Brian Foley writes,
"I'm not sure what the overall point is here: To decipher the
evolutionary tree of organisms, or the tree of Heat Shock proteins?"
The point is to discuss the HSP70 dendrogram. I presented two different
versions of the tree and tried to point out that one of them was in
conflict with the data.
"If you wish to decipher the tree of oprganisms, you should look at
many genes from different families, and not just one. Cammarano et
al in J. Mol. Evol. 34: 396-405 (1992) used EF-G and EF-2 sequences
and compared their results to results obtained using EF-1alpha/EF-Tu
sequences. They discuss why the two genes/proteins result in different
phylogenetic trees. Braithwaite and Ito in Nucleic Acids Research 21:
787-802 (1993) used the DNA polymerase proteins to build phylogenetic
trees. Just as classical taxonomists use a weighted average of many
features, and not just one feature like "wing shape" to generate
phylogenetic trees, molecular evolutionists must consider many
gene or protein sequences, and not just one or two.
I agree with the point that the dendrograms of ALL available genes need to
be taken into account when constructung species phylogenies. I look forward
to the time when supporters of the Three Domain Hypothesis follow your advice.
As you know, there are several workers who are content to publish extensive
phylogenies based on a single gene (ie. SSU rRNA).
"If it is a tree of heat shock proteins which you seek, beware
that some gene and protein sequences are mis-labelled and this might
result in circular logic. Many genes are clones and sequenced based on
sequence similarity (probing libraries, designing PCR primers, etc...)
and many others are sequenced at random and then assigned a name based
on sequence similarity (sequence a random piece of DNA, compare it to
the database, it is most similar to HSP70, so name it HSP70). For
example in the study of EF-1alpha/EF-Tu one could be mislead by the
fact that many EF-1alpha proteins are named as "the" EF-1alpha for that
species, when in fact there are a whole family of EF-1alpha proteins
expressed at different developmental stages. One may then end up
comparing the Drosphila embryonic EF-1alpha to the Xenopus adult liver
EF-1alpha without realizing it. One must be careful to analyze how much
is known about the uniquenss of a gene in a genome. If more than one
member of a gene family is present, then intra-genomic recombination
could occur which would allow for an evolutionary tree to be
Inter-genomic recombination between say an archaebacterium and a plant,
is much less likely to occur, but still possible."
The HSP70 database of sequences is fairly large and it is relatively easy to
distinguish orthologous and parologous genes. There certainly are examples
of genes that are mistakenly identified but these can be recognized. In the
case of bacterial dnaK genes the dendrogram is far more reliable than the
DNA polymerase trees and probably better than the EF-Tu/EF-2 trees. Any
arguments against the reliability of the HSP70 trees can also be applied to
all other trees, including those that support a different version of the
relationship between archaebacteria and other organisms. I wonder why these
objections never come up when discussing these other trees?
Laurence A. Moran (Larry)