As Lynn Margulis elegantly explained, some eukaryotic organelles — such as mitochondria and chloroplasts — are the product of an ancient endosymbiosis event. Free living prokaryotes were absorbed by primitive eukaryotes and, over many generations, become entangled in an obligate host-symbiont interaction. There are other examples of such interactions between eukaryotes and intracellular symbionts, such as those found within deep sea tubeworms, sponges, and plant roots.

The structures we currently call organelles were, at one point, merely endosymbionts. Where do we draw the line between endosymbiont and organelle? Ursula Theissen and Bill Martin lay out the ground rules for what we can call an endosymbiont and what we can call an organelle:

  • Formerly endosymbiotic organelles are double-membrane-bound intracellular structures.

  • All of the cytosolic proteins in an endosymbiont are encoded by the endosymbiont’s genome. Most organellar proteins are encoded by nuclear genes and translated by the host’s ribosomes.

  • Organelles must import most of their proteins using evolved protein importation apparatuses. Theissen and Martin argue that this is “rate-limiting step in the transition from endosymbionts to organelles”.

Merely being an obligate host-parasite interaction does not make an endosymbiont an organelle. Only when the endosymbiont abandons most of its genes in favor of those encoded by the nuclear genome does it gain the status of organelle. In order to do so, it must be able to import proteins across its double membrane.

Theissen and Martin’s article is inspired by four publications that describe an obligate endosymbiont as a plastid (1, 2, 3, 4). Two authors from the previous studies reply to Theissen and Martin, pointing out that the host in question regulates the number of symbionts within the cell as well as their replication and segregation. They argue that this justifies calling the endosymbiont a plastid. In the end, it’s all just semantics, right?


  1. #1 Jonathan Vos Post
    December 24, 2006

    “Rate-limiting step” is old-school. It is a glib generalization of old static analysis of Michaelis-Menten kinetics in a sequence of enzymatic reactions in a metabolism. In the old approach, it was assumed that one enzyme is the “rate limiting step” or “bottleneck.” Selective pressure was presumed to be towards increasing the Vmax or Km of that enzyme.

    My dissertation research (1975-1977) included a demolition of that view. Rather, the metabolism is dynamic, and responds to environmental perturbation in many frequencies. What matters is how the dynamics of the metabolism behave in Laplace Transform chemical phase space. What matters most for the alleged “bottleneck” enzyme is a lumped parameter combining Vmax, Km, and other parameters, which is the eigenvalue associated with the eigenfunction of waves passing through that phase space.

    I can send you a list of my publications establishing that, but it’s perhaps too esoteric for the blog venue.

    The distinction between Endosymbionts and Organelles is indeed interesting. But let us attack it in a contemporary way. We are no longer stuck with setting all derivatives to zero in a differential equation model, and calculating a steady-state ratio of product and substrate by algebra. We have computers, software, and a methodology of emergent behavior on the edge of chaos.

    Let’s not attack a real problem with ancient and clumsy approaches.

    – Professor Jonathan Vos Post

  2. #2 Nick Dellhall
    December 25, 2006

    Can someone explain the third bullet?

    Organelles must import most of their proteins using evolved protein importation apparatuses.

    What exactly are ‘evolved protein importation apparatuses?”

  3. #3 RPM
    December 25, 2006

    Nick, see here and here.

  4. #4 apalazzo
    December 27, 2006

    One problem – the centrosome could be the remnant of an endosymbiotic fusion. And it is definitely an organelle. Although most of the old ideas that Margulis had about centrosomes are obsolete, the discovery that centrsosomes may have their own nucleic acids and the fact that they undergo a duplication process (reminiscent of semi-conservative replication) argues for this idea.

    Don’t forget that double membranes are essential for chloroplast and mito function. Other endosymbionts may have contributed to our genome however they did not contribute a membrane of any functional importance.