The Cell That Still Needs Someone to Feed It

I have spent five years asking fusion physicists when. This week the question landed in biology instead, and I want to say plainly, before anything else, that this is not my beat. My doctorate is in plasma confinement, not membrane biophysics, and a science correspondent who wanders outside her lane owes the reader that admission before she opens her mouth. I am bringing an outsider's skepticism to this story, not a specialist's, and you should weigh what follows accordingly. But the question I know how to ask travels fine across the border: what, precisely, was demonstrated, and on what timescale does it happen again without somebody holding its hand.

On Wednesday, a team led by Kate Adamala at the University of Minnesota, working with co-lead Aaron Engelhart, announced they had built a cell from scratch — not modified, not stripped down from an existing organism, but assembled from roughly 150 to 200 non-living molecular components with no undefined ingredients anywhere in the mixture. They call it SpudCell, a small joke at Sputnik's expense: a first launch, not a finished mission. Fed continuously and held at a steady 86 degrees Fahrenheit, the droplets grew by fusing with each other, copied their DNA, divided into daughters, and did it again — reportedly across five generations, with roughly twelve hours between each division. Some of those generations even competed: engineered variants reportedly outcompeted their unmodified siblings for scarce nutrients, a small but real flicker of selection, the mechanism that makes evolution more than a word.

Adamala's own framing of the result is worth quoting rather than paraphrasing, because paraphrase tends to sand the edges off a claim this large. "We've replicated in chemistry what only used to be possible in biology," she has said: "the complete set of behaviors of a cell." And more pointedly still: the work shows that "the most fundamental functions of life, like growth and replication, do not need a mysterious magical spark." That second sentence is the one doing the philosophical work, and it is worth pausing on. Origin-of-life research has always circled a version of this question — whether life required some irreducible, un-recreatable ingredient, or whether it is chemistry all the way down, arranged well enough. Adamala is arguing for the second answer, on the strength of one lab result, five generations deep, not yet reviewed. That is a much bigger claim than "we built a cell," and it deserves to be weighed as its own separate assertion, not smuggled in as a footnote to the engineering.

What "built from scratch" actually means

Synthetic biology has spent two decades building minimal cells from two different directions. One is top-down: take a living organism — in the most famous case a bacterium at the J. Craig Venter Institute — and strip its genome down to the smallest set of genes that will still let it live and divide. That 2016 project, JCVI-syn3.0, got the count to 473 genes, and even that stripped-down cell is still, unambiguously, alive, descended from something that was always alive; John Glass, who led it, has spent the years since trying to figure out what half of those 473 genes actually do. The other direction is bottom-up: start with no organism at all, just chemistry — lipids, nucleotides, enzymes, the vocabulary of biochemistry rather than biology — and see how much of what a cell does you can rebuild without borrowing a living thing's machinery wholesale. Jack Szostak's lab has pursued a version of this for the better part of thirty years, chasing the question of how the first cells on Earth could have assembled themselves before evolution existed to help. Bottom-up synthetic cells have existed for years under that program; they have metabolized, they have grown, a few have even split in two. What none of them had done, reportedly, until this week, was the whole cycle in one system: feed, grow, replicate the genome, divide, and do it again across generations, with the better-suited variants measurably outcompeting the others. That is the claim, and it is worth sitting with before we complicate it, because if it holds up it is a genuinely significant result — evidence that the specific behaviors we call "alive" are not, in Adamala's own phrase, evidence of "a mysterious magical spark," but outcomes of chemistry arranged precisely enough.

The numbers, because they are the whole story

• Roughly 150 to 200 molecular components — against the low billions of distinct molecules in an ordinary living cell
• A 90-kilobase-pair genome, reportedly split across roughly seven separate DNA plasmids, smaller than the 113-kilobase-pair genome once estimated as the theoretical floor for a living cell
• Five generations of replication reported in the disclosed experiments
• A roughly twelve-hour replication cycle, held at a steady 86°F (30°C)
• Ribosomes borrowed from E. coli and supplied externally at every feeding — SpudCell does not yet make its own

Demonstrated, not yet reviewed

Here is where the fusion reporter's habits turn out to travel well, because this is exactly the seam I've learned to check first: what was demonstrated, and what was merely announced. As of Wednesday, the SpudCell paper had not been peer-reviewed. According to STAT News, the manuscript — reportedly some 190 pages — was submitted to the journal Cell and rejected, with at least one reviewer objecting, in terms blunt enough to be worth quoting exactly: that SpudCells "were not real biology." Adamala's team then shared the manuscript with journalists under embargo ahead of posting it to the preprint server bioRxiv, which means the public got the announcement before independent biologists working the formal, adversarial review process got their say. None of that makes the underlying chemistry fake. But it means every superlative in this piece — first, complete, demonstrated — currently rests on the authors' own account of their own experiment, checked so far by briefed journalists and consenting outside scientists, not yet by the anonymous process biology built specifically to keep itself honest. "Not real biology" is a testable objection, not an insult, and it deserves an answer with data attached, not a press cycle. I would also note, gently, that going to journalists before going to bioRxiv is a choice, and it is not the choice a team makes when its first instinct is to let the data speak before the story does.

The most honest sentence in a cell-biology press release is usually the one about what the cell still cannot do without you.

What "reproduces" means here

It is worth being exact about the word "reproduces," because it is doing a great deal of work in every headline this week, including, probably, mine. SpudCell reproduces the way a wind-up toy walks: genuinely, mechanically, and only for as long as someone keeps intervening. Growth here means individual lipid droplets fusing with others in the surrounding broth, a process the researchers supply by hand. Division was reportedly assisted rather than autonomous — an initial split induced with the protein streptavidin, later divisions pushed along mechanically through a fine mesh. And the ribosomes, the machines that read RNA and build protein, arguably the hardest piece of cellular machinery to fake, are not made by SpudCell at all. They are E. coli ribosomes, harvested from a living organism and added at every feeding, and by the researchers' own account they function for a while and then stop, replaced at the next feeding rather than renewed by the system itself. That single detail is, to me, the clearest measure of the gap between demonstrated and announced: a cell that cannot yet build the one machine every cell needs to build anything else is a cell that has not yet closed its own loop. Selection across generations is real and worth taking seriously — it means variation plus differential survival, the actual mechanism, not a metaphor for evolution — but it happened inside a vessel a person was feeding on a schedule, at a temperature a person was holding constant. Five generations under continuous, hand-delivered life support is a real result. It is not yet, on any timescale anyone involved has stated, a system that keeps itself alive.

The safety case, which I believe, for one specific reason

There is a place in this story where I would normally reach for my sharpest skepticism, and it is the one place I do not need to: the researchers' safety claims. Adamala's team says SpudCell poses no biosafety risk and could not be adapted into a biological weapon, and for once the claim and the mechanism are the same fact stated twice. A system that cannot make its own ribosomes, cannot divide without a researcher supplying essentially every component including protein-synthesis machinery borrowed from a different organism, and has never been shown to survive a missed feeding, has no path to uncontrolled replication outside a lab actively sustaining it. That is not a promise about the future; it is a description of the chemistry as built. I will take that one at face value, the same way I would take a fusion company's confinement-time number at face value if I could see the diagnostic data behind it — it is a specific, checkable claim about what the system does, not a claim about what it might someday do. Notably, the same dependency that makes SpudCell safe is the dependency that makes it, for now, not really alive by the ordinary meaning of the word. Those two facts are the same fact, viewed from two directions, and I trust a claim more when I can see it cut both ways like that.

Openness, and the odd shape it took

The team is not keeping this to itself. Adamala and partners outside the university are launching Biotic, described as a public-benefit research and engineering institution meant to build shared technical infrastructure for synthetic-cell work and keep it open to researchers anywhere, rather than locked behind one lab's patents. That is a genuinely good instinct, and it is the kind of move I've learned to credit in my own beat, where companies that stop hoarding their data usually turn out to have more confidence in it, not less. It sits a little oddly, though, next to the sequence of this particular announcement — journalists briefed under embargo, a manuscript already once rejected, a preprint posted only after the news cycle had already formed its first impression. Openness about the tooling and openness about the evidence are not the same kind of openness, and this week the field got a clear dose of the first while still waiting on the second.

What the field is actually arguing about

None of this should read as a debunking, because it is not one. Jack Szostak, whose own lab has spent decades on exactly this bottom-up problem, says he does not know of another effort to assemble an artificial cell from biological components that has gotten this far. John Glass, who led the top-down JCVI minimal-genome work, called combining all of these functions in a single system "a staggering technical accomplishment" — a compliment that carries real weight coming from someone on the other branch of the family tree. And the evolutionary biologist Michael Lynch offered the most useful kind of caution, the kind that credits the work while still drawing the line: he called it "a synthetic biology tour de force" and, in the same breath, warned against overhyping a system that still cannot sustain itself. That is roughly where I land too, reading this from outside my own field — three real, credentialed scientists, none of them cranks, disagreeing not about whether something happened but about what to call it. The framing that SpudCell shows "most of the hallmarks of life," which has circulated widely this week, is probably the most defensible headline available, precisely because "most" is doing honest work in that sentence instead of hiding from it.

On what timescale

Adamala named this thing after a satellite for a reason: Sputnik did not put anyone on the moon, it announced that the moon was now a place people could plausibly go. If SpudCell survives peer review roughly as described — and I want to be honest that "roughly as described" is still an if, not a when — it will mark the year bottom-up synthetic biology stopped being a promising set of separate tricks and became, briefly, on a bench, in a warmed dish, a genuine cycle. What it will not mark, not yet and not on any timescale its own creators have stated, is the year a cell built by hand learned to let go of the hand. Adamala says the questions in her lab now are about what else biology can do outside biology's original packaging, and that is a good question, possibly the right one. My question, the one I would ask a fusion company and the one I will keep asking here, is smaller and less poetic: how many feedings before this needs no one. Nobody at the University of Minnesota has put a date on that, and I'd trust the work more, not less, if it stays that way for a while. Until there is an answer, the honest version of this story is the one where a remarkable, real, chemically exact cell cycle happened five times in a row, and every single time, someone was standing there with the next meal.