Progress in Biophysics and Molecular Biology 165 (2021) 66e71

Contents lists available at ScienceDirect

Progress in Biophysics and Molecular Biology

journal homepage: www.elsevier.com/locate/pbiomolbio

Cellular Darwinism: Regulatory networks, stochasticity, and selection

in cancer development
Denis Noble
Department of Physiology, Anatomy & Genetics, University of Oxford, OX1 3PT, UK

a r t i c l e i n f o a b s t r a c t

Article history: There are strong parallels between the evolutionary origin of species within populations of organisms
Received 29 December 2020 and new concepts for the origin of cancers within cell populations in the tissues of the body. The analogy
Received in revised form is that cancers can be regarded as a new somatic species developing within the host organism. In both
8 June 2021
cases, understanding the processes involved requires a multi-scale analysis, including higher-level
Accepted 14 June 2021
control of genetic and epigenetic changes. A key to developing successful therapeutic strategies will
Available online 17 June 2021
be to identify the processes that control heterogeneity in tissues. These include processes outside the
currently dominant theory of evolution, i.e. the Modern Synthesis. Specifically, organisms can partially
Keywords:
Cancer evolution
direct both genetic and epigenetic changes through the harnessing of chance. The loci and rates of
Harnessing of chance mutation and of genome reorganisation are forms of targeted functional reorganisation of genomes. They
Harnessing of stochasticity are more likely to result in functional reorganisations compared to the slow accumulation of point
Hypermutation in evolution mutations.
Tissue heterogeneity © 2021 Elsevier Ltd. All rights reserved.
Central dogma

1. Introduction I will show that the reverse must also be the case. Cells and
organisms can control and use the stochasticity within them, and in
A major theme of this Special Issue on Cancer and Evolution is their environments, so that stochasticity is harnessed (Noble, 2017).
that there are strong parallels between the evolutionary origin of Development and evolution then become partially directional. Or-
species within populations of organisms and new concepts for the ganisms can influence their futures and inheritance. The organism
origin of cancers within cell populations in the tissues of the body. is therefore not just a vehicle for its genes but, together with its
The analogy is that cancers can be regarded as a new somatic ancestors, can be considered the author of its own genome
species developing within the host organism. But how, in fact, do (Shapiro, 2017).
new species develop? As a historical aside, this idea is closer to Darwin's (1859)
That question is not as obvious as it may seem. Because, to fully “Artificial Selection”, and his later work on “Sexual Selection”
appreciate the parallel between the origins of cancers and the or- (Darwin 1871). In both cases Darwin's meaning is real (not blind)
igins of species, we need to turn the current popular theory of selection by organisms themselves. Natural selection, by contrast, is
evolution, the Modern Synthesis (Huxley, 1942; Futuyma and more like a filter. The idea of selection in “Natural Selection” is
Fitzpatrick, 2018; Coyne, 2009), on its head with regard to therefore metaphorical. This important distinction is very clear at
random genetic variations. That theory, also called Neo-Darwinism, the beginning of the Origin of Species (Darwin 1859).
is based on the assumption that inherited variations are always My conclusion will be that there are strong reasons to think that
purely random. The better survival of fitter organisms is then cancerous tumours are doing something similar. The biological
attributed to natural selection, which is blind in the sense that it mechanisms enabling them to do so exist.
does not purposefully chose which particular variants should be
generated. Variations are undirected even if evolutionary adapta-
2. Terminology
tion appears directed. This idea was famously popularised in
Dawkins’ (1986) The Blind Watchmaker.
There has been considerable confusion over what is meant in
evolutionary biology by saying that genetic changes are or are not
random. When randomness is harnessed by higher-level functional
E-mail address: [email protected]. systems, they can be both. It then depends on the level from which

https://doi.org/10.1016/j.pbiomolbio.2021.06.007
0079-6107/© 2021 Elsevier Ltd. All rights reserved.
D. Noble Progress in Biophysics and Molecular Biology 165 (2021) 66e71

one views what is happening. It is easy to illustrate the difference parallel:

by considering what happens in the immune system, or in any
“The apparently predetermined gene expression patterns that
other physiological process involving guided somatic hyper-
characterise the defined cell phenotypes are the results of a
mutation. At the hyper-mutating sites e the variable part of the
selective stabilization of some patterns through interactions
immunoglobulin sequence in the case of the immune system e the
between the cells. This way of framing one of the modern bio-
variations will be seen as stochastic. At the supra-cellular level of
logy's central questions calls for the same reasoning Charles
selection the very same process will be viewed as non-random
Darwin proposed to explain the evolution of biological species.
since it favours the successful cells that produce a match for the
The idea that spontaneous variation followed by selective sta-
antigen. Yet the success of those cells depends on a random process
bilization of some of these variants can account for the emer-
at the molecular level. Harnessed randomness is then no longer
gence of new cell types during ontogenesis places the evolution
purely random (undirected) from the viewpoint of the organism.
of the species and cell types on the same theoretical ground.”
The guidance can be either active or passive. As a physiologist I
would define the difference as being whether the guidance is
purely structural or whether it depends on a decision process. The The emphasis is mine and it speaks for itself. Kupiec's idea is
immune system clearly involves a higher-level choice between the precisely the key concept of the Cancer and Evolution comparison.
results thrown up by hypermutation by targeting the process to the Note though that the influence is not just one way. The study of
DNA that encodes the antibody-binding variable regions of im- cancer benefits from concepts developing currently in evolutionary
munoglobulins. Also, the somatic hypermutation process is trig- processes that lie outside the narrow confines of the Modern
gered only in activated B cells that have entered “germinal centers” Synthesis (Shapiro and Noble, 2021, this volume). In turn, inter-
in lymph nodes and the spleen. It is targeted. preting tumour development, and particularly rapid metastasis, as
A good example of passive guidance would be the photosyn- a rapid evolutionary radiation strongly reinforces the need to break
thetic process. Plants and cyanobacteria passively absorb the pho- out from those narrow confines.
tons that happen to hit them, yet the resulting storage of energy is
clearly not random e it is a canalisation of the radiation energy. But 4. Harnessing of chance
there is no active selection of photons. This difference is important
to keep in mind when applying the canalisation concept to tumour By harnessing chance variations organisms canalise develop-
development. A key question will be the extent to which flexible ment and evolution in directions that favour maintenance of their
physiological control processes are involved in cancer radiation of functions. Canalisation by higher-level (meaning epigenetic, i.e.
cellular forms and what those might be. The answer to that ques- above the genome) organisation was first described by Conrad
tion will also be relevant to possible therapeutic approaches. If the Waddington (1957). Moreover, as Huang (2021) also notes, the
guidance is through active physiological control processes then we meaning of ‘epigenetic’ here is not the modern meaning of
need to understand how to work with their flexibility to nudge “epigenetic regulation by DNA methylation and covalent chromatin
those processes towards better outcomes. This idea will become marks.” It is the canalisation process, which includes transitions
clearer when I introduce the concept of the epigenetic landscape between attractor states with no necessary mutations or even
(Fig. 2). changes in genome or histone marking.
A further terminological clarification is that what I usually refer The evolutionary process is therefore not completely blind. It
to as “physiological regulatory networks” corresponds with what can be characterised as a form of feeling forwards, the one-eyed
many authors call “Gene Regulatory Networks (GRNs).” I don't use watchmaker rather than a blind one (Noble and Noble, 2017).
that terminology because the regulation is not by genes, it is Many of the physiological control processes by which organisms
regulation at a higher level. The genes themselves are what are achieve this kind of control are now fairly well understood. The
regulated. GRN terminology sometimes seems ambiguous in rela- regulatory networks by which cells maintain their integrity effec-
tion to this distinction. The processes involved float largely free of tively buffer higher levels against molecular-level random varia-
the genome since, as Huang (2021) argues: tions (Noble, 2011). This process protects the organism from many
possibly deleterious effects of genomic change by developing
“The GRN is the master system through which a genome im-
robustness in the functional networks, so ensuring that vital
poses genome-wide constraints in gene expression, such that
physiological functions are maintained. Yet harnessing still permits
despite random molecular noise, activities of gene loci are not
variation to be used when the organism needs to do so. Harnessing
independent but globally coordinated. This coordination results
of chance therefore serves a dual role in living organisms: pro-
in recurrent gene expression patterns that are stable to
tecting function from disorder, while also selecting and targeting
perturbations”
disorder at the cellular and molecular levels when they need to
create novelty.
As a bonus, this view of the relationship between regulatory
3. Cell selection in development networks and the genome automatically explains the very low as-
sociation scores in Genome-wide association studies. These are so
A theory of Darwinian selection amongst random variations in low that it is clear that very many genes must be involved in each
populations of cells in developing tissues was first developed by phenotype. Some even go so far as to conclude that perhaps all
Kupiec (1983, reprinted 2020), although it had a partial precursor in genes are involved in every function: now known as the omnigenic
Weismann's idea of selection amongst germ-line cells (Weismann, hypothesis (Boyle et al., 2017). This is what we would expect if the
1896). Kupiec (2009) emphasised the role of stochasticity in cellular functional networks operating above the level of the genome are
development. Elsasser (1984) was also one of the early pioneers of a integrating, as they must, the activity of all the gene products
non-reductionist model of cellular function leading to similar involved. In a completely interconnected system, perturbations of
conclusions. His approach was non-reductionist precisely because all components may produce change in overall behaviour.
reductionists tend to avoid stochasticity and espouse determinism. This issue is of practical importance since it can be used to
Writing a commentary on the 2020 republication of Kupiec's distinguish between association and causality. By connecting ge-
1983 paper, Andras Paldi (2020) has succinctly expressed the nomics data to physiological modeling of functional processes it
67
D. Noble Progress in Biophysics and Molecular Biology 165 (2021) 66e71

may become possible to introduce causal interpretations of

Genome-wide association studies e GWAS (Noble and Hunter,
2020). The low association scores are not only what would be ex-
pected if the functional levels above the genome are strongly
robust, they are also the clues to establishing causality rather than
just association. As I have shown in the case of the robustness of the
cardiac rhythm generator (Noble, 2011), a highly functional
component gene or protein can show a low or even zero association
score. Even a process as central as circadian rhythm can survive the
knockout of a key gene (Debruyne et al., 2006), while studies of
yeast showed that as many as 80% of knockouts can be functionally Fig. 1. Central Dogma diagram (bottom row of causation) embedded in interactions
silent under standard growth conditions (Hillenmeyer et al., 2008). with the system of functional causal networks (often called GRNs). The forms of
Yet most of those knockouts were found to be functional under causation in Crick's 1958 formulation are represented by the blue arrows: DNA makes
conditions of nutritional stress. RNAs and proteins that then form the networks. All causation is upward. The 1970
reformulation after the discovery of reverse transcription is represented by the orange
Living organisms can therefore be seen as living on the
arrows. The downward green arrows represent the processes of natural genetic en-
boundary between order and disorder (Noble, 2021a). Contrary to gineering in response to environmental influences (downward black arrows). Purple
Schro€dinger's (1943) proposal that organisms create high-level arrow represents the influence of network structure on protein folding and function
order from molecular-level order, life resembles physics, particu- (Noble, 2021b).
larly thermodynamics, in creating order from disorder. I have
explained the details of this contrast with Schro€dinger's theory, and
why that theory cannot be correct, in a previous article in this chromosomes exist in a liquid-crystalline state (Chow et al., 2010),
journal (Noble, 2021c). Furthermore, as already noted above, but how these organisms access and replicate their DNA is un-
determinism and randomness are opposites. So why do we need to known. Clearly, the liquid crystalline state must somehow permit
reject Schro€dinger's theory, and its subsequent inspiration for the that.
Central Dogma of molecular biology (Crick, 1958, 1970)?
The reason is that it is physically impossible for DNA to replicate 5. Stochasticity in cell populations
like a crystal (which was the metaphorical basis for Schro €dinger's
theory) since, inside a cell, it does not actually form a crystal. In Sui Huang (2009, 2013; Pisco et al., 2013; Zhou et al., 2014) and
living organisms it remains at all times a thread, wound around its his colleagues have demonstrated extensive stochasticity of protein
mirror image to form the double helix, which is in turn wound expression even in clonal cell populations. The range of expression
around the chromatin histone proteins. This thread-like character levels can be as much as 1000 fold different from cell to cell. Their
is necessary for it to be possible for parts of the thread to become work also shows that the form of distribution within the population
loosened enough from the condensed attachment to the histones to is a population level attractor. If cells are cloned from high ex-
be read. The sequence is not locked up in the rigid 3-dimensional pressers, they initially form a population of high expressers, but
structure of a crystal. Replication cannot therefore use the deter- over a few days the distribution of expression reverts to the original
minate crystal method of self-templating. DNA is not a crystalline distribution. This is true even in populations showing bi-modal
self-replicator. distributions. Cells cloned from one of the two peaks initially
Schro€dinger could not have known that in 1942. He would have show the peak from which they were cloned. Over several days the
had to foresee the much later discoveries on the remarkable cellular bi-modal distribution becomes re-established.
machinery of DNA error-correcting processes that can reduce a Huang uses a diagram inspired by Waddington’s (1957) original
natural copying error rate from 1 in 104 base pairs to only 1 in 1010. epigenetic landscape diagram to explain his results, as shown in
What happens then to the Central Dogma of molecular biology? Fig. 2. In the landscape representation, individual cells will shift
It loses its status as a general principle of one-way causation in around amongst the various stable positions and the landscape of
biology, and reverts to being a more restricted molecular biological their probabilities will determine the overall population distribu-
fact about templating: DNA sequences form templates to make tion. The epigenetic landscape can therefore be regarded as a
proteins, but protein sequences do not form templates for making thought-tool embodying the concept of harnessing the random-
DNA. That is what the molecular biology showed. No more than ness, and canalising development towards the population attractor.
that. As an attractor the landscape can persist through cell divisions, just
The processes by which the organism changes its DNA do not as happens once cellular differentiation has occurred in the
rely on template information. Fig. 1 illustrates that the flow of developing embryo. The same genome can be used to generate
causation then is from higher-level functional networks that can hundreds of different cell types.
sense when the organism is stressed in its environment and when What might the function of such stochasticity be? The general
it can use the processes of natural genetic engineering (Shapiro, function is that it permits selection amongst cells with different
2011 2021). gene expression profiles, just as the immune system selects
As a side-issue, if DNA does not form a crystal in living cells, how amongst cells with different hyper-mutation outcomes. The dif-
was the crystallographic data for deducing the double helical ference lies in whether the stochasticity is genetic or epigenetic. In
structure of DNA obtained? The answer in the case of Rosalind conditions of stress this difference may not matter. As Waddington
Franklin's work, which was used by Watson and Crick, was that she (1957) first showed, epigenetic changes can canalise development
produced X-ray diffraction patterns from fibrous DNA, the famous over a sufficient number of generations to enable genetic assimi-
Photo 51 (Pederson, 2020). 3D crystals formed from DNA were not lation to occur. If rapid mutation is already occurring as a reaction
studied until Alexander Rich's much later work on Z-DNA (Wang to stress, the existing genetic variability could be enough to permit
et al., 1979; Ha et al., 2005; Schimmel, 2015). DNA can be made rapid genetic assimilation (see explanation of this process in Noble,
to form 3 dimensional crystal lattices (Paukstelis et al., 2004) but 2016, pp 216e219). Speciation then begins. Possible markers for
these are not the forms in which it exists in a living cell. There is a early stage cancer might therefore depend on identifying the
partial exception in the case of dinoflagellates where the relevant parts of the regulatory networks that are responsible for
68
D. Noble Progress in Biophysics and Molecular Biology 165 (2021) 66e71

Fig. 2. An epigenetic landscape diagram to illustrate Huang et al.‘s observation of population attractors in gene expression dynamics. (A) A projected epigenetic landscape with two
attractors [low X (LX), high X (HX)] and their sub-attractors, which contribute to heterogeneity. Each circle represents a network state (i.e. a cellular phenotype) determined by the
level of X as indicated by the position on the horizontal axis (i.e. one state space dimension, trait X, of the high-dimensional state space). The vertical axis displays the ‘potential’ V
(X). The height of the accumulation of circles reflects the density distribution as a function of X. (B) Associated flow cytometry histograms of cell population distributions with
respect to X. Subpopulation sorting can reveal the reversibility and the transition rates between the subpopulations (from Huang, 2009).

the architecture of the epigenetic landscape attractor. That is a big attractors, leads to the view that mutations are largely followers in
challenge since it will require deep understanding of the architec- cancer evolution. This also is a parallel between cancer evolution
ture and complex dynamics of cellular networks. and species evolution (West-Eberhard, 2003).
The process of genetic assimilation, guided by epigenetic

69
D. Noble Progress in Biophysics and Molecular Biology 165 (2021) 66e71

6. Tissue stochasticity in cancer and the development of drug Huang concludes:

resistance
“Of practical relevance is that recognition of the phenomenon of
drug-induced escalation of malignancy opens a new opportu-
That is the question tackled by Zhou et al. (2014). They ask
nity for therapeutic approaches designed to stifle therapy
whether the models of population genetics that have been exten-
resistance: if development of resistance is not a Darwinian
sively developed within the framework of the Modern Synthesis
evolutionary progress fueled by random mutations but a
could be used. The problem is that “these models tacitly assume a
response shaped by network dynamics that involves attractor
one-to-one mapping between genotype and phenotype and as-
transitions, then it is mediated by the molecular signaling
sume random genetic mutations as the mechanism for cell
pathways that, in the healthy part of the landscape, facilitate
phenotype innovation.” This is a serious drawback since genome-
state transitions during development but now misdirect tran-
wide association studies have clearly falsified this assumption in
sitions into the wrong attractors. These pathways could serve as
favour of multigenic or even omnigenic theories (Boyle et al., 2017).
therapeutic targets.”
New approaches are therefore required (Huang, 2009; Altschuler
and Wu, 2010). The requirements all fall outside the Modern Syn-
thesis framework. Zhou et al. identify the three requirements: I believe this could be the game-changer in cancer treatment. If
we can nudge epigenetic landscapes in the right direction by
 Reversible state transitions: these transitions are often approxi- intervening before genomic assimilation occurs then we have the
mately symmetrical while mutations in the traditional model justification for early intervention. As I hope I have made clear, this
are typically irreversible; requires abandoning the Modern Synthesis framework as the sole
 Frequent state transitions: transition rate often is in the same model for development and evolution. Non-genetic cell plasticity
time scale as division time or even faster, while mutation rate makes it impossible to explain within the limits of somatic muta-
per locus is much slower than the division rate. tion theory (Shen and Clairambault, 2020). Switching to a more
 Transitions are not strictly non-Lamarckian. Since they are due to productive approach to cancer therapy is urgently needed since we
changes of gene expression and not mutations in the genome are facing not only an impasse in developing successful therapies,
they can be induced in a controlled (purposeful) or uncontrolled but also a financial one (Fojo et al., 2014).
manner while mutations are randomly directed and their rates
not easily tunable. 7. The role of intercellular communication

I agree with these requirements, but it is important to note that The existence of tissue-level attractors that determine the sto-
the statement “mutations are randomly directed and their rates not chastic distributions in cell populations requires intercellular
easily tunable” is really just an assumption of standard evolu- communication. Can cells communicate their epigenetic states to
tionary theory. In fact, growth conditions can influence both the each other?
number and spectrum of mutations, and the frequencies are One of the ways in which this can be achieved is via intercellular
exquisitely sensitive to stresses, including factors like exposure to exchange of information within extracellular vesicles, EVs
sunlight, as well as factors like mitotic errors. In E. coli the presence (Edelstein et al., 2019). These contain the RNAs and other molecules
or absence of a DNA polymerase (DNA Pol IV) can dramatically that both give a portrait of the epigenetic state of the cells from
change the mutation rate (Wagner and Nohmi, 2000). which they come, and which can also influence the epigenetic state
These requirements require modeling the directed harnessing of of the cells that absorb them. The implications for the development
stochasticity as argued earlier in this article. Zhou et al. (2014) of cancers are important. Those implications are the subject of
specifically conclude “By uniting the dynamics of reversible tran- another presentation in this issue of the journal (Bonner and
sitions between distinct cell phenotypes with their growth rates, Willms, 2021).
we can capture the contributions for the evolution of the cell Cells can also communicate via their electrophysiological
population from both Darwinian selection and Lamarckian induc- properties, either directly when they are electrically coupled, or via
tion.” Detailed examples of such modeling are given in their article. the ion exchanges with the tissue fluid. These aspects are addressed
Can this approach help in clarifying the relative roles of muta- in another article in this Special Issue (Levin, 2021).
tions and epigenetic transitions in cancer development? This is the
issue tackled in Pisco et al. (2013) who have combined quantitative 8. Conclusions
measurements and mathematical modeling to investigate possible
mechanisms of multi-drug resistance in H60 leukemic cells. The The main conclusion of this article is that the active role of
results clearly show that the speed with which such resistance physiological regulatory networks (often called Gene Regulatory
develops is better explained by epigenetic transitions. This is Networks when they refer to intracellular regulation) can help
important for two reasons. First, their arguments match well with explain why so little advance has been made in the treatment of
parallel arguments in relation to evolutionary biology. Epigenetic late stage metastasis. Rapid evolution, of species and tumours, re-
transitions will always enable faster change since there is no need quires more processes than the gradual accretion of point muta-
to wait for single mutations to slowly spread through a population. tions. Mutations are more likely to be followers of the development
The environmental factors that can canalise the relevant epigenetic of new attractor states at levels above the genome. Therapeutic
transitions will influence the whole population, not just single cells. strategies would be best targeted at that level of organisation. This
Lamarckian evolution is therefore inevitably much faster conclusion is compatible with the concept of two-phased cancer
(Burggren, 2016). In any situation in which both mechanisms can evolution, comprising a genome reorganisation mediated punctu-
operate, the epigenetic one will win out. ated phase and a gene-mutation mediated stepwise phase, as
The good news is that, for the same reasons, epigenetic variation outlined by Heng and Heng (2020).
could be more readily reversed since the changes are not neces- Epigenetic control of genomic and evolutionary changes could
sarily assimilated into the genome. Pisco et al.‘s work therefore play a role in both phases, particularly if we include Waddington's
reveals a “new opportunity for early therapeutic intervention early (1957) definition of epigenetics, which was the canalisation of
against the development of drug resistance.” In a 2013 review genomic change by the phenotype followed by genetic
70
D. Noble Progress in Biophysics and Molecular Biology 165 (2021) 66e71

assimilation. The modern definition, by focusing on genome and Galhardo, R.S., Hastings, P.J., Rosenberg, S.M., 2007. Mutation as a stress response
and the regulation of evolvability. Crit. Rev. Biochem. Mol. Biol. 42 (5), 399e435.
chromosome marking has narrowed the concept down compared
Ha, S.C., Lowenhaupt, K., Rich, A., Kim, Y.G., Kim, K.K., 2005. Crystal structure of a
to Waddington's work. Both forms of epigenetic process must be junction between B-DNA and Z-DNA reveals two extruded bases. Nature 437
involved to fully understand how the physiological regulatory (7062), 1183e118.
networks at all levels carry out their role. Heng, J., Heng, H.H., 2020. Genome chaos: creating new genomic information
essential for cancer macroevolution. Nov 13:S1044-579X(20)30224-8 Semin
A second conclusion is that the harnessing of stochasticity may Cancer Biol.. https://doi.org/10.1016/j.semcancer.2020.11.003. Epub ahead of
play a major role in tumour development. Genome reorganisation print. PMID: 33189848.
of tissues under stress would explain the rapid radiation of cell Huang, S., 2009. Non-genetic heterogeneity of cells in development: more than just
noise. Development 136, 3853e3862.
forms in late stage cancer. Huang, S., 2013. Genetic and non-genetic instability in tumor progression: link
I acknowledge the previous work, more than a decade ago, by between the fitness landscape and the epigenetic landscape of cancer cells.
Susan Rosenberg showing that the occurrence of random genetic Canc. Metastasis Rev. 32 (3e4), 423e448.
Huang, S., 2021. Reconciling Non-genetic Plasticity with Somatic Evolution in
mutations may be a tightly controlled, and, in a sense, “purposeful” Cancer (Submitted to Trends in Cancer.
response to stressful environments. She and her team presented Huxley, J., 1942. Evolution. The Modern Synthesis. OUP.
good evidence of stress-induced mutagenesis programs (Galhardo Hillenmeyer, M.E., Fung, E., Wildenhain, J., Pierce, S.E., Hoon, S., Lee, W., et al., 2008.
The chemical genomic portrait of yeast: uncovering a phenotype for all genes.
et al., 2007). Rosenberg's hypothesis was that stress-inducible Science 320, 362e365.
mutagenesis mechanisms can “accelerate adaptive evolution in Kupiec, J.-J., 1983. A probabilistic theory for cell differentiation, embryonic mortality
populations specifically when organisms are maladapted to their and DNA c-value paradox. Speculations Sci. Technol. 6, 471e478.
Kupiec, J.-J., 2009. The Origin of Individuals. World Scientific.
environments, i.e., when they are stressed, and then return ge-
Kupiec, J.-J., 2020. A probabilistic theory for cell differentiation, embryonic mor-
nomes to low mutation rates in rare adapted mutants that thrive in tality and DNA c-value paradox, Organisms. J. Biol. Sci. 4 (1), 8e10. https://
the new environment and so are stressed no longer". doi.org/10.13133/2532-5876/16955, 2020.
Levin, M. (2021). (Article in this issue).
Noble, D., 2011. Differential and integral views of genetics in computational systems
Author statement biology. Interface Focus 1, 7e15.
Noble, D., 2016. Dance to the Tune of Life. Biological Relativity. Cambridge Uni-
This article has not been submitted elsewhere. versity Press, Cambridge.
Noble, D., 2017. Evolution viewed from physics, physiology and medicine. Interface
No funding supported this work, and I have no conflict of in- Focus 7, 20160159.
terest to declare. Noble, D., 2021a. Function forms from the symmetry between order and disorder.
Function (in press).
Noble, D., 2021b. The illusions of the modern synthesis. Biosemiotics (in press).
Acknowledgements Noble, D., 2021c. The role of stochasticity in biological communication processes.
Prog. Biophys. Mol. Biol. 162, 122e128. https://doi.org/10.1016/
I thank James Shapiro, Azra Raza and Sui Huang for many helpful j.pbiomolbio.2020.09.008.
Noble, D., Hunter, P.J., 2020. How to link genomics to physiology through epi-
comments in writing this article. genomics. Epigenomics 12. https://doi.org/10.2217/epi-020-0012.
Noble, R., Noble, D., 2017. Was the watchmaker blind? Or was she one-eyed?
References Biology 6, 47.
Paldi, A., 2020. Stochastic or deterministic? That is the question. Organisms. J. Biol.
Sci. 4, 77e79. https://doi.org/10.13133/2532-5876/16963.
Altschuler, S.J., Wu, L.F., 2010. Cellular heterogeneity: do differences make a dif-
Paukstelis, P.J., Nowakowski, J., Birktoft, J.J., Seeman, N.C., 2004. Crystal structure of
ference? Cell 141, 559e563.
a continuous three-dimensional DNA lattice. Chem. Biol. 11, 1119e1126.
Bonner, S., Willms, E., 2021. Extracellular vesicle in metastasis, and their evolu-
Pederson, T., 2020. The double helix: “Photo 51” revisited. FASEB J. 34, 1923e1927.
tionary aspects. Prog. Biophys. Mol. Biol. (In this Special Issue).
https://doi.org/10.1096/fj.202000119.
Boyle, E.A.L., Li, Y.I., Pritchard, J.K., 2017. An expanded view of complex traits: from
Pisco, A.O., Brock, A., Zhou, J., Moor, A., Mojtahedi, M., Jackson, D., Huang, S., 2013.
polygenic to omnigenic. Cell 169, 1177e1186.
Non-darwinian dynamics in therapy-induced cancer drug resistance. Nat.
Burggren, W., 2016. Epigenetic inheritance and its role in evolutionary biology: Re-
Commun. 4, 2467.
evaluation and new perspectives. Biology 25, 24.
Schimmel, Paul, 2015. Alexander Rich (1924e2015) Biologist who discovered
Chow, M.H., Yan, K.T.H., Bennett, M.J., Wong, J.T.Y., 2010. Birefringence and DNA
ribosome clusters and 'left-handed' DNA. Nature 521 (7552), 291.
condensation of liquid crystalline chromosomes. Eukaryot. Cell 9 (10),
Schro€dinger, E., 1943. What Is Life? OUP.
1577e1587.
Shapiro, J.A., 2011. Evolution: A View from the 21st Century. Pearson Education Inc,
Coyne, J.A., 2009. Why Evolution Is True. OUP.
Upper Saddle River, NJ.
Crick, F.H.C., 1958. On protein synthesis. Symp. Soc. Exp. Biol. 12, 138e163 pmid:
Shapiro, J.A., 2017. Living organisms author their read-write genomes in evolution.
13580867.
Biology 6 (4). https://pubmed.ncbi.nlm.nih.gov/29211049/.
Crick, F.H.C., 1970. 'Central Dogma of molecular biology. Nature 227, 561e563.
Shapiro, J.A., 2021. What can evolutionary biology learn from cancer biology? Prog.
Darwin, C., 1859. The Origin of Species. John Murray, London.
Biophys. Mol. Biol. Article in this Special Issue.
Darwin, C., 1871. The Descent of Man, and Selection in Relation to Sex. John Murray,
Shen, S., Clairambault, J., 2020. Cell plasticity in cancer cell populations.
London).
F1000Research 9, 635. F1000 Faculty Rev.
Dawkins, R., 1986. The Blind Watchmaker. Norton & Company.
Wagner, J., Nohmi, T., 2000. Escherichia coli DNA polymerase IV mutator activity:
Debruyne, J.P., Noton, E., Lambert, C.M., Maywood, E.S., Weaver, D.R., Reppert, S.M.,
genetic requirements and mutational specificity. J. Bacteriol. 182, 4587e4595.
2006. A Clock Shock: mouse CLOCK is not required for circadian oscillator
Wang, A.H., Quigley, G.J., Kolpak, F.J., Crawford, J.L., van Boom, J.H., Van der Marel, G.,
function. Neuron 50 (3), 465e477.
Rich, A., 1979. Molecular structure of a left-handed double helical DNA fragment
Edelstein, L.R., Smythies, J.R., Quesenberry, P.J., Noble, D., 2019. Exosomes: A Clinical
at atomic resolution". Nature 282 (5740), 680e686.
Compendium. Academic Press.
Weismann, A., 1896. On Germinal Selection as a Source of Definite Variation. Open
Elsasser, W.M., 1984. Outline of a theory of cellular heterogeneity. Proc. Natl. Acad.
Court, Chicago.
Sci. Unit. States Am. 81, 5126e5129.
West-Eberhard, M.J., 2003. Developmental Plasticity and Evolution. Oxford OUP.
Fojo, T., Mailankody, S., Lo, A., 2014. Unintended consequences of expensive cancer
Zhou, J.X., Pisco, A.O., Qian, H., Huang, S., 2014. Nonequilibrium population dy-
therapeuticsdthe pursuit of marginal indications and a me-too mentality that
namics of phenotype conversion of cancer cells. PloS One 9 (12), e110714.
stifles innovation and creativity. The john conley lecture. AMA Otolaryngol
https://doi.org/10.1371/journal.pone.0110714Ed.
Head Neck Surg 140, 1225e1236. https://doi.org/10.1001/jamaoto.2014.1570.
Futuyma, D., Fitzpatrick, M., 2018. Evolution, fourth ed. OUP.

71