5

Assessment of Concerns Related to Production of Chemicals or Biochemicals

Metabolic engineering of microorganisms is a decades-old discipline that has been used to enable manufacturing of a variety of products including fuels, commodity and specialty chemicals, food ingredients, and pharmaceuticals. The core tenets and successes of metabolic engineering are based on the observation that biological systems are inherently chemical systems. A functioning cell, whether of microbial, human, or other origin, is essentially a collection of biochemical reactions taking place within a confined physical space as defined by a cell wall, cytoplasmic membrane, or other enveloping feature. These reactions produce structures that provide both physical form and function. Metabolic engineers have exploited biochemical pathways both to increase the production of compounds an organism naturally produces (e.g., upregulating the production of ethanol by yeast cells) and to coax an organism to produce compounds that are novel to the organism (e.g., rerouting the ergosterol biosynthesis pathway in yeast to produce a plant terpenoid [Kampranis and Makris, 2012]).

Synthetic biology concepts, approaches, and tools have allowed metabolic engineers to pursue an increasingly complex array of chemical products, typically following the overall workflow conceptualized in Figure 5-1. Westfall et al. (2012), for example, engineered yeast to produce artemisinic acid, an antimalarial drug native to the Artemisina annua plant. Galanie et al. (2015) added more than 20 genes encoding enzymes nonnative to yeast to the yeast genome in order to produce a variety of plant-based opioids. Microbes have even been engineered to produce compounds for which no naturally occurring biological pathways have been elucidated, such as 1,4-butanediol (Yim et al., 2011), a common industrial chemical also used as a recreational drug.

As the field of synthetic biology endeavors to “improve the process of genetic engineering” (Voigt, 2012), there is a concerted effort across the metabolic engineering community to demonstrate the biological production of increasingly complex molecules while simultaneously developing tools and approaches that reduce the resources required to achieve specific production metrics (e.g., titer, rate, and yield) (NRC, 2015). Hence, it is worth considering how this technology could be misused to produce chemicals or biochemicals for malicious purposes. Such products are likely to fall into one of three categories:

• Toxins.¹ Toxins are molecules produced by biological systems that are known to be harmful to humans or other animals. Toxins exhibit wide structural diversity and include small molecules as well as peptides.

___________________

¹ The word biochemical is used throughout the report to include toxins.

Given that toxins are known to cause harm, they are obvious candidates for engineered synthesis by an actor aiming to do just that.

• Antimetabolites and small-molecule drugs. Antimetabolites are compounds that interfere with the normal functioning of cellular metabolism. Although some antimetabolites can be used for therapeutic purposes, as in the use of chemotherapeutic drugs to disrupt metabolic pathways in cancer cells, compounds that target normal functions in healthy tissues can lead to dysfunction or disease. Chemically synthesized small-molecule drugs can also cause dysfunction in healthy tissues. Both antimetabolites and small-molecule drugs may be amenable to synthesis by biological systems.
• Controlled chemicals. Synthetic organic chemistry has given rise to a wide variety of chemical compounds with no known biological origin. Many have been essential to advances in human quality of life, whereas others have been used to produce explosives, chemical weapons, and other types of dangerous compounds.

While these categories of compounds are instructive in considering end uses, for the purposes of this report it is also useful to differentiate between naturally occurring products (those that are generated in a non-engineered biological host) and manmade products (those that have been chemically synthesized). This distinction affects both the experimental approach and the technical difficulty of using synthetic biology to produce a given target compound. In addition, it is useful to consider the mode of production. For example, target compounds could be produced in small quantities in a laboratory, at large scale in bioreactors (analogous to the industrial production of bio-based chemicals), or even in situ in the human host, such as the production of a toxin by a microbe in the gut microbiome. These various modes offer different challenges with regard to production, delivery, and opportunities for mitigation.

Considering the different types of potential target compounds and the different ways synthetic biology technologies might be exploited to produce them, three main types of activity were identified that are of potential concern: manufacturing chemicals or biochemicals by exploiting natural metabolic pathways, manufacturing chemicals or biochemicals by creating novel metabolic pathways, and making biochemicals via in situ synthesis of target compounds. This chapter assesses the relative level of concern warranted for each of these potential capabilities based on the four framework factors: Usability of the Technology, Usability as a Weapon, Requirements of Actors, and Potential for Mitigation.

MANUFACTURING CHEMICALS OR BIOCHEMICALS BY EXPLOITING NATURAL METABOLIC PATHWAYS

Biochemical compounds naturally produced by plant and microbial cells have been used for centuries as medicinal compounds. These products have been prepared as both plant extracts, in which the active ingredient is one of numerous chemical structures in the formulation, and as high-purity single compounds, made by cultivating the producing organism in large-scale bioreactors and then purifying the output. Such products have been used to treat diseases ranging from microbial infection to hypertension. The opioids, used as analgesics, are now accessible by microbial fermentation, as well, though optimization of the “home-brewing” process has not been rigorously explored (Endy et al., 2015; Galanie et al., 2015).

Each naturally occurring biochemical is the result of a series of chemical reactions that transform simple feedstocks such as glucose into the end products of interest. These transformations are mediated by enzymes encoded by the host organism’s DNA. Because biotechnologies allow the DNA encoding the necessary enzymes to be exploited independent of the original host, it is now possible to make such products without relying on the organism that naturally produces them.

The assessment of concerns related to manufacturing chemicals or biochemicals by exploiting natural metabolic pathways is summarized here and described in detail below.

Usability of the Technology (High Concern)

While the production of natural products in microbial hosts is not a trivial endeavor, the core technology required to complete one iteration of the Design-Build-Test cycle for metabolic pathway engineering of a target molecule is readily accessible and relatively easy to use with a basic level of molecular and microbiology expertise. Therefore, the level of concern with regard to this factor is relatively high. Assuming an actor has access to a tractable host organism (e.g., Escherichia coli, Saccharomyces cerevisiae, Pseudomonas putida), the ability to design gene cassettes and insert them into the host, the ability to culture the recombinant host and (as necessary) induce gene expression, and the ability to analyze the resulting products, attempting to engineer a metabolic pathway to produce a target toxin or other chemical or biochemical is, on the whole, a relatively straightforward proposition. Although success after one iteration of the Design-Build-Test cycle is probably unlikely, repeated cycles of effort frequently yield improvements in performance.

Of critical importance is whether the pathway, that is, the specific series of chemical reactions leading from a specified starting substrate to the final product, has been fully elucidated. If the pathway is not fully known, this can create a substantial bottleneck or barrier, because a combination of both bioinformatics and experimental techniques would be needed to identify the missing enzymes and reaction steps, necessitating a more advanced level of expertise, more time, and more scientific resources. Difficulty will also increase if a chemical or biochemical is not well tolerated by the host organism engineered to produce the pathway. The difficulty of metabolic engineering also depends on the complexity of the molecule of interest; engineering a pathway to produce structurally simpler molecules will generally be more feasible than engineering a pathway for more complex molecules. For example, the complete biosynthetic pathway for the anticancer drug Taxol remains elusive some five decades after its first discovery in the Pacific yew tree.

Once the pathway is known—and once the genes that encode the pathway enzymes have been specified—the next step is functional expression of the enzymes. This step is often challenging because enzymes transferred from one host to another may lose local structural features that are associated with activity, or they may be separated from essential accessory proteins. The tools of synthetic biology could be used to address these lost structural functions or to provide alternative pathways, but this makes for a more complicated proposition, as discussed below under Manufacturing Chemicals or Biochemicals by Creating Novel Metabolic Pathways. However, if post-translation modifications absent in the new host are essential for enzyme activity, this likely represents an insurmountable hurdle, at least in the near term.

Usability as a Weapon (High Concern)

More than offering new delivery mechanisms or modes of administration, metabolic engineering simply affords access to more material. In short, metabolic engineering in and of itself does not facilitate weaponization, but rather provides a potential means to access larger quantities of harmful material over shorter time frames.

Simply introducing a series of functional enzymes into a suitable host to produce chemicals or biochemicals does not ensure sufficient productivity to warrant concern. Three metrics are essential to assessing the effectiveness of product formation in an engineered organism: productivity (amount of product made per unit of time), titer (concentration of the product external to the engineered organism), and yield (amount of the available feedstock that is converted to product). Whereas such metrics are inconsequential in the native environment (because most biochemicals and peptides are naturally produced in small amounts), these parameters are important to the weaponization of a chemical or biochemical that requires large-scale production. For example, if a toxin is deadly to humans at a concentration of 50 mg/kg, producing that toxin to a titer of 5 mg/L would require someone to ingest at least 10 L of fermentation broth per kilogram of body weight. At a titer of 10 g/L, only 5 mL of broth per kilogram of body weight would need to be ingested. Achieving higher titers allows effective doses to be manufactured in smaller bioreactors, potentially requiring fewer resources. Productivity, titer, and yield determine the volume of cell growth and feedstock needed to make a useful (i.e., harmful) amount of compound, as well as the length of time required for production.

Generally speaking, engineering an organism to increase productivity, titer, and yield becomes progressively

more difficult. At present, engineering microbes to produce toxic small-molecule products in excess of 1 g/L would likely require the dedicated effort of trained metabolic engineers with access to a modern molecular biology laboratory, while a lower titer might be attainable with less expertise and fewer scientific resources. As a result, it can be expected that high-potency toxins would be more desirable targets for malicious actors. However, from the actor’s perspective there may also be a trade-off between the relative difficulty of producing a given chemical or biochemical and the amount needed to cause harm. Purity and productivity, as well as the complexity of the target molecule, will also factor into this trade-off. If a compound must have high purity to be effective as a weapon, the difficulty of achieving this level of purity in production or downstream processing (e.g., purifying from lysates) can potentially create a barrier. Low productivity is often related to insufficient substrate concentrations and/or low activity (i.e., the reaction rate is too slow); if enzymatic activity is not sufficiently high to achieve the turnover rates required, even when enzymes are expressed at high levels, additional iterations of the Design-Build-Test cycle may be required to achieve the desired level of productivity.

Once an actor is able to produce a sufficient quantity of a target chemical or biochemical, the predictability of results is likely to be high, assuming the actor has selected a target chemical or biochemical that is already known to cause harm. For example, mass production of botulinum toxin would not require testing of the fermentation product because the effects of its exposure are already known. Indeed, an actor could probably have greater confidence in the effectiveness or lethality of a chemical or biochemical whose pathway is well understood and is produced using synthetic biology as compared to a synthesized pathogen. The latter would definitely require testing to verify that the desired phenotypic results would be achieved.

The scope of casualty expected from a chemical or biochemical compound produced in this way would depend on the amount produced, the potency, and delivery. Chemicals, biochemicals, and toxins do not spread on their own the way pathogens do, and so, effecting a large-scale attack would require delivering a sufficient amount to targeted populations, even if the compound is highly potent. However, there are many potential delivery mechanisms for chemicals or biochemicals, which do not tend to degrade when exposed to the environment the way that pathogens do, and thus would remain potent in a broader array of delivery scenarios than would a pathogen.

In summary, engineering a microorganism to produce a chemical or biochemical by exploiting a natural pathway is considered to pose a relatively high level of concern with regard to usability as a weapon, primarily because of the predictability of the results: Producing a known toxic substance will result in a product with a known toxicity. In addition, chemical or biochemical products are more stable than pathogens. These considerations outweighed the fact that the difficulty of scaling up production to produce large amounts of a substance is a bottleneck or barrier, because there are a number of substances that are highly potent and thus toxic in very small amounts.

Requirements of Actors (Medium Concern)

Generally speaking, the core capabilities for executing a Design-Build-Test cycle in metabolic engineering require a relatively low level of metabolic engineering expertise, especially for a natural metabolic pathway that is already fully elucidated. However, the expertise required depends on the complexity of the pathway and target molecule. Achieving high-level synthesis, especially for difficult targets, does require more expertise and experience; for example, in many cases an actor would need working knowledge of how to knit pathways together into a functioning whole. To fill in the gaps in an incompletely elucidated metabolic pathway, an actor would need access to bioinformatics capabilities in order to analyze genome and transcriptome data, as well as experimental capabilities to detect and identify intermediates. For these reasons, manufacturing chemicals or biochemicals by exploiting natural metabolic pathways is considered to pose a medium level of concern with regard to this factor.

The organizational footprint required depends on the amount of product that is desired (which in turn depends on factors such as potency and titer). Small batches of a chemical or biochemical of interest could be achievable with a relatively small organizational footprint, but scaling up to produce large quantities in a bioreactor would require a larger organizational footprint and more resources.

Potential for Mitigation (Medium-High Concern)

Overall, there is a medium-high level of concern with regard to this factor, primarily driven by the fact that countermeasures are not available for a number of toxins. Lessening the concern slightly is the fact that an attack would be expected to be readily recognized. This assessment assumes that an actor would endeavor to use metabolic engineering to produce compounds with known properties. Because most known biochemicals that could potentially be misused for an attack would naturally be present in very small amounts, the emergence of disease would be a strong indication of purposeful release, thus enabling rapid identification of an attack. However, because the end product would be a chemical or biochemical that is purified away from the organism that produced it, organism-associated signatures would not be available to determine whether the attack resulted from an organism intentionally engineered to produce a dangerous chemical or biochemical, and attributing an engineered organism to a specific actor would be even more difficult.²

The capacity for consequence management depends on the chemical or biochemical used. Governments have developed medical countermeasures to respond to attacks using a subset of known toxins, but there are other toxins that have not been the focus of such efforts. The countermeasures and public health response would be expected to be the same for naturally occurring chemicals or biochemicals and for those created using synthetic biology.

MANUFACTURING CHEMICALS OR BIOCHEMICALS BY CREATING NOVEL METABOLIC PATHWAYS

While nature has provided a wide array of biochemical compounds that could be exploited for targeted synthesis, enzyme-mediated conversions also can be used to produce chemicals that organisms do not naturally create. Biocatalysis has long been used to produce pharmaceutical intermediates and active ingredients not found in nature (Bornscheuer et al., 2012). It is not always necessary to use living microbial organisms in these processes; instead, purified enzymes can be used in reaction vessels in a manner analogous to traditional organic synthesis. At its core, designing a new biosynthetic pathway involves specifying a series of enzymatic steps that can convert a set starting substrate to the desired end product. In practice, the starting substrate is often a known primary metabolite (e.g., acetyl-CoA) (Savile et al., 2010), and the proposed reaction steps are based on known enzymatic chemistry.

Engineered metabolic pathways that do not follow an existing natural blueprint have been exploited to commercialize biological production of chemical compounds (Yim et al., 2011). The true limits of biological synthesis are unknown, and advances in protein design and engineering are rapidly expanding the repertoire of enzyme-catalyzed reactions (Siegel et al., 2010; Kan et al., 2017). Researchers have also shown that materials typically present in very small amounts in biological systems, such as halogens, can be incorporated into natural products by merging plant and microbial biosynthesis machinery (Runguphan et al., 2010). These examples suggest that the range of molecules that may be accessible by biological synthesis is far larger than what has been demonstrated to date.

The assessment of concerns related to manufacturing chemicals or biochemicals by creating novel metabolic pathways is summarized here and described in detail below.

___________________

² However, note that the use of isotope ratios for chemical and biochemical attribution has been explored by the Federal Bureau of Investigation (Kreuzer-Martin and Jarman, 2007).

Usability of the Technology (Medium-Low Concern)

Producing a novel metabolic pathway is likely to be significantly more technically challenging than synthesizing a natural metabolic pathway and is likely to require multiple iterations of the Design-Build-Test cycle. Therefore, the level of concern is medium-low with regard to the usability of the technology. The technical challenge stems largely from the fact that engineering novel pathways typically requires engineering enzyme activity, either through rational (computational) design or through directed evolution, to achieve both the activity and specificity required for the pathway of interest. In addition, the enzymes in many cases may be acting on substrates not encountered in nature; in such cases, the likelihood of success is greater if it is structurally similar to the natural substrate of the enzyme being used (Hadadi et al., 2016). For some reactions, it may simply be technologically infeasible to generate high enzymatic activity, but this is likely to be unpredictable, and it may require many Design-Build-Test cycles to determine that one has reached a dead end. Generally speaking, the level of difficulty is likely to be lower if the goal is to engineer a novel pathway that is based on an existing pathway, as opposed to engineering a pathway that is wholly new.

Usability as a Weapon (High Concern)

Considerations related to weaponization, scale-up, predictability of result, delivery, and scope of casualty for novel metabolic pathways are largely similar to those for natural metabolic pathways, and so large-scale production is a barrier or bottleneck. Scaling up production may present additional challenges in the case of novel metabolic pathways if the product is toxic to the cells used to produce it, creating another barrier or bottleneck. In the context of delivery, it may be possible for chemicals created through novel metabolic pathways to be more stable for storage and transport compared to natural biochemicals.

Requirements of Actors (Medium-Low Concern)

While computational tools and established methodologies exist for creating new metabolic pathways, metabolic engineering is still largely an “art” rather than a “science.” Because intuition continues to play a significant role in the successful execution of experimental designs, creating functional novel metabolic pathways is likely to require a higher level of expertise and experience than exploiting natural pathways would. In particular, if a novel pathway requires enzymes to act on novel substrates, expertise in protein engineering (which is beyond the typical skill set of an experienced metabolic engineer) would also be required. Both the knowledge about how to design novel pathways and knowledge of how to engineer enzyme activity are bottlenecks or barriers in this space. Therefore, the level of concern with regard to this factor is medium-low.

Potential for Mitigation (Medium-High Concern)

Considerations related to mitigation capabilities for chemicals or biochemicals manufactured by creating novel metabolic pathways are largely similar to those for chemicals or biochemicals created through natural metabolic pathways.

MAKING BIOCHEMICALS VIA IN SITU SYNTHESIS

The human microbiome, particularly the gut microbiome, has been a target for metabolic engineering. Gut microbes influence the metabolism of their host and are capable of producing a wide variety of biochemicals. While the extent of the influence of the microbiome on host metabolism remains an active research area, there has already been significant progress toward engineering gut microbes for therapeutic purposes. Engineered microbes are currently being prepared for clinical trials for the treatment of metabolic disorders (Synlogic, 2017), although engineering high flux through a metabolic pathway remains undemonstrated.

As this research gains steam, it is worth considering whether the human microbiota could be exploited to

make biochemicals (within the cells of commensal organisms) and deliver them to human hosts to cause harm. In addition to the gut microbiome, the skin microbiome could be another potential avenue for in situ synthesis of such compounds. Related concepts include the manipulation of the human microbiome to cause dysbioses or as an avenue for horizontal gene transfer (see Chapter 6, Modifying the Human Microbiome). Environmental dispersion of a microorganism capable of producing toxins, antimetabolites, or controlled chemicals may also be considered a potential in situ delivery mechanism, one whose outcome would be difficult to predict. The basic principles of pathway engineering in a microbe are the same whether the intention is to culture the organisms in large vessels followed by purification of the molecules of interest or to introduce the organisms into the environment or a human host for in situ production and release of a biochemical. However, the scope of the engineering effort can vary substantially since manufacturing in vessels is likely to require that much higher production titers be achieved. For example, nanograms of a sufficiently toxic material delivered in situ could be sufficient to produce a harmful effect compared to tens of grams per liter needed for cultivation in and purification from fermentation vessels. This difference is important to consider in assessing concerns.

The assessment of concerns related to making biochemicals via in situ synthesis is summarized here and described in detail below.

Usability of the Technology (Medium-High Concern)

From an engineering perspective, creating a microbe capable of in situ biological synthesis of a biochemical presents many of the same opportunities and challenges as engineering metabolic pathways for the production of chemicals or biochemicals in a bioreactor, though there are some additional challenges, as well. While productivity, titer, and yield can typically be measured in the process of manufacturing a chemical or biochemical product in a bioreactor, conditions in the microbiome, for example, are quite different from those present in the laboratory. This makes it difficult to predict and control whether productivity, titer, and yield measurements in the laboratory will translate to similar numbers once the microbe is delivered to the microbiome (or environment). Many Design-Build-Test cycles, including a substantial amount of testing in both cell cultures and in animal models, are currently needed to obtain engineered gut microbes with functional gene circuits (Lu et al., 2009; Kotula et al., 2014, Mimee et al., 2015; Matheson, 2016). One potential way to expedite development and reduce the need for multiple rounds of resource-intensive in vitro and in vivo testing would be to expose human subjects to large libraries of prototype microbes, then sequence the microbiome content to identify the successful prototype microbes if toxicity is observed. However, this library approach has important limitations. For example, a prototype microbe capable of producing high titers of a toxin if introduced to the gut as a monoculture could be effectively diluted by the presence of large numbers of ineffective prototype microbes, making it difficult to detect and identify the successful prototype microbe. In addition, it is possible that a microbe that produces high titers of a toxin would grow more slowly than prototype microbes that produce little or no toxin, making it difficult to separate signal from noise. Finally, the current state of the art in gut microbiome sequencing and assembly does not guarantee that a successful prototype strain could be correctly constructed and differentiated from all other introduced library strains. Nonetheless, the fact that many organisms harbor their own toxins as part of their infective life cycle means that it should not be impossible to align pathogenicity and evolutionary fitness, and indeed one of the easiest means of establishing a toxin in situ may be via an already known pathogen, as discussed under Usability as a Weapon, below, and in Chapter 4, Box 4-2.

Overall, the knowledge needed to manipulate organisms in the gut and skin microbiome remains limited, as further discussed in Chapter 6, Modifying the Human Microbiome, and it is possible that unforeseen challenges in

producing biochemicals in situ will emerge in the coming years. However, the field has been advancing quickly. Already, researchers have demonstrated the ability to manipulate some human gut microbes, and the use of the microbiome for delivery of pharmaceuticals is an active area of research. Thus, the high rate of development and investment in this field leads to a medium-high level of concern with regard to this factor. It will be important to monitor for research breakthroughs that exacerbate opportunities for misuse in this area, as well as breakthroughs in understanding.

Usability as a Weapon (Medium Concern)

Usability as a weapon is considered of medium concern, largely due to current limitations in the ability to make introduced microbes persist in the microbiome. However, microbiome engineering is an active area of research, and significant advances, such as a demonstrated ability to cause persistent changes in the gut microflora, would cause the level of concern to rise.

The gut microbiome is known to host thousands of gene clusters, and products of these clusters have been shown to be present in the gut at high micromolar concentrations (Donia and Fischbach, 2015). Therefore, it should be possible to engineer gut microbes to produce harmful small molecules at similar levels. However, despite the presence of these natural pathways in the microbiome, the principles behind engineering similar pathways to produce other products in situ have not been determined. Engineering the production of a toxin with sufficient titer, produced over a long enough time to be harmful to the host, is not necessarily straightforward. Furthermore, after being delivered into the host microbiome, the engineered microbe would need to colonize and persist to have a long-term effect. Experiments with attenuated vaccine strains suggest that it is necessary to eliminate some existing microbes in order to allow an introduced microbe to persist in the gut, adding to the complexity of purposefully infiltrating a host microbiome. A perhaps more likely scenario is that existing gut or skin microbes could be manipulated to increase their natural production of a harmful compound or to resist antibiotics or other countermeasures, thus allowing delivery of an agent without the barrier of infiltrating the native microbiome with a new microbial species. In addition, it is possible that a pathway lodged on a broad-host-range vector might be horizontally transferred to native species following transient introduction on a microbe that was otherwise unlikely to colonize; the horizontal transfer of in situ engineered pathways is further considered in Chapter 6, Modifying the Human Microbiome.

Although the chemical product would be manufactured by cells, bioreactors or flasks would likely be required to produce a sufficient number of cells to enable delivery to the target human population. Microbes engineered to secrete highly potent biochemicals, which could cause greater damage in smaller quantities, would warrant greater concern than those engineered to produce lower-potency chemicals. But effectively delivering engineered microbes to the human target would still present significant barriers. Cold War–era studies on the weaponization of bacteria remain relevant to this concept. Contamination of food could be an efficient method of dispersal, but could be thwarted by standard food safety measures such as cold storage, cooking, and mechanisms to limit the spread of contaminated food. The scope of casualty from in situ biosynthesis would be expected to be relatively low, because the agent would need to be delivered to each individual and then persist in the gut or skin long enough to cause harm. That said, the ability to slightly or gradually modify human physiology and behavior via even low-level production of compounds could be extremely debilitating to a modern nation-state.

Requirements of Actors (Medium Concern)

Engineering microbes to actively secrete products in the microbiome would generally require a higher level of expertise than engineering a natural metabolic pathway but less sophistication than designing a novel metabolic pathway, leading to a medium level of concern with regard to this factor. Because multiple iterations of the Design-Build-Test cycle would be needed, actors would likely require access to significant laboratory resources over a long period of time. On the other hand, in situ synthesis presents fewer barriers with regard to scale-up and downstream processing than the production of chemicals or biochemicals in a bioreactor, and once a sufficient

engineered microbe is developed, producing and delivering a small quantity would not require a great deal of technical expertise.

Potential for Mitigation (High Concern)

The challenges of attribution and the difficulty of identifying and stopping an attack based on in situ synthesis of biochemicals lead to a relatively high level of concern with regard to this factor. Policies and procedures related to the containment of natural foodborne pathogen outbreaks should transfer well to the containment of engineered toxin-producing gut microbes. Indeed, the presence of strong public health infrastructure for food safety and response to contaminated-food outbreaks may deter skilled actors from pursuing an attack with engineered gut bacteria in favor of other attack vectors. In addition, while engineering microbes to resist traditional countermeasures (such as the use of broad-spectrum antibiotics) could increase the casualty rate, containment and isolation of contaminated facilities would be expected to limit the spread of such agents. However, the delivery of engineered microbes to the gut via food is not the only potential attack vector or means of delivery. The development of an engineered microbe that could infiltrate the skin microbiome, or the development of a high-efficiency method of delivering gut microbes, could be less vulnerable to existing mitigation measures and thus significantly increase the level of concern warranted. However, these delivery modes are currently theoretical.

Regardless of the effectiveness of public health infrastructure for containing an attack, it could be extremely difficult to recognize an attack—that is, to differentiate between a natural disease outbreak and an intentional introduction of engineered microbes into the microbiomes of affected people. This difficulty is the primary driver of the relatively high level of concern related to the potential for mitigation. Some types of attack would be easier to recognize than others; for example, the presence of an unlikely gut toxin or extremely high resistance to available countermeasures may be more easily recognized as signs of an attack, while tracing an effect that is not a classical gut problem (e.g., opioids made in the gut) to engineered gut microbiota would be a substantial task.

In contrast to the other applications of metabolic engineering discussed in this chapter, the genetic material of the engineered microbe would in the case of in situ synthesis remain present in the weaponized product. Sequencing clinical samples of impacted individuals could allow investigators to identify the genetic sequences or organisms used in an attack. However, such an effort would face significant technical challenges. First, if the engineered microbe is present in low abundance, most of the sequence data in a sample would come from non-engineered commensal microbes. Compounding this, only a small amount of the genome of an engineered microbe would be expected to contain new DNA. For example, an engineered Escherichia coli genome could contain fewer than 10 heterologous genes, which would need to be detected within the rest of genome, which contains more than 4,000 genes. The high complexity and variability of the gut microbiome composition increases the potential that uncharacterized genes present in the sequencing data could be confused with transgenes.

Even if the sequence of an engineered pathway could be identified in a clinical sample, it may still be difficult to trace the attack to the actors responsible. One potential approach would be to attempt to identify the vendor that produced the synthesized DNA. However, with DNA synthesis technology becoming increasingly accessible, it may become difficult to query all companies capable of producing synthetic DNA. Furthermore, assembly of synthetic DNA from nucleotides could obviate the need for DNA synthesis from a commercial provider. While investigative work in tracing the engineered microbes to their source is likely to be more informative than focusing on the transgenic DNA sequences, the sequences would be extremely important to connecting suspected actors to the weapon material, if matching materials in the actor’s laboratory were available.

SUMMARY

This chapter considers various ways in which synthetic biology technologies could potentially be applied to produce chemicals and biochemicals such as toxins, antimetabolites, small-molecule drugs, or controlled chemicals for use in an attack. Broadly, the use of microbes to synthesize agents in situ presents the greatest level of concern, the synthesis of agents using naturally occurring metabolic pathways warrants a medium to high relative level of concern, and the engineering of novel metabolic pathways poses a medium level of concern.

It will be important to continue to monitor developments in the manipulation of the human microbiome because efforts in the pharmaceutical arena are likely to propel advances and reduce bottlenecks and barriers as the field continues to progress (see Table 5-1). Although the level of certainty around the in situ manufacture of biochemicals via the gut or skin microbiome is lower than the level of certainty involved in the other metabolic engineering processes described in this chapter, manipulation of the microbiome is an active and quickly advancing area of research. Overall, this potential capability warrants a higher level of concern, because an attack effected through manipulation of the human microbiome could be difficult to recognize and trace. However, understanding of microbiome dynamics is still relatively limited, and it would likely take a relatively high level of expertise and many iterations of the Design-Build-Test cycle to develop a microbe capable of colonizing the human host microbiome, manufacturing the biochemical in sufficient quantities, and persisting long enough to cause harm.

The primary drivers of the medium to high relative level of concern for the potential exploitation of naturally occurring metabolic pathways are the relatively high level of knowledge available, the relatively low level of technical expertise required, the availability of multiple delivery mechanisms, and the difficulty of tracing the source of an attack. Exploitation of naturally occurring pathways could be an option for attackers because it is easier, in general, to use microbes to manufacture complex chemicals or biochemicals than to use chemical synthesis techniques. However, scalability remains a bottleneck, and manufacturing large enough quantities of the chemical or biochemical to effect a large-scale attack would require a large organizational footprint. Given this, a more likely application of this approach may be to manufacture drugs, such as opioids. The difficulty of this approach also depends heavily on the complexity of the chemical or biochemical of interest and of the metabolic pathway for producing it. For some target chemicals or biochemicals, an actor may conclude that cultivating the native host organism may be more feasible than using metabolic engineering to produce a biochemical in a bioreactor (e.g., cultivating Clostridium botulinum instead of heterologous production of botulinum toxin).

The development of novel metabolic pathways to produce chemicals is a technically challenging proposition that would require expertise in both metabolic engineering and protein engineering in order to develop the necessary enzymatic activities, and further efforts to make the novel pathway yield a sufficient amount of product for

TABLE 5-1 Bottlenecks and Barriers That Currently Constrain the Capabilities Considered and Developments That Could Reduce These Constraints^a

^aShading indicates developments that are likely to be propelled by commercial drivers. Some approaches, such as combinatorial approaches and directed evolution, may allow bottlenecks and barriers to be widened or overcome with less explicit knowledge or tools.

an attack. Multiple iterations of the Design-Build-Test cycle would be required. The difficulty would be reduced if the novel metabolic pathway were to use steps, enzymes, or substrates from a naturally occurring pathway, and indeed, recent advances in protein design and engineering have rapidly expanded capabilities for engineering novel metabolic pathways. The most feasible metabolic routes will be those that have been already demonstrated elsewhere (e.g., in the academic literature), because recapitulating an engineered pathway is substantially more tractable than developing a pathway from scratch. However, even where biological synthesis is feasible for producing controlled chemicals or other products, traditional chemical synthesis may prove to be a more reliable, cost-effective, and surreptitious means to do so when the involved pathways are novel. An actor skilled in the art of metabolic engineering who is capable of engineering high-titer strains and has access to the right scientific resources is expected also to be sufficiently skilled to access, and potentially opt for, these other options.

Relevant developments to monitor for each of these capabilities are summarized in Table 5-1.