What makes a good egg?

Desiderata for in vitro oogenesis

As we get closer to being able to grow human eggs from stem cells, it’s important to be sure that the eggs are high quality and safe enough to use for reproduction.¹

So, let’s discuss: what does an egg need to have in order to develop into a healthy baby after fertilization? How can we tell if an egg is good or not? And what does this mean for in vitro oogenesis?

Good genetics

At the most basic level, an egg must have the correct number of chromosomes. Chromosome pairing is established during meiosis I, and chromosomes are distributed into the embryo during meiosis II.² Chromosome spreads³ are a simple yet effective method to check whether chromosomes are distributed correctly in individual cells. In metaphase of meiosis I, chromosomes should be present as tetrads: groups of four chromatids connected at their centromeres and at a crossover site. If chromosome pairing and recombination happened correctly, a spread at metaphase I will show exactly 23 tetrads per cell, with no chromosomes left unpaired.

An example of chromosome tetrads (also known as bivalents) in a recent mouse in vitro oogenesis paper from the Saitou lab (Nosaka et al 2025). I wrote about this paper in July, it’s a great study! Their chromosome pairing looks wonderful; in humans there should be 23 tetrads but here there are 20 since it’s mouse cells.

This is a destructive test that can’t be used on embryos before implantation, but it’s a good way to do a quality control check to see if an in vitro oogenesis method is reliable enough to safely use.

Additionally, the chromosomes can’t have harmful mutations. Adult somatic cells often accumulate mutations throughout their genome, with this effect being worse in cells exposed to the environment (like skin)⁴ or cells which divide rapidly. Most of these mutations won’t matter, but some could be harmful if they disrupt important cellular functions. In order to check for mutations, whole genome sequencing can be performed on the starting stem cell lines. Starting with a low-mutation cell type, on average the best of five stem cell lines will have fewer mutations than natural reproduction.

Both the overall number of chromosomes, as well as any mutations, can be examined using pre-implantation genetic testing. A small number of cells are removed from the trophectoderm (the outer surface of the embryo) and sequenced. This is the same concept as what companies like Orchid Health use for embryo screening. Although this method is not perfectly reliable,⁵ combined with extensive sequencing of the starting cell lines, it can make the overall process safer than natural reproduction.

Good epigenetics

Epigenetic marks (chemical modifications to DNA or histones) are crucial for controlling gene expression throughout development. In particular, DNA methylation present in the egg can persist all the way into adult life.

So, in order to develop properly, an egg needs to have the correct epigenetics. This means completely erasing the epigenetic marks present in the starting cell, and also writing the egg-specific marks. This has been a challenge for the field: it’s likely that the low developmental potential of in vitro grown mouse eggs is largely due to epigenetic issues.

Examining epigenetics in natural eggs and embryos is challenging due to low numbers of cells, which makes it difficult to establish standards for what the epigenetics of in vitro grown eggs should look like. Some DNA and histone modification patterns have been characterized for human oocytes and embryos, but overall, more data are needed here. I am confident that recent advances in low-input epigenetic analysis will allow for improved datasets in the near future.

Good everything else

In order to develop properly, an egg needs to have grown to a large enough size, and have stored up the correct RNAs and proteins for embryonic development. It also needs to have enough mitochondria (several hundred thousand per egg). Egg growth is enabled by ovarian supporting cells, but for proper downstream development, the important thing is that the egg itself ends up looking like a natural egg, with the correct levels of RNA and protein expression.

Due to the large size of eggs, single-cell transcriptomics and proteomics is actually not too difficult with them. Fun fact: the first-ever scRNAseq paper came out of Azim Surani’s lab all the way back in 2009! This looked at mouse eggs and early embryos, and the ~100 million SOLiD sequencing reads they generated probably cost them several tens of thousands of pounds. Today, proteomic, and especially transcriptomic, quality control data sets are readily accessible for eggs and early embryos. Transcriptomic data on early embryos can also shed light on epigenetic processes like zygotic genome activation. Overall, in vitro grown eggs should be as similar to natural eggs as natural eggs from different donors are to each other.

Implications for in vitro oogenesis

Any useful method for growing eggs in vitro must ensure that the eggs are consistently good. I emphasize consistency here because it’s often the case that a differentiation protocol works for certain stem cell lines, but not for others.⁶

At this point I want to mention a recent paper from Shoukhrat Mitalipov’s lab (Gutierrez et al. 2025). In this paper, the researchers took eggs from human donors, removed the chromosomes, and inserted nuclei of skin cells. The eggs were then stimulated to divide, and randomly distributed their chromosomes such that on average the resulting embryos ended up with 23 of them. This approach was an extension of their work with mouse eggs which I wrote about previously.

Although the researchers provided an accurate assessment of the method’s limitations in the paper itself, popular news media reported it as “researchers create human eggs from skin cells”. This can hardly be true if the researchers used human eggs as starting material! Plus, the method has no way to control chromosome distribution.

Figure 4 of the paper shows that their chromosome distribution is random.

And furthermore, the epigenetics will correspond to a skin cell, not an egg cell.⁷ I am actually a big fan of the Mitalipov lab’s work, since they have uncovered some very interesting biology, but I am disappointed in the media response.

More broadly, any in vitro oogenesis company needs to have rock-solid quality control. A proof-of-concept human egg would be a great achievement, but to actually benefit patients, the method must be both scalable and reliable.

Thoughts on speed

The one big advantage of the Mitalipov lab’s method is its speed: taking an adult cell and putting it into a pre-existing egg means that a patient only needs to wait a few days for the egg to develop into an embryo, rather than weeks to months for growing a new batch of eggs.

Quality still comes first: no patient wants to use a procedure that has a substantial risk of making a baby with developmental issues. But assuming quality control is solved, the method that will win in the market will be the method that’s fastest (and relatedly, cheapest). And there is not necessarily a tradeoff between speed and quality: in fact, faster methods will actually be easier to optimize for quality due to shorter experiment cycle times. Methods that try to perfectly re-create natural ovarian development will be stuck waiting many months (and potentially several years) for their eggs to grow, and in the end, only 30%-50% of natural eggs develop normally after fertilization, which places an upper limit on how good their eggs can be.

At Ovelle, we focus on using regulatory factors to directly drive developmental processes such as epigenetic erasure, meiosis, and ovarian follicle formation and growth. This lets us fast-forward through development, achieving in weeks (and in some cases, days) what takes others months. At the same time, we continue to optimize the quality of our cells in gene expression, chromosome pairing, and epigenetics. We believe that this is the best way to make in vitro oogenesis work for everyone who needs it.

1. No reproduction is ever completely safe, but our goal at Ovelle is to have our eggs be safer to use than natural eggs. This is actually not a very high bar, given that natural eggs are >30% aneuploid, with the rate increasing at older ages.

2. Meiosis II is not actually completed until after fertilization. The chromosomes that don’t make it into the embryo are extruded into the polar body.

3. There are two common types of chromosome spreads: sucrose spreads in prophase of meiosis I, and metaphase spreads in meiosis I and II. The protocols are optimized for use at different stages of the cell cycle (prophase vs metaphase).

4. The mutation rate is many times worse in skin exposed to the sun, but even in areas which don’t see sun exposure, there are on average hundreds of mutations per cell.

5. The trophectoderm cells may not be representative of the inner cell mass, and whole genome sequencing on a small number of cells may detect many false-positive mutations due to amplification errors.

6. Notably, the Saitou lab has seen this for germ cell induction protocols: https://academic.oup.com/biolreprod/article-abstract/96/6/1154/3769376 https://www.cell.com/cell-reports/fulltext/S2211-1247(21)01382-6

7. Although SCNT cloning proves that this epigenetic barrier is not insurmountable, the efficiency of cloning is quite low and many cloned embryos develop abnormally.