Saturday, March 15, 2014

How does molecular biology overthrow the Modern Synthesis?

I think the hardened version of the Modern Synthesis is inadequate to describe 21st century evolutionary biology. I think that it didn't adequately recognize Neutral Theory and random genetic drift and it didn't place enough emphasis on macroevolution and the possibility of hierarchical modes of evolution.

There are a whole host of scientists who want to overthrow the Modern Synthesis for a variety of other (stupid) reasons. Most of them have no idea that the Modern Synthesis has (or should have) been replaced 40 years ago.

Here's another example from last week's issue Science (March 7, 2014). Susan M. Rosenberg and Christine Queitsch have an article entitled "Combating Evolution
to Fight Disease" (Rosenberg and Queitsch, 2014). They begin with ....
Traditional evolutionary biology began in the 1930s with the "modern synthesis," which fused Darwin’s theses on phenotypic variation and selection with Mendel’s concepts of genetic inheritance to explain the source of biological diversity. This synthesis predated knowledge that genes were made of DNA and of the structure of DNA and how it replicates. Thus, molecular mechanisms could not be integrated into concepts about how phenotypic variation is generated. Instead, assumptions had to be made about the origins of the variation that drives evolution. Among the cornerstone assumptions were that mutations are the sole drivers of evolution; mutations occur randomly, constantly, and gradually; and the transmission of genetic information is vertical from parent to offspring, rather than horizontal (infectious) between individuals and species (as is now apparent throughout the tree of life). But discoveries of molecular mechanisms are modifying these assumptions.
This is fairly typical of modern critics of evolution. They begin with the assumption that evolutionary theory hasn't changed since the 1930s. What that tells us is that, right from the beginning, they don't know what they are talking about.

So, what are the "new" revelations that molecular mechanisms have uncovered?
In at least two ways, heritable variation can be generated by proteins, not DNA (1). Spontaneously self-aggregating alternative conformations of some proteins—prions—can flip into their aggregated state and change a cell’s phenotype in an environmentally responsive manner with no change to DNA. The change is transmissible vertically, parent to offspring cell, as well as horizontally, to other cells in which the proteins come in contact. Another mechanism involves chaperones such as heat shock protein 90 (Hsp90), proteins that massage subideal (mutant) proteins into functional conformations but abandon their regular client proteins during heat and other stresses that destabilize proteins. This causes a stress-inducible release of phenotypic diversity, which may drive evolution (with phenotypes ultimately stabilized by subsequent genetic changes). Both of these molecular mechanisms of protein-based inheritance are major departures from the modern synthesis views of solely mutation directed variation, solely genetic inheritance, and independence of the generation of variation from environmental conditions.
Yawn. We've known about prions for 25 years. The proteins are encoded by genes and certain variants in certain species can adopt different conformations depending on conditions within the cell. Move along folks, there's nothing to see here.

We've known about the role of molecular chaperones for thirty years. We've known for almost twenty years that chaperones can assist in protein folding so that some polypeptides that cannot fold spontaneously on their own in a reasonable length of time can nevertheless function due to chaperones. Chaperones are very ancient proteins. This means that "subideal" folders have been around for several billion years. In times of stress these "subideal" proteins may not be able to fold correctly. If the proteins are important then this could be lethal and the genes will be eliminated. If the function of particular, properly folded, proteins is not essential then cells can survive short periods of stress.

This has nothing to do with "protein-based inheritance" and the existence of prions and chaperones does not challenge the old-fashioned 1930s view of evolutionary theory. It also doesn't challenge the updated view of evolutionary theory that arose after 1968. Anyone who thinks that prions and chaperones have a serious impact on our understanding of evolutionary theory is really grasping at straws.

If, at the same time, they fail to mention molecular clocks, the evidence of junk from genome sequences, and the massive amount of data supporting Neutral Theory as an explanation for variation, then they have not only grasped the wrong straws but failed to notice the haystack.
Similarly, transient errors in mRNA synthesis can also cause heritable non–DNAbased phenotypic change. This is observed when low-abundance transcriptional regulators are affected by transcription errors. This disruption can cause a cell to alter its gene expression, resulting in a phenotype that may be heritable (2).
There are times when errors in transcription or translation can cause problems. These are quite rare since most of the time there are dozens of copies of functional RNAs and the loss of one doesn't make a difference. Similarly, the loss of a single protein is very unlikely to make much of a difference except under highly unusual circumstances. It's difficult to imagine scenarios where errors in transcription and translation could routinely affect the evolution of a population.
Even the assumption that mutations are random, constant, and gradual has been revised on the basis of molecular mechanisms of mutagenesis. For example, in bacteria, responses to environmental stress can activate mutagenesis mechanisms that increase mutation rate, which can potentially increase the ability of a cell to evolve, specifically when it is poorly adapted to its environment (when stressed). Most of a 93-gene network that promotes mutagenesis in Escherichia coli is devoted to sensing stress and activating stress responses that direct the bacterium to mutate when stressed (3). Stress responses also up-regulate mutagenesis in yeast (4) and human cancer cells (5) and underlie mutations induced by antibiotics that cause resistance to those very drugs, and others (6).
It's true that transient changes in mutation rate caused by mutations in repair and replication genes are commonly observed in evolving populations that are under extreme selection pressure. (For example, in Lenski's long-term evolution experiment.)

It's also true that the vast majority of observed changes during evolution are consistent with the idea that mutations rates are roughly constant over millions of years. I don't think that the 1930s version of the Modern Synthesis necessarily assumed that "mutations are random, constant, and gradual" but even if it did, that's still pretty much what we observe today. (Where "random" means "non-directed.")

It's not obvious to me that we need to revise our 21st century understanding of evolution in order to account for some increases in mutation rates from time to time.
Mutations are also nonrandom in genomic space—for example, forming hot spots at DNA double-strand breaks, as demonstrated in bacteria (7) and suggested by local clusters of mutations in cancer genomes (8, 9).
We've known for at least 45 years about hot spots and we've known for almost as long that certain DNA sequences are more susceptible to mutation than others. The classic example is adjacent pyrimidines, especially TT. There's only one reason why this would cause a fuss and that's if you believe that the probability of mutation is exactly the same for every single nucleotide in a genome. If that's what you believe then you've probably got lots of serious issues about understanding evolution and you should go back and read the literature on mutagenesis from the 1970s.
The long-standing assumption of random, constant, and gradual mutagenesis is refuted by observations that mutations occur more frequently when cells are maladapted to their environments, together with the discoveries of mechanisms by which mutations are targeted to specific genomic structures. These modifications of the modern synthesis assumptions could not have been predicted or found without exploration of molecular mechanisms.
As far as I know, it was never a requirement of the Modern Synthesis that all mutations are completely random and occur all the time at a constant rate. It WAS part of the Modern Synthesis to claim that evolution occurred gradually over long periods of time. That may not be strictly true but it's not because of mutation rates.

Rosenberg, S.M. and C. Queitsch (2014) Combating Evolution to Fight Disease. Science 343:1088-1089. [doi: 10.1126/science.1247472]


  1. Hi Larry,

    This is not the first instance when Susan Rosenberg has left me scratching my head in bewilderment.

    As mentioned here:

    Susan Rosenberg suggested that: “Cells actually decide to turn up their mutation rate when they are poorly adapted to the environment.”

    I don’t get it. I question whether higher mutation rates can themselves be considered an adaptive response to stress. Her logic to me appears most circular.

    Just the same – the constancy of molecular clocks as reliable evolutionary timepieces, would appear at face value, quite questionable especially after what you yourself posted about the Elephant Shark.

    Here is how I summarize it all for my students...

    Molecular clocks are not constant:

    1. Different proteins within one organism can evolve at different rates.
    2. Identical proteins in two different related species can evolve at considerably different rates (say between two closely related species of birds)
    3. Different categories of genome evolve at different rates (say rRNA vs AA sequences)
    4. Identical categories of genomes can evolve at different rates between lineages (The paper cited by Larry here )

    All of which makes molecular clock calibration problematic but not necessarily impossible.

    Please tell me that I am not misleading my students.

    1. You are seriously misleading your students.

    2. Hi again Larry

      I was worried that you would say that... please correct me - I am about to borach the topic next week in class.

      best and grateful regards

    3. I think that the points you make are true, as far as they go. However, there are some other considerations. A majority of the changes "read" when evaluating molecular clocks are not subject to natural selection (and thus widely varying rates of change). They're synonymous changes in codons, or changes that substitute amino acids with similar properties at parts of the protein where close is good enough. Or very often they're changes in non-functional DNA. The rate at which these changes happen may fluctuate, but probably not in a consistent way for very long periods of time. Even the very slow or very fast rates of change in some DNA sequences undergoing selection won't bias the molecular clock as strongly as it seems at first because, as you pointed out, different proteins even within one species will evolve at different rates.

      Therefore a molecular clock would be expected to work, though it would be sloppy. And the prediction of effective molecular clocks has been supported. Once calibrated with the fossil record at some known points, molecular clock analysis can roughly "predict" other known points. Therefore it can be applied with some confidence to cases in which we don't have other evidence about when branch evolutionary branch points occurred.

    4. Tom,

      I think a balanced view of the pros and cons of molecular clocks would aid your students' critical thinking. 'Teach the controversy', I say!

    5. Hi Allan

      We just cross-posted - with the identical Ayala reference.

      I guess great minds think alike... so to speak.

    6. Hi Barbara,

      I am at a loss to see how I was at all inconsistent given I can (after a little google-whacking) buttress each of these four points with posts made by quoting Larry himself.

      For example - Larry: "There are many problems with molecular clocks, including the fact that they tick at different rates for different proteins as shown below. This is because the sequences of some proteins, like cytochrome c, are highly constrained by natural selection (i.e. conserved). Other proteins, like fibrinopeptides, can tolerate many more changes."

      … which seems to me to beg the whole notion of “neutrality”. Clearly, I am yet again missing something.

      Larry - please do not take me wrong! I remain in your debt and am currently in the throes of rewriting my original naïve Abiotic Origins of Life Assignment I used to hand out to my students. (Big and grateful Thank you!!!... when I arrive in Toronto, it would be my honor to treat you to lunch!)

      I sincerely do not want to repeat such agonizing history by yet again back-peddling in class after obtaining a tardy clarification on a misconception I inadvertently delivered in class.

    7. Here are my thoughts on the molecular clock/

      Can some genomes evolve more slowly than others?
      The Modern Molecular Clock
      Estimating the Human Mutation Rate: Phylogenetic Method

      Here are a couple of good reviews.

      Bromham, L., and Penny, D. (2003). The modern molecular clock. Nature Reviews Genetics 4, 2 [doi: 10.1038/nrg1020]

      Kumar, S. (2005). Molecular clocks: four decades of evolution. Nature Reviews Genetics 6, 654 [doi: 10.1038/nrg1659]

      The bottom line is that there is an approximate molecular clock just as predicted from populations genetics. The fact that different proteins/genes evolve at different rates is irrelevant.

    8. The fact that different proteins/genes evolve at different rates is irrelevant.

      Hardly! People working with the clock are well aware of all the issues, and the things they need to do to correct for them. If they're happy, I'm happy. But teaching about methodological difficulties, the need to be on the lookout for artefacts etc, seems relevant for a class that needs a general intro (assuming the class is beyond the blotting-paper-and-peas stage!).

      Having a grasp of the cons as well as the pros helps, for example, with evaluating claims that the protein in the ribosome is older than the RNA.

    9. @Allan Miller,

      The molecular clock only applies to nearly neutral substitutions fixed by random genetic drift. The fact that some proteins had a greater percentage of conserved amino acids than others may have been a bit of a surprise back in the 1960s but not today.

      We know why histones evolve much more slowly than fibrinogens. It's not a "methodological difficulty," it's perfectly normal. Nobody in the 21st century expects every protein/gene to have the same number of nearly neutral sites.

    10. True, but nor does anybody in the 21st century expect near-neutral sites to have great penetration at depth, nor ignore the confounding effects on near-neutral sites - hitch-hiking, Ne effects on the efficacy of s, etc. Even neutral sites show some biases, molecular and organismal, that modify (not 'invalidate') the underlying assumption that DNA polymerase's error rate minus mismatch repair = a steady 'tick'.

      These too are perfectly normal, but also "methodological difficulties". I know that people are aware of and correct for these, but they are real considerations for others to be aware of.

      The clock is typically taken to mean more than just neutral substitution. Then, we're into differentials, and histones and fibrinogens are but extremes on a continuum.

    11. I want to endorse the statement that variation of evolutionary rate from molecule to molecule is not, has never been, a violation of the molecular clock. Also to add that approximate clockness can in principle also be achieved if there are episodes of natural selection that are randomly enough distributed.

      People like to Disprove The Molecular Clock a lot. There is no precise molecular clock, and rates of change vary between different organisms. So should we never make use of the assumption of a molecular clock? In fact, all work on coalescent genealogies within species uses the molecular clock, and most work on closely-related species does too. It is a useful and fruitful approximation.

      As the organisms get more and more different, one first finds work using models of "relaxed" molecular clocks in which rates of evolution change randomly in each lineage. And then, with even more difference, people simply stop using clocks at all and use branch lengths rather than times on their molecular trees. Anyone who uses the molecular clock on a tree that contains Bacteria, Archaea, and Eukarya is simply being silly. They are too different in their biology to come anywhere near changing at similar rates.

      So the molecular clock is dead, long live the molecular clock!

    12. I read Tom as wanting to discuss confounding factors (he'll have that Purim hangover to deal with, so may not surface for a while). There'd be little point in presenting a method to a class then saying "...but it doesn't work"!.

      Of course, this is one that gets mangled to death by Creationists. But screw 'em. Discussing limitations and ways round them is little different from emphasising the importance of the right isotope for the period when radiometric dating.

    13. Hi again everyone - I thank yu all for your patient indulgence at my dilettante naivete

      OK – I confess I have a bit of a fog in my cranium right now… but I would like to repose the question of the reliability of any putative molecular clock in the light of “punctuated equilibrium” as understood by EvoDevo.

      For example, the enhancer HACSN1 seems to tick along pretty reasonably when comparing Rhesus monkeys to different mammals or even to a chicken...

      ... but HACSN1's clock runs completely out of whack when comparing Humans to Chimpanzees - a difference of an amazing 13 nucleotides within an 81 nucleotide stretch, a far larger number of changes than would be expected had the mutations been the result of drift rather than selection.

      Once the enhancer mutations have been fixed in a population – I can easily imagine a further rush of changes in the genome becoming fixed at an accelerated rate to “catch up” with those earlier enhancer mutations (subject again to Natural Selection)

      That at least is how I wrap my head around the cogency of “punctuated equilibrium”.

      So the molecular clock can surge – and the number of fortuitous surges can greatly affect accuracy of the clock which very sensitive to Natural Selection. Or at least that is what I have been teaching my students.

      I welcome any suggestions for improvement or correction for the following worksheet I provide my students in a Grade 11 non –AP Bio class.

      I bid adieu… I will need to refocus on this question later on.

  2. Hi again Larry,

    I guess where my problem is in reconciling these two sentences of yours:

    “It's also true that the vast majority of observed changes during evolution are consistent with the idea that mutations rates are roughly constant over millions of years.”

    “As far as I know, it was never a requirement of the Modern Synthesis that all mutations are completely random and occur all the time at a constant rate.”

    I have always been a fan of Fransico J. Ayala’s work… and remain intrigued by his continuing criticisms of the molecular clock and its major premise of "neutrality".

    A contentious religious footnote for any interested – Ayala is a former Roman Catholic priest and remains very interested in all matters theological and sees no problem in reconciling science and religion… and no that does not at all mean he endorses ID.
    Interesting interview:

    1. We just cross-posted - with the identical Ayala reference.

      Hee hee! Yes, I respect Ayala. I think the clocks are useful, particularly when you have good estimates of the number of mitoses in a period, and a genuinely neutral marker. But such probes soon saturate or disappear in deep time, so you need selected sites, with the confounding variables that that entails.

    2. @ Allan

      Re: "...But such probes soon saturate or disappear in deep time, so you need selected sites, with the confounding variables that that entails."

      Maybe not if we are to take Kevin Peterson at his word regarding miRNAs whcih appear remarkably refractory to saturation over deep-time.

  3. Tonight is Purim and the Talmud clearly teaches that one is obliged to drink enough alcohol to the point of not knowing the difference between Haman and Mordechai.
    Who says religious rituals possess no value? Gotta run…


  4. "Mutations are also nonrandom in genomic space—for example, forming hot spots at DNA double-strand breaks, as demonstrated in bacteria (7) and suggested by local clusters of mutations in cancer genomes (8, 9)."

    Hot-spots in genomic space: Benzer Scientific American January 1962, as cited in the Srb, Owen & Edgar, General Genetics, textbook 1965.

  5. Give it up, the whole thing needs to be thrown out, and their is a huge need to start OVER....