3

URoL: Building a Synthetic Cell

In 2018, NSF published a call for proposals under URoL:Building a Synthetic Cell.¹ Proposals were received in two stages. Applicants from a broad range of disciplines first applied to participate in an intensive 5-day NSF-sponsored Ideas Lab workshop, designed to foster collaborative thinking and generate highly innovative research projects. As described in the NSF solicitation:

Researchers at the Ideas Lab were then invited to submit full proposals to NSF, based on ideas and collaborations developed from the Ideas Lab. The charge from NSF to researchers submitting the full proposal was as follows:

The workshop to review projects relevant to URoL:Building a Synthetic Cell was held on March 21, 2023. Seven PIs participated in the live discussion. These PIs were joined by three moderators—Corey Wilson (Georgia Institute of Technology), Kate Adamala (University of Minnesota), and Eric Gaucher (Georgia Institute of Technology)—all of whom conduct multidisciplinary research in synthetic biology. Forty individuals watched the live webcast, and 7 have viewed the recording as of May 10, 2023.

Adamala summarized written responses to the 2022 questionnaire by PIs from projects represented in the workshop. These comments have been incorporated into the sections below.

SCIENTIFIC ADVANCEMENTS

Participants described their research projects and briefly outlined their scientific findings to date. A brief overview of these research projects is provided here.

___________________

¹ See https://www.nsf.gov/pubs/2018/nsf18599/nsf18599.htm (accessed April 10, 2023).

Building an RNA Chromosome from Scratch

How big of an RNA genome can one build using genetic engineering? Jef Boeke (New York University Langone Health) is collaborating on a project that was pitched in the Ideas Lab by Joel Bader (Johns Hopkins University School of Medicine) and Kaihang Wang (California Institute of Technology), based on observations of RNA replicons that emerged from in vitro transcription reactions.² The group approached this question by starting with a narnavirus, a small and simple naked RNA virus with a genome of roughly 2.5 kilobases, which they aimed to expand to a replicon of 50 kb or more.

To their surprise, the researchers discovered a unique structural feature of narnaviruses that prevents their genomes from being extended by as little as 10 bases. The narnavirus genome is highly structured throughout its sequence, a feature the researchers call “pervasive RNA folding,” and even slight disruptions to this structure cause viral RNA to be degraded in the host cell. To do this study, Bader developed new ways of visualizing and predicting RNA secondary structure. The narnaviruses are revealing themselves to be a diverse clade with multiple subgroups possessing a range of biological attributes, said Boeke. The group is still trying to expand an RNA genome, starting with a mammalian RNA replicon based on Sendai virus. Asked about the possible role of protein binding in stabilizing the narnavirus RNA genome, Boeke answered that, other than the translational machinery, the only host protein known to be required for narnavirus replication is the replicase, but the group is searching for others.

Building a Neuron

Allen Liu (University of Michigan) leads a team of seven PIs who are working to build synthetic structures with the electrical properties of neurons.³ Whereas some members of the group are taking a top-down approach, introducing neuronal ion channels into non-neuronal cells, Liu’s project involves building a synthetic neuron from the bottom up, reconstituting the minimum set of ion channels needed to generate an action potential in a cell-free system. Liu uses a HeLa-based cell-free expression system⁴ to assemble four types of channel proteins (ligand-gated channel and voltage-gated Na+, K+, and Ca+ channels) from DNA. In a neuron, activation of each of these channels in sequence leads to a series of ion fluxes and membrane depolarization events that culminate in the release of a neurotransmitter.

For the ligand-gated channel, which is the first to be activated in the series, Liu is using channel rhodopsin, in order to trigger the action potential with light. He has found that synthetic channel proteins made in the HeLa lysate are able to insert directly into the membrane; by encapsulating the cell-free reaction in a lipid bilayer vesicle, he can ensure that all proteins insert in the correct orientation, which is crucial for generation of a membrane potential. Incorporating multiple channels into the system makes it difficult to test functionality, so each channel needs to be tested independently before being added to the system, he noted. Although he has yet to show that these channels are functional—that remains “the next hurdle” —Liu has demonstrated activity for other types of reconstituted ion channels.

___________________

² See https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7445081 (accessed May 1, 2023).

³ For a brief description of how cells like neurons generate electrical signals, see https://www.nature.com/scitable/topicpage/ion-channel-14047658 (accessed May 1, 2023).

⁴ This is a method for generating proteins in a test tube (rather than in a cell), using DNA and enzymes.

One finding that has emerged from this work is the realization that a transmembrane protein can be synthesized and incorporated into a lipid bilayer without a lot of complicated cellular machinery, said Liu. Evidence suggests that the endoplasmic reticulum-derived microsomes in HeLa cell lysate are sufficient to get direct insertion of these membrane proteins in the bilayer, albeit with low efficiency. Liu is doing similar work in living cells, expressing ion channels in non-electrically active cells to see whether he can make these cells electrically active, which is an extension of previous work in this field. Liu has been able to change cells’ spiking activities by expressing different types of voltage-gated channels.

Building a Cellular Contractile Machine

Mary Elting (North Carolina State University) described her group’s work, which emerged from a larger question regarding how to introduce synthetic or non-native parts into a cell in a way that would allow all the new parts to integrate effectively into the cellular machinery. Their approach is neither top-down nor bottom-up, said Elting, but “starting at the middle, introducing synthetic parts and then adding more and more … until you have something that is wholly synthetic in the long term.” The team focused its efforts on identifying the components that are necessary and sufficient for generating a contraction in a purified system. The long-term goal of this approach is to develop an in vitro system of purified proteins that can generate force with the addition of Ca+ and that can function as a “prosthetic cytoskeleton” in a cell that is not normally contractile (or in an entirely synthetic cell).

One of nature’s fastest-known motors is found in the contractile machinery of some unicellular ciliates,⁵ which can contract to half their body length in under 5 milliseconds, said Elting. The mechanics and underlying biochemistry of this process are very poorly understood; her group is studying it as a potential source of their synthetic cytoskeleton. They have been able to generate a Ca+-dependent contraction in vitro using proteins purified from the ciliate Spirostomum. The group is building a theoretical model to describe this contraction, which appears to be very different from actin-myosin or kinesin-microtubule motors.⁶

Spirostomum’s cytoskeleton is “non-canonical,” said Elting; it is not powered by adenosine triphosphate (ATP) but is instead triggered by a calcium wave that travels from one end of the cell to the other. Using ultrafast imaging of whole cells, her group has found that contractile speed is limited by both the viscoelastic environment and the calcium wave. The force that this cytoskeleton exerts on its cell is “hugely greater than muscle contraction … if I put that into a mammalian cell which doesn’t usually get contracted by a factor of two in 5 milliseconds, can [it] do this behavior that it normally can’t do, or do I just rip it apart?” she asked, adding, “How can we imagine putting synthetic parts together in a cell that’s more like ourselves? Can we give ourselves new abilities that they don’t have by taking abilities from another cell?” For example, she raised the possibility of anchoring a reconstituted protein system on the outside of the cell membrane to form a contractile ring. The group also developed a “magnetic tweezers” approach to move magnetic particles inside cells, which could help deliver synthetic components to specific compartments and modulate cell division. (In answer to a question, she noted that, while bacteria can proliferate through a primitive process of membrane synthesis in the absence of any cytoskeleton, some type of cytoskeleton is required for cell division if, as in the case of eukaryotes, genetic copy number is important.) Elting envisions

___________________

⁵ Ciliates are single-celled organisms covered in hairlike structures called cilia.

⁶ These are two well-studied mechanisms for generating movement in eukaryotic cells.

developing intracellular machines that could be manipulated without disrupting existing cellular processes.

Booting Up a Mirror Cell

In nature, DNA is encoded entirely by D-nucleic acids and proteins by L-amino acids.⁷ The group led by Farren Isaacs (Yale School of Medicine) in collaboration with Adamala seeks to engineer and produce biomolecules that have mirror chirality to their naturally derived counterparts, with one goal being to chemically synthesize D-proteins and encode D-amino acids at the ribosome. Their long-term vision is to construct synthetic mirror cells whose nucleic acids and proteins exist in the opposite chiral states, creating “an entire mirror system that can coexist almost orthogonally from the natural living world,” said Isaacs. This effort will lead to a deeper understanding of the role of chirality and the ways in which ribosomes, polymerases, and translation machinery impose constraints on the stereochemistry of proteins, he added, noting the roles that chirality plays in macromolecular aggregation, organelle formation, patterning, and development. This work could enable the production of new kinds of drugs that are resistant to naturally produced proteases or nucleases.

The team’s recent advances have been propelled by a method developed by Neal Devaraj (University of California, San Diego), called lipid-facilitated native chemical ligation,⁸ which improved the group’s ability to chemically synthesize large stretches of D-peptides and D-proteins. Devaraj has adapted this method further to synthesize transmembrane proteins, using split intein-mediated ligations.⁹ Irene Chen (University of California, Los Angeles) has found that ribozymes¹⁰ do not demonstrate a strong preference for self-aminoacylation of L- over D-amino acids. Other members of the research team have divided into subgroups to engineer components of the biomolecular machinery needed for generating D-proteins. These include efforts to create synthetases capable of charging tRNAs¹¹ that can stably deliver D-amino acids to the ribosome, which then raises questions related to EF-Tu-mediated ribosome binding,¹² accommodation by the catalytic core of the ribosome, and amide bond formation, said Isaacs. The cytoplasm contains proteases that degrade D-amino acids, so another aspect of his group’s work is focused on identifying these proteases and creating knockout strains.

The group has developed several proofs-of-concept, said Isaacs. They now have the ability to charge tRNAs with D-amino acids inside cells and to drive molecular evolution of “tethered ribosomes,” which are being engineered through selection to preferentially accommodate D- over L-amino acids. “We’ve developed capabilities to engineer the entire ribosome, including the peptidyl transferase center, and have also developed an in vivo fitness landscape of this ribosome and have shown how surprisingly malleable [it is],” he added. Adamala noted that her laboratory’s work evolving the ribosome in vitro is complementary to the in vivo work of Isaacs. The researchers have not yet succeeded in getting two D-amino acids to form an amide bond in vivo, though the literature has shown this in vitro, added Isaacs. Cells

___________________

⁷ D- and L- refer to right-handed (dexter) and left-handed (laevus) configurations of a molecule. Nucleic acids and amino acids are chiral, meaning they can exist in either D- or L- forms. D- and L- forms of the same molecule are stereoisomers (mirror images).

⁸ See https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7380508 (accessed April 10, 2023).

⁹ See https://www.nature.com/articles/s41467-020-16299-1 (accessed May 1, 2023).

¹⁰ A ribozyme is an RNA molecule that can catalyze a chemical reaction, like a protein enzyme.

¹¹ A charged tRNA is bound to an amino acid, which it delivers to the ribosome during protein synthesis.

¹² Elongation factor Tu (EF-Tu) transports charged tRNAs to the ribosome.

likely contain multiple layers of safeguards that protect against encoding of the mirror-image D-amino acids, so “we have to approach the problem from all of those dimensions simultaneously, and then try to put them together as a single system and do systems engineering and optimization,” he added.

In answer to a question, Isaacs noted that D-amino acids generated in his group’s system are able to bind L-amino acids and that Chen’s work indicates flexibility in at least some of the machinery; nonetheless, having a chiral structure is clearly critical both functionally and evolutionarily.

Stitching Up a Wounded Cell

If the aim is to have a synthetic cell that is resilient against physical damage to its membrane, there may be no better model than the giant, pond-dwelling ciliate Stentor coeruleus. Stentor cells, each of which possesses roughly a dozen “micronuclear nodes” containing complete copies of its genome, are known for being able to regenerate after being cut into tiny pieces, which can be as small as 1/27 of the original cell size. In addition, said Sindy Tang (Stanford University), Stentor can heal membrane wounds that cut through more than half the entire cell surface area, with some cells able to heal a cut through 80%. It takes 5 to 10 minutes to close the wound, during which at least 20% of the cytoplasm is lost and the cell may endure influxes of external calcium, toxic chemicals, or radicals. “We’re very interested in extracting any principles that we can find from this cell … to have this wound healing capacity in synthetic cells,” said Tang, whose group is studying both wound-healing and regenerative processes in Stentor.

Tang’s group has built tools to introduce precise cuts in the cell, “which is more challenging than what you might think because the cell is very motile … it’s a strong swimmer,” she said. These tools have enabled the researchers to precisely control wound size and characterize the wound-healing capacity of the cell, including maximum wound size and healing rate. The group has found that, as in other wound-healing models, lipids are delivered from the inside of the cell to the wound site by vesicle-mediated transport. In addition, Stentor uses large-scale mechanical motions—folding up around the wound, contracting, twisting, pulling, and swimming—to physically reduce the size of the wound before it is patched up. “That could be the rule of life moment, where these large-scale mechanical behaviors may be of greater importance than we thought … especially for single cells that are free-living and also might have ways to swim around, like these cilia,” said Tang. This could have a potential application to the design of synthetic cells, she added, noting the many efforts to engineer synthetic cells with cilia or other drivers of mechanical motion. In addition, by studying the cell’s response to cytoplasmic leakage, the researchers hope to identify the minimum set of components required for healing and regeneration.

Unraveling the Engineering Constraints for Synthetic Cells

In his collaboration with Adamala and John Glass (J. Craig Venter Institute), Christopher Kempes (Santa Fe Institute) applies tools of theoretical physics, evolutionary biology, and development to elucidate the constraints that govern the design of bacterial cells and small organisms, by searching for “theories that allow us to take the diversity of life and project those onto simple axes or extract general laws.” Kempes described how he turned these efforts to the problem of engineering synthetic cells. Theoretical approaches to the diversity of life tend to fall at either of two extremes, he said, either “everything’s the same” when viewed from a broad

macromolecular perspective, or “everything’s different” when comparing the particular evolutionary niches and adaptations of individual species. Instead, he suggested that the theories needed for designing synthetic cells must inhabit a middle ground so as to account for similarities among bacterial species while “still tell[ing] us what the key … trade-offs are across that diversity,” he said.

In particular, Kempes’s group is working to derive scaling laws—rules linking a cell’s size to its composition, energetics, and physiology—that account for observed similarities and differences among bacterial species. Noting that “a scaling law is really just cell size … to some exponent,” he said, the researchers have found strong scaling relationships that relate key physiological and metabolic processes to bacterial cell size. In many cases, they have derived theories to predict the values of these exponents, with organisms “evolving to hard and fast physical constraints … really, it’s physics predicting what you should see in the physiology,” he said.

These scaling laws are able to account for radical differences in the physiology and composition of large versus small cells, said Kempes. For example, both theory and observation predict that the smallest cells will mainly consist of DNA and expressed proteins, with a “handful of RNA components;” but as a cell gets bigger, protein and DNA concentrations are rapidly diluted, and RNA becomes much more abundant. “This tells you that if you’re building a small cell, you need to build it a fundamentally different way than if you’re building a large cell, because the ratios of macromolecules … and the overall concentrations are dramatically changing,” he said.

As a result of these differences, small cells are much more phosphorus-dominated and large cells more nitrogen-dominated, said Kempes, which indicates different requirements for growth and therefore the need to design different environments depending on cell size. In collaboration with Sara Walker (Arizona State University), Kempes found that these differences extend to the composition of the genetic information, with certain enzyme categories having strong scaling relationships with genome size.¹³ “Putting all of that together, we start to get a picture of what types of trade-offs you need to have in building a synthetic cell,” he said, suggesting that the next step would be to test these theories by implementing various designs at different scales.

For many of the laws, said Kempes, there are strong constraints on structure at both small and large extremes. “We think that what motivates major evolutionary transitions is that a certain architecture starts to break down,” he added. He also noted that, while cell size is often the dominant variable against which all others are measured, “in a more general context … we want to take the interconnected scaling between all sorts of pairwise physiological features … and get a high dimensional space of trade-offs.” This is particularly important for a synthetic or highly engineered organism, in which one parameter may drastically change, so it would be important to determine “what else comes along for the ride … ask where are the rigid dimensions and where are the flexible dimensions for these perturbations,” which may require more complicated theoretical work.

Building Cells Without Lipids

Noting that living cells are composed of four types of biomolecules—carbohydrates, lipids, nucleic acids, and proteins—the ProteoCell group attempted to construct a synthetic cell

___________________

¹³ See https://pubmed.ncbi.nlm.nih.gov/35217602 (accessed April 10, 2023).

from the ground up without one of these molecules, leaving out lipids, said Vincent Noireaux (University of Minnesota). To form the cell boundary, instead of a lipid bilayer, Noireaux and colleagues are using the polymers polyethylene glycol, dextran, and Ficoll. In a buffer, these polymers spontaneously form cell-sized droplets ranging from 1 to 40 microns in diameter, establishing a liquid-liquid phase separation. Co-PIs Cheryl Kerfeld (Michigan State University) and Christine Keating (Pennsylvania State University) introduced purified structural proteins from bacterial microcompartments into the mix; these proteins spontaneously travelled to the phase separation interface, where they “replaced a phospholipid bilayer … [with] a membrane made of proteins,” said Noireaux. Together with group members Millicent Sullivan (University of Delaware) and Giovanna Ghirlanda (Arizona State University), Noireaux was then able to incorporate cell-free transcription and translation, obtaining “cells” by liquid-liquid phase separation that were able to carry out cell-free gene expression. “It’s a relatively complex solution, but we are able to now do it,” he said, adding that he has “beautiful images” showing co-localization of expressed proteins.

One protein that has been expressed in the droplets is Escherichia coli MreB, a cytoskeletal protein that localizes to the cell membrane. In the cell-free expression system, MreB spontaneously localizes to the interface of the droplet, said Noireaux, who is hoping that differential localization of proteins in the droplets may shed light on a rule of life.

LARGER THEMES AND RULES OF LIFE

Synthetic biology is a clash between “engineering … and the biologist’s intuition that life will never yield to engineering,” said Boeke. Developing rules of life may seem at odds with building a synthetic cell, but serious efforts at synthetic biology depend on understanding an array of cellular processes on a level that has not, in many cases, been achieved. Participants discussed their progress toward identifying biological rules that underlie the processes they are attempting to manipulate.

Development of Tools

Many of the advances to date have been in the realm of building the systems or tools that may be needed to find or test a rule of life, participants said. For example, Liu’s group is developing methods for expressing ion channels in synthetic or non-electrically-active cells. If researchers succeed in demonstrating that the propagation of electrical information can be mediated purely by ion channels—itself a rule, perhaps—then they will be in a position to explore this process at a fundamental level. Similarly, by developing a toolkit that enables generation of a strong contractile force ectopically, in a cell that normally does not contract in that way, Elting and colleagues seek to build a system for interrogating the rules of life that govern mechanical force generation: “How can cells generate mechanical force across evolution, and how many different ways have cells evolved to generate mechanical force?” she asked. This group’s development of magnetic tweezers provides an additional avenue for generating force within cells and extends the potential for discovery to include intracellular trafficking and cell division. Furthermore, Elting noted that, because Spirostomum’s method of force generation is so poorly understood, the act of building is itself an avenue for discovery.

Isaacs said that the proof-of-concept and cell-free efforts being pursued by his group, including the development of new methods of protein synthesis in vitro, are establishing a path toward a mirror cell. “We’re basically learning all the skills and technologies we need, not only

for the opposite chirality building blocks but also for all different kinds of unnatural monomers, including different kinds of amino acid chemistries,” added Ademala. This is a long-term endeavor, Isaacs noted: “Being able to polymerize entire proteins for all 20 D-amino acids is a daunting, daunting task. We’re years away from that, but we want to try to at least establish feasibility.” Much of Tang’s work to date has centered around developing the tools needed to precisely control Stentor wound size and characterize its wound-healing capacity. “The development of liquid-liquid phase separation/coacervate chemistry with a biological building material (protein-based membranes) is a foundational advance that will prove its significance,” wrote one PI.

Flexibility in Design

The tractability of biological systems was a recurring theme during this discussion. One rule emerging from his group, said Isaacs, is that the machinery responsible for producing proteins is flexible in its stereoisomeric preference. “I think there’s a lot of potential,” he said, “to go beyond just what we’re talking about in this context and open up a new possibility of repurposing ribosomes to create entirely new kinds of polymers and materials.” Adamala concurred, noting “how generalizable the technology is … we’re basically learning all the skills and technologies we need, not only for the opposite chirality building blocks but also for all different kinds of unnatural monomers, including different kinds of amino acid chemistries.”

The researchers’ explorations of nonclassical systems of motility and wound repair, including Spirostomum’s ATP-independent contractions and Stentor’s twisting and folding, have opened up new potential pathways for designing synthetic cells. “This might be the rule of life that a lot of people have missed, especially in the wound-healing literature,” said Tang. “There’s nothing [in existing literature] that’s macro-scale involving a whole cell that folds around a wound.” In the case of Noireaux’s group, cells’ flexibility is being pushed to the extreme by eliminating one of the “essential” components altogether. “The rule of life for this project is that we do not rule out the possibility of making a synthetic cell without one of the four types of biomolecules, which is lipids…. We absolutely don’t see any major issue, any experimental evidence that it’s impossible…. So far, it’s absolutely possible,” he said.

Nonetheless, not all biological systems are open to manipulation. One rule that has emerged from his group’s work, “much to our chagrin,” is the narnavirus’ dependence on pervasive RNA folding “from the first base to the last,” said Boeke. This has led his team down a novel path, developing methods to study structure-based recognition of RNA viral genomes, as distinct from the sequence-based recognition of DNA, wrote another PI.

Broad Rules Are Starting to Emerge

The interspecific scaling relationships discovered by his group constitute a major rule of life, said Kempes. These scaling laws reflect the physical constraints governing evolution of diverse life forms. Using one set of exponents, the group derived scaling laws that apply broadly across the diversity of bacterial cells. Using different exponents, they also derived scaling laws for unicellular eukaryotes, metazoans, and large vascular plants. In each case, there is a strong theoretical explanation as to “why certain physiological features are scaling in the way that they do with the overall size of an organism,” he said, adding that “this is very useful for designing small synthetic cells.”

Liu’s group has made non-electrically-active cells active through the expression of ion channel proteins. By swapping out different types of voltage-gated channels, the team has been able to characterize channel-specific differences in spiking activities. This work revealed that the ratios of ion channels matter more than their absolute concentrations when it comes to shaping spikes, which he suggested may be an emerging rule.

MULTIDISCIPLINARY RESEARCH/MULTI-TEAM RESEARCH: EXAMPLES OF CHALLENGES AND STRATEGIES FOR SUCCESS

Gaucher asked participants to share examples of particular challenges posed by multidisciplinary research and strategies they used to address them.

Merging Diverse Approaches to Engineering Cells

Liu’s team includes both bottom-up and top-down (or middle-out) approaches to engineering cells. “Merging these two systems is not easy … that presents a challenge,” he said, though the group did include both approaches in a recent publication. Elting said that her group addressed the challenge of integrating multiple projects by developing a set of questions designed to maintain everyone’s focus on the big picture.

Cross-Disciplinary Research May Work Best Where It Goes Deep

Interdisciplinary projects can take either of two paths, said Kempes. He described the simple approach as “the collage or gluing or stapling together version … you do this, and I’ll do this, and then we’ll put them together in a paper.” He suggested that the more valuable collaborative path is to share enough knowledge over time so that each partner grows to understand the other’s work in sufficient depth to be able to ask hard questions. Experimentalists could challenge theorists about their assumptions and point out when the biology is inconsistent with the model, while theorists could understand the experiments in sufficient depth to evaluate data quality. “Each half of the team is really asking the other half difficult questions,” he said. This can lead to an emergent third path, when “you get to a place where you start to co-design things, where you say, here’s what’s easy in experimentation, let’s build theory to deal with that. Then here’s what’s easy in the theory, are there any experiments that can test that?” This approach has taken hold among both PIs and trainees on Kempes’s team.

One complication of combining theoretical and experimental work, said Wilson, is that “the theorist wants to use their tool, and perhaps their tool isn’t the best tool to solve the problem.” Kempes agreed, noting that experimentalists need to push theorists to ask whether their tools adequately reflect the data they can obtain and if not, to pursue a different intellectual path, though it might take time to develop the right tools. “That sort of honest self-questioning and cross-questioning is really essential,” he added.

EDUCATION AND TRAINING

Gaucher asked participants to discuss the challenges or opportunities that had emerged regarding training students in multidisciplinary teams, acknowledging that the COVID-19 pandemic overlapped with these projects, limiting in-person interactions among team members.

The Value of Exposure to Diverse Approaches

There is a clear benefit to interdisciplinary training, said Liu, particularly for a program that aims to train transdisciplinary scientists. His group meetings bring together physicists, engineers, social scientists, and biologists. “I think the students do learn a lot about how different science works,” he said, noting that “I personally have learned a great deal about thinking about how synthetic cells might benefit society.” Elting concurred, noting the “great opportunity for students to really learn how to talk across disciplinary boundaries.” “You have to make the effort to understand the other discipline, how they think, what they are interested in,” said Noireaux, adding that in his team’s case, it has been a very positive experience for both trainees and PIs.

NSF contributed significantly to the success of these projects; Ideas Lab activities taught “really helpful skills,” said Elting, which she and the other PIs in her group passed on to their trainees. The Ideas Lab also brought together Elting’s team, none of whose members had worked together before. Tang suggested hosting more events like the current meeting and including trainees.

The Importance of In-Person Interaction

Although the PIs in Elting’s group got to know each other at the NSF Ideas Lab, the COVID-19 pandemic struck before their trainees could have a similar opportunity. They finally met and conducted experiments in person at the Marine Biological Laboratory in Woods Hole in the summer of 2022, which “was really key … to have them all be in the same physical space together,” said Elting. “Because our team is so diverse, it took a little while to get to where we all had shared language … an in-person retreat for a couple of days probably would have gone a long way toward that,” she added.

Laboratory visits are “really critical in overcoming barriers across disciplines or to help learn certain techniques,” said Isaacs. His group attempted to overcome travel restrictions with “lots of more targeted subgroup meetings and collaborative efforts … to share resources and knowledge.” He credited his students with taking initiative to reach out across disciplines; “the missing piece was not being able to physically visit other labs for training, which obviously was unique to the recent circumstances.” The proximity of the two laboratories on her team, at the University of California, San Francisco, and Stanford University, facilitates in-person meetings and laboratory visits, said Tang. It also enables the joint postdoctoral researcher, a cell biologist, to immerse herself among the mechanical engineers in Tang’s laboratory and tie the engineering and biology sides together, which is key, she said.

According to several participants, although the pandemic caused real disruptions to work and (in particular) in-person interactions, trainees remained highly committed to their projects and used virtual meeting platforms to maintain active collaborations. However, Gaucher noted that younger faculty members who were trying to learn new skills may have disproportionately suffered, compared to older faculty members who were drawing on established approaches.

Multidisciplinary Training in the Laboratory and Classroom

Participants considered the differences in how graduate students are trained in biology versus engineering. Perhaps in a reflection of the interdisciplinary nature of synthetic cell

research, they did not perceive the divisions between these fields as strongly as had the microbiome researchers. Most students in Isaacs’s laboratory gain both experimental and computational skills, regardless of their formal training. Isaacs also noted increasing efforts at Yale “to forge more interactions and collaborations between the biological sciences and engineering, through coursework, collaborations, and joint mentorships.” Nonetheless, Isaacs noted fundamental differences in how the two fields approach questions, while Wilson described the major difference between his engineering and basic science students as one of motivation, with the engineers driven more toward application than basic research.

Several participants said that courses aimed at training undergraduates to do interdisciplinary work are particularly valuable. As examples, Boeke cited the “build-a-genome” course at Johns Hopkins University and the synthetic biology course at New York University, which combines bench work with computation. “Those kinds of courses are super, super useful for undergrads to prepare them for this kind of work,” he said. Noireaux is an instructor in the Cold Spring Harbor synthetic biology course, which “is an incredible experience for most of the participants,” he said, adding that there are many courses like this where trainees can develop new skills. Noireaux suggested developing additional courses, perhaps as summer programs or year-long immersions after completing an undergraduate degree. These could extend the reach of multidisciplinary education, for example, by enabling a newly minted physics graduate to spend 1 year developing skills in biology or chemistry.

One caveat, said Boeke, is that trainees vary in their ability to function in an interdisciplinary way. As a result, he said, “there’s no simple answer to what’s the best way.” A case can be made for developing true expertise in a single area while being aware of other dimensions that can be explored through interdisciplinary research, he added. In addition, it can be challenging to recruit trainees into an interdisciplinary project, particularly one that is starting from scratch, said Elting. Tang experienced a similar challenge, perhaps owing to the fact that “Stentor is a little weird organism that is definitely far from any of the mammalian systems that people are more used to, [although] once we get someone on board, I think they are excited about the project.”

Empowering Trainees to Reach Out

Trainees should feel empowered to reach out to one another and to PIs across teams, said Kempes. Noting that trainees in Adamala’s laboratory would email him to propose ideas, share drafts, and ask hard questions, he said, “I think that’s a really good sign that the trainees have become invested and feel some ownership over the project.” This is also a sign that PIs have created a comfortable environment conducive to such efforts, noted Gaucher.

SOCIETAL IMPACT AND SIGNIFICANCE OF WORK

Noting that his own synthetic biology research tends to be received with a mixture of awe and terror—and that how society perceives this work is important—Wilson asked participants to briefly discuss the significance of their research to society and to address the social issues raised by these innovative projects.

Education Through Novel Forms of Outreach

“A lot of our efforts are aimed at really pushing the limits of highly conserved enzymes and molecular machinery … to do unnatural things and encode unnatural molecules,” said Isaacs.

This has led to his work being lampooned on late-night television as Frankensteinian. Such exposure, he said, can be “an opportunity to put innovations in science in the forefront … to connect with people and broader society and explain to them why we do what we do … and perhaps even inspire them to participate in some way.” “When you’re on the verge of publishing something complex, interesting, but likely to attract attention, it’s important to prepare yourself” with clear descriptions of the work and its significance, said Boeke, whose synthetic yeast project has also elicited comparisons to Frankenstein.

Explaining fundamental research to the public is difficult but necessary, said Noireaux; people need to know that “we are just at the beginning of exploring the potential of living material [and] how it can be engineered and reshaped to create new applications, new material.” Isaacs’s group is seeking to educate non-scientists by enhancing the fundamental understanding of chirality and its role in biology. As part of their outreach effort, researchers from his team are engaging artists to illustrate chirality using relatable ideas, such as mirror images or handedness. Boeke emphasized the importance of “wherever possible, using analogies that anybody can understand” when communicating the complexities of research. It would be useful to draw on past examples from other fields, in which work on purely fundamental questions led to extraordinary applications, said Noireaux. One long-term societal impact of her project, said Elting, would be to broaden the general understanding of biology across the breadth of the evolutionary tree, beyond its current focus on “a few narrow branches.”

Technological Advances to Benefit Society

Manipulating biological molecules to do unnatural things “lends itself to the development of powerful technologies,” said Isaacs, including systems for engineering new types of drugs that could address “long-standing challenges” such as antibiotic resistance and instability.

Tang’s group is seeking to identify laws regarding wound repair mechanisms, which could be applied to synthetic cells or soft micro-robots, to make them more robust and enable self-repair in harsh environments. Elting anticipates similar benefits by developing methods to synthetically generate force, either by cells or micro-robots. Further in the future, Tang envisions the potential use of synthetic cells for biochemical manufacturing. Self-repairing cells would be more resilient to harsh industrial conditions, she said.

Technology development can sometimes lead to unexpected benefits. When the pandemic shut down laboratory research in New York City, Kempes’s group contributed its high-throughput polymerase chain reaction (PCR) facility to the COVID-19 testing effort, teaming up with a local robotics company to build the largest testing operation in New York City that ultimately performed 11 million tests. NSF facilitated the laboratory’s ability to address this immediate social need by continuing to pay trainees, who pivoted from their research to COVID-19 testing, Kempes said.

Runaway Nanotechnology and Other Unintended Consequences

Wilson expressed the concern that Tang’s research might ultimately produce engineered cells that never die. The research is still in its early stages, just trying to understand the basic

science, said Tang. Nonetheless, she raised the idea of providing a “kill switch” that could turn off the self-repair process if an engineered cell became dangerous. The theme of biocontainment and kill switches is becoming increasingly important to funding agencies, noted Wilson.

“With any new technology, you always have some risk of unintended consequences,” said Kempes. It can be impossible to predict how a complex system will interact with everything else around it, he added, citing the example of chlorofluorocarbons and the difficulty of anticipating their long-term effect on the ozone layer when they were introduced as a coolant. A concern raised by biological innovation, he said, is that “simply in the process of trying to build the thing, you might produce problematic things along the way … it’s like someone handed us a very complicated steam engine … and we don’t understand the whole thing. We start tinkering with it, and sometimes it explodes.” Kempes’s team seeks to address this societal concern by developing theories that describe the design constraints from the ground up, “building steam engines from scratch … [so] we understand the hard limits.” Although this process takes time because it requires a basic understanding of the system, it helps to mitigate risks, he said: “Knowing the design principles helps us build things very specifically, and that decreases certain unintended consequences.”

Wilson cautioned that one failure—like an exploding engine—could stimy further development. He wondered whether it was possible to have an open, honest dialogue about failures that could avoid this outcome. Researchers need to recognize the unique challenges of this technology, said Kempes. Tinkering with an advanced machine, the product of billions of years of evolution, is far more challenging than bottom-up technologies that people built in the past, and it requires honest conversations about these unique challenges and risks, he added.

Engaging Social Scientists, Historians, and Bioethicists

Bioethicists on the team need to challenge the scientists on the social and ethical dimensions of their research, said Boeke. He advocated engaging a bioethicist who can illuminate the array of potential societal impacts, such as identifying those groups who will (and will not) benefit from technology once it is available. Elting suggested that non-scientists who might offer alternative perspectives need to be involved in the team from the start.

Elting noted that her team benefited from the inclusion of science historians, who explored how language shapes the public perception of science and how the connotations of words like “synthetic” and “artificial” have changed over time. When communicating about research, it is important to recognize how language can evoke an emotional response in people who lack the scientific context for these words, she said.

Social scientists on Liu’s team conducted interviews with members of the public to gauge their responses to the development of synthetic neurons. They found that people were very enthusiastic and accepting of new technology aimed at benefitting human health, said Liu, who also noted that very few people know, and even those working in the field sometimes disagree, about what synthetic cells are.