杏吧原创

San Francisco

AT HOME in the human body, Mycoplasma genitalium enjoys an
enormously varied diet of delicious chemicals. Outside the body, it demands no
less. Clyde Hutchison describes its daily meal as 鈥渉orribly complex鈥�, a
bacterial banquet including several ingredients that most microbiologists call
鈥渦ndefined鈥�, and Hutchison calls 鈥渃ow bits鈥�.

But while M. genitalium鈥檚 diet is complex, its inner workings are
simple. A model of efficiency, M. genitalium gets by with only 468
genes, making it the simplest free-living organism known. Viruses are far
simpler, but they need help from the proteins of a host cell. M.
genitalium makes all the proteins that it needs.

Hutchison, who works at the Institute for Genome Research (TIGR) in
Rockville, Maryland, believes he can simplify things further. He is dismantling
the bug, one gene at a time, in an attempt to work out the minimum genome, the
smallest group of genes capable of sustaining independent life. Others are
taking a bottom-up approach to the same end. Researchers like Evgeni Selkov at
the Russian Academy of Science in Pushchino are using computers to model the
minimum number of biochemical pathways, and, by working backwards, the minimum
number of genes, needed to create a virtual bug.

With such information, Hutchison, Selkov and their colleagues may finally be
able to understand how everything in a cell鈥攖hat鈥檚 every single one of the
hundreds of thousands of molecules鈥� interacts with everything else to keep
life ticking over. The path to this ultimate understanding could help to create
superefficient, living chemical factories, and give hints about how life emerged
from the primordial slime.

One of the participants calls it 鈥渁 basic research project of breathtaking
beauty鈥�. But not everyone is so enthusiastic. One sceptic calls the whole
concept of a minimum organism 鈥渇allacious鈥�.

Genetic jumbling

Minimum organism research is only possible because of genome
projects鈥攅fforts to decode the entire DNA encyclopedias of various
bacteria, yeast and (by 2005) humans. TIGR scientists were the first to blast
their way through an entire genome, initially for a usually harmless resident of
our lungs called Haemophilus influenzae, and then for the diminutive
M. genitalium, which infects the lung and the genital tract. As these
genomes became available, Eugene Koonin, a computer scientist at the National
Center for Biotechnology Information near Washington DC, lined them up against
each other. Any genes that the two bugs had in common are genes that have
weathered the genetic jumbling of the 1.5 billion years of evolution since their
last common ancestor. Such genes are likely to be vital for survival since
natural selection has preserved them unchanged for so long. Koonin, aping
nature, kept the shared 240 genes and threw out the rest.

Next, where more than one gene could do the same job, Koonin disposed of the
surplus. He also threw out the genes that allow the two bacteria to cling to
human cells, figuring that they are only essential for infecting a human body,
not life. Finally, it turned out that 22 genes that look completely different in
the two bug species actually perform the same essential tasks. Although these
genes had been eliminated in the initial analysis, Koonin added back the M.
genitalium versions of the genes. That gave him 256 genes, the final
minimum set that Koonin says is 鈥渘ecessary and sufficient鈥� for the life of a
modern cell. Gone were all the genes for consuming food of all but the simplest
sort, gone were the genes for manufacturing anything that could be scavenged
from the environment. The theoretical bug could barely repair any damage that
its DNA sustained, and it had lost the ability to fine-tune the activity of each
of its remaining genes. 鈥淭he take-home message is how little metabolism there is
[in a minimum organism],鈥� says Koonin.

Out of action

Koonin鈥檚 theoretical bug is a starting point, but it is still just a list of
genes. Nobody knows for sure whether, if it were a real bug, it could survive.
The next step is to chop surplus genes out of a real organism. Hutchison is
putting random M. genitalium genes out of action by inserting bits of
DNA called transposons into them. Already he has destroyed 70 genes that were
not essential. He calculates that the job is only half done, so another 70
should follow. 鈥淲e鈥檙e thinking the minimum gene set may be a little bigger than
[Koonin] predicted,鈥� says Hutchison. 鈥淏ut so far things are in remarkable
补驳谤别别尘别苍迟.鈥�

Hutchison is creating lots of strains of the bacterium, each with one gene
missing. His next job is trickier. He has to make one strain with all of the
genes missing.

One way to target and destroy specific genes is a technique called homologous
recombination. Homologous recombination is temperamental, with a ballpark
success rate of one in a million. It works best within the single-celled yeast,
Saccharomyces cerevisiae. So Hutchison will remove the DNA from M.
genitalium and insert it into yeast, and then employ a reusable marker
called URA3 to label the genes that have been successfully destroyed.

Others have used homologous recombination to destroy genes (albeit not so
many at once), and put largish pieces of DNA into yeast鈥攕o that part of
the project has a good chance of working. The hard part will be transferring the
pockmarked DNA back into an M. genitalium cell. But the concept 鈥渋s
certainly not outlandish鈥�, says Patrick Brown, a genome researcher at Stanford
University in Palo Alto, California. If the technique works, Hutchison will have
created the simplest living cell in the world.

But Hutchison acknowledges that those who scorn the concept of creating the
simplest cell in the world have a point. That鈥檚 partly because of genetic
redundancy鈥攎ore than one gene can do a given job. This doesn鈥檛 means that
a very simple organism cannot be constructed, rather it means there is no one,
ultimate microbug. For instance, if you remove gene 1, M. genitalium
may be kept alive by gene 2; but if you started by removing gene 2, the bug may
survive using gene 1. 鈥淎s long as you realise that [the concept] has certain
flaws,鈥� says Carl Woese, an evolutionary biologist at the University of Illinois
in Urbana-Champaign, 鈥渋t will be useful.鈥�

The ultimate Tamagotchi

That doesn鈥檛 mean that minimum bugs are going to be easy to understand. A
cell is a bag of seemingly ever-changing chemical reactions and, even for a
minimum cell, the only thing capable of keeping track of them all will be a
computer.

With that end in mind, Hutchison has teamed up with computer scientist Masaru
Tomita of Keio University in Fujisawa, Japan, who by 2000 plans to have created
a 鈥渧irtual鈥� bacterium鈥攁n electronic entity that Tomita compares to a
Tamagotchi, the Japanese virtual pet.

But Tomita鈥檚 toy will be a little more complex. For a start it will require a
supercomputer to run it. At the International Conference on Intelligent Systems
for Molecular Biology, held in Greece in June, Tomita described a prototype
M. genitalium called E-CELL. At the moment, E-CELL is pretty good at
metabolism鈥攆or example, working out how to convert a sugar molecule into
energy, cell building blocks, and waste products. It also handles the conversion
of genes into proteins quite well. But it is less reliable when it comes to
transporting chemicals across membranes, replicating its DNA, or dividing to
form two daughter cells.

Tomita wants his final virtual M. genitalium to be an approximation
of all the biochemical pathways that make up the living cell. He predicts it
will comprise perhaps 2000 interacting virtual genes, proteins, chemicals, and
so on, governed by several thousand rules that control all of the myriad
possible interactions between these components.

Meanwhile, Selkov at the Russian Academy and Ross Overbeek at Argonne
National Laboratories in Illinois have devised a computer program that
constructs rough models of networks of biochemical pathways directly from the
genome sequence of an organism. If the program comes across a gene whose
function is unknown, it matches it up with a gene with a similar sequence whose
function is already known and slots its protein product in a pathway
accordingly. So far, Selkov and Overbeek have simulations of the main
energy-producing circuits of several organisms, including M. genitalium.
But their virtual circuits can act up, perhaps failing to respond to an
environmental change. This is a sign of a missing link in the virtual chain of
biochemical reactions. As more connections are added, the model should gradually
come to resemble a real cell.

Both the Selkov and Tomita teams plan to use their virtual bacteria to work
out what happens when different genes are removed, generating other versions of
the minimum organism.

But few are convinced that such ambitious models can be made to work. Harley
McAdams, an independent researcher in Palo Alto, California, is one such
sceptic. He has created a computer model of the rise and fall of viral protein
production in a bacterium when a virus called lambda infects it. This process
involves few genes, and yet, each simulated infection took considerable
supercomputer muscle. 鈥淥ne could say pretty quickly that one couldn鈥檛 model a
[whole] bacterium with that approach,鈥� says McAdams.

A bacterium is more than a group of linear chemical reactions. In the cell,
chemical pathways interconnect, layer upon layer, forming criss-crossing complex
鈥渃ircuits鈥�. 鈥淭he cell has layers and layers of feedback that keep things
stable,鈥� says McAdams. 鈥淚f you take account of all these effects, you end up
with such a complicated mess that it鈥檚 hard to model.鈥�

Thick skins

But the way the cell creates that stability could come to the modellers鈥�
rescue. Princeton biophysicists Stanislaws Leibler and Naama Barkai reported in
the 26 June issue of Nature that they had constructed a computer model
of the biochemical pathway that allows bacteria to sense food. When Leibler
changed the characteristics of the virtual proteins to change the speed of the
chemical reactions in the food-sensing pathway, the bacteria often remained
surprisingly good at seeking out food. Leibler concluded that, rather than
fine-tune every protein in every pathway, evolution has designed the pathways so
that they are insensitive to certain changes鈥攖he changes that might occur
naturally as genes mutate and code for slightly different proteins.

Tomita takes this as a sign that creating a virtual bacterium is a manageable
project鈥攎odellers can get away with making a few errors. 鈥淥ne can easily
make any problem unsolvable,鈥� adds Selkov, by demanding an unreasonable level of
perfection.

A cell鈥檚 ability to control its internal environment also interests
鈥渕etabolic engineers鈥�, who see minimum organisms as a new type of chemical
factory. Regular-sized bacteria like Escherichia coli have been used
for decades to make edible molecules, such as monosodium glutamate, and they are
now being genetically engineered to make industrial chemicals like
1,3-propanediol, used in polyesters.

The trick is to get the bacterium to make the right chemical, and stop it
from wasting energy and contaminating the end-product by making too many others.
One solution is to add enzymes to drive the reaction you want. But the cell,
with its vigilant environmental surveillance, usually notices and retaliates by
shutting down the overactive pathway. It can also siphon off chemicals for other
uses, or modify chemicals in unexpected ways. A minimum bacterium could be the
basis for a souped-up bacterial factory in which production lines are
ultra-efficient, and wasteful and modifying pathways absent.

Cells are designed to cope with an ever-changing natural environment.
Industrial fermenters are, by contrast, eminently predictable. Consequently, the
genes that deal with changing temperatures and irregular food supplies could be
stripped out, and genes for enzymes that make the chemical of interest added.
鈥淭he rest of the cell [would be] there for the purpose of supporting those
enzymes,鈥� says Gregory Stephanopoulos, a chemical engineer from the
Massachusetts Institute of Technology in Cambridge.

Metabolic cost-cutting can go too far, however. If having fewer genes were
always better, all chemical companies might opt to use tiny bacteria like M.
genitalium. But M. genitalium needs everything but the kitchen
sink in order to grow, whereas bugs used in industry live off simple mixtures of
salts and a sugar. This is another argument against a single 鈥渕inimum organism鈥�.
What constitutes the smallest gene set also depends on what you鈥檙e prepared to
do for the bug. Witness M. genitalium鈥攁 bacterium with a less
complicated diet is going to need a slew of extra genes to be able to
manufacture the missing nutrients.

Downsizing

Karl Sanford, a microbiologist at Genencor International in Palo Alto,
California, is kicking off his genome-reduction campaign with E. coli
and another industrial microbe, Bacillus subtilus. Sanford is not
giving much away, except that he has started by deleting the genes for the
cellular pumps and channels that let in nutrients that are not present in your
average industrial reactor.

Meanwhile, Woese is identifying the genes that are common to the growing
number of genomes from all three kingdoms鈥攖he Bacteria, other
single-celled organisms called the Archaea, and the Eukaryota鈥攖o create
yet another minimum genome, in this case an ancient one. The project is allowing
him to make educated guesses about how the ancestors common to all three
kingdoms lived. Controversially, he has concluded that RNA molecules were not
responsible for the emergence of primordial life
(鈥淟et there be life鈥�, New 杏吧原创, 6 July 1996, p 22).

Now he has shown that when it comes to DNA duplication, only the Archaea and
the Eukarya show similarities鈥攂acteria go about it in their own,
idiosyncratic way. This suggests, says Woese, that the different strategies
arose after the kingdoms split apart, perhaps because the ancestor organism did
not have the genome structure that is now common to all three kingdoms. Rather
than long stringy chromosomes, the first sort of genetic material might have
been small fragments of DNA.

The final, minimum gene set of Woese鈥檚 ancestor organism will be tentative,
and he believes the same is true for any minimum organism that could be created.
鈥淚t鈥檚 not something that you go to the top of the mountain and proclaim as the
word of God,鈥� he says. But the minimum genome researchers are confident that any
small genome will be better than none. 鈥淵ou can鈥檛 avoid it,鈥� says Overbeek.
鈥淲e鈥檙e finally going to get a sense of how the cell works.鈥�