Biology Reference
In-Depth Information
with 'curly arrows'. For many molecular biologists, the current measure of understanding is
an ability to summarize a process in blob-and-arrow diagrams in which signalling pathways,
and so on, are shown as chains of interaction and causation. The best-studied morphogenetic
mechanisms can already be summarized in this way (and some have been earlier in this
topic), and an intensification of research into developmental cell biology promises that
more will join their ranks soon. Is that level of understanding enough?
The problem with blob-and-arrow diagrams, useful as they are, is that they present a one-
dimensional view of what is a three-dimensional process (four-dimensional, if time is
included). They provide a useful summary of key processes, but have almost no predictive
power except in the crude sense of predicting what knockouts would cause morphogenesis
to fail. A model would have a much more convincing claim to be an 'understanding' if it had
the power to predict the actual movements or shapes that would result from the action of
a morphogenetic system. This involves quantitative modelling (Chapter 26), and quantitative
modelling that is more than guesswork requires at least some quantitative measurements to
be made from the system under scrutiny. Fortunately, the nature of the feedback loops that
give self-organizing properties to morphogenetic mechanisms means that when modelling
a system with 30 different types of molecules, it is probably not necessary to measure the
binding and diffusion constants of every molecule; feedback controls the system in a way
that is almost independent (within limits) of the precise quantitative characteristics of indi-
vidual components. Useful predictive modelling of morphogenesis ought to be achievable
using only the characteristics of higher level integrons (although a 'complete understanding'
will require more detailed information).
Quantitative modelling
d
which nowadays means computer modelling
d
of the morpho-
genetic processes in real embryos promises to be one of the most powerful means of under-
standing morphogenesis that we will have in the near future. It does, however, require a great
deal of work if it is to be done properly, which means using mechanisms that have been iden-
tified by real molecular cell biology and using key quantitative parameters that have been
measured from life. Guessing mechanisms and parameters, and then showing that a life-
like shape results on the computer screen, is of very limited use unless it leads directly to
the design of real experiments. This is because it is too easy to tweak a large set of parameters
until the right shape happens to result. As Von Neumann is alleged (be Fermi) to have said
'
Give me four parameters and I can fit an elephant. Give me five and I can make it wave its trunk
'.
One very important aspect of practical research into morphogenesis, indeed development
in general, is to identify components that control or regulate particular activities. This is
partly for more detailed knowledge but mainly because finding key regulators makes model-
ling and high-level understanding easier. Metabolic biochemists, for example, have a much
better predictive understanding of the dynamics of complex, linked pathways when they
have identified the rate-limiting steps (which is where control is exerted in this context). It
is therefore unfortunate that the word 'regulator' is so often misused in modern biology.
In particular, it has become fashionable for researchers who have shown that knocking out
a gene abolishes an event to claim that they have identified a 'regulator' of that event (often
a 'key regulator', in the hyperbole of our times). This is based on completely flawed logic that
confuses regulation with mere necessity.
26
The fact that something is required does not make
it a regulator: the propeller shaft and gears of an automobile are required for it to move, yet
neither of these is a regulator of its motion in the sense that the accelerator pedal is. Gene
