91av

KARY MULLIS was driving along Highway 1, heading north from San Francisco
to Mendocino, when he experienced the kind of enlightening brainstorm that
most scientists only ever fantasise about. He pulled over to the side of
the road, and started doing mental arithmetic. ‘Ten cycles would get me
a thousand, and 20 would get me a million,’ mused Mullis. He had just invented
the polymerase chain reaction, PCR. It is not hyperbole to say that PCR
has shaken molecular biology to its foundations, enabling scientists to
do undreamt-of things with DNA. By combining the target-seeking ability
of a bloodhound with the multiplicative prowess of a rabbit, PCR would give
genetic engineers a tool of unbelievable power.

Sitting at the side of the road, though, Mullis knew only that it would
be a long weekend before he could get back to his laboratory at the Cetus
Corporation near San Francisco, so he swung back into the traffic and carried
on to his destination. He did not sleep much as he went through his thought
experiment over and over. All the steps were known to work, but Mullis found
himself assailed by doubts. If what he proposed was that simple, why had
nobody done it before? And if they had done it, something must have gone
wrong, or the world would already know about it.

Unwilling to waste his time, Mullis’s first task when he got into Cetus
early on Monday was a search of the literature. He found nothing, so he
decided to give it a try. Mullis picked a short stretch of a piece of well-known
DNA as his target, and set to. There were a score of different parameters
that he just guessed at, but the very first time he performed a polymerase
chain reaction, it worked. Mullis swears he really did shout ‘Eureka’ as
he gazed on his first successful results.

What he had done was to amplify – make more copies of – a small stretch
of DNA hidden within the genome, the entire genetic material of an organism.
The technique depends on knowing the DNA sequence on either side of the
target. Heat denatures the DNA into its two component strands. Primers,
small molecules that match the end sequences, seek out the target. An enzyme,
DNA polymerase, then builds a new strand, starting with the primer and moving
across the target towards the other primer. After another round of denaturing,
the new strand becomes a template for a further strand, and so on and so
on. The magic of exponentiation turns a single target into a million copies
in just 20 cycles .

To begin with, the technique, valuable though it was, was time-consuming
and laborious. The temperature that denatured double strands of DNA also
destroyed the polymerase enzyme, so that the researcher had to add a fresh
squirt of polymerase on every cycle. The answer was to find a polymerase
that could stand the heat, and the obvious place to look for it was among
the bacteria that live in hot springs. Mullis knew that one of these, called
Thermophilus aquaticus, had been investigated, because it was possible to
buy other enzymes from this bacterium, nicknamed Taq. But no one had purified
Taq’s polymerase. Worse, no one in Mullis’s department at Cetus wanted to
do so, either.

It took Mullis a while to convince anyone at Cetus that he was onto
a winner. His department had been searching for ways to diagnose point mutations.
These are single changes to the genetic code that underlie certain inherited
diseases, such as sickle cell anaemia. Mullis had been wondering what might
go wrong with a new approach to the detection of point mutations that he
was planning to try when that first vision of PCR forced him to pull over.
But when he went back to the head of department, and told him he had a method
that might help, he discovered that ‘people don’t usually like to have somebody
else solve their problems’. Rather grudgingly, the diagnostic department
handed Mullis some of their samples. He went off, amplified the stretch
containing the haemoglobin gene, where the sickle-cell mutation lies, and
returned with a warning: ‘I told them, this is going to be like 100 000
times your normal signal, so don’t use it all.’

What he meant was that there might be 100 000 times as much DNA as the
diagnostic department was used to working with. To make the DNA visible
to the naked eye, they would first use ultraviolet light and a fluorescent
marker, ethidium bromide, to reveal the position of the bands of DNA on
the gel used to separate different lengths of DNA electrically. If that
looked good, they would use the radioactive label previously incorporated
into the DNA, laying the gel onto an X-ray film so that the radioactivity
would make a permanent record of the position of the DNA bands. Too much
of the labelled DNA would obscure the bands on the ‘autoradiograph’.

The diagnostics department paid no heed to Mullis’s warning. They used
the whole lot he brought them. Under ultraviolet light the gels had been
beautiful, but the autoradiographs were completely black. Just as Mullis
had predicted, there was an overwhelming quantity of DNA. When the team
repeated the test with just a small portion of the amplified DNA, a fine,
clear picture of the DNA appeared. It enabled them to test for the presence
of the sickle-cell gene with very much smaller amounts of DNA.

The idea of using a heat-stable polymerase met with the same sort of
opposition from Mullis’s colleagues. The head of Mullis’s department now
had a PCR system that, no matter how cumbersome it might be, worked. He
did not see the need to streamline it. Indeed, he spent months building
what Mullis describes as ‘a ridiculous machine’ to automate the tiresome
procedure. Just about the time the machine started working, Mullis persuaded
someone in the microbiology department at Cetus to get hold of some Taq
and culture them. David Gelfland, a colleague, purified the polymerase from
Taq and brought it over to Mullis. It worked first time.

The importance of Taq polymerase was twofold. First, it removed the
necessity to hang about adding polymerase after each cycle. That made automation
a cinch, and within months the hardware to perform PCR was available off
the shelf. More important, from the scientific point of view, was the fact
that because Taq polymerase could withstand high temperatures, it enabled
researchers to carry out the annealing, when the small primer molecules,
the oligonucleotides, bind to their targets, at a higher temperature. The
higher the temperature, the more precise the match between two strands has
to be for them to bind together. Any mismatch weakens the bond between the
strands, which come apart at lower temperatures. This meant that researchers
could be more confident that the oligonucleotides had found their intended
targets – not some other sequence that was close but not a perfect match.
And that improved the sensitivity of the technique.

Once they realised the potential of PCR, the group at Cetus set about
defining some of its capabilities. They measured its sensitivity: how many
copies of the target did the technique need to get going? By diluting known
amounts of DNA, Cetus quickly discovered that PCR would work when there
was just a single copy of the target in the test tube. And with the correct
conditions for annealing and polymerising, it remained extremely specific,
amplifying only the target sequence and not others with similar ends.

Other improvements came rapidly. A team at Stanford University School
of Medicine in California devised a technique they called ‘anchored PCR’,
which required the researcher to know only one end of the target. Elwyn
Loh and his colleagues attached a tail of guanines, one of the four bases
that make up DNA, to the unknown end of the fragments they were interested
in. The primer for that end consisted of a stretch of cytosines, which bind
to guanine, linked to a sequence that forms the site of a ‘cutting’ enzyme.
Such enzymes, known as restriction enzymes, cut DNA at specific sequences
that they recognise, enabling the researchers to release the fragments they
had amplified. Anchored PCR is proving extremely valuable to scientists
who want to investigate variable portions of the DNA, such as the genes
that code for the immune system.

Duplications in any direction

One drawback of ‘standard’ PCR is that it works towards the target.
That is, the polymerase builds a strand from one primer, across the target,
towards the other. By the very nature of the process, the two primers define
the ends of the target, and only the stretch between them is amplified.
But often the regions of interest lie on either side of the target. They
may be promoters, responsible for switching genes on or off, or processing
signals downstream of the gene, but in either case they are invisible to
ordinary PCR. To overcome that limitation, Howard Ochman and his colleagues
at the Washington University School of Medicine in St Louis devised a technique
they call inverse PCR.

The first step is to cut the DNA with a restriction enzyme that does
not recognise any sequences within the target or its flanking regions. Many
restriction enzymes produce a ‘staggered’ cut, with a short single-stranded
piece of DNA sticking out. These strands will link with one another, and
under the right conditions, two ends of a single fragment come together
to form a circular molecule. Polymerase is set to work on these circles,
with primers that match the ends of the target sequence, but face the other
way. The polymerase works around the circle, away from the target and through
the flanking regions. After sufficient cycles – Ochman used 30 – the circles
are broken open with a different restriction enzyme that cuts within the
target. Conventional PCR can then be used again to amplify the flanking
regions, which are now sandwiched between the two portions of the original
target (see Figure).

When Mullis originally devised PCR he was working on a problem of diagnosis
– how to find the point mutation that leads to sickle-cell anaemia. That
remains an area in which PCR has immediate and easily recognised benefits,
most notably by speeding things up. Conventional DNA analysis – sequencing,
for example, or mapping – requires relatively large amounts of DNA. Before
PCR, there were two ways to obtain those amounts: either take a large sample
to begin with, or take a small sample and hope to be able to multiply it.
The two techniques that were then available for multiplying DNA were to
grow the cells containing it, which was not always possible, or to excise
the area of interest and insert it into cells that would grow. Both took
time and trouble. With PCR, researchers can amplify the area of interest
and then analyse it within a matter of hours, rather than days or weeks.

One area in which this is particularly important is in the diagnosis
of genetic diseases. The earlier a positive diagnosis can be made, the better,
and PCR has enabled researchers reliably to detect specific genes in a single
cell taken from a young human embryo. Couples undergoing in vitro fertilisation
could choose to have only those embryos that do not carry the gene in question
returned to the mother. Even for conventionally conceived embryos, speedy
procedures mean earlier diagnosis, with the possibility of safer and less
traumatic elective abortion.

It is PCR’s ability to find and amplify small tiny amounts of DNA that
makes it so useful. For example, HIV, the virus that causes AIDS, can be
present in the body in low numbers, hidden within a person’s own DNA. PCR
will ‘pull’ it out and allow its identity to be confirmed with conventional
techniques. This has proved especially useful in diagnosing the presence
of HIV in newborn babies. It also enables physicians to assess drugs to
see whether they can actually get rid of HIV.

In forensic medicine too, PCR has quickly found a niche. It enables
one to obtain genetic fingerprints from the minute amount of DNA in the
root of a hair, or from degraded samples of semen, skin, and blood. Indeed,
PCR is even being put to work in the emerging discipline of molecular archaeology,
pulling out fragments of usable DNA from 8000-year-old mummies and mammoths
frozen in ice.

But the very sensitivity that makes PCR so useful also threatens to
render it useless. The primers are very particular about the sequence they
will bind to, but they cannot distinguish a genuine sample from an impostor.
Contamination is a devilish problem, especially when PCR is being used to
amplify sequences common to all people. One solution is scrupulous cleanliness:
staff have to swab down vessels and surfaces with enzymes that destroy DNA.
The drawback is that any traces of these enzymes could also destroy the
target. Physical segregation of various tasks in the laboratory is important
too, and many laboratories that use PCR for detecting, say, HIV, have had
to redesign their facilities. Technicians who prepare samples for amplification
are kept out of areas where amplified samples are analysed. That way there
is less chance of someone carrying even a single molecule back to contaminate
the amplification process. Good experimental design helps too, with appropriate
controls to ensure that the source of the amplified DNA is indeed the sample.

These are the problems of any sensitive technique, and while some scientists
may have been led down the garden path by results that were not what they
seemed, they generally realise their error before too long. Meanwhile, it
is no exaggeration to say that PCR has changed the face of modern molecular
biology, perhaps almost as much as restriction enzymes did nearly two decades
before. The journal Science recognised this when it gave DNA polymerase
its first annual Molecule of the Year award in December 1989. The interesting
question, one that no one anywhere has addressed, is why PCR happened when
it did. After all, the molecular components – primers, DNA polymerase, denaturing
and annealing – had all been in place for years. It was the conceptual leap,
the construction of an entirely new process from familiar ingredients, that
made PCR the tool it has become.

How did Mullis make that leap? ‘At the time,’ Mullis told me, ‘I was
writing a lot of computer programs and I was used to the idea of iterative
loops, whereas I think most biochemists think in terms of one reaction.
They would probably not have thought that, hell, you can do something over
and over and over again if you want to. And if you do something that doubles
something over and over again, then that thing increases exponentially.
. . . It seemed so simple that I couldn’t believe that nobody had ever done
it.’ They hadn’t, he did and it worked.

* * *

1: PCR – what it is and how it works

THE POLYMERASE chain reaction takes advantage of the ability of a small
piece of DNA to find and bind to a target sequence on another stretch of
DNA. It then uses an enzyme called polymerase to make a new piece of DNA
corresponding to the region adjacent to the target. By repeating the process
through several cycles it exponentially amplifies the target sequence.

DNA normally exists as the renowned double helix, two strands wound
round one another. Each strand is a string of so-called bases, adenine,
thymine, cytosine and guanine, denoted by their initial letters A, T, C
and G. Because of their structure, the bases form pairs. Adenine on one
strand is always opposite thymine on the other, while cytosine pairs with
guanine. The bonds between pairs of bases hold the two strands together
in the double helix.

Base pairing is the key to DNA’s success as a molecule for storing and
copying hereditary information; each strand contains all the information
needed to build its opposite number. This base-pairing means that for every
sequence of DNA there is a complementary sequence – ATCG is complementary
to TAGC – and complementary sequences of single-stranded DNA will tend to
bind together to form a double strand.

In the original form of PCR, the researcher had to know the sequence
of bases on either side of the target DNA before making short stretches
of DNA. These ‘primers’, called oligonucleotides, are complementary to these
sequences. The oligonucleotides, along with a soup of enzymes and raw materials,
are mixed with the DNA containing the target and heated.

Heat separates the two strands of the double helix of DNA, which are
normally held together by bonds between complementary bases. As the mixture
cools, the short oligonucleotides find their complementary sequences on
the long strands of DNA and bind to them. The researcher now adds an enzyme
called a polymerase.

The polymerase joins nucleotides, the building blocks of DNA, into a
new strand of DNA, attached to the oligonucleotide primer. The new strand’s
sequence is complementary to the sequence of the long strand.

After a while, when the researcher can be reasonably sure that the new
strand goes past the primer sequence at the other end of the target DNA,
the mixture is heated again. The DNA again separates into its component
single strands, but where before there were, say, just two copies of the
target, now there are four. Again, as the mixture cools, the primer binds
to the target sequence, and the polymerase builds a new strand of DNA on
the target. Now there are eight copies of the target. Each new cycle of
heating, cooling and polymerisation doubles the number of copies of the
target DNA.

More to the point, the ends of all of those copies are the same, defined
as they are by the oligonucleotide primers. This means that after PCR, the
researcher can purify the amplified sequence by running the DNA through
an electrophoretic gel. This sorts DNA out by size; all the copies of the
target, because they are all the same length, will lie in a neat band on
the gel.

Under ideal conditions, the number of copies doubles with each round
of denaturing, annealing, and polymerisation. After 30 cycles there will
be more than a billion copies. PCR is a chain reaction, based on the polymerase
enzyme.

* * *

2: The legal battle for polymerase power

JUDGING by the number of corporate lawyers fighting over it, PCR is
as much of a financial gold mine as it is a scientific breakthrough. In
a case of Goliath pursuing a rather well-armed David, the chemicals giant
Du Pont filed a lawsuit in San Francisco last year against Cetus. Du Pont
is challenging two of four patents Cetus has been awarded covering the principles
and materials underpinning the technology. (Cetus has filed for some 40
patents in all concerning PCR.) The case is now scheduled for trial in November.

The suit followed Cetus’s refusal to grant Du Pont a licence to sell
PCR-related products, having already granted an exclusive one for medical
diagnostics to another big drugs company, Hoffmann-La Roche. The stakes
are high: financial analysts put the value of PCR diagnostics for the next
decade at well over $1 billion.

Du Pont has decided to proceed anyway; on 2 March, the company began
selling a test for HIV that employs PCR, charging researchers $150 for each
kit. Cetus’s associate counsel, Peter Staple, says Du Pont’s entry into
the marketplace could be cause for a countersuit. Du Pont’s case hinges
on two scientific papers published in 1971 and in 1974. The papers, the
first in the Journal of Molecular Biology and the second in the Journal
of Biological Chemistry, sketch a concept quite like PCR. In the first paper,
H. Gobind Khorana and Christopher Joyce associates describe the principles
of making multiple copies of pieces of DNA. In the second, Khorana and Amos
Panet add details on how to use primers and alternate cycles of heating
and cooling.

American law holds that any public description of a technique detailed
enough to allow others to duplicate it, if made a year or more before a
patent is applied for, invalidates the application. Du Pont argues that
the two papers predate Cetus’s applications by more than a decade, and thus
the patents describe nothing novel. And Khorana’s work can hardly be dismissed
as obscure, says Du Pont lawyer George Frank – Khorana won a Nobel prize
in 1968 for his work at the University of Wisconsin on the structure of
DNA.

Cetus counters that the papers did not ‘teach’ the technique well enough
for others to duplicate it; otherwise, say Cetus’s attorneys, people would
have been doing it before Cetus unveiled PCR in 1985. But Du Pont has secured
the opinion of another Nobel-prize winner, Arthur Kornberg of Stanford University,
to deny that claim. An affidavit from Kornberg filed with the patent office
on 9 March states that Khorana’s papers were quite adequate to teach the
technique to anyone with ‘ordinary skills’ in the discipline.

The curve of legal complexity shows no sign of levelling off. Hoffmann-La
Roche, trying to protect its grip on the market, has asked the patent office
to reexamine Cetus’s two disputed patents. The aim is self-protection, says
Kassy McGourty, a spokeswoman for Hoffmann-La Roche; the company wants to
be sure it stands on solid ground. In February the patent office agreed
to re-examine some of the claims made in the patents, including several
related to the content of the two Khorana papers. The review could take
as long as 18 months, according to the patent office.

True to type, lawyers for both Cetus and Du Pont find succour in the
patent office’s decision. By implication, says Cetus’s Staple, those patent
claims that will not be reviewed are solid, and a single valid claim can
support a patent. Du Pont counters that the re-examination lends credence
to the notion that Khorana stole a march on Kary Mullis and his colleagues
at Cetus. For good measure, Du Pont also has asked for a reexamination of
the patents by the patent examiner, according to George Frank, an attorney
for Du Pont.

The tug of war over diagnostic applications of PCR does not affect its
use in basic research, which constitutes a fraction of the diagnostics market.
Kits now sold to scientists by Cetus and its partner, Perkin- Elmer, come
equipped with written permission as part of the price.

Dr Jeremy Cherfas is European correspondent for Science.