Gene-sequence similarity not good match for phyla of bacteria?

Robert Maas rem at shell.netmagic.net
Wed Dec 9 04:06:51 EST 1998


(I tried to post this 1998.Nov.22, but as of Dec.08 it hadn't yet
appeared in DejaNews, so now I've flushed the moderated newsgroup
sci.bio.evolution and am re-trying to the one non-moderated newsgroup:)

Regarding SCIENCE, Vol.282 P.282 (1998.Oct.09):

Comparing the maximum-likelihood tree for 16S rDNA against the
classification by phyla in Lynn Margulis, FIVE KINGDOMS, I notice the
tightest bunch around the three target strains (obviously belonging to
phylum Chemoautotrophic Bacteria because of their methane-consuming
metabolism, except for the fact they apparently need vitamins unlike
the rest of that phylum) includes genera Rhodopseudomonas (phylum
Anarobic Phototrophic Bacteria) and Beijerinckia (phylum
Nitrogen-fixing Aerobic Bacteria). Other 16S rDNA in the chart NOT as
closely bunched to these are mostly in those same three phyla. Thus 16S
rDNA seems to be mostly orthogonal to (not correlated with)
classification by phylum.

For bacteria that exchange genes so much that the different genes
within any given strain of bacteria really have NOT co-evolved for a
long time, maybe it's not appropriate to classify the bacterial strains
themselves, but rather to classify each kind of gene separately and
then indicate the strain by the mix-and-match (chinese-restaurant menu,
one from column A and one from column B) composition from the various
gene classifications.

The theory presented in the article, that Rhodopseudomonas and
Beijerinckia both were methane-using bacteria but lost the genes for
that ability and magically obtained genes to function as members of the
other two phyla they are in now. Were those genes always present, but
inactive, when at some point in time those became inactive while the
mmo (methane monooxygenase) genes became inactive, so they suddenly
switched to new phyla per change in chemical lifestyle? Wasn't it more
likely that genes were transferred from one strain of bacteria to
another, so that depending on which genes are analyzed the clustering
is different? We need to sequence the genome of at least one of the
three new strains, and the genomes of both Rhodopseudomonas and
Beijerinckia, and the genomes of some of the regular methane-using
bacteria (Methylosinus, Methylocystis, Methylococcus), identify the
various genes in each, match up corresponding genes and do a separate
clustering analysis for each gene separately. Then see what
correlation, indicating a long time of co-evolution, there is between
various pairs of genes. We may then decide whether co-evoution has been
very long for ALL the genes, which implies the sudden jumps of phylum
are due to a tiny mutation turning on one gene allowing another to turn
off non-lethally, or whether co-evolution is longstanding ONLY for
sub-units of the whole genome, where at least one sub-unit has been
transferred. Perhaps we'll discover a truly major amount of transfer,
causing a gross mix-and-match of the various genes.

P.S. I'm very lonely. I don't have any friends. If anybody who lives
near Mountain View, CA, likes my ideas that I post to the net, and
wants to consider becoming my friend, please let me know. Otherwise
I'll have to assume that nobody around here likes me enough to care
whether I live or die.



More information about the Cellbiol mailing list