Eukaryogenesis

Speculative If the merger that produced the eukaryotic cell can be reconstructed as a sequence of steps rather than named as an event, the prize is a rate: how often a prokaryotic biosphere produces a cell able to carry organelles. The condition comes first because it is not close to met. Two Asgard archaea are in culture worldwide, the closest living relatives of the host lineage, and everything anybody has directly observed about the host side of eukaryogenesis comes from them.

Established The magnitude is best stated as the gap between what the field can grow and what it needs to grow. Promethearchaeum syntrophicum strain MK-D1 took twelve years from deep-sea sediment to a maintained co-culture, passes through a lag phase of 30–60 days, doubles every 14–25 days and needs more than three months to reach full growth at roughly 10⁵ 16S rRNA gene copies per millilitre. Against Escherichia coli at a 20-minute doubling that is a factor of about 1,400 in generation time, and the whole experimental literature on the archaeal half of this problem rests on it and on one Vienna culture doubling every 7–14 days.

Frontier The second prize is astrobiological and it is a denominator. Eukaryogenesis is recorded once in a prokaryotic ocean that ran for at least a billion years before it, and stem-lineage biomarkers span 1,640 to about 800 million years ago before the crown group displaces them. A mechanism would convert that single event into a probability with error bars, which is the number every estimate of complex-life frequency currently substitutes a guess for. It would also tell synthetic biology whether a stable engineered endosymbiosis is a design problem or a waiting problem.

Handwave Who collects this is nobody commercial, and that is a fact about the field rather than a complaint. The gains accrue to astrobiology mission design, to a small cultivation community at JAMSTEC, Uppsala and Vienna, and to the framing of every claim about how common complex life is. The condition on all of it: the upside is real only if the surviving Asgard lineages retain the ancestral traits, and section 5 is about why nobody can currently test that on fewer than decade timescales.

Established Eukaryogenesis is the origin of the cell type that has a nucleus, an endomembrane system, a dynamic actin and tubulin cytoskeleton, and mitochondria — and every eukaryote alive descends from one lineage that had all of it. That last eukaryotic common ancestor, LECA, is reconstructed as already complex: one domain-level analysis places roughly 12,780 genes in it, nearly double the proto-eukaryotic complement, from 10,259 tree nodes inferred to represent an ancestral protein domain.

Established The modern framing is a two-domain tree. Eukaryotes are not a third primary lineage beside Bacteria and Archaea but a branch inside the archaeal domain, specifically within or beside the Asgard archaea, a group first recognised from metagenome assemblies at Loki’s Castle on the Arctic mid-ocean ridge and then expanded into a superphylum carrying hundreds of genes previously thought eukaryote-specific.

Frontier Those genes are the field’s central evidence and its central ambiguity. Eukaryotic signature proteins — actins, profilins, gelsolins, small GTPases, ESCRT components, ubiquitin machinery, tubulin homologues — are present in Asgard genomes in combinations found nowhere else in the prokaryotic world. What they do in an archaeal cell is a separate question from what they became, and until 2023 nobody had seen any of them assembled inside a cell.

Established The second parent is bacterial. Mitochondria derive from an alphaproteobacterial endosymbiont; plastids came later and separately from a cyanobacterium and are not part of this brief. The disagreement is not whether the merger happened but when in the sequence it happened and what the host already was.

Speculative Three families of model compete, and they differ in what the host is doing before the bacterium arrives. The hydrogen hypothesis has a hydrogen-dependent archaeon feeding on a hydrogen-producing bacterium, so the endosymbiosis is the founding metabolic event. The inside-out model has an archaeon extruding blebs around an epibiotic partner, so the endomembrane system is built from the outside in. The entangle–engulf–endogenize model, proposed with the first Asgard culture, has protrusions and membrane vesicles surrounding a partner without phagocytosis, followed by metabolic delegation. All three are compatible with the genomic evidence, which is the problem.

Established Two other transitions are routinely bundled with this one and should be separated. Plastids came later and independently from a cyanobacterium, and multicellularity arose many times in several eukaryotic groups. Eukaryogenesis is the single event, and its singularity is the whole reason it is interesting.

Speculative The eukaryotic features are not a package that must have arrived together. Nucleus, endomembrane system, cytoskeleton, mitochondrion, sex and linear chromosomes with telomeres are separable innovations, and a model that explains one does not automatically explain the rest — which is why no single account currently on the table covers the whole list.

Handwave The word that does the most unearned work in this subject is “then”. Nearly every published narrative is a sequence, and almost none of the orderings has an observational discriminator attached. Section 4 separates the orderings that could in principle be distinguished from the ones that are currently rhetorical.

Established The measured record begins with a culture that took twelve years. Imachi and colleagues recovered strain MK-D1 from deep-sea methane seep sediment of the Nankai Trough off the Kumano area of Japan at 2,533 m water depth, ran a bioreactor pre-enrichment, and then carried out a final seven-year in vitro enrichment. The published growth characteristics are specific: a lag phase of 30–60 days, a doubling time of approximately 14–25 days, and more than three months to reach full growth at around 10⁵ 16S rRNA gene copies per millilitre.

Established The cell is small, has protrusions, and has no organelles. MK-D1 cells are small cocci, 300–750 nm in diameter with an average of 550 nm across fifteen measured cells, producing membrane protrusions about 80–100 nm in diameter that are long and often branching, plus membrane vesicles of 50–280 nm. The published description states that the isolate has no visible organelle-like structure. It degrades ten amino acids and disposes of electrons as hydrogen and formate.

Established It is not a monoculture and cannot be one. MK-D1 is an obligate syntroph: it depends on a hydrogen- and formate-consuming partner, and what the twelve years produced was a two-member association with Methanogenium sp. strain MK-MG after the sulfate reducer Halodesulfovibrio sp. MK-HDV was removed. The authors describe the result as a pure co-culture, which is the honest phrase and is not the same thing as an isolate in the ordinary sense.

Established In July 2024 the organism was validly named, and the name changed. The formal description establishes Promethearchaeum syntrophicum gen. nov., sp. nov., with family Promethearchaeaceae, order Promethearchaeales, class Promethearchaeia, phylum Promethearchaeota and kingdom Promethearchaeati; the type strain MK-D1 is deposited as JCM 39240 and JAMSTEC 115508. The description records a circular chromosome of 4.32 Mb at 31.1 mol% G+C, growth between 4 and 30 °C with an optimum near 20 °C, chemoorganotrophy on amino acids, peptides and yeast extract, and cell yields up to 6.7 × 10⁶ 16S rRNA gene copies per millilitre on the best substrate. The 2020 Candidatus spelling, Prometheoarchaeum, is superseded.

Established The second culture exists, and it came from a different lineage and a different laboratory. Candidatus Lokiarchaeum ossiferum, strain Loki-B35, was enriched at the University of Vienna from coastal sediment; growth was first detected after 140 days and selective transfers reached enrichments of 25–80%. Generation times are about 7–14 days. Cells are coccoid at 0.3–1.0 µm with extensive branched protrusions carrying frequent constrictions and bulbous expansions.

Established That culture produced the first direct structural evidence for an archaeal actin cytoskeleton. Cryo-electron tomography of Ca. Lokiarchaeum ossiferum resolved twisted double-stranded filaments running through cell bodies and protrusions, frequently bundled, with helical parameters reported as a rise of 27.9 Å per subunit and a twist of −167.7° per subunit — described by the authors as highly similar to eukaryotic F-actin. This is the single strongest cell-biological result in the subject, and it is five years old.

Frontier The structural programme has since moved to individual protein families, mostly outside cells. An Asgard ESCRT-III system has been shown to form helical filaments and remodel eukaryotic-like membranes in vitro; microtubule-like structures have been reported in Asgard archaea; and cryo-electron microscopy has resolved open and closed Asgard chromatin assemblies. None of these papers could be retrieved in full for this brief and each is cited by name below, so the claims here are stated at the level the titles and abstracts in circulation support and no further.

Frontier Where eukaryotes attach to the Asgard tree was treated as settled in 2023 and was reopened in 2025. The 2023 analysis placed eukaryotes as sister to Hodarchaeales, nested inside Heimdallarchaeia. In June 2025 a study using 411 Asgard genomes — 136 newly recovered from Chinese coastal wetland sediments, only two of them from cultures — placed eukaryotes outside Heimdallarchaeia altogether, as its sister lineage.

Established The 2025 re-analysis identifies a concrete mechanism for the earlier result rather than merely disagreeing with it. Njordarchaeales genomes used in the earlier work are argued to be chimeric assemblies: 24–60% of their contigs are assigned to TACK archaea against only 3–35% assigned to Asgard archaea, with differential coverage across metagenomes indicating mixed populations. On the expanded sampling, support for the Hodarchaeales-sister topology falls and support for Heimdallarchaeia-sister rises, with bootstrap support above 95% after site exclusion on two marker sets and Bayesian posterior probabilities of 0.69, 1.0 and 0.8 across the S67, S97 and S150 datasets; re-running the earlier study’s own 57-marker set on the expanded data returns posterior probability 1.0 for the new topology.

Frontier The same study attaches dates, and they are early. Relaxed-clock analysis places the last Asgard common ancestor at about 3.38 Ga with a 3.72–3.06 Ga interval, the Heimdallarchaeia–Lokiarchaeia ancestor at about 3.16 Ga (3.54–2.82 Ga), and crown Heimdallarchaeia between 3.12 and 2.26 Ga — that is, eukaryotic stem lineages diverging before the Great Oxidation Event. Ancestral reconstruction makes the last Asgard–eukaryote common ancestor an anaerobic hydrogen-dependent acetogen with a complete archaeal Wood–Ljungdahl pathway, which the authors read as support for the hydrogen hypothesis.

Frontier The independent record from rocks says the crown group is late. Mid-Proterozoic sediments carry abundant protosteroids — protosteranes, monoaromatic lanosteroids and triaromatic steroids, early intermediates of the modern sterol pathway — from about 1,640 Ma to about 800 Ma, after which they fade as red algae proliferate into the Tonian. The reading is a long-lived stem-eukaryote biota displaced by the crown group some 800 million years after it first appears in the record.

Frontier That reading is disputed at its weakest joint, which is the source organism. Gordon Love has argued in print and in reporting that bacteria could have produced these protosteroids; the response is that modern bacteria make them only in restricted settings such as hydrothermal systems, so widespread bacterial production in Proterozoic oceans is implausible. Nobody has run the abiotic and bacterial baseline experiment that would close the argument, which is why it appears again in section 7.

Frontier On the ordering of mitochondria against host complexity, three positions are on the table and each has a published objection. The mito-late position rests on stem-length phylogenomics reporting that alphaproteobacterial, mitochondrion-targeted proteins have the shortest distances to their closest prokaryotic relatives, implying acquisition after the host had accumulated genes of mixed bacterial ancestry. A direct methodological critique argues the stem-length signal is inverted relative to other phylogenetic work, that many families assigned mixed bacterial ancestry actually resolve with aerobic proteobacteria, and that fast-evolving parasitic eukaryote lineages distort the trees.

Frontier The duplication-timing analysis lands between the two and calls itself that. Dating ancient gene duplications against the endosymbiosis gives a mitochondria-intermediate scenario: cytoskeletal and nucleolar duplications predominantly precede the acquisition, while signal transduction and transcriptional regulation expand after it, on an estimated LECA gene complement of roughly 12,780 with a multiplication factor of 1.8.

Speculative What the whole record does not contain is an intermediate. There is no organism, living or fossil, with an archaeal host and a bacterial endosymbiont short of a mitochondrion. Every reconstruction above is an inference from the two endpoints, and the field’s two cultured archaea are modern organisms with roughly two billion years of their own evolution behind them.

Frontier Question one: does the 2025 topology survive? Two analyses, both using site-heterogeneous models and both published in the same journal two years apart, place eukaryotes inside Heimdallarchaeia and outside it. The 2025 paper offers a specific and testable diagnosis — chimeric Njordarchaeales bins — and the discriminator is whether the earlier topology returns when those bins are replaced by clean genomes from culture or single-cell sequencing rather than removed.

Speculative Question two: what were the protrusions for? Both cultured Asgard archaea make long branching protrusions, and in both the organisms are syntrophs whose partners must be within diffusion distance for hydrogen transfer. Whether the protrusions are an ancestral feature that was later repurposed for engulfment or an ordinary adaptation to interspecies electron transfer is unsettled, and it matters because the entangle–engulf–endogenize model assumes the former.

Speculative Question three: could the host engulf anything at all? The genomic reading is that Asgard archaea lack phagocytic machinery, and a pre-mitochondrial cell is argued to lack the energy budget for phagocytosis; MK-D1 at 550 nm is also far too small to engulf a bacterium in the ordinary way. The E3 model answers this by proposing engulfment without phagocytosis, via protrusion and vesicle fusion. No one has observed it.

Frontier Question four: which alphaproteobacterium? Placing the mitochondrial ancestor within, or outside, the sampled alphaproteobacteria has flipped more than once as taxon sampling and substitution models have changed, and the answer conditions what metabolic story is available at the moment of merger. This brief treats the identity as open.

Speculative Question five: were the eukaryotic signature proteins vertically inherited? The alternative is that some arrived in Asgard lineages by lateral transfer, in which case their presence is evidence about archaeal ecology rather than about eukaryotic ancestry. Recent reviews flag this explicitly as unresolved, and it is the kind of question that a dense sampling of Asgard genomes with confident gene trees could answer and has not.

Frontier Question six: how do these cells divide? Whether division runs on FtsZ, on the ESCRT-III/Cdv system, or on both was posed as an open question when the first Asgard culture was still a preprint and has not been answered by direct imaging of a dividing Asgard cell. It is the most tractable item on this list and among the most informative, because eukaryotic cytokinesis and ESCRT-mediated abscission both trace back through it.

Speculative Question seven: does the timing work with oxygen? A last Asgard common ancestor at about 3.38 Ga and crown Heimdallarchaeia possibly as early as 3.12 Ga put the relevant divergences before the Great Oxidation Event, while the E3 model gives the proto-mitochondrion an oxygen-scavenging role under microaerobic conditions. Those are not inconsistent — a long anaerobic stem is compatible with a microaerobic merger — but nothing currently dates the merger itself, only the lineages either side of it.

Handwave Question eight: why once? The honest state is that a single observed occurrence supports almost no inference about rate. It is consistent with a step of very low probability, with a step of ordinary probability that could only take hold in a narrow environmental window, and with multiple occurrences of which one lineage happened to survive and erase the others. This brief does not know, and neither does anyone who says otherwise in a press release.

Handwave Question nine: is the two-domain tree itself an artefact? This is the minority position and it is weak, but it deserves naming rather than omission: the deepest splits in the tree of life are estimated from a handful of universally conserved proteins with saturated substitution histories, and the same long-branch pathologies that produced the Njordarchaeales problem operate on the eukaryote branch too. The reason to hold this lightly is that the two-domain result is recovered under a wide range of models and datasets; the reason to hold it at all is that so was the 2023 topology.

Established The binding constraint is that the organisms grow on a timescale longer than a graduate degree. A 14–25 day doubling with a 30–60 day lag means a single growth curve is a season and a genetic screen is a career. The second culture at 7–14 days is better by a factor of two and is still three orders of magnitude slower than a laboratory bacterium.

Established Obligate syntrophy compounds it. Neither cultured Asgard archaeon can be grown alone, so every experiment is an experiment on a consortium, and every perturbation has to be disentangled from its effect on the partner. Removing Halodesulfovibrio from the MK-D1 culture was itself a multi-year result.

Speculative There is no genetic system. No transformation protocol, selectable marker, replicating plasmid or knockout has been published for any Asgard archaeon. Without one, every functional claim about a eukaryotic signature protein is either a sequence homology, a heterologous expression in a distant host, or an in vitro reconstitution — three kinds of evidence that are collectively weaker than one deletion mutant.

Established Cell yields are low enough to constrain method choice. Full growth at 10⁵ to 6.7 × 10⁶ 16S rRNA gene copies per millilitre is a working density that suits cryo-electron tomography and single-cell imaging and does not suit bulk biochemistry, proteomics at depth, or anything requiring grams of cells.

Frontier The genomic evidence base is mostly metagenome-assembled and its quality is the live methodological risk. Uneven community representation, incomplete bins and contamination are acknowledged in the field’s own reviews, and the 2025 chimeric-Njordarchaeales finding — 24–60% of contigs assigned to a different superphylum — is the demonstration that this is not a hypothetical worry. Of the 411 Asgard genomes in the largest recent analysis, two came from cultures.

Frontier Molecular clocks on this problem have almost nothing to calibrate against. There is no Archean fossil of an archaeon, so dates for the last Asgard common ancestor inherit their calibration from distant eukaryotic and bacterial nodes and from assumptions about rate variation. A 3.72–3.06 Ga interval on a 3.38 Ga estimate is the honest expression of that, and it is wide enough to accommodate several incompatible narratives.

Speculative The biomarker record has a source-organism problem that instrumentation cannot fix. Protosteranes in a 1.6 Gyr-old rock are a molecular fossil of a biosynthetic capability, not of a clade, and the objection that bacteria could make them is a statement about biochemistry rather than about analytical chemistry. Better mass spectrometry does not resolve it; culturing candidate bacteria and looking does.

Speculative The two cultures are not independent replicates of the question. Both are Lokiarchaeia-adjacent sediment organisms grown as syntrophs; the traits they share may be lineage traits rather than Asgard traits, and there is no way to tell with n = 2. That is a sampling bottleneck disguised as a biological result.

Frontier Nothing in the cultivation record is published as a base rate. The successes are documented in detail and the failures are not, so nobody outside the handful of laboratories doing this can say whether a twelve-year enrichment is typical, lucky or unusually fast.

Handwave The deepest bottleneck is that the target event left no intermediate to study. Every other frontier in this corpus has a system that can be perturbed. Here the phenomenon happened once, roughly two billion years ago, and the objects available are its two endpoints plus a handful of modern relatives that have been evolving ever since. That is a structural limit, not a funding one, and it is the reason this brief sits in Category XI.

Frontier This subject waits first on anaerobic cultivation capacity that nobody counts as research. Decade-long continuous enrichments, high-pressure incubation, low-temperature incubators held at 20 °C for years, and the technicians to subculture at the exact point of maximum yield are the input that produced both existing Asgard cultures. It is infrastructure with a publication rate of roughly one paper per decade per line.

Speculative Second, on archaeal genetic tools reaching a group that has none. Genetic systems exist for methanogens, halophiles and some thermophiles; extending one to an obligately syntrophic, slow-growing, mesophilic Asgard archaeon is a distinct programme and is not currently anybody’s funded objective.

Frontier Third, on phylogenomic methods that can be adjudicated. The 2023 and 2025 results differ not because one used a worse model but because they used different genomes, different marker sets and different site-removal treatments. What is missing is an agreed benchmark — simulated data with a known topology, and a fixed set of high-confidence genomes — against which competing pipelines can be scored before they are applied to the real question.

Speculative Fourth, on cryo-electron tomography throughput. The Lokiactin result required cryo-ET on a slow-growing enrichment culture; the questions that follow it — division machinery, protrusion dynamics, whether anything is ever internalised — require time-resolved imaging of many cells, which is a different instrument regime.

Frontier Fifth, on Precambrian geochronology and biomarker standards. The protosterol story depends on absolute ages for mid-Proterozoic sections and on contamination controls strong enough to survive the field’s long history of retracted ancient biomarkers. This dependency is shared with Biosignature Standards, which owns the general problem of how a life-detection claim is adjudicated.

Speculative Seventh, on lipid biochemistry that neither side of the merger shares. Archaeal membranes are isoprenoid ether lipids and bacterial membranes are fatty-acid esters; eukaryotes have the bacterial kind. Any account of the merger has to explain a membrane replacement, and the reconstitution experiments in section 7 are only as informative as the lipid composition they are run in.

Handwave Sixth, and least tractable, on a theory of what makes an endosymbiosis stable. The field can describe two successful primary endosymbioses and cannot state the conditions that distinguish them from the many transient intracellular associations that never became organelles. Without that, every model of eukaryogenesis is a narrative about one outcome with no control group.

Speculative The decisive experiment is to reconstitute a stable archaeon–alphaproteobacterium syntrophy in culture and demonstrate engulfment. Concretely: take Promethearchaeum syntrophicum or Ca. Lokiarchaeum ossiferum, replace the methanogenic partner with a hydrogen- and formate-consuming alphaproteobacterium, establish that the archaeon grows on it, and then image the association by cryo-electron tomography over enough cell cycles to say whether a partner cell ever ends up inside the archaeal cell envelope. That is the one result that would most change this brief’s assessment, because every competing model of eukaryogenesis makes a different prediction about whether it can happen at all.

Handwave What would count as success has to be fixed before the experiment, or it will be argued about afterwards. Engulfment means a bacterial cell fully enclosed by host membrane with no cytoplasmic continuity to the exterior, observed in more than one cell, with a stated frequency per host cell per generation and a control culture in which the partner is a bead or a non-partner bacterium. Without the frequency and the controls, the result is a micrograph, and this field has been moved by micrographs before.

Speculative The second half of the commissioned test — culture a second independent Asgard lineage — has already been done, and the field should update on that. Ca. Lokiarchaeum ossiferum was enriched at Vienna and published in 2023, from a different lineage and a different environment, and it independently confirmed the branching protrusions and added the actin cytoskeleton. The live version of this test is now the third and fourth lineage, ideally from Thorarchaeia or Odinarchaeia, because two cultures cannot distinguish an Asgard-wide trait from a Lokiarchaeia-wide one.

Speculative A genetic system in any Asgard archaeon would reorganise the whole subject. The minimum useful version is a replicating vector plus one selectable marker, tested by expressing a fluorescent protein. The first target after that is a knockdown of the actin homologue, to test whether the protrusions depend on it — which would convert the strongest structural result in the field from a correlation into a mechanism.

Frontier Cryo-electron tomography of a dividing Asgard cell would settle the FtsZ-versus-ESCRT question. The question was posed publicly in 2019 and is still open. It requires no new organism and no genetics, only enough cells caught mid-division, which is a sampling problem on a culture that doubles every one to three weeks.

Frontier A pre-registered phylogenomic adjudication would settle the topology dispute or expose it as unsettleable. The design: a fixed panel of high-confidence Asgard genomes with chimeric bins excluded by differential-coverage criteria agreed in advance, simulated datasets with known topologies to score each pipeline’s recovery rate, and the competing marker sets applied blind. The 2025 diagnosis — that the earlier result came from bins with 24–60% TACK contigs — is exactly the sort of claim a benchmark can confirm or refute.

Speculative The protosterol source-organism experiment is cheap and nobody has run it. Culture a panel of bacteria known or suspected to make sterol-pathway intermediates, under conditions resembling Proterozoic marine settings rather than hydrothermal ones, and measure protosterane yields. A positive result weakens the stem-eukaryote reading of the 1,640–800 Ma record; a null strengthens it. Either way it converts a dispute between two geochemists into a measurement.

Speculative A systematic metagenomic search for an archaeon with an intracellular bacterium is a natural experiment waiting in existing data. If any modern archaeon anywhere hosts an endosymbiont, it is the closest thing to an intermediate that will ever be available, and the search cost is compute against sequence archives rather than new sampling.

Frontier Heterologous reconstitution of Asgard eukaryotic signature proteins is the tractable functional route while genetics is unavailable. ESCRT-III filaments remodelling eukaryotic-like membranes in vitro is the template; the same treatment applied to Asgard profilins, gelsolins and tubulin homologues, with the archaeal lipid composition rather than a convenient one, would establish what these proteins can do before anyone argues about what they did.

Handwave The experiment that would matter most and cannot be designed is a second eukaryogenesis. Nothing in the record says the transition is repeatable on a laboratory timescale, and no proposal exists with a stated duration, selection scheme or endpoint. Saying so is part of the assessment: a brief that lists only feasible tests would misrepresent how far this subject is from closure.

Speculative The negative results worth publishing here are cultivation failures. The community knows informally that many Asgard enrichments have been attempted and abandoned; none of that is in the literature, so nobody can estimate the base rate of success or the conditions that predict it. A registry of attempted enrichments with substrate, source, duration and outcome would be a genuine research asset and costs nothing to build.

Frontier Finally, a direct test of the syntrophy premise: measure whether MK-D1 grows at all with an oxygen-scavenging partner. The E3 model has the proto-mitochondrion supplying oxygen removal and building blocks under microaerobic conditions. That is a growth assay with a defined partner and a defined oxygen tension, and its outcome is a fact about a living organism rather than about a reconstruction.

Established The apparatus here is unglamorous and is the whole programme. Anaerobic glove boxes, sealed serum-bottle and continuous-flow bioreactor lines held at 20 °C for years, high-pressure vessels for deep-subsurface conditions, and quantitative PCR against 16S rRNA genes as the routine readout because cell counts at 10⁵ per millilitre are impractical by microscopy alone.

Frontier Sample supply is a ship. Both cultured organisms began as marine sediment, MK-D1 from 2,533 m water depth at a Nankai Trough methane seep. Access to deep-sea sediment is a research-vessel and drilling capability held by a small number of national agencies, and it is the physical input this subject cannot substitute.

Frontier Structural work requires cryo-FIB milling and cryo-electron tomography on cells that are scarce and slow. The Lokiactin filament parameters — 27.9 Å rise, −167.7° twist — came from tomography on a 25–80% enrichment, which means most of what the instrument sees is something else. Throughput improvements matter more here than resolution improvements.

Speculative The computational requirement is unusual for a biology subject. Bayesian inference under CAT+GTR with SR4 recoding on hundreds of genomes and hundreds of markers, repeated across progressive fast-site removal treatments, is a high-performance-computing workload, and the fact that the 2023 and 2025 results differ means the field needs to run many more of these, not fewer.

Speculative The engineering ambition beyond analysis is a synthetic endosymbiosis. Stable intracellular bacteria in a eukaryotic host have been engineered; the archaeal version does not exist, and the tooling for it — archaeal transformation, membrane-compatible vectors, selection under syntrophy — overlaps heavily with what Synthetic Biology already treats as its design-versus-fabrication gap.

Frontier Quantitative PCR against a single-copy marker is the field’s standard yardstick and it is an indirect one. Growth curves reported as 16S rRNA gene copies per millilitre depend on copy number and extraction efficiency in a consortium, and converting them to cell counts requires assumptions that are rarely stated. The 14–25 day figure inherits that uncertainty.

Speculative The one piece of kit that would change the pace is a fast anaerobic imaging platform. Live, long-duration, oxygen-excluded light microscopy of a culture doubling every one to three weeks would convert questions about protrusion dynamics and division from tomography snapshots into time series, and no such instrument is standard anywhere.

Handwave No part of this stack has a commercial pull, which sets its pace. Every instrument named above was built for another purpose and is borrowed. That is survivable for structural work and fatal for cultivation capacity, which nobody else needs.

Frontier Synthetic Biology is the nearest existing page and owns the construction problem this brief deliberately does not touch. That brief’s subject is that fabrication became an engineering discipline while design did not — 19 of 88 designed genome segments lethal in the 57-codon Escherichia coli project, a third of the minimal cell’s 473 genes without known function. This brief is about a transition that happened once without a designer; the overlap is the shared observation that nobody can yet predict which cellular configurations are viable.

Speculative Prebiotic Chemical Networks and Homochirality Origins sit on the other side of this event. They ask how the first cells arose; this page starts two billion years later, with prokaryotic life established and the question being why complexity appeared once. The methodological rhyme is strong: in all three, the laboratory demonstrations require an experimenter supplying a step the Earth had to supply on its own.

Frontier Biosignature Standards owns the adjudication machinery the protosterol dispute needs. The question of whether a 1.6 Gyr-old molecule is evidence of eukaryotes is the same class of question as whether a spectral feature is evidence of life, and that brief’s finding — a seven-level confidence scale that no journal or funder requires anyone to use — applies directly.

Speculative Assembly Theory is adjacent by ambition rather than by method. Both subjects want a quantitative statement about how improbable a biological configuration is; that brief documents what happens when such a measure is published before its blind test exists, which is a caution this one should read.

Speculative Xenobiology and Microbiome Engineering carry the practical inheritance. Engineered orthogonality and engineered consortia are the two places where a mechanistic account of stable endosymbiosis would be used, and both briefs document how much harder stable multi-organism systems are than their proposals assume.

Frontier Synthetic Ecosystems carries the closest empirical analogue to the stability problem. Closed multi-organism systems that do not stay closed — BIOS-3 and Biosphere 2 — are the same class of failure as a transient intracellular association that never becomes an organelle, one level up in scale.

Speculative Basal Cognition is the sibling page in this category and shares a specific failure mode. Both subjects attach a large interpretive claim — “this is the origin of complexity”, “this is cognition” — to a small set of well-measured observations in organisms that are hard to grow, and in both the measured layer is considerably more secure than the interpretation resting on it.

Frontier The institutional fact that explains this field’s shape is that its central method takes longer than its funding cycle. Twelve years to a maintained co-culture, seven of them a single continuous enrichment, does not fit a three-to-five-year grant with annual milestones. Both existing Asgard cultures came from institutions able to carry a line without deliverables: JAMSTEC in Japan and the University of Vienna.

Speculative What is missing is a funding instrument for cultivation as infrastructure rather than as a project. The closest analogues are long-term ecological research sites and astronomical facilities, which are funded on the understanding that the output arrives late and cannot be scheduled. Nothing of that kind exists for anaerobic microbial cultivation.

Established Culture collections are load-bearing and are set up for a different kind of organism. The type strain of Promethearchaeum syntrophicum is deposited as JCM 39240 and JAMSTEC 115508, but the depositable object is an obligate two-member co-culture, and the collections’ procedures, quality criteria and distribution terms are built around axenic isolates.

Established Nomenclature governance did work here and it took four years. The 2020 Candidatus designation became a validly published name with a kingdom, phylum, class, order and family in July 2024. The lag is not a scandal; it is the cost of a system that requires a deposited, distributable type strain, and it is why two spellings of the genus are in simultaneous circulation in the literature.

Frontier Deep-sea sampling access is a sovereignty question that this field has not had to answer yet. Both cultures came from sediments within national reach, and the subject has been small enough to avoid the access-and-benefit-sharing disputes that have shaped marine genetic resources more broadly. That is a statement about its current size rather than about its future.

Speculative There is no registry, consortium or standing meeting for Asgard cultivation. Two laboratories on two continents hold the entire experimental capability, and the reagents, media recipes and subculture timings that matter are distributed informally. A single retirement is a material risk to the field, which is not a sentence one can write about most research areas.

Handwave The publication system rewards the phylogenomic half of this field and not the cultivation half. A metagenomic study can produce 223 new genomes in one paper; a cultivation line produces one organism per decade. Until the incentive difference is addressed by something other than exhortation, the ratio of genomes to cultures — 411 to 2 in the largest recent analysis — will keep widening.

Speculative The ethical content here is mostly epistemic, and the main risk is overstatement in public communication. “Scientists culture our ancestor” is the headline this subject reliably generates, and it is wrong in a specific way: MK-D1 is a modern organism related to the host lineage, not the host. A field whose public claims outrun its evidence spends credibility it will need when a genuinely decisive result arrives.

Frontier The second-order use of this work is in arguments about how common complex life is, and those arguments carry policy weight. A confident “eukaryogenesis is a once-in-a-galaxy fluke” feeds rare-Earth reasoning, which feeds arguments about the value of searching for life and, at the margin, about how seriously to treat humanity’s uniqueness. A single observed event does not license that confidence, and this brief declines to supply it.

Speculative Biosafety here is close to nil and should be stated so rather than performed. The organisms are slow-growing obligate anaerobes from deep-sea sediment that cannot grow alone, at 20 °C, with no known pathogenicity. Treating them as a containment concern would be theatre; the real containment concern in this neighbourhood belongs to engineered organisms and is Xenobiology’s subject.

Speculative Sampling equity is a live question in principle and a quiet one in practice. Deep-sea sediments are a shared resource, the capability to sample them is held by a handful of states, and the cultures that result are held by the institutions that made them. Nothing has gone wrong yet; the governance is simply untested at any scale.

Frontier Naming is a small ethical surface here and it behaved reasonably. A four-year gap between a Candidatus designation and valid publication produced two competing spellings in the literature, which costs readers time; the alternative — validating names without deposited, distributable type material — costs more.

Handwave There is a worldview question underneath, and pretending otherwise is worse than naming it. An account of eukaryogenesis as a contingent accident and an account of it as an inevitable consequence of prokaryotic ecology have different implications for how readers understand humanity’s place, and both are currently underdetermined by the evidence. The appropriate posture is to hold both open, which is uncomfortable and correct.

Speculative The civilizational consequence that actually turns on this research is the prior on complex life elsewhere. Every estimate of how often a biosphere produces large, energetically expensive organisms passes through a term nobody can currently populate: the probability per unit time that a prokaryotic ocean produces a cell capable of carrying organelles. A mechanistic account with a rate would replace that guess in every such calculation.

Frontier That feeds directly into mission design and into how a detection would be read. If complexity is a bottleneck, the sensible search targets are biosignatures of simple metabolism and the interesting question becomes how long biospheres persist without complexity. If it is not a bottleneck, the absence of unambiguous complex-life signatures becomes a harder fact to explain.

Speculative The applied consequence, if it comes, is engineered endosymbiosis. A design rule for when an intracellular partnership stabilises rather than collapsing into parasitism or digestion would apply to engineered consortia, to nitrogen-fixing partnerships in crops, and to any attempt to install a new organelle-scale function in a cell. The magnitude is unstated here because nobody has measured it; that is section 1’s point restated.

Handwave There is a temptation to claim this reorganises how a civilization thinks about itself, and the brief declines it. The measured content of this subject over the next fifty years is likely to be a handful of cultures, a resolved topology, and a better-constrained date. Those are real and they are not civilizational.

Speculative A second consequence would be to the reading of the Proterozoic itself. If the protosterol biota is stem-eukaryotic, then the eukaryotic branch existed for roughly 800 million years as an ecologically minor presence before the crown group displaced it, and the interesting historical question shifts from origin to why the delay.

Speculative The one genuinely long-range implication is about method, not content. This is a field where the rate-limiting step is patience — a twelve-year enrichment, a seven-year continuous culture — and it is a working demonstration that some scientific questions are gated by institutional continuity rather than by instruments or ideas. Any civilization that wants answers to questions of this shape has to build organisations that outlive the careers of the people who start them.

These horizons track cultures, topologies and dates, because those are the three quantities this subject actually moves:

• 10 yr: Frontier Expect a third and possibly a fourth Asgard lineage in enrichment culture, on the evidence that the interval between the first and second was three years and that at least two groups are now equipped for it. Speculative Expect the topology dispute to be argued rather than settled, with the discriminator being whether clean Njordarchaeales genomes recover the 2023 result. Speculative Expect more in vitro reconstitution of Asgard signature proteins, and expect it to outpace anything done inside a living Asgard cell. Handwave A genetic system in an Asgard archaeon inside ten years would be a surprise and would be the most consequential single development available on this horizon.
• 25 yr: Speculative This is the window in which the decisive experiment plausibly runs: an archaeon grown on an alphaproteobacterial partner, imaged over enough cell cycles to state an engulfment frequency. Speculative A positive result would make the E3 family of models the working account and would put a number on a step that currently has none. Speculative A clean null — no internalisation at any frequency, under any partner — would be equally informative and would push the field toward inside-out and metabolic-first accounts. Handwave Expect the mitochondria-early/late/intermediate question to resolve by accumulation of better gene trees rather than by any single result.
• 50 yr: Speculative On this horizon a mechanistic account of endosymbiotic stabilisation is plausible, and it is the thing that would convert eukaryogenesis from a historical narrative into a design rule. Handwave Whether that account generalises to a probability per unit time is the step nobody has shown how to take, and stating it as a plan would be exactly the overclaiming section 11 warns about.
• 100 / 250+ yr: Handwave Beyond useful forecasting. The defensible shape is that a subject with two cultured organisms either acquires many more, in which case it becomes an ordinary comparative discipline, or it does not, in which case it stays a phylogenomic argument indefinitely. Handwave Any specific claim about engineered eukaryogenesis on this horizon is an assertion, and this page does not make one.

• Depends on Nothing on this map blocks this subject, and the absence is worth stating plainly: the inputs are cultures, sediment and phylogenomic compute, none of which another brief here produces. Prebiotic Chemical Networks and Homochirality Origins are upstream in time and supply context rather than prerequisites, because this transition begins with established prokaryotic life. Biosignature Standards supplies the adjudication framework the protosterol dispute needs but does not gate any result here.
• Requires (not on this map) Six conditions, in the order of the tokens above. A genetic system. No transformation protocol, marker or vector exists for any Asgard archaeon, so no functional claim about a eukaryotic signature protein can currently be tested by deletion in the organism that carries it. Cultivation capacity. Anaerobic lines held at 20 °C for a decade, subcultured at maximum yield, with no publications in the interim — the input that produced both existing cultures and the one that scales worst. A phylogenomic benchmark. Simulated data with known topologies plus an agreed panel of non-chimeric genomes, so that competing marker sets can be scored before they are applied; without it the 2023 and 2025 results cannot be adjudicated, only alternated between. Sediment access. Research-vessel and drilling capability reaching methane-seep sediment at depths around 2,500 m, held by a small number of national agencies. A biomarker baseline. Measured protosterane yields from cultured bacteria under non-hydrothermal conditions, which is the only thing that will settle whether the 1,640–800 Ma record is eukaryotic. A collection mandate. Culture collections whose deposit, quality and distribution terms accommodate an obligate two-member co-culture rather than an axenic isolate, since that is the only form the organisms come in.
• Enables A mechanistic account would give Biosignature Standards a prior for complex life rather than a placeholder, and would give Synthetic Biology and Xenobiology a stability criterion for engineered intracellular partnerships that neither currently has. No typed enabling edge is claimed, because a field with two cultured organisms enables nothing on a schedule.
• Adjacent Within this map: Synthetic Biology for the design-versus-fabrication gap, Assembly Theory for the hazards of publishing an improbability measure before its blind test, Microbiome Engineering for engineered consortia, Synthetic Ecosystems for closed multi-organism systems that do not stay closed, and Basal Cognition as the sibling Category XI page with the same measured-layer-versus-interpretation structure. Outside the map: archaeal cell biology, cryo-electron tomography, Precambrian organic geochemistry and deep-biosphere drilling supply nearly every number on this page.

Handwave “Asgard archaea are our ancestors.” They are the closest sampled prokaryotic relatives of eukaryotes, which is a statement about a tree, not about descent. Promethearchaeum syntrophicum is a living organism with roughly two billion years of its own evolution since the common ancestor, and its traits are evidence about that ancestor only to the degree they are shared across lineages — which, with two cultures, is largely untestable. Established The tree claim itself is well supported; the ancestry claim as usually phrased is not what the tree says.

Established “The first Asgard archaeon was isolated in pure culture.” It was not, and cannot be. MK-D1 is an obligate syntroph maintained as a two-member co-culture with Methanogenium sp. MK-MG, after twelve years of work whose main achievement was removing the third member. The published phrase is “pure co-culture”. The 2024 formal description calls it the first isolate of the Asgard lineage, which is correct in the nomenclatural sense — there is a deposited type strain — and misleading without the qualifier.

Speculative “The host engulfed a bacterium by phagocytosis.” Asgard genomes are read as lacking phagocytic machinery, a pre-mitochondrial cell is argued to lack the ATP budget for phagocytosis, and a 550 nm coccus is too small to engulf a bacterium the ordinary way. Every serious current model works around this rather than through it, which is what the entangle–engulf–endogenize and inside-out proposals are for. Handwave Note also that the model was called entangle–engulf–enslave in the preprint and entangle–engulf–endogenize in the published paper; both spellings circulate and refer to the same proposal.

Frontier “Eukaryotes are 1.6 billion years old, because the biomarkers say so.” The protosteroid record from about 1,640 to 800 Ma is read as stem-group eukaryotes — organisms on the eukaryotic branch before the last common ancestor of all living eukaryotes. The crown group, on that reading, rises late, after 800 Ma. Conflating stem and crown turns a careful result into a wrong number. Frontier And the eukaryotic attribution is itself contested, on the ground that bacteria might make the same molecules.

Frontier “The phylogeny is settled: eukaryotes are Heimdallarchaeia.” That was the 2023 position and it was reversed in 2025 by a study that identified 24–60% TACK contamination in the Njordarchaeales bins the earlier analysis relied on. The current published state is eukaryotes as sister to Heimdallarchaeia rather than inside it, with the earlier result attributed to an artefact. Anyone quoting either as settled is quoting a snapshot.

Speculative “Lokiactin proves Asgard archaea have a eukaryotic cytoskeleton.” Cryo-electron tomography resolved twisted double-stranded filaments with a 27.9 Å rise and a −167.7° twist, closely matching eukaryotic F-actin geometry, running through cell bodies and protrusions. That is a homologous polymer in the right place. Whether it performs the dynamic, motor-coupled, regulated functions that make the eukaryotic cytoskeleton what it is remains untested, because there is no way to perturb it in the organism.

Frontier “Mitochondria were necessary because eukaryotic genomes need energy per gene that prokaryotes cannot supply.” This bioenergetic argument is influential and has been directly contested on the ground that the relevant scaling comparisons do not hold when made carefully across genome sizes and cell volumes. The dispute is live and neither side has produced a measurement that ends it; both positions are named in the reading list and neither is adopted here.

Handwave “It happened once, so it is essentially impossible.” One observation supports a very weak rate estimate, and the observation is not independent of the observer. A low-probability step, an ordinary step with a narrow environmental window, and several occurrences of which one lineage survived and outcompeted the rest are all consistent with what is on the record. The single-occurrence fact is real; the inference usually drawn from it is not licensed.

Speculative “The nucleus came from a virus.” Viral eukaryogenesis proposals have been in circulation for decades, and they are not absurd: large DNA viruses build membrane-bounded replication factories, which is a genuine structural analogy. They are carried here because omitting them would leave the reader to meet them elsewhere without a flag. The evidential position is weak — the analogy has never been converted into a phylogenetic signal that survives scrutiny — and no current Asgard result bears on it either way.

Speculative “Asgard archaea live only at deep-sea vents.” The first genomes came from a hydrothermal setting and the name stuck, but the second culture came from coastal sediment and the 2025 expansion recovered 223 new genomes from coastal wetlands. They are a widespread sediment group, and the environmental range matters because it is what makes a third and fourth culture plausible.

Handwave “A genome tells you what a cell looks like.” The strongest single lesson of the last six years is the opposite. Asgard genomes were available from 2015; nobody predicted the long branching protrusions, the membrane vesicles, the cell size, or the 14–25 day doubling time from them, and every one of those facts required growing the organism. The inferences in sections 3 and 4 that rest on sequence alone should be read with that record in mind.