More Recent Comments

Friday, January 25, 2008

Junk DNA Poll

 
Just a reminder to vote in the junk DNA poll seen in the left sidebar. Check out A Junk DNA Quiz and comments for more information.
Take the junk DNA quiz in the left sidebar to let me know what you think of your genome. How much of it could be removed without affecting our species in any significant way in terms of viability and reproduction? Or even in terms of significant ability to evolve in the future? In other words, how much is junk?
It's important to register your choice now. You'll get another chance to vote on a similar topic in February and it will be fun to compare the two polls.


[Image Credit: The junk DNA icon is from the creationist website Evolution News & Views.]

21 comments :

Anonymous said...

George in Oregon

This is a hard question for me to answer. While a great deal of human DNA may have no control or encoding function and thus the sequence is not of consequence, you ask how much could be cut out without impact?

What I have no idea about, is the accomodation of the bulk. To what extent have we accomodated the useless bulk (junk) DNA such that reproduction and transcription do rely to some extent on that bulk?

TwoYaks said...

I'm remaining agnostic about this. The number of studies performing megabase or greater deletion studies is fairly limited. I'd like to see someone go through and just start to mangle the heck out of the mouse genome before I have any sort of answer.

The Monkeyman said...
This comment has been removed by the author.
The Monkeyman said...

Once you start to think of the genome in 3 and 4 dimensions, how can you possibly think consider junk DNA in terms of being well....junk? It is just unthinkable to me.

Larry Moran said...

the monkeyman says,

Once you start to think of the genome in 3 and 4 dimensions, how can you possibly think consider junk DNA in terms of being well....junk? It is just unthinkable to me.

That's because your mind is too limited. I think of the genome in seven or fourteen dimensions (depending on the day of the week) and it's obvious that junk DNA is common.

Georgi Marinov said...

Mmmmm, this question can not be answered without doing the experiment and it's exactly because of the 3D problem. There's a lot of regulation (although I don't think anybody knows how much exactly) going on at the levels of 3D nuclear structure and interactions between distant genomic regions and if you start deleting the sequences that have no function, you might end up disrupting the proper patterns of regulation even though you do not delete anything that is meaningfully transcribed...

My guess would be that you could delete all junk DNA (f course, provided that there is nothing important hidden in its vicinity and still get viable cells, but you will not be able to produce a human organisms because of those effects I mentioned

Larry Moran said...

Georgi Marinov says,

There's a lot of regulation (although I don't think anybody knows how much exactly) going on at the levels of 3D nuclear structure and interactions between distant genomic regions and if you start deleting the sequences that have no function, you might end up disrupting the proper patterns of regulation even though you do not delete anything that is meaningfully transcribed...

You make a good point.

For the benefit of lurkers, here's the gist of what George is saying.

It's well known that transcriptional regulation involves the binding of regulatory proteins to short sequences of DNA (about 10bp or less). The bound proteins then interact with each other, or with RNA polymerase, via a loop of DNA. What this means is that the DNA between the binding sites is necessary even though its exact sequence is irrelevant.

Similar considerations apply when thinking about regulation in terms of nucleosomes and nucleosome positioning. Some important sequences have to be spaced so that they are one nuclesome apart (200bp).

Furthermore, there are long range interactions that regulate gene expression. They probably have to do with the supercoiling of DNA loops and the positioning of nucleosomes. The most common examples are the silencing regions in the yeast genome but there are others as well.

All this has been known since the early 1980's. The question before us is how much DNA has to be devoted to these regulatory functions, including the spacing problem?

We know the answer in yeast because the genome is relatively compact. Less than 1000bp per gene (five nucleosomes) seems to be a good estimate.

If we assume that regulation in mammals is much more complicated (an unwise assumption in my opinion) then there could be as much as 2000bp of regulatory sequence associated with each gene. To put this into context, the average protein-encoding gene has about 2000bp of coding region (most of the exons).

There are about 21,000 genes in our genome. If each one needs 2000bp of regulatory DNA then this translates to about 1.3% of our genome.

Most people who raise this objection seem be be thinking that the percentage is much higher, but even if it was ten fold higher (an unreasonable assumption) it would still only account for 10.3% of the genome.

You can't wish away junk DNA with this sort of argument.

Anonymous said...

What about meiotic linkage between intrachromosomal genes? Recombination unlinks such genes at a rate of approximately 1% per meiosis per megabase (average human spacing per gene = 3E9 bp / 20000 genes = 150 kb, average intrachromosomal length = 3E9 bp / 23 = 130 Mb). If the human genome is shrunk by 90%, then genetic linkage increases 10x. Is the supposition that this increase in linkage would not be important "in terms of significant ability to evolve in the future"?

Georgi Marinov said...

Chromosomal looping is important for proper gene expression, it is regulated in a cell type specific manner, and it is the size and position of the loops that matter and change during differentiation. T-helper cell differentiation seems to involve reorganization of chromatin loops (SATB1) and I've seen reports for a related protein involved in differentiation of osteoblasts (SATB2). And there are examples of interchromosomal interactions regulating gene expression, especially with imprinted regions (H19 immediately comes to mind). And it might also matter where a gene is physically located in the nucleus.

Now these are still isolated examples and I don't think it's known how extensively these mechanisms are used. High-throughput sequencing (preferably with longer read lengths) will allow doing genome-wide chromosomal conformation capture analysis. We'll know the answer when we do that for different tissues and cell types.

But if you delete all the junk at the same time, it will probably affect these interactions, just because the spacing is disturbed. There are vertebrate genomes that are much smaller and much bigger than ours so it is not as if the 3D structure is fixed and extremely important, but those did not arose by sudden removal or addition of large amounts of junk. And I do not fully agree that yeast is the best organism to compare with, because yeast cells do not differentiate into a complex organism. It is junk, it has absolutely no function as information carrier, it just might be the case that the system has adapted to having bulk DNA around and might no be robust enough to still be able to regulate expression so precisely, if it's removed all at once. Again, there is no way to tell the answer without going out and deleting every major stretch of junk DNA in the genome and observing what the mouse will look like.

Larry Moran said...

anonymous asks,

If the human genome is shrunk by 90%, then genetic linkage increases 10x. Is the supposition that this increase in linkage would not be important "in terms of significant ability to evolve in the future"?

That would be my assumption. Mixus (shuffling of genes by recombination) doesn't seem to be very important. It might even be a selective advantage to avoid breaking up successful combinations.

Anonymous said...

LM: It might even be a selective advantage to avoid breaking up successful combinations.

Yes, I agree. Note however that the question addresses whether the change would be significant without regard to whether that significance represents a reduction in fitness. I take then that at a minimum we agree that increasing meiotic linkage 10-fold would fundamentally change the observable biology of meiosis? There certainly would be no difficulty detecting this phenotype experimentally. At least this potentially suggests a possible experimental handle for addressing the question of the significance of genetic distance between loci.

Larry Moran said...

anonymous asks,

I take then that at a minimum we agree that increasing meiotic linkage 10-fold would fundamentally change the observable biology of meiosis?

No, we do not agree.

I don't know what you mean by "observable biology of meiosis." I don't know what you mean by "increasing meiotic linkage." I don't know whether the total number of recombination events (crossovers) would change if we shrink the chromosome. Finally, I don't know whether any of this would have a significant effect on the species—although I doubt it.

One additional point. Recombination is not confined to meiosis. You know that, don't you?

Anonymous said...

LM:
I don't know what you mean by "observable biology of meiosis." I don't know what you mean by "increasing meiotic linkage."

Simply that the genetic distance between loci (frequency of crossing over) tends to vary in proportion to the physical distance. As you know, this is very well established. Unlike other phenomena such as chromatin looping (addressed above), the distance scale relevant to meiotic crossing over is clearly in the multi-megabase range. Compacting the genome 10x is going to have an effect.

I don't know whether the total number of recombination events (crossovers) would change if we shrink the chromosome.

Current indications are that in most organisms one crossover per chromosome is required for correct meiotic segregation, and then additional crossovers occur stochastically albeit distributed with more regular spacing that would be expected (crossover interference). The expectation is that shrinking the chromosome will change the amount of crossovers (but not to less than one), in accordance with the observation that there is a reasonable correlation between human chromosomal physical lengths and genetic lengths.

One additional point. Recombination is not confined to meiosis. You know that, don't you?

Yes. I thought the potential meiotic phenotype of decreasing the genetic length of human chromosomes would be more likely to have an evolutionarily significant impact than a corresponding somatic recombination phenotype.

Finally, I don't know whether any of this would have a significant effect on the species—although I doubt it.

Fair enough.

Larry Moran said...

anonymous says,

Simply that the genetic distance between loci (frequency of crossing over) tends to vary in proportion to the physical distance. As you know, this is very well established. Unlike other phenomena such as chromatin looping (addressed above), the distance scale relevant to meiotic crossing over is clearly in the multi-megabase range. Compacting the genome 10x is going to have an effect.

The Indian muntjac (Muntiacus muntjac) has only three chromosomes. It seems to be a typical mammal in spite of this fusion. Its close relative, the Reeves muntjac has 22 chromosomes.

We know that the smallest human chromosome (#22) is only 49 Mb so this sets a lower limit on the size of a chromsome. (The real lower limit is probably smaller.)

The average size of Indian muntjac chromosome is 1000 Mb so if we eliminate 90% of the junk DNA the chromosomes will still be 100 Mb in size. This is larger than the smallest human chromosome so it seems likely that the Indian muntjac would survive this deletion.

The same reasoning applies to the largest human chromosomes. I guess your concern is that reducing the smallest human chromsomes to only 5 Mb might cause a problem.

I still don't see why this is a problem but let's modify the original question so that we have only three or four chromosomes after eliminating all the junk DNA.

Would this satisfy your objections?

It has the added advantage of eliminating lots of centromeric and telomeric DNA so we can delete even more "junk."

Anonymous said...

LM:
The same reasoning applies to the largest human chromosomes. I guess your concern is that reducing the smallest human chromsomes to only 5 Mb might cause a problem.

I didn't say it would be a problem, I just pointed out that reducing chromosome length decreases the genetic distance between loci as well as the physical distance.

I still don't see why this is a problem but let's modify the original question so that we have only three or four chromosomes after eliminating all the junk DNA.

You mean reduce the human genome from 3000 Mb to 300 Mb, but maintain approximate mammalian-normal physical and genetic chromosome lengths by having say, 4 chromosome pairs of 75 Mb each, instead of 23 pairs of from 25 Mb down to 5 Mb?

If you do this, the diversity of gametes from random segregation of parental homologs in meiosis I is going to be reduced from 2^23 combinations to 2^4 which is a 500000x reduction.

Meiosis generates genetic diversity by both random segregation of parental alleles (dependent on chromosome number) and by recombination (dependent on genetic distance). I don't see how the genome can be reduced in size 10-fold without affecting either or both of these processes. The evolutionary significance of these changes could still be debated of course.

Rob said...

I liked the wording of the poll. The impact on the species would be interesting I think, when you get up into the high %'s.
The individual would essentially be unable to mate as it would be essentially it's own species. The chromosomes would no longer have homologous counterparts in a mate. This might be the only phenotype, I have no idea.
The species, however, would be limited in it's ability to evolve as raw material for evolution is taken away from it. For example pseudogenes would be unavailable for further mutation and potentially eventual usefulness.

Larry Moran said...

anonymous says,

Meiosis generates genetic diversity by both random segregation of parental alleles (dependent on chromosome number) and by recombination (dependent on genetic distance). I don't see how the genome can be reduced in size 10-fold without affecting either or both of these processes.

Chromosome segregation and recombination have no effect whatsoever on population diversity except that segregation is the process underlying random genetic drift.

All that recombination does is shuffle alleles so that different combinations appear in different individuals. The linkage, or lack of it, is transient. New combinations are broken up just as often as they are created. The process is called mixus.

The evolutionary significance of these changes could still be debated of course.

If there's no evolutionary significance to recombination then what's the problem?

Larry Moran said...

rob says,

The species, however, would be limited in it's ability to evolve as raw material for evolution is taken away from it. For example pseudogenes would be unavailable for further mutation and potentially eventual usefulness.

That's why we have the word "significant" in the poll question. I'm well aware of the fact that the anti-junk crowd can make this argument in favor of junk DNA. The question isn't whether it's true, it's whether it's significant.

The facts suggest not. There are very few example of junk DNA that has become useful. From a long-term evolutionary perspective a species without junk DNA will likely survive just as long as one with lots of junk DNA.

Anonymous said...

LM:Chromosome segregation and recombination have no effect whatsoever on population diversity except that segregation is the process underlying random genetic drift.

Indeed? Segregation and recombination affect the rate at which diversity is generated. This is going to matter to any population that does not have all loci in steady-state equilibrium already, including of course, humans. Are you asserting that there is no linkage-disequilibrium in the human population?

Larry Moran said...

Anonymous says,

Segregation and recombination affect the rate at which diversity is generated.

Perhaps I don't understand what you mean by "diversity."

When I use that word I mean the total number of alleles in a population. It's hard for me to see how recombination affects that value and it's even harder to see how it affects the rate at which diversity is generated.

The amount of diversity is continuously increased by mutation and continuously decreased by random genetic drift. Natural selection also decreases diversity and that works much faster than drift.

Are you referring to the hitchhiker effect where selection at one locus results in a decrease in allele frequency at a closely linked locus?

Anonymous said...

LM:Perhaps I don't understand what you mean by "diversity."

Recombination and segregation increase the rate at which genotypic change spreads through a population not already in full Hardy-Weinberg equilibrium. This is particularly true for alleles under strong selective pressure, or alleles with significant negative epistasis, but can also arise through new mutation, genetic drift, non-random mating etc. Even in the absence of selection, recombination increases the rate at which the linkage disequilibrium arising from founder effects in periodically bottlenecked populations (including humans -- see the HapMap data) is removed. This increases "diversity" in the sense that recombination (in these circumstances) increases the rate of change of overall population genotypic content with respect to earlier generations. "Diversity" is probably the wrong word in this context.

Does recombination affect the ability of a species to evolve? If "evolve" means change population-level genetic content with time, then yes (as described in the preceding paragraph). We know from classical genetics that the amount of recombination is related to the physical length of the chromosome. This being the case, reducing chromosome length by 90% is going to reduce recombination and thereby slow down evolution (of populations), regardless of the underlying mechanism of evolution (mutation, drift, selection etc.).