Growing an animal muscle protein in a salad leaf is no longer a figure of speech. A team at Imperial College London has just obtained stable, heritable accumulation of porcine myoglobin in the chloroplasts of tobacco and lettuce — a first in higher plants, where only transient expression had been reported until now. The result is clear-cut on protein quantity; it is far less so on what makes myoglobin interesting in the first place, namely its heme.

Source: frontiersin.org

In plain terms

Myoglobin is the protein that gives red meat its colour and a good part of its taste. Until now, to make it without an animal, it was produced by microbial fermentation in tanks; in plants, only a fleeting production had been achieved, one that was not passed on to the offspring. Here, the researchers inserted the gene directly into the chloroplast, the cell's little green factory: it is the chloroplast that captures light energy, and the chloroplast again that assembles the protein.

It works: the leaves accumulate a measurable quantity, the trait is transmitted to daughter plants, and photosynthesis in the transformed plants is not measurably degraded. But there is a snag. Myoglobin is coloured and "tasty" only if it encloses at its centre an iron pigment called heme. Yet in the protein purified from tobacco leaves, only 35% of the molecules had captured it — the others are empty shells, correctly folded but colourless. The authors think the pigment may be in short supply; their own figures nonetheless show that the plant makes more of it than would be needed, which invites a look at how it is put in place as well. This is therefore not a product ready to sell, but a demonstration of feasibility accompanied by a well-delineated bottleneck.

Discovery

Technical explanation

1. Why target the chloroplast rather than the nucleus — The chloroplast descends from an endosymbiotic cyanobacterium and has retained a bacterial-type genome, the plastome. As a reminder, this compartment combines three properties that make it a good production workshop: it is present in a very high copy number per cell, it transcribes its genes with bacterial σ-factor machinery, and it lacks the post-transcriptional silencing mechanisms that shut down nuclear transgenes. The authors do not measure these properties — they are established in the literature — but they do measure their consequence: across the 37 nuclear transformants they analyse, myoglobin accumulation is systematically lower, by a factor of at least 3, than in the transplastomic plants. The comparison is made against two strong promoters of the plant repertoire, the 35S of cauliflower mosaic virus and the ubiquitin promoter of Arabidopsis thaliana. It is methodologically clean because it pits the same myoglobin ortholog against itself in the same organism — which, as we shall see, is not the case for the comparison with the alga.

2. The expression cassette, and what each part does in it — The details of the construct are instructive, because they show that the gain does not come from the compartment alone. In tobacco, the coding sequence — synthesised and optimised for chloroplast codon usage — is placed under the promoter of the ribosomal RNA operon (Prrn), with a synthetic 5' untranslated region and a terminator borrowed from the small ribosomal subunit gene. In lettuce, the team substitutes the endogenous promoter and 5'UTR of the psbA gene, one of the most strongly expressed in the plastome. In both cases, the insertion targets the same intergenic region, between trnfM and trnG. This choice is anything but incidental: the transgene is flanked by sequences homologous to the plastome, and it is homologous recombination — active in the chloroplast, unlike in the plant nucleus — that installs it at this defined site rather than at random.

3. What verifying homoplasmy proves — A plant cell contains thousands of copies of the plastome. After bombardment, only a few carry the transgene: the plant is heteroplasmic, and the trait would be unstable. The authors therefore run successive cycles of regeneration under spectinomycin — 50 mg/L for lettuce — until every copy has been converted, then verify this by Southern blot and by genome sequencing. It is this step, and not the mere detection of the protein, that licenses the word "stable": it demonstrates that no wild-type copies remain that could regain the upper hand over the generations.

4. Myoglobin is not a protein, it is a protein plus a pigment — As a reminder, myoglobin is a single-chain globin of about 153 amino acids folded into eight α-helices, which harbours in a hydrophobic pocket a heme b (iron–protoporphyrin IX). The iron is coordinated there by a proximal histidine, while a distal histidine stabilises the bound dioxygen:
   $\text{Mb} + \text{O}_2 \rightleftharpoons \text{MbO}_2$Mb+O2​⇌MbO2​
   The red colour and a decisive share of the meaty aromas generated on cooking are due to the heme iron — with Maillard reactions and lipid oxidation also contributing. A myoglobin without heme — apo-myoglobin — is colourless and sensorially mute. Heme occupancy is therefore not a detail of yield, but the variable that decides whether the product is useful at all:
   $\theta = \frac{[\text{holo-Mb}]}{[\text{holo-Mb}] + [\text{apo-Mb}]} \approx 0{,}35$θ=[holo-Mb]+[apo-Mb][holo-Mb]​≈0,35

5. Three different assays, three different questions — The measurements must be carefully distinguished. The quantity of myoglobin accumulated is estimated by immunoblot: the polypeptide is revealed with an antibody and the signal compared with a standard curve — a technique that counts apo and holo forms indifferently. The authors moreover attach an explicit caveat to these values: since quantification was performed on a limited number of biological samples, the figures should be regarded as rough estimates, and they call for validation by an independent method, antibody detection being liable to bias from epitope accessibility. The fraction actually loaded with heme is measured separately, by a pyridine hemochromogen assay on the purified protein: it gives the ≈ 35%, versus 80% for the same myoglobin produced in E. coli — but only 20% in S. cerevisiae, which places the plant between the two microbial platforms rather than below both. Finally, the total heme of the tissue is assayed by a third method, using apo-horseradish peroxidase, and shows a doubling in the producing tobacco line. The authors specify that the purified protein is correctly folded; the failure therefore concerns the loading of the pigment, while noting that a heme deficit could in turn limit total accumulation, since cofactor availability conditions stability and folding.

6. The arithmetic that shifts the suspicion — This is the most interesting passage of the discussion, and it complicates the simple reading that "there is not enough heme". The authors push the calculation through. With 94 mg/kg fresh weight of myoglobin loaded at 35%, the heme already carried by the protein represents about 1.93 μmol/kg; 100% loading would require 5.5 μmol/kg. Yet the increase in total heme actually measured reaches 6 μmol/kg fresh weight — that is, more than complete saturation would call for. (Editor's reading from this point on.) In aggregate, the supply of pigment does not appear manifestly insufficient, which directs the suspicion towards the step of delivering and inserting heme into apo-myoglobin rather than towards the raw capacity of the biosynthetic pathway. The authors, cautiously, keep all three candidate causes on the table: a demand for exogenous hemoprotein that exceeds the capacity of the endogenous pathway, competition with chlorophyll biosynthesis for shared tetrapyrrole precursors, and inefficient insertion into the apoprotein. They also propose remedies: overexpressing ferrochelatase 1 (FC1), the terminal enzyme of heme b synthesis, or supplementing with precursors such as aminolevulinic acid. As a reminder, the pathway concerned takes place in the plastid: glutamate is converted there into δ-aminolevulinic acid via the so-called C5 pathway, then the chain leads to protoporphyrin IX, the branch point where Mg-chelatase commits towards chlorophyll and ferrochelatase towards heme.

7. Beware of denominators — and of platforms — Two metrics are in circulation, derived from the same quantification expressed on two bases. As a fraction of total soluble protein: 2.7% for tobacco, 1.5% for lettuce. Related to dry matter: 800 ± 110 mg/kg for tobacco, 810 ± 60 mg/kg for lettuce — intervals that overlap widely, hence two hosts that are indistinguishable on this basis, despite the apparent gap in protein fraction. A second precaution comes from the authors themselves: the ortholog expressed is not the same in the plant (porcine) and in the alga (bovine), and intrinsic differences between orthologs may affect accumulation, stability or heme incorporation. They conclude that the study demonstrates feasibility in each host but does not license a direct quantitative comparison of performance between platforms. There remains the perspective offered by muscle: beef contains 8.10 mg/g dry matter for psoas major (fillet) and 11.16 mg/g for longissimus dorsi (rib steak), according to Rickansrud and Henrickson (1967). The authors summarise this as "about 10-fold"; the muscle-by-muscle calculation gives a factor of 10 (psoas) to 14 (longissimus). Note that the comparison, as framed in the article, crosses species: a porcine myoglobin produced in a leaf, set against a bovine muscle content.

Why It Worked

The success rests on a choice of compartment coupled with a carefully designed construct. By placing the gene in the plastome, the team combines a high copy number, bacterial transcription and the absence of silencing; by optimising codon usage and borrowing psbA's expression signals in lettuce, it exploits the most active machinery of the compartment. The gain can be read in the figures: a factor of at least 3 across 37 nuclear transformants. A second result, measured rather than assumed: across the three parameters monitored — effective quantum yield of photosystem II, electron transport rate and non-photochemical quenching — no significant difference appears between transformed plants and controls. The fear that a foreign, heme-hungry hemoprotein would disturb the photosynthetic apparatus does not materialise.

The gap between the announcement and the proof must nonetheless be stated. What the study demonstrates: higher plants can stably and heritably accumulate an animal myoglobin, at a level above that reached by nuclear expression — where the state of the art was limited to transient expression in Nicotiana benthamiana. What it does not demonstrate: that this myoglobin confers on a food the expected colour, taste or nutritional value, nor that a plant platform outperforms an algal platform, a comparison the authors explicitly rule out for themselves. At 35% occupancy, nearly two molecules in three are empty shells, and the authors write that commercial viability will depend not only on total protein accumulation but also on holo-myoglobin yield — with progress required on both fronts at once.

The heaviest caveat is statistical in nature, and it comes from the authors themselves: a single independent transplastomic line was obtained, for tobacco as for lettuce, and they write that validation on additional lines would strengthen the generality of the conclusions. All the figures quoted above therefore rest on one clone per host. Asked by the Science Media Centre, Derek Stewart (James Hutton Institute), who declares no interest, points to this same fragility and adds that tobacco naturally produces bioactive alkaloids that would pose a real extraction and food-safety problem — which gives weight to the lettuce arm, an edible host. Rodrigo Ledesma-Amaro (Imperial College London) — who declares that two authors belong to the community of Imperial's Bezos Centre for Sustainable Protein, that he was not involved in the study and that the project was not supported by the Centre — considers that the potential will depend on improving heme incorporation, and that food functionality, safety, field performance, processing requirements and economic and environmental benefits at scale remain to be confirmed.

The framework of interests, as the article declares it, deserves to be reported in full. Two of the co-authors, Kyoko Morimoto and Mistianne Feeney, are employed by Kyomei Ltd, a private company in the sector; the same company supplied the sK23 lettuce line and partly funded the doctoral work of the first author, Alexia Groff, alongside the EPSRC Centre for Doctoral Training in BioDesign Engineering. Finally, senior author Peter J. Nixon declares that he was a member of a Frontiers editorial board at the time of submission, specifying that this bore on neither the peer review nor the final decision. None of this disqualifies the results, published open access and verifiable; it is the frame within which to read them.

Causal Chain

Endosymbiosis of a cyanobacterium → retention of a bacterial-type plastome, in a very high copy number per cell → development of chloroplast transformation by biolistics and homologous recombination → observation that this compartment escapes the transgene silencing that limits nuclear expression → industrial demand for hemoproteins for meat substitutes, met until now by microbial fermentation → first attempts in plants, limited to transient expression in Nicotiana benthamiana → insertion of the codon-optimised porcine myoglobin gene between trnfM and trnG in the plastome of tobacco and then lettuce → spectinomycin selection to homoplasmy, verified by Southern blot and sequencing → stable and heritable accumulation at 2.7% and 1.5% of total soluble protein, with no measurable degradation of photosynthesis → but heme loading capped at ≈ 35%, even though the measured increase in total heme (6 μmol/kg FW) exceeds what complete saturation would require (5.5) → the authors retain heme availability as a possible bottleneck, while their arithmetic also points to insertion of the cofactor into the apoprotein → avenues proposed by the authors: overexpression of ferrochelatase 1 or supply of aminolevulinic acid → a bottleneck to be lifted before any food prospect.

Anecdote

The industrial precedent for this story comes not from a leaf but from a root. Impossible Foods built its burger on soy leghemoglobin — the hemoprotein the plant makes in its root nodules to buffer free oxygen and thereby protect nitrogenase — and had, unable to extract enough of it from soy itself, to have it produced by fermentation in a yeast, Pichia pastoris. The molecule was plant-derived; its production, industrial. The study published this week attempts the reverse: keep the animal protein, but make the production plant-based. And it runs up against the same cofactor as the one that had made leghemoglobin interesting — heme, decidedly the scarce part in all this chemistry.

Legacy and Current Data

One must resist here the temptation of the market figure. What the data allow one to assert is limited to laboratory orders of magnitude, which the authors themselves present as rough estimates: 2.7% of total soluble protein in tobacco, 1.5% in lettuce, i.e. 800 ± 110 and 810 ± 60 mg/kg dry matter, to be compared with the 8.10 and 11.16 mg/g of bovine muscle — a factor of 10 to 14. No food trial, no field trial, no cost analysis accompanies these results, and the whole rests on a single line per host.

What remains to be established is therefore explicit: the heme occupancy attainable after optimisation, reproducibility across independent lines, quantification by a method independent of immunodetection, the behaviour of transplastomic lettuce under real growing conditions, the cost of extraction and purification, and regulatory status — a transplastomic lettuce intended for food would in all likelihood fall under GMO authorisation in jurisdictions that regulate the product on the basis of the process, but none of the sources consulted addresses this point. The only structural regulatory advantage lies in the maternal inheritance of the plastome in most angiosperms, which limits dissemination of the transgene by pollen; this is a reminder of general biology, not a result of this study.

The researcher's view — open questions

The authors already propose two avenues — overexpressing ferrochelatase 1, or supplementing with aminolevulinic acid. The questions below are extensions formulated by the editor, and do not figure among the study's results or proposals.

• Test insertion first, not supply: the authors' arithmetic suggests that the pigment produced could be enough to saturate the myoglobin accumulated. Before increasing the flux further, it would be decisive to supply exogenous heme to detached leaves or to transplastomic cultures: if occupancy does not rise, the bottleneck really is insertion into the apoprotein, and overexpressing FC1 will change nothing.
• Separate the two levers: a crossed trial — FC1 alone, then FC1 combined with a heme-insertion chaperone — would distinguish a production limitation from a delivery limitation.
• Decouple light and heme: measure occupancy in transformed non-chlorophyllous tissues, or in etiolated plants, where Mg-chelatase no longer drains protoporphyrin IX. This is the most direct test of the competition hypothesis, and it requires no new construct.
• Compare the platforms properly: use a single myoglobin ortholog in all three hosts, which the authors themselves identify as the work needed to settle relative performance.
• Robustness: regenerate several independent lines per host, a caveat raised both by the authors and by Derek Stewart, and without which the yield remains the property of one clone.

Sources

References verified during the fact-checking audit, on 6 August 2026, against the full text of the article.

• Groff A., Lu Y., Feeney M., Whitelegge J., Shao S., Morimoto K., Nixon P. J., "Sustainable production of myoglobin meat protein in plant chloroplasts", Frontiers in Plant Science, vol. 17, Plant Biotechnology section — accepted 22 June 2026, published 6 August 2026. Peer-reviewed article, open access. DOI: 10.3389/fpls.2026.1876707
• Science Media Centre, "Expert reaction to study looking at growing myoglobin animal muscle protein fibres in chloroplasts of lettuce and tobacco plants", 6 August 2026 — comments from Rodrigo Ledesma-Amaro (Imperial College London, declared interest) and Derek Stewart (James Hutton Institute, no declared interest).
• Rickansrud D. A. & Henrickson R. L., "Total pigments and myoglobin concentration in four bovine muscles", Journal of Food Science, 1967, 32(1), 57–61 — reference contents for bovine muscle, cited by the article under discussion. DOI: 10.1111/j.1365-2621.1967.tb01957.x

Background references

These references support the reminders of general mechanism and do not come from the study under discussion.

• Bock R., "Engineering plastid genomes: methods, tools, and applications in basic research and biotechnology", Annual Review of Plant Biology, 2015, 66, 211–241 — plastome copy number, homologous recombination, absence of silencing, maternal inheritance. DOI: 10.1146/annurev-arplant-050213-040212
• Tanaka R. & Tanaka A., "Tetrapyrrole biosynthesis in higher plants", Annual Review of Plant Biology, 2007, 58, 321–346 — C5 pathway, protoporphyrin IX branch point between Mg-chelatase and ferrochelatase. DOI: 10.1146/annurev.arplant.57.032905.105448
• Ordway G. A. & Garry D. J., "Myoglobin: an essential hemoprotein in striated muscle", Journal of Experimental Biology, 2004, 207(20), 3441–3446 — eight-helix architecture, proximal and distal histidines, oxygen binding. DOI: 10.1242/jeb.01172