More Recent Comments

Tuesday, December 31, 2019

Are introns mostly junk?

There are many reasons for thinking that introns are mostly junk DNA.
  1. The size and sequence of introns in related species are not conserved and almost all of the sequences are evolving at the rate expected for neutral substitutions and fixation by drift.
  2. Many species have lost introns or reduced their lengths drastically suggesting that the presence of large introns can be detrimental in some cases (probably large populations).
  3. After decades of searching, there are very few cases where introns and/or parts of introns have been shown to be essential.
  4. Researchers routinely construct intronless versions of eukaryotic genes and they function normally when re-inserted into the genome.
  5. Intron sequences are often littered with transposon and viral sequences that have inserted into the intron and this is not consistent with the idea that intron sequences are important.
  6. About 98% of the introns in modern yeast (Saccharomyces cerevisiae) have been eliminated during evolution form a common ancestor that probably had about 18,000 introns [Yeast loses its introns]. This suggests that there was no selective pressure to retain those introns over the past 100 million years.
  7. About 245/295 of the remaining introns in yeast have been artificially removed by researchers who are constructing an artificial yeast genome suggesting that over 80% of the introns that survived evolutionary loss are also junk [Yeast loses its introns].
Defenders of intron functionality will find ways of quibbling with all of these statements but if you step back and look at the big picture, their arguments look a lot like ad hoc rescue.1

Lately there's a new argument that seems to be taken seriously. Several labs have looked closely at the small number of essential introns in yeast and determined that some of them are essential because they contain other genes, such as snoRNA genes. There's another subset that contain some sort of regulatory sequences that help regulate the growth of yeast cells under starvation conditions (Morgan et al., 2019; Parenteau et al., 2019) [Yeast loses its introns].

The bottom line, as far as I'm concerned, is that something like 99.7% of all the introns in the ancestor of modern Baker's yeast were junk but there's a small number that have secondarily acquired a function.

Two of the authors of one of the papers cited above have looked at the same data and reached a very different conclusion.
Parenteau, J., and Elela, S. A. (2019) Introns: Good Day Junk Is Bad Day Treasure. TRENDS in Genetics. 35:923-934 [doi: 10.1016/j.tig.2019.09.010]

Abstract: Introns are ubiquitous in eukaryotic transcripts. They are often viewed as junk RNA but the huge energetic burden of transcribing, removing, and degrading them suggests a significant evolutionary advantage. Ostensibly, an intron functions within the host pre-mRNA to regulate its splicing, transport, and degradation. However, recent studies have revealed an entirely new class of trans-acting functions where the presence of intronic RNA in the cell impacts the expression of other genes in trans. Here, we review possible new mechanisms of intron functions, with a focus on the role of yeast introns in regulating the cell growth response to starvation.
Here's how they explain their stance in the paper's introduction.
Clearly, removing introns correctly from pre-mRNAs is important but cannot explain their ubiquitous preservation in genomic DNA across the course of evolution. Certainly, they are not simply disposable junk. In organisms with large genomes, introns are indispensable for the process of alternative splicing, which is essential for regulating gene expression and function. However, introns are also preserved in organisms where alternative splicing and splicing-dependent regulation of gene expression are rare. Why would single cells with compact genomes tolerate the energetic cost of transcribing, splicing, and degrading introns if their only function is associated with the act of their removal? Could introns serve functions other than regulating the expression of host genes? In the past few years, sequencing advances and our ability to delete introns from eukaryotic genomes have started to provide answers to some of these questions. Introns are now known to provide a reservoir of small noncoding (nc)RNAs that act in trans on other genes. Recent studies have shown that intronic RNAs also function as direct regulators of multiple cell-response genes during nutrient depletion.
This isn't a case of viewing a glass as either half full or half empty. It's a case of viewing a glass as either >99% empty (i.e. mostly junk) or ~1% full (i.e. mostly functional). The authors are ignoring all the evidence suggesting that introns are mostly junk and concentrating on the relatively few examples where they have acquired a function. They then extrapolate from these few examples to declare that introns in general are not junk.

Here's how they put it in their conclusion.
Modern introns are not junk DNA, but rather they are noncoding sequences that have gained functions through evolution to merit their maintenance. It is possible that a population bottleneck, or inability to eliminate introns, provided the initial reason for their preservation and allowed introns to develop these functions. However, it is becoming increasingly clear that these introns have rapidly gained inherent functions beyond regulating their host gene that further supported their maintenance through evolution.
It seems obvious that none of the reviewers of this paper had a problem with such a conclusion. Am I the only one who sees a problem?

1. Alternative splicing is a common reason for thinking that introns are essential but we now know that alternative splicing isn't common so it can't account for the vast majority of introns.

Morgan, J.T., Fink, G.R., and Bartel, D.P. (2019) Excised linear introns regulate growth in yeast. Nature [doi: 10.1038/s41586-018-0828-1]

Parenteau, J., Maignon, L., Berthoumieux, M., Catala, M., Gagnon, V., and Elela, S.A. (2019) Introns are mediators of cell response to starvation. Nature. [doi: 10.1038/s41586-018-0859-1]


Rosie Redfield said...

It's like arguing with creationists. They already 'know' the answer, and no amount of evidence will change their minds.

Mikkel Rumraket Rasmussen said...

I have to wonder if these people have ever really done anything to familiarize themselves with the evidence for junk-DNA. They're also exhibiting rather poor reasoning skills. Their whole case seems to be based on a hasty generalization and yet they show zero awareness of that.
It should be simple enough to see if it is pointed out to them.

Joe Felsenstein said...

I agree with Larry's general conclusion here. However it is worth noting that when an intron is removed in the lab and the resulting cell seems unaffected, that is weak evidence. It does not rule out modest deleterious effects, ones which would be sufficient to keep the intron in the genome. Lack of conservation in evolution is stronger evidence. Basically, nature can find the time and the funding to do a much bigger experiment than we can.

Larry Moran said...

I understand your point but I disagree that removal of an intron is "weak" evidence that it's junk. One can always argue that viability in the lab may not be sufficient to detect small beneficial effects but that's still special pleading. It's the kind of argument you make if you don't want to accept the strong evidence for junk.

But the yeast experiments go way beyond the removal of just one intron. Several hundred introns were removed without noticeable effects. That's a little more difficult to defend, no?

Joe Felsenstein said...

My statement was about "an intron", so one of them. You are right that seeing no effect of "several hundred" yeast introns is not-so-weak evidence. If we figure that we could see an effect on the selection coefficient of, say 0.001, and if Michael Lynch's "drift barrier" were, say s = 0.0000001, then you might need to have 10,000 introns (0.001/0.0000001) removed to be likely to see the effect.

Unknown said...

I wonder, how many of the papers' author were also part of the ENCODE project?

S Johnson said...

The principle that natural selection ensures all features are adaptive is being applied to the genome, as the gene (no matter how defined) is the unit of selection. (Not the unit of heredity.) Therefore, the positive functions "explain" how natural selection has ensured the survival of so many introns. Nonsurvival of course reflects the power of negative selection. Drift of course is acknowledged somewhere else but isn't considered as a hypothesis, not even as explaining why not all features are adaptive. It's an old story, isn't it? This sort of thing is why lay people tend to think the Modern Synthesis needs some work. This site keeps telling us the Modern Synthesis actually incorporates random drift etc. but these kind of articles contradict this claim, so far as I can see.

João said...

Oh, it's really sad to see people not looking at the big picture, cherry picking here and there. Boring.

Since we are talking about junk DNA/RNA... Larry, are you still writing your book?

João said...

So you are saying that one possibility to be serious considered about the results of the "removal" studies Larry addressed is that no effect was observed because none could even be detected?

João said...

*seriously considered

Typo. Sorry.

daedalus2u said...

What if a large amount of 'junk' in a genome is not a 'bug', but a 'feature'?

A 'feature' so that when typical errors of DNA duplication happen randomly, they are less likely to occur at an essential site of an essential gene?

A 'feature' that would allow DNA duplication errors to inactivate homing endonuclease genes while not inactivating essential genes?

Isn't that what the spliceosome is for?

Joe Felsenstein said...

Does your argument assume that the total number of errors in DNA duplication is constant, and when you add 5% more DNA you don't get 5% more errors of replication?

Larry Moran said...

Book is still a work in progress but I've found a publisher.

Larry Moran said...

Let's assume that DNA replication makes one mistake for every 1000 bp. Let's assume that the genome contains 1000 bp of DNA so that there's one mistake every time the genome is replicated.

Now let's add 1000 bp of junk DNA. Now the genome size is 2000 bp and DNA replication introduces two mutations every time the genome is replicated. How does that help?

Joe Felsenstein said...

... and with that proportionality, which makes biological sense, the chance of a mutation landing in an essential site is the same after the extra DNA is inserted as it was before.

Joe Felsenstein said...

Good to hear. Maybe we could hold a contest here for a title of more reasonable length.

João said...

This is amazing!!! I'm so happy you found a publisher! I really want to read your book.

Larry Moran said...

@Joe: You want something that's shorter than "What's in Your Genome"? What do you suggest?

Joe Felsenstein said...

The title you gave earlier was longer than that, more than twice as long. This one is much better.

João said...

I think "90% of your DNA is junk" is the subtitle.

Joe Felsenstein said...

I'm relieved to hear that.

Unknown said...

Thank you professor Moran for this excellent blog. I am not sure that “alternative splicing isn't common”. I do not have the numbers, but after looking at hundreds/thousands of gene structures in NCBI's Sequence Viewer my feeling is that more than half of human genes have mRNA variants, and many of these (maybe half of them) are generated by alternative splicing. Still, introns look unnecessarily large and probably much shorter versions of them would still be functional for alternative splicing. What I tell my students is that most parts in introns can be considered junk DNA.

Larry Moran said...

Here's a figure showing the number of RNA variants produced from the triose phosphate isomerase (TPI) gene according to the ECgene database. There are 33 of them.

Debating alternative splicing (part II)

Almost all of those variants have been rejected by genome annotators. The current entry for the TPI gene has only two RNA variants.

Are multiple transcription start sites functional or mistakes?

There is no evidence that the predicted proteins from those variants are functional in any way - in fact, they have not been detected in any cell type. Here's a link to other databases showing additional RNA variants from the TPI gene and a discussion about why the two variants that have survived annotation are probably not examples of alternative splicing.

Splice variants of the human triose phosphate isomerase gene: is alternative splicing real?

There is abundant circumstantial evidence that the few remaining variants in the gene databases are NOT examples of biologically relevant alternative splicing. There are only a small number of human genes that have been proven to produce different functional products by alternative splicing.

The problem here is that there are still multiple variants associated with most human genes but in the vast majority of cases there is no evidence whatsoever that the non-standard variants are biologically relevant. The mere existence of an RNA variant - even one that has survived the initial culling by annotators - does not prove that it is an example of alternative splicing. You need evidence to establish that claim.