control.

Screening would not be a catch-all

Gene-length DNA screening would not “detect” the synthesis of those pathogenic organisms not requiring de novo DNA synthesis. For example, rather than requiring the chemical synthesis of DNA to create a pathogenic organism, some pathogenic organisms can be synthesized via a method known as polymerase chain reaction (PCR), using the natural isolate as a template.

Too many questions, not enough answers

Chief questions regarding the selection of sequences for the reference database, the maintenance of DNA databases, and the categorization of risk-associated sequences, remain unanswered (Bernauer et al., 2008; Garfinkel et al., 2007; Maurer et al., 2006; NSABB, 2006).

Security cracks remain

Even with the advent of efficient screening technology, it is conceivable that regulation by individual vendors could be circumvented in several ways. These include using different companies to synthesize parts of a dangerous pathogen’s sequence; synthesizing a novel pathogenic sequence that cannot be detected by regulatory software; or using two or more individuals to order various parts of a pathogenic sequence from different gene synthesis companies. Mandatory customer registration would partly alleviate these concerns, but it could not act as a sufficient safeguard. It is imaginable, for instance, that a scientist aligned with a terrorist group could be registered with a gene synthesis company and thus have ready access to pathogenic sequences.

Given these challenges, it is clear that existing screening processes do not yet offer a sufficient international safety net against the potential biosecurity risks associated with the synthetic life sciences. While current regulatory proposals for the synthetic life sciences, such as gene-length DNA screening, are problematic, alternative regulatory mechanisms should be considered. A central DNA clearinghouse is an alternative regulatory mechanism that addresses most of the concerns regarding the holes in the screening regulation.

A central DNA clearinghouse

Given the weakness of any screening system that is operated through individual vendors, it is worth considering the merits of establishing a facility for the screening of all gene-sequence orders, such as a central clearinghouse. A clearinghouse could act as a way to detect when different companies are used for multiple orders; a clearinghouse ensures that all registered users’ orders are processed by all individual vendors and cross-checked in a central (either national or international) database. It can also check all users against all vendors, so that a registered user would not be able to order sequences from multiple companies. A central clearinghouse would have the added benefit of acting as a mechanism for screening individual orders that would be done inconsistently or badly if left to individual companies. This centralized and uniform approach to screening would also: allow increased transparency in what would otherwise be a bias toward determining what genes are harmless; provide a hub of screening expertise; and create a more expansive network to share customer orders, thereby possibly increasing the deterrence of terrorists.

Because DNA sequences can be ordered across state and national borders, such a facility would need to work at an international level due to both the global nature of synthetic life sciences and the corresponding global public health issues. Undoubtedly, a number of obstacles will emerge in the establishment of such an international facility, but there are a number of precedents for such a proposal. For example, the Comprehensive Test Ban Treaty is an endeavor to outlaw nuclear test explosions on an international level; more than 100 international monitoring stations will present information in both real-time and upon request. Harvey Rubin, the director of the Institute for Strategic Threat Analysis and Response at the University of Pennsylvania, presents a strong second example. Rubin proposed an enforceable and comprehensive four-part international agreement between states, the private sector, and other stakeholders that will limit and control known, newly discovered, or deliberately created infectious diseases (Rubin and Arroyo, 2007).

Considering the obvious merits of an international clearinghouse, it is surprising that the international governance institutions have not more seriously considered a clearinghouse scheme. It is dismissed because it is found to be highly unlikely that a nefarious scientist would have the complex understanding of bioinformatics and expertise that would be required to place multiple orders to successfully escape detection by vendor screening (Maurer et al., 2006). There is an additional concern that a clearinghouse could breach customer confidentiality and threaten trade secrets, and would therefore be unpopular among commercial entities or research institutions (Bernauer et al., 2008). We also note that a clearinghouse would face significant technical, political, and funding obstacles, including those associated with developing software that can efficiently compare and analyze processed orders and with establishing a central database able to hold all registered users.

Global would not be simple

An international facility would be difficult to establish. However, national clearinghouses can still offer some advantages (even if they are unable to detect the splitting of orders across borders), and an international collaborative agreement could still be implemented.

A clearinghouse has limitations

The clearinghouse would not provide further regulatory benefits for detecting highly conserved pathogenic sequences (as discussed above, even small stretches of conserved DNA sequences would trigger obvious biosecurity concerns during the screening process). Detection of those less conserved pathogenic sequences, which could benefit from a clearinghouse strategy, would require the identification of a threshold of sequence homology to activate concern (i.e., the degree of similarity between the pathogenic sequence and other sequences in a database would have to be greater than an established amount to raise concern). Although the identification of such a threshold would be cumbersome, current homology searching software, like BLAST, already relies on similar methods of comparison.

Customer confidentiality could be compromised

Although data could be presented to the clearinghouse in ways that conceal customer identities and mask individual orders, this would only protect customer confidentiality if the investigated gene sequence were not identified as being potentially pathogenic (Bernauer et al., 2008). Once flagged, universities, research organizations, and commercial firms would have to be prepared to open their facilities to inspection by a regulatory body. Such inspections would breach any commitment to maintain customer confidentiality and significantly delay research dependent upon access to the sequences under investigation. This may not be unreasonable, however, as the interests of industry and the concerns they have for the protection of their customer secrets may not be more important than protecting society against biosecurity threats. Given threats to global security, industry expectations may need to change. Moreover, if such screening of DNA sequences is enforced and becomes the moral norm, then those companies complying with screening regulations would in all likelihood enjoy a competitive advantage. Furthermore, these types of regulatory restrictions may not be as problematic as alternative forms of regulatory restrictions on scientific freedom, such as proposals to regulate the publication and dissemination of research (Samuel et al., 2009).

While a clearinghouse would have advantages for detecting and preventing biosecurity threats, it may be compromised in a number of ways. Several buyers could split orders among a number of commercial firms (Tucker and Zilinskas, 2006). And even if all commercial gene-synthesis firms were registered with a clearinghouse, that still leaves noncommercial, state-run gene synthesis outside the scope of regulatory influence. This means that a central database would function to reduce poor science and low-level terrorism, but might not be effective against terrorism supported by well-resourced and highly organized states or countries. Thus, although a central database may augment any system for regulating the biosecurity threats arising from the synthetic life sciences, it may offer little protection against major terrorist organizations, state-sponsored terrorism, or state-run biological weapons programs.

The question, therefore, is whether these limitations mean that the benefits of a clearinghouse would be small; that the benefits would be far outweighed by the economic, political, and bureaucratic costs; and/or that a clearinghouse would ultimately only be effective in regulating the activity of those who submit to regulation and would be largely ineffective against major bioterrorist threats. Many of these potential shortcomings, it should be noted, would also apply to other proposed regulatory mechanisms.

Conclusion

While the synthetic life sciences potentially offer enormous environmental, biomedical, and commercial benefits, they also create significant biosecurity and bioterrorist threats. Given the risk that information in DNA databases may be used to synthesize pathogenic DNA sequences—and that these may be used to develop bioweapons—there seems to be little question that some sort of industry regulation is necessary. Of the possible regulatory options, most attention has focused on screening customer orders of potentially dangerous DNA sequences, which are placed with gene synthesis firms; limiting the sale of DNA sequences to registered personnel in companies, universities, and other establishments; the retention of records of gene-length sequences that may be accessed in the event of bioterrorist attack; and the licensing and regulation of bench-top synthesizers.

Irrespective of what regulatory mechanisms are adopted, it is important to recognize that the success of each option demands a high degree of technological capability. While technological limitations may restrict what type and degree of regulation is currently achievable, as technology evolves it may be possible to impose more stringent regulatory mechanisms and better control the biosecurity risks posed by synthetic life sciences. In light of this, we need to define what is realistically feasible from a technological standpoint both now and in the future. This means that we must acknowledge that the present regulatory technological capability for the synthetic life sciences is, at best, embryonic—the efficiency of existing screening software is low (generating too many false positives); screening processes are too easily evaded; and there is no agreed procedure for investigating and managing suspect sequence orders. Given these regulatory and technological limitations, the adoption of a number of specific regulatory responses, at least in the short term, makes sense. A regulation that limits the sale of DNA sequences to registered personnel in companies, universities, and other establishments should, and can, be imposed; efforts should be made to establish internationally agreed protocols following the identification of potentially pathogenic sequences (for the time when more efficient screening software is developed); a reference DNA sequence database should be established, determining how to keep this database current and how to categorize dubious sequences; and although implementing mandatory screening of customer orders for potentially dangerous DNA sequences currently only provides limited protection against biosecurity threats (due to inefficient software and limited policy in other areas), screening systems must still be developed and standardized for the time when software can more accurately and efficiently detect suspect sequences. Targeting the functional potential of gene synthesis (via identification of potentially pathogenic DNA sequences), rather than targeting select agents, is ultimately likely to represent the most effective means for regulating the life sciences (NSABB, 2006). While the establishment of a DNA clearinghouse is theoretically very attractive and may be feasible in the future, there is such limited regulatory technological capacity in this area that it is difficult to justify its establishment. Though the limitations of existing technology mean that a clearinghouse may not make sense now, the advantages of this approach to regulation should arguably drive the development of relevant technology.

Given that there are some reasonable, but imperfect, regulatory mechanisms available for application to the synthetic life sciences, we should not rely on the goodwill of scientists and/or the biotechnology industry to ensure that progress is made in the development of screening technology and bioinformatics. The biosecurity threats posed by the synthetic sciences are very real, and yet efforts to counter these risks are hindered by limitations in existing technology and by failure to develop biosecurity responses that cross national borders and bypass national interests. It is crucial that political and financial support is made available to advance public policy in this area and to hasten the development of better regulatory technology.