快猫短视频

WHEN it comes to deciphering the secrets of life, the genome is just the
beginning. The difference between a heart cell that can beat out a cardiac
rhythm and a nerve cell that can help you memorise a Gershwin tune isn鈥檛 found
in your genes. Genetically, these cells are identical twins. The same genes can
build a whole range of cellular machines.

And that鈥檚 not all. The mysteries of diseases such as cancer, especially our
very individual responses to disease, will not be solved by studying genes
alone. Many of these secrets will only be revealed when we know what individual
cells are doing with that genetic information.

Each cell uses the blueprint in its genes to manufacture its own set of
proteins, which fold into complex shapes, join forces to create molecular
machines and factories, and collectively form the skeleton and inner workings of
the cell. 鈥淎 lot of what we call a cell鈥攊ts size, appearance and
function鈥攊s the work of proteins,鈥� says Emanuel Petricoin, a biochemist at
the Food and Drug Administration in Washington DC. 鈥淭hey are its essence.鈥�

And proteins are as essential to medicine as they are to biology. About 99
per cent of all drugs are either proteins themselves, or act by binding to
proteins. Maybe that鈥檚 not so surprising when you consider that disease is
usually caused by the breakdown of the cell鈥檚 protein machinery鈥攅ither an
internal failure or an attack by microbial proteins.

Until a few years ago, the sheer number and variety of proteins sustaining
each cell seemed so impenetrable that no one even thought to bestow a collective
name on them. Well now they have: the proteome. Buoyed by new technology and
information from sequencing the human genome, biologists are busy gathering the
money, tools and chutzpah they need to begin studying proteomes in earnest.

But naming the beast hasn鈥檛 made the task of taming it seem any less
daunting. The title of a meeting next month contains a hint of this post-genomic
anxiety: 鈥淗uman Proteome Project. Genes Were Easy鈥�. The idea is to paint a
portrait of the hundreds or thousands of different protein species working in
each type of cell and see how that picture changes in response to disease or
environmental influences.

Estimates vary, but human chromosomes are thought to harbour between 50,000
and 150,000 genes. In contrast, there may be millions of different proteins.
Even in a single, healthy cell, the number and selection of proteins is
ever-changing. 鈥淭he genome is static, the proteome is dynamic,鈥� says Petricoin.
鈥淚t changes based on the stage of a cell鈥檚 development, its growth and its
别苍惫颈谤辞苍尘别苍迟.鈥�

Look at how proteins are built, and it becomes clear why the proteome is so
complex. When a gene is switched on, it generates a mobile RNA copy of the DNA,
called messenger RNA. The cell鈥檚 protein factories, the ribosomes, use this mRNA
as a template to determine the sequence of amino acids that link up to make a
protein. But segments of mRNA can be removed, or spliced out, before a protein
is even made. And each 鈥渆dit鈥� creates a different protein. One extreme example
was recently discovered by fruit fly geneticists鈥攁 gene capable of
encoding 38,000 different proteins thanks to mRNA splicing
(快猫短视频, 24 June, p 20).

Even after the editing stage, there are many other transformations proteins
can go through. They may be sliced up, installed in membranes or decorated with
sugars or phosphate or any of more than 200 chemical groups. All these changes
can alter the function of the protein and therefore the workings of the
cell.

No wonder biochemists have had to spend years wringing out the secrets of one
protein at a time. But now, instead of painstakingly creating an intimate
portrait of each protein in turn, the focus has switched to taking panoramic
photographs of the shifting molecular landscapes of the cell, involving hundred
or thousands of proteins. This means the whole analytical process has had to be
vastly accelerated.

One incredibly powerful tool that proteome detectives have at their disposal
is gene sequences. Genome databases contain the sequences that encode nearly all
human proteins, and complete sets from several microbes. Biochemists can use
these to identify a characteristic mass 鈥渇ingerprint鈥� of a protein. They do this
by first using enzymes to chop up proteins into a predictable set of fragments,
then determining the size of the fragments by feeding them through a machine
called a mass spectrometer. These are then compared with the database to work
out what the protein is. When there is ambiguity the mass spec analysis can be
done again to identify a few amino acids鈥攁ll in a few seconds.

But before such an analysis can be done, a protein must be separated from all
the other proteins in the cell. Ironically, the technology on which the vast
majority of proteomics is now based is not new but an old workhorse of protein
biology called 2D gel electrophoresis. This technique involves separating
proteins in one dimension by their charge and in another dimension by size on a
square layer of a jelly-like material about 20 centimetres on each side. 鈥淲e鈥檝e
been doing this for donkey鈥檚 years,鈥� says Robert Burns of Oxford GlycoSystems in
Abingdon, Oxfordshire. 鈥淭he difference is we are going to an industrial scale,
doing for proteins what Henry Ford did with the Model T.鈥�

Drugs in the pipeline

To turbocharge their analysis, the OGS researchers automated the process. To
compare normal and diseased tissues, for example, proteins from cells of each
tissue type are separated on gels, stained with dye, photographed and then
compared by computers for any changes in protein production. Robots then cut out
the important regions of the gels, purify the protein and send it to a mass
spectrometer for analysis. OGS now analyses thousands of gels every week,
searching for proteins involved in Alzheimer鈥檚 disease, atherosclerosis,
diabetes and respiratory diseases.

While there are the whisperings of drugs in the pipeline inspired by these
methods, the approach has yet to produce a blockbuster. But some less flashy
successes have been reported. Sandra Steiner and her colleagues at Large Scale
Proteomics in Rockville, Maryland, credit the technology with providing insights
into the toxic effects of cyclosporin A, a drug widely used to suppress the
immune system after transplant surgery. It turns out that the drug causes the
loss of a totally unsupected calcium-binding protein and so allows calcium
levels to fluctuate wildly, leading to the destructive build-up of insoluble
calcium salts in kidney tubules. 鈥淭his protein was never before considered
relevant for toxicity,鈥� she says. 鈥淏ut this approach makes you look at ideas you
would never consider otherwise.鈥�

Several other proteome studies are revealing why some drugs or pollutants are
only toxic to certain tissues. For example, one study has suggested why some
substances stimulate the growth of liver tumours. Another has looked at the
toxic effect on the liver of certain viruses used to ferry genes into cells in
gene therapy. Proteome studies are also starting to show how effective
prospective drugs may be on a patient. For example, at least two groups of
researchers are using serum proteome profiles to predict the efficacy of
non-steroidal anti-inflammatory drugs (which include aspirin and many
painkillers). Studies of microbial proteomes are also starting to reveal how
antibiotics work, and so may give advance warning of bacteria developing drug
resistance.

Besides guiding the development of new drugs and understanding how old ones
work, protein markers that indicate toxicity in the body or disease could also
be of great diagnostic value. In the same way doctors monitor blood cholesterol
levels now, they may one day keep track of the levels of hundreds of proteins to
gauge the side effects of drugs or determine the general health of their
patients. Steve Martin of the Proteome Research Center in Framingham,
Massachusetts, suggests that looking into the protein core of cells may even
allow clinicians to predict which types are in danger of failing鈥攁 new
concept he calls prognostics. 鈥淲e鈥檒l be able to say you aren鈥檛 sick yet, but you
will be if we don鈥檛 do something about this protein level,鈥� he says.

William Haseltine of Human Genome Sciences in Rockville, Maryland, thinks
that monitoring large numbers of proteins will also help us find new drugs and
drug targets. His company focuses on secreted proteins, and carries out massive
screens for their biological function by growing cells in 10,000 separate
chambers and bathing each with a different protein. The cells are then rapidly
analysed for changes in the levels of mRNAs and proteins, and the information is
automatically fed into computers. 鈥淯nderstanding a protein鈥檚 effect on cell
physiology is going to be the best way to find new, interesting proteins with
medical utility,鈥� says Haseltine.

But for proteins whose work is done inside cells, Gavin MacBeath and Stuart
Schreiber at Harvard University have developed a way to do protein biochemistry
in a very small space. Taking advantage of the technology used to create DNA
chips, they developed a way to allow robots to place spots of 10,800 proteins on
a single microscope slide鈥攖wo or three of which would easily fit across
the palm of your hand鈥攚hile retaining the shape and biological activity of
each protein (Science, vol 289, p 1760).

MacBeath plans to put proteins made by all 7000 or so genes from the malarial
parasite Plasmodium falciparum on a single slide. 鈥淭hen, instead of
screening for drugs against one protein, you could screen the whole proteome
simultaneously,鈥� he says.

But before many proteins can act, they need to link up with other proteins to
form molecular machines. By working out which proteins interact with which
others, you can start to work out what those machines are, and how they can go
wrong. Last February, Stanley Fields of the University of Washington in Seattle
and his colleagues reported the biggest search yet for such protein
partnerships. Fields is famous for having pioneered a genetic screen for such
interactions. Using one protein as 鈥渂ait鈥�, they allow it to interact with a
large library of other protein 鈥減rey鈥�. If the two interact, they form an active
protein machine that gives the cell a special ability, such as a change of
colour, or the ability to grow without a particular nutrient
(see Diagram).
Using this technique, Fields鈥檚 group screened a library of all 6000 genes of the
yeast Saccharomyces cerevisiae. They report that they discovered 957
possible pairs involving 1004 proteins, almost doubling the number of known
interactions (Nature, vol 403, p 623).

Biochemist Donny Strosberg, head of Paris-based Hybrigenics, says his company
has pushed the technique even further in a study of Helicobacter pylori,
a bacterium involved in the development of ulcers and stomach cancer. They
carve up the genes for the bait and prey proteins into smaller pieces, and this
allows them to test not only if two proteins interact but which part of the
molecules are involved. If proteins were people, this would be the equivalent of
determining whether they were holding hands or exchanging kicks. Using 260
H. pylori proteins as bait, they found 1200 interactions. With this
information, they were able to find groups of proteins that seemed linked,
suggesting they cooperate to form complexes or on a biochemical pathway.

Strosberg says they then focused on proteins that formed a part of many
biochemical pathways or complexes, suspecting that their multiple roles make
them indispensable. His team has begun disrupting the genes that make these
proteins and has already shown that some of these multi-purpose molecules are
essential for cell survival, and so would be potentially excellent targets for
new antibiotics. 鈥淲e don鈥檛 get the whole story this way,鈥� says Strosberg. 鈥淏ut
we get many pathways and how they are affected by different triggers. We believe
that鈥檚 extremely powerful information.鈥�

Nevertheless, once proteomics has identified potential drug targets or
biological roles, more traditional analysis will always be necessary to
completely understand any biological process. 鈥淭hey need to work in tandem since
the level of understanding each provides is complementary,鈥� says Fields. Which
is why he thinks the full potential of proteomics will only be reached when many
of these techniques become part of the modus operandi of every biology lab.
鈥淵ou鈥檒l know when genomics and proteomics have made a lasting impact, because
science magazines will stop doing specials on them.鈥�

GIVEN the potential of proteomics, it鈥檚 not surprising researchers want to
find ways of pushing it to the limit. Craig Venter, head of Celera, the biotech
upstart that raced with the public efforts to sequence the human genome, has
vowed to build a proteomics facility with the capacity to analyse a million
proteins a day. The secret will be automation.

Part of that automation will come from a technology developed by Denis
Hochstrasser and his colleagues at the University of Geneva in Switzerland,
which dispenses with the complicated process of cutting proteins out of gels.
Instead, proteins are transferred by an electric current to a membrane studded
with enzymes, which digest and prepare them for analysis. A laser then moves
along the membrane, like an electron beam on a TV screen, to feed all the
proteins into a mass spectrometer (Analytical Chemistry, vol 71, p 4800
and p 4981).

Keith Williams, head of Proteome Systems in North Ryde, New South
Wales鈥攖he researcher who coined the term proteomics鈥攈as come up with
yet another innovation. His proteins are also transferred from gels to a
membrane, but then a 鈥渃hemical printer鈥� scans across the surface, depositing a
number of different protein-digesting enzymes and chemicals on each protein
spot. This allows more information than just the protein sequence to be derived
from each spot, such as whether other chemical groups are attached.

Another development has come from Ruedi Aebersold鈥檚 group at the University
of Washington in Seattle. The technology, known as Isotope Coded Affinity Tags,
eliminates the need for gels to compare the relative amounts of proteins from,
say, diseased and healthy cells (see Diagram).
In ICAT, proteins in the two samples are treated with two versions of a molecular tag,
each containing an isotope of a different weight, before they are digested with enzymes
and purified.

Because they are differently labelled, the samples can then be mixed and fed
into the mass spectrometer at the same time. By measuring one isotope and then
the other, the machine can distinguish between the fragments, measure the
relative amounts in each sample, and then identify the fragments (Nature
Biotechnology, vol 17, p 994). 鈥淲e can also deal with proteins that are too
big, too small, or too greasy to enter a gel,鈥� says Aebersold.