A Prediction Tested

Several themes have been doing the rounds lately. The origin of organelles, standards of evidence, common descent, the role of phylogenetic analysis, and the meaning of ‘prediction’ in science. Here’s a case study/rambling discourse that links a few themes.

Researching an answer to a separate question (I do try), I was struck by a thought about RecA. RecA (also going by the names RAD51, Dmc1 and RADA in different groups, for historical reasons) is a ubiquitous group of proteins involved in homologous DNA repair. That’s a process whereby a break in DNA can be ‘patched’ if a homologous sequence can be located. Matching sequence either side of the gap is aligned (by nothing more sophisticated than the binding energy of DNA complementarity) and then a DNA polymerase template-copies from the intact strand to the broken one between the two complementary sequences. Both accidental and deliberate breaks are repaired by this, hence it is involved both in maintaining DNA integrity and in the more ‘orchestrated’ process of crossover formation in meiosis.

Because this process relies on quite a high degree of complementarity, it works best on sister chromosomes – those recently replicated, within the current cell cycle, and hence clearly commonly descended. This is all a prokaryote has to work with, outside of instances of LGT. In eukaryotic diploids, the donor for repair can be the homologous diploid chromosome (that’s a terminological confusion: the chromosome pair with the greatest amount of homology is actually not the homologous pair, but the sister pair). But even in diploids, the sister is ‘preferred’ for repair – when not available, the normal repair pathway is ‘nonhomologous end joining’, which simply splices the break. An exception to this is during crossover of meiosis. Most crossovers form between homologues, not sisters.

So, thinks I, if chloroplasts and mitochondria evolved from bacteria, their RecA equivalents should be more like those of bacteria than archaea. To the internet!

It so happens that all these proteins are, in the modern eukaryote, held in nuclear DNA. So in a plant, you’ve got your RAD51s, plus Dmc1 specific to meiosis, but you’ve also got RecA proteins targeting, respectively, mitochondria and chloroplasts. RECA1 heads for chloroplasts, RECA3 for mitochondria. There’s also RECA2 which goes to both.

More specific and comprehensive phylogenetic analysis reveals quite a complex picture. Nonetheless, the Lin paper notes a ‘striking’ sequence similarity between the recA genes of plants and protists and those of the bacteria from which they are presumed to have come. There is a healthy 61% sequence match between RECA1 and the RecA of cyanobacteria. Sequence identity for RECA3/bacteria is not so high, but interestingly, Arabidopsis RECA3 can complement E. coli deficient in bacterial recA. E. coli are Gammaproteobacteria, not Alphaproteobacteria as is thought to be the group from which mitochondria came, but still not a million miles away. That’s not conclusive of a common origin, but is a noteworthy fact, consistent with structural conservation.

And that, really, is where I was headed. I started from the hypothesis that mitochondria and chloroplasts originated as bacteria. A prediction of that hypothesis is that organelle-targeted proteins would be expected in general to align more closely with bacterial than with archaeal or non-organellar eukaryote proteins. That prediction has been borne out. The hypothesis has been strengthened by that observation. How can that be? I did the same a few days ago with N-formyl methionine as translation initiator. Maybe I’m cherry-picking, but there are no searches I’m not mentioning that drew a blank. This is the sum total of my ‘research’: two things that it occurred to me to look for, and I found them both.

Also of interest to me, given the conviction explored in my ‘Evolution of Sex’ paper that meiosis is foundational to the modern eukaryote clade, is the finding that Dmc1 apparently evolved very early during what it pleases me to call ‘eukaryogenesis’. Whether it preceded or succeeded the mitochondrial endosymbiosis is not clear, which is one reason I don’t think of endosymbiosis as definitively the origin of the eukaryotic cell, whatever lols may accompany someone finding an author who does just that (Hi, Mung!).

Another point to ponder: homologous recombination relies upon a physical analogue of the algorithmic alignment performed during sequence comparison. It is only by anchoring matching sequences that ‘differences’ – in repair, the missing vs the intact sequence – can be located. Molecular differences between taxa are trumpeted by Creationists, but they are located in much the same way. The question remains: where does the alignment come from? In the case of sister chromosomes, it is non-controversially common descent – the sisters arise in the same cell cycle. In the case of diploid homologues, again not too controversial – the bases of the haploid chromosomes in gametes can reasonably be assumed to have a common origin in template-copies originating in an ancestral cell. But somehow, for the Creationist, this logic breaks down somewhere not clearly specified outside of the species. Alignment suddenly stops being common descent and becomes the completely indistinguishable ‘common design’. I don’t see why.

206 thoughts on “A Prediction Tested

  1. Therefore, the similarity of their proteins could be partly due to convergent evolution.

    – An Alternative Hypothesis for the Origin of Mitochondria

  2. Mung,

    (quote) Therefore, the similarity of their proteins could be partly due to convergent evolution.

    Nah. There are no good examples of convergence at the molecular level covering more than ‘a few’ positions. You don’t get to 61% through convergence. Not without cumulative selection anyway [pause for hearty belly-laugh].

    At the very least, it would be a very surprising result, since gradual erosion of similarity is a far more likely cause than its gradual acquisition.

  3. Allan Miller: At the very least, it would be a very surprising result, since gradual erosion of similarity is a far more likely cause than its gradual acquisition.

    Indeed! In the absence of selective bias, silent mutations will creep in at a rate regular enough to date bifurcation from a shared ancestor.

  4. I am surprised to see how extensive the literature on organelle-to-nuclear genome DNA transfers actually is. The phenomenon is so frequent it has been detected in real time for both plastids and mitochondria.

    And there have even been experiments done on it where researchers inserted plastid-defective-expression-genes into plastids, and could later detect them through their functional expression when they had been transferred the nucleus.

  5. An interesting extra layer of complexity. It appears, on phylogenetic evidence (that thing we deny for the Hell of it), there are Chlamydia genes in plants. Many of them are targeted to plastids.

    In animals and protists, Chlamydia is an intracellular parasite – an endosymbiont, indeed, much like Wolbachia, and likewise definitely a bacterium. Chlamydia does not infect plants, so one possibility is that it was an early endosymbiont that has been lost from the plant lineage, beyond the Cheshire-cat smile of a few residual transferred genes. So we have an interesting tie-up between definitive and putative bacterial endosymbionts.

Leave a Reply