Caviar is considered a luxury food, fetching up to hundreds of dollars per ounce for the most exclusive grades. But if you think caviar is expensive, consider that human eggs cost around $3,000 each at the low end, and much higher if you want to use a donor with good genetics.

Recently, Mitinori Saitou’s lab at Kyoto University published a paper (Nosaka et al., Developmental Cell 2025)^[1] describing a method to mass-produce mouse eggs from stem cells, at up to 400,000 eggs per batch. Although the eggs are low-quality and cannot reach a fully mature stage, this is still a remarkable advance. Surprisingly, this method works without ovarian supporting cells, and this has important implications for in vitro meiosis and in vitro oogenesis.

How does this work?

The basic approach is as follows:

1. Starting from mouse pluripotent stem cells, differentiate them to primordial germ cell-like cells (PGCLCs) through treatment with signaling factors.
2. Take the PGCLCs and culture them on m220-5 feeder cells, which promote cell proliferation and epigenetic erasure. The PGCLCs also start forming intercellular bridges, which are important for meiosis to happen properly.
3. Add a short pulse of retinoic acid, followed by BMP (a signaling protein), to activate meiosis. The Saitou lab previously published a similar meiosis induction protocol for mouse cells. In the current paper, the main advance was finding that a short retinoic acid pulse was better than more prolonged stimulation.
4. After 13 days of culture, a lot of cells died off, but about 70% of the surviving cells completed prophase I of meiosis and arrested in the diplotene stage. Now the cells are dissociated (breaking up any intercellular bridges) and cultured as single cells for an additional 21 days, during which time they activate oocyte-specific gene expression and grow to a large size.

The researchers performed extensive optimization experiments to identify the best culture conditions for oocyte growth, identifying antioxidants and activators of important signaling pathways (PI3K/mTOR, cAMP, WNT, SCF, FGF)^[2] as key media components. The gene expression (at the RNA level) in the oocytes was quite similar to in vivo, and the oocytes grew to a size of 60 - 65 µm (compared to about 80 µm for a fully grown mouse egg). Upon stimulation, the oocytes resumed meiosis, completing meiosis I. However, they were not able to progress to meiosis II.

How good are these eggs?

The fact that they were able to generate eggs, even at an immature stage, without ovarian supporting cells is highly impressive. However, these eggs are not very good, probably due to the lack of supporting cells.

The biggest issue is that the eggs can’t do meiosis II, and thus can’t be fertilized. This is probably a result of the fact that they can’t grow to a fully mature size. The researchers also found increased mitochondrial activity in these oocytes compared to in vivo, probably as metabolic compensation for lack of supporting cells.

Additionally, there are epigenetic issues with the eggs. Although the DNA methylation erasure from stem cells to meiotic cells was pretty good (erasing all but 9.2% of genome-wide methylation compared to 2.8% in vivo), the re-establishment of methylation in female-specific imprint patterns during oocyte growth was defective. However, the X-chromosome activation status of the eggs was pretty normal.^[3] The researchers also examined histone modifications, and found them to be reasonably close to normal, implying the main defect was with DNA methylation.

What’s special about m220-5 feeder cells?

Although the researchers were successful in generating eggs without ovarian supporting cells, they still needed feeder cells to support their culture system. The cell line they used was m220-5, a variant of the m220 cell line. m220 cells were engineered all the way back in 1993, through transfecting Sl/Sl4 mouse hematopoetic stromal cells^[4] with an expression vector for a membrane-bound isoform of the growth factor SCF. In order to use these as feeder cells, they need to be mitotically inactivated, or else they will overgrow the culture. The straightforward way to do this is treating them with mitomycin C, a DNA damaging agent, but unfortunately m220 cells will completely die off when treated with mitomycin, instead of just halting proliferation. So, the Saitou lab screened 242 subclones of m220 (which had accumulated random mutations over prolonged culture) to find ones that survived mitomycin treatment. The result was m220-5.

Besides the current study, the Saitou lab also used m220-5 cells in their paper last year about epigenetic erasure in human PGCLCs. Feeder cells which overexpress membrane-bound SCF are a valuable resource.

The question now is: what besides SCF do the feeder cells provide? There are likely other important signaling pathways involved in promoting meiosis and egg growth. Anything which is expressed both by these feeder cells and by the ovarian supporting cells would be a likely candidate.

What does this imply for in vitro meiosis and oogenesis?

The first key takeaway is that meiosis can happen properly in 2D culture,^[5] even without ovarian supporting cells. Although some kind of feeder cells (e.g. m220-5) are still required, this is a lot easier than using ovarian supporting cells. This is good for iterated meiosis, because it’s a lot more scalable to do meiosis in 2D culture rather than an ovarian organoid 3D culture system.

The second key takeaway is that the eggs didn’t grow to full maturity without ovarian supporting cells. So, I think ovarian supporting cells will still be required to make high-quality, fully grown eggs. But perhaps the ovarian supporting cells could be combined with the egg precursors after the egg precursors arrest in diplonema.

Overall, this is a groundbreaking study that overturns previous ideas in the field: ovarian supporting cells are not actually required for meiosis nor for the early stages of oocyte growth. I’m quite surprised it didn’t end up in Cell instead of Developmental Cell – I guess even Mitinori Saitou might get harsh peer reviewers sometimes.

So, if things go well at Ovelle, maybe in a few years we’ll be eating a lot more caviar!

1. ^{^}

   They also presented a slightly earlier version of this at a conference in May.

2. ^{^}

   PI3K/mTOR and cAMP pathways are known to be important in promoting oocyte growth and preventing premature resumption of meiosis, respectively. Antioxidants help mitigate the mitochondrial stress caused by increased metabolic load due to lack of supporting cells.

3. ^{^}

   One interesting finding was that a lot of the PGCLCs lost an X chromosome prior to meiosis. It seems that these X0 cells died off during or shortly after meiosis though.

4. ^{^}

   These cells don’t express the normal form of SCF due to a mutation, so the membrane bound SCF is the only kind they express. The culture media is also supplemented with non-membrane-bound SCF.

5. ^{^}

   The hard part of meiosis is getting synapsis and crossing over in meiosis I. The current study didn't get meiosis II to work, but I'm pretty confident of some ways to force meiosis II to happen given completion of meiosis I.