Cell biology

The persistence of memory


Live imaging reveals that whether or not a daughter cell proliferates is influenced by two molecular factors inherited from its mother, providing insight into how the behaviour of a newly born cell can be predetermined. See Letter p.404

A fundamental principle of cell theory is that all cells arise from pre-existing ones. Every cell, except sperm and eggs, inherits an essentially identical copy of its mother's genome, which it then passes on to two daughters when it divides. But it can also inherit a variety of other 'memories' from its mother cell, in the form of proteins, RNA and other biochemical keepsakes. Identifying these molecular memories and understanding how they influence cell behaviour has been a long-standing puzzle. On page 404, Yang et al.1 tackle the question of how molecular memories acquired from the previous generation of cells influence whether daughter cells proliferate or enter a reversible resting state known as quiescence.

Proliferation drives both the development of an organism and the maintenance of its tissues. In response to growth signals, proliferating cells proceed through an initial phase of growth (known as G1), after which they begin DNA synthesis (S phase). Following a second growth phase (G2), the mother cell divides its contents into two daughter cells through a process called mitosis. Not all cells proceed swiftly through these phases, however. Instead, some temporarily withdraw from the cell cycle before S phase, entering quiescence2.

How does a cell 'decide' between proliferation and quiescence? A study in the 1970s suggested that this decision is made during G1, before a cell commits to DNA synthesis3. According to this model, each cell is a clean slate, able to make an independent decision on the basis of the signalling molecules to which it is exposed. However, this idea was challenged in 2013 by the discovery that some cells are born predisposed to rapidly enter S phase4. For these cells, the decision is influenced by the experience of the mother during its G2. Precisely how this memory is transmitted from mother to daughter has remained elusive.

Yang et al. exposed mother cells to different combinations of growth signals and DNA damage. They then withdrew the signals and used live imaging to chart the proliferation–quiescence status of the daughters. They found that newly born cells 'remembered' the signalling history of their mothers. Specifically, cells from mothers exposed to growth signals had high levels of the protein cyclin D1, which promotes progression from G1 to S phase5. By contrast, cells from mothers exposed to DNA damage had high levels of the p21 protein, a potent inhibitor of G1 progression6. In fact, the balance between these two factors was highly predictive of whether a cell would undergo proliferation or quiescence. At the molecular level, cyclin D1 and p21 compete to control phosphorylation of the retinoblastoma protein, which acts like a switch that determines whether cells enter S phase.

However, p21 and cyclin D1 have short lifetimes, making it unlikely that inheritance of these factors is the basis of cellular memory. The authors therefore reasoned that cells needed a more persistent form of memory that would last from the previous G2, through mitosis and well into the daughter cell's G1. The messenger RNA molecule that encodes cyclin D1 is much longer-lived than its protein product. Similarly, the stress-response protein p53, which is an activator of p21, becomes stabilized when it is activated by DNA damage. The researchers found that, in this activated form, it can last roughly ten times as long as when it is in its inactivated form.

Directly visualizing activated p53 and cyclin D mRNA as they are passed from mother to daughter is technically challenging. To work around this difficulty, Yang and colleagues demonstrated that both long-lived factors are generated in the mother cell and detectable soon after daughter-cell birth. Moreover, changes in the levels of these factors in the mother influenced daughter-cell fate. Cyclin D1 mRNA and p53 protein therefore represent opposing molecular memories that alter the proliferation–quiescence decision of daughter cells (Fig. 1). To our knowledge, this is the first identification of molecular factors that directly compete for influence over daughter-cell fate.

Figure 1: Making cellular memories.

Dividing cells progress through a growth phase called G2, and then undergo mitotic cell division. Daughters either undergo another growth phase, G1, before committing to proliferation, or become quiescent. The addition of growth factors to the mother cell leads to the accumulation of cyclin D1 messenger RNA, whereas DNA damage leads to activation of the protein p53; the sizes of the red arrows indicate the amounts of DNA damage and of added growth factors to which the mother cells were exposed. Yang et al.1 report that cyclin D1 mRNA and p53 persist through mitosis into daughters, where cyclin D1 mRNA is translated into protein and p53 promotes the production of the p21 protein. The authors show that the ratio of cyclin D1 to p21 is an accurate predictor of whether a daughter will proliferate or enter quiescence.

An unexpected secondary finding of this study is that levels of just two molecules can predict a single cell's behaviour with exceptional accuracy. The proliferation–quiescence decision shows an ultrasensitive response to changes in the ratio of cyclin D1 to p21 — tiny changes in this ratio could dramatically switch the fate decisions of daughter cells. This finding might indicate that cyclin D1 and p21 represent the end of a complex molecular funnel that compresses multiple proliferation-promoting and -inhibiting signals carried by upstream factors into a single output. It also opens up the possibility that other binary cell-fate choices (such as a stem cell's decision to self-renew or differentiate) is predetermined by a relatively small set of inherited, competitive memory signals.

Another interesting aspect of Yang and colleagues' work is that, in the case of DNA damage, only the memory of damage — and not the damage itself — is passed from mother to daughter. This finding contrasts with recent reports showing that replication stress in mother cells leads to DNA damage that persists through mitosis, causing quiescence in daughter cells7,8. A possible explanation for this discrepancy is that Yang et al. induced high levels of DNA damage, instead of looking at the less abundant breaks that occur naturally. Greater DNA damage can more efficiently trigger a response9 that temporarily halts the cell cycle at G2, forcing cells to stop and repair the damage before entering mitosis. In either case, a history of DNA damage seems to be an important factor in a cell's proliferation–quiescence decision. Both DNA damage and the memory of such damage, by inducing quiescence, reduce the accumulation of potentially cancer-causing mutations in growing tissues.

The concept of competing molecular memories is attractive, but raises questions about the behaviour of individual cells. For example, why do some pairs of daughter cells make different decisions from one another? One possibility is that p53 and cyclin D1 mRNA are not equally distributed between daughter cells during division. This hypothesis could be tested by comparing the relative levels of these factors between sister cells immediately after division. Another question is how molecular memories cooperate with external signals — extra DNA damage during G1, signals from neighbouring cells and mechanical forces10, for instance. These external factors probably have a role in vivo, where the heterogeneous make-up of complex tissues could act either to strengthen or repress the memories of individual cells.


  1. 1

    Yang, H. W., Chung, M., Kudo, T. & Meyer, T. Nature 549, 404–408 (2017).

    CAS  Article  ADS  Google Scholar 

  2. 2

    Temin, H. M. J. Cell Physiol. 78, 161–170 (1971).

    CAS  Article  Google Scholar 

  3. 3

    Pardee, A. B. Proc. Natl Acad. Sci. USA 71, 1286–1290 (1974).

    CAS  Article  ADS  Google Scholar 

  4. 4

    Spencer, S. L. et al. Cell 155, 369–383 (2013).

    CAS  Article  Google Scholar 

  5. 5

    Musgrove, E. A., Caldon, C. E., Barraclough, J., Stone, A. & Sutherland, R. L. Nature Rev. Cancer 11, 558–572 (2011).

    CAS  Article  Google Scholar 

  6. 6

    Karimian, A., Ahmadi, Y. & Yousefi, B. DNA Repair 42, 63–71 (2016).

    CAS  Article  Google Scholar 

  7. 7

    Arora, M., Moser, J., Phadke, H., Basha, A. A. & Spencer, S. L. Cell Rep. 19, 1351–1364 (2017).

    CAS  Article  Google Scholar 

  8. 8

    Barr, A. R. et al. Nature Commun. 8, 14728 (2017).

    CAS  Article  ADS  Google Scholar 

  9. 9

    Mankouri, H. W., Huttner, D. & Hickson, I. D. EMBO J. 32, 2661–2671 (2013).

    CAS  Article  Google Scholar 

  10. 10

    Gudipaty, S. A. et al. Nature 543, 118–121 (2017).s

    CAS  Article  ADS  Google Scholar 

Download references

Author information



Corresponding authors

Correspondence to Katarzyna M. Kedziora or Jeremy E. Purvis.

Related links

Related links

Related links in Nature Research

Systems biology: Defiant daughters and coordinated cousins

Systems biology: Molecular memoirs of a cellular family

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Kedziora, K., Purvis, J. The persistence of memory. Nature 549, 343–344 (2017). https://doi.org/10.1038/nature23549

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.