Epigenetics
Somatic and hereditary disease
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
two mo
. . . . . . .
1. Introduction
Understanding the relationship between the apparently rigid
coding sequence in the genotype and the much more flexible
cellular phenotype in higher organisms is an enduringly important
Contents lists availab
Progress in Biophysics a
Progress in Biophysics and Molecular Biology xxx (2011) 1e7E-mail address: keith.baverstock@uef.fi.3.1.2. Molecular biological considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
3.1.3. A theoretical consideration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
3.1.4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
3.2. Potential problems with the IA model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
4.1. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
Conflict of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
Funding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . .
2. Materials and methods . . . . . . . . . .
2.1. The models . . . . . . . . . . . . . . .
3. Results . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1. Discriminating between the
3.1.1. The experiment . . .0079-6107/$ e see front matter
2011 Elsevier Ltd.
doi:10.1016/j.pbiomolbio.2011.01.002
Please cite this article in press as: Baverstock
phenotype, Progress in Biophysics and Moleacts as the interface between the cell and its environment. Environmental sources of stress can serve to
trigger attractor and therefore phenotypic, transitions without entailing genotypic sequence changes.
It is asserted that the evidence from cell and molecular biological research and logic, favours the
second model. If correct there are important implications for understanding how environmental factors
impact on evolution and may be implicated in hereditary and somatic disease.

2011 Elsevier Ltd. All rights reserved.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
dels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00Speciation
Genomic instabilitygene products (mainly proteins) and independent of the genomic sequence, are descended from a pre-
cellular state fromwhich life originated. As this model is based on the cell as an open system the attractorA comparison of two cell regulatory models entailing high
dimensional attractors representing phenotype
Keith Baverstock
Department of Environmental Science, University of Eastern Finland, Kuopio Campus, 70211 Kuopio, Finland
a r t i c l e i n f o
Article history:
Available online xxx
Keywords:
Phenotype represented by attractor
Cell regulation
a b s t r a c t
Two models for mammalian cell regulation that invoke the concept of cellular phenotype represented by
high dimensional dynamic attractor states are compared. In one model the attractors are derived from an
experimentally determined genetic regulatory network (GRN) for the cell type. As the state space
architecture within which the attractors are embedded is determined by the binding sites on proteins
and the recognition sites on DNA the attractors can be described as “hard-wired” in the genome through
the genomic DNA sequence. In the second model attractors arising from the interactions between activeReviewjournal homepage: www.elsevier .com/locate/pbiomolbioAll rights reserved.
, K., A comparison of two cell
cular Biology (2011), doi:10.1le at ScienceDirect
nd Molecular Biologyregulatorymodels entailing high dimensional attractors representing
016/j.pbiomolbio.2011.01.002

s anproblem in biology that is only partly solved. That a single genotype
can give rise to more than 200 terminally differentiated cell
phenotypes and several times that number of intermediate
phenotypes, in the human, is understood to be due to the selective
use of gene products but how that selection is achieved is far from
clear. Additionally, although these cellular phenotypes are discrete
and can generally only make unidirectional transitions spontane-
ously, that is, for example, from stem to differentiated cell, within
a specific lineage they can be reset to a pluripotent state by specific
manipulation including nuclear transfer (Takahashi et al., 2007). It is
also the case that the environmental conditions, diet, for example,
during some stages of development can influence phenotype not
only in the developing organism but also for future generations,
with implications for human health (Bateson et al., 2004; Gluckman
et al., 2009). In other words cells have a degree of plasticity not
reflected in the genotype. Furthermore, ionising radiation can
induce in the cells of a stable species a phenotypic transition towhat
is known as genomic instability (Kadhim et al., 1992), a phenotype
that appears to be novel, that is, not associated with processes such
as normal differentiation. A possibly related phenomenon has been
observed in genetically engineered bacterial cells, which can adopt
adaptive phenotypes in response to environmental stress
(Kashiwagi et al., 2006)without the benefit, because they have been
genetically modified, of a “genetic programme”.
The question addressed here is “what are the processes that can
account for these phenotypic properties?”; in essence, “how is the
cell regulated?” and the starting point in answering that question is
to understand what exactly is the nature of cellular phenotype. The
commonly accepted answer, based on the Central Dogma of
molecular biology which postulates information transfer only in
one direction, from DNA to protein, proposes that the sum of the
properties of these deterministically translated proteins constitutes
the cellular phenotype. This proposition raises many questions
outside of genetics, such as how the sequences to be translated are
specified and, of course, the problem, still unsolved, of predicting
a specific folded protein from its peptide amino acid sequence. In
2001, with the completion of the sequencing of the human genome,
it became clear that that deterministic aspect of the Central Dogma
was even more problematic. The number of gene coding sequences
in the genotype was substantially less than the number of func-
tional products they could produce (Carninci, 2008). This further
underlines the importance of non-genetic processes vital to the
translation of genotype to phenotype.
The realisation that the extent of chromatin marking in
eukaryotic cells far exceeded that required for the permanent
imprinting of alleles suggested to some that regulation of the
second by second transcription of coding sequences, based on such
marking, might be the primary mechanism by which phenotypic
expression was regulated (Jaenisch and Bird, 2003). However,
according to Huang (Huang, 2009) chromatin marking lacks the
necessary stability and locus specificity necessary for it to have
a regulatory role in gene expression. I draw attention to additional
short-comings below.
In 2000 the idea that phenotype could be represented by a self-
organised high dimensional attractor state was proposed indepen-
dently in two publications (Baverstock, 2000; Huang and Ingber,
2000). The idea that attractors might have an important role in
biology was not new. Max Delbrück, in his intervention in
a discussion on a paper by Sonneborn given at a genetics conference
in Paris in 1949, was probably the first to express the concept.
Sonneborn had attributed a particular phenomenon to the repro-
duction of genes that were either favoured or inhibited by envi-
ronmental factors. Delbrück noted that “many systems in flux
equilibriumare capable of several equilibria under identical conditions.
K. Baverstock / Progress in Biophysic2They pass from one stable [i.e. ordered] state to another under the
Please cite this article in press as: Baverstock, K., A comparison of two cell
phenotype, Progress in Biophysics and Molecular Biology (2011), doi:10.1influence of transient perturbations” (Delbruck, 1949). However, well
before that Darwin, in Chapter III of the Origin of Species, had in
effect articulated the same principle in terms of a stable ecology
(Darwin,1859). The subject prior to 2000 is reviewed by Emlen at al
(Emlen et al., 1998). Huang and I proposed that phenotype could be
represented by a high dimensional attractor in order to explain very
specific features of cell biology. Huang et al were concerned to
understand how the fates of neighbouring cells in the developing
embryo were determined. Neighbouring cells, possibly in contact,
may have very different fates, for example, apoptosis, differentiation
or proliferation. It was suggested that these different fates were
determined by environmental influences on the cell, for example,
soluble growth factors in the extra-cellular matrix, which caused
transitions between self-organised attractors in the regulatory
network. My argument was based on an attempt to explain the
phenomenon of radiation induced genomic instability.
Genomic instability was uncovered by Munira Kadhim and
colleagues at the MRC Radiobiology Unit at Harwell (Kadhim et al.,
1992). Explanted bone marrow cells were subjected to low alpha-
particle fluences (w1 passage per cell on average) and the survivors
plated out singly and grown as clones. Subsequent karyotypic
analysis revealed, within a single clone, chromosome aberrations in
some cells while others exhibited no damage. As it is expected, on
the basis of the prevailing radiobiological dogma, that after the first
cell division following irradiation any molecular damage will be
replicated in all future generations the only conclusion to be drawn
was that damage that was expressed in the later cell divisions was
in some way “hidden” in the earlier divisions. The term genomic
instability was coined to describe this non-clonal generation of
molecular damage. The question I addressed was “what is the
inheritance mechanism and the source of the latent molecular
damage in the cell progeny?”
Both the above, essentially proposals for the regulation of the
cell, have been further developed over the past decade, again
independently, and two models (Baverstock and Rönkkö, 2008;
Huang, 2009) are now available to compare in the light of the
evidence derived from experimental cell and molecular biology
research accruing over that period.
2. Materials and methods
2.1. The models
It is helpful to describe Huang’s model first as it is based on the
familiar original ideas of Monod and Jacob in 1961 (Monod and
Jacob, 1961), namely the concept of the genetic regulatory
network (GRN) (Babu et al., 2004). This model is referred to here as
the “GRN model”. The GRN is seen as “orchestrating” or regulating
the process of transcription to produce a profile of active gene
products, mostly proteins, which can be presented as an attractor.
The attractors and other features of the GRN architecture are
contingent on the structure of proteins and the target DNA
sequences and are therefore “hard-wired” in the genome.
Knowledge of the interactions encoded in the GRN can be
derived from experimental data on the transcriptome and the
dynamics of the network can be described in terms of ordinary
differential equations (ODE), also experimentally derivable. In this
way the stable (attractor) states of the system are defined as are the
potential transitions between attractors; some, as in the case of
differentiation, inherent and others, for example, transitions
between lineages, excluded by dynamic gradients and barriers in
the state space architecture. Influences external to the cell can
cause the attractor tomake transitions to allowed states in an “all or
nothing” manner if the perturbation is sufficiently “strong” to
d Molecular Biology xxx (2011) 1e7overcome the basin of attraction surrounding the attractor.
regulatorymodels entailing high dimensional attractors representing
016/j.pbiomolbio.2011.01.002