Archaebacteria and the Three Domains

L.A. Moran lamoran at
Thu Apr 14 14:01:47 EST 1994

Paul Lepp writes,

     "There do seem to be several problems with the Gupta and Golding 
      paper (1993). I could be persuaded that the Archea are more closely
      related to the Eubacteria than to the Eukarya, as long as the Archea
      remain monophyletic. What I find most disturbing is the polyphyletic
      origins of Archea that Gupta and Golding propose. If this were the 
      case There would have to be a hell of a lot of horizontal transfer 
      to explain cell wall constituents. More importantly it would require 
      at least 3 instances of convergent evolution to produce the ether 
      linked lipids of the CM."

I agree that there are many problems with the Gupta and Golding paper but
the dendrogram is not one of them. You seem to be under the impression that
all archaebacteria have the same cell walls and that these differ from those
of the eubacteria. Not so. In some cases the archaebacteria cell walls
are similar to those of the eubacteria except that pseudomurein replaces
murien. In other cases the cell walls are composed almost entirely of 
protein. Still other archaebacteria have cell walls made of polysaccharide.
Looks like the archaebacteria are composed of separate lineages that have
diverged independently from other bacteria if one examines cell walls.

All archaebacteria contain distinct lipids with ether linked isoprenyl
chains (Koga et al. 1993). The eubacteria and eukaryotes have lipids
with ester linked fatty acids. Within the archaebacteria there is quite
a variety of different types of lipids and the three groups of archaebacteria
(methanogens, sulfur dependent thermophiles, extreme halophiles) can be
distinguished on the basis of their lipids. It looks like the presence of
specific types of lipids is one of the few remaining characters that unite 
the archaebacteria.

The question before us is whether it is more reasonable to postulate an
separate origin of similar lipids in three groups (ie. convergent evolution)
or half a dozen different gene transfers in order to come up with a phylogeny.
At this point I don't really care which phylogeny anyone prefers as long as
they retain an open mind. What I object to is the increasing tendency to
assume that Three Domains has been proven and, especially, the tendency
to treat skeptics as reactionaries,

Paul Lepp continues,

    "An analogous situtation of conflicting phylogenies occurred when
     examining the phylogeny of chloroplasts. Morden and Golden (1989)
     examined the phylogeny of chloroplasts using the psbA genes and 
     found them most closely related to prochlorophytes. This conclusion 
     was at odds with the 16S data (Turner, et al; 1989) and the psbA
     phylogeny generated by Kishino, et al (1990). It turns out Morden 
     and Golden's analysis included a seven amino acid domain that was
     missing in Prochlorothrix, much like the Gupta and Golding paper. 
     The original analysis treated each of missing seven a.a. as
     individual evolutionary events, resulting in chloroplasts being
     grouped with the Prochlorophytes. Reanalysis treating the missing
     domain as a single event or excluding it entirely produced a tree
     concurring with those of Turner and Kishino (Morden and Golden;
     1989).  So, the treatment of a seven a.a. domain, much smaller
     than signature in HSP70, resulted in a substantially different 
     phylogeny.  To their credit Gupta and Golding did exclude shared
     G+/Archea domain upon reanalysis and got the same phylogeny. But
     was excluding the sequence enough?  It is unclear from the paper
     if the deleted domain was treated as a single event or multiple
     events in the analysis. Perhaps they should try treating it as a
     single event."

I have constructed trees where I treated the deletion as a single event or
ignored it completely. It doesn't make any difference. Remember that these
are highly conserved genes and the aligned region stretches from N-terminus
to near the C-terminus (>600 aa). A simple visual inspection of the
aligned sequences is more than convincing (see Boorstein et al. 1994).
If you E-mail me an address I will send you my database of 140 aligned
HSP70's. (Note that the aligned regions of EF-Tu's and many other proteins
are much smaller than 600 aa and there are many unresolved segments. In
other words the HSP70 data is much, much better.)

Just for the record, in spite of what Gupta and Goldberg say, the HSP70's
are not related to MreB. Furthermore, there is no evidence for an internal
duplication in HSP70 genes. The missing region in archaebacteria and gram
positive bacteria is a deletion that occurred in those specific lineages
but Gupta and Golding claim that all other sequences have acquired an
insertion. There is no evidence for internal repeat polypeptides as Gupta
and Goldberg suggest in their Figure 3. The scheme for the evolution of
HSP70 genes proposed by these workers is not correct.

Laurence A. Moran (Larry)

Koga, Y., Nishihara, M., Morii, H., and Akagawa-Matsushita, M. (1993)
    Ether polar lipids of methanogenic bacteria: structures, comparative
    aspects, and biosynthesis. Microbiol. Rev. 57, 164-182.

More information about the Mol-evol mailing list