Hacker News Reader

Comment by klmr | Hacker News Reader

Sure, and that happens all the time in unicellular organisms. But it’s fundamentally different in multicellular organisms because they have distinct cells forming the germ line. So any (random) changes to DNA in other cells won’t be transmitted. And while mutations in the germ line do happen, they are also random. So far there’s no known mechanism for systematic, guided DNA changes, and no good reason to assume that such mechanisms exist (on the contrary). Furthermore, the article is discussing epigenetic changes which, by definition, are not encoded in DNA and are instead transient (and thus not inherited trans-generationally).

View on HN

> and thus not inherited trans-generationally

I'm not sure what you're referring to here? There appears to be quite a significant body of literature on epigenetic inheritance -- just to pick a few: [1,2,3,4]. The term "Transgenerational epigenetic inheritance" even has its own wikipedia page: https://en.wikipedia.org/wiki/Transgenerational_epigenetic_i...

[1] https://www.nature.com/articles/ng1199_314

[2] https://www.nature.com/articles/nrg1834

[3] https://www.nature.com/articles/nature05917

[4] https://doi.org/10.1016/j.cell.2014.02.045

Yes, I’m well aware of TEI, it being my former field of research. However, as your link (4) shows there are all kinds of caveats, and usual epigenetic inheritance isn’t transgenerational (in fact, it’s a ubiquitous, “mundane” biological mechanism to maintain cellular state). Your first example — agouti mouse coat colour — is essentially the only firmly established example of TEI in mammals. And, as I explained in another comment, TEI tends to attenuate over generations, and it probably functions very differently in non-mammalian organisms. To date, the existence of pervasive TEI in mammals remains a point of contention. Outside of the agouti locus there’s almost no high-quality, replicated evidence for its existence, and the hypothesis is fundamentally hindered by the lack of a plausible mechanism [1]. Most epigenetics research focuses on other things, and there’s general scepticism that TEI in mammals is all that important [2].

[1] https://www.theguardian.com/science/sifting-the-evidence/201...

[2] https://twitter.com/ewanbirney/status/1014494822525296640

Ok, thanks for clarifying! Is that part of why you got out of the field?

It was a reason, yes. But more generally I started a new job at the time (~ 2 years ago) that took me out of biological research altogether for the moment.

> So far there’s no known mechanism for systematic, guided DNA changes, and no good reason to assume that such mechanisms exist (on the contrary).

Why are we significantly less likely to have mutations on our mitochondria than in our fingers? It seems obvious to me that our DNA has error correction mechanisms, and some subsystems are more stringently error corrected than others. This amounts to a control on how much different aspects are allowed to deviate from the last generation. I would be incredibly surprised if a crocodile, who has been in a state of relative homeostasis for millions of years, has the same rate of mutation as a human. Note that I haven't suggested that a directional bias exists, just a set of mechanisms for changing variability of different traits. I wouldn't be particularly surprised by a mechanism for directional bias in some cases, but I don't assume there is one and I certainly don't feel comfortable making the case for it.

> Why are we significantly less likely to have mutations on our mitochondria than in our fingers?

I’m afraid this is simply not a biologically meaningful statement to begin with. Mitochondria are subcellular organelles. Fingers are complex organs that are formed from spatial arrangements of multiple cell types. Furthermore, mutations happen on the level of DNA, not on the level of either organelles or organs.

Apart from that, mutation rate is (not quite, but more or less) uniform across the whole genome (the exceptions, due to e.g. GC bias, are not relevant here). Viable mutations are not uniformly distributed because of selection; that’s precisely the striking insight that Darwin had (though he was unaware of DNA and genes). Importantly, this does not require a guided error correction mechanism. Simply put, even slight variations in the oxidative phosphorylation pathway of the mitochondrion are likely to kill you during gestation; whereas similarly extreme variations in finger morphogenesis, at worst, make you unable to hold a tool.

> some subsystems are more stringently error corrected than others

This is virtually certainly not the case, because of what I’ve just described. Selection is sufficient to describe the outcome, and there is no evidence whatsoever to point to the mechanism you postulate.

> I would be incredibly surprised if a crocodile, who has been in a state of relative homeostasis for millions of years, has the same rate of mutation as a human.

Evolutionary mutation rate is governed by effective population size and generation time. That’s why fruit flies evolve fast and crocodiles evolve slowly. This, too, doesn’t require variable mutability of different traits, and no biologist has yet seriously suggested such a mechanism.

> [...] Furthermore, mutations happen on the level of DNA, not on the level of either organelles or organs.

This is all besides the point. Mitochondria are just an easy example because of the marked difference in mutation rate, and my loosely worded bit about fingers evolving should obviously be taken as "some parts of the genome associated with the development of fingers."

> Apart from that, mutation rate is (not quite, but more or less) uniform across the whole genome (the exceptions, due to e.g. GC bias, are not relevant here).

Not my field, but I'm pretty sure you're incorrect. Here's a citation:

https://www.ncbi.nlm.nih.gov/pubmed/11057667?dopt=Abstract

>> A few patterns have been found: proteins involved in antagonistic co-evolution (for example, immune genes, parasite antigens and reproductive conflict genes) tend to be rapidly evolving, and there is a correlation between the rate of protein evolution and the mutation rate of the gene. Here we report a new highly statistically significant predictor of a protein's rate of evolution, and show that linked genes have similar rates of protein evolution.

>This is virtually certainly not the case, because of what I’ve just described. Selection is sufficient to describe the outcome, and there is no evidence whatsoever to point to the mechanism you postulate.

This is virtually certainly not the case, because of what I've just cited.

> Evolutionary mutation rate is governed by effective population size and generation time.

I'm not arguing that these aren't factors, but we do have good reason to believe that the rate of change for some subsets of the genome can itself change.

Here's another source:

https://www.ncbi.nlm.nih.gov/pubmed/10469563?dopt=Abstract

>> Our results provide the first substantial statistical evidence for the existence of a regional variation in the synonymous substitution rate within the mammalian genome, indicating that different chromosomal regions evolve at different rates. This regional phenomenon which shapes gene evolution could reflect the existence of 'evolutionary rate units' along the chromosome.

You fundamentally misunderstand the article you cite. It describes population-level events, and it’s entirely compatible with what I’ve said: variable mutation rates, in modern evolutionary biology, are entirely explained by variable selective pressure (see “purifying selection”, which is what I’ve described in my previous comment using your example).

As for the second article, I’ve made reference to that in my previous comment, too (that’s GC bias). And, as mentioned, this isn’t relevant here.

> You fundamentally misunderstand the article you cite.

That seems entirely possible.

> As for the second article, I’ve made reference to that in my previous comment, too (that’s GC bias). And, as mentioned, this isn’t relevant here.

Can you unpack why you think this isn't relevant? The rate of change varying across regions of the genome seems highly relevant to me, even if our current knowledge of it were hypothetically confined to a specific type of change.

GC content (the local ratio of G and C nucleotides, i.e. (#C + #G) / (#A + #C + #G + #T)) varies in patches across certain mammalian genomes. GC content correlates with stability because stacked C–G base pairs are chemically more stable than A–T base pairs. However, while gene density correlates with GC content, gene function is uncorrelated. This means that genes (which tend to be in high-GC regions) in general tend to be (very, very slightly) less susceptible to mutations than non-genic DNA (by contrast, they are vastly less susceptible to mutation than non-genic DNA due to negative selection^1). But the difference in GC content between different genes is purely stochastic, and there’s no mechanism for changing the GC content of a given gene, except by random (!) mutation. This can happen, and it indeed improves the stability of a gene, but the same is true for all genes to the same extent, and it isn’t directed.

^1 I’ve never seen anybody explicitly quantify this but the relative impact of GC content and negative selection on mutation rate must be several orders of magnitude different … at a guess at least thousandfold, more likely millionfold.

> But the difference in GC content between different genes is purely stochastic, and there’s no mechanism for changing the GC content of a given gene, except by random (!) mutation. This can happen, and it indeed improves the stability of a gene, but the same is true for all genes to the same extent, and it isn’t directed.

That seems fine. If some genes are more stable than others, and this can vary by normal random mutation, it can be selected for. I'm still failing to see why this isn't relevant to the current discussion.

Because it’s just stochastic. There’s no mechanism for systematic, guided DNA changes to mutate some genes more than others, which is exactly what I wrote in my initial comment that you took umbrage at. “It can be selected” for only through the blind process of chance, which isn’t really selection at all.

> “It can be selected” for only through the blind process of chance, which isn’t really selection at all.

Selecting is not blind, selecting is sexual or based on survival. The mutation that changes the rate of other mutations is random, but once that mutation occurs it can be selected for via normal means.

Iʼm aware of how evolution works, thanks. I do have a PhD in genetics after all. But this isnʼt what you were talking about here, which is the hypothetical existence of a biological mechanism enacting differential, directional selection.

Evolutionary selection (regardless of whether natural, sexual, artificial or whatever) happens in aggregate over multiple generations, it canʼt account for guided mutations in the germ line that encode “learned” behaviour (which, as my initial comment explained, simply donʼt exist).

I appreciate that you are more well versed in your field than I am. That doesn't mean that you're not making a misstep here. I've known some PhDs and while they often have good points to make in their fields I've also seen some of them eat crow more than once. We're all human.

> But this isnʼt what you were talking about here, which is the hypothetical existence of a biological mechanism enacting differential, directional selection.

It should be noted that I never said directional -- I actually made it explicitly clear in my first post that I was not arguing for a direction, only for modified rates of change: "Note that I haven't suggested that a directional bias exists, just a set of mechanisms for changing variability of different traits." So, I think this is what I've been talking about from the beginning. Perhaps it's due to my lack of familiarity with domain specific language, but I'm not sure that you're understanding what I've been arguing for. Please don't get bogged down in the specifics of this example (it's the pointing finger, not the moon), but I'm not suggesting that cheetahs evolved a tendency to evolve faster running, I'm suggesting that they might have had a period of evolution where a variety of genes related to running were more likely to change in both beneficial and non-beneficial ways, but because faster running was such a beneficial trait it actually made sense for some portion of the population to have offspring who were more likely to have both beneficial and harmful changes happen over the relevant sections of genetic code. In times of less rapid environmental/adversarial change, where stagnation is good enough and the reward for beneficial deviation is more outweighed by the risk of harmful deviation, the rate of change can be toggled back down (again through normal evolution).

1) Do we agree that random mutations of the germline can result in differing rates of change across subsets of the genome?

2) Do we agree that this can affect the viability of offspring?

3) If yes to both of the above, is any component missing for evolution to select for different rates of variability in different subsets of the genome?

> [Being an expert] doesn't mean that you're not making a misstep here.

That’s correct, but we’re not discussing the cutting edge here but relative basics. I’m clearly not articulating myself well but I can assure you that this discussion wouldn’t take place between two biologists because there’s nothing here.

> It should be noted that I never said directional -- I actually made it explicitly clear in my first post that I was not arguing for a direction

You did say that, but, respectfully, you implied directionality — otherwise it would be completely unclear what your initial comment wanted to say, because it seems to specifically have objected to my assertion that no directed changes of the germ line exist.

> Do we agree that random mutations of the germline can result in differing rates of change across subsets of the genome?

Not within a single generation, no. Across evolutionary time, by virtue of selection, yes (that’s in fact the very basis of evolution). But this is irrelevant in the context of this comment thread, started by my initial comment: I asserted that there is no mechanism to encode learned behaviour in the germ line. This would require a direct mediator of mutations on the molecular level, not across evolutionary time but in the here and now. A molecule in the gametes (or their precursor cells). Such a molecule does not exist. Stochastic variation that shapes patterns of mutation across multiple generations is fundamentally a red herring here.

> You did say that, but you meant directionality — otherwise it would be completely unclear what your initial comment wanted to say, because it seems to specifically have objected to my assertion that no directed changes of the germ line exist.

No, I did not mean directionality. I meant what I said, and was very explicit about it. In my first post I quoted the exact bit of your comment I was replying to, which went like this: "So far there’s no known mechanism for systematic, guided DNA changes, and no good reason to assume that such mechanisms exist (on the contrary)." I wasn't even sure if there would be disagreement between us when I originally replied to that, because I'm not sure how broad the definitions you're using for "systematic" and "guided" are. I could see arguing that evolution changing the variability of a subset of the genome is "guided" in some sense, as it can "guide" the process towards more changes in a particular subsystem, but I could also see not arguing that since it lacks direction. Language is mushy. I was trying to qualify your statement, and I posed questions and made statements of my own to that end, but I wasn't sure if my positions were in conflict with yours until you replied. Now I think our positions probably weren't in conflict, but you made assumptions about what I meant to convey based on the larger context of your own post rather than the actual content of mine.

So, to clarify, in that comment by me, “guided” (and, in fact, my whole comment) definitely implies “directional”. But let’s not get hung up on this one word. The comment was about encoding learned experience in the DNA of gametes (i.e. Lamarckism), and the fact that this requires a (nonexistent) mechanism.

Reply on news.ycombinator.com