Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.


Developing cells remember where they came from, thanks to keratin filaments

The processes by which a single cell — the fertilized egg — gives rise to all the different cell types that make up an adult organism remain some of life’s great mysteries. We know that it takes time for cells in an embryo to settle on a fate, because a single embryo that splits during early development can give rise to twins, triplets and more. But how are cell-fate decisions made, and how do cells coordinate their choices with their peers? Researchers have suggested numerous mechanisms that influence the paths taken by cells in early mammalian embryos. Writing in Nature, Lim et al.1 describe a surprising role for a protein polymer, keratin, in the first of these decision-making processes.

Two of the main challenges of early development are to increase the number of cells through repeated rounds of cell division, and to ensure that these cells assume distinct forms and functions at the right time and place to generate functional tissues and organs. The two processes can be coupled through ‘asymmetric’ cell divisions. These are divisions that give rise to two sibling cells with distinct identities, either as a result of the asymmetric segregation of material, or in response to local differences in the extracellular environment that the cells encounter after division.

It is during the 8- to 16-cell transition that cells in early mammalian embryos first become asymmetrically organized — with subsets of proteins becoming concentrated at opposite cell poles, a feature called apical–basal polarity. Cell identity remains plastic at this stage, but daughter cells that end up at the periphery (termed the trophectoderm) of the 16-cell embryo give rise to the placenta, whereas daughter cells that end up inside the embryo contribute to the fetus.

The observation of apical–basal polarity at the 8- to 16-cell transition led to the proposal that the future identity of these cells is determined by the asymmetric inheritance of the outward-facing apical domain2, which is rich in actin (a component of the cell’s ‘skeleton’) and polarity proteins3. But, using fast live imaging, the group that performed the current study showed previously3 that the apical domain is transiently lost during mitotic cell division, before re-forming in the daughter cells on the embryo’s periphery. This puzzling observation suggested the existence of other factors that act as a memory of polarity during divisions. Following up hints from the old literature on mouse embryos4, Lim and colleagues have now homed in on keratin, a type of intermediate filament protein.

Imaging keratin, the team observed a few short keratin polymers in a subset of cells in the 8-cell embryo. As these filaments grew during the part of the cell cycle between divisions, called interphase, they became preferentially associated with the apical, actin-rich cortex — a layer of proteins just inside the cell membrane. When the apical domain became disassembled during mitosis, these keratin filaments remained in place (Fig. 1a).

Figure 1

Figure 1 | Keratin in early embryos. a, In 8-cell mouse embryos (not all cells shown), keratin filaments are expressed stochastically in a subset of cells, before associating with a ‘cortex’ on one side of the cell (called the apical side). The cortex is rich in the protein actin. Lim and colleagues1 show that, as these cells divide, the cortex disassembles but keratin filaments remain apically localized. The identity of daughter cells is determined by the position of keratin relative to the axis along which division occurs. Daughter cells that lack keratin end up inside the embryo and go on to form the inner cell mass. Daughter cells that inherit the mother’s apical region also inherit keratin, which helps to re-establish the cortex at the apical pole. These cells contribute to the trophectoderm, from which the placenta arises. b, In most other settings studied, asymmetric division involves coordination between multiple polarized cues (such as polarized proteins) and a specific division axis. c, In other species, such as flies, asymmetric inheritance of protein aggregates provides a memory of the mother cell’s state.

Although this might seem unexpected, other intermediate filaments have been shown to remain associated with the cortex during mitosis5,6. Lim et al. found that apical retention of these polymers depends on their slow diffusion, which is limited by their large molecular weight and the cytoplasmic actin meshwork in which they are embedded. As a result, keratin filaments are inherited by daughter cells that retain an outward face. So, once positioned at one end of the cell, these relatively inert stable polymers act as a physical memory of polarity.

The authors went on to show that, as cells of the new 16-cell embryo exit mitosis, inherited keratin filaments accelerate the repolarization of the apical cell cortex, which biases the cell towards becoming trophectoderm (through signalling pathways that involve Yap and Hippo proteins7). In turn, this bias is associated with high levels of keratin expression. So, over a period of hours, positive feedback in the system reinforces the accumulation of keratin in peripheral cells, and inhibits its expression in cells at the embryo’s centre. By the 32-cell stage, when cell fate is more firmly established, the embryo itself is clearly polarized, with an outer, keratin-rich supporting cell layer, and inner cells that lack keratin.

Given its well-established role in stiffening epithelial cells8, in an embryonic context keratin might both prevent outer cells from becoming internalized by apical constriction and help to give the trophectoderm its near-perfect spherical shape. Conversely, keeping keratin levels low in cells in the centre might help them retain the flexibility in shape that they require to generate a multilayered embryo.

By using keratin filaments to stably mark the peripheral cortex, mammalian embryos (in which patterns of cell division differ widely between individuals) can ensure that cells fated to become trophectoderm are always formed in the outer layer of the cell cluster, irrespective of the orientation of divisions. Keratins play a part as asymmetrically inherited fate determinants only in these relatively rare ‘inside-out’ divisions. The early mammalian embryo therefore differs from most other systems in which asymmetric division has been studied (Fig. 1b). In those cases, in order to impose a reproducible division asymmetry, the mitotic apparatus itself is oriented so that daughter cells inherit different complements of cortically localized cell-fate determinants9.

In the coming years, it will be important to reconcile Lim and colleagues’ data with suggestions of roles for the unequal segregation of messenger RNA encoding the Cdx2 protein10 (one function of which is in forming the trophectoderm), or for differential contractility of the actomyosin protein complex11, in the symmetry-breaking events that occur at this stage in mouse embryos. The fate of dividing cells that do not express keratin at the 8-cell stage also remains to be studied.

Taking a broader perspective, this work shows how the cellular function of a protein such as keratin can emerge from its physical characteristics. In early mouse embryos, keratins provide a physical memory of polarity that is relatively independent of cell-division events. In other organisms, from bacteria to multicellular animals, other proteins that polymerize or form aggregates have also been found to provide a physical memory of cell state during asymmetric divisions12 (Fig. 1c). So Lim and co-workers’ study provides another intriguing example of nature exploiting the material properties of a protein.

Nature 585, 352-353 (2020)



  1. 1.

    Lim, H. Y. G. et al. Nature 585, 404–409 (2020).

    Article  Google Scholar 

  2. 2.

    Johnson, M. H. & Ziomek, C. A. Cell 24, 71–80 (1981).

    PubMed  Article  Google Scholar 

  3. 3.

    Zenker, J. et al. Cell 173, 776–791 (2018).

    PubMed  Article  Google Scholar 

  4. 4.

    Jackson, B. W. et al. Differentiation 17, 161–179 (1980).

    PubMed  Article  Google Scholar 

  5. 5.

    Serres, M. P. et al. Dev. Cell 52, 210–222 (2020).

    PubMed  Article  Google Scholar 

  6. 6.

    Duarte, S. et al. Nature Commun. 10, 4200 (2019).

    PubMed  Article  Google Scholar 

  7. 7.

    Nishioka, N. et al. Dev. Cell 16, 398–410 (2009).

    PubMed  Article  Google Scholar 

  8. 8.

    Yamada, S., Wirtz, D. & Coulombe, P. A. J. Struct. Biol. 143, 45–55 (2003).

    PubMed  Article  Google Scholar 

  9. 9.

    Knoblich, J. A. Nature Rev. Mol. Cell Biol. 11, 849–860 (2010).

    PubMed  Article  Google Scholar 

  10. 10.

    Skamagki, M., Wicher, K. B., Jedrusik, A., Ganguly, S. & Zernicka-Goetz, M. Cell Rep. 3, 442–457 (2013).

    PubMed  Article  Google Scholar 

  11. 11.

    Maître, J.-L. et al. Nature 536, 344–348 (2016).

    PubMed  Article  Google Scholar 

  12. 12.

    Otzen, D. & Riek, R. Cold Spring Harb. Perspect. Biol. 11, a033860 (2019).

    PubMed  Article  Google Scholar 

Download references


Nature Careers


Nature Briefing

An essential round-up of science news, opinion and analysis, delivered to your inbox every weekday.

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing


Quick links