Some heritable epigenetic marks may be functionally neutral—i.e., set up in early development
but simply mechanically copied at each cell division. Because the maintenance methylation machinery has a finite error rate [1 in 25 cell divisions
per CpG, although this has only been measured in
certain contexts ( 56)], every cell may harbor a
unique code of methylation sites that would allow tracking of its developmental trajectory. This
acts as if lineage were marked by DNA mutations
(either natural ones or induced) (Fig. 4B). This
may allow noninvasive lineaging in the future
without genetic manipulation, which might be
particularly useful in human studies.
We have highlighted the different time scales
of variation of these different layers of the epigenome, as well as their interdependencies. It
is important to recognize that most of these are
from indirect measurements or inferences. In
due course, we may connect epigenome dimensions by pseudotime measurements, allowing
us to formulate temporal connections and dependencies. However, what is yet to materialize
are real-time in vivo recording systems of epigenetic states, ideally at a single-locus level. Hence
the single-cell epigenomics revolution has additional challenges to overcome. Our existing methods are already allowing us to zoom in on new
concepts of “cell fate”—for example, in developmental systems where cell history can be recorded
in epigenetic marks. Yet their actions at key decision points require yet unknown mechanisms
( 57, 58). This presumably requires new epigenomic codes for cell plasticity and future potential. Deeper insights into these rules will provide
not only a better understanding of living biological systems but also new tools and new ways of
thinking about changing cell fate experimentally.
At the other end of the spectrum, we anticipate information regarding the presumed degradation of cell fate during aging. Models
involve either clonal competition or exhaustion
and hence a potential loss of cell heterogeneity in an aging tissue. Conversely, an increase
in heterogeneity may occur with a concomitant loss of coherence of transcriptional networks ( 59). Interestingly, programmed changes
of the epigenome during aging, particularly of
the DNA methylome, accurately record chronological age. However, this “methylation aging
clock” can be accelerated or decelerated by biological interventions that shorten or lengthen
life span, respectively ( 60–62). It remains to be
seen how this methylation clock plays out at the
single-cell level. As many human adult diseases,
including cancer, are associated with altered epigenome patterns, individual cells may gradually
and in a potentially programmed way acquire
disease risk via changes in epigenetic marks
during aging. Conversely, single-cell multi-omics
methods may identify hidden cell states with
potential for tissue repair or rejuvenation.
As large-scale efforts are mapping all human
cells transcriptionally and spatially [e.g., the
Human Cell Atlas ( 63)], there is the prospect in
the future that epigenomics measurements, in
particular, will add unprecedented layers of in-
formation about memory of past experiences and
about future potential of cells in the human body.
Imagine that we had at our disposal the techniques for single-cell multi-omics, including the
ability to identify all key epigenetic modalities,
robustly and at an affordable cost. Imagine similarly that we had the computational tools to
unravel and visualize connections between the
different molecular layers within and between
cells. From such advances, we anticipate answering
many questions in embryonic development (
including comparisons of various organisms). We would
like to know any epigenetic determinants of cell
fate and lineage decisions and their timing and/or
memory of such decisions.
Travelling back in time (i.e., generating iPSCs)
or across tissues (via transdifferentiation), we will
be able to see how each cell responds in terms of
erasing epigenetic memory and acquiring new
cell fate trajectories, especially those not part of
the normal developmental repertoire. We also anticipate unraveling tissue-level heterogeneity.
Highly multiplexed methylome sequencing can
already identify cell types in a complex tissue such
as the brain with similar accuracy as transcriptome
sequencing ( 27).
Finally, we aim to discover links between
epigenetic and genetic heterogeneity, showing
to what extent epigenetic change (particularly
in disease) is driven by underlying changes in
DNA sequence such as copy-number variation,
mutations, and rearrangements in cancer, or
the mobility of selfish DNA elements. Conversely,
primary epimutations may underlie the initiation of some diseases but may subsequently elicit
more permanent genetic change that stabilizes
the disease phenotype.
These advances have implications for diagnosing and understanding disease progression. We
envision that precancerous cell states may be
detected at an early stage in tissues by their single-cell epigenome signatures, and other chronic
diseases may also reveal unique signatures of
progression. Single-cell epigenomic analyses might
allow for a biopsy of only a few cells or by capturing
small amounts of cell-free DNA in circulation. Such
tools may also reveal cell populations in tissues
with the greatest potential for regeneration and
REFERENCES AND NOTES
1. R. Hooke, Micrographia: Or Some Physiological Descriptions of
Minute Bodies Made by Magnifying Glasses, with Observations
and Inquiries Thereupon (Courier Corporation, 2003).
2. M. J. T. Stubbington, O. Rozenblatt-Rosen, A. Regev,
S. A. Teichmann, Science 358, 58–63 (2017).
3. E. Lein, L. E. Borm, S. Linnarsson, Science 358, 64–69 (2017).
4. C. Gawad, W. Koh, S. R. Quake, Nat. Rev. Genet. 17, 175–188
5. C. D. Allis, T. Jenuwein, D. Reinberg, Epigenetics (CSHL Press,
6. W. Jin et al., Nature 528, 142–146 (2015).
7. J. D. Buenrostro et al., Nature 523, 486–490 (2015).
8. D. A. Cusanovich et al., Science 348, 910–914 (2015).
9. F. Guo et al., Cell Res. 27, 967–988 (2017).
10. S. Pott, eLife 6, e23203 (2017).
11. S. J. Clark et al., bioRxiv 138685 [Preprint] ( 17 May 2017).
12. A. Rotem et al., Nat. Biotechnol. 33, 1165–1172 (2015).
13. T. Nagano et al., Nature 502, 59–64 (2013).
14. H. Guo et al., Genome Res. 23, 2126–2135 (2013).
15. S. A. Smallwood et al., Nat. Methods 11, 817–820 (2014).
16. M. Farlik et al., Cell Reports 10, 1386–1397 (2015).
17. D. Mooijman, S. S. Dey, J. C. Boisset, N. Crosetto,
A. van Oudenaarden, Nat. Biotechnol. 34, 852–856 (2016).
18. C. Zhu et al., Cell Stem Cell 20, 720–731.e5 (2017).
19. I. C. Macaulay, C. P. Ponting, T. Voet, Trends Genet. 33,
20. I. C. Macaulay et al., Nat. Methods 12, 519–522 (2015).
21. S. S. Dey, L. Kester, B. Spanjaard, M. Bienko,
A. van Oudenaarden, Nat. Biotechnol. 33, 285–289 (2015).
22. C. Angermueller et al., Nat. Methods 13, 229–232 (2016).
23. Y. Hu et al., Genome Biol. 17, 88 (2016).
24. Y. Hou et al., Cell Res. 26, 304–319 (2016).
25. C. Angermueller, H. J. Lee, W. Reik, O. Stegle, Genome Biol. 18,
26. M. Frommer et al., Proc. Natl. Acad. Sci. U.S.A. 89, 1827–1831
27. C. Luo et al., Science 357, 600–604 (2017).
28. R. M. Mulqueen et al., bioRxiv 157230 [Preprint]
(2 June 2017).
29. L. F. Cheow, S. R. Quake, W. F. Burkholder,
D. M. Messerschmidt, Nat. Protoc. 10, 619–631 (2015).
30. J. R. Peat et al., Cell Reports 9, 1990–2000 (2014).
31. X. Wu, A. Inoue, T. Suzuki, Y. Zhang, Genes Dev. 31, 511–523
32. M. R. Corces et al., Nat. Genet. 48, 1193–1203 (2016).
33. S. Preissl et al., bioRxiv 159137 [Preprint] ( 6 July 2017).
34. J. Kind et al., Cell 163, 134–147 (2015).
35. I. M. Flyamer et al., Nature 544, 110–114 (2017).
36. T. Nagano et al., Nature 547, 61–67 (2017).
37. V. Ramani et al., Nat. Methods 14, 263–266 (2017).
38. T. J. Stevens et al., Nature 544, 59–64 (2017).
39. J. van Arensbergen, B. van Steensel, Mol. Cell 66, 167–168
40. S. Gravina, X. Dong, B. Yu, J. Vijg, Genome Biol. 17, 150
41. W. Zhang, T. D. Spector, P. Deloukas, J. T. Bell, B. E. Engelhardt,
Genome Biol. 16, 14 (2015).
42. J. Ernst, M. Kellis, Nat. Biotechnol. 33, 364–376 (2015).
43. F. Buettner et al., Nat. Biotechnol. 33, 155–160 (2015).
44. G. Ficz et al., Cell Stem Cell 13, 351–359 (2013).
45. A. McKenna et al., Science 353, aaf7907 (2016).
46. J. P. Junker et al., bioRxiv 056499 [Preprint] (1 June 2016).
47. S. D. Perli, C. H. Cui, T. K. Lu, Science 353, aag0511 (2016).
48. R. Kalhor, P. Mali, G. M. Church, Nat. Methods 14, 195–200
49. K. L. Frieda et al., Nature 541, 107–111 (2017).
50. M. Nishizawa et al., Cell Stem Cell 19, 341–354 (2016).
51. X. Huang, J. Wang, Cell Stem Cell 20, 741–742 (2017).
52. C. D. Laird et al., Proc. Natl. Acad. Sci. U.S.A. 101, 204–209
53. L. Zhao et al., Genome Res. 24, 1296–1307 (2014).
54. F. von Meyenn et al., Mol. Cell 62, 983 (2016).
55. P. Giehr, C. Kyriakopoulos, G. Ficz, V. Wolf, J. Walter,
PLOS Comput. Biol. 12, e1004905 (2016).
56. T. Ushijima et al., Genome Res. 13, 868–874 (2003).
57. H. J. Lee, T. A. Hore, W. Reik, Cell Stem Cell 14, 710–719
58. H. Mohammed et al., Cell Reports 20, 1215–1228
59. C. P. Martinez-Jimenez et al., Science 355, 1433–1436 (2017).
60. S. Horvath, Genome Biol. 14, R115 (2013).
61. G. Hannum et al., Mol. Cell 49, 359–367 (2013).
62. T. M. Stubbs et al., Genome Biol. 18, 68 (2017).
63. A. Regev et al., bioRxiv 121202 [Preprint] ( 8 May 2017).
64. B. Lake et al., bioRxiv 128520 [Preprint] ( 19 April 2017).
W.R. thanks I. Herraez, T. Stubbs, S. Clark, C. Alda, H. Mohammed,
M. Eckersley-Maslin, S. Rulands, W. Dean, J. Marioni, and B. Simons
for discussions or comments on the manuscript. Thank you to
V. Juvin (SciArt Work) for artwork. Work in W.R.’s laboratory is
supported by the Wellcome Trust, the Biotechnology and Biological
Sciences Research Council (BBSRC), and the Medical Research
Council (MRC): G.K. is supported by the BBSRC and the MRC;
O.S. is supported by European Molecular Biology Laboratory core
funding, the Wellcome Trust, and the European Research Council.
W.R. is a consultant and shareholder of Cambridge Epigenetix.