More stuff

[kai/lca12.git] / script.txt

Thank you xy for the kind introduction.
Hi! Welcome to the biology section of LinuxConf.AU. If you want to learn about
how new antibiotics are discovered, you've come to the right auditorium.

I'm going to present antiSMASH, the software I'm developing as a Ph.D. project.
It's open source software under the GNU GPLv3 (or later) and we're also running
a public instance for the scientific community to use.

But before I start talking about the software I'm working on, let me give you a
short primer on the biology side of things. Without that background, the rest
of the talk will be much harder to follow. Feel free to interrupt with
questions at any time.

As you might have seen on the first slide, I work in the Division for
Microbiology/Biotechnology at the Microbiology Institute of the University of
Tübingen, Germany. So, biotechnology, what is this all about?

The United Nations "Convention on Biological Diversity" defines biotechnology
as "Any technological application that uses biological systems, living
organisms, or derivatives thereof, to make or modify products or processes for
specific use". Quite a mouthful. But let me use a metaphor to build my
explanations on.

In biotechnology, we use biological systems such as bacteria or yeast, and then
turn them into little factories to produce things we want. A popular example
would be... beer. It's one of the oldest biotech applications on the planet. We
use a certain kind of yeast (Saccharomyces cerevisiae) to turn sugar into
alcohol and carbon dioxide. Another widespread example is the use of a
bacterium (Escherichia coli) to produce human insulin to treat people suffering
from diabetes.

Now, what's so nice about using those tiny organisms to produce these
substances instead of going for an all-chemical full synthesis? Well, the first
is that in some cases, like yeast producing ethanol, nature has already built
that functionality into the organism. It's much easier to just let the yeast do
it's thing that it would be to do the synthesis from scratch.

Using bacteria to produce human insulin is a different story. The bacteria
involved don't naturally produce insulin, they were engineered to do so.
However, there's another reason we're using biological systems to produce
things. Unlike a real factory, bacteria are self-reproducing. So if you provide
enough food, a tiny amount of starter bacteria will multiply, and then you have
a lot of little factories running your production line. This turns out to be
much more efficient than harvesting animal insulin from pigs or other large
animals.

As in a big factory, our little biofactories need machines to build their
products. In biology, these machines are called enzymes. Some of these perform
the complex chemical reactions needed to build up products. Others act as
sensors that tell the cell about it's environment. Regulators act on the input
from these sensors and allow the cell to adapt to changes or find food. Last
but not least, there's special machines that build new machines. Those are
called ribosomes, and we'll have a closer look at them in a minute.

Because living organisms need to keep up with an ever-changing environment,
nature provided them with a wide variety of tools. It would not be efficient to
keep all those machines around even if they're unused. Instead, the cells only
carry the blueprints for the vast array of machines they can build. When the
cell needs a specific machine, it will select a blueprint, copy it, and then
build the machine it needs. The biological term for such a blueprint is "gene".

Using the instructions stored in a gene, the ribosomes build up other molecules
called proteins. Proteins that perform some sort of chemical reaction are the
enzymes I was talking about a bit earlier. Ususally, if your focus is on what
the thing is made from, you'd call it a protein, and if your focus is on the
function, you'd say "enzyme". So let's have a look at how proteins are made.

As mentioned before, the instructions on how to build a protein are stored in a
blueprint, the gene. Genes are encoded on nature's universal storage system, a
molecule called "desoxyribonucleic acid", or in short DNA. DNA was discovered
in 1869 by Friedrich Miescher at the University of Tübingen, in this lab in the
basement of the Castle of Tübingen.

DNA consists of a linear backbone (the desoxyribose). This backbone carries the
actual information-containing molecules, the nucleobases or bases in short.
There are four different bases in DNA, adenine, thymine, guanine and cytosine,
abbreviated as A, T, G, and C respectively. DNA turns out to be an efficient
and robust storage for information. This is partly because in nature a DNA
strand always comes together with a backup copy, the so-called complement
strand. The complement strand is an inverse copy of the original strand, with
adenine being complemented by thymine and guanine being complemented by
cytosine. Even if only one of the strands is present, this can be used to
recover the complete set of information. The two DNA strands usually wind
arournd each other in the twisted double helix you usually see when people talk
about DNA.

In bioinformatics, you usually only store one strand because calculating the
complement strand is trivial. So all you need to store is a (potentially pretty
long) sequence of As, Ts, Gs and Cs. To give you a rough number, a virus is
about 15000 bases or 15kb in size, a bacterium is in the low Megabase range,
and a human has about 3 Gb worth of genome.

Genes are encoded with this four letter alphabet using a word size of three. In
biology these words are called codons. This means that there are four to the
power of three, or 64 possible encodings. However, nature builds proteins from
only 20 different ammino acids, the building blocks all proteins are made from.
And because it would be a shame to let the remaining 44 encodings go to waste,
multiple different codons encode the same ammino acid. This is called a
"degenerate" code and adds even more protection against changes to DNA. The
translations of codons into the corresponding amino acids often visualized in a
codon wheel, like this. Going from the center to the out side, we can see for
example that A-T-G encodes for Methionine. The three special cases are TGA, TAA
and TAG, all three telling the ribosome to stop.

Once a cell decides it needs a specific machine, it makes a copy of the
gene and sends it to a ribosome to build a new proteins The copy is made
from ribonucleic acid or RNA in short. It is similar to DNA but has some
chemical differences to the backbone and one of the nucleobases, but those
aren't really important for this part of my talk. What is important is that in
contrast to DNA, RNA usually does not come with a complement strand.  This
means that it's usually less stable, but much easier to process.

Because the RNA copy of a gene is used to tell the ribosome what to produce, it
is called messenger RNA, or mRNA. The flow of information from DNA to mRNA to
protein is called the central dogma of molecular biology. For a long time it
was believed to be the absolute rule at the foundation of the flield. Of
course, like for all absolutes, there's always an exception. Still, it's a good
rule of thumb to go by.

Blueprints that are usually read together are often stored close to each other
on the genome. These genes are said to be in a gene cluster. A common way to
illustrate how the genes are organized in a cluster is this kind of picture,
where the genes are coloured arrows. The arrow directions show which DNA strand
each gene is encoded on. Remember, DNA comes in two strands, and one is acting
as the backup copy of the other. There is no clear distinction which strand is
the original and which the backup, both strands carry blueprints and backups.

The processes required by a cell to carry on living are called the metabolism.
The metabolism is all about feeding, growing and reproducing. Central parts of
it are present in pretty much every living organism. Because living means
running the metabolism, it's going on all the time. When yeast is eating sugar
under low-oxygen conditions, any ethanol it produces is actually a waste
product. So if you're drinking a beer, you're acutally recycling what a yeast
cell would consider toxic waste.

Many microorganisms and plants also have something called the secondary
metabolism.  Opposed to the basic or primary metabolism, the secondary
metabolism deals with building up substances that are not strictly required for
living. Examples include substances like pigments that colour the petals of
flowers. If the plant would be unable to produce a pigment, it wouldn't die
right away. The same applies for the secondary metabolites that I'm interested
in professionally: antibiotics.

Many antibiotics are produced by bacteria. About 70% of the antibiotics on the
market are produced by Streptomycetes. When grown on agar plates, they form
these wrinkled colonies that often have colored pigments. Streptomycetes also
produce the molecules people usually associate with the smell of earth on a
freshly tiled field. Because these bacteria are such important producers of
antibiotics, we're focusing much of our work on them.

How do antibiotics work anyway? If we look at how a cell works, there are a
couple of key parts the cell absolutely requires to function. The cell wall,
which the cell not only needs to keep all the other parts together but also
because the electrical potential between the inside and the outside is how the
cell powers itself. Many antibiotics target the cell wall integrity. The group
of penicillin-like antibiotics is the most widespread here. Food additives like
Nisin also target the cell wall of bacteria and poke holes into it. Also, if we
remember the way the cell produces proteins, pretty much every step is the
target of some antibiotic. Quinolones disrupt the enzymes that unwind the DNA
for replication. Antibiotics like Rifampicin target the enzyme that makes the
mRNA copies. Aminoglycoside antibiotics target the ribosomes and stop them from
producing proteins. Sulfonamides inhibit some proteins in central metabolism
pathways. Remember, running the metabolism means living, so if the metabolism
stops, the cell dies.

It would be very hard to come up with substances that hit all these diverse
targets when starting a clean slate design. Fortunately, bacteria have been
waging wars against each other for countless milennia already. All we need to
do to identify new antibiotics is to screen if bacteria we have discover
inhibit the growth of bacteria we want to kill. A common way to run these tests
is by using a screening assay. In a screening assay you grow the target
bacteria on an agar plate. On that agar plate, you put little paper discs with
substances you want to test. The larger the clear inhibition zone around the
paper disk, the more effective the substance you put on the paper disk is
against the tested bacteria. On this picture from the US Center of Disease
Control, this substance is the least effective, and this substance is the most
effective.

This technique is a systematic repetition of Alexander Fleming's accidental
discovery that a Penicillium mould would inhibit the growth of nearby
Staphylococcus bacteria. Even though Fleming's discovery was over 80 years ago,
systematic bioassays are still done this way. It's probably a good idea as
well, considering that penicillin and related substances are some of the most
versatile antibiotics known, with activity against a broad range of
microorganisms.

If penicillins are so great, why do we need more antibiotics? Unfortunately,
with the widespread use of antibiotics, we have been directing the evolution of
bacteria towards antibiotic resistance. If you look at this map of europe, you
see the percentage of Staphylococcus bacteria that were identified in clinics
that were resistant to all penicillin-related antibiotics we know. Ranging from
a really low number in scandinavia, the percentage rises the further you go
south. In pretty much all of the mediterranean states, at least every fourth
patient with a Staph infection can't be cured by using penicillins anymore.
I didn't find nice visual data for Australia, but a 1999 ABC report cited a
number betwen 20 and 40 percent of the clinical Staph isolates were resistant
to penicillins. This number likely has risen in the last ten years.

How do bacteria get resistances in the first place? Some bacteria will always
carry a mutation that makes them less suspecitble to a given antibiotic. If
suddenly you speed up evolution by killing off all the more vulnerable
bacteria, you're left with the resistant ones. And because they now don't have
much competition for room and food, they thrive even better. In the end, the
average resistance level in the population has risen. That's just what's
happening in clinics all over the world since the introduction of antibiotics.

A really nasty feature in this respect is that bacteria are able to transfer
genetic materials between different species, so even if the surviving bacteria
from this example are harmless, there's a possibility that the resistance
mechanisms will be transferred to a more harmful bacterium. It is believed that
many of the more complex resistance mechanism have spread by such transfers
from the original producer of an antibiotic. Obviously, the bacterium producing
an antibiotic has to be resistant against it's own product, or it would kill
itself off.

You can speed up this process by using sublethal doses of antibiotics, which
often happens when antibiotics are misused. In countries where you can buy
antibiotics off the shelf, like in the US, antibiotics misuse is widespread.
For example, I was able to buy this tube of triple-antibiotic ointment at
Wallmart for less than three dollars. If I misused this, I'd have a good shot
at creating bacteria resistant to three different antibiotics.

Remember how the central dogma of molecular biology went? From DNA to mRNA to
protein. Now let me show you one of the exceptions I was talking about earlier.
Some bacteria and moulds have a completely different way of building proteins.
There is no blueprint for the product, no mRNA involved and the ribosome never
sees anything in the process. Instead, the cell builds a huge megaenzyme that
works just like a factory production line. Many different modules perform a
well-defined reaction. Then they get the next piece of work where they perform
the exact same reaction again, rinse, repeat.

So why the heck does the cell bother with a whole new way of producing
proteins? First of all, compared to the proteins produced by a ribosome, the
factory-made proteins can contain unusual building blocks. The ribosome is a
multipurpose machine that can deal with 20 amino acids without requiring any
changes. A module in the production line megaenzyme is specialized on dealing
with a single amino acid, but can be designed to deal with non-standard
amino acids as well.

Also, the production line approach can produce a much higher amount of product
per timeframe. While the ribosome is building up products at one step at a
time, the production line performs all the steps at every cycle. So using a
production line megaenzyme, the cell can pump out a lot of product really fast.
Because this system allows the cell to build peptides without involving a
ribosome, this is called non-ribosomal peptide synthesis. The megaenzyme is a
non-ribosomal peptide synthase, or NRPS in short.

With the biological background part out of the way, let's talk about how I'm
using antiSMASH to identify gene clusters involved in the production of
antibiotics. Remember the biology part, I'll be handing out a graded test at
the end of the talk. Sorry, giving talks in university lecture hall triggers
teaching reflexes.

antiSMASH, the antibiotics and secondary metabolites analysis shell, is a
modular pipeline that uses a host of exsiting bioinformatics tools to search
genomes for secondary metabolite gene clusters.