INTRODUCTION
Researchers at EMBL Barcelona have derived muscle, neural and vascular cells from one bovine embryonic stem-cell source. The work matters because it could reduce the assembly needed to make structured cultivated beef.
Nature Communications published the open-access research on 2 September 2026. The authors report that the cells formed tissue-like arrangements in both flat cultures and three-dimensional systems. Muscle fibres developed sarcomeres and showed calcium activity linked to nearby neurons. Endothelial cells formed vessel patterns within the cultured material. The protocol did not use serum, according to the paper. Those results address several technical tasks at once, rather than treating muscle, nerves and vessels as separate manufacturing streams.
The result remains a laboratory demonstration. Cultivated X and vegconomist reported on 3 September 2026 that each aggregate was about 0.6 mm wide. They also said the method used Matrigel, which comes from mice, and cost between €0.39 and €0.99 for each aggregate. The research therefore shows a biological route towards more complex tissue, not a practical process for producing a steak. Scale, maturation, acceptable materials and input costs still stand between the experiment and food.
CORE FINDING
The paper’s central conclusion concerns co-development. The researchers guided bovine embryonic stem cells into skeletal muscle, neurons and endothelial cells while allowing those lineages to organise together. Nature Communications says the approach generated innervated and vascularised muscle rather than a collection of independently prepared cell types. This distinction matters for manufacturing. Conventional tissue-engineering routes can require purified populations, scaffolds, printed bioinks or stacked hydrogel units. The study tests whether developmental biology can perform part of that assembly instead. The evidence supports self-organisation on a small scale. It does not establish that the same process will remain controlled in food-sized tissue.
The cell source also separates this work from common cultivated-beef methods. The Nature Communications paper says producers often begin with myosatellite cells or mesenchymal stem cells taken from bovine biopsies or bone marrow. Those adult cells lose growth capacity during extended culture. Their possible fates are also narrower. Myosatellite cells mainly produce muscle, while mesenchymal cells can form muscle, bone, fat and cartilage within limits set partly by their tissue of origin. The authors chose embryonic stem cells because these can keep multiplying and produce cell types associated with all three embryonic germ layers. Unlike induced pluripotent cells, embryonic cells do not require genetic reprogramming, according to the paper.
EVIDENCE
The muscle protocol followed the order seen in embryonic development. Nature Communications reports that the team first pushed the starting cells towards presomitic mesoderm, the tissue that later contributes to skeletal muscle. For the opening two days, the culture received CHIR, basic fibroblast growth factor, SB431542 and DMH1. These inputs activate WNT signalling while inhibiting TGF-beta and BMP pathways. By day two, most cells expressed TBX6, which the researchers used as a presomitic mesoderm marker. Markers associated with pluripotency, including NANOG, OCT4 and SOX2, had fallen before that point. BRACHYURY and TBX6 expression rose instead.
A molecular rhythm provided another check on the induced tissue’s identity. The researchers monitored HES7, a gene involved in the developmental clock that governs the timing of body-segment formation. Nature Communications reports clear HES7 oscillations around day two. The bovine cycle lasted about four hours. Medium changes reset the phase of that rhythm, according to the paper. Three independent experiments produced similar oscillation patterns. This evidence does not show food performance. It does indicate that the cells passed through a developmental state resembling bovine presomitic mesoderm rather than merely expressing one selected muscle marker.
The team then changed the culture conditions to favour muscle formation. Nature Communications says cells received hepatocyte growth factor, insulin-like growth factor and basic fibroblast growth factor for four days. The researchers removed the last of those inputs and continued with the other two for a further nine days. MYF5 and MYOD1, markers associated with myoblasts, rose near day six. MYOG, MYF6, MYH7 and MYH4 appeared around day 10. Fibres positive for MYOG and sarcomeric alpha-actinin emerged at about the same time. Their presence reached its highest observed level around day 14.
By day 15, the fibres contained repeating contractile structures. The Nature Communications team measured an average sarcomere length of 2.1 micrometres across 88 fibres from four independent experiments. The paper says this measurement resembles values previously reported for skeletal muscle inside cattle. Seven independent experiments showed similar staining for myocyte and fibre markers. The authors describe the overall induction period as slightly shorter than a published human-muscle protocol. They suggest that the difference may reflect faster bovine embryonic development. That explanation remains an interpretation from the paper, rather than a direct production advantage.
NEURAL ACTIVITY
The cultures also showed fluctuating calcium levels, which can indicate cell activity. Nature Communications reports that the frequency and size of those changes differed between cells. The authors say density and maturity may explain some of that variation, although the supplied evidence does not settle the cause. Curare, an acetylcholine-receptor blocker, reduced the calcium pulses. The researchers used a concentration of 10 micromolar in the reported comparison. Since curare interferes with signalling at neuromuscular junctions, the response suggested communication between the induced neurons and muscle. The experiment points to functional interaction, but not to fully developed nerves or mature junctions.
Staining supplied structural evidence for that interpretation. The paper reports TUJ1-positive neurons mixed among fibres carrying sarcomeric alpha-actinin by day 15. SOX2 expression increased between roughly days 10 and 13. More mature neural markers, including SMI-32 and NEFM, appeared in patterns aligned with muscle markers such as MHC and DESMIN. Alpha-bungarotoxin staining produced small rounded signals where nerves and fibres sat close together. The authors describe these sites as early junction-like structures because their shape remained immature. They propose that stimulation from the co-induced neurons might eventually reduce reliance on external electrical or mechanical conditioning. The present study does not test that manufacturing claim.
VASCULAR ROLE
Blood vessels present a separate problem for thick cultivated tissue. Cells need oxygen and nutrients, which cannot diffuse indefinitely through a dense mass. The Nature Communications paper says engineered pores can offer an alternative route, while vessel networks could support further growth. In this study, endothelial cells arose through the same broad presomitic mesoderm route as the muscle lineage. They produced relatively even networks inside the experimental tissues, according to the authors. The presence of endothelial patterns does not mean blood flowed through them. Nor does the supplied source establish that the networks kept a large construct alive. It shows that vascular cell organisation can emerge alongside bovine muscle under the reported conditions.
PROCESS
The proposed production benefit lies in removing several assembly stages. The paper lists bioprinting with purified cells and bioink, seeding cells onto edible porous supports, and stacking muscle-bearing hydrogel pieces among existing approaches. Each route gives engineers control over placement, but also adds operations and materials. The EMBL Barcelona method instead changes signalling factors over time so that one initial population follows several developmental paths. Cells then recognise neighbours and form local structures without manual placement. Cultivated X and vegconomist describe this as avoiding separate preparation and later assembly. That account agrees with the research paper, although no source provides a manufacturing comparison of labour, yield or equipment.
The absence of serum removes one common animal-derived culture input, but it does not make the full system animal-free. Both trade publications identify Matrigel as part of the current procedure. That material comes from mice and would need replacement before the method could form the basis of an animal-free food process. The sources do not report tests of any edible substitute. They also do not provide data on whether changing the matrix would preserve differentiation, neural activity or endothelial organisation. Serum-free therefore describes the medium protocol. It should not be read as a claim that every material in the experiment meets food-production requirements.
COMPARISON
A porcine study offers the closest comparison in the supplied sources. Nature Communications published that work on 2 March 2026 as article 3347 in volume 17. Yixuan Yao, Gaoxiang Zhu, Minglei Zhi and colleagues used stable porcine pregastrulation epiblast stem cells. They directed those cells into muscle, fat and vascular endothelial lineages. The progenitors then recognised one another in a scaffold-free, three-dimensional suspension system and formed spheroids. Unlike the bovine work, the porcine process deliberately included adipose tissue, a major component of meat’s flavour, nutrition and texture. The bovine study instead added neural cells and examined their interaction with muscle.
The porcine team also reported a system without serum or animal components during differentiation. According to its Nature Communications paper, the resulting multitissue material had textural properties similar to conventional pork and allowed changes to its nutrient profile. The researchers reported greater springiness and a comparatively high proportion of polyunsaturated fatty acids. Those are claims from the study, not evidence that consumers would judge the material equivalent to pork. The supplied text gives no tasting results. It also gives no direct cost comparison with the bovine process. The two papers therefore share a pluripotent-cell strategy but test different outputs and different measures of performance.
The pig work gives more detail on fat-cell development. Its authors activated WNT signalling with CHIR99021 and blocked TGF-beta signalling with SB431542. Nature Communications says the strongest fall in NANOG, OCT4 and SOX2, alongside increased BRACHYURY, supported a three-day mesoderm induction period. From day three to day 12, markers including PDGFRA, ZNF423 and VIM rose while the adipogenesis inhibitor DLK1 declined. The team then allowed another six days for progenitor maintenance and expansion. A further six-day phase raised PPARG and CEBPA, while later genes including FABP4 and LPL increased as cells accumulated lipid droplets.
The porcine comparison also shows why pluripotent cells attract attention. The researchers compared their derived fat cells with porcine fibroadipogenic progenitors isolated from muscle. By passage 13, those tissue-derived progenitors formed fewer lipid droplets and expressed lower levels of genes linked to fat development, according to the paper. The epiblast-derived cells retained a normal karyotype during differentiation. RNA sequencing found increased expression of PPARG, CEBPA, ADD1, PLIN3 and AGPAT2 against undifferentiated controls, with the paper reporting p values below 0.01. These data address lineage stability. They do not resolve commercial expansion costs or regulatory acceptance.
LIMITS
Physical scale remains the clearest limit for the bovine result. Miki Ebisuya, formerly a group leader at EMBL Barcelona and now a Humboldt Professor at PoL-TU Dresden, told Cultivated X that steak-like production would require much larger constructs and more mature vessels capable of sustaining growth. The study used only one cell line, according to Cultivated X and vegconomist. That leaves unanswered whether other bovine embryonic lines would behave in the same way. The team has also filed a patent covering the approach, the two publications report. Neither source states its scope, ownership terms or possible effect on access.
Economics present an equally plain obstacle. Cultivated X and vegconomist attribute the current aggregate cost chiefly to growth factors and other reagents. A unit priced at up to €0.99 while measuring less than one millimetre offers no credible basis for a food-cost claim. The sources provide no estimate for a larger piece, no yield per vessel and no projection for cheaper media. They also do not report long-term storage, contamination rates, repeated batch performance or nutritional composition for the bovine tissue. Marina Sanaki-Matsumiya, now an assistant professor at the University of Tsukuba, says the purpose was to let several lineages develop and arrange themselves together. The evidence supports that narrower claim.
CONCLUSION
The bovine study replaces some manual tissue assembly with controlled development from one stem-cell source. Its value now rests in proving that the biology can survive larger scale, cheaper inputs and food-compatible materials, tests that the supplied research has not yet passed.