DISCUSSION
We have cloned and sequenced the three histone H3 genes from Tetrahymena thermophila. Thus, T. thermophila is one of the few eukaryotes for which the sequence of its entire histone gene family
is known (see Table 1). The T. thermophila H3 genes encode the major replication-dependent H3 and the replication-independent (replacement) variant hv2, whose predicted amino acid sequence
is very different from those of animal and plant H3 replacement variants. Phylogenetic analyses of H3 protein sequences by the neighborjoining and parsimony methods suggest that the three types of
replacement variants arose independently at least two and probably three times in eukaryotic evolution. This uncertainty is due to a puzzling change in the topology of the NJ tree caused by the addition of the A.nidulans H3 sequence. We had constructed the H3 NJ tree several times with increasingly larger data sets and found the branch order highly resistant to the addition of new sequences and uniform in its placement of animal
H3.3 and fungal H3s on the same branch (data not shown). We do not know why the addition of the Aspergillus H3 changes the placement of H3.3, since the H3s from Aspergillus and Neurospora differ at only 4 positions and at 3 of those 4 positions the Aspergillus sequence is more like the other fungal H3s than Neurospora (data not shown). Given that animal H3.3 and fungal H3s were monophyletic for all parsimony analyses carried out
on the entire H3 data set and on several different subsets, and that NJ trees constructed from the same subsets all placed H3.3 and fungal H3s on the same branch, we feel it is reasonable to conclude that H3.3 evolved after the branching of plants and animals, and thus independently of plant H3.IJJ and Tetrahymena