A Future Where Milk is "Harvested" from Plants, Not "Squeezed" from Cows

When thinking of where white milk comes from, many people imagine farms and dairy cows. However, in the future, "fields" might be added to that answer.

A research team centered at the Hebrew University has successfully conducted an experiment to make plant seeds produce β-casein, one of the key proteins that make up milk. What caught the researchers' attention was not just the detection of the target protein. The β-casein did not enter the initially planned storage location within the cells but instead accumulated in an unexpected structure closely associated with oil droplets in the seeds.

For the researchers, this was a result that did not go "as planned." However, this phenomenon, which appeared to be a failure, suggested a potential shortcut for storing complex animal-derived proteins in plants.

This achievement is not an improved version of almond milk or oat milk available in stores. Rather than mimicking milk with plant flavors and components, this research aims to have plant cells produce the very proteins found in milk. This difference is significant when considering the future of alternative milk and plant-based cheese.

Why β-Casein is Important

Milk contains water, fat, sugar, minerals, vitamins, and several proteins. Among them, casein plays an important role not only as a source of nutrition but also in creating the texture characteristic of dairy products.

Especially in cheese, casein forms structures that encapsulate fat and moisture, resulting in the formation, elasticity, smoothness, and melt-in-the-mouth feel. The reason plant-based cheese is often described as "hard to melt," "doesn't stretch," or "greasy" is that it lacks the same casein structure as real dairy products.

Current plant-based foods combine starch, vegetable oils, thickeners, and proteins from beans and nuts to recreate a texture similar to dairy products. Although technology is advancing, it is difficult to make them behave the same as milk in terms of stretch and coagulation when heated or behavior during fermentation.

If β-casein can be stably extracted from plants, it could add functions to foods that are difficult to replicate with plant-based ingredients alone. Future applications are not limited to beverages. Cheese, yogurt, protein foods, infant nutrition, and medical and elderly care foods are all potential candidates.

However, this research does not mean that "milk has been completed from plants." What was produced is one type of protein that makes up milk, and the taste, aroma, digestibility, safety, and processing suitability as a food have not been confirmed.

The value of the research lies not in the finished product but in gaining new insights into the mechanism of producing and storing milk proteins within plants without breaking them down.

Turning Seeds into Small Biofactories

The research team used Arabidopsis, a plant widely used in plant research. Known as a model organism for plants, it is small, has a fast generation turnover, and is easy to study genetically.

The researchers introduced genetic information for producing bovine β-casein into the plant and further attached a part of a protein called oleosin, which exists on the surface of oil droplets in plants.

Oleosin plays a role in stabilizing the oil droplets in seeds. By associating the target protein with oil droplets, it was expected to be stably accumulated within the seeds, making post-harvest separation and purification easier.

Additionally, to change where the synthesized protein is transported within the cell, multiple "destinations" were tested. The endoplasmic reticulum, chloroplasts, and vacuoles, which are cellular compartments for folding, modifying, and storing proteins, were targeted.

Among them, the accumulation of β-casein was confirmed in the line designed to send it to the vacuole. However, a detailed look with an electron microscope revealed that β-casein was not present inside the vacuole. Instead, it gathered in a spherical high-density structure that appeared in the cytoplasm, surrounded by numerous small oil droplets.

Using immunogold labeling to pinpoint the location with antibodies and gold particles, both β-casein and oil droplet-associated proteins were confirmed in the aggregate.

The research team suggests that the fusion protein may have formed an aggregate of protein and oil reminiscent of natural casein micelles while altering the structure of the oil droplets.

In other words, the plant did not transport the cargo to the designated warehouse. Instead, it created a new warehouse-like structure on its own and housed the foreign protein.

Why "Unexpected" Holds Value

When making plants produce recombinant proteins, preserving them without breaking down within the cells is as important as making them.

Cells have mechanisms to degrade proteins deemed unnecessary or abnormal. Even if produced in large quantities, if they are quickly degraded, yields will not increase. Moreover, if proteins interfere with cellular functions, plant growth and seed formation could deteriorate.

In the most successful line, recombinant β-casein accounted for about 1.26% of the total soluble protein in the seeds. The research team evaluates this as a competitive level compared to past studies on casein production in plants.

Furthermore, the germination rate of this line was 91%, indicating that at least the germination ability was maintained.

The unexpected aggregates may represent a mechanism by which plant cells isolate foreign proteins and preserve them while minimizing damage to themselves. If the conditions for formation can be elucidated, it might be applicable not only to β-casein but also to other high-value-added components such as food enzymes, vaccine materials, antibodies, and nutritional proteins.

"Plant molecular farming," which uses plants as production facilities, is attractive because it can utilize existing agricultural infrastructure for sowing, growing, and harvesting, without relying solely on large fermentation tanks.

Dried seeds are easy to store and may allow for the distribution of production areas. It might establish a supply model different from precision fermentation, which requires electricity and capital investment.

Safflower, Strong in Arid Regions, as a Candidate for Practical Use

While Arabidopsis is convenient for research, it is not a crop suitable for commercial production. Therefore, the next step involves transitioning to crops with high yields and established cultivation, harvesting, and oil extraction systems.

According to the university's announcement, researchers have already progressed to the stage of introducing milk protein genes into safflower.

Safflower is a crop from which oil is extracted from seeds, relatively resistant to heat and drought, and could be a candidate for cultivation in arid regions. There is also the potential to utilize existing processing facilities for oilseed crops and to simultaneously produce oil and high-value-added proteins.

The important point here is not plant-derived proteins, but using plants as manufacturing devices to produce proteins with the same sequence as those derived from cows. The boundary between "plant-based" and "dairy-derived" becomes complex due to the manufacturing method rather than the raw materials.

If milk proteins are harvested from safflower fields in the future, dairy farming and plant agriculture may not be in simple competition.

Traditional milk might be used for drinking, regional brands, and high-quality cheese, while plant-produced proteins could be directed toward processed foods and nutritional materials, potentially leading to a division of labor based on usage.

Surprise at the "Futuristic Feel" on Social Media, but Few Direct Reactions So Far

The article introducing this research on Phys.org was just published on July 19, 2026. Social media reactions are still limited.

While the article page shows 43 shares, there were zero public comments. LinkedIn posts also had few reactions at the time of confirmation, and posts on Hacker News were on a small scale.

On the other hand, on Threads and Facebook, posts highlighting the unexpected nature of the research, such as "seeds produced bovine β-casein" and "accumulated near oil droplets instead of the targeted vacuole," were shared.

On Instagram, there was also a future-oriented presentation showing that plants have taken the "next step" in food technology.

At this stage, rather than a fierce divide in opinions, the strength of the phrase "plants produce real milk protein" is leading the initial reactions. Surprise, curiosity, and a futuristic feel akin to science fiction are at the center of these early responses.

However, the fact that there are few comments on this research itself does not mean there are few social issues.

Looking at past social media posts about animal-free milk proteins produced by fermentation, consumer reactions can be broadly divided into four categories.

Expectations for Environment and Animal Welfare

The first is the voice of welcome from the perspective of animal welfare and environmental impact.

There is an expectation to try dairy products that are close in taste and texture to the real thing without raising cows. Posts evaluating animal-free milk protein ice cream and beverages as "more dairy-like than regular plant-based products" and expressing a desire to try new products can be seen.

Plant molecular farming might also gather expectations for reducing farmland, water, and greenhouse gases.

However, just because it is made from plants does not necessarily mean it will have a lower environmental impact. It is necessary to evaluate the entire process, including fertilizers, agricultural machinery, drying, transportation, extraction, and purification.

Confusion Between "Animal-Free" and "Dairy-Free"

The second is confusion over the ambiguity of terms.

"Animal-free," "dairy-free," "plant-based," "vegan," and "contains milk protein" do not mean the same thing.

Even if manufactured without using cows, if the finished molecule is milk protein, it is close to milk in terms of components. It may be accepted by those avoiding dairy for ethical reasons but could be dangerous for those with milk protein allergies.

On past social media, confusion has been repeated with questions like "Why is it a milk allergen if no animals are used?" and "Is it for vegans or is it a dairy product?"

If this technology is commercialized, it will be essential to clearly communicate what it contains and who needs to avoid it, not just with attractive catchphrases.

Caution Towards Genetic Modification

The third is caution towards genetic modification and "artificial foods."

Even if produced by plants, as long as genes are introduced for production, consumers will demand safety evaluations, cultivation management, prevention of crossbreeding with other crops, and transparency in labeling.

Scientifically demonstrating that there are no issues as food may not be enough. It is necessary to clearly explain which genes are used, where they are grown, how they are purified, and whether plant DNA or other components remain in the final product.

In plant molecular farming, production management to prevent contamination with regular food crops will also be important. Containment mechanisms, including cultivation locations, harvesting machines, transport containers, storage facilities, and processing equipment, will be required.

Realistic Questions About Price and Availability

The fourth is questions about price and availability.

There is a significant gap between being able to produce in a laboratory and being able to purchase in large quantities with stable quality by food companies. Even if the taste is good, if the price is high, the market will be limited.

In past social media posts about animal-free dairy products, alongside expectations for taste, there were many questions like "Where can I buy it?" "How much will it cost?" and "Can it be sold at the same price as conventional products?"

While plant production might utilize agricultural equipment, there are cost factors different from fermentation tanks, such as yield fluctuations due to weather, cultivation periods, and construction of purification facilities.

Whether it truly becomes widespread depends not on whether it can be scientifically made, but on whether it can be made in the necessary quantities at the necessary price.

"Not Using Cows" and "Addressing Milk Allergies" Are Separate Issues

If this technology is commercialized, the most careful explanation will be needed for allergies.

Lactose intolerance is a condition where it is difficult to break down lactose, the sugar in milk. On the other hand, milk allergy is when the immune system reacts to proteins like casein or whey.

Even if β-casein produced within plants does not contain lactose, it is not necessarily safe for those who need to avoid β-casein itself.

Therefore, expressions like "not using cows" or "produced in plants" are insufficient. While it may be animal-free ethically, it could still correspond to milk components as an allergen.

In future product labeling, it will be necessary to clearly indicate the manufacturing origin and allergen information separately.

The same issue relates to vegan labeling. While some may consider it vegan because no animals are directly used, others may not accept it since it is milk protein based on bovine genetic information.

This is not an issue that can be decided by science alone; it is an area where ethics, certification systems, and consumer values intersect.

"Cheese from Fields" Hasn't Been Achieved Yet

Precisely because this research holds great promise, it is necessary to accurately see its current limitations.

Firstly, production volume is still low. Although the figure of 1.26% of soluble protein is higher than past research, it is only a small part of the total seed weight.

To make it price-competitive as a food ingredient, expression levels must be further increased, and losses during extraction must be reduced.

Secondly, in the line that produced the most β-casein, seed yield and soluble protein content were lower compared to the wild type.

In the comparison published in the paper, the average seed weight of the wild type was 514 milligrams, while the line with high β-casein accumulation was 246 milligrams. Even if the germination rate is good, if the number of seeds obtained from one plant decreases, production per unit area of farmland will not increase.

It is necessary to design a way to increase protein content without placing excessive burden on the plant.

Thirdly, it has not yet been proven that the aggregates confirmed this time have the same food functions as