In article <4l05om$hhh at News.Dal.Ca> "Andrew J. Roger" <aroger at ac.dal.ca> writes:
Maybe it was a case where the ancestral eukaryotic GAPDH could not
work well in concert with a eubacterial TPI (and the enzymes above
it in the pathway)- so for a while several enzymes existed and
the one which worked best eventually was retained while the
other was lost.
This seems unlikely since, as far as I know, TPI and GAPDH don't
interact in any way; a product of one is just a substrate for the other.
But, more importantly...
The problem is that there are several classes of eubacterial GAPDHs.
There is a huge bacterial assemblage which is quite distant to
the eukaryotes (in fact the new archaebacterial sequence from
Haloarcula vallismortis is a little more similar to the eukaryote
enzymes than this eubacterial group is). Then there are GAPDH
enzymes found in cyanobacteria (Anabaena and Synechocystis) and
enzymes found in gamma-proteobacteria (ranging from E.coli to
Haemophilus) which fall RIGHT in the eukaryote clade (they are
far more similar to euk enzymes than either archaebacterial ones
or the big eubacterial assemblage). Moreover, there are multiple
copies of GAPDH in E.coli and the cyanobacteria- the above mentioned
ones and ones which fall in the eubacterial assemblage.
So the explanation is likely either:
1) cyanobacteria and E.coli received these eukaryotic-like GAPDH
genes from lateral transfer from eukaryotes
2) eukaryotes received their GAPDH genes from bacteria (this
scenario invokes a minimum of 2 transfers if one pays attention
to the topology of the GAPDH tree)
There is an important other possibility which does not involve horizontal
transfer at all. What you're calling the 'eukaryotic group' and the
'bacterial group' may in fact be the results of a duplication which
occurred prior to the divergence of eukaryotes and eubacteria. Everything
you said about GAPDH is consistent with this scenario. The existence
in some organisms of enzymes from both clades supports the possibility
of an advantage to maintaining both (perhaps one is specialized in
some way for glycolysis, and the other for gluconeogenesis, for example).
One should always consider this possibility befor jumping to a conclusion
of horizontal transfer. Think of the havoc that could be wreaked by
someone naively considering trypsin sequences from some animals and
chymotrypsin sequences from others.