More Recent Comments

Tuesday, December 15, 2009

Does Excess Genomic DNA Protect Against Mutation?

Many eukaryotic genomes have a large amount of "excess" DNA that doesn't have any of the functions we normally assign to DNA (protein-coding, regulatory, origins of replication, centromeres, RNA genes etc.). Many of us think this is junk DNA. It has no function and could easily be dispensed with.

One of the adaptive explanations for this excess DNA is that it protects the functional DNA from mutations. Ryan Gregory thinks this is a serious scientific hypothesis even though he's skeptical. He has a wonderful post that reviews the history of the idea and how the hypothesis should be tested [Does junk DNA protect against mutation?].

The bottom line is that this hypothesis is not taken very seriously by the scientific community for some very good reasons.

First, most spontaneous mutations in the germ line seem to be due to errors in DNA replication. The overall rate of evolutionary change is consistent with the mutation rate of DNA replication + repair, suggesting that it is the dominant form of mutation. This mutation rate is based on the number of nucleotides replicated. What this means is that the rate of mutation in functional DNA is independent of how much other DNA is being replicated. Excess DNA offers no protection from the spontaneous error rate of DNA replication.

THEME

Genomes & Junk DNA
However, the protection hypothesis may be applicable to other kinds of mutation such as those caused by chemicals or ionizing radiation. In multicellular organisms such as animals, fungi, and plants, this possible protection may prolong the lifetime of somatic cells or prevent them from becoming deregulated (e.g., cancer).

The idea is that excess DNA may shield the functional DNA from the effects of these mutagens but this would only work if the excess DNA was specifically organized so that it surrounded the functional DNA and provided physical shielding. There's no evidence that this is the case and, furthermore, it doesn't make much sense. The functional DNA in a nucleus is already shielded by lots of proteins, lipids and membranes so it's unlikely that a bit more DNA is going to make a difference.

Not only that, but some kinds of DNA damage caused by these mutagens will cause strand breakage. What does that mean? It means that the larger the genome the greater the chance that damage will occur. In other words, excess DNA leads to greater rates of mutation, not lower rates of mutation, for those types of mutagens. Ryan Gregory shows results from several studies during the 1970s that establish that fact.

I sympathize with Ryan's call for experimental support of the hypothesis but I'd also like to point out that not only does it not have direct evidence to back it up but it's not even theoretically feasible. It's just a bad hypothesis based largely on a misunderstanding of mutations and how they arise.

Also, the protection hypothesis doesn't pass The Onion Test which is one of the first requirements for an adaptive explanation of junk DNA.


22 comments :

agmartin said...

Could the junk DNA make it easier for DNA to be packaged into chromosomes? Perhaps the specific patterns in the coding DNA don't interact as well with the histones as the junk DNA. If this was the case then the junk DNA would only be providing space rather than information.

Ford said...

Larry, I agree, tentatively, with your stance on the nonadaptive value of "junk DNA". But it does not follow, therefore, that it could be easily dispensed with. Genome size effects such things as cell size and cycle rate which, the alterations of which could be quite profound.

Alex said...

If I recall correctly, histones associate with the DNA backbone. While base composition does subtly affect the shape of the backbone, I'm pretty sure histones associate well with any DNA helix. Moreover, many yeast species have very compact genomes, and they don't seem to have a problem with chromosome packaging.

To Ford, I'm not sure about how genome size affects cell cycle rate. Do you think that a larger genome takes longer to replicate? This isn't a problem in eukaryotes (some yeast exempted) where origins of replication occur ubiquitously. If I have misinterpreted, I apologise.

Vene said...

From what I know, genome size influences replication rates in prokaryotes, because it is a limiting factor, so it's an advantage for replication to occur as fast as possible. But, with eukaryotes, the replication rate isn't what limits generation time, so it doesn't really matter if the genome is large.

PaulM said...

Eric, your explanation wouldn't explain the large differences in genome size between similar species (see e.g.s in the 'onion test' link), nor the difference in genome size between Drosophila and humans, despite only a moderate difference in number of functional genes.

I would subscribe to Michael Lynch's idea that genome expansion reflects a non-adaptive drift process, where purifying selection fails to remove a slightly deleterious duplication.

Under this hypothesis, relatively small populations (i.e. eukaryotes vs prokaryotes; multicellular eukaryotes vs unicellular; small, abundant animals vs apex predators etc) are prone to genome expansion, where prokaryotes are not. He goes further to propose threshhold sizes at which genome expansion could occur.

DiscoveredJoys said...
This comment has been removed by the author.
DiscoveredJoys said...

Here's another hypothesis - on the assumption that non-functional DNA carries at least some extra cost of production, the amount of non-functioning DNA should be smaller in unicelled organisms because the mother cell carries the full extra cost when splitting into daughter cells. This extra cost could be critical when subject to natural selection. In the population of similar organisms each mother cell (on average) must survive to replicate if the population does'nt go extinct.

In multicelled organisms the costs per cell are the same, but the impact of natural selection on sex cells is buffered over the full range of cells in the organism. The number of sex cells surviving to replicate does not have to be 100%, merely sufficient to produce the next generation.

Once organisms contain many cells, or are particularly fecund, the deleterious effect of non-functional DNA will diminish greatly, so the amount of non-functional DNA in any population becomes more determined by chance events. Given sufficient data points you could test the hypothesis by looking for a correlation between the ratio of successful sex cells to total number of body cells vs the amount of non-functional DNA. I don't know if we have enough data points to draw probable conclusions.

Do viruses have non-functional DNA? How do you define non-functional DNA for a virus?

Divalent said...

One possibility is that it minimizes deleterious errors during crossing over. A typical chromosome will experience a couple of such events during meiosis, and the selective advantage is that it gradually unlinks genes on the same chromosome. But errors can occur at the exchange site due to a slight alignment mismatch, with one chromosome getting a bit more, and the other a bit less. If this occurs in the middle of necessary DNA, the outcome can be severe.

I’m not well read in the process, but it seems to me that if 90+% of the DNA is junk, then most of any such mismatch errors will be harmless.

Bryan said...

the amount of non-functioning DNA should be smaller in unicelled organisms because the mother cell carries the full extra cost when splitting into daughter cells

While this seems a nice theory on the surface, the data doesn't hold out. Many "simply" unicellular eukaryotes have more junk DNA than we "complex" humans. For example, the single-celled amoeba Amoeba dubia has several hundred times more junk DNA than humans do.

DiscoveredJoys said...

Huh. Back to the drawing board.

Thanks for your info Bryan.

Larry Moran said...

Divalent says,

One possibility is that it minimizes deleterious errors during crossing over.

Recombination is ubiquitous. There is an error rate but it can't be very significant on an evolutionary scale or else there would have been strong selective pressure to make the process more accurate.

It's unlikely that decreasing the mutagenic effect of recombination errors would be a selective advantage. Besides, it doesn't pass The Onion Test.

A typical chromosome will experience a couple of such events during meiosis, and the selective advantage is that it gradually unlinks genes on the same chromosome.

The idea that there's a significant selective advantage to mixus is controversial. It's been very difficult to find experimental support and modeling studies aren't very convincing. You should not assume that there is a selective advantage in your argument.

P.S. Recombination happens all the time. It is not restricted to meiosis.
 

qetzal said...

@Divalent,

I think that would only work if crossover frequency is a function of chromosome number (e.g. each chromosome has about the same number of crossovers, regardless of size). However, my recollection is that crossover frequency is more a function of chromosome length. If that's true, adding junk DNA would proportionately increase the total number of crossovers, such that the same number would still take place within 'required' sequences.

However, I'm not well read in meiotic crossover frequencies either, so I could be wrong.

Larry Moran said...

Ford says,

Larry, I agree, tentatively, with your stance on the nonadaptive value of "junk DNA". But it does not follow, therefore, that it could be easily dispensed with. Genome size effects such things as cell size and cycle rate which, the alterations of which could be quite profound.

I understand the distinction but it's important to distinguish between adaptive effects and non-adaptive effects that are just unintended consequences of having lots of junk in your genome.

For example, it may be that having more DNA leads to large nuclei and large cells. This does not mean that there was selection for large cells. If removing excess DNA results in smaller cells then this only counts as "profound" (IMHO) if there are significant consequences. If there are significant "bad" consequences then this is an adaptationist argument for larger genomes.

If the result isn't "profound" then the DNA was probably junk.

BTW, I don't think that large genomes have much effect on the cell cycle as was pointed out by others.
 

Divalent said...

The “Onion Test” is flawed, because it implies that any function of junk DNA *also* has to also make sense as a reason for the existence of junk DNA. This implication is unwarranted. A true pluralist would recognize that. :-)

Perhaps junk DNA arose for non-adaptive reasons (ERVs gone wild, etc) or for an adaptive reason not yet elucidated, but regardless of why it exists, it now is part of the genetic landscape and a factor in many cellular processes. And so quite visible to natural selection.

Recombination can accomplish for genes located on the same chromosome what sex accomplishes for genes located on different chromosomes: producing new gene combinations. I assume that if sexual recombination is advantageous, so is crossing over.

Given that 90%+ of our DNA *IS* junk, it is trivially true that for any given error rate, the presence of junk DNA means that we have fewer deleterious mutations occurring than would be the case if there was no junk DNA (for the same number of crossing over events per chromosome). Yes, selective pressure could lower the error rate for crossing over events, but given that 90% of our DNA *IS* junk, the error rate *could* be 10 times greater than the error rate in a similar species with no junk, and yield the same overall mutation rate.

In other words, whatever selection pressure these errors exert on the process that controls crossing over, that selection pressure is lower because of the presence of junk DNA.

Larry said: It's unlikely that decreasing the mutagenic effect of recombination errors would be a selective advantage.

I’m curious. Why do you say that? Given the nature of the errors (unlikely to be beneficial or even neutral if they occur in a coding region), I would assume that there would always be a selection pressure against such errors. (It may be a small advantage, but I don’t see it ever being zero).

Ford Prefect said...

http://jcs.biologists.org/cgi/reprint/34/1/247.pdf

Here is a link to an older paper by Cavalier-Smith which discusses the effects of c-value on cell size and cycle length, even in eukaryotes.

qetzal said...

Divalent:

"Given that 90%+ of our DNA *IS* junk, it is trivially true that for any given error rate, the presence of junk DNA means that we have fewer deleterious mutations occurring than would be the case if there was no junk DNA (for the same number of crossing over events per chromosome). "

Except that the number of crossovers per chromosome does *NOT* appear to be the same. It appears to be directly proportional to chromosome length.

I plotted genetic length (in centimorgans) vs. physical length (in megabases) for all human chromosomes (using data at http://www.ncbi.nlm.nih.gov/SCIENCE96/). They're very tightly correlated: cM = 0.98Mb + 23, R^2 = 0.957.

Longer chromosomes have more crossovers than shorter ones. This implies that going from a junk-free chromosome to junk-filled one that's 10 times bigger will just increase the number of crossovers about 10-fold as well. The chance of a crossover within the essential sequences wouldn't change.

Doesn't disprove your idea, obviously, but makes it pretty unlikley, IMO.

Anonymous said...

I assume that if sexual recombination is advantageous, so is crossing over.
Crossing over is mechanistically required in meiosis to both identify parental homologs (only appreciated in the last 10 years or so) and for correct mechanical separation of homologs (known for a long time). The genetic reshuffling may (or may not) be an unintended consequence of the mechanics of meiosis.

Except that the number of crossovers per chromosome does *NOT* appear to be the same. It appears to be directly proportional to chromosome length.
This varies by organism. See C. elegans for a counterexample.

The idea is that excess DNA may shield the functional DNA from the effects of these mutagens but this would only work if the excess DNA was specifically organized so that it surrounded the functional DNA and provided physical shielding.
For practical purposes, this is exactly the manner in which junk/non-junk is organized, with genes surrounded by large stretches of non-genes both two-dimensionally along chromosomes and in 3-dimensional space with higher-ordered chromatin packaging in the nucleus.

The functional DNA in a nucleus is already shielded by lots of proteins, lipids and membranes so it's unlikely that a bit more DNA is going to make a difference.
Proteins, lipids and membranes don't provide any shielding against 260 nm UV light for example.

Alex said...

Well, Ford, that paper is too big for me to be interested for long-- I just skimmed the first couple pages. This is something an expert like Ryan Gregory should pick up on. I'll note that one of his hypotheses is that C-value controls cell size, and cell size controls cell cycle rate, thus the limiting factor is for a cell to get to the appropriate size for division. However, my earlier point is still valid in that the synthesis step of the cell cycle probably doesn't vary enormously among eukaryotes.

Larry Moran said...

Anonymous says,

Crossing over is mechanistically required in meiosis to both identify parental homologs (only appreciated in the last 10 years or so) and for correct mechanical separation of homologs (known for a long time).

Interesting. How do male Drosophila get along without crossing over?

Crossing over is also common during mitosis. Why?

Be careful about extrapolating results from the yeast Saccharomyces cerevisiae to other species.
 

Larry Moran said...

Divalent says,

The “Onion Test” is flawed, because it implies that any function of junk DNA *also* has to also make sense as a reason for the existence of junk DNA.

Ryan Gregory brought up The Onion Test as a reality check for those people who propose adaptive explanations for the origin of junk DNA. That's a perfectly valid test.

Given that excess DNA exists, it's interesting to ask whether it has consequences and whether those consequences are significant. I don't accept that "consequences" are the same as "function." When you say that something has a "function" you are implying that the function is beneficial and useful and if that's true then it amounts to an adaptive explanation for its origin.

Perhaps junk DNA arose for non-adaptive reasons ... it now is part of the genetic landscape and a factor in many cellular processes. And so quite visible to natural selection.

This makes no sense. If excess DNA is "visible to natural selection" then it will either be eliminated because it's deleterious or enhanced because it's advantageous. Presumably you think it's beneficial in which case it would have been selectively advantageous for an organism without excess DNA to acquire it, no?

Recombination can accomplish for genes located on the same chromosome what sex accomplishes for genes located on different chromosomes: producing new gene combinations. I assume that if sexual recombination is advantageous, so is crossing over.

It's not clear whether sexual reproduction is advantageous (prokaryotes don't do it very often) and even if it is, it's not clear whether mixus due to recombination is one of the advantages. This is a very controversial field and that's why sex is still considered one of the most important problems in biology.

You can postulate that mixus is advantageous and then go on to assume that crossing over is also advantageous but be careful about assuming that something is proven when, in fact, it is simply a hypothesis without strong experimental support.

Given that 90%+ of our DNA *IS* junk, it is trivially true that for any given error rate, the presence of junk DNA means that we have fewer deleterious mutations occurring than would be the case if there was no junk DNA (for the same number of crossing over events per chromosome).

No, it is not "trivially true." It's only true is you also make the assumption that the number of crossovers per chromosome is constant. If, instead, it is correlated with the amount of DNA then increasing the size of the genome simply means mor cross-overs.

Please be more careful about your assumptions. You may be correct that the number of crossovers is constant but you must include that assumption in your statemetns before you declare something to be "trivially true."

Yes, selective pressure could lower the error rate for crossing over events, but given that 90% of our DNA *IS* junk, the error rate *could* be 10 times greater than the error rate in a similar species with no junk, and yield the same overall mutation rate.

So how do you account for those millions of species that don't have 10X as much junk in their genomes? Are they just happy with a higher error rate?

In other words, whatever selection pressure these errors exert on the process that controls crossing over, that selection pressure is lower because of the presence of junk DNA.

See above.

Larry Moran said...

Divalent says,

I said that it's unlikely that decreasing the mutagenic effect of recombination errors would be a selective advantage.

I’m curious. Why do you say that? Given the nature of the errors (unlikely to be beneficial or even neutral if they occur in a coding region), I would assume that there would always be a selection pressure against such errors. (It may be a small advantage, but I don’t see it ever being zero).

I say that because there are many species with very little junk DNA—most prokaryotes, for example, and species such as yeast—and these species don't seem to be hurting because of the high error rate of recombination.

Anonymous said...

Interesting. How do male Drosophila get along without crossing over?
It's a question about which I ask my students to speculate. My guess is that with only 4 chromosomes, there is less of a homolog identification problem in Drosophila than in other species where an increased number of chromosomes is a greater combinatorial problem to solve with respect to meiotic pairing. Pombe, with only three chromosomes also seems to have less of a requirement for crossing-over, than for example, cerevisiae with 16 chromosomes, or mice with 20. C. elegans, with 6 chromosomes is an interesting intermediate case.

Crossing over is also common during mitosis. Why?
Probably an accidental result of recombination-mediated replication fork restart. The current view is that crossover recombination in eukaryotes is greatly suppressed relative to non-crossover recombination. You can look at what happens in humans when crossover suppression is relaxed by, for example, loss of the Bloom syndrome protein.

Be careful about extrapolating results from the yeast Saccharomyces cerevisiae to other species.
I completely agree with this admonishment.