Transforming Men Into Mice

Hello.

Today, we'll talk about genome rearrangements, and we'll be

focusing on the question, how to transform man into mice.

Ten years ago when my daughter was just ten

years old, she asked what I am doing at work.

And I told her, I'm transforming men into mice.

She wasn't satisfied, and she asked for a clarification,

and therefore I had to tell her a story.

Men and mice

may look different, but genetically they are very similar.

For nearly every gene in man, there is a similar gene in mouse.

And, mouse of course, outperform us with respect to smell genes,

but we outperform mice a little bit with respect to brain related genes.

But overall, you can say that men and mice

have roughly the same set of genes.

But these genes are arranged differently in human and mouse genomes.

And that's why explaining to my daughter how man and mice differ genetically,

I was telling her take 23 human chromosome, cut them

into 280 pieces, shuffle these pieces and glue them together

in a new order in 20 mouse chromosomes. You will get a mouse genome.

She seemed satisfied, but she asked me if you can transform

men into mice, can you transform mice into man as well.

And I responded, of course it's very easy.

You just need to reverse this operation of cutting and gluing,

and you will get, starting from mice, you will get man.

So today we'll focus on a slightly more simple problem of

transforming mouse X chromosome into human X chromosome.

X chromosomes in mammals are special because genes do

not jump from the sex chromosome to other chromosome.

And therefore you can think about X chromosomes in mammals as

separate sub-genomes. Making it a little bit easier to analyze.

And it turns out that human and mouse X chromosome, despite

the fact that they are very long strings, 150 million nucleotides long, they can

be thought of as just sequences of 11 large segments.

Each of these segments may contain hundreds of genes

but within each segment the genes are very similar.

However, these segments, called synteny blocks,

are arranged differently in mouse and human.

And a number of questions arise. First, how can we transform a

long strand consisting of 150 million nucleotides into just

11 synteny blocks.

And what is the evolutionary scenario that nature used to transform

mouse arrangement of blocks into human arrangement of blocks.

You may notice that I show every block as a directed block, which is oriented

either to the left or to the right. And I will explain later

what precisely these directions mean, but you may recall that

two complimentary strands of DNA ran opposite to each other.

And depending on what strand a gene is located, we may assign

the gene's orientation, left or right, or plus or minus,

as we show in this slide. Now nature doesn't use this

dramatic cut and glue together operations that I described

when I was explaining the process to my daughter.

It uses a simpler operation called "reversal".

And reversal simply takes a segment of the genome and flip it over like

this, reversing the directions of full blocks within the segments.

Lets try to see step

by step of what this particular evolutionary

scenario for transforming mouse into human, amounts to.

At the first step, we simply revent the orientation of block 6.

In the next step we revert the orientation of block 9.

Then we take two blocks and

reverse their orientation, and continue, continue, continue,

until we transform the mouse gene arrangement into the human gene

arrangement on the X chromosome. I emphasize that this

is just a hypothetical scenario. Nobody knows today what

really happened during 75 million years of evolution while

nature will has been transforming, mouse gene arrangement

into human gene arrangement. But if the scenarios that I've

showed here were correct, then one of the intermediate

arrangements of blocks would correspond to arrangement of

blocks on the X chromosome of the human-mouse ancestor.

That's shown here, and we have to realize that

on the way from mouse to the human-mouse ancestor, we're actually moving in back in

time, and then from the human-mouse ancestor to human, we are moving

forward 75 million years in time. Now,

rearrangements are of course dramatic events happening within genomes.

And you can think about rearrangements as earthquakes,

because many bad things may happen.

For example, every rearrangement, every reversal has two end points.

And these endpoints, after a reversal happens, may actually

disrupt the gene, or they may bring a gene

to a completely foreign territory and put it under

the influence of the wrong transcription factor, thus disrupting gene regulations.

Now, if rearrangements can be compared to earthquakes, and we

know that earthquakes are not happening just at random points,

for example, where I live near Los Angeles earthquakes, are

common, but in most other places they would be extremely rare.

Well what about genomes?

Are there any rearrangement hotspots or fragile regions in mammalian genomes

and human genomes in particular? And also, so far, we have been talking about

evolutionary rearrangements that happened at the million-year scale.

But rearrangements also happened at much smaller scale during human development.

And we also may ask a question, do rearrangement hot spots, if they exist,

in mammalian evolutions, correlate with human rearrangement hotspots

in fragile regions in humans that we will be talking about a bit later.

Now, let's talk a little bit whether, in

the scenarios that we described, there are rearrangement hotspots.

And let's proceed step by step.

So this is how our first reversal. And there

are two earthquakes corresponding to the end points

of this reversal that are shown in this slide.

Next one, two more earthquakes.

And again every time we perform

a reversal, there are two earthquakes happening.

Note that I marked the positions of these

earthquakes and points of reversal by vertical

gold streaks, and now we have two gold

streaks in the same regions, this is a rearrangement hotspot.

And by the time we finished transformation of the mouse X chromosome

into a human X chromosome, we will actually count three regions

where there will multiply repeated earthquakes in

the corresponding areas.

Can we deduce from here that there are fragile regions in the human genome where

this rearrangements are happening over and over again? Of course not.

This is just a hypothetical scenario.

The real scenario may be completely different.

And also, statistically, we cannot make judgment based on such a small sample.

40 years ago, the prominent biologist Susumu Ohno, came up

with a random breakage model of chromosome evolution.

Ohno argued that since rearrangements are so rare, then they must occur at

random positions in the chromosome, implying that there are

no fragile regions In human genome.

Honestly, Ohno hardly had any

information to support this model. But 30 years ago Nadeau & Taylor

generated the first statistical argument in favor of the Random Breakage Model.

You may be wondering how one can suggest statistical argument supporting something

that happened millions of years ago,

without even knowing the evolutionary scenario

that describe transformation of one genome into another.

Nadeu and Taylor asked a question,

does random breakage model have predictive power,

despite the fact that according to this model,

rearrangements are happening at random places.

And they suggested the following sort of experiment.

Let's apply

N random reversals to a fake chromosome consisting

of M genes. Can we predict how many blocks of a given

length will be generated as a result of this random experiment?

If we can predict it, will this prediction fit what we observe in real genomes?

And if they do fit what we observe in real genomes?

Then we have an evidence in support of the random breakage model.

For example, what would you expect after applying N random reversals

to a chromosome? Would we expect roughly the

same number of blocks of every length, or would we expect

something like this, where the number of blocks of each length

will be variable. It turned out that we expect

something that is similar to this last slide, and it turned out

that despite the fact that reversals occur at random positions,

we can predict roughly how many blocks of each length will be generated.

For example, if we repeat this experiment 100 times,

then the average number of blocks of given size will follow this distribution.

And this distribution is very well approximated by the exponential

distribution and an approximation curve as shown on this slide.

So when Nadeau and Taylor figured out how random breakage

model would look like at the simulated example, they compared

the synteny block distribution for real human and mouse data, despite the

fact that in 1984 when their work came out, there was

very little information about the lengths of human mouse synteny blocks.

But even in this time, the lengths

of known synteny blocks fit exponential distribution quite well.

10 years later, after the Nadeau and Taylor work,

when the amount of human-mouse data on comparative

architecture increase tenfold, people built the same diagram

and saw that it fits exponential distribution even better.

So Nadeau and

Taylor prediction had almost prophetic power.

As a result, starting 1990s, random breakage model was

embraced by biologists and has become de facto theory of chromosome evolution.

And after we described the random breakage model, we will

look into algorithmic aspects of genome rearrangement, and

talk about sorting by reversals.

Comparing Genes, Proteins, and Genomes (Bioinformatics III)

Conoce a los instructores

Explora nuestro catálogo

Inscríbete de manera gratuita y obtén recomendaciones personalizadas, actualizaciones y ofertas.