Endogenous Retroviruses in the Case for Common Ancestry

medicodon take a look at this: http://www.evolutionnews.org/2011/05/more_points_on_ervs046761.html mk

I am looking for documentation in the scientific literature of a particular ERV integrating at EXACTLY THE SAME LOCUS IN EXACTLY THE SAME POSITION in two different species. Could you (or anyone else) direct me to an article or study that shows this? Best regards, MD (board certified radiologist) Medicodon

Dave, your article of May 10 only recently came to my attention. I disagree with your statement that "As it turns out it probably isn’t very likely at all for sperm cells to be either infected with a provirus or survive the infection. ....... They are also very active cells and even if infected would likely be hobbled enough to not be successful at fertilizing an egg." Sperm cells of virtually all animal species are axtremely permeable to foreign DNA or RNA sequences that can be taken up after ejaculation when spermatozoa are exposed to the environment and delivered to oocytes at fertilization. Exploiting this spontaneous feature, spermatozoa have been used as vectors to introduce new genetic traits in various animal species. In addition, spermatozoa are endowed with an endogenous reverse transcriptase activity that allows to reverse transcribe in cDNA copies exogenous RNA molecules that are taken up. If these events can take place occasionally in nature they can be regarded as a potential source of mutation in the host genome with some relevance for the evolutionary processes. You may have a look at the following review articles: 1. C. Spadafora Sperm cells and foreign DNA: a controversial relation. (1998) BioEssays 20, 955-964 2. K. Smith and C. Spadafora Sperm-mediated gene transfer: applications and implications (2005) BioEssays 27, 551-562 corrado

This suggests to me that there is no *particular* location where the SAME ERV is located itself. –PaV You've basically answered my question when you said earlier that "we can talk abstractly about a single spot, but the data concerns lots of ERVs having fixed in lots of spots." As far as fixation, I appreciate the explanation and the statistical means used. You then say: "So ERVs that are present in both macaque and humans at the same spot will be extremely likely to be fixed in both species)" In the paper I read, which used the LTR's in assessing integration times, etc, it didn't seem to me like they were comparing ERV's that are present at the same spot in both humans and macaques. There'e no explicit indication of that. Is that simply to be implied? In the paper they compare the 5' and the 3' LTR's. It seems that the 5' LTR is from the "host" and the 3' LTR is inserted with the ERV. Is that a correct inference? PaV

The language that has been used concerning the ERV’s has suggested a.) that we’re dealing with fixation --PaV Agreed. b.) and..that we’re dealing with fixation at a *particular* “hot spot”. Not really. We can talk abstractly about a single spot, but the data concerns lots of ERVs having fixed in lots of spots. And the experimental (cell culture) data concerns lots of viruses inserting in lots of different spots. So focusing on a single "hot spot" is not very helpful IMO given that we are dealing with how ERVs generally behave in genomes and populations. If we’re dealing with limited genomic sequencing—which I suspect to be the case—then in what way can we say that “fixation” has taken place? Sampling size can be an issue for some purposes, but for the most part does not affect the inferences we're making here. The genome sequence is a single or at most an amalgam of a few individuals. So it is indeed a small sample. Yet fixation is determined or inferred by a number of means. In some cases a large sample of the population is tested for the locus. If all of 100 geographically diverse people have the insert, its very likely that its fixed. Not 100% certain, but pretty likely. In other cases, if two distant species share the same insert, and the (presumed) time since speciation is very long, then fixation is inferred because (barring very rare circumstances) an allele can (statistically speaking) only remain polymorphic for a limited period of time before it is either lost in one population or fixed. So if a long time has elapsed and the species share an insertion, fixation is inferred. In addition to the speciation time, a long interval of time can also be inferred by the accumulated mutations in the ERVs themselves. (In humans the average coalescence time at a given locus (i.e. how long before its completely replaced by some individual allele) is around a million years if I'm not mistaken. This is a function of population size. So ERVs that are present in both macaque and humans at the same spot will be extremely likely to be fixed in both species) This suggests to me that there is no *particular* location where the SAME ERV is located itself. --PaV I don't understand this last point, perhaps you could try restating it some other way. great_ape

Looks like a post is in the spam filter. Before addressing your response, there is a clarification or two that is needed. The language that has been used concerning the ERV's has suggested a.) that we're dealing with fixation, and b.) that we're dealing with fixation at a *particular* "hot spot". As to a.), I've read a paper that is basically using information that comes from the human genome project, and other sequencing work. But how representative is that data for entire species, whether human, baboon, orangutan, or chimp? If we're dealing with limited genomic sequencing---which I suspect to be the case---then in what way can we say that "fixation" has taken place? It seems to me that the most one can say is that the *allelic* frequency is at least such and such---and no more. As to b.) it would appear that certain ERV's just happen to appear in certain chromosomal locations, and, based on a specific sequence, and a BLAST survey, these locations are identified. This suggests to me that there is no *particular* location where the SAME ERV is located itself. If so, this completely changes the kind of thinking that's needed. I would appreciate your take on these two points before we proceed any further. PaV

(continued from above) #3. Another way of saying that is that in nature we see, more or less, stable populations. What that means is that over significant periods of time, on average, any two mating animals will give rise to two members of the population. --Pav a) This isn't true of animal populations today nor is there any reason to believe it has been true in the past. Hence extinction. b) Even if it were true, it would be largely irrelevant for the purposes of our discussion. Gametes are sampled each generation by and can produce drift even under stable population sizes. #4. “...Well, if you have 2N insertion events, then one of them will become fixed. The problem with dealing with insertions, though, is that we don’t know really know what the mechanism is for these kinds of insertion. We know a fair amount about how viruses insert. Certainly there are some details left to work out, but arguing that this rules out the 2N insertions yields one fixed events is an extreme example of retreating into a gap in knowledge to prevent yourself from accepting the inevitable conclusion. There are studies examining patterns of viral integration, both in the wild and in cell culture. #5 [assundry calculations] Isn’t this proof that insertions and SNP’s are entirely different? The short answer is no, but I'll elaborate a bit. When your math involves one generation a year and still demonstrates a SNP would take 40 billion years to fix in a population size of 10k, it's safe to assume a wrong turn was taken somewhere along the way. There are a number of ways to approach this issue, some more complex than others, I give the crude abridged version below. To calculate the rate of the same SNP occurring at the very same location in the genome... Why exactly would you want to calculate it for the same location in the genome? Where the SNP or ERV insertion occurs is irrelevant. This is the first indication that something has gone wrong. Let's use the human per site mutation rate from Nachman and Crowell (2000) of 2.5 x 10-8 mutations per nucleotide site. (175 novel mutations per diploid genome per generation). The number of base replications is already taken into account in the mutation rate so your second 3 billion term is redundant. The probability of mutating to the same base is taken into account already in the way observed mutation rate is calculated, so that term is also redundant also. (The math for nucleotide transition models that calculate probability of identity and nonidentity at a locus as a function of time is more complex; see a good popgen or molecular evolution text for details on transition models if you're interested). If you calculate that out, you get roughly 175 novel nucleotide (SNP) mutations in each newborn individual. Crudely, that's about 1,750,000 novel nucleotide mutations in a given generation in a population of 10k breeding individuals. So clearly that's enough such that we'd expect at least a number of those new mutations (assuming its neutral) to ultimately fix in the population/species. (1.75 million / 20,000 yields 87.5 lucky nucleotide variants that will reach fixation) Now, of course, there are many such generations and not every change will be neutral. Now what's the situation for ERVs/SINEs/LINEs. Mutation rates are less well defined and certainly much more variable than the biochemical processes underlying nucleotide mutation. Current estimates for LINE/SINE insertions are on the order of 1 insertion per 50 to 100 births. These estimates are made, in part, by observations of the number of de novo insertions disrupting genes and causing disease, but other methods are used as well. The number of fixations when looking at chimp/macaque, etc. are concordant with that figure. ERV data is less clear. There are about 3500 ERVs that fixed in humans since the human lineage diverged from macaque. I'm guessing less than 1000 since chimp, but I'm not certain. Let's use 750 as our number of fixed ERV insertions since the . What would the ERV germline insertion rate have to be to come up with this figure since, say, human and chimpanzee diverged? Roughly 750 * 20k = 15 million insertions. Sounds daunting. But let's run the numbers. How many folks do we have to work with? That's 240,000 twenty-five year generations over roughly 6 million years or 10k humanish things per generation crudely yields 2400000000 births. So that roughly 1 ERV insertion in every 160 births (or, rather in the germ cells that yield the birth) to support current genomic observations. Not crazy, particularly given that viral outbreaks could have caused the germline insertion rate to shoot up wildly during some parts of history. I ran through this quickly, so I may have screwed up something, but the take home message is that you don't need astronomical numbers/time to get the kind of fixation rates we're talking about. great_ape

Pav, Here's my assessment of your analysis: #1 Well, one of the things that separates the insertion from a SNP is that the insertion can only take place at certain sites (apparently) --PaV Yes, but that's still leaves a lot of possible sites for insertion by any given ERV sequence. It's very rare for a virus to have a specific/unique target preference in the DNA. We can recover insertion loci from experimentally infected cells. That is, we can find out where they inserted. The integration process is still effectively random with respect to location. You needn't take my word for it. Look up papers that have experimentally examined where retroviruses inserts. There can be hotspots, yes, but in general it's pretty sporadic. You must be able to make the distinction between something not being random with respect to every factor, yet still being random with respect to the issue at hand. This is a common sticking point. I used to think people were conflating the different meanings of random for rhetorical purposes--and I'm sure some folks do--but I now think that many people are just confused. #2 ...according to what you’re proposing, the ONE individual in this population of 100,000 that has a particular insertion at a particular locus on one of its two chromosome pairs is, randomly going to overtake the 99,999 individuals who, at this very same locus, have the very same nucleotide base a) Not exactly. That one inserted allele of that particular genomic locus is going to overtake all the other non-inserted locus. Because of recombination, we're not dealing with individuals in population genetics. b) the effective breeding population size of humans is estimated to have been around 10,000 longterm in the relevant time period. c) you don't have to believe me; it can be demonstrated mathematically (P=(1-e^(-4Nsq))/(1-e^(-4Ns)) (shown by Kimura; where q is the initial frequency of the allele, N is effective population size, and s is the selection coefficient (for our purposes it's zero)). and by simulation that the probability of any one such allele beating out all other such alleles is 1/2N (in other words, its initial frequency). So, for every 20,000 germline insertions of ERVs (that go on to form a zygote), roughly one would fix simply by neutral drift in ancestral humans. (continued) great_ape

according to what you’re proposing, the ONE individual in this population of 100,000 that has a particular insertion at a particular locus on one of its two chromosome pairs is, randomly going to overtake the 99,999 individuals who, at this very same locus, have the very same nucleotide base --Pav Pav, Going thru your math and reasoning is going to take a moment, but I'll address where I believe you have gone astray later tonight. Hopefully my earlier post that's stuck in the filter will get thru by then. great_ape

Population genetics is the study of the allele frequency distribution and change under the influence of the four evolutionary forces: natural selection, genetic drift, mutation, gene flow and selective mating Take away mutations and what have you left? --Pav You do realize that we can still read the text that isn't bolded, don't you? I count 3 forces remaining when mutation is removed. And I don't see any specification of "random" mutation as opposed to any other kind. That's because the study of population genetics is about the polymorphic marker, not its origin per se. So, all pedantry aside, your original statement is still very much wrong regardless of which parts of the definition of population genetics that you bold for emphasis. great_ape

hmmm.. My latest post may have gotten hung up in the spam filter because of the refs included... great_ape

ERV phylogenies and some references: It occurred to me after my earlier post that, in addition to PCR-amplifying the same locus from different species to assess insertion status of ERVs, the PCR products are also sequenced in almost every such study. The sequence shows accumulated nucleotide differences that are themselves informative and can rule out contamination issues, etc, that might otherwise confound the results. So I think that, with the notable exception of the smaller sample size, these studies are just as good for proving the point. As we get a denser sampling of full genomes, the sort of in silico comparisons you're looking for will be available (see macaque paper below), but below I list the closest thing so far that I've come across. I think there is more from the rice genomes comparisons as well. This one is a bit dated but freely available. It's not a whole genome study, but it is a good example of what has been done: Constructing primate phylogenies from ancient retrovirus sequences Proc Natl Acad Sci U S A. 1999 Aug 31;96(18):10254-60. Here is a companion paper to the macaque genome that did a whole genome comparison of ERVs, SINEs, and LINEs Unfortunately it requires a Science subscription. Some Highlights, though: "Similar to the human genome, the rhesus macaque genome contains over half a million recognizable copies of endogenous retroviruses (ERVs) and their nonautonomous derivatives, with the great majority being present or fixed before the hominoid-OWM split" In other words, most of the ERV are shared at the same location in both humans and macaques. (The same goes for LINEs and SINEs.) For ERVs, both human and macaque show ~3500 new insertions that have fixed since the separation of the lineages. For any of these (500,000 - 7000) shared ERV insertions, the complete and unambiguous absence of the insert at the orthologous location in the chimpanzee genome would, in principle, be evidence against common descent. (There's a YEC research program for you, by the way, any takers?) At the end of the day you'll find 99.99% of inserts support common descent. Sometimes stubs are left behind where the LTRs are recombined out. Sometimes the entire locus may have been lost in one or more species making analysis impossible. You might find a few blips, but the overwhelming majority will be consistent with the hypothesis of common descent. Hard to ask for much more in the way of support. reference: Mobile DNA in Old World Monkeys: A Glimpse Through the Rhesus Macaque Genome Science Vol. 316. no. 5822, pp. 238 - 240 Here is the the great ape phylogeny from the same group using SINE insertions: Alu elements and hominid phylogenetics. Proc Natl Acad Sci U S A. 2003 Oct 28;100(22):12787-91. Epub 2003 Oct 15 Here is an old world monkey phylogeny based on SINE insertions. It is in agreement with the phylogeny based on nucleotide divergence. Common descent is the only plausible explanation for this. A mobile element based phylogeny of Old World monkeys. Mol Phylogenet Evol. 2005 Dec;37(3):872-80. Epub 2005 Jun 3. Perhaps there's someone lurking, also, that knows of additional refs that would be useful outside the mammalian world. great_ape

hi all, tied up at work for the moment; will post reply to your questions later this evening. great_ape

I am ignorant, and ask for help. If two separate cells from the same individual were to become infected by the same type of ERV, how would the genetic material from the ERV become incorporated into the host DNA, and would the ERV infect the same loci in the host genome of both cells? What gets the ball rolling, precisely? jaredl

great_ape, is there anything wrong with the mathematics I used above? I'd be interested in where I might be wrong. PaV

Orthologous loci in different species might be explained by the same virus infecting different species at the same preferred insertion point but there doesn't seem be much room for dispute in how ERVs find their way into heritable DNA - the same way they get into somatic DNA - by invading a cell and reverse transcription. It's a classic smoking gun. Means, motive, opportunity, and weapon. An eyewitness is all that's lacking. That's still enough to get a conviction and a death sentence. DaveScot

DS, True, but if there are problems with the "only game in town" presumed explanation (assuming that PaV's assesment is correct), then I'd rather stick to the "I don't know at this point" approach. Maybe that's just me. (Like John Baez, I leave that town for the time being.) Atom

atom How about “we don’t know yet”? I think I've been careful to call it a hypothesis rather than a theory. In the literature it's often qualified as a presumption. Just because something isn't proven doesn't mean we can't presume something is correct. Nobody actually saw South America and Africa connected together at one time. Nobody saw meteors making all the craters on the moon. Nobody has gotten a sample of the earth's core to prove it's molten iron, and so forth. DaveScot

A caveat to what I just posted is the fact that there appears to be numerous preferred (non-random) sites where ERV's can happen. So even if there were a massive infection taking place in a population, in order to arrive at a fixed ERV at a fixed site, we would probably further have to add that the particular infection that invaded the germ line, also, in some way, had a preference for a particular site. While this is plausible, I wouldn't want to say it's probable. I have to run..... PaV

"If not germline infection then how do you propose all those ERVs became fixed by the thousands in the genomes of each of many vertebrate genomes examined so far?" DaveScot I've been scratching my head a lot (and still continue to do so), and the only thing that makes sense to me---and I think it's reasonable to think so---is that you have an infection that effects the majority of the population all at once. I'm thinking that if, e.g., 75% or more of the germ line was infected all at once, that from there, just due to randomness, it's possible for an ERV, at a particular site, to become fixed, and thus serve as a 'marker'. That's the best I can think of. As I said, I continue to scratch my head. (It would make a big difference, of course, if we knew how the original, let us say, "infection" of the germ line occurred. I'm not aware that anyone knows that.) PaV

DS

I’ve yet to hear any other plausible explanation and absent that the germline infection hypothesis is the only game in town.

How about "we don't know yet"? In the words of John Baez:

But, the "only game in town" argument is still flawed. Once I drove through Las Vegas, where there really is just one game in town: gambling. I stopped and took a look. I saw the big fancy casinos. I saw the glazed-eyed grannies feeding quarters into slot machines, hoping to strike it rich someday. It was clear: the odds were stacked against me. But, I didn't respond by saying "Oh well - it's the only game in town" and starting to play. Instead, I left that town. It's no good to work on string theory with a glum attitude like "it's the only game in town." There are lots of other wonderful things for theoretical physicists to do. Things where your work has a good chance of matching experiment... or things where you take a huge risk by going out on your own and trying something new.

Atom

PaV If not germline infection then how do you propose all those ERVs became fixed by the thousands in the genomes of each of many vertebrate genomes examined so far? I've yet to hear any other plausible explanation and absent that the germline infection hypothesis is the only game in town. DaveScot

Oops! I really should be more careful! The final number in #92 should be 0.4 x 10^9, or 400 million years. Again, an almost virtual impossibility for the fixation of an ERV insertion into a mammalian line, and a death knell, it seems to me, of the neutral theory. PaV

Population genetics is based on random mutations.–PaV great_ape: "Not so. Population genetics is based on the behavior of allele/character states in populations." I was contrasting the non-randomness of the insertions with the randomness of either recombination events or SNP's, etc. I ask you, would Kimura have written his book on the Neutral Theory if random mutations didn't exist? Here's what Wikipedia says: "Population genetics is the study of the allele frequency distribution and change under the influence of the four evolutionary forces: natural selection, genetic drift, mutation, gene flow and selective mating." Take away mutations and what have you left? So, I hope we can stop with the pedantry. (That includes you Mung) PaV

great_ape: "Once it inserts, it becomes an allelic character, and whether neutral or deleterious, it behaves very much like a SNP, microdeletion, etc, in population genetics." Well, one of the things that separates the insertion from a SNP is that the insertion can only take place at certain sites (apparently). SNP’s can occur anyplace. Leaving this distinction aside for the moment, for fixation of this insertion to take place, according to what you’re proposing, the ONE individual in this population of 100,000 that has a particular insertion at a particular locus on one of its two chromosome pairs is, randomly going to overtake the 99,999 individuals who, at this very same locus, have the very same nucleotide base. How is it all reasonable to assume that this ONE different component---and NEUTRAL at that---is going to overtake the other 99,999 homozygous nucleotide bases that exist at the same locus? If you start saying…..“Well, if this ONE has a lot more offspring than the others……” Well, isn’t it possible, and reasonable to assume, that there will be lots of the 99,999 individuals without the insertion that will have even more offspring (as paired mates) than this ONE? Really, it strikes me as preposterous to assume that ONE in 200,000 is going to swamp the other 199,999 rather than the 199,999 swamping the ONE. And bear in mind that one of the things that Fred Hoyle quickly points out is that populations have to be "normalized". Another way of saying that is that in nature we see, more or less, stable populations. What that means is that over significant periods of time, on average, any two mating animals will give rise to two members of the population. Seen in this fashion---which corresponds to nature itself---on average, the ONE insetion will give rise to ONE insertion. But because of stochastics, it should quickly disappear. That's why the probability of extinction is so high. The way population geneticists deal with the huge improbability of fixation is to say, “Well, if you have 2N insertion events, then one of them will become fixed. The problem with dealing with insertions, though, is that we don’t know really know what the mechanism is for these kinds of insertion. Yes, it is supposed to be through the maternal cell line, but we don’t have an insertion rate into the genome. This makes it almost impossible to assess the situation. So, let’s leave the insertion to the one side, and simply ask ourselves what the numbers look like for the fixation of a SNP at a particular locus in the genome since we have some idea what the mutation rate is for SNP’s. The mutation rate is roughly 1 x 10^-8. To calculate the rate of the same SNP occurring at the very same location in the genome (the condition for building up to the 2N needed mutation events) is: (1 locus/3 x 10^9 nucleotide bases) x (10^-8 mutations/base replication) x (3 x 10^9 nucleotide bases per replication=generation) x ¼ nucleotide base options= 2.5 x 10^-9 identical nucleotide base replacement at locus “x”/generation. For an entire population, there would be 2N such replacements taking place per generation, so that we have a rate of replacement, for the population, of 2N x 2.5 x 10^-9 replacements/generation, which, for N= 10^5 means 5 x 10^-4 replacements/population/generation. But we need 2N such replacements for fixation. So how many generations of this population are necessary for these 2N replacements to take place? Well, for a population of 10^5, that means 2 x 10^5 replacements x 1 generation/5.0 x 10^-4 replacements = 4 x 10^10 generations. Assuming, conservatively, a generation takes one year, then this represents 40 billion years. Thus, we conclude that if we treat this insertion as a SNP, then it is impossible for the insertion to become fixed. Isn’t this proof that insertions and SNP’s are entirely different? PaV

Davescot, Such studies have been conducted for SINEs/LINEs in apes...I think ERVs as well, but I need to rescan the literature to refresh my memory. I think Eichler's group did an ERV study a few years back. I'll double-check the literature and post back here. One reason such completely in-silico studies might be scarce is that it requires at least 3 species' sequences from the same Order. We now have that with human-chimp-macaque, for instance, but that's an unusual case. Two species don't really allow you to infer much without consulting a third, and this until recently was done via PCR so the number of loci examined were modest. great_ape

great_ape I was hoping you could respond to my comment 72 asking for whole genome comparisons of othologous loci. The only thing I was able to find in how orthologous loci for the same ERV in different species are identified was DNA amplification using primers with viral LTRs and flanking sequences thought to not be part of the provirus. This was done before whole genome sequences were available. The same search for orthologous loci can now be done entirely with something like BLAST using whole genome databases and nothing more. Has this been done? It could reveal flaws in the assumptions underlying the inferences drawn from othologous loci or it could serve as even stronger evidence in support of them. It would certainly be superior to DNA amplification techniques. DaveScot

Patrick, One of my comments to you just got released from the spam filter. Sal scordova

The insertion of the ERV is non-random.

Irrelevant.

Population genetics is based on random mutations.

False. Population genetics is based on initial frequency of the allele, the size of the population, and the "fitness" or "selective value" of the allele.

How can the same phenomena explain both situations?

What a confusing (confused?) question. From Wikipedia:

By the explanandum, we understand the sentence describing the phenomenon to be explained (not that phenomenon itself); by the explanans, the class of those sentences which are adduced to account for the phenomenon (p.152).

Population genetics is based on the behavior of allele/character states in populations. These alleles can be neutral, nearly neutral, deleterious, or beneficial. These alleles can be introduced into the population by random mutation, migration, or even transgenics. Doesn’t matter.

That's more like it. Mung

But does any of this data point to speciation by any particular mechanism? --Jerry To my knowledge, none of this data speaks directly to mechanisms of speciation. And I certainly am not claiming that ERVs/SINEs provide a strong argument against ID itself--particularly since we still may have much to discover about their biological properties--I'm simply saying that they are a powerful argument for common descent and against YEC. A decisive argument in my opinion. ...or does it just show that they seemed to have a common ancestor at one point in time and how they diverged and formed different characteristics is still at best a guess? I think that's a fair assessment. To qualify, though, sometimes (we believe based on data) ERV/SINE insertions are commandeered and are a component of the divergence process. great_ape