This appendix describes a core set of current synthetic biology concepts, approaches, and tools that enable each step of the Design-Build-Test (DBT) cycle, focusing particularly on areas in which advances in biotechnology may raise the potential for malicious acts that were less feasible before the age of synthetic biology. Although the examples presented are intentionally quite broad and somewhat arbitrary—and do not represent an exhaustive list of all technologies or all possible applications of synthetic biology—they provide useful context for understanding how specific tools or approaches might enable the potential capabilities analyzed in Chapters 4–6. In addition, while the main known concepts, approaches, and tools at the time of writing are captured, this list will need to be updated and modified to stay relevant as the science advances. The relative maturity of the different technologies is described in Table A-1 to give a sense of which technologies are in widespread use, which are just in development, and which are somewhere in between.
Concepts, approaches, and tools most closely aligned with the Design phase of the DBT cycle are those that enable researchers to envision and plan the engineering of biological components. This report takes a broad view of Design to include both the technologies that enable design and design objectives; as such, this grouping includes both synthetic biology technologies and examples of the types of applications that they might enable.
Automated Biological Design
Engineering biological components can be a challenging proposition; organisms are complex, and scientific understanding of biology remains incomplete. Designers must consider the effects of a large array of potential variables, including DNA bases, codons, amino acids, genes and gene segments, regulatory elements, environmental context, empirical and theoretical design rules, and many other elements. Automated biological design, known in the field as bio-design automation, lowers the barrier to designing genetic constructs by automating some decisions and processes that would otherwise require a high level of expertise or a long time to carry out. This automation is enabled by tools such as computer algorithms, software environments, and machine learning.
Some automated design tools help researchers specify the desired function of the biological construct or how the parts in the construct will be organized. Other tools help to transform these specifications into collections of realizable DNA constructs; many software tools, for example, help manage and visualize synthetic DNA sequences
TABLE A-1 Summary of Relative Maturity of Selected Synthetic Biology Concepts, Approaches, and Toolsa
|In Development||In Use by Developers of the Technology||In Use by the Synthetic Biology Community||In Use by the Molecular Biology Community||In Use by Amateur Biologists|
|Multiplexed genome editing (MAGE/CRISPR)|
|DNA synthesis and assembly|
|Multi-input logic circuits|
|Combinatorial DNA assembly|
|Automated DNA assembly|
|De novo protein structure prediction|
|Broad-spectrum horizontal transfer vectors|
|Xenobiology (incorporation of nonnatural nucleotides or amino acids)|
aFor each column, darker shading indicates routine use for that community, lighter shading indicates emerging use, and white background indicates little or no use. Adoption flows from left to right in most cases.
as they are being designed. Computer software can greatly enhance the designer’s ability to predict a design’s function and performance, making it more feasible to engineer increasingly complex biological functions and potentially reducing the time and resources required to generate and test designs. Some predictive components of these tools are fairly straightforward, such as the virtual translation of a gene’s DNA sequence into the corresponding chain of amino acids. Other functions are more complex, such as the predicted cross-interaction of transcription factors in a genetic circuit.1 There has been significant progress, for example, in the automated compilation of in vitro and in vivo transcription-dependent or translation-dependent genetic circuits starting from high-level functional or performance specifications (Brophy and Voigt, 2014). Software can also allow designers to create
1 “Genetic circuits” in synthetic biology are analogous to electronic circuits. Just as electronic circuits are comprised of individual electronic components (e.g., resistors, transistors) assembled together to perform a desired function (e.g., sensing, actuation), genetic circuits are constructed from the assembly of biological components. These components are encoded in the DNA and may include, for example, DNA binding sites, promoters, or transcription factors. As an example, a genetic circuit could be constructed to detect (sense) a particular metabolite and to initiate expression of a protein once the metabolite concentration crosses a certain threshold (actuate).
large libraries of combinatorial variants quickly and use machine learning to converge on optimal solutions. This allows for higher levels of design abstraction and the use of standards to exchange information globally between software frameworks.
In addition to aiding biological design, automation tools are used in other phases of the DBT cycle, as well. For example, researchers can use automated assembly tools to plan how to physically create their designed constructs most efficiently or to send designs created in silico directly to remote manufacturing facilities. These designs can be distributed across locations to massively parallelize the construction process. Once a construct is assembled, automated testing tools can be used to verify that it functions as designed. Taken together, a greater predictive capacity, automated assembly, and rapid testing can be expected to facilitate the engineering of increasingly difficult biological functions. Some example applications of automated biological design that are useful to consider in the context of biodefense include design of genes and proteins and bioprospecting and pathway design.
Design of Genes and Proteins
Automated design programs can create thousands of genetic design variants by combining libraries of genetic “parts” in various ways, an approach known as combinatorial library design. The developers of such programs typically build certain design rules into the algorithm to increase the chances that the designs created will be functional from a biological standpoint. Once the program is in use, the variants it creates can be used to improve design rules via machine learning or statistical analysis. Through this learning process the programs are able to refine subsequent designs; the process also could ultimately remove human designers from the design process, allowing DNA design, assembly, and verification equipment to explore large genetic design spaces automatically. The results of combinatorial library design programs can be stored and shared electronically for researchers to validate each other’s designs, merge multiple designs, or otherwise manipulate the outputs.
Computer-aided design is also being applied to engineer protein structures, which are crucial to many biological processes. Examples of key protein functions being pursued include folding into a desired structure, binding to another protein or to a small molecule, and catalyzing a chemical reaction. Researchers have already made significant progress toward the predictive design of protein structures and engineering existing peptides and proteins for new functionalities. Automated design tools could facilitate the pursuit of more complex protein engineering, such as designing a new protein or enzyme capable of functioning with a level of specificity similar to that of natural proteins.
Bioprospecting and Pathway Design
Software can also enable designers to search for existing enzymes or biochemical pathways that could be incorporated into genetic designs to produce chemicals of interest. This type of searching is known as in silico bioprospecting. Using this approach, researchers systematically screen a large body of DNA sequence data to identify genes or protein domains that encode enzymes capable of performing a desired chemical reaction. After identifying hundreds of candidate genes, researchers produce selected genes synthetically and test their functions in vitro or in vivo. Additional software tools can be used to engineer more complex biochemical pathways by helping the user visualize those pathways, including their connections to the larger metabolic network of the cell, and estimate how different factors affect the levels of the various compounds produced. In this way, simulation and modeling tools can help to identify where adjustments might be most impactful, such as by increasing the expression of one gene product or by deactivating or downregulating a gene involved in a competing pathway.
Metabolic engineering involves the manipulation of biochemical pathways within a cell, frequently with the objective of producing a desired chemical. The desired chemical may be new or one that the cell already makes, and it may be simple (e.g., ethanol) or more complex (e.g., polypeptide or polyketide antibiotics). Based on a detailed understanding of the network of biochemical reactions within the cell, researchers can identify the
genes involved in crucial steps in the network of biosynthetic pathways and then adjust them to improve yields. This process is rarely as simple as increasing the expression of all enzymes in the pathway, which can lead to overconsumption of cellular resources and harm the cell’s ability to grow and produce effectively. In addition, some intermediate chemical products of the pathway may be toxic to the cell, in which case it can be important to carefully regulate how rapidly such compounds are produced and consumed. Other pathways that compete with production of the final product may also need to be adjusted. Because biochemical pathways are often complex, engineering them frequently involves the use of sophisticated computer software. Metabolic engineering could potentially be used to produce toxins, narcotics, or other products relevant to biodefense. For example, yeast has already been engineered to produce opioids in minute quantities (Thodey et al., 2014). It is also conceivable that these techniques could be used to engineer organisms in the human microbiota to produce compounds that alter human health, perception, or behavior.
The phenotype of an organism can be affected by multiple genetic components. While there are some phenotypes for which it is possible to identify specific genes or circuits that would need to be added or altered in order to achieve a particular outcome, such as the capability for horizontal transfer (the movement of genes from one organism to another, as opposed to the vertical transfer of genes from parent to offspring) and transmissibility (the ability to pass from one organism to another), in many other cases it is difficult to determine the multiple genetic components that may impact phenotype. In the past, an organism’s phenotypes were manipulated largely by the accumulation of sequential mutations, which in many cases led to local rather than global optimizations of function. More recently, the explosion of sequence information and accompanying systems biology characterizations of multiple organisms have provided a cornucopia of possibilities for engineering phenotypes that involve much more complex networks of genetic components. In parallel, the rise of DNA construction and genome editing technologies could facilitate the construction of multiple variants that involve alterations to multiple genes in an organism. By applying high-throughput screening or selection to these variant libraries, it may be possible to isolate pathogens with dramatically modified phenotypes relevant to their potential weaponization, such as environmental stability, resistance to desiccation, and ability to be mass produced and dispersed.
Horizontal Transfer and Transmissibility
The spread and impacts of a given pathogen are closely tied to its ability to replicate and be transmitted to naïve hosts. Synthetic biology technologies could potentially be applied to make a pathogen’s genes more easily transmitted, such as by enabling or enhancing the horizontal transfer of genes. Genes, circuits, or episomes (pieces of genetic information that can replicate independently of the host) can already be engineered to be horizontally transferred by exploiting commonalities in replication and transformation machinery; for example, the introduction of invasin genes has been used to alter the host ranges of bacteria (Palumbo and Wang, 2006; Wollert et al., 2007). New research aims to combine multiple such techniques to create near-universal horizontal transfer vectors with expanded functionality; if successful, this work could broaden the potential areas of concern (Fischbach and Voigt, 2010; Yaung et al., 2014). Combinatorial methods that are available via library synthesis and either high-throughput screening or directed evolution may also potentially be used to alter or expand horizontal transfer and transmissibility. Past research has demonstrated that even low-throughput directed evolution of functions can be used to enhance airborne transmission of H5N1 influenza virus between mammals (Herfst et al., 2012; Imai et al., 2012).
Xenobiology refers to the study or use of biological components not found naturally on Earth (Schmidt, 2010). A simple example is the engineered incorporation of a new amino acid (one not typically found in living cells) into a cell’s proteins. Recent research has demonstrated that it is possible to engineer cells to employ a genetic code different from that shared by most life on Earth, or to incorporate nonnatural DNA bases (beyond adenine,
thymine, cytosine, and guanine) into a cell’s DNA (Chen et al., 2016; Feldman et al., 2017). Such approaches could potentially be used to block infection by viruses or prevent undesired horizontal transfer of gene function. Cells with alternative DNA bases, codons, amino acids, or genetic codes may also be able to evade detection based on standard methods such as polymerase chain reaction (PCR), DNA sequencing, or antibody-based assays.
While past considerations of biodefense concerns have largely been focused on pathogens, synthetic biology raises new possibilities for modifying a person’s physiology or environment in ways that may lead to dysfunction, disease, or increased susceptibility to disease. For example, altering the makeup or functions of the gut microbiome could either enhance a person’s health or cause dysfunction. Modulation of the immune system—the body’s defense against pathogens—is another hypothetical possibility worthy of consideration, as is epigenetic modification (changes in how cells express genes but not changes to the DNA sequence itself). In short, there is now a large amount of information available about the human form that could potentially inform phenotype modulation in different ways.
Technologies and applications most closely aligned with the Build phase of the DBT cycle are those that are used to physically create actual biological components. Synthetic biology is often pursued in an iterative fashion, blurring the lines among the Design, Build, and Test phases, and some technologies can play a role in multiple phases. Considered here are technological capabilities and advances related to specified changes and to the construction of libraries for high-throughput screening or directed evolution.
Factors that may impact the level of concern related to Build capabilities include cost, time, and ease of access for DNA construction; the complexity of libraries that can be generated for directed evolution; and the difficulties inherent in rendering the DNA “operable” (i.e., the ability to create a synthetic DNA sequence that actually functions within a living system).
DNA construction refers to technologies that can be used to produce a desired DNA molecule de novo. The general and overlapping terms “DNA synthesis” and “DNA assembly” are included in this category. Much of modern biotechnology depends on having DNA molecules of defined sequence; synthetic DNA has been used, for example, to advance understanding of the basic workings of the genetic code, to enable modern DNA sequencing, and to develop and enable common use of PCR. In addition, gene editing technologies such as zinc finger nucleases, TALENs, and CRISPR/Cas9 each depend on some amount of synthetic DNA. Decreasing costs and increased production scales have made it far more feasible to use synthetic DNA for a variety of purposes. Before DNA construction technologies became available, the only way to obtain a particular DNA segment of interest was to find it in an organism. Now, nearly any DNA—whether natural or designed—can be obtained by simply ordering the sequence to be synthesized from one of many commercial suppliers or by making it on a laboratory DNA synthesizer. While DNA is the most common product of DNA construction technologies, these technologies can also be used to create synthetic RNA molecules and chemical modifications to DNA or RNA.
This access is tremendously enabling for the many beneficial uses of biotechnology, but also has ramifications for potential malicious use. For example, DNA construction could conceivably be leveraged to make toxins, enhance a pathogen, re-create a known pathogen, or even create an entirely new pathogen. Generally speaking, ready access to synthetic DNA allows designers to construct, test, and revise their designs more easily. Many DNA synthesis companies have agreed to screen orders in accordance with guidelines from the U.S. Department of Health and Human Services (HHS, 2015), although limitations of these guidelines have been described (Carter and Friedman, 2015).
Factors that may impact the level of concern related to DNA construction capabilities include cost, time,
ease of access, and difficulty of rendering the DNA “operable.” The size of a segment of synthetic DNA (a DNA construct) is typically described in base pairs for double-stranded DNA and nucleotides for single-stranded DNA. DNA constructs can range from a few nucleotides to several thousand base pairs to entire genomes. Generally speaking, longer DNA constructs are more difficult to produce (or assemble) and using them requires additional laboratory skills compared to shorter constructs. The following examples describe potential uses of DNA construction in ascending order of length and complexity.
Oligonucleotides (Several to Hundreds of Nucleotides)
In its most basic form, DNA construction produces oligonucleotides (oligos), single strands of user-defined sequence that can range in length from a few nucleotides to a few hundred. Oligos can be combined to construct longer DNA sequences. Oligos are extremely useful for a wide variety of research tasks that involve manipulating and analyzing DNA, including sequencing and PCR, as well as site-directed mutagenesis and genome-scale gene editing (e.g., using multiplexed automated genome engineering, or MAGE; Gallagher et al., 2014). Although oligos are typically too short to form the types of protein-encoding genes necessary to support more complex biological functions, they can be used to encode regulatory regions (such as promoters or enhancers), certain short polypeptide-based toxins, transfer RNA, and guide RNA molecules such as those employed for gene editing.
Genes (Hundreds to Thousands of Base Pairs)
Most genes range from a few hundred to a few thousand base pairs in length. Synthetic genes are available commercially as either cloned DNA (in which the product is verified as correct and pure, and typically delivered as part of a general circular plasmid DNA vector) or uncloned linear fragments of DNA (which typically contain some amount of undesired mutations). Potential uses for synthetic genes are at least as diverse as the range of genetic functions found in nature. Genes could be used for a wide variety of malicious purposes, for example, to enhance the pathogenicity of an organism or to produce a toxin.
Genetic Systems (Thousands to Hundreds of Thousands of Base Pairs)
Genetic systems are groups of genes that work together to achieve a more complex function but fall short of supporting an entire cell. For example, genetic systems could be used to encode a biosynthetic pathway or to form engineered genetic circuits that combine operations such as sensing, computing, and actuation. Viral genomes can also be considered as genetic systems, and the genomes for several viruses have already been synthesized and used to produce fully infectious virions (Blight et al., 2000; Cello et al., 2002; Tumpey et al., 2005). Viral genomes can vary from thousands to hundreds of thousands of base pairs in length; large viral genomes (e.g., orthopox viruses) are currently more challenging to synthesize than small ones (e.g., polio).
Cellular Genomes (Millions of Base Pairs)
DNA construction can also be used to assemble the genome for an entire single-celled organism. In 2010, researchers synthesized and assembled the DNA genome of the bacterium Mycoplasma mycoides and used that genome to produce a self-replicating cell (Gibson et al., 2010). This was a difficult, time-consuming, and costly process. At about one million base pairs, the synthetic genome was also one of the smallest known in the microbial world. Nevertheless, this feat demonstrated that it is possible to re-create a living, reproducing organism based on its genetic data. In this case, researchers “booted” their synthetic genome by inserting it into the cell body of a closely related organism, leading to complete replacement of its natural genome with the synthetic one. It remains to be seen how generalizable this approach can be for larger microbial genomes and other types of cells. Other researchers are currently pursuing the construction of bacterial and yeast genomes ranging from 4 to 11 megabase pairs in length; these efforts also use an existing close relative, replacing or “patching” the natural genome with large fragments of the synthetic genome (Richardson et al., 2017). Concerns have been raised about the possibility
of using whole-genome construction to generate dangerous organisms that otherwise could not be obtained without attracting attention (or might not be obtainable at all).
Editing of Genes or Genomes
A variety of technologies allows the modification of specified bases or genes within a pathogen, vector, or host. Such technologies could potentially be utilized to imbue pathogens with new functions; for example, site-directed mutagenesis capabilities could allow the construction of viral variants with novel properties such as altered immunogenicity or species range. Examples include oligonucleotide-meditated mutagenesis, recombination-mediated genetic engineering (“recombineering”) and related techniques (Murphy and Campellone, 2003; Ejsmont et al., 2011), CRISPR/Cas9-based genome editing approaches, and MAGE. Most significantly, newer gene editing platforms such as CRISPR/Cas9 enable the modification of a wide range of organisms. Both the ease with which pathogens can be modified and the types of possible phenotypes that could arise from such modifications would be relevant to an assessment of vulnerabilities related to gene or genome editing.
In the past, genome engineering was a painstaking process that required individual genes to be modified serially. Now, however, multiple genes can potentially be modified in parallel and iteratively. For example, with MAGE, multiple synthetic oligos are created that differ from the existing host genome in at least one base pair. These synthetic oligos are then inserted into a population of cells, where they essentially overwrite the targeted portion of DNA in the cells. MAGE has been used to optimize metabolic pathways, turn off sets of genes, tune gene activity up or down, and engineer a microbial genome with an altered genetic code.
While the biochemical mechanisms MAGE relies on are common throughout both simple and complex organisms, MAGE has primarily been demonstrated in Escherichia coli, and the work required to adapt MAGE to a new species may prove cumbersome. In contrast, genetic engineering and CRISPR/Cas9-based technologies may allow engineering in many new species, providing convenient paths to the further identification of altered phenotypes via either high-throughput screening or directed evolution of organisms with radically new phenotypes and genome-wide sequence changes.
One of the watershed differences that has been enabled by improvements in DNA construction is the ability to generate large libraries of genetic variants. Such libraries can be sieved for improved phenotypes without knowing precisely what variants will arise. This contrasts with the more deliberate process of gene and genome engineering described above (Editing of Genes or Genomes), but there are overlaps between the two approaches because an increased knowledge of how genotype relates to phenotype can guide library design and thereby improve the probability that a given phenotype will be achieved. As an analogy, library construction techniques allow the construction of many more “darts,” and knowledge of genotype-to-phenotype relationships, gained through experiments with gene and genome editing, provides an increasingly larger “target” at which to throw those darts. In particular, the ability to construct degenerate oligonucleotides in a wide variety of ways, including by codon mutagenesis or with nucleotides that are inherently mutagenic, provides a means to construct both large and relatively targeted libraries.
Because DNA can span thousands or even millions of base pairs, designers typically prioritize which parts to vary based on analyses and educated guesses about which changes are most likely to yield the desired results. For example, a designer may use protein structure analysis and visualization software to identify specific parts of a protein that might affect the desired function, such as its enzymatic specificity, build proteins with random variation in those specific parts, and then test how each random variation affects enzymatic specificity.
Booting of Engineered Constructs
With some exceptions, synthesized DNA (or RNA) does not perform biological functions on its own. The process of inducing raw genetic material to perform biological functions is known as “booting,” a term borrowed
from computer technology, where booting refers to the ability to execute functions on digital information by taking it out of storage and putting it into an active state. Booting a synthetic construct is most relevant to the Build and Test phases of the DBT cycle. In the context of biodefense, booting may also be important for a malicious actor’s ability to deliver a bioagent to a target.
Booting in biological systems can take many forms. In the context of viruses, booting may be broadly considered to mean that viral nucleic acids are delivered to cells, where the viral nucleic acids are subsequently able to replicate. A few viruses have been booted by merely delivering their genetic material into host cells, whereas others require additional genetic components expressed separately in host cells in order to produce infectious viral particles. In the context of bacteria, researchers have successfully booted synthetic bacterial genomes by replacing part or all of the genetic contents of natural or synthesized cells with a partial or full synthetic genome. Booting a fully functioning, self-replicating bacterium is significantly more complex than booting a virus.
Perhaps the simplest example of booting engineered constructs is through the use of episomes, pieces of genetic information that can autonomously replicate but typically cannot be readily transferred between cells. Plasmids (typically found in prokaryotes) and extrachromosomal linear arrays of DNA (typically found in eukaryotes) are examples of episomes. Episomes are the most common vector that synthetic biologists use to boot engineered constructs, and there are many available techniques to boot episomes. Although episomes in general are not as complex as full viral or bacterial genomes, they can be used to, for example, introduce a viral genome into a cell and then use the host cell’s transcription, translation, and replication machinery to boot the virus. It may even be possible to use a similar approach to boot a free-living organism. It is also possible for some episomes to spread through a microbial population and between individuals, albeit in general more slowly than a viral infection would.
Testing is used to determine whether a design or biological product created with synthetic biology tools has the desired properties. Tests are typically performed at many stages of a project; for example, a researcher might use computer models to determine if a design is likely to work, then perform tests to validate that the correct DNA construct has been synthesized, then boot the construct to verify that it is capable of performing the intended biological functions. Testing might involve the use of cell cultures, model organisms in laboratory conditions, organisms in the wild, or even potentially human populations.
Test results can be used to further refine a design based on information gained from experimental measurements and observations, and the DBT cycle begins again. In general, state-of-the-art synthetic biology efforts require a great deal of testing in order to yield organisms with the desired properties, making Test both a crucial step and a substantial bottleneck in the DBT cycle. It is a matter of debate whether malicious actors could skip the Test phase and still successfully carry out a biological attack. While a test can be applied to a single variant, in practice it is often more desirable to carry out multiple tests in parallel (high-throughput screening) or to have organisms “test” themselves (directed evolution).
Automation provides the means to screen thousands to billions of individual variants of an organism for function or phenotype. High-throughput testing in cell cultures is a type of screening test commonly used in synthetic biology. Such tests can be used to answer more specific questions (e.g., did this precise genomic change yield the desired phenotypic alteration?) or more exploratory questions (e.g., did any of these 100,000 combinatorial variants in one viral protein yield the desired phenotypic alteration?). Technologies for cheaper and faster screening are in high demand across the biological and biomedical communities, in particular for “-omics” approaches that are agnostic to the type of organism being tested, such as genomics, transcriptomics, metabolomics, and proteomics.
Screening-based tests are performed serially, evaluating different designs or biological products one at a time. Using multiplexing and automation, researchers have developed high-throughput screening–based tests capable of screening tens to thousands of prototypes. On the other hand, selection-based tests (see below, Directed Evolution) are more difficult to design than screening-based tests, but allow much higher throughput.
In nature, the process of evolution selects the best performers from a genetic pool that includes some degree of random variation. Researchers can use a similar process to create prototype biological components representing multiple competing variations and then select among them for the phenotypes that best match the desired outcomes. Prototypes can vary based on smaller changes—different DNA bases, codons, or amino acids, for example—or based on larger-scale differences such as the configuration of multiple genes within a genetic circuit. Like automated biological design, directed evolution is a synthetic biology technique that spans all three phases of the DBT cycle. By building and evolving constructs with random variations, researchers use directed evolution to refine new designs through an iterative approach. The primary difference between high-throughput screening and directed evolution is that in directed evolution, individual organisms compete for the ability to replicate. For example, genomic variations could be introduced into a modified pathogen to produce a large library of variant organisms, which could then be tested for the ability to grow in the presence of an antibiotic. Directed evolution can thus be used to evaluate millions of prototype biological components in parallel, though typically, only one or a few variants would ultimately emerge as successful.
This approach can allow a researcher to sidestep the need for predictive design by creating libraries of millions or more variants and then selecting or screening them to find those few that have a desired set of properties. For example, a researcher could randomly alter residues within specific genes or across an entire genome and then select for a desired phenotype, such as growth, tropism, or lysis. Importantly, the selection can be carried out directly in a host organism, thus allowing for the selection of host-related phenotypes, such as transmissibility (ability to move from an infected to an uninfected host) or pathogenicity (e.g., necrosis within particular tissues). The most promising variants that emerge can be refined further through additional iterations of rational design or selection, following the DBT cycle. Many of the same methods used for library construction and high-throughput screening can also be used for directed evolution, and these different approaches can be combined. For example, a researcher could conduct a high-throughput screen of variants created by a CRISPR/Cas9 library, MAGE, or DNA shuffling (a technique whereby a set of related genes or genomes is broken down into smaller pieces that are randomly reassembled). The variants selected by the screen could then be selected for growth on a novel substrate, potentially identifying both a gene and an organism whose sequence was not fully included in any of the original precursor genes.
Blight, K.J., A.A. Kolykhalov, and C.M. Rice. 2000. Efficient initiation of HCV RNA replication in cell culture. Science 290(5498):1972–1974.
Brophy, J.A., and C.A. Voigt. 2014. Principles of genetic circuit design. Nature Methods 11(5):508–520.
Carter, S.R., and R.M. Friedman. 2015. DNA Synthesis and Biosecurity: Lessons Learned and Options for the Future. J. Craig Venter Institute. Available at: http://www.jcvi.org/cms/fileadmin/site/research/projects/dna-synthesis-biosecurity-report/report-complete.pdf. Accessed January 31, 2018.
Cello, J., A.V. Paul, and E. Wimmer. 2002. Chemical synthesis of poliovirus cDNA: Generation of infectious virus in the absence of natural template. Science 297(5583):1016–1018.
Chen, T., N. Hongdilokkul, Z. Liu, D. Thirunavukarasu, and F.E. Romesberg. 2016. The expanding world of DNA and RNA. Current Opinion in Chemical Biology 34:80–87.
Ejsmont, R.K., P. Ahlfeld, A. Pozniakovsky, A.F. Stewart, P. Tomancak, and M. Sarov. 2011. Recombination-mediated genetic engineering of large genomic DNA transgenes. Methods in Molecular Biology 772:445–458.
Feldman, A.W., V.T. Dien, and F.E. Romesberg. 2017. Chemical stabilization of unnatural nucleotide triphosphates for the in vivo expansion of the genetic alphabet. Journal of the American Chemical Society 139(6):2464–2467.
Fischbach, M., and C.A. Voigt. 2010. Prokaryotic gene clusters: A rich toolbox for synthetic biology. Biotechnology Journal 5(12):1277–1296.
Gallagher, R.R., Z. Li, A.O. Lewis, and F.J. Isaacs. 2014. Rapid editing and evolution of bacterial genomes using libraries of synthetic DNA. Nature Protocols 9:2301–2316.
Gibson, D.G., J.I. Glass, C. Lartigue, V.N. Noskov, R.Y. Chuang, M.A. Algire, G.A. Benders, M.G. Montague, L. Ma, M.M. Moodie, C. Merryman, S. Vashee, R. Krishnakumar, N. Assad-Garcia, C. Andrews-Pfannkoch, E.A. Denisova, L. Young, Z.Q. Qi, T.H. Segall-Shapiro, C.H. Calvey, P.P. Parmar, C.A. Hutchison III, H.O. Smith, and J.C. Venter. 2010. Creation of a bacterial cell controlled by a chemically synthesized genome. Science 329(5987):52–56.
Herfst, S., E.J. Schrauwen, M. Linster, S. Chatinimitkul, E. de Wit, V.J. Munster, E.M. Sorrell, T.M. Bestebroer, D.F. Burke, D.J. Smith, G.F. Rimmelzwaan, A.D. Osterhaus, and R.A. Fouchier. 2012. Airborne transmission of influenza A/H5N1 virus between ferrets. Science 336(6088):1534–1541.
HHS (U.S. Department of Health and human Services). 2015. Screening Framework Guidance for Providers of Synthetic Double-Stranded DNA. Available at: https://www.phe.gov/Preparedness/legal/guidance/syndna/Pages/default.aspx. Accessed May 22, 2017.
Imai, M., T. Watanabe, M. Hatta, S.C. Das, M. Ozawa, K. Shinya, G. Zhong, A. Hanson, H. Katsura, S. Watanabe, C. Li, E. Kawakami, S. Yamada, M. Kiso, Y. Suzuki, E.A. Maher, G. Neumann, and Y. Kawaoka. 2012. Experimental adaptation of an influenza H5 HA confers respiratory droplet transmission to a reassortant H5 HA/H1N1 virus in ferrets. Nature 486(7403):420–428.
Murphy, K.C., and K.G. Campellone. 2003. Lambda Red-mediated recombinogenic engineering of enterohemorrhagic and enteropathogenic E. coli. BMC Molecular Biology 4:11.
Palumbo, R.N., and C. Wang. 2006. Bacterial invasin: Structure, function, and implication for targeted oral gene delivery. Current Drug Delivery 3(1):47–53.
Richardson, S.M., L.A. Mitchell, G. Stracquadanio, K. Yang, J.S. Dymond, J.E. DiCarlo, D. Lee, C.L.V. Huang, S. Chandrasegaran, Y. Cai, J.D. Boeke, and J.S. Bader. 2017. Design of a synthetic yeast genome. Science 355(6329):1040–1044.
Schmidt, M. 2010. Xenobiology: A new form of life as the ultimate biosafety tool. Bioessays 32(4):322–331.
Thodey, K., S. Galanie, and C.D. Smolke. 2014. A microbial biomanufacturing platform for natural and semisynthetic opioids. Nature Chemical Biology 10:837–844.
Tumpey, T.M., C.F. Basler, P.V. Aguilar, H. Zeng, A. Solórzano, D.E. Swayne, N.J. Cox, J.M. Katz, J.K. Taubenberger, P. Palese, and A. Garcia-Sastre. 2005. Characterization of the reconstructed 1918 Spanish influenza pandemic virus. Science 310(5745):77–80.
Wollert, T., B. Pasche, M. Rochon, S. Deppenmeier, J. van den Heuvel, A.D. Gruber, D.W. Heinz, A. Lengeling, and W. Schubert. 2007. Extending the host range of Listeria monocytogenes by rational protein design. Cell 129(5):891–902.
Yaung, S.J., C.M. Church, and H.H. Wang. 2014. Recent progress in engineering human-associated microbiomes. Methods in Molecular Biology 1151:3–25.