Feb. 9, 2016: I have edited the paragraph beginning with "Exciting..." to remove details of mutation rates because my initial posting was probably wrong about coding vs. non-coding mutation rates. To fix that requires much more nuance than is relevant for the point I'm making in that paragraph, not to mention much more nuance than I'm capable of grasping immediately! Cheers and thanks to Daniel and Ken in comments below and to everyone who chimed in on Twitter.
And it's with that appreciation for constant, unpredictable, but tolerated mutation—of evolution's momentum, of a lineage's perpetual change, selection or not—on top of a general understanding of population genetics that just makes adaptation seem astounding. It makes it difficult to believe that adaptation is as common as the myriad adaptive hypotheses for myriad traits suggest.
That's because this new raw material for adaptation, this perpetual mutation, really is only a tiny fragment of everything that can be passed on. But, what's more, each of those itty bitty changes could be stopped in its tracks before going anywhere.
The good, the bad, and the neutral, they all need luck to pass them onto the next generation. That's right. Even the good mutations have it rough. Even the winners can be losers! Here are the ways a mutation can live or die in you or me:
This view of mutation fits into that slow and stately process that Darwin described, despite his imagination chugging away before he had much understanding of genetics.
Of course, bottlenecks or being part small populations would certainly help our rogue underdogs proliferate, and swiftlier so, in future generations.
***
I always account for virally-induced mutation when I imagine the evolution of our genome. That's because I'll never forget this quote. Who could?“Our genome is littered with the rotting carcasses of these little viruses that have made their home in our genome for millions of years.” - David Haussler in 2008Or this...
"Retroviruses are the only group of viruses known to have left a fossil record, in the form of endogenous proviruses, and approximately 8% of the human genome is made up of these elements." (source and see this)Exciting virus discoveries aside, we're constantly mutating with each new addition to the human lineage. Thanks to whole genome sequencing, the rate of new mutation between human parent and offspring is becoming better known than ever before. We each have new single nucleotide mutations in the stretches of our DNA that are known to be functional (very little of the entire genome) and that are not (the majority of the genome). These are variants not present in our parents’ codes (for example, we might have a ‘T’ where there is a ‘A’ in our mother’s code). And there are also deletions and duplications of strings of letters in the code, sometimes very long ones. Estimates vary on parent-offspring mutation rate and that's because there are different sorts of mutations and individuals vary, even as they age, as to how many mutations they pass along, for example. Still, without any hard numbers (which I've left out purposefully to avoid the mutation rate debate), knowing that there is constant mutation is helpful for imagining how evolution works. And it also helps us understand how mutations even in coding regions aren't necessarily good nor bad. Most mutations in our genome are just riding along in our mutation-tolerant codes—where they will begin and where they will go no one knows!
And it's with that appreciation for constant, unpredictable, but tolerated mutation—of evolution's momentum, of a lineage's perpetual change, selection or not—on top of a general understanding of population genetics that just makes adaptation seem astounding. It makes it difficult to believe that adaptation is as common as the myriad adaptive hypotheses for myriad traits suggest.
That's because this new raw material for adaptation, this perpetual mutation, really is only a tiny fragment of everything that can be passed on. But, what's more, each of those itty bitty changes could be stopped in its tracks before going anywhere.
The good, the bad, and the neutral, they all need luck to pass them onto the next generation. That's right. Even the good mutations have it rough. Even the winners can be losers! Here are the ways a mutation can live or die in you or me:
 |
| The Brief or Wondrous Life of Mutations, Wow. |
This view of mutation fits into that slow and stately process that Darwin described, despite his imagination chugging away before he had much understanding of genetics.
Of course, bottlenecks or being part small populations would certainly help our rogue underdogs proliferate, and swiftlier so, in future generations.
Still, trying to imagine how any of my mutations, including any that might be adaptive, could become fixed in a population is enough to make me throw Origin of Species across the room.
By "adaptive," I'm talking about "better" or "advantageous" traits and their inherited basis ... that ever-popular take on the classic Darwinian idea of natural selection and competition.
For many with a view of mutation like I spelled out above, it's much easier to conceptualize adaptation as the result of negative selection, stabilizing selection, and tolerant or weak selection than it is to accept stories of full-blown positive selection, which is what "Darwinian" usually describes (whether or not that was Darwin's intention). One little error in one dude's DNA plus deep time goes all the way to fixed in the entire species because those who were lucky enough to inherit the error passed it on more frequently, because they had that error, than anyone passed on the old version of that code? I guess what I'm saying is, it's not entirely satisfying.
But what if a mutation could be less pitiful, less lonely, less vulnerable to immediate extinction? Instead, what if a mutation could arise in many people simultaneously? What if a mutation didn't have to start out as 1/10,000? What if it began as 1,000/10,000?
That would certainly up its chances of increasing in frequency over time, and quickly, relative to the rogue underdog way that I hashed out in the figure above. And that means that if there was a mutation that did increase survival and reproduction relative to the status quo, it would have a better chance to actually take over as an adaptation. This would be aided, especially, if there was non-random mating, like assortative mating, creating a population rife with this beneficial mutation in the geologic blink of an eye.
But how could such a widespread mutation arise? This sounds so heartless to put it like this, but thanks to the Zika virus, it seems to me that viruses could do the trick.
 |
| Electron micrograph of Zika virus. (wikipedia) |
*albeit a horrifying one, and unlikely to get passed on because of its effects, so it's not adaptive whatsoever.
If viral mutations get into our gametes or into the stem cells of our developing embryos, then we've got germ-line mutation and we could have the same germ-line mutation in the many many genomes of those infected with the virus. As long as we survive the virus, and we reproduce, then we'll have these mutant babies who don't just have their own unique mutations, but they also have these new but shared mutations and the shared new phenotypes associated with them, simultaneously.
Why not? Well, not if there are no viruses that ever work like this.
We need some examples. The mammalian placenta, and its subsequent diversity, is said to have begun virally, but I can't find any writing that assumes anything other than a little snowflake mutation-that-could.
Anything else? Any traits that "make us human"? Any traits that are pegged as convergences but could be due to the mutual hosting of the same virus exacting the same kind of mutation with the same phenotypic result in separate lineages?
I've always had a soft spot for underdogs. And I've always given the one-off mutation concept the benefit of the doubt because I know that my imagination struggles to appreciate deep time. What choice do you have when you think evolutionarily? However, just the possibility that viruses can mutate us at this larger scale, even though I know of no examples, is already bringing me a little bit of hope and peace, and also some much needed patience for adaptationism.
***
Update: I just saw this published today, asking whether microcephaly and other virus-induced birth defects are congenital. Answer = no one knows yet: http://www.nytimes.com/2016/02/09/science/zika-virus-microcephaly-birth-defects-rubella-cytomegalovirus.html?partner=IFTTT&_r=1
It's very possible that when people write about viruses causing mutations in our past, they actually mean in this way I've imagined in the post But I can't find any explicit discussion of this either in the primary lit or the dissemination of it. That could be my limitation. It's possible that all I need is the right textbook!
ReplyDeleteThe debate about the importance of horizontal transfer (as opposed to classical parent-offspring vertical transfer) of genes is an old one and involves the classic instances, that seem rather clear, of mitochondria, chloroplasts, and plasmids. At least as important are transposable elements, that are all over the genome, and at least many seem to have been originally from incorporated viruses. There are instances of such sequences that have been shown to have some function, such as use as regulatory elements.
ReplyDeleteHowever, there are problems with making this an important general contributor to functionally adaptive evolution. Primarily, I think, the incorporation of a virus into the genome is roughly random, like other mutations, relative to any potential function. Secondly, to be expressed, the viral DNA must be near appropriately active regulatory elements (or must be located so as to serve as a regulatory element to some useful nearby gene. Thirdly, it has to have some useful effect. This might be to activate a regular vertically originated gene, in a useful context.
The idea of thousands of incorporations as a means of making adaptive change more rapid etc. seems forced, to me. The reason is that the incorporation sites of viruses would largely be different in each infected person in a population, so that in essence the event would serve as a kind of random mutation that would only have adaptive effects in a given person, not a population--any value would accumulate over generations in the usual evolutionary (selection plus drift) way, I think.
This doesn't mean there can't have been instances where your idea occurred, of course. Incorporation sites aren't 100% random and there could be some preferred integration sites that could in principle do what you suggest, but even then it would affect one gene, say, rather than the complex of genes usually required for adaptive evolution. So it would be a contributor, but how much would it change the adaptive pace? Maybe sometimes in a detectable way, in small demes, etc.
There is the other problem of the prevalence of viral epidemic infections in small ancestral hunter-gatherer (or earlier primate) demes. I don't know enough about viral and population ecology to know how prevalent such things might have been.
So, overall, while this could be a factor certainly, I would need to see more to be convinced that it was a major one.
Finally, there is a very energetic debate about how important horizontal transfer has been in the evolution of the major branches of life. How branch-like vs interwoven is the tree of life? I don't know enough to have an opinion, but feelings run high among those who do!
Very interesting hypothesis, Holly. But, I think there are many possible ways that a virus might affect a developing fetus that don't require it to insert itself into gametes -- and remember that for your evolutionary scenario to work, the viral sequence has to be heritable, which means it has to insert itself into the DNA of the sperm or egg cell, not just any old cell in the already developing fetus.
ReplyDeleteIn the case of Zika, for example, it seems perfectly credible to imagine that if the virus crosses the placenta, it might manage to insinuate itself into precursor cells and interrupt cell division somehow, or interrupt gene regulation and thus cell division, and thus the growth of developing brain/skull cells somehow (as far as I know, it's still not clear whether the brain drives the growth of the skull, or vice versa). Or, as the NYT piece you link to suggests, perhaps even if it doesn't cross the placenta there's an immune response that somehow disrupts cell growth.
In any case, it seems to me that plausible scenarios for how fetal growth might be disrupted require nothing more than normal viral messing with cell function -- disrupting gene regulation and cell growth. Given that, as Ken says, insertion of viruses into DNA seems to be essentially random, it seems unlikely, to me, to be a clear, predictable explanation for rapid adaptive evolution.
thanks Ken and Anne. All I'm suggesting is a way for a mutation to start out with a better chance at going to fixation than the present conception offers. It seems plausible to me that a virus could leave the same genetic signature behind in its hosts. How do we know that viruses insert randomly within each host and not systematically?
ReplyDeleteLast sentence is clearer if written: how do we know that a given virus inserts randomly within each host and not systematically?
DeleteThere must be at least something known about viral incorporation sites. Of course, that will likely depend on the virus, and the polymorphisms among host individuals. There are some repeatable rearrangements and probably some preferred incorporation sits. I think that incorporation involves some target DNA sequence elements, and so they are likely to be many in mammalian genomes and in the same location in members of the same species.
ReplyDeleteYour idea of group evolution, if I can call it that, is a clever and interesting one, and worth asking how plausible it is as more than an occasional or incidental contributor. Since most complex adaptations seem to involve many different genes, over long periods so that the various combinations can be screened for fitness, I would expect the contribution to be minor for complex adaptations, but perhaps more important for highly specific, such as single-gene, ones. Certainly the effect of infection on specific traits in a pathogenic sense is very well-known! But the usual argument is that they disrupt something that is highly organized and that is easier than to build something new. But why not the occasional beneficial one? And, in particular, as you note, we have to keep in mind the slowness of real adaptive evolution and no need to suggest that a viral infection suddenly causes some new complex trait to arise. Patient thinking about evolution as a patient process is difficult but important. That is a point that I think you made clearly.
Anyway, it's something very much worth thinking about! Probably a lot of implicit or perhaps even some explicit relevant data exist in the general biology/evolution/genetics literature, if one searches for it.
Thanks Ken. The glory of a blog is that you can think out loud before doing much digging. I did dig a bit though and am finding plenty of studies of viruses evolving, but none that answer/address this particular question about systematic effects in hosts' genomes. I'm probably just not digging properly yet.
DeleteSearch for viral incorporation sites or mechanisms, for viruses that work by incorporating in the genome. I don't remember if RNA or DNA viruses would be more likely. Look for the identity of transposable elements and how they work. Many or most repeat elements in genomes are of these types. They have, as I recall, some stereotypical incorporation sites. Since most people will have similar sites in similar locations there must be some preferential incorporation sites. Something must be known about this---but not by me....
ReplyDeleteThere's a lot known about retroviral insertion from years of gene therapy research, as the idea is to introduce corrected genes into the cell via retroviruses and people used to worry that if insertion was totally random, it might induce cancer. Now it seems it's nonrandom to the extent of retroviruses selectively inserting near transcription sites, but there are many many such sites, so not all copies choose the same site, just the same flavor of site.
ReplyDeleteAnd, for example, here's a 2011 paper reporting random viral insertion into the shrimp genome. Here's one reporting random insertion of the Hepatitis B virus into tumor and non-tumor cells of patients with liver cancer.
But I don't think this kills your idea! I'm sure there are papers showing nonrandom insertion, as well. I assume it's like everything else in evolution -- if something can happen, it will!
A huge amount is known about certain insertion patterns, such as Alu inserts and others. Much of the literature is about endogenous proliferation after some initial insertion event. But there may be evidence of independent exogenous insertion. It may be that insertion is largely random but if in an active transcription region in an appropriate tissue, the viral signal is produced. In such instances, infection may lead to disease in multiple exposed individuals, but for individually specific insertion reasons. There just has to be a literature on this.....
ReplyDeletePlus, of course, mothers would have to survive the infection, and whatever trait the newly inserted virus was responsible for would have to be non-lethal.
ReplyDeleteMaybe you've thought up a new mass extinction explanation!
Here's an even more extreme thought - due to HGT a positively selected gene could appear in many individuals not just within the same population but across species, even clades, resulting in (especially if the adaptive gene effect is large) concerted evolutionary change across swaths of the biota! But not likely, to say the least, at least not in eukaryotes.
ReplyDeleteLong aside: I understand the dismay over Just-So Stories: the assumption that every trait is an adaptation produced by natural selection, assigned specificity of function that is unwarranted, and provided an imaginative scenario of how it evolved. I call the alternative Just-Cause Stories: it's assumed that the trait cannot be an adaptation, and provided an imaginative scenario of how it emerged (developmental constraint, heterochronic linkage, phylogenetic inertia, phenotypic neutralism etc.). The problem is not just with adaptation and selection, but the assuming and the storytelling, adaptive/selective or not they're still Just Stories. Well, those things aren't really problems if they're tools in an rigorous investigation comparing models, but they are problems if they're asserted as reliable knowledge without much evidence.
How can positive selection work to produce adaptations? These come to mind (I should provide citations for each one but...):
- Functional equivalence: whether because the genetic variation is silent or the different proteins that are produced do virtually the same thing, there more chances for phenotypic fixation if there are alternate genetic pathways. Even if some are lost to drift etc., it's not necessary for that one special snowflake to persist because other snowflakes, even if slightly different, are special in the same way.
- The mutation reappears over and over again because fixed population genome is already one step away - whether that step is site mutation or chromosomal duplication to anything in between; the conditions are in place for the positively selected mutation to be one several 'attractors'. Of course, many of those attractors are nonadaptive - and typically selected against.
- The adaptive mutation is fortuitously linked with a genetic element with crossover bias (meoitic drive).
- A very large population is held together over large area for long time, which reduces effect of drift.
- Populations split and then fuse back together, allowing subpopulations to explore possible 'genospace' with combination of drift (which makes drift perversely the accomplice of adaptation), mutation, & selection, then novel adaptive genes can spread in the re-fused expanding population (e.g. neanderthals and modern humans - but we also got some of their bad genes in the bargain).
Please don't confuse patience for adaptationism as support for it :)
ReplyDeleteMicrocephaly not (yet) apparent at a distance from Brazil. Is virgin or mutated Zika + Brazilian (unique?) genome reverse engineering evolution of the brain case / brain? Stuff for science fiction.
ReplyDeleteI meant we need not to think of major adaptations as happening rapidly, as if a single event causes them as a rule. And I'd be the last person to assert support for blanket adaptationism, if that's what you meant!
ReplyDeleteTo anyone reading: this thread reads like we're talking past each other because it doesn't show how many of these are replies. If you read this on the mobile app, you can see the nested replies. Here you can't because we had to change the settings to make the comments more welcoming to users with different browsers.
ReplyDelete"We each have somewhere around 100-150 single nucleotide mutations in the stretches of our DNA that code for proteins."
ReplyDeleteThis is about two orders of magnitude too high. This number is closer to what we actually expect for the genome as a whole. For the exome, the expected (and observed) number of mutations per generation is slightly less than 1.
Integration can certainly be nonrandom - take homing endonucleases for example. This principle's already been employed to create extremely strong genetic drivers using a CRISPR system (Google "mutagenic chain reaction").
ReplyDeleteI don't think it's yet possible to integrate that kind of system into an infectious virus particle though - for which we should all be *profoundly* thankful.
Daniel: It's funny you should say so. I used to assume that number was describing the whole genome but was corrected a while back. So I did write that based on good authority. Could you provide references to back up your comment, please? That's a really important error if indeed it is one. And I'd like to fix it.
ReplyDeleteThere have been various estimates of how many new mutations arise in any gamete. Roughly, at a rate of 10^-8 per nucleotide and 3x10^9 nucleotides in the genome, one would expect around 10, mostly not in coding regions and even those might be in redundant (synonymous) sites. Double that for diploid inheritance. There are also many other sorts of mutations that are much more common, such as copy-number changes and various sorts of slippage (as in microsatellites).
ReplyDeleteI've seen estimates that we each carry around 100 or more dead or dsyfunctional genes, some substantial fraction of which are homozygous. Other estimates have been smaller, as I recall.
There is another thing to consider in this. Not all function is 'coding' in the protein sense, so that the number of functionally relevant mutations per gamete could be much higher than the number in exomes. The fraction of the genome that has 'function' is hotly debated, of course.
But I'm personally not clear how or if the details of these numbers changes the basic ideas one way or the other....
I got some very insightful feedback via Twitter when I posed this question, "Hey #genetics Twitter: Do we have 100-150 single nucleotide mutations across the whole genome or just in the protein-coding %? #help #Refs" But I didn't get a very satisfying answer maybe because it's the wrong question, b ut I know it's because the answer is both complicated and partially unknown, and also 140 characters + genetics acronyms = confusing. So... I think I'm going to go up and edit out any numbers in that paragraph. Without them, it's still accurate.
ReplyDeleteAs I tried to indicate in my previous comment, it's a complex subject. What counts as a 'mutation'? The usual idea, though only a kind of informal one, is that this relates to single-copy DNA (i.e, not microsatellites, telomers, centromeres, and so on), and also that it means comparing parent and offspring. But what cells in the parent and offspring? Even a single sequenced sperm or egg only would reveal differences from some arbitrary parental sample (blood? cheek swab?). And if one asks whether non-coding DNA should count, or non-repetitive, then there are various legitimate answers. I don't think these issues change the points you are trying to make in your post.
ReplyDeleteIt's becoming clearer. If you only count single nucleotide changes, then we have only about 1 in our entire protein-coding genome. However, that portion of the genome has copy number variants too and if you count those nucleotides... you'd have a higher mutation number but is that a meaningful or useful number? Who cares for now. I took out the numbers. All that matters is (a) we have constant mutation and (b) that mutation rate is not so low as merely a single ("~1) A, T, C or G in all of our protein coding genome because there are other ways to mutate and to count it up and estimate rate.
ReplyDeleteThank you to all!
Not really related to the above, however related to the content on this site...ready to groan at this sci-news post?
ReplyDeletehttp://dannyboston.blogspot.com.au/2016/02/neanderthal-genes-make-us-depressed.html
Yes, well, exaggeration and overstatement are the current way of doing business, mutually by journals, investigators, and popular media. Every day needs some dramatic new discoveries!
ReplyDelete