杏吧原创

THE text would fill a thousand fat telephone directories. It is written in a
cryptic alphabet of just four letters, in a barely understood language. It is
subtle and self-referential and filled with allusions to other works, and it has
kept an army of scholars working for years in the hope of being able to read it
through just once.

This nightmarish document is of course the human genetic code, carried in our
DNA. Government labs and their private-sector competitors say they鈥檒l have rough
drafts of the human genome sometime next year, and more precise sequences a few
years after that. And the first sequence of a whole chromosome (one of the 23
pairs of molecules of DNA in each of our cells) will be published in
Nature next month.

But sequencing the genome once will only be a start. Tiny differences in
genes from person to person cause thousands of hereditary diseases, and
influence our risk of developing killers like cancer and heart disease. So the
payoff will come when we understand how variations in genes lead to illnesses.
For that, we鈥檒l want to map the genetic code of thousands or even millions of
individuals鈥攊mpossible with today鈥檚 technology.

But one group of scientists have a radical technique that could work ten
thousand times as fast as present-day gene sequencers. It鈥檚 as simple as
dragging a strand of DNA through a hole. As well as saving lives, it could
provide a wealth of genetic information to solve some deep biological
mysteries.

DNA鈥檚 information is stored in its sequence of the four bases: adenine,
thymine, guanine and cytosine (A, T, G, C). Bases on the two backbones of the
DNA double helix pair up, adenine always with thymine and guanine always with
cytosine鈥攕o one strand of DNA always complements the other.

The fastest conventional sequencing methods rely on chopping a
single-stranded DNA molecule into small pieces at a known base. You can then
measure the lengths of the fragments by gel electrophoresis鈥攍onger pieces
diffuse slowly through a gel, shorter pieces more quickly. But there are several
other fiddly steps involved, and the best machines based on this process take
about a third of a second per base pair, or thirty years for all 3 billion.

The idea that could change all that appeared a few years ago. In the early
1990s, John Kasianowicz of the US National Institute of Standards and Technology
(NIST) had been experimenting with ion channels, the tiny pores in a cell鈥檚 wall
that regulate the flow of chemicals in and out of the cell. Kasianowicz wanted
to use ion channels as sensors for detecting very low concentrations of
chemicals.

Pulling through

Then in February 1994, Dan Branton of Harvard University and David Deamer of
the University of California at Santa Cruz came to ask him whether DNA could be
dragged through the channel. 鈥淚t was just an idea in our heads,鈥� says Branton,
鈥渁nd we weren鈥檛 set up to do that kind of work.鈥� But if the different bases
could somehow be distinguished as they went through, it would be a damned fast
DNA sequencing method.

They used an ion channel called alpha haemolysin, plucked from the bacterium
Staphylococcus aureus. First they stretched a cell membrane across a
hole separating two compartments containing a potassium ion solution. When they
added alpha haemolysin, the tiny protein structure inserted itself into the
membrane. Then they added small sections of single-stranded DNA to one side and
applied a voltage. To their delight, the voltage dragged the negatively charged
DNA through the channel.

These channels are tiny鈥攁bout a nanometre and a half in
diameter鈥攁nd yet the experimenters found they could measure the current of
potassium ions flowing through a single channel. Could that current be used to
read the genetic sequence? The bases on a single DNA strand should periodically
block up the opening, like a string of ping-pong balls being stuffed down the
plughole of an emptying bath. If the four bases all reduce the current to a
different degree, you ought to be able to read off the genetic code just by
measuring the current.

But it鈥檚 not easy. Bases are not ping-pong balls, and it is difficult to read
the tiny changes in current. At the moment, the molecular chain ploughs through
the hole much too fast, and only about 100 potassium ions make it through during
the passage of one base pair鈥攙ery close to the limits of detectability. So
the DNA needs to be slowed down, letting more ions through to produce a robust
signal.

To do that, the Harvard team are planning to use double-stranded DNA, but
tease just one strand of it through the hole. The DNA should unzip, popping the
base-pair bonds as one strand is sucked through the pore
(see Diagram),
and because energy is going into breaking apart the base pairs, the passage of
DNA through the pore should be slower. Another possibility is to increase the
signal current by using a flow of electrons transmitted by quantum tunnelling, a
process by which the electrons can hop from one object to another across an
intervening barrier.

The teams at Harvard, Santa Cruz and NIST reported some of their latest work
at the Biophysical Society meeting in Baltimore this February. So far they鈥檝e
spotted differences between the two classes of base, purines (adenine and
guanine) and pyrimidines (thymine and cytosine), as they pass through a
haemolysin channel. Synthetic polymers based on cytosine reduced the current by
95 per cent, while those based on adenine reduced it only by 85 per cent. So the
researchers can see the transition from one type to the other as a chain slides
through the channel. But they can鈥檛 do it with real DNA yet.

To make a real gene sequencer, one that unzips DNA by dragging it through a
hole, much tougher nanopores are needed. The haemolysin channels tend to get
damaged. Far better would be a nanopore made of some durable nonorganic
material.

So Branton teamed up with Jene Golovchenko of Harvard鈥檚 physics department.
They decided to try carving nanometre openings in silicon nitride, an ultra-hard
ceramic. But no one has made a hole just 2 nanometres across before. 鈥淐urrent
technology can reproducibly make holes 40 to 50 nanometres in diameter,鈥� remarks
Branton, 鈥渟o we needed a new method.鈥�

They have adapted an old trick used by machinists for making tiny holes. 鈥淚f
you want a 1-micron hole in a sheet of copper, you take a ball-peen hammer and a
centrepunch and give it a tap,鈥� he explains. 鈥淭his makes a small depression, and
then you turn it over and start sanding it down.鈥� Eventually you break through.
That way you can make a hole much smaller than any drill bit.

Golovchenko and Branton are developing similar techniques on the nanoscale
using tools such as ion beam etching. So far, the group has been able to fashion
4-nanometre pores, and they think they can get down to 2 nanometres鈥攖he
minimum requirement for a working nanopore sequencer.

Branton estimates that it will be another three years before he has a robust
sequencing instrument based on nanopores. If it works, the sequencing rates will
be off the map, perhaps as high as 1000 bases per second. If you could persuade
a lot of these nanopores to work in concert, say around 500 channels all
sequencing together in a single machine, then you鈥檇 be up to half a million per
second. All 3 billion base pairs of the human genome could be read in about a
day.

Lloyd Smith, a chemist at the University of Wisconsin, sounds a note of
caution. 鈥淚t is a very cool idea, totally orthogonal to other approaches out
there. But this is early stage, super-high-risk work,鈥� he says, warning that
鈥渢his [nanopore] sequencing rate may be a massive extrapolation from too little
诲补迟补.鈥�

Even assuming that nanopores live up to expectations, they鈥檒l probably arrive
too late to get the first human genome sequence. According to Jeff Schloss,
director of the technology development programme of the US National Human Genome
Research Institute, about 12 per cent of the genome has already been deposited
into public databases. 鈥淭he complete sequence is expected not later than the end
of 2003,鈥� he says. But that is hardly the end of the story.

Unweaving the genome

Reading out the genome is akin to decoding a message of 3 billion letters,
but not having a clue about what language it is written in. To understand what
genes mean, we need to compare the genetic message with the outward
characteristics of people and other organisms.

Subtle traits like appearance, intelligence and personality are almost
certainly affected by hundreds or thousands of different genes, rather than one
or a few, so the only way to get a handle on what genes influence these
characteristics will be to compare thousands of individuals鈥攑erhaps
something that only nanopore technology will be able to do.

Another puzzle is how humans have migrated around the planet over the
millennia. By comparing the genes of different populations, assuming that the
more similar they are the more recently the groups diverged, we can reconstruct
these human tides and answer some crucial questions about our past. Who first
settled the Americas? Did Homo sapiens ever interbreed with
Neanderthals? We know that we came out of Africa, but when and where, and was it
a single emigration or a more complicated process?

Nanopore sequencers could churn out genetic libraries for other species too.
鈥淚f the cost could come down, you could sequence most organisms of interest on
Earth,鈥� says Schloss. 鈥淚t would be comparative genomics taken to its limit.鈥� We
could construct a complete and reliable family tree for all species. We could
learn which genes make backbones, brains or warm-bloodedness, and which make us
subtly but significantly different from chimpanzees
(New 杏吧原创, 15 May, p 26).

And the medical benefits would be huge. For example, geneticists believe that
there are 3000 to 4000 hereditary diseases. Gene therapy could one day cure some
of them, but only when all the details of their genetic basis are known
(New 杏吧原创, 3 October 1998, p 24).

And if you could have your own genome scanned, doctors could prescribe drugs
tailored to your genetic quirks
(New 杏吧原创, 14 November 1998, p 32),
or tell you whether you were predisposed to heart disease, so you could
change your lifestyle to ward it off.

Bad reactions to prescription drugs alone kill tens of thousands of people
every year and lead to other complications. For example, a number of
anti-depression drugs can lead to a build-up of toxins in people who have defects in
a gene called CYP2D6. This is a member of the cytochrome P450
gene family, which is involved in the metabolism of around 20 per cent of
commonly proscribed drugs
(New 杏吧原创, 14 November 1998, p 32). In
cases like this, rapid gene sequencing should identify those at risk so they
could take a different drug.

Disease-causing bacteria could also find their secrets being exposed.
According to William Nierman of The Institute for Genomic Research in Rockville,
Maryland, such a speedy technique as nanopores 鈥渨ould allow the complete genome
sequencing of all human pathogens, revealing simultaneously all of the vaccine
candidates and therapeutic targets in each organism鈥�.

And germs couldn鈥檛 even escape this analysis by mutating. 鈥淲e are studying
bacteria in our lab that kill patients over a long time,鈥� says Maynard Olson, a
medical geneticist at the University of Washington. 鈥淎nd it鈥檚 clear that the
bacteria that start the infection are different from those that kill the
patient. They evolve over time and we are severely limited in our ability to
sequence these changes.鈥� HIV and other retroviruses also evolve quickly, but
they would all be outpaced by a lab using a fast nanopore sequencer.

For most medical applications, however, nanopores aren鈥檛 the only fancy new
technology. Stan Fields of the University of Washington in Seattle says that DNA
chips have a substantial head start. These are arrays of DNA segments which
light up when the complementary DNA segments comes along to bind to them. To
build these chips, the human genome must first be sequenced many times in
different people to identify the crucial points of variation between them. But
once that is done the chip could identify an individual鈥檚 quirks in minutes
(New 杏吧原创, 14 November 1998, p 46).

Another contender is 鈥渓ab on a chip鈥� technology, in which the biochemical
tools of DNA sequencing are reduced to a substrate about the size of a credit
card. MIT鈥檚 Whitehead Institute recently announced a device to do DNA sequencing
by electrophoresis, but electrophoresis in microchannels machined into a wafer
of glass. A team at Imperial College, London, and researchers at Lawrence
Livermore National Laboratory in California have amplified DNA by the polymerase
chain reaction on a chip. All this miniaturisation could speed up conventional
sequencing methods.

But Schloss says that the nanopore idea is the most radical proposal he鈥檚
seen. 鈥淚t鈥檚 too early to know if it鈥檒l work in its present incarnation,鈥� he
says, 鈥渂ut most people find it only a little bit science-fictiony鈥�. The
thirst for gene sequencing will only increase, he believes, and departures like
the nanopore idea will push the technology in directions that we can鈥檛 even
anticipate. 鈥淒o we need that much sequencing?鈥� asks Schloss. 鈥淭here is no
question we do.鈥�