The challenges of developing synthetic pathogens

The announcement in January 2008 that a team of researchers at the J. Craig Venter Institute had constructed the first complete bacterial genome, and that they intend to create a viable bacterium from it, has spotlighted the power of gene synthesis technology. As the ability to design, construct, and biologically activate long strands of genomic material improves, the benefits in terms of research in medicine, energy, materials, and other applications will grow. However, so could concerns that the technology might lead to accidental or deliberate harm.

Many of these apprehensions anticipate synthetic genomics lowering the barriers a bioterrorist might otherwise confront when trying to obtain and disseminate a highly pathogenic agent. Indeed, technological trends already give cause for concern. Biotechnology is growing ever more powerful, ever more available geographically, and ever more familiar as it penetrates the economy in a host of applications, including many having little to do with medicine or drug development. There are also few technical barriers to bioterrorism. Essentially all of the materials, technology, and skills required to produce biological weapons have legitimate applications for the commercial economy or scientific research, and many capabilities that once required the resources of a state could now be accessible to non-state groups. True, there are no legitimate applications that integrate all of the requirements that a bioweaponeer would need, but essentially any of the individual prerequisites can be mastered by those who have the technical ability, the dedication, and the resources to pursue them.

The most serious potential scenario would be if individuals trained sufficiently broadly in biological science or biotechnology–and who therefore already have the expertise to develop biological weapons–become sympathetic with or recruited by terrorist groups. Terrorists becoming biologists is less of a concern than biologists becoming terrorists. For these individuals, gene synthesis offers three paths to obtaining a dangerous pathogen. The most straightforward would be to order an entire genome from a commercial vendor, drawing on an emerging set of firms that will produce a vial full of DNA or an organism containing a specific piece of DNA of any desired sequence. Alternatively, a genome could be constructed by stitching together a number of smaller oligonucleotides–single-stranded strings of DNA typically less than 100 base pairs in length. These oligonucleotides can be obtained from commercial providers, who can manufacture them in any desired sequence. A third option would be to synthesize oligonucleotides using commercially available DNA synthesizers, rather than purchase them, and stitch them together to form a genome.

The difficulty of producing a genome via any of these pathways depends on the length and nature of the desired sequence, and the difficulty of using that genome to construct a viable organism depends greatly on the type of organism. For some viruses, such as poliovirus, the virus’s genetic material is directly infectious, and introducing it into a susceptible cell will result in the production of complete virus particles. Other categories of viruses, including the flu or smallpox, have genomes that are not in themselves infectious. When these viruses infect organisms, they carry with them key enzymes not found in the host, and synthesizing these viruses by constructing their genetic material also requires reproducing the function of these extra enzymes.

In addition to the difficulties in synthesizing a pathogen genome and in creating viable pathogens from it, a third factor to be considered by anyone attempting to synthesize a pathogen is the fidelity of the database from which that pathogen’s genome is taken. Although the genomes of many pathogens have been sequenced and placed in publicly accessible databases, even a fully faithful synthesis may not lead to a virulent pathogen. First of all, published genome sequences contain errors, some of which may be completely disabling and others of which would attenuate a resulting organism’s virulence. Second, a sequenced genome may not have come from a virulent “wild type” virus, but rather from a culture that has spent many generations in the laboratory. Cultures raised in this way do not have to overcome a host organism’s immune system, and they therefore do not face selective pressure that would keep them virulent in the presence of attenuating mutations. For both of these reasons, anyone synthesizing a pathogen from a publicly available database could not be confident that the result would be fully virulent, or even functional.

Bacteria are far more complicated than viruses, and synthesizing their genomes, which are generally far larger than viral genomes, is a more difficult and time-consuming process. The same goes for introducing a bacterial genome into a cell where it can become active. So, although the Venter Institute’s recent work with the genome of the bacteria Mycoplasma genitalium raises a number of questions for societal discussion regarding the future applications, consequences, and governance options appropriate to this new technology, the security implications of this particular experiment are not immediately apparent. In general, synthesizing bacteria is not likely to be the preferred route for a scientist seeking to obtain them for harm.

Given how complicated and expensive it is to synthesize pathogens, compared to other means of acquisition, only a few viruses might today be more easily obtained through synthesis than through other means of acquisition, according to a 2007 report, “Synthetic Genomics: Options for Governance,” funded by the Alfred P. Sloan Foundation. In five to ten years, however, as synthesis costs continue to drop and performance continues to rise, the situation may reverse; synthesis may be easier than obtaining a pathogen through other routes. Indeed, synthesis appears to be following a trajectory familiar to other useful techniques: Originally accessible only to a handful of top research groups working at state of the art facilities, synthesis techniques are becoming more widely available as they are refined, simplified, and improved by skilled technicians and craftsmen. Indeed, they are increasingly becoming “commoditized,” as kits, processes, reagents, and services become available for individuals with basic lab training. Even the database uncertainties will become less significant over time as more and higher fidelity pathogen sequences are uploaded into databases, and as the ability to sequence cultures taken directly from patients increases.

The viruses for which synthesis might make acquisition appreciably easier include extinct viruses, viruses of unknown origin in nature, and those that are kept in laboratories under some degree of security or control. Two such examples would be Ebola and Marburg hemorrhagic fever viruses, whose natural reservoirs are unknown. Constructing these viruses would be moderately difficult to difficult at present. The genomes of these viruses are relatively small, but they are not directly infectious, and producing viable organisms from them would be challenging.

The variola virus, which causes smallpox, is another example of an organism that could be more easily acquired by synthesis than by other means because it is extinct in nature and is known to exist only in two high-security laboratories. But synthesizing variola would be difficult because its genome is one of the largest of any virus, and because that genome is not directly infectious. Like variola, 1918 flu is extinct in nature and exists only in controlled laboratories. Therefore, anyone seeking to obtain this virus, which killed tens of millions of people early in the last century, would likely synthesize it. Even though its genome has been reconstructed and published and is small, constructing this virus would be moderately difficult because the genome is not directly infectious.

One other virus that could be acquired through synthesis is Foot-and-Mouth-Disease (FMD). This virus is an animal pathogen that does not infect humans but could be economically devastating to agriculture. It has been eradicated from the United States but is endemic in many other countries around the world. Although an infectious amount could be smuggled into the United States with very little chance of detection during an X-ray or baggage examination, a smuggler might be concerned that his or her identity might be known to authorities or that he or she might otherwise raise suspicions at the border. Therefore, the smuggler might seek to avoid border crossings by choosing to synthesize FMD virus within the United States. Since this virus has a small genome that is directly infectious, its synthesis and construction would be relatively easy.

Some observers have expressed concern that beyond synthesizing current threat viruses, synthetic genomics might be used to create novel “designer viruses” or “superpathogens” that are more virulent, more contagious, or in other ways even more suited for malicious use than organisms that already exist or that could be constructed using conventional genetic engineering techniques. Without ruling out the development of such future pathogens, not enough is known today about viral pathogenesis immune responses, or how changes in genomic sequence affect viral function for such a genome to be designed from scratch and synthesized for that purpose. Given the accelerating rate of advance in the relevant fields of biology, however, it is difficult to predict with confidence what will remain outside our realm of capability five to ten years out.

The 2007 Sloan study concluded that synthetic genomics, like other potentially dual-use technologies, can pose safety and security risks, but these risks do not merit a halt to further development of the field. Given how embedded synthesis techniques have become in modern biology, and how accessible the underlying materials and technologies are, it is not clear how such a halt could be implemented, particularly to constrain those who are not concerned with following rules or standards. Nevertheless, the study did identify policies that might help manage these risks. No set of policies or controls can eliminate the risk of illegitimate or malicious use, but measures can be taken to increase barriers to development and the probability of detection. Among the study’s suggestions: having commercial providers of oligonucleotides or genes screen their orders, to see if dangerous pathogens are being constructed, or seek assurances that their customers are performing legitimate activities that are known to their host institutions. Controls might also be placed on equipment and reagents that individuals might use to synthesize genomes and construct organisms on their own. Synthetic genomics will also benefit from whatever measures are put in place to deal with security concerns posed by the advance of bioscience and biotechnology more broadly, such as are being considered by the National Science Advisory Board for Biosecurity.

Any attempt to build a governance system for a technology as nascent as synthetic genomics is fraught with difficulty. At its outset, a technology is so immature, so specialized, and so limited in distribution that it’s hard to assess its benefits and risks, much less estimate the relative costs and advantages of potential governance approaches. On the other hand, once a technology becomes ubiquitous, the opportunities to institute effective governance become much more limited; it is always more difficult to “retrofit” a set of policies than to build them in from the beginning. The hope, in addressing the safety and security risks of synthetic genomics, is that governance measures exist that are both useful (effective at mitigating risk) and feasible (able to be implemented at acceptable cost), and that there is sufficient room between “too early to tell” and “too late to change” to develop and institute them.