This piece is an engineer's take on SpudCell. Enough people have asked us what we made of it that it seemed worth writing down properly. We saw a lot of the commentary on this work ask if it's 'alive.' We tend to ask a different question: what modules did they build, how were they reused or composed, and what's still missing before they can be composed further. We hope you enjoy the potato-themed memes along the way!
The Adamala and Engelhart labs have built "SpudCell," the most advanced simulacrum of life to date: a synthetic cell, physically built up from parts, that replicates and expresses a genome, grows when fed, divides using genetically encoded machinery, and evolves over generations. To do so, they built a synthetic genome (90.3 kb, seven plasmids) that encodes 34 of 38 genes used in their synthetic cytosol "pREP" (PURE + Phi29 replicase), which implements Central Dogma (transcription, translation, and DNA replication). They also built and integrated two genetically encoded modules (feeding and division), developing and documenting generally useful techniques in the process.
What they achieved was a composed system that underwent five replication cycles, albeit under generous conditions with active manual manipulation. This was a huge engineering effort that meaningfully advances the field and declares immediate next steps. This announcement was also paired with the launch of Biotic, a nonprofit whose mission is to conduct and coordinate further development of chemically- and functionally-defined synthetic cells for the public benefit.
Let's get into it.
How SpudCell Works
To implement the "Central Dogma," information in DNA must be copied and passed down (DNA replication) and flow into RNA (transcription) then proteins (translation). SpudCell implements these three functions with pREP (Libicher et al. 2020), integrating two existing systems. PURE is a defined composition of 36 proteins, ribonucleics (ribosomes and tRNAs), and small molecules (e.g., rNTPs, amino acids, etc.) that reconstitute transcription and translation (Shimizu et al. 2001). Phi29 replicase, borrowed from a B. subtilis phage, can copy DNA and is frequently used in this field as well for DNA replication (van Nies et al. 2018). Integrating these two modules and adapting them for this implementation required the authors to substantially reformulate their system (e.g., more energy in creatine phosphate, less in rNTPs, doubling the protein concentration, etc.).
The authors encoded Phi29 replicase, alpha hemolysin (aHL; usage described below), and most (31 of 36) proteins in PURE on a seven-part synthetic genome (90.3 kb).  Notably absent are ribosomal proteins and rRNA; tRNAs; all metabolic enzymes; and, most notably, EF-Tu, which makes up about half the protein (by mass fraction) in PURE. Caveats aside, the authors were able to demonstrate that their system could reliably replicate the full synthetic genome (by qPCR) and that pREP retained translational capacity, using GFP and αHL as fluorescent reporters. The authors note that, without active plasmid maintenance, genome partitioning (i.e., the distribution of the genome's seven plasmids to daughter cells) occurs stochastically causing variance in plasmid distribution throughout the population over generations (more later).
If the SpudCells can't make all of their proteins, or any ribonucleics or small molecules (e.g., lipids), then how do they grow? Simple: feed the SpudCell with smaller synthetic cells!
The authors developed "Feeder Cells:" nanovesicles filled with pREP reaction mixture with membranes decorated with an affinity group (18:1 DGS-Ni-NTA). SpudCells then express a membrane-bound protein (aHL-6xHis) that binds these affinity groups, fusing with Feeder Cells and eating their membranes and innards. For SpudCells, the incorporation of new material is growth. As an added benefit, the high protein content, the ribosomes and tRNA, and the rich synthetic cell media used (essentially the small molecules of the reaction mixture) greatly reduce the metabolic burden of transcription, translation, and small molecule metabolism, reserving more of SpudCellâs resources for other functions.
The authors verified feeding by directly measuring membrane fusion (by FRET) and cytosol mixing (by GFP expression upon DNA template delivery). The authors also verified that feeding was limited by the amount of aHL-6xHis expressed and not by Feeder Cells which are typically in excess (relevant later). Together, these results show that SpudCells can eat and incorporate Feeder Cells, growing in both size and content.
Having demonstrated growth, the authors sought to develop a genetically-encoded cell-division module. The authors note that, at this stage, mechanical extrusion provides higher division yields enabling them to better study the cell-cycle and evolution: add Feeder Cells to SpudCells in synthetic cell media, incubate for 12 h, extrude (2 ”m), and repeat (up to five times). With the experimental set up established, the authors then developed their genetically-encoded division module.
Typically, cell division (like genome partitioning) requires a cytoskeleton and motor proteins. The cytoskeleton provides a rigid structure for motor proteins to exert energy against, else partitioning happens stochastically rather than actively. With SpudCell having none of these, the authors got creative with three components: (1) aHL-FLAG (expressed and self inserting) decorating the SpudCell membrane, which binds (2) an antiFLAG-biotin linker (added directly to the media), which in turn binds (3) the large protein (52 kDa) streptavidin (also added to media). As more streptavidin molecules bind, they increasingly bump into each other, pushing each other apart and inducing membrane curvature. Membrane size grows with feeding (as Feeder Cell lipids are added to the SpudCell membrane), in direct tension with the streptavidin-induced curvature. The result is that daughter cells bleb off their parents in a stochastically-driven process.
To assay this behavior, the authors got creative, immobilizing parent SpudCells with "click chemistry" to magnetic beads and assaying the daughter cells that float away into bulk solution. They then used qPCR to track cells in solution (by counting their genomes) and a custom "cell cycle counter" to observe which generation of Feeder Cells each generation ate.
With their replicating system in place, the authors tested if SpudCell was subject to evolution. They introduced "T7MAX," an allele with a stronger promoter that over-expresses aHL (used in both growth and division modules) and measured how mixed populations of SpudCells with T7MAX and without (here, "WT") changed over time (by flow cytometry).
T7MAX indeed outcompeted, becoming approximately 60% of the total population after five generations (with mechanical cell-division using extrusion). Under stronger selective conditions (0.1x feeder cells, mimicking starvation), this enrichment was even higher (70%). These results, together, demonstrate (partial) component regeneration, growth, division, and evolution in a synthetically-composed, human-designed system.
What SpudCell Isnât (Yet)
This work is ambitious, complicated, and groundbreaking, but be careful: with all its nuance and detail, its easy to misinterpret or overstate these results.
SpudCell does grow, but it does not reproduce. It does not produce itself anew. It lacks any small molecule metabolism and is entirely auxotrophic for the majority of its constitutive parts, including EF-Tu, the largest component of its proteome (by mass fraction). Moreover, it cannot make ribosomes or tRNAs at all, which may require an additional 100 genes or so. Feeder Cells are designed to mimic the composition of SpudCell (minus the genome) and small molecules can enter from the rich synthetic cell media through the aHL pores. Thus the burden of providing new SpudCell parts is largely on the experimentalist rather than on the cellâs metabolism.
Moreover, SpudCell does divide, but not autonomously. Many of these experiments used mechanical methods to coax division, including the majority of the work demonstrating evolution. Even the genetically-encoded division module is still auxotrophic for externally provided molecules (antiFLAG-biotin and streptavidin). DNA replication is arguably the most compelling part of the self-replication story, maintaining a synthetic genome over multiple generations. Still, after five generations without active genome segregation (i.e. only random partitioning of plasmids into daughter cells), only 30% of cells retain their full genome, and one plasmid was only present in half of the population.
A Few Things We Noticed
Taken together, SpudCell is a watershed in the history of biological integration engineering. Underneath this remarkable scientific narrative, are a few features that stood out to us in particular, especially as people who spend a lot of our time thinking about how emerging synthetic cell capacities get reused: legible protocols, remixing of functional modules, processes that take existing modules in new directions. More specifically:
- All the protocols, specifications, and materials are released publicly and in detail with the original preprint and on the Adamala Lab Website.
- The authors remixed and reused two "established" modules (pREP and aHL), invented two new ones (growth and division), and integrated them into a functional system that should open the door for others to explore their technological implications.
- The authors introduce or extend useful techniques for the field, including encapsulation by mechanical extrusion (extended to prepare cells of two different sizes in the same experiment), their microscale "mother machine" that uses beads to trap parent cells and tracks only daughter cells across generations, a FRET-based membrane fusion assay, and a novel RNA-based cell cycle counter.
Where This Leaves Us
This work is rich. Every section spills over with details and lessons learned. The implications here are profound, and the data and experiments support them. This work is undoubtedly a landmark for the field. For us at b.next, weâre excited to see a world where others build on these protocols and integrate the new capacities documented here with existing ones.
The release of this paper coincides with the launch of Biotic, a nonprofit organization that will further this work by directly developing, and coordinating efforts to build, chemically- and functionally-defined cells that make their own parts and sustain themselves. Moreover, they are declaring that their work will be exclusively for public-benefit, with the explicit goal of keeping no single actor in control of the results. "The machinery of life must be common" they say, and we couldn't agree more: technology built from the machinery of life is likely too useful to risk being gatekept by a handful of entities.
We launched Nucleus nearly three years ago out of frustration with the slow pace of progress: âwhy canât we solve the hard problems with this technology alreadyâ. Building new cells from scratch shouldnât mean reinventing every module from scratch too. We recognized that proficiency in this field required fluency across many other fields. Our premise for making Nucleus an open source project was simple: no single entity could assemble all the technologies needed to fully unlock this field. And even if they could, they shouldnât: everyone should be able to solve the hard problems using lifeâs machinery. Concretely, that's part of why contributions to Nucleus, including b.next's own, are made available under acknowledged open-source licenses allowing others to use and modify Nucleus technology for any purpose.
Even so, many of us were drawn to this young field by how easily engineering mingles with the philosophical. Every day we contemplate constructing complex functional systems that are useful, built from lifeâs building blocks. This paper makes that dialogue apparent, inspiring many to ask: "are SpudCells alive?" The authors deftly withhold ultimate judgement, saying they'll "know it when [they] see it". Further, we think the question is poorly posed. Life, and being alive, is increasingly thought of as a spectrum, or a set of qualities, rather than a single binary property. Take viruses, long argued whether "alive" or "not" and clearly somewhere in between. We'd ask you instead to reframe this question as an engineer, asking instead "what life-essential functions do SpudCells demonstrate, and which do we build next?"
The authors have identified the next critical modules on the path to increasingly self-sufficient cells. We need cellular organization, genome segregation, and cell division using a synthetic cytoskeleton. Â We need regenerative metabolism, reproducing key enzymes and, over time, all small molecules, from simple feedstocks. Â And we need ribosome biogenesis to express and assemble working ribosomes (to this, we add tRNA expression and maturation as well).
However, the question âwhat to build nextâ is ultimately for the whole community to work out together. These are exciting times for our field, and we at b.next are grateful to be part of such a vibrant and growing community. Congratulations to the Adamala and Engelhart labs for their "hot potato". We applaud Biotic for its vision, approach, leadership, and commitment to this field. Letâs build together!
â
â