Why our intuition about mutation rates was wrong
18 February 2016
Tweet
There’s a general feeling among many of us that work on rates of evolution (well, at least with the folks that I regularly chat to) that DNA replication is one of the main sources of new mutations. A great new paper in PLOS Biology [1] points out that this conclusion was premature.
The intuition comes from a diverse set of observations which show correlations between rates of DNA replication (or proxies of it) and rates of molecular evolution. For example, it has been demonstrated many times in many different taxa that generation time correlates with long-term substitution rates (e.g. [this paper], [and this one]). The thinking goes something like this: all else being equal, species with shorter generation times will copy their DNA more often, and so (given a roughly constant mutation rate per copying event across species) accrue more mutations per unit of time. Other observations in the same vein include differences in rates of evolution between chromosomes (chromosomes that spend more time in males tend to evolve faster, and there’s more DNA copying per unit of time in the male germ line, e.g. [this paper]); links between rates of evolution and height in plants (shorter plants evolve quicker, and copy their DNA more often e.g. [this paper]); and the accumulation of germ line mutations in humans (the number of mutations in a child increases roughly linearly with the age of the father, not so with the mother, e.g. [this paper]). Taken together, it seems like all of these results can be explained simply: DNA replication is an important source of new mutations.
Now, studies like the ones I mention above are all correlative. So of course we have to be careful - just because rates of DNA replication correlate with rates of evolution doesn’t give us concrete evidence that one causes the other. Correlations often have nothing to do with any causal link between the two measured variables. But we kept seeing things that linked rates of DNA replication to rates of evolution, and so a causal link between the two seemed likely to explain these observations, even if the correlations don’t prove it. Nevertheless, all sensible studies that report correlative evidence should have a long list of potential causes of the observed correlation, and most do. DNA replication will be prominent on the list, because it’s an obvious candidate. But the list should contain other things too. What the new PLOS study points out is that we had prematurely crossed something off our list of potential causes: endogenous and exogenous mutagens.
Cells are full of mutagens which damage DNA all the time. Most of this damage is repaired, but some certainly leads to mutations. The point is that until now we thought that rates of mutation that result from these mutagens would not be linked to rates of DNA replication. And because of that, we didn’t think that mutations caused by mutagens could explain correlations between rates of DNA replication and rates of evolution. We were wrong.
The PLOS paper points out, rather beautifully, a gaping hole in our reasoning. Once you fix this hole, it’s clear that we should often expect rates of mutation from mutagens to correlate with rates of DNA replication. The paper provides a mathematical model which shows exactly how this can work, as well as intuitive explanations for the phenomenon. The key point to me is the realisation that that DNA replication converts DNA damage into irreversible mutations. Because of that, in general, the more DNA replication you have, the more frequently you convert damage into mutations, and the higher your mutation rate. In retrospect, this seems like a simple observation, but this belies the fact that the study of rates of molecular evolution is at least 50 years old (counting from Zuckerkandl & Pauling’s book chapter in 1962), and nobody figured it out until now. Perhaps it seems simple because the paper is so well written. You should read it.
A simple example might help. Imagine a mutagen that causes DNA damage at a constant rate. The cell will repair the damage, but not perfectly and not immediately. Unrepaired damage has a chance of causing a mutation during DNA replication. If you leave a cell long enough it will reach an equilibrium amount of damage: i.e. after a while you’ll find that there’s a roughly constant number of damaged nucleotides awaiting repair. If DNA replication happens quickly relative to the rate of DNA damage, most damaged nucleotides will result in mutations. As you slow down the rate of DNA replication, you give the cell more time to repair DNA damage, so a smaller proportion of the damaged nucleotides will result in mutations. Because of this, as the rate of DNA replication slows down, so does the mutation rate attributable to mutagens. Moreover, as the rate of DNA replication slows, the correlation between rates of DNA replication and rates of mutagen-derived mutation strengthens. Once the rate of DNA replication is much slower than the time it takes a cell to reach the DNA repair equilibrium above, any further slowing of DNA replication will lead to a perfect correlation between the rate of DNA replication and the rate of mutagen-derived mutation.
What does this mean? It means we have to go back to the drawing board for how we interpret many of the observations of links between rates of DNA replication and rates of molecular evolution. Specifically, we have to include mutagens in our list of possible causes of many observed correlations. This, in turn, means that we need to think of new ways to distinguish the relative importance of DNA replication and mutagens in causing new mutations. That’s the fun part.
[1] Gao Z, Wyman MJ, Sella G, Przeworski M. 2016. Interpreting the Dependence of Mutation Rates on Age and Time. PLoS Biol 14:e1002355.
There’s a general feeling among many of us that work on rates of evolution (well, at least with the folks that I regularly chat to) that DNA replication is one of the main sources of new mutations. A great new paper in PLOS Biology [1] points out that this conclusion was premature.
The intuition comes from a diverse set of observations which show correlations between rates of DNA replication (or proxies of it) and rates of molecular evolution. For example, it has been demonstrated many times in many different taxa that generation time correlates with long-term substitution rates (e.g. [this paper], [and this one]). The thinking goes something like this: all else being equal, species with shorter generation times will copy their DNA more often, and so (given a roughly constant mutation rate per copying event across species) accrue more mutations per unit of time. Other observations in the same vein include differences in rates of evolution between chromosomes (chromosomes that spend more time in males tend to evolve faster, and there’s more DNA copying per unit of time in the male germ line, e.g. [this paper]); links between rates of evolution and height in plants (shorter plants evolve quicker, and copy their DNA more often e.g. [this paper]); and the accumulation of germ line mutations in humans (the number of mutations in a child increases roughly linearly with the age of the father, not so with the mother, e.g. [this paper]). Taken together, it seems like all of these results can be explained simply: DNA replication is an important source of new mutations.
Now, studies like the ones I mention above are all correlative. So of course we have to be careful - just because rates of DNA replication correlate with rates of evolution doesn’t give us concrete evidence that one causes the other. Correlations often have nothing to do with any causal link between the two measured variables. But we kept seeing things that linked rates of DNA replication to rates of evolution, and so a causal link between the two seemed likely to explain these observations, even if the correlations don’t prove it. Nevertheless, all sensible studies that report correlative evidence should have a long list of potential causes of the observed correlation, and most do. DNA replication will be prominent on the list, because it’s an obvious candidate. But the list should contain other things too. What the new PLOS study points out is that we had prematurely crossed something off our list of potential causes: endogenous and exogenous mutagens.
Cells are full of mutagens which damage DNA all the time. Most of this damage is repaired, but some certainly leads to mutations. The point is that until now we thought that rates of mutation that result from these mutagens would not be linked to rates of DNA replication. And because of that, we didn’t think that mutations caused by mutagens could explain correlations between rates of DNA replication and rates of evolution. We were wrong.
The PLOS paper points out, rather beautifully, a gaping hole in our reasoning. Once you fix this hole, it’s clear that we should often expect rates of mutation from mutagens to correlate with rates of DNA replication. The paper provides a mathematical model which shows exactly how this can work, as well as intuitive explanations for the phenomenon. The key point to me is the realisation that that DNA replication converts DNA damage into irreversible mutations. Because of that, in general, the more DNA replication you have, the more frequently you convert damage into mutations, and the higher your mutation rate. In retrospect, this seems like a simple observation, but this belies the fact that the study of rates of molecular evolution is at least 50 years old (counting from Zuckerkandl & Pauling’s book chapter in 1962), and nobody figured it out until now. Perhaps it seems simple because the paper is so well written. You should read it.
A simple example might help. Imagine a mutagen that causes DNA damage at a constant rate. The cell will repair the damage, but not perfectly and not immediately. Unrepaired damage has a chance of causing a mutation during DNA replication. If you leave a cell long enough it will reach an equilibrium amount of damage: i.e. after a while you’ll find that there’s a roughly constant number of damaged nucleotides awaiting repair. If DNA replication happens quickly relative to the rate of DNA damage, most damaged nucleotides will result in mutations. As you slow down the rate of DNA replication, you give the cell more time to repair DNA damage, so a smaller proportion of the damaged nucleotides will result in mutations. Because of this, as the rate of DNA replication slows down, so does the mutation rate attributable to mutagens. Moreover, as the rate of DNA replication slows, the correlation between rates of DNA replication and rates of mutagen-derived mutation strengthens. Once the rate of DNA replication is much slower than the time it takes a cell to reach the DNA repair equilibrium above, any further slowing of DNA replication will lead to a perfect correlation between the rate of DNA replication and the rate of mutagen-derived mutation.
What does this mean? It means we have to go back to the drawing board for how we interpret many of the observations of links between rates of DNA replication and rates of molecular evolution. Specifically, we have to include mutagens in our list of possible causes of many observed correlations. This, in turn, means that we need to think of new ways to distinguish the relative importance of DNA replication and mutagens in causing new mutations. That’s the fun part.
[1] Gao Z, Wyman MJ, Sella G, Przeworski M. 2016. Interpreting the Dependence of Mutation Rates on Age and Time. PLoS Biol 14:e1002355.