¿ìè¶ÌÊÓƵ

THIS SUMMER sees the inauguration of the most ambitious enterprise planned
so far in the drive to map and sequence the human genome. It concerns a
worm, the nematode Caenorhabditis elegans. The ultimate aim is to produce
the complete DNA sequence of the organism’s genetic blueprint. The project
is impressive enough, simply in the magnitude of its goal, but it takes
on a special significance because it is the first venture on this scale
that is truly international.

The task is to be divided equally between the Medical Research Council
Laboratory of Molecular Biology, Cambridge, and Washington University, St
Louis, Missouri. ‘It will be a test case of the technology and the approach
we are taking,’ says James Watson, who, as head of the US National Institutes
of Health’s Office of Genome Research, has played a key role in promoting
the genome initiative in the US (see ‘In the beginning was the genome’,
¿ìè¶ÌÊÓƵ, 21 July).

Although the worm is a modest creature by any standards – it measures
less than a millimetre long and its body is composed of a mere 959 cells
– its genome is 400 times as large as the biggest genome that has been sequenced
so far. Packed into six small chromosomes, some 100 million nucleotide bases
constitute the nematode’s genetic blueprint, instructions for making the
mature organism from a single cell, and for maintaining the individuals
in their daily lives. It is this set of instructions that the joint venture
hopes eventually to read from beginning to end. The researchers confidently
expect that once the instructions are deciphered, they will yield insights
into the fundamental molecular machinery not only of the nematode itself
but of all multicellular organisms, including humans. ‘It was this argument
that persuaded the MRC to become involved in the project,’ says John Sulston,
joint head of the Cambridge end of the operation.

That involvement is by no means trivial, especially in these days of
tight British science budgets. The first three years of the endeavour –
labelled as a pilot study – are expected to cost a total of $6 million,
to be divided equally between the two sides of the Atlantic. The MRC is
to be responsible only for two-thirds of the Cambridge funding.

In a characteristic display of deft politicking and bureaucratic gamesmanship,
Watson, who discovered the structure of DNA with Francis Crick, persuaded
the National Institutes of Health (NIH) in the US to pick up the rest of
the tab for the Cambridge effort in addition to all of the St Louis funding.

Even though the NIH is relatively flush with money for the Human Genome
Program – its budget request for fiscal year 1991, for instance, is $108
million – it is unusual for the agency to fund research in laboratories
outside the US. ‘It was important to get some real international collaboration
going on,’ explains Watson. ‘This was a good opportunity.’ It is fair to
say, however, that without the MRC’s support of the Cambridge laboratory’s
work on the nematode over the years, and especially in its recent gene mapping
efforts, there would be no project for the NIH to fund.

Although the instigators of the American human genome initiative focused
originally on mapping and sequencing human DNA only, the idea emerged early
on that other genomes should be included. Information from other genomes
would be important in terms of comparative biology, and would help to make
better sense of what eventually would be readable in the human genetic instructions,
ran the argument. The idea to include other organisms in the overall venture
was Watson’s, something he describes as ‘my most important contribution
to the project’. The simplest of these so-called model organisms is the
bacterium Escherichia coli and the most complex is the mouse. The nematode,
also on the list, comes somewhere in between.

The nematode owes its presence on the list of model organisms to a wildly
ambitious idea dreamt up in the early 1960s by Sydney Brenner, then at the
MRC Laboratory of Molecular Biology in Cambridge. The plan was as simple
as it was breathtaking: the complete description of the developmental biology,
genetics and molecular biology of a single organism. Brenner chose Caenorhabditis
elegans as his candidate organism. ‘People thought I was crazy,’ recalls
Brenner. ‘Jim Watson said at the time he wouldn’t give me a penny to do
it. He said I was 20 years ahead of my time.’ Ironic, now that, almost three
decades on, Watson is pushing $6 million at the worm project, and that’s
just for starters.

Brenner was not discouraged, and in 1963 initiated an enterprise that
has become one of the most admired in all of experimental biology. The upshot
has been that the nematode is now the most completely understood of all
multicellular organisms, just as Brenner hoped it would be. When the genome
project came along, the nematode was a natural candidate as a model organism
for complete DNA sequencing.

In the early 1980s, John Sulston, a leading figure in Brenner’s team,
had just completed the monumental – and some said impossible – task of tracing
the fate of every cell in the developing nematode, from fertilised egg to
mature organism. Work on the molecular genetics of the little worm was going
on apace around him, steadily clarifying the function of the organism’s
several thousand genes. Gene by gene, Brenner’s target was being neared.
‘It was all very impressive, very exciting, but I wanted to contribute in
a more global way,’ remembers Sulston. ‘That’s when we decided to do the
physical map.’ Sulston initiated the project in 1983 with colleague Alan
Coulson, who earlier had developed DNA sequencing with the MRC laboratory’s
Fred Sanger.

Simply stated, a physical map is a description of the genome in which
landmarks and key genes are identified along its entire length, as if the
chromosomes have been laid end to end. As it turned out, the existence of
a physical map is also a perfect starting point for launching an effective
sequencing operation. This was another reason why Watson enthusiastically
championed the Cambridge/St Louis collaboration.

‘You have to remember that although mapping is relatively commonplace
these days, when John and Alan embarked on the C. elegans map, it was a
novelty,’ says Robert Waterston, head of the St Louis end of the collaboration.
‘It was a tremendous challenge.’ Working with fragments of the nematode
genome of about 40 000 bases long, Sulston and Coulson identified genetic
landmarks on sets of related fragments. Although 2500 such fragments could
in theory span the entire genome if placed end to end, a much larger number
must overlap in order to create a map. This means that you have to break
up the genome in an essentially random way to produce different spectra
of fragments.

By the summer of 1985, the Cambridge team had some 7000 cosmids, which
in theory represented the nematode genome three times over. In practice,
however, only about 60 per cent of the genome was truly mapped, leaving
about 850 gaps, genetica incognita. Sulston and Coulson could not fill these
gaps, because for some reason the DNA in them simply refused to be cloned.
They had encountered the problem of ‘closure’, a spectre that would haunt
all genome mappers who followed in their footsteps.

Closure is still something of a puzzle, but perhaps has to do with the
way that genome mappers routinely clone the DNA they’re interested in: specifically,
they use the bacterium Escherichia coli. In order to have enough material
to work with, mappers insert their fragments of DNA, suitably tailored,
into E. coli, where it is multiplied millions of times. Mostly this works
very well. But not always. The reason may be that there are regions of eukaryotic
DNA – DNA from creatures like you and me and the nematode – that E. coli
cannot process easily. It was at this point, summer 1985, that Waterston
went to the Cambridge laboratory for a year.

In fact, it was a return trip for Waterston, who had been part of the
Brenner team as a research fellow between 1972 and 1976. His postdoctoral
work had been on the genetics and molecular biology of nematode muscle,
a subject he continued with when he left Cambridge and went to Washington
University. He was still working on these problems when he took the sabbatical
in Cambridge, but he had also become interested in mapping. One reason was
that Maynard Olson, one of America’s innovators in the mapping arena, worked
in the laboratory next door. ‘We talked a lot, and I was very aware of the
problems and the techniques,’ says Waterston. Waterston’s interest in mapping
and association with Olson was to be crucial in Sulston and Coulson’s worm
project – and ultimately in the drive for the sequence.

‘When Bob was here on sabbatical we discussed the mapping problems,
the problem of closure and so on,’ says Sulston. ‘Bob became very involved’.
In fact, Waterston had brought with him a newly developed technique, pulsed-field
gel electrophoresis, that was useful in handling large pieces of DNA. And
that would be necessary to close the gaps, to cover the enigmatic, unclonable
regions. With luck, some of the large pieces of DNA that could be isolated
with this technique might span some of the recalcitrant gaps. But the real
break came in the summer of 1986, when Waterston returned to St Louis. He
had decided to join in the mapping effort, ‘even though we hadn’t actually
sorted out how extensive the collaboration would be,’ says Waterston. ‘Soon
after I got back I heard about the idea of YACs that Olson’s lab was working
on. I hoped this could solve our problem, the problem of closure.’

YAC stands for yeast artificial chromosome, developed in Olson’s laboratory.
You take a very large piece of the DNA you’re interested in – a quarter
of a million bases at least – and attach to it various elements from a yeast
chromosome. You put this genetic chimera – the YAC – into a yeast cell,
and it is treated just like a normal chromosome: the YAC, which contains
the piece of DNA you’re interested in, will be replicated to experimentally
useful quantities . The most obvious advantage of YACs is that you can work
with pieces of DNA at least five times as large as the usual cosmids. But,
even more important, yeast is a eukaryotic organism, and so is likely to
be able to handle DNA from other eukaryotes. The problem of gaps might be
minimised.

‘Bob called us as soon as he heard about YACs,’ recalls Sulston, ‘and
we were therefore able to take advantage of the technique right away.’ After
some initial technical problems, the YAC technique quickly made an impact
on the nematode map. Within two years virtually all the genome was represented
in YACs, and the number of gaps in the map had plummeted to about 140. Although
the closure of the remaining gaps would take time, the researchers wondered
what to do next.

‘Sequencing the genome seemed the logical next step, even though the
project would be on an entirely new scale,’ says Waterston. Having the complete
sequence is rather like the ultimate physical map. The molecular geneticists’
task would be transformed, because everything would be in front of their
eyes. All they would have to do would be to read the messages. But could
the sequence be done? ‘With talk of the human genome program in the air,
we saw a chance of getting it started,’ says Waterston.

The Cambridge and St Louis researchers began to work out a plan, and
were joined in discussions by Robert Horwitz, a nematode worker at Massachusetts
Institute of Technology (MIT). They approached Jim Watson with the idea
in May 1989, at a gathering of nematode researchers at Cold Spring Harbor
Laboratory. In addition to his role in the Human Genome Program, Watson
is director of the Cold Spring Harbor laboratory. ‘Watson liked the idea,’
says Waterston, ‘especially the international aspects.’ Watson offered some
practical advice about how funding might be obtained, and this included
involving the MRC rather than just seeking NIH money, as had been the initial
notion.

In fact, Watson more than liked the idea. He thought it was tremendous.
A month later, at a meeting of the NIH advisory committee on human genome
research, Watson was enthusiastically promoting the notion of a collaboration,
arguing that the NIH might even agree to help fund research in the Cambridge
laboratory. He had a receptive audience on the committee. ‘It’s a wonderful
idea,’ said Phillip Sharp, of MIT. ‘I’m all for it,’ was the response of
Nancy Wexler, a pioneer in research on Huntingdon’s disease. ‘After that,
Watson was going around the country talking about the project as if it were
all arranged,’ says Sulston. ‘It put a lot of pressure on us, but we’re
not complaining.’

A joint proposal was formulated, and submitted to the MRC and the NIH
at the end of last year. By spring both the agencies had given preliminary
approval, with details to be worked out during the summer. ‘It all happened
very fast,’ says Sulston. ‘We were swept along by events.’

One of the pieces of advice that Watson had offered to the C. elegans
quartet at the Cold Spring Harbor meeting was that they should initially
ask for funding for a pilot project, leaving the all-out sequencing effort
for a second stage. Going for the whole thing all at once would be likely
to fail, he said. The three-year, $6-million pilot project is designed to
see how a large-scale sequencing effort might be established. Each laboratory
will have about 10 people devoted to the project. ‘We want to test the technology,
develop software for data handling, and work out the organisation that would
be needed to go to full-scale sequencing,’ says Sulston. ‘We really don’t
know how it will work out.’

One of the prime tasks of the pilot is to determine what kind of sequencing
machinery will work best for scaling up, and just how much automation will
be possible in the short term. Automation is what in the end will bring
down the cost of sequencing, but the initial concern is with accuracy. Even
a 99 per cent fidelity of sequencing is unacceptably low. During the first
year, both laboratories will be trying out the various sequencing options,
producing only about 100 000 bases on each side of the Atlantic . This rate
will build up to 400 000 bases in the second year, and 1 million in year
three, giving a total of 3 million in all.

Measured against the biggest sequencing effort achieved so far – the
240 000-base sequence of cytomegalovirus completed over a period of five
years by Bart Barrell and his colleagues at the MRC laboratory in Cambridge
– these nematode numbers look impressive. But set against the ultimate goal
of the 100 million bases in the entire genome, they represent barely a start.
One of the major challenges of the last year of the pilot project would
therefore be to work out a scale-up of the operation, from 1 million bases
in each laboratory each year to 10 million.

‘We hope to achieve significant economies in the scale-up,’ says Waterston.
These economies might drop the price to something like 50 cents a base –
a realistic aim with the projected technology. Even so, that still represents
a hefty figure for the final bill, some $50 million over five years.

If the pilot project is a technical success, there is little doubt that
the funds will be available for continuing the American end of the effort.

But what of the Cambridge contribution? ‘Oh, I bet the money will come,’
says Watson, with his typical bluff confidence. ‘It’s a very important project,
a very important part of the human genome program as a whole.’ Tony Vickers,
who is overseeing the project for the MRC, says that it is too early to
guess what the funding prospects might be. Currently, however, the British
government has set itself limited goals, by American standards at least.
With a budget for all human genome research of some $19 million over the
next three years, the prospects of Britain taking a significant share of
a $50-million, five-year project do not look bright at the moment. But,
as Watson says: ‘It depends what the British government wants here.’

* * *

BUILDING ARTIFICIAL CHROMOSOMES IN YEAST

IN THE three years since the yeast artificial chromosome (YAC) technique
hit the pages of the scientific press, the business of genome mapping has
been transformed. At a stroke, the technique allows mappers to march in
giant leaps along the genome.

In the pre-YAC era, mappers were limited to using DNA fragments up to
about 50 000 bases in length. These fragments – cosmids – had several benefits.
First, because the cosmids could be replicated in the bacterium Escherichia
coli, large quantities could be obtained through the standard process of
cloning. In addition, if a DNA sequence of interest was known to be located
on a particular cosmid, it was not difficult to pin down: there would be
only 50 000 bases to look through.

Cosmids have disadvantages, however, principally because they are smaller
than even the smallest genome, not to mention the 3-billion-base genome
of humans. Trying to map the human genome with cosmids would be tedious.
‘Think of the human genome as a jigsaw,’ explains Eric Green, of Washington
University, St Louis. ‘If you try to do a jigsaw with tiny pieces, you have
a very difficult time. That’s what cosmid technology gives you.’ Jigsaws
are always easier if you have larger pieces. That’s what YACs give you.
The YACs provide jigsaw pieces varying from 250 000 to a million bases,
a significant improvement on cosmids.

The size limit of cosmids is determined solely by how large a piece
of foreign DNA can be inserted into the vector, which is then introduced
into E. coli to be processed. The jump in size allowed by YACs is achieved
because the vector is so much bigger, and this can be handled by yeast.
Yeast chromosomes vary in size from 250 000 to a couple of million bases.

What the YAC technique does is take a large piece of DNA and disguise
it as a yeast chromosome. Simply introduce the YAC into the yeast cell,
and the cell’s molecular machinery will replicate it just like genuine chromosomes.

Like most good ideas, it sounds deceptively easy, but it took the genius
of Maynard Olson of Washington University to make it a reality. There are
three basic elements of the necessary disguise. Two of them are fundamental
components of all chromosome architecture, a centromere (for the centre),
and two telomeres (for the ends). The third element is called the ARS, or
autonomous replicating sequences, which is necessary to the initiation of
DNA replication. These elements had been well characterised by yeast researchers
over the years, and so could be pulled off the shelf. The centromere, telomeres,
and ARS form the basis of the DNA carrier system, the vector.

One or two other elements were also necessary for the YAC technique
to work well, developed in Olson’s laboratory by graduate students David
Burke and Georges Carle. To introduce the foreign DNA into the YAC, you
have to split the vector in two, giving a right arm and a left arm. Mix
the foreign DNA with split vectors, and attach the two arms to the ends
of the DNA fragment. You have to be certain, of course, that the YAC has
a right arm and a left arm, not two right or two left. So, Olson and his
colleagues fixed this by adding a gene to each arm, say gene A on the right
arm and gene B on the left arm. The trick is that the strain of yeast used
will grow only if genes A and B have been introduced together on the YAC.
Any other combination, and the yeast simply refuses to grow. It’s a form
of selectivity well known to geneticists. Other genes are often present
on the YACs, to allow variations on this theme of selectivity.

Having fooled the yeast cell that it has got an extra chromosome, and
having provided the missing genes necessary for growth, DNA replication
can get under way. Although the yeast cell system can handle large pieces
of foreign DNA, it cannot be fooled into producing huge quantities of this
DNA to the exclusion of its own genome, as can be done with the bacterial
system. So the YACs impose limits on how much DNA can be produced, and there
are more problems with separating the YAC DNA from the yeast chromosomes.
Nevertheless, the large size of the foreign DNA fragments that can be cloned
makes it worthwhile.

In addition, because yeast is a eukaryotic cell – a cell with a nucleus
– it has less problem replicating certain difficult sections of DNA from
other eukaryotic species. In many cases, E. coli appears unable to replicate
all of a eukaryotic genome, leaving unavoidable holes in any gene map.

The ultimate aim with genome mapping is to have a series of overlapping
YACs that represent the entire genome from end to end. The Cambridge/St
Louis collaboration is reaching this goal. At this point it is possible
to lay out the YACs in order as a grid, a physical representation of the
entire genome sequence. In the case of the nematode, more than 95 per cent
of the genome is covered by some 950 YACs (selected from several thousand).
These YACs can be laid out over a single piece of filter paper, the size
of a postcard, making it the largest organised collection of YACs in existence.
The actual number of YACs here is dwarfed by the St Louis collection from
the human genom – 70 000 – but these are not yet organised as sequences
in the genome.

* * *

DNA SEQUENCING MADE SIMPLE

THE human genome initiative relies on the ability to sequence quickly,
accurately and cheaply. One of the central goals of the Cambridge/St Louis
col laboration is to put current sequencing technology to the test. The
basic method of much modern sequencing remains the same as the one that
earned Fred Sanger at Cambridge a share of the 1980 Nobel Prize for Chemistry.
Recent modifications have focused on ways of automating the process.

In a short DNA strand of 500 bases, the usual size of strand worked
with, you need to identify each base, from position 1 through to position
500. The Sanger method for doing this actually works backwards. You take
single-stranded DNA, and make the complementary strand. This second strand
gives you the required information.

A short primer is added to the single-stranded DNA, and this is then
bathed in a pool of the four bases – adenosine, thymidine, cytosine, and
guanosine – under correct enzymic conditions for growing the second strand.
But in the synthesis a small proportion of one of the bases is in a form
– the di-deoxy form – that halts chain elongation. Because of the mix of
normal to aberrant forms of this base in the reaction, chains may grow only
a short way before being stopped by the di-deoxy base, sometimes proceeding
all the way to the end, sometimes stopping in between.

With the correct mixture of reagents, the finished pot will contain
a mixture of chains, with chain elongation halted at every position in the
DNA strand that this particular base occurs. If the base concerned here
is adenosine, for example, then the reaction pinpoints all the positions
in the DNA strand at which adenosine appears.

Run the DNA products of the reaction mixture down a gel that separates
chains out according to their length, and you will find a ladder-like separation
of the different elongated chains. If you find strands at positions that
correspond to lengths of, say, 3, 5, 6, 9, 12 and so on, you know that these
are the positions at which adenosine appears in the DNA. Do this same reaction
separately for the other three bases, and you can read off the positions
of all four bases throughout the DNA strand. That’s the sequence.

A key part of this operation, of course, is being able to detect the
position of the chains of different length on the gel. Until recently, this
was done using radioactivity to tag the chains. The gel was exposed to a
piece of photographic film, and the four ladder patterns read off from there.
Four years ago, Leroy Hood, of the California Institute of Technology, introduced
the idea of using fluorescent dyes to tag the chains, an approach that many
believe now leads the field. Fluorescent dyes allow band detection by laser,
a trick that could become both accurate and automated. Reliable computerised
data collection is a must for the genome project.

The Cambridge and St Louis laboratories will be testing two automated
sequencers, both based on the fluorescent dye approach. One is that invented
by Lee Hood and developed commercially by Applied Biosystems Incorporated,
California. Wilhelm Ansorge of the European Molecular Biology Laboratory,
Heidelberg, devised the second, and this is sold by Pharmacia, Sweden. In
the ABI machine, four different fluorescent dyes are used, one for each
base. The reaction products are then mixed together, and passed down the
gel in a single lane, not in the four lanes required when a radioactive
tag is used. A laser detector scans the four colours as the DNA comes down
the gel, and records the sequence automatically. The Pharmacia machine uses
only one fluorescent dye, and it keeps the products of the four different
reactions separate. Four lanes on the gel have to be run for each DNA, as
in the traditional radioactive method. But the advantage is that the laser
detector is tuned to a single colour, and, in principle, can be more robust
than Hood’s technique.

Each of the two nematode labs will have an ABI machine and a Pharmacia
machine, allowing parallel testing under different conditions. Because the
ABI machine needs just one gel lane for each DNA to be sequenced, it can
run more clones simultaneously than the Pharmacia machine, which needs four
gel lanes for each clone. A total of 24 clones (of about 500 bases each)
can be run simultaneously on the ABI machine, giving total sequence collection
of 12 000 bases in 10 hours. Although the Pharmacia machine can run only
10 clones at a time, it is faster, completing a run in five hours. Theoretically,
it could sequence 10 000 bases in 10 hours, but squeezing two runs in for
the ABI’s one would be pushing the machine and personnel to the limit.

Because the Cambridge and St Louis teams have the physical map in hand,
they will be able to approach sequencing in a targeted way. Specific regions
will be parcelled out between the two laboratories, and each will be able
to produce biologically useful sequence data very quickly. With the map,
sequence data immediately slot into a context, rapidly expanding to large
sequenced sections whose genetic information can then be analysed. Without
the map, sequence data often remain enigmatic, isolated in fragments until
much of the genome is completed. The researchers plan to sequence both DNA
strands, as a check on the fidelity of the data.