The poll is now closed and here are the final results (see sidebar on the left-hand side of this page). The question was "How much of our genome could be deleted without having any significant effect on our species?"The results are surprising to me. I would have thought that a far higher percentage would have voted for 50% or more. As it turns out, half of you think that 50% of our genome is essential. That's not right.
Here's the plan. Over the next few days I'm going to try and post a bunch of articles on the composition of our genome. I'll try and explain why most of it is junk. Then I'll re-do the poll to see how many I've convinced to change their minds. Deal?
If anyone else wants to join in, send me the link to your posting and I'll put it on Sandwalk.
60 comments :
Scrambling perhaps, but 90% DELETED could certainly have an effect. Remember, polyploidy, as in 100% addition, frequently produces phenotypes.
In many clades it holds that the duration of meiosis is predictable by the amount of DNA, regardless of what you may argue as biochemist ; and then, of course, we have the prediction of genome size and from cell size in cross-species comparisons.
I agree with neutralism, you should know that, but MERE BULK of DNA is the issue for me. I find it very difficcult to think everything in the cell will be the same with 10 times less DNA.
A lot of the junky DNA may be necessary simply as spacer, to preserve regional patterns of expression (between insulator elements), or to promote reliable packing during chromosome condensation. The specific sequence might not matter, even if the mass per se does matter. We might call such DNA "crap" DNA -- you need some crap in there, but it dnesn't really matter what crap it is. What we need is a minimal mouse that, like the minimal E. coli described in Science last year, has as much DNA as possible removed. It will be quite a while before we have good answers, but we really don't understand yet how important large-scale chromosome structure is (or is not) in metazoan biology. If we can say one thing about biology, it is that simplifying assumptions are usually proven wrong once the data come in.
I would have voted for >50%, but it's a tricky to narrow it down more than that. For example, one could ask, how much of our introns could be completely deleted without adversely affecting alternative splicing? More than 35% of our genome falls into currently annotated introns; it's hard to say how much of that 35% could be deleted without serious problems.
According to this, humans have about 3.50 pg of DNA; the bent-winged bat has about 1.7 pg. Going on the assumption that mammals in general have about the same amount of functional genes and regulatory sequences, this suggests that the amount of junk DNA is at least 50%.
Look forward to the posts, particularly the ones that will categorically demonstrate that loss of X% of our "junk" DNA will not effect us as a species (this to me suggest a generational element not included in the "if we deleted 75% of the crap it would not cause a phenotype in the individual" approach).
Maybe I didn't read the rules carefully enough. Like sanders and perhaps others, I balked at the word 'deleted.' Your posts, as fantastic as I'm sure they'll be, are unlikely to convince me otherwise. (Though I'll read carefully.) 50% of the human genome is indeed evolutionary debris, but that doesn't mean that deleting it from a cell would be without deleterious effects.
But I look forward to your posts, and will cite them in my own series which is targeting specific creationist misinformation.
I thinks it's important to specify what is meant here by essential. Does it mean for example, a stretch of DNA that encodes an essential gene(s)? Or could it also include regions of DNA that serve as a scaffold for the assembly of critical regions of chromatin structure? What if a piece of the genome has absolutely no function at all in a cell at the present time, but is essential for genes to shuffle around and therefore essential for the ongoing process of genome evolution?
It's also quite possible for the criticalness of a given stretch of the genome to be context dependent. An example might be a structural region of a chromosome that is essential in the natural configuration but could be eliminated if you designed an artificial chromosome/genome from scratch.
I look forward to seeing any article detailing what happens to viability when you delete most of the DNA from a mammalian organism. That was the question you asked us, wasn't it?
I can accept that over 95% of individual bases might be unnecessary (on a base by base deletion experiment) but deleting them all at once is entirely a different matter. The mammalian genome is not a simple linear string of genes, its normal functioning frequently relies on three dimensional interactions between regulatory regions and distant targets that would be disrupted by removing the intervening sequence (whether this intervening sequence is 'junk' or not).
Thanks Larry. I'm looking foward to the artciles. I'll try and follow them.
anonymous says,
A lot of the junky DNA may be necessary simply as spacer, to preserve regional patterns of expression (between insulator elements), or to promote reliable packing during chromosome condensation.
I intend to address that question in an upcoming post. Before I do, why don't you give me your estimate of the amount of DNA that you think serves this role? Keep in mind that the smallest vertebrate genome (pufferfish) is only 10% of the size of the human genome and 50% of our genome is known to consist of degenerate transposons.
mike says,
For example, one could ask, how much of our introns could be completely deleted without adversely affecting alternative splicing?
We know what determines splice sites. The sequences are usually 10 bp or less. That means that 90% of most human introns is non-essential. For individual introns we can ask whether the lengths are similar in different species or whether they drift all over the map.
The point is that a great deal more is known about these things than the average person assumes.
More than 35% of our genome falls into currently annotated introns; it's hard to say how much of that 35% could be deleted without serious problems.
It is surprisingly difficult to estimate the amount of our genome found in introns. Part of the problem arises because of different methods of making gene predictions. The Celera team, for example, throws in all kinds of small exons that are far away from the main body of the gene. This gives rise to predictions of extremely long introns at each end of the gene (especially the 5' end). Most of these are artifacts, in my opinion.
If we confine our estimates to well-characterized genes, there are 7.1 introns per gene with an average length of 3750 bp. This gives a total of 553 Mb of intron DNA or 17% of the genome.
It is not that hard to say how much could be deleted since these intron sequences contain the same kinds of junk found in intergenic regions. I'd say that >90% is dispensible.
The average intron size in Drosophila is 800 bp and in Arabidopsis it's 160 bp. Does it seem reasonable to you that mammals would have a evolved a system where they need five times more in their introns than a fly?
George in Oregon
I selected 0%. But not because I disagree that a very large portion of human DNA has no sequence importance for coding RNA or controlling transcription.
Rather, I just do not know how that you could cut it out without disrupting replication and transcription. That is a bulk physical impact.
If the question was worded how much of human DNA has a sequence that is necessary, I would say about 2%.
the lorax says,
Look forward to the posts, particularly the ones that will categorically demonstrate that loss of X% of our "junk" DNA will not effect us as a species
With that kind of attitude, you'll be wasting your time reading Sandwalk.
Are you the kind of person who won't accept evolution unless you see a detailed step by step history of every species from 3 billion years ago to the present time?
George in Oregon says,
I selected 0%. But not because I disagree that a very large portion of human DNA has no sequence importance for coding RNA or controlling transcription.
Rather, I just do not know how that you could cut it out without disrupting replication and transcription. That is a bulk physical impact.
You are exactly the kind of person I'd like to persuade. We actually have a pretty good idea of what is required for transcription and DNA replication. Most of the objections to junk DNA are just uninformed hand waving and wishful thinking.
We know what determines splice sites. The sequences are usually 10 bp or less. That means that 90% of most human introns is non-essential. For individual introns we can ask whether the lengths are similar in different species or whether they drift all over the map.
I know I should probably wait for your posts on the subject, but this particular comment looks oversimplified, from my (admittedly superficial) knowledge of splicing mechanics.
The slice regulatory motifs, which I asume is what you refer to, are indeed small. But how they work depends on their structure -- they need to be single stranded, and if you move them, they stop working. The other bits of the intron are important in determining that. Some of the bases in those sections can be swapped and scrambled without any effect, but deletion would have a big effect. One could perhaps engineer or evolve a gene with miniature introns, but merely cutting the
Perhaps the disagreement over the effect of deleting junk DNA is just because we're imagining the thought experiment differently? You bring up the examples of organisms that have much smaller genomes, which is fine, but their non-junk DNA has been evolving in the context of the junk DNA-lite genome, while ours has evolved in the context of a junk DNA-rich genome. So whether "deleting" our junk DNA could produce a viable result perhaps depends on whether we're deleting it in a single generation, or over many generations, allowing the rest to catch up.
But again, you've probably given this much more thought than me, and I should have just waited.
Larry, just because a sequence was once a 'degenerate transposon' doesn't mean it will always be useless junk. The lactose tolerance locus, for instance, is an example of a repeat element that has taken on new function down the evolutionary line (line, geddit?). There are undoubtedly many more such regulatory segments buried within apparent repetitive derived sequence that will become apparent when mass snp analysis is performed on a large scale.
MartinC says,
Larry, just because a sequence was once a 'degenerate transposon' doesn't mean it will always be useless junk.
Nobody is denying the fact that some former junk DNA can accidentally evolve into something that has a function.
But that cannot be a reason for declaring that all junk DNA has a function. It cannot be a rational argument against junk DNA because, if it is, then nothing can ever be junk as long as it has even the remotest chance of being converted into something useful over the next few billion years.
That doesn't make sense.
Once we identify a broken gene, whether it be a former transposon or any other type of gene, then it's perfectly reasonable to declare that that pseudogene is junk. It serves no useful purpose in the genome and can be deleted without causing any harm to the individuals who lack it.
If you want to argue against that logic on the grounds that there is some minuscule chance that the gene will be accidentally reactived at some time in the future then we can't have a rational discussion about genome organization.
Larry, you are not very convincing, to put it mildly.
The phenotype is in part a result of the effect of higher-level genomic architecture on gene expression ; chromatin & histones, complexed, compacted or not....others have mentioned the importance of arrangement and relative position of different components affecting gene expression. All of these mechanisms have been shown to have heritable effects on the phenotype.
Can you get the same architecture if you delete 90% of that junk DNA fill-in material? I doubt it. Would you have an altered phenotype? Quite probably.
To bypass architecture and arrangement,as if the small percentage of functional genes would give a phenotypic result regardless of their ordered position in genomic structre, is absurdly reductionism of the pea-brained kind.
You need to stop pontfying and start getting a litttle bit more conving, Larry. How about a REAL explanation about your DNA replication argument?
You know, many calculations on paper just crackle in the fire of the "in vivo" evidence.
"Nobody is denying the fact that some former junk DNA can accidentally evolve into something that has a function."
I like this quote. Is the term "accidentally" to be understood in contrast to all the stuff that evolved non-accidentally? :) Just teasing.
Larry,
I know you haven't made the full case yet, and I'm interested to read it all.
But I have to side with the other guys on the issue of degenerate transposons and pseudogenes and the like. These are not "junk DNA", these are spare parts. Surely Larry you can agree that there is an important difference between the two?
Larry,
I certainly wouldn't claim that all or even most (or even a high percentage) of junk DNA has a specific function in the way the lactose tolerance locus functions. It demonstrates, however, that one cannot simply dismiss all transposon derived DNA as being functionless. Perhaps we can agree that while most transposon derived DNA has no specific function, some of it has and these functional sequences - many of which WOULD have a significant effect on the viability of humans if deleted - have not been completely discovered yet. I would hazard a guess that it is the nonspecific function of junk DNA that is causing people to come to a different conclusion to you regarding the percentage of the genome you can delete.
Perhaps we can turn the question around. Are there any mammalian species with a more minimal genome, who have acquired a rampant transposon based increase in junk DNA (doubling their DNA for instance) without affecting their viability?
Perhaps it might even be an experiment that its possible to perform now using a model small genomed vertebrate such as Fugu (knock out the piwi pathway and see what happens).
Sanders says,
Larry, you are not very convincing, to put it mildly.
The phenotype is in part a result of the effect of higher-level genomic architecture on gene expression ; chromatin & histones, complexed, compacted or not....others have mentioned the importance of arrangement and relative position of different components affecting gene expression. All of these mechanisms have been shown to have heritable effects on the phenotype.
None of this is news to me. For over thirty years I've been teaching about and arguing that chromatin is important. I'm delighted that everyone has come over to my side.
But the question before us is how much DNA is required for this kind of regulation. I'll be addressing this issue soon but while we're waiting, why don't you give us your estimate?
Remember, there are 20,500 genes in the genome. How much regulatory DNA is required, on average, for each one? Would it be 1000 bp, or 0.6% of the genome? Would it be 10,000 bp, or 6% of the genome, or could it be 100,000 bp (60% of the genome)?
Please give me a number so we can discuss it. Handwaving arguments about regulatory DNA don't cut it unless you put your money on the table.
Most of the available evidence says that regulatory sequences are less than 1000 bp. Do you have a different number in mind?
martinc says,
I certainly wouldn't claim that all or even most (or even a high percentage) of junk DNA has a specific function in the way the lactose tolerance locus functions.
Good. Because it's very clear that a large percentage of our genome has no specific function. I'm glad we can agree on that.
It demonstrates, however, that one cannot simply dismiss all transposon derived DNA as being functionless.
We agree on that as well. I was shocked when you decided to challenge me for reminding another commenter to take into account the fact that 50% of our genome is composed of degenerate transposons.
Let's agree that we both know something about this subject, OK? You will assume that I already know about the small amount of transposon-derived DNA that has acquired another function. You don't need to keep bringing this up as though it were an important point.
Likewise, I will assume that you know about pseudogenes and fragments of transposons that litter out genome. I will assume you know that these are junk.
Perhaps we can agree that while most transposon derived DNA has no specific function, some of it has ...
Yes. We both agree that when DNA has a function it is not junk DNA.
... and these functional sequences - many of which WOULD have a significant effect on the viability of humans if deleted - have not been completely discovered yet.
Yes, you can assume that I'm not an idiot. There are certainly some parts of our genome that we haven't yet learned about. The question is how much falls into this category?
I would hazard a guess that it is the nonspecific function of junk DNA that is causing people to come to a different conclusion to you regarding the percentage of the genome you can delete.
I'm sure you're right. At the very least I hope to stimulate people to propose something concrete instead of just advancing fuzzy wishful thinking about the topic.
What kind of function are they actually thinking about and what evidence do we have it is important? Most of the specific examples that have been given can be shown to be incorrect.
I suspect that the biggest problem for people who disagree with me is that they have a hard time accepting the concept that our genome may be incredibly sloppy and poorly designed. Most people tend to think that everything in biology has a purpose. To them, the very term "junk" DNA contradicts their understanding of evolution.
bayman asks,
But I have to side with the other guys on the issue of degenerate transposons and pseudogenes and the like. These are not "junk DNA", these are spare parts. Surely Larry you can agree that there is an important difference between the two?
You need to be very careful with this argument. Are you suggesting that species with large genomes are better at evolving than those with small genomes? Are you trying to make a case for function or purpose of junk DNA or are you just pointing out one of the potential accidental benefits for the species of having something that is useless for the individual? Are you talking about a spandrel? Or an epiphomenon? Or real function that has been selected?
Unlike others, I am not doubting that most intron sequence is junk. I do not doubt that trasposons are mostly junk and that ony a small percentage of them have been co-opted to a certain function. That is, as far as sequence composition or "information" with any possibility of producing is very very low. As I said, if you had said "scrambling", I'd go for values >90%
But then, how much DNA is required for the higher level genomic architecture involved in regulation? Given that there are many such higher level mechanisms, and most importantly, because we are still learning about these mechanisms, estimates are made through a very murky glassball indeed, whether it's you or me who is making them.
For what it's worth, my estimate was 50-75%. I may perfecty be wrong. I think it may be too much; you certainlyl fail to convince it is too little.
You own brand of handwaving is making great displays of confidence and palying high and mighty, but how about acytually laying out the argument?
To begin with, I'm pretty sure there would be different ways of removing DNA, different genomic regions, where the effect of deleting would be more disturbing than others, in terms of modifying the genomic architecture.
Sanders said,
For what it's worth, my estimate was 50-75%. I may perfecty be wrong. I think it may be too much; you certainlyl fail to convince it is too little.
You own brand of handwaving is making great displays of confidence and palying high and mighty, but how about acytually laying out the argument?
Back in 1989 we decided to do an experiment to find out. We took the regulatory region from a well-known gene and fused it to a reporter.
Then we created transgenic mice to see if the regulatory region was sufficient to regulate the genes properly in various chromatin contexts [Kothary et al. 1989]. About 800 bp was sufficient to drive synthesis of the hybrid gene no matter where it integrated in mouse chromosomes.
Since then lots of similar experiments have been done and the results mostly indicate that the regulatory regions are found in less than 1000 bp at the 5' end of the gene.
Your turn.
Or, regulatory regions may be a million bp away (Dev Dyn. 2005 232:345). Shh could be a genuinely interesting gene to look at under various chromatin contexts (rather than genes we already know are stuck to regulatory sequences).
Indeed, very often regulatory sequences are found in inmediate vicinity of genes, so I can understand you saying this is the majority of cases, though a quantification or calculation of some kind would be be much better
But regardless of that argument, there are other considerations. For instance, those higher level aggregates of DNA and protein that are chromosomes themselves. You can chop or fuse the DNA into a different quantity of chromosomes and your possibilities of combining phenotypic traits would change.
The chromosome level is necessary to understand genetics, beyond the "operon lac" level. What do you think would happen to chromosome structure with a >90% deletion? do you think that would be inconsequential? I would be surprised if the cell manages to make chromosomes bulky and normal enough to divide in regular cytoplasmic "machinery".
I know fugu is champion of reductions, with a 90% reduction compared to other teleosts, but is there any evidence that this happenend in a single step, and without significant phenotypic change? I'll have to look into that, I can't download Ventakesh's paper on it right now, rats.
We must consider that observations made in some clades do no repeat well in others; for instance, polyploidy may have large phenotypic effects in one clade, but not much in another. The amount of DNA may correlate well with meiosis time in one clade, but not another. So I tend to think that losing junk may have a great phenotypic effect in one clade, but not in another.
I do not know why would that be so, and much less what kind of clade do humans belong to.
About introns: If you take the current total bp for the RefSeq genes at the UCSC Genome browser and subtract out the exons, you get about 35% of the human genome in introns.
I agree that most of that sequence is probably dispensable, although we still don't know the extent of regulatory elements in introns that control alternative splicing. Comparisons of intron sizes with other animals, like you suggest, imply that much of this 35% is not absolutely necessary for function.
Sanders says,
Indeed, very often regulatory sequences are found in inmediate vicinity of genes, so I can understand you saying this is the majority of cases, though a quantification or calculation of some kind would be be much better.
Well, I'm not going to do any "quantification" by combing through the literature and adding up all the examples.
I tend to rely on what makes sense from an understanding of the basic concepts. If there are a number of experiments that confirm that understanding then I have more confidence that it's correct.
In this case we know that in the vast majority of cases regulatory proteins bind to specific DNA sequences and they control the rate of transcription. They do this by interacting with the transcription initiation complex or by making other binding sites sites accessible (or inaccessible). In oder to accomplish this task they have to simultaneously bind to their DNA binding site and something else, forming a DNA loop. (There are other possibilities but it doesn't change the basic theory.)
The basic theory comes from studies of the lac operon. We know that the loop can't be too large or you lose all the advantages of the system (i.e. increasing local concentration).
Besides, if binding sites that are far away could activate transcription, then there would be a lot of spurious transcription since all specific binding proteins have a fairly high degree of non-specific binding. If anything bound within one million base pairs has an effect then at any given time there will almost certainly be some regulatory protein bound to that DNA (non-specifically).
In eukayrotes, the same rules apply, although they are complicated a bit by the presence of nucleosomes. Thus, in addition to experimental evidence, we have excellent theoretical reasons for expecting most regulatory sequences to be closely linked to the genes they control. There are always exceptions, of course, but you don't make general rules out of exceptions.
It seems quite reasonable to assume that you only need about 1000 bp per gene for regulation. That's less than 1% of the genome.
Hmm, Sander's has mentioned this briefly, but in a bit more detail;
Interfering with mitosis will lead to a cell line that is unable to divide properly, and interfering with meiosis leads obviously to not being able to reproduce.
So Larry might well be right in saying that 90% of the human genome can be deleted and still have a cell that functions, but I'm fairly sure that said cell wont be able to undergo mitosis and meiosis without somehow rebuilding the systems that are vital to cell division.
Mind you, this is recalling under-grad molecular genetics from two years ago and my text books aren't proving all that helpful on the subject...
Anyhow, I agree more with Sanders on this, with reservations about the impact of deleting 50-75% of the genome. But I'm looking forward to Larry's posts on this, it should be most enlightening.
Nick says,
So Larry might well be right in saying that 90% of the human genome can be deleted and still have a cell that functions, but I'm fairly sure that said cell wont be able to undergo mitosis and meiosis without somehow rebuilding the systems that are vital to cell division.
Why do you say this? Why are you "fairly sure" that cells can't divide with less DNA? What did they teach you in undergraduate molecular genetics?
Centromeres and telomeres aren't junk so they won't be deleted. Would it help if I changed the question to allow for construction of new chromosomes with the remaining 10% DNA? How about if we make only three medium sized chromosomes? Would that ease your mind?
LM:Would it help if I changed the question to allow for construction of new chromosomes with the remaining 10% DNA? How about if we make only three medium sized chromosomes? Would that ease your mind?
This doesn't ease my mind. (see comment 21 on http://sandwalk.blogspot.com/2008/01/junk-dna-poll.html)
Two points I'd like specifically addressed in future posts on the subject of Junk in the Genome:
1) What are your views on decreasing the physical length of the human genome decreasing the genetic length of the human genome? Will decreasing meiotic recombination and/or meiotic chromosomal random segregation have evolutionarily significant consequences in terms of decreasing the rate of removing linkage disequilibrium and decreasing the rate of spread of alleles under selection?
2) I'd like you to address, if you would, Larry Chasin's work on the issue of Alu elements in the evolutionary generation of new conditionally expressed human exons. Specifically, I'd appreciate your perspective on:
Proc Natl Acad Sci U S A. 2006 Sep 5;103(36):13427-32.
Comparison of multiple vertebrate genomes reveals the birth and evolution of human exons.
Zhang XH, Chasin LA.
( http://dx.doi.org/10.1073/pnas.0603042103 )
Where they conclude:
This work suggests that highly repeated sequences, rather than being parasitic invaders and junk, play an important evolutionary role in the evolution of new genes. The documentation of a number of Alu exonization events led Sorek et al. to propose that exaptation of Alus may have "promoted speciation of the human lineage." Our data support this idea and extend it to additional classes of repeats and to other mammals.
For me, these are the stickiest two points to address in the genome reduction discussion.
anonymous asks,
1) What are your views on decreasing the physical length of the human genome decreasing the genetic length of the human genome? Will decreasing meiotic recombination and/or meiotic chromosomal random segregation have evolutionarily significant consequences in terms of decreasing the rate of removing linkage disequilibrium and decreasing the rate of spread of alleles under selection?
My view is that chromosome 22 at 49 Mb works just as well as chromosome 1 at 247 Mb. That's already a five-fold difference. Therefore, the length of chromosomes is not important.
My view is that the frequency of shuffling genes (mixus) is irrelevant. There's no reason to think that Homo sapiens enjoys an evolutionary advantage over yeast (no junk DNA), or Drosophila melanogaster (~5% the size of the human genome), or Arabidopsis (~4% of the human genome), or pufferfish (~10% of the human genome).
Why do you think this is important? There's no theoretical reason to believe that the rate of random shuffling has an effect on evolution and, as far as I know, there's no scientific evidence that it does either.
This topic has been thoroughly discussed in the debates over the advantages of sex. Most experts agree that mixus is not important, or if it is, then the effect is small and difficult to measure.
anonymous asks,
I'd like you to address, if you would, Larry Chasin's work on the issue of Alu elements in the evolutionary generation of new conditionally expressed human exons. Specifically, I'd appreciate your perspective on:
Proc Natl Acad Sci U S A. 2006 Sep 5;103(36):13427-32.
Comparison of multiple vertebrate genomes reveals the birth and evolution of human exons.
Zhang XH, Chasin LA.
( http://dx.doi.org/10.1073/pnas.0603042103 )
Where they conclude:
This work suggests that highly repeated sequences, rather than being parasitic invaders and junk, play an important evolutionary role in the evolution of new genes. The documentation of a number of Alu exonization events led Sorek et al. to propose that exaptation of Alus may have "promoted speciation of the human lineage." Our data support this idea and extend it to additional classes of repeats and to other mammals.
I think this is pretty typical of the kinds of fuzzy adaptationist thinking that permeates the field of genomics.
Think carefully about the implications of what they're saying. Their claim is that the expansion of Alu sequences in the human lineage may not have been beneficial to any individual but it may have benefited the species in the long run.
If we take their words literally then it's consistent with the concept of junk DNA (exaptation). However, the clear implication is that the "junk" DNA confers a long-term selective advantage and therefore it has a function. It's there because it will be needed in the future.
Is that what you want to argue?
Larry, it might be helpful if you were less fuzzy yourself about what you are proposing. You seemingly dismiss transposon derived sequences as junk in one post and then later say you of course know that there are a number of regulatory sequences that located within these sequences.
Let me put it like this:
Are you proposing a thought experiment in which we subtract all the repeat derived sequences (minus the small proportion of bases that regulate gene expression), and most of the the non regulatory intergenic sequences, reducing intron and intergenic sizes by, say 90%, such that we end up with a small number of short chromosomes (say 50 Mb in size - since we know the current division processes can cope with such a size) but keeping the necessary telomeric and centromeric repeats. This would be done with all members of the human species such that all the current copy number polymorphisms and regulatory snps are maintained and all genes are allowed up to 2000bp of regulatory sequences (although we will allow a proportion of genes to have larger regulatory regions). The ultimate result is a human genome that is say, 90% smaller than the current one. You suggest that the current cellular machinery would be able to cope with this change, such that there would be no viability or evolutionary disadvantage compared to the current situation.
Is this correct?
I'm not trying to catch you out Larry, just suggesting that it would be a good idea for us to hear more specific details about what you suggest this minimal genome would look like.
LM:Their claim is that the expansion of Alu sequences in the human lineage may not have been beneficial to any individual but it may have benefited the species in the long run. [...] However, the clear implication is that the "junk" DNA confers a long-term selective advantage and therefore it has a function. It's there because it will be needed in the future. Is that what you want to argue?
I just asked for clarification of your views. Apparently, you sometimes do and sometime do not have a problem with long-term vs short-term selection. Let me remind you of your words on human ribosomal RNA genes:
LM:However, concerted evolution of these genes is essential in the long run and the mechanism of concerted evolution and gene conversion requires lots of copies. Thus, all copies are necessary for the species even though only half may be required for any one individual.
Is this an example of "fuzzy adaptationist thinking"?
--
Regarding the importance of meiotic recombination
as a driver of evolution, as you mention, this generally comes up in research on advantages of sex, for which there is no clear consensus (for those new to the topic two recent accessible reviews are given below (1)). Your view that recombination is unimportant as an evolutionary factor seems the minority opinion (but see (2) below for an interesting contrasting view). The topic is clearly of contemporary interest. Probably the best-studied species with the least amount of meiotic recombination is C. elegans (3) with one crossover only per chromosome per meiosis. Genetic disruption of crossover interference in C. elegans will enable exactly the experiment necessary to resolve the issue of evolutionary significance of meiotic recombination in this organism.
The issue of whether Homo sapiens enjoys an evolutionary advantage over yeast, Drosophila, Arabidopsis, or pufferfish is irrelevant, as you well know, since they are all equivalently evolved. The issue at hand is whether Homo sapiens as it exists now would have a difference in long-term fitness (either positive or negative) over the genomically shrunken version you propose.
As an additional point of information, human chromosome 21 at 4.7 Mb is shorter than chromosome 22 (4.95 Mb), and the term is 'mixis' (from the Greek) rather than 'mixus' and is only rarely used in any event. Those in the field generally refer to this as "meiotic recombination".
(1)
Nat Rev Genet. 2007 Jan;8(1):23-34.
An evolutionary view of human recombination.
Coop G, Przeworski M.
http://dx.doi.org/10.1038/nrg1947
Nat Rev Genet. 2007 Feb;8(2):139-49.
The evolution of sex: empirical insights into the roles of epistasis and drift.
de Visser JA, Elena SF.
http://dx.doi.org/10.1038/nrg1985
(2)
Genome Res. 2008 Jan 16 [Epub ahead of print]
No effect of recombination on the efficacy of natural selection in primates.
Bullaughey KL, Przeworski M, Coop G.
http://www.genome.org/cgi/pmidlookup?pmid=18199888
(3)
Curr Biol. 2003 Sep 16;13(18):1641-7.
Chromosome-wide control of meiotic crossing over in C. elegans.
Hillers KJ, Villeneuve AM.
http://dx.doi.org/10.1016/j.cub.2003.08.026
anonymous says,
As an additional point of information, human chromosome 21 at 4.7 Mb is shorter than chromosome 22 (4.95 Mb), ...
I picked chromosome 22 because I thought most readers would accept this as the smallest chromosome. While it's true that the published sequence of #21 is shorter than that of #22, it's also true that both chromosomes have significant gaps at both the heterochromatic regions and the rRNA gene loci.
The best estimates are that the gaps in #22 are larger than the gaps in #21 so it looks like chromsoome #21 is the smallest chromosome.
... and the term is 'mixis' (from the Greek) rather than 'mixus' and is only rarely used in any event. Those in the field generally refer to this as "meiotic recombination".
I realized just as I posted that I had made the same mistake I've made many times before. That's for pointing out that the correct term is "mixis."
I think it's a much better term than "meiotic recombination" for two reasons.
1. Recombination is not confined to meiosis. It takes place during mitosis as well.
2. The term "mixis" is not a synonym for recombination. It is used to describe one of the possible roles of recombination. The other important one is DNA repair.
What you are discussing here is the possible role of junk DNA in promoting mixis. I presume you're not suggesting that junk DNA affects whether recombination can repair DNA.
Recombination will occur now matter how big the genome is. What you want to know is whether the frequency of shuffling of alleles (mixis) is relevant.
martinc says,
Larry, it might be helpful if you were less fuzzy yourself about what you are proposing. You seemingly dismiss transposon derived sequences as junk in one post and then later say you of course know that there are a number of regulatory sequences that located within these sequences.
I'm sorry if I haven't made my position clear.
It certainly isn't for lack of trying.
I'm saying that the vast majority of transposon-derived sequences are junk. Some of the original transposon-derived sequences have been co-opted over the past 100 million years and they now carry out essential functions. Those sequences are not junk.
martinc says,
Are you proposing a thought experiment in which we subtract all the repeat derived sequences (minus the small proportion of bases that regulate gene expression), and most of the the non regulatory intergenic sequences, reducing intron and intergenic sizes by, say 90%, such that we end up with a small number of short chromosomes (say 50 Mb in size - since we know the current division processes can cope with such a size) but keeping the necessary telomeric and centromeric repeats.
Yes, that's pretty much what I'm saying although the real question is about how much junk there is in our genome. I'm a little annoyed that the nitpickers are turning this into a highly specific experiment so they can quibble about it and avoid discussing the real issue.
I guess this is pretty much what one should expect when you get a bunch of smart people speculating about a difficult subject.
This would be done with all members of the human species such that all the current copy number polymorphisms and regulatory snps are maintained ...
I see no reason to preserve known copy number variations. Most of them serve no purpose. I certainly see no reason to preserve all known single nucleotide polymorphisms in the thought experiment but if it's important to you then we'll keep them.
... and all genes are allowed up to 2000bp of regulatory sequences (although we will allow a proportion of genes to have larger regulatory regions).
Ideally, we will only keep the actual regulatory sequences. The question we need to discuss is how much of that there is likely to be. We aren't even close to knowing the exact boundaries of regulatory sequences for 90% of the genes.
In spite of that, there are people who are making wild speculations that the amount of regulatory DNA in humans is 50X greater than the actual scientific evidence suggests.
The ultimate result is a human genome that is say, 90% smaller than the current one. You suggest that the current cellular machinery would be able to cope with this change, such that there would be no viability or evolutionary disadvantage compared to the current situation.
Is this correct?
Yes, that's correct in its essence but maybe we should modify the thought experiment in order to eliminate the nitpicking.
Let's say we remove 1% at a time and allow 10,000 years for humans to adjust between each step. I'd like to suggest that after one million years we will be perfectly happy with a genome that is only 10% of our present genome.
Would that be a better thought experiment in order to avoid irrelevant distractions?
I'm not trying to catch you out Larry, just suggesting that it would be a good idea for us to hear more specific details about what you suggest this minimal genome would look like.
It would be a genome without any junk DNA. What part of that confuses you?
LM: What you are discussing here is the possible role of junk DNA in promoting mixis. I presume you're not suggesting that junk DNA affects whether recombination can repair DNA.
Correct. It's clear that in mitotic cells with a functional DNA mismatch repair system, the high-copy genomic elements have an insufficiently high degree of sequence identity for non-sister-chromatid recombination to be generally allowed (this is also true meiotically). There are sporadic reports in the literature of high-copy elements causing cancer initiating mitotic chromosomal translocations (with presumed recombinational etiology), but these are the clear minority of cases. The role of high-copy elements in mismatch repair defective cell cancer progression (eg: hereditary non-polypolis colorectal cancer) is largely unexplored, but likely not relevant to a general discussion of the role (if any) of recombination in evolution, which is to say, meiotic recombination.
Recombination will occur now matter how big the genome is. What you want to know is whether the frequency of shuffling of alleles (mixis) is relevant.
Exactly.
Separately, I'd also like to know whether you disagree with Chasin's conclusion that "de novo recruitment rather than shuffling is the major route by which exons are added to genes, and that species-specific repeats could play a significant role in recent evolution".
If you disagree, on what basis? Do you dispute their data, or just their interpretation? If their data, how do you fault their methodology? If their interpretation, what is your alternative? Or would you argue that de novo addition of exons to protein coding regions is not relevant to evolution?
anonymous asks,
Separately, I'd also like to know whether you disagree with Chasin's conclusion that "de novo recruitment rather than shuffling is the major route by which exons are added to genes, and that species-specific repeats could play a significant role in recent evolution".
If you disagree, on what basis? Do you dispute their data, or just their interpretation? If their data, how do you fault their methodology? If their interpretation, what is your alternative? Or would you argue that de novo addition of exons to protein coding regions is not relevant to evolution?
That debate belongs on another thread.
The identification of alternative splicing and unusual exons is based largely on the EST database. As far as I'm concerning that database is mostly artifact.
I've looked carefully at the so-called alternative splice variants for my favorite genes—the HSP70 gene family. None of the variants make any sense at all. The HSP0 genes are the most highly conserved genes in all of biology and the structure of the protein is known.
According to the EST data there are alternative splice variants that remove the central hydrophibic core of the most highly conserved protein. There are other variants that insert 10 or 20 amino acids into highly conserved regions in the middle of the protein.
If you look at other well-characterized proteins you see the same kind of nonsense. That's why annotators have ignored the splice data for such proteins.
If the EST data is flawed for well-characterized proteins then I assume that it's flawed for all proteins. Thus, I have no confidence in scientists who make predictions about the birth and death of new exons based on the Unigene EST data.
Let's assume, for the sake of arguments, that the EST data is flawed. What it's picking up is aberrant splicing that never leads to a functional protein. In the case of new "exons" the additions will by and large come from intron sequences that were mostly junk DNA. These will contain all sorts of repeat sequences and they will not be conserved because they are junk.
If that's what so-called "new exons" look like then I suspect we have the answer. It's an artifact.
LM: The identification of alternative splicing and unusual exons is based largely on the EST database. As far as I'm concerning that database is mostly artifact.
Well, that's certainly a bold statement. :)
If the EST data is flawed for well-characterized proteins then I assume that it's flawed for all proteins.
Even supposing that to be the case (for the sake of argument), why would there be such a large bias for spurious Alu containing cassette exons in the newest human genes? You'd have to conclude that the UniGene database is somehow strongly preferentially flawed for more evolutionarily recent genes only, and not just uniformly artifactual.
Larry, I would take issue with your assertion regarding copy number variations. Variation in CCL3L1 copy numbers, for instance, has clearly been shown to be associated with susceptibility to HIV and SLE. The whole field of CNVs and disease is very new. I cannot claim to be an authority on it but I do know that it is being taken very seriously in my field of cancer research in terms of susceptibility to particular malignancies - and I suspect the same is true of other branches of disease based studies.
I have a question regarding the timing of your 1% per 10,000 year deletion of junk DNA. Why do you need any time at all to do this deletion? Surely, if there is no advantage to the 90% of junk DNA, then we could remove it all at once without ANY harmful effects. Wasn't that your original point?
LM:Why do you say this? Why are you "fairly sure" that cells can't divide with less DNA? What did they teach you in undergraduate molecular genetics?
It's based off recollections of how homologous chromosomes line up for recombination in meiosis, which I'm still sure would be disrupted by deleting 90% of the genome.
As for Mitosis, finally found what I was looking for in Molecular Biology of the Gene. I'm no longer concerned so much with the effects of the deletion on mitosis, but only if origins of replication sites are maintained in the chromosome structure and recombinational repair is still possible...
As for my education, my earlier comment is was based off logical extensions of the know functions of chromosomal organisation. Mildly rusted though from having insomnia all last year and I should have searched for the right information...
Centromeres and telomeres aren't junk so they won't be deleted. Would it help if I changed the question to allow for construction of new chromosomes with the remaining 10% DNA? How about if we make only three medium sized chromosomes? Would that ease your mind?
If beyond the realms of theory, and mitosis/meiosis are fully capable, and there's no issues with gene regulation as anonymous and sanders have brought up, then by all means I agree, but still with reservations.
Whatever about recombination sites, facilition of novel gene duplications, and long range genetic regulation, I do have one issue that I'm in disagreement about regarding deleting such large amounts of DNA. What Larry seems to be suggesting is that in the hundreds of millions of years that vertebrate genomes have existed with huge amounts of junk DNA there have been no major changes in the basic cellular machinery that existed in the small genome era that would now cause the cell difficulty if all the junk sequence was suddenly removed. We could always modify those pathways, such that they could again handle a small genome, but, of course, that is a separate issue and just underscores the point that deleting so much bulk DNA would cause problems to the cells we have today.
I guess I'm confused by MartinC's last comment. What specific mechanisms to handle large amounts of junk are you talking about? And if the junk is just noise that the cell needs special tools to prune, wouldn't it be fair to assume that losing the noise would just make those tools superfluous?
The few possibilities I'd consider probable are:
1. Silencing. Would we see something like position effects writ large, with excessive methylation in a genome stripped of anything to keep those enzymes occupied?
2. Dosage. There may be gene products that are upregulated to compensate for some competitive junk transcripts. Removing some junk might lead to relative changes in effective dose.
The thing is, though, that these aren't functions for junk DNA, these are side effects and possible contingent compensatory mechanisms that have evolved with them. It's like suggesting that if I have a massive growth on the left side of my body, it's removal would hint at a functional role for that growth because I might have some trouble keeping my balance afterwards.
Fine. Life probably has evolved details to cope with large amounts of junk DNA. It seems backwards to me to argue that this implies it has a function...which function is to give the compensatory mechanisms something to do.
I don't buy the idea that the junk is a necessary bulking agent, either. Sure, DNA/cytoplasm ratios are a regulator of mitosis, so removing junk might mean cell volumes would be smaller. Would this be significant for multicellular animals? I don't know. I suspect it wouldn't be a deleterious effect (necessarily), but it might cause some changes in morphology. I don't think the expectation is that removing all the junk would have no effect on phenotype, but that it would cause subtle effects, nothing like, say, stripping out those boring histone genes.
One could also consider the human Y chromosome:
If ever there was a chromosome that has the genetic potential and biochemical capacity to reduce its bulk length, this is the one. However, it still weighs in at a relatively beefy 58 Mb. Why? More to the point, since the Y chromosome is transmitted in an essentially asexual manner, why doesn't the Y length show greater person-to-person variability? Individual Y chromosome lengths should have drifted all over the place, but it's clear from cytogenetics that they have not. On the face of it, the Y chromosome is an argument that human chromosomes much smaller than 50 Mb are not long-term tolerated in the population.
Can the genomicists in the audience provide species-specific information on Y chromosome sizes both in absolute terms, and as a percentage of the smallest autosomal chromosome length within a given species? (Clearly species without a Y chromosome need not apply.)
PZ, I didn't say this bulk DNA was doing anything functional, rather I was directly addressing the original question posed by Larry;
"How much of our genome could be deleted without having any significant effect on our species?"
I personally don't buy the idea that most of the genome (or even more than a small percentage of it) is functional - in the way we might say a coding sequence, promoter sequence or structural, tRNA or microRNA gene is functional. What I do suggest is that we could not delete 90% of the DNA and not see a "significant effect on our species".
OK, Craig Venter will probably get around to producing a minimal genome mouse at some stage (named Craig, no doubt) and we will get an idea of the mammalian cell can cope with the size reduction but the types problems I can forsee at the moment are things like the regulation of particular long genes - what will happen to a gene that previously took an hour to transcribe, due to very long introns lengths, and now takes two minutes? Might that not have a significant effect on the organism?
PZ: I'm with you 100%, and I think a lot (perhaps even most) of the people who have commented would have no problem with what you say. But the example of the removal of the growth gets right to the centre of the issue: the question was not whether the junk is an adaptation or doing something useful. It's about straight deletion of the lot.
Larry asks,
Are you trying to make a case for function or purpose of junk DNA...?
Yes Larry, I was indeed suggesting that you consider the possibility that much of our "junk DNA" may have the function of conferring "evolvability" or plasticity upon our genomes. This indeed could be a "real function that has been selected".
Why do I propose such a preposterous theory?
It is a hypothesis that answers the following question: How did the merciless process of natural selection produce human cells that devote 99% of the energy invested in DNA replication to the propagation of genomic parasites?
Maybe the parasites are doing something for us. Maybe that something is helping our genomes evolve.
I can't think of a better answer to this question. Clearly you can't either or you wouldn't have given up already and settled for relegating the DNA in question to the conceptual trash can.
Larry asks,
You need to be very careful with this argument. Are you suggesting that species with large genomes are better at evolving than those with small genomes?
Not really. First of all comparing any random species with a small genome to one with a larger genome is not relevant. It's a question of whether primates with more of certain "junk DNA" elements (ie Alu or transposons), for example, are more likely to achieve beneficial mutation than those with less of these elements. So maybe one prediction would be that primates with more Alu elements produce novel exons faster. Or that primates with more transposons produce gene duplications faster. As a result of this increased rate of copying and shuffling, bona fide protein-encoding genes with novel, fitness enhancing functions are more likely to emerge in individuals who carry more of this "junk" in their DNA.
In addition to carrying a lot of mobile elements around, is the human genome not also one of the most duplicitous (in fact, multiplicitous) with respect to the prevalence of gene paralogues? Certainly the preponderance of mobile elements in the human genome compared to say, yeast, might explain why the human genome exhibits vastly more para-ology than such organisms? And if there has been evolutionary selection for paralogues, does this not mean that the process that generated them was also selected for?
What I do suggest is that we could not delete 90% of the DNA and not see a "significant effect on our species".
Yes, and this problem with its formulation was hashed out at the time the question was first (ill)posed.
I am not willing to make an estimate simply because I don't know enough to do so and -- given that neither you nor I are making the huge series of deletion mutations in mice to find out -- I'm not going to in any foreseeable future. Another way of putting this: I (in agreement with others here) think the question is ill-posed, and as such no interesting answer will emerge. Thus, trying to come up with an answer is no more than mental self-gratification. It is not science.
Note that the question IS being answered for at least several bacteria and for Saccharomyces. THAT is science.
George Smiley says,
I am not willing to make an estimate simply because I don't know enough to do so and -- given that neither you nor I are making the huge series of deletion mutations in mice to find out -- I'm not going to in any foreseeable future. Another way of putting this: I (in agreement with others here) think the question is ill-posed, and as such no interesting answer will emerge. Thus, trying to come up with an answer is no more than mental self-gratification. It is not science.
I completely disagree. Trying to understand nature is science no matter how you do it. You don't sit around waiting for definitive experiments (that are never going to happen) before speculating about what it all means.
What you are doing is dismissing all of theoretical biology as non-science.
It is very important to try and understand what all that DNA in our genome is doing. Yes, experiments are important, but so is applying our knowledge of the big picture of biology to the particular problem of genome organization.
People are doing it all the time. It's real science when the ENCODE project announces that 70% of the genome is transcribed and it's also real science to debate what that means. The authors of the paper didn't hesitate to put their particular spin on the result. Are they right? Are they being scientists in the Methods and Results section of the paper but not in the Discussion?
It's real science to discover that almost half of our genome is transposable elements and it's real science to ask whether those transposable elements are harmless parasites or whether they perform an essential function in our species.
Note that the question IS being answered for at least several bacteria and for Saccharomyces. THAT is science.
No, that's your kind of science. There are others who think that science includes theory, the formation of hypotheses, thought experiments, and critical evaluation of the scientific literature.
There's a heavy emphasis these days on "doing" rather than "thinking" in science. While nobody will discount the importance of experiments and observations they still have to be put into the proper context. I think we're in serious trouble. Too many scientists have become adept at doing complicated experiments (e.g., systems biology) without understanding what their results mean.
As you know, there are huge numbers of very competent technicians out there who don't understand basic concepts like evolution. That's partly because of the attitude you express here where you dismiss thinking as not being part of science.
I do not agree that I am dismissing all of theoretical biology. There's a razor's edge to walk, though. There is an enormous amount of really interesting theoritical biology going on in (for example) protein folding and cell motility. The salient feature of the most important of these theories is that they are being vigorously tested with ongoing experiments. The rapid cycling of experiment and theory drive progress. Crick was, of course, the second greatest exemplar of this approach in the history of biology.
Major examples of falling on the wrong side of the razor's edge can be found in embryology, where a rich -- and largely incorrect -- theoretical biology literature grew from the 1950s to the mid-1980s, when it became evident that developmental genetics of Lewis, Nusslein-Vollhard, and Weischaus had blown the lid off the problem. Similar difficulties are still prevalent in neuroscience.
So there are two difficulties here. One is in formulating a problem with sufficient precision, something that nether I nor several others here think you've done. The second is getting too far ahead of the data. Predictions that can't easily be tested, and that don't explain a large body of existing data, are not so interesting. Your project as so far described has, in my opinion, both problems.
George Smiley says,
So there are two difficulties here. One is in formulating a problem with sufficient precision, something that nether I nor several others here think you've done.
The question I'm asking is whether a large amount of our genome is junk. It's a question that's been discussed for 35 years in the scientific literature. If you choose to ignore the question because it's not precisely enough formulated then that's perfectly fine with me.
What I won't accept is your opinion that the discussion is not science.
The second is getting too far ahead of the data. Predictions that can't easily be tested, and that don't explain a large body of existing data, are not so interesting.
I don't agree that the question of junk DNA is too far ahead of the data. We have lots of data starting with the renaturation studies of the early 1970's. The real problem is that many people don't know about the data. They assume we are completely ignorant about our genome.
I've already explained that one third of our genome is composed of degenerate transposable elements. How is that being too far ahead of the data?
We also know that most of the DNA in our genome is accumulating mutations at a rate that indicates the sequence is unimportant. How is that being too far ahead of the data?
The genetic load argument has been around since the late 1960's. It makes the point that most of our genome cannot be essential or else we would be dead. Don't you think that's useful information? Why is it "too far ahead of the data" to recognize that point?
The idea that evolution can result in large amounts of useless DNA in our genome is part of a much larger picture of biology and evolution. Once we recognize that lots of things aren't "designed" we will be in a better position to explain a large body of existing data.
The stakes are serious. The concept that everything has a function is under attack. I can understand why defenders of that point of view are looking for ways to make the entire question illegitimate and out-of-bounds.
I'm not sure where you stand on this issue? Do you believe that most of the DNA in our genome should have a function?
Your project as so far described has, in my opinion, both problems.
No problem. :-)
You don't have to participate in the discussion if you don't want to.
BTW, I expect you to be equally critical of people who make claims that most of our DNA has a function. Many recent scientific papers have made those claims. Have you criticized those papers for not being scientific? Can you point me to some examples of your criticism?
"BTW, I expect you to be equally critical of people who make claims that most of our DNA has a function."
I am.
That is a position with even less experimental support than yours. As I wrote above, the question for humans is interesting but untestable. The question for mice is interesting but will not be tested for many years. The question for model organisms with smaller genomes and more facile genetic manipulation is being aggressively pursued and we have or will have answers on a reasonable time scale (a decade or two for metazoans and plants).
As for your project not being science, my language was too strong. But I think that your approach is closer to polemic than to theory. This may be understandable given the amount of, um, junk that has been generated by various adaptationist camps in the genomics community. It has to be frustrating to you. But I think that your claims are couched in language that is overly broad -- broader than the claims that the existing evidence can support. More tightly defined claims using more precise language would really help. ("Junk" is a classic example of an emotionally charged term with a commonly understood meaning in regular speech that is not the same as its technical meaning. Such situations at best lead to confusion and at worst become playgrounds for mischievous misuse of the terrminology.)
Perhaps semantics, but all that junk DNA, the non- used parts, are most likely endogenous retroviruses which have been in the genome so many millions of years that they are obliterated by mutations. See recent reincarnation of fossil virus.
Delighted to find this blog.
Peggy Wolfson
Perhaps semantics, but all that junk DNA, the non- used parts, are most likely endogenous retroviruses which have been in the genome so many millions of years that they are obliterated by mutations. See recent reincarnation of fossil virus.
Delighted to find this blog.
Peggy Wolfson
Hi Peggy, welcome to Sandwalk.
You might enjoy reading some of my earlier postings on Genomes and Junk DNA.
Post a Comment