Protein Society Meeting -- Reductionism in Biology

Terry M. Gray (
Wed, 29 Jul 1998 00:00:29 -0700

To the group(s),

I'm at the annual meeting of the Protein Society, a gathering of around
2000 researchers from academia and industry including many post-doctoral
fellows and graduate students. I hope to write a more extensive report
with some opinion on items relevant to our discussions on these lists, but
for now I wanted to pound this out while the thoughts were fresh on my

Greg Petsko, a chemist/protein crystallographer from Brandeis University
(nearing his 50th birthday, as he noted), was the award banquet after
dinner speaker. His title was "Structural Biology in the Name of
Genomics." One of the themes of the meeting this year was how to deal with
the vast amount of DNA sequence information (and hence protein sequence
information) coming out of the various genome projects with complete
genomes now coming out at the rate of one every six months. Over and over
I heard at this meeting that we need to develop methods for determining
structures of these protein either through more rapid or strategic
experimental structure determination or by homology modeling--determining
the structure of a protein whose amino acid sequence is similar to another
sequence whose structure is already know--or other computational structure
determination methods. The implicit assumption, sometimes even stated
explicitly, is that if we know the structures of all the proteins encoded
by genomes of organisms that we understand the biology of the organism.

It's obvious that the meeting was populated by a bunch of biochemists,
biophysicists, chemists, and physicists who have no sense of things

Don't get me wrong. I'm a structural biologist myself and am vastly
interested in the structure of biological macromolecules. I've also used
the genomics argument in the significance part of research proposals that I
have written concerning my own research in protein folding and the
importance of cracking the folding problem. But someone needs to remind us
biochemists and biophysicists that living things are complex systems of
molecules interacting in complex ways and that these complex interactions
don't simply drop out of a structure determination.

In comes Greg Petsko, one of the patriarchs of structural biology with a
fine antidotal (and anecdotal) lecture. He had several messages--one that
as more protein structures come and that as the structure determination
techniques become more routine that the very difficult task of protein
structure determination will soon go the way of small molecule
crystallography. Nowadays small molecule crystal structure determination
is in the realm of the department support staff crystallographer--a central
service that everyone utilizes as part of their routine chemical analysis
of their favorite molecule that they are studying for other reasons. People
seldom regard small molecule crystallography as a worthy research program
in its own right. Protein crystallography is soon going that way--merely a
technique to get at structural information that is only part of the story
that the biologist is interested in. Incidentally, at National Institutes
of Health grant workshop that I attended, it was announced that there was a
new program for supplemental grants for structure determination of the
protein that you are working on--intended to encourage
non-crystallographers to pursue structure determination of their protein as
a rather routine part of their biochemical/biological analysis.

Petsko's main message had to do with the relationship between structure and
function. He reminded us of dictum by Francis Crick from the 1950's.
"Study function, but if you can't study function, study structure." Petsko
noted that determining the sequence of the genome tells us nothing about
function (of course, if we see that the DNA or protein sequence is similar
to another sequence where we do know the structure, we may be able to make
an educated guess. He noted that structure really tells us nothing about
function either (at least with current methods). There are experimental and
computational methods that may tell us where binding sites are and know
what binds to our protein may give us a hint about its structure. This is
a step in the right direction. He pointed out that TIM barrels (a fairly
common protein fold) is the architecture for several different enzyme
mechanisms. Knowing the fold only leads us to a set of several and often
very diverse possibilities. Then he made the outrageous claim that knowing
the function tells us little or nothing about the function! The point was
that many proteins have multiple functions and that knowing one function
still leaves us in the dark relative to the other functions. Crick's
dictum is no longer good enough, Petsko said. Study function, but if you
can't study function ... well ... don't study. He then urged us not simply
to train structural biologists, but structural biologists who are
interested in function and who can and will use all the repertoire of the
methods of biology to understand function.

While Petsko didn't explicitly say this, it is a natural consequence of his
argument. Biology can't be reduced to chemistry and physics. Listen
carefully here--I'm not suggesting a return to vitalism nor am I suggesting
that chemistry and physics don't have a crucial role to play in
understanding things biological or that the substratum for living systems
is anything but physical-chemical entities. But, aren't we really
interested in understanding biology--how living things work. And as I said
earlier, living things are complex systems of molecules interacting in
complex ways. Dissecting these complex interactions necessarily involves
doing genetics, cell biology, developmental biology, physiology, and
ecology. Interestingly, it was the genome sequencers and molecular
modelers who most explicitly spoke the reductionistic language. To give
credit where it is due, the protein crystallographers always put their
structures into the biological context: their role in gene expression,
development, metabolism, physiology and reported their results in the
broader organismic framework showing how their structure provided molecular
detail for results from the more explicilty biological fields of genetics,
physiology, and developmental biology.

This discussion is not new. Many biologists and philosophers of biology
(Ernst Mayr, Stephen Rose, Steve J. Gould inter alia) have made these point
before. But as we're on the verge of seeing whole genomes of organisms
coming out at an unprecedent pace (the worm, C. elegans, within a year,
Drosophila within two years, and the human genome within 5 years), the
point needs to be made anew. Organisms aren't their genomes or even their
genomes with their gene products. No. They are complex systems these
structures and the information are interacting in complicated spatial and
temporal ways. It is those very complicated spatial and temporal
interactions that make organisms unique. Organisms are wholes where the
whole is more than the sum of the parts.

I used to start my biochemistry course by chopping up a whole marigold
plant in a Waring blender. This is often the first step in our biochemical
analysis. (Thankfully, I have done most of my research over the years with
bacteria instead of cows.) As much as we might gain by cataloging all the
molecules in the marigold, surely we all recognize that something was lost
in the process of chopping it up. (Nervous excitement crept over my
classes when I had a goldfish in a beaker or laboratory mouse in a cage up
on the front desk next to the blender. I suppose the mouse and fish were
nervous too... Never fear ... I never actually did it.) Interestingly, at
the same NIH workshop mentioned above there was also a program announcement
calling for proposals on higher order systems. We need to understand
better how all the parts interact with each other spatially and temporally.
We need to put the marigold back together!

Terry M. Gray
July 28, 1998

Terry M. Gray, Ph.D., Computer Support Scientist
Chemistry Department, Colorado State University
Fort Collins, Colorado 80523
phone: 970-491-7003 fax: 970-491-1801