Archaebacteria and the Three Domains

L.A. Moran lamoran at gpu.utcc.utoronto.ca
Thu Apr 14 14:02:29 EST 1994


Earlier I wrote,

     "(You may have done this in Hilario and Gogarten (1993) - I can't 
      find this paper; but I haven't given up yet.)"

and Peter Gogaten kindly replied,

     "If you let me know your mail address, I'll send you a reprint."

I was able to find your paper (Hilario and Gogarten, 1993). That's an amazing
issue of BioSystems! All of the papers are part of the "Swansong" of Lynn
Margulis - she's ending her term as editor. There are lots of papers about
Gaia and various versions of endosymbiosis. I had a little trouble under-
standing Margulis' (1993) paper. Did you read it? She seems to be saying 
that all bacteria on the planet are just one species that constantly
exchange genes between individuals. When various bacteria fuse, new species
are formed, including eukaryotes.

Your paper was quite a surprise. You readily admit that there are now two
eubacterial species whose ATPase genes cluster with the archaebacterial
sequences on the V-ATPase side of the tree. Furthermore, your own analysis
of the Methanosarcina (archaebacterium) sequence confirms that it falls 
within the eubacterial cluster on the F-ATPase side. To me this strongly
suggests that many bacteria contain both types of gene (or perhaps their
common ancestor did and one has been lost in individual species).

If this is true, and the archaebacteria do not form a separate monophyletic
group, then you will get the trees that you have published. Both the 
V-ATPases and the F-ATPases would form independent trees with eubacteria 
and archaebacteria interspersed.

However, you suggest that ancestral bacteria cannot have contained copies of
both genes because the V-ATPase and F-ATPase trees do not show the eubacteria
and archaebacteria as separate monophyletic clusters! You say,

     "If the bacterial A/V-type ATPases are remnants of an A/V-ATPase
      already present in the last common ancestor, and if the A/V-
      ATPase catalytic subunits are considered as markers for the
      organismal evolution, then the root would be given by the non-
      catalytic F-ATPase. Therefore, Fig. 2 would lead one to conclude
      that Bacteria evolved from a mesophilic Archaeum. Clearly, this
      conclusion runs counter to the traditional bacterial systematics
      (e.g., Woese, 1987; Woese et al., 1990), and, as discussed below,
      is also at odds with the conclusion obtained from the analysis
      of the archaeal F-ATPase."

This seems to be circular reasoning. Your original dendrogram of ATPase
sequences is always cited as one of the two papers that most obviously
support the Three Domain hypothesis. Now you show that the latest sequences
do not support this hypothesis. However, now that Three Domains has been
widely accepted ("traditional") it must be correct. Therefore, any 
sequences that do not fit into Three Domains must has arisen from horizontal
gene transfers. You postulate that your anomalies arise from such transfers.
This reasoning seems contrived to me.

Suppose that the trees that you now have were published in 1989. Do you
think that Woese would have been so eager to take them as evidence for
his concept? I don't think so. In my opinion your most recent paper
invalidates your original conclusions in Gogarten et al. (1989) and suggests
that the Three Domain hypothesis is in even more trouble than I originally
imagined when I began this discussion.

Do you believe that it is still proper to cite Gogarten et al. (1989) as
one of the main bits of evidence in support of Three Domains and the close
relationship of archaebacteria and eukaryotes?

For the benefit of anyone else who might be reading this thread I will try
to reproduce the trees below. I hope Peter Gogarten will correct me if
I misinterpret his data. (Is anyone else reading this?)

Here is the original tree in Gogarten et al. (1989). The sequences are from
the catalytic subunits of two types of ATPases. Note that the eubacterial
sequences (B) group together and are called F-ATPases. They are distantly
related to the archaebacterial sequence (A) and the eukaryotic genes (E) 
(called V-ATPases). This tree supports the Three Domain hypothesis and also
suggests that archaebacteria are more closely related to eukaryotes.


                               |-------- A Sulfolobus           V-ATPases
                 |-------------|       
                 |             |   |------ E carrot
                 |             |---|  
                 |                 |----- E Neurospora
                 |
                 |                  |------ B Rhodospirillum beta
                 |                  | 
                 |                  |
                 |------------------|  |----- B E. coli beta
                                    |  | 
                                    |--|                         F-ATPases
                                       |--B Anabaena beta


Here is the latest version of the tree in Hilario and Gogarten (1993). Note
that the V-ATPases now consist of representatives from all three "domains".
The F-ATPases have acquired an archaebacterial sequence. This tree suggests
that there are two paralogous genes, each of which generates a complete tree
consisting of representatives of all three groups. Hilario and Gogarten
suggest that this tree has arisen from gene transfers between the archae-
bacteria and Thermus and Enterococcus and between the eubacteria and
Methanosarcina (beta subunit).


                                       |-- A Methanosarcina 
                                     |-|
                                   |-| |-- A Halobacterium
                                 |-| |------ B Enterococcus
                                 | |
                               |-| |--- B Thermus
                               | |------ A Sulfolobus           V-ATPases
                 |-------------|       |--- E carrot
                 |             |   |---|
                 |             |---|   |-- E cow
                 |                 |----- E Neurospora
                 |
                 |                      |-- B Rhodpblastula beta
                 |                    |-|
                 |                  |-| |-- B Rhodospirillum beta
                 |                  | |----- B E. coli beta
                 |                |-|  |- B Anabaena beta
                 |              |-| |--|                         F-ATPases
                 |--------------| |    |-- A Methanosarcina beta
                                | |-- B PS3 beta
                                |
                                |---- B Thermotoga beta

Forterre et al. (1993) raise some interesting points. Their analysis of
DNA polymerase genes, topoisomerase I genes, and topoisomerase II genes
do not lend support to the Three Domain hypothesis. Of more interest is
their paralogous hypothesis that explains the evolution of the ATPases
(Figure 6 of their paper). They say,

     "It is most likely that the ATPase trees used to demonstrate
      the eubacterial rooting of the universal tree also contain
      a mixture of orthologous and paralogous proteins. Indeed,
      two new eubacterial ATPases discovered recently in Thermus
      thermophilus and Enterobacter hirae are more similar to the
      archaebacterial and eukaryotic ATPases (V-type like) that to
      the eubacterial Fo/F1 ATPases which have been taken as eubacterial
      representatives in the composite phylogenetic trees of Iwabe
      et al. (1989) and Gogarten and co-workers (1989). This suggests
      that Fo/F1 and V-type ATPases are paralogous and that the 
      eubacterial rooting of the ATPase tree does not reflect the
      actual rooting of the universal tree of life, but the result
      of a fortuitous combination of orthologous and paralogous proteins
      in the same tree (Figure 6)."

I realize that you have rebutted their suggestion by hypothesizing one
or more horizontal genes transfers to explain the data but to me their
schematic dendrogram in Figure 6 looks more convincing.

I also find the Forterre et al. (1993) criticism of the EF-Tu trees to
be thought-provoking. I hadn't really appreciated how weak the elongation
factor alignments were.


Laurence A. Moran





More information about the Mol-evol mailing list