麻豆传媒

IMAGINE a device no bigger than a credit card that could extract your DNA
from a drop of blood and map your entire genetic code while you wait. In an
instant, you鈥檇 have an estimate of your risk for developing cancer or diabetes,
perhaps, or find out how your cells were responding to drug therapy.

A miracle device of the 22nd century? Hardly. Biologists and engineers will
have one working in just a few years, because the tool that makes it possible, a
genetic microarray known as the 鈥淒NA chip鈥�, already exists. Able to scrutinise
tens of thousands of genes at once, the DNA chip鈥檚 astonishing abilities are
astounding biologists.

鈥淯sing chips, people have in one afternoon confirmed work that took several
years using conventional gene-sequencing processes,鈥� says Ed Hurwitz of
Affymetrix, the Silicon Valley company that pioneered the technology. And the
chips aren鈥檛 just about increased speed. Using them, researchers can do things
that were previously almost impossible, such as uncovering the genetic
machinations behind the complex biochemistry of organisms. With a yeast cell,
for example, they can represent virtually all of its 6200 genes on just four
chips. It鈥檚 then possible to take 鈥渟napshots鈥� that reveal which genes are
active, which are dormant and how these patterns change during the organism鈥檚
life cycle.

Today yeast, tomorrow us鈥攆or the chips should soon be able to handle
all 100 000 human genes. 鈥淭hey should radically change our ability to discover
drugs, infectious processes and even disease processes that we didn鈥檛 know about
before,鈥� says Richard Young of the Whitehead Institute for Biomedical Research
at MIT. What silicon chips did for computers, DNA chips may do for biological
research. 鈥淪ystems biology will be the challenge of the 21st century,鈥� says
Leroy Hood of the University of Washington Medical School in Seattle. 鈥淎nd the
only way to understand the biology of systems will be to use tools such as these
肠丑颈辫蝉.鈥�

The similarities between silicon chips and their gene-oriented counterparts
begin with the way they鈥檙e made. Like computer chips, most DNA versions are
produced by computer-controlled microprinting. In computer chips, the process
lays down microscopic circuits and switches; in DNA chips, it puts down the
stuff of genes.

Chips and chains

To make the chips, manufacturers first coat a surface such as glass or silica
with a sticky chemical. Then, at precise locations, they attach nucleotides that
include the four 鈥渂ases鈥� that lie behind the unique properties of DNA: adenine
(A), cytosine (C), guanine (G), and thymine (T). These are linked to form short
chains with different sequences. Some chip makers build the chains first and
then stick them to the surface, others build them onsite by adding nucleotides
one by one. When finished, the microarray is dotted with anything from a few
dozen to tens of thousands of short strands of nucleotides
(see Diagram, p 49).

This repertoire of sequences lets the chip do its work鈥攕earching out
specific sequences from an organism鈥檚 genome. The method works because DNA
consists of two strands along which A on one strand always binds with T on the
other, and C on one strand always binds with G on the other. This gives the
double helix a 鈥渃omplementary鈥� nature: if one strand reads ATCGTT, the other
reads TAGCAA. So each single strand on a DNA chip is the lonely partner of a
unique complementary strand鈥攁nd stands ready to bond with it should that
strand happen by.

Fortunately, to pick out a gene, the strings on the DNA chip don鈥檛 have to be
terribly long. A typical gene may contain 10 000 pairs of nucleotides, or more.
But within a gene鈥檚 chain there鈥檚 usually a short sequence, no more than 25
bases long, unique to that gene. By encoding that short, unique chain on a chip,
geneticists can represent the whole gene. This shorthand makes the entire
process manageable.

To use the chip, a physician or lab technician extracts a sample of an
organism鈥檚 DNA鈥攜ours, for example鈥攆rom a bit of blood or tissue. The
DNA is purified, replicated, split into single strands, and finally cut up by
enzymes into small pieces. Each piece in this DNA hash is then tagged with a
fluorescent molecule. At the moment lab technicians do this work, but all these
processes could soon be carried out in computer-controlled reaction chambers
squeezed onto the same chip as the microarrays (see 鈥淲ater gets weird鈥�, New
Scientist, 11 July, p 28).

This mixture is then washed over the chip. If a DNA fragment in the hash
meets a complementary counterpart on the chip, the matching sections zip
together to form a double strand. The better the match, the more bonds there
will be between the strands, and the stronger the join. Next, the array is
flushed with a chemical solution that breaks apart all but the best-matched
double strands. In theory, the pairs that remain are perfect fits. In practice,
there鈥檚 about a 5 per cent error rate. So a typical chip has built-in
redundancies鈥攁dditional test sites with the same unique DNA fragment, each
offering a second opinion on the results of the others.

When the flushing is complete, a computer reads the location of any
fluorescent tags shining from the chip鈥檚 surface and matches those locations to
its record of the nucleotide chains deposited at those points. The
result鈥攁 catalogue, warts and all, of your genetic ingredients.

Disease specific

Chips can also specialise in specific maladies, replicating the handful of
specific genes implicated in a particular illness from arthritis to HIV. For
example, Larry Brody and Joseph Hacia at the National Institutes of Health near
Washington DC are studying the human BRCA1 and BRCA2 genes,
which are implicated in hereditary breast cancer. When breast cancer runs in a
family, as many as 80 per cent of the women affected show mutations in either or
both of these genes. Those mutations make a woman鈥檚 chance of developing the
disease as high as 85 per cent.

The two scientists are testing chips that lay out the complete sequence of
鈥渉ealthy鈥� BRCA1 and BRCA2 genes鈥�5000 and 10 000 base
pairs, respectively. The researchers flood a chip with a sample of BRCA
genes from a woman whose family has a history of the disease. If her genetic
sample matches point by point鈥攖hat is, if complementary sequences bind to
all the strings on the microarray鈥攕he has normal BRCA genes and a
clean bill of health. If there are mismatches鈥攐ne or more strings don鈥檛
find complementary partners鈥攕he鈥檚 at risk.

鈥淲ithout the chips, it鈥檚 relatively easy to screen for a mutation when you
know what to look for,鈥� Brody explains. 鈥淏ut BRCA1 has more than 500
known mutations and more are being found all the time. When every third patient
has a variation that you鈥檝e never seen, you can鈥檛 build a conventional test.
That鈥檚 the beauty of these things.鈥�

Louis Staudt, a researcher at the National Cancer Institute near Washington
DC, appreciates it, too. 鈥淲hen you have to use more laborious methods, you tend
to study those locations where you expect that there might be an important
variation,鈥� he says. 鈥淲ith chips, you can use this large amount of data to tell
you what鈥檚 unusual instead of looking only at what you expect to be
颈苍迟别谤别蝉迟颈苍驳.鈥�

This capability should also help to tailor medicines to the specific genetic character of any one individual
(see 鈥淛ust for you鈥�).
In a few years, the human genome project will be complete, and biologists will have a rough idea of
the average genetic make-up of a human being. But specific variations from the
norm are often crucial in determining a person鈥檚 susceptibility to disease or
their response to drugs. Chips will enable doctors to take a quick snapshot of a
person鈥檚 genes to see which treatment is best for them鈥攁nd to avoid drugs
that might be dangerous to them.

The chips will also be a powerful tool for basic research鈥攆or
understanding how the pattern of gene expression in different kinds of cells
leads to their unique characteristics, for instance. A gene expresses itself
when it acts as a template to make its own distinctive protein. The strength of
a gene鈥檚 expression depends on how much of that protein it causes to be made.
Having an easy way to measure the strength of a gene鈥檚 expression would be
extremely useful to biologists.

But a gene doesn鈥檛 make its proteins directly. Instead, the cell extracts the
information needed to make a protein and dispatches it to protein-making areas
in the form of RNA, a single-stranded molecule very similar to DNA. The more RNA
blueprints created, the more protein manufactured. So by measuring the amount of
RNA in a sample, researchers can easily figure out the amounts of protein being
produced.

How active are your genes?

To measure gene expression on a chip, designers again arrange snippets of DNA
on the wafer that match various genes. This time, however, each gene is
represented by thousands of identical strands. Researchers then isolate the RNA
from a patient鈥檚 sample, chop it up, tag it with a fluorescent flag, and wash it
over the chip. Each RNA fragment binds to the strand of the gene from which it
was created. So if a gene is strongly expressed, RNA fragments might bind to
most of the strands representing that gene on the array, whereas those of a
weakly expressed gene will gather only a few bits of RNA. In this way,
biologists can calculate the proportions of each version of RNA
represented鈥攁nd learn how strongly each gene is expressed.

鈥淲e can, for example, use gene expression to study what the body is doing to
fight a bacterial infection鈥攐r what the bacteria are doing to survive in
the host,鈥� says John Keller of SmithKline Beecham in Philadelphia. Researchers
there grow a strain of bacteria in a Petri dish and gauge the expression
strengths of its genes. Then they inject the bacteria into a living host and see
which genes are turned up or down in both organisms.

鈥淭he chip lets you read those genetic patterns in the normal and infected
states 100 to 1000 times faster than previously,鈥� Keller says. 鈥淭hat gives you a
hint of which genes are in play during the infection process and lets you know
which genes in the bacteria you want to attack to inhibit its growth. It also
can show you which of the body鈥檚 responses are strongest and, therefore, which
you might want to find ways to support.鈥�

Being able to read expression patterns also enables scientists to distinguish
between people with apparently identical diseases. Staudt is developing genetic
profiles of cancer cells from a group of people with lymphoma who all received
an identical diagnosis. 鈥淲e know that patients vary considerably in their
response to treatment,鈥� he says. 鈥淥ne possible reason is that the diagnostic
category that pathologists have assigned to this disease actually lumps together
more than one illness.

Studying the expression patterns of tens of thousands of genes can
distinguish such illnesses, by revealing subgroups of patients with different
patterns of gene expression. 鈥淓ssentially,鈥� says Staudt, 鈥渨e鈥檙e looking for
diseases within diseases.鈥�

These tiny differences can be a matter of life and death, for each subgroup
may respond differently to the same treatment regime. By understanding the
genetic behaviour of a specific form of a disease and its reaction to different
treatments, therapies can be more carefully designed for each individual
patient. 鈥淲e鈥檙e finding that tumours sort into different classes depending on
their gene expression patterns,鈥� agrees Stephen Friend of Rosetta Inpharmatics
in Kirkland, Washington. 鈥淧eople are now using these expression patterns to
determine if patients are at high risk and need quick and radical treatment, or
will respond to less drastic therapy over a longer period.鈥�

Fine-tuning

Chips could be used to fine-tune drug-design as well as for medical
diagnosis. Drugs companies test a huge number of potential drugs against a
spectrum of biochemicals to determine their interactions. They also try
thousands of variations of any promising compound to see which ones maximise
benefits and minimise side effects. Chips can speed these screening processes by
years.

鈥淟et鈥檚 say you take some tissue from the knee of a person with arthritis,鈥�
says Hurwitz at Affymetrix. 鈥淵ou read the pattern of gene expression in that
tissue and compare it with the expression pattern of genes in a non-arthritic
knee. Suddenly you see that, say, a dozen specific genes are more strongly
expressed in the arthritic knee. You could then concentrate on developing drugs
that would keep those genes from expressing themselves so strongly.鈥�

And this approach isn鈥檛 limited to humans. Agricultural biologists are using
the devices to read plants鈥� designs as well. Shauna Somerville, a plant
biologist at Stanford University, is part of a team using the chips to explore
the detailed architecture of thale cress (Arabidopsis thaliana), a weed
related to broccoli that has had more of its genes sequenced than any other
plant.

鈥淥ur early experiments compared gene expression in plants infected with a
particular disease versus uninfected plants, and then in susceptible plants
versus resistant plants,鈥� she says. 鈥淲e knew which genes should be involved, so
we wanted to make sure the chips showed us what we expected.鈥� The tests gave
them more than confidence in the technology. 鈥淲e have, in fact, identified new
markers or new genes correlated with infection that we wouldn鈥檛 have anticipated
based on the literature,鈥� she reports.

Stress busters

Botanists hope to use the chips to understand how plants cope with the stress
caused by drought, heat, excess rain or the lack of certain nutrients. By
charting genetic responses to specific stresses, they may be able to engineer
more versatile plants that thrive in more varied environments. Researchers also
hope to use the chips to read the genetic signatures of the changes in plants
bred for specific advantages鈥攄rought tolerance or cold hardiness, for
example鈥攁nd then engineer the same changes in other strains.

DNA chips may even help to turn plants into chemical factories. 鈥淭here has
been some work on producing biodegradable plastic from plants,鈥� Somerville
notes. 鈥淏ut we don鈥檛 know the impact on the plant of that kind of
engineering鈥� diverting so much carbon to a novel product that the plant
has never seen before. Microarrays will help us understand how a plant鈥檚
metabolism copes with that鈥攁nd whether that kind of engineering will ever
be viable.鈥�

Ultimately, chips will give Somerville and her colleagues a much better
understanding of the processes going on inside plant cells. 鈥淭hat should allow
us to make better choices in the modifications we make to plants in developing
hardier and more productive strains.鈥�

It shouldn鈥檛 take long. As demand grows for DNA chips and the computers that
read the results, the cost of both should plummet. Chips now cost as much as
$2000 each to make, and some of the equipment needed to make and read
them is 10 times as expensive. Within a few years, however, the price of many
chips could be low enough to rival that of disposable needles for syringes. And
as they become widely available, geneticists will be steadily analysing the
genes of everything from bacteria to your Aunt May.

鈥淚n three to five years, virtually every scientist will have access to these
chips in an economic way,鈥� says Leroy Hood. 鈥淭hat will revolutionise every
aspect of biology.鈥�