Lake, J. A. 2009. Evidence for an early prokaryotic endosymbiosis. Nature 460: 967-971.
Endosymbioses have dramatically altered eukaryotic life, but are thought to have negligibly affected prokaryotic evolution. Here, by analysing the flows of protein families, I present evidence that the double-membrane, Gram-negative prokaryotes were formed as the result of a symbiosis between an ancient actinobacterium and an ancient clostridium. The resulting taxon has been extraordinarily successful, and has profoundly altered the evolution of life by providing endosymbionts necessary for the emergence of eukaryotes and by generating Earth’s oxygen atmosphere. Their double-membrane architecture and the observed genome flows into them suggest a common evolutionary mechanism for their origin: an endosymbiosis between a clostridium and actinobacterium.
Point 1: If such an endosymbiosis had occurred, then both partners would have probably had cell walls (peptidoglycan layers) outside their membrane (the only single-membraned eubacteria lacking cell walls are Mollicutes, which are parasites of eukaryotes*). Why did the external partner lose its cell wall rather than the internal partner?
*I originally referred to Mollicutes as intracellular - thankfully, Elio Schaechter put me right (they're extracellular). I did say that my brain had stopped working.
Point 2: The endosymbiosis scenario requires that the genetic complement of the external partner be entirely lost or transferred to the inner partner (as well as the greater part of its cytoplasm). While unusual, this would not be unique. Transfer of genes from one partner to the other is a common (if not universal) event in endosymbioses. The hydrogenosomes found in some anaerobic eukaryotes are generally accepted to be derived from mitochondria, but most have entirely lost their genomes (as far as we know). Also, in most secondary or tertiary chloroplasts, the eukaryotic genome of the endosymbiont has disappeared. However:
Point 3: In all established cases of endosymbiosis, gene transfer or loss has been mostly (if not entirely) to the cost of the internal partner. This applies not just to mutualist endosymbionts, but also to intracellular parasites. Lake's proposed endosymbiosis requires the transfer to happen the other way, at the cost of the external partner (offhand, the same objection applies to scenarios that propose an endosymbiotic origin for the eukaryotic nucleus).
Point 4: Even if Lake's premise that double-membraned bacteria carry genes from two phylogenetically separate ancestors is correct (I have to confess that I don't feel knowledgeable enough to critique that point), that doesn't necessarily require an endosymbiosis. A simple symbiosis might possibly be sufficient. Many bacteria form closely-linked ectosymbiotic consortia, and the well-established propensity of bacteria to swap genetic material like trading cards could result in a substantial transfer in such an arrangement over time. Also:
Point 5: Among eukaryotes, the endosymbiosis theory receives something of a boost from the point that eukaryotes are absolutely lousy with endosymbionts at all stages of interdependence. Lake mentions the proteobacterium Buchnera in aphids; there are also zooxanthellae in corals and clams, Perkensiella in amoebae, a whole universe of intracellular parasites... the list goes on. Prokaryotes, in contrast, just don't seem to carry endosymbionts to the same degree. Lake mentions in his support that the aforementioned Buchnera carries its own endosymbiont; what he doesn't mention is that this is the only well-established case of an endosymbiont inside a prokaryote. Lake claims that the Chlorochromatium consortium is very close to an endosymbiotic relationship. It's still ectosymbiotic nonetheless. Contrast this with the extreme general diversity and versatility of prokaryotes, which is leagues ahead of that of eukaryotes - you'd think that if they could easily do it, they would be.
Point 6: The presence of a nucleus of sorts in Planctomycetaceae indicates that bacteria are not incapable of developing new membrane systems.
Point 7: I really, really hate this paragraph:
In fact, the membrane organization of double-membrane prokaryotes fundamentally differs from that found in single-membrane prokaryotes. In the former, the peptidoglycan layer is sandwiched between the outer and inner membranes, so that it surrounds the inner membrane: in contrast, in the latter there is no inner membrane, and the peptidoglycan layer, located outside the cell, surrounds the outer membrane. Also, double-membrane prokaryotes contain their flagellar motors in the inner membrane, whereas single-membrane prokaryotes contain their flagellar motors in the outer membrane. And the photosynthetic apparatus in double-membrane prokaryotes is in the inner membrane, rather than in the outer membranes as in single-membrane prokaryotes. In other words, the organization of the inner membrane of the double-membrane prokaryotes resembles that of the outer membranes of typical single-membrane prokaryotes. The inner membranes of double-membrane prokaryotes are organized almost as if they were derived from the outer membrane of an engulfed single-membrane prokaryote.
This, ladies and gentlemen, is a classic case of semantic silly buggers. Lake obfuscates the difference between single- and double-membrane bacteria by noting that double-membrane bacteria have an outer and inner membrane, then referring to the membrane in single-membrane bacteria as the outer membrane even though it is positionally comparable to the inner membrane (pause to wipe foam from frothing mouth and allow bulging eyeballs to return to their sockets). His referral to the supposed difference between flagella of the two types of bacteria looks a little different when you consider that what he is saying is that both anchor their flagella on the membrane inside the cell wall. (noooo! the family curse!)
There is currently much discussion of the prokaryotic ‘tree of life’, but there are few points of agreement regarding its topology, except that it is not a tree.
While the idea that a tree is not an appropriate expression of prokaryote evolution is increasingly popular, I think it's jumping the gun a little to present it as a consensus. Take a look at the number of trees in an average issue of the International Journal of Systematic and Evolutionary Microbiology, for a start (and yes, now I am just picking at minor details).
Point 9: And just on a final quibble, this line from the supplementary info:
I propose the name Domain Synergia (Gr. Synergia – joint work) for those prokaryotes that possess the Gram negative, double membrane organization, and are derived from large, statistically significant gene flows from both the Actinobacteria (as defined in Table S1) and the Clostridia (also as defined in Table S1).
Not only would I consider it fairly unacceptable to bury the publication of a new taxon name within the online-only supplementary info of a print-based article, but under Lake's proposed scenario this "new" taxon would be circumscriptionally equivalent to the already-available name of Didermata.