192s Biochemical SocietyTransactions ( 1 994) 22
Metabolic gradients as the source of differentiation and morphogenesis
YORAM SCHIFFMANN
Department of Applied Mathematics and Theoretical Physics, University of Cambridge, Silver Street, Cambridge CB3 9EW, U. K.
The notion that quantitative differences in the household (glycogen) metabolism are the driving force for the creation of the
qualitative differences between cell types and for morphogenesis
may appear strange. Indeed, differentiation involves the creation
of order via low-entropy states characterised by a nonuniform distribution of luxury proteins. The heterogeneity provides for 'division of labour' in a true Spencerian spirit. In contrast, the household energy metabolism seems to be the antithesis of
specialisation, and its concomitant functional and structural complexity. This energy metabolism is in fact common to all cells.
Also, why should a spatially-differential metabolism lead to a differential gene expression? Why would a quantitative spatial difference in glycogen depletion be used for differentiation and not the
more 'qualitative' difference between, for example, glycogen metabolism and amino acid metabolism? Why should differential metabolism result in morphogenesis, e.g. in a deformation and movement of an ephithelial sheet? If differentiation and morphogenesis
are based on differential energy metabolism, is it not possible for
disruptive interference with other energy-requiring processes to
occur? Is it compatible with the requirement of ATP homeostasis?
Then there is the question of why differences in the intensity of
!he same metabolism should arise in the first place. What is the
primary cause' - localisation - of nonuniform metabolism? And
can the nonuniformity not arise by phase transition?
On the other hand, perhaps the centrality of energy metabolism
in development is to be anticipated.Already the Spencerian notion
of development, and evolution in general, as a transition from 'incoherent homogeneity' to 'coherent heterogeneity' conjures up the
image of 'active transport' which implies the activation of the energy metabolism. Furthermore, if all biochemical processes are to be
correlated in space and time, and all throughout development, we
need a system - and the energy metabolism is such a system that operates throughout development, and it is also necessary that
the correlating molecules are not only able to be distributed nonuniformly, but are also ubiquitous and universal and able to affect
all biochemical processes. cAMP and ATP come to mind as appropriate candidates which additionally can diffuse in the tissue,
even after cellularisation, through gap junctions. We also note that
we cannot obtain the ordered lowentropy states - the nonuniformities - by lowering the temperature or increasing the pressure,
i.e. by phase transitions.The biological system is an isothermal
system that presents not only localised molecules as in phase transitions, but also localised biochemical processes. Thus the spatial
order we look for is unlike an equilibrium structure such as a crystal but involves simultaneous localisation and coherence of both
structure and process. The two are inter-dependent. Nonuniformity in process and structure is a reflection of the struggle
of life against the universal drive to molecular chaos, randomness,
in short against the tendency of entropy to increase. It thus makes
sense to anticipate that the success of this struggle will depend on
a biochemistry tailored to minimise energy dissipation. It indeed
turns out that the condition for the spontaneous appearance of localised structure and process upon the activation of glycogen catabolism is the requirement for energy efficiency. The cause of
the above activation is an homogeneous extracellular signal; we
have self-organisation manifesting symmetry-breaking that can explain epigenesis. The metabolic gradient affects all types of endogenous proteins via the action of reduction, phosphorylation and
non-phosphorylative ATP-hydrolysis, and can explain various localisation phenomena such as the localised assembly of the cytoskeleton and localised electric potential; both localisations may be
responsible for the localisation of morphogenetic determinan:s
such as maternal RNAs.
Recall that polymerisation of the cytoskeleton is a dynamic,
energy-requiring process involving both ATP-based polymerisation and depolymerisation; therefore the possibility for locally increased synthesis of ATP, and locally increased (CAMP, ATP)dependent phosphorylation, is precisely what is needed for the dynamic localised assembly of the cytoskeleton. The metabolic field
can also determine the direction of the polymerisation and thus
the direction of the filament; such a directionality is an antientropic effect.
Recalling that glycolysis involves dehydrogenation but respiration involves the separation of hydrogen atoms into protons and
electrons, we can argue that spatially-differential respiration is responsible not only for a spatiallydifferential reduction potential,
but it can also be the basis for the explanation of various bioelectric phenomena. Thus, the theory explains why positive currents
around oocytes or embryos enter the sites of high metabolism
since these are also the sites, according to our theory and existing
experiments, of high electronegativity. We also suggest that it is
the spatial nonuniformity in the intensity of electron-transfer that
is responsible for the gradients in the elecmc potential - again an
anti-entropic, endergonic effect, which in turn can explain selfelectrophoresis of endogenous charged proteins. The explanation
of bioelectrical phenomena on the basis of symmetry-breaking instability in a (CAMP, ATP) Turing system replaces the usual explanation based on a nonuniformity of pumps and leaks in the plasma membrane. A spatially-nonuniform pattern of activation of
pumps and channels is provided by the metabolic field and its associated phosphorylation field. Since according to our theory, oxygen is required for the creation of a reduction field and an electric
potential gradient, both of which are important for differentiation,
the 'Cambrian explosion' in biological complexity can be explained by invoking the fact that atmospheric oxygen became
available during this period.
A succession of metabolic patterns with an increasing number
of nodes is predicted by the theory and indeed confirmed by experiments. The sites of high metabolism lead the development and
also correspond to the 'organisers' and 'differentiation centres' in
the classical literature. Thus the high-point of the graded nuclear
uptake of the dorsal protein in Drosophilu is in the ventral side.
And this site is also the high point of metabolism for a number of
insects. Furthermore, the activation of the dorsal complex does indeed involve its reduction and its (cAMP,ATP)-dependent phosphorylation. The localisation of maternal cytoplasmic determinants
in Drosophilu and other organisms involves true 'symmetrybreaking' because of the absence of a systematic extracellular differential and because both the receptor and its ligand remain homogeneous. It is the differential metabolism that dictates the differential mechanical and elastic properties, e.g. the ventral furrow
associated with the activation of the dorsal, and in general it dictates the invagination involved in gastrulation. Thus, the differential metabolism is responsible for both differentiation and morphogenesis.
One conclusion from the nonlinear bifurcation analysis of reaction-diffusion systems performed by the author in the early seventies was that it is largely true that the shape of the Turing patterns, such as the metabolic patterns, depends on the geometry of
the system and not on the chemical mechanism assumed. This is
in some sense a fulfilment of the goal of rational morphology,
which is to find universal laws of form. But at the heart of our explanation of epigenesis - self-organisation and symmetrybreaking - is the source of the instability of the homogeneous
state. For the last few decades an intensive search for the real-life
biochemical Turing-instability mechanism has been raging. The
failure to identify the Turing morphogens was largely responsible
for the appearance of other 'symmetry-breaking' theories, where
the all-important instability is based on mechanical or elastic or
electrical properties, and not on a chemical-genetic mechanism.
As a result of this, some scientists came to believe that there is no
genetic program for development, and that natural selection plays
no major role in development. By contrast, we claim that cAMP
and ATP are the Turing morphogens responsible for the universal
instability and the resulting differential metabolism. Furthermore,
this differential metabolism is a low-entropy dissipative structure
of such a nature that it can couple and correlate all biochemical
processes. Both the instability-of-the-homogeneous condition, and
the coupling of the resulting morphogenetic field to all biochemical processes, are dictated in an unambiguous manner by the base
sequence of the DNA. Hence there is a genetic program for development and it is chemically a highly improbable and specific one,
and natural selection is paramount in the evolution of this program. In our quest to understand how the specific one-dimensional
order in the DNA is responsible for an equilibrium structure such
as a virus or a ribosome, we invoke a principle such as 'selfassembly'. Similarly, in our quest to understand how the specific
one-dimensional order in the DNA is responsible for the threedimensional nonequilibrium structure-process that is the developing organism, we invoke the principle of the instability of the homogeneous and the coupling of the resulting field to all biochemical processes. For further details, see a forthcoming paper.