Earlier, I presented some evidence that the green algae, Volvox, appears to possess a homlog of beta-catenin. We saw that when this hypothetical protein was aligned with human beta catenin, there was a 27% sequence identity over the entire 525 amino acids of the Volvox protein with an E value of 1e-25. Also, the Conserved Domain Database representation predicts an almost identical domain arrangement of the two proteins (although beta catenin has an extra 100 amino acids on both ends).
Yet it is quite possible that the homology between this Volvox protein and beta catenins is merely a reflection of the fact that they share Armadillo (ARM) repeats. The ARM domain is about 40 amino acids in length and functions in many different proteins simply to mediate protein-protein interactions. Thus, the hypothetical Volvox protein could simply be a protein with multiple ARM repeats yet possess no beta catenin-like function. It would simply stick to some other proteins using a homologous domain.
Unfortunately, we won’t be able to resolve this issue until biochemical evidence is provided that a) establishes the hypothetical protein is truly expressed and b) shows some level of beta-catenin like function. However, we can use a phylogenetic approach to compare the Volvox sequence with both beta catenins (and alpha importins) and a random selection of ARM repeat proteins. If the Volvox protein has a beta catenin-like function, we would expect it to nest with the known beta catenins. If it is just an ARM repeat protein, we would not expect it to nest with beta catenins and instead find itself situated among the randomly selected ARM repeat proteins. So let’s give it a try.
First, I obtained the sequence of human beta catenin, sponge beta catenin, and Aardvark, a known beta catenin homolog found in Dictyostelium discoideum. I also used the sequence from human alpha importin and Dictyostelium alpha importin since, as I have argued (see the link above), there is evidence the beta catenins and alpha importins are homologs. Since alpha importins are highly conserved and quite ancient, they would serve as nice outgroup to help root the trees.
For the random selection of ARM repeat proteins, I went to this page.
If you scroll down the page, there is a little wheel which shows the distribution of ARM proteins among various groups of organisms. If you click on any group, a table pops up and if you click on FASTA, you’ll be provided with a (very) convenient list of ARM protein sequences ready to be aligned and analyzed.
So here is what I did. I chose both the Eukaryotes and Other Eukaryotes groups and selected proteins. I then used these sequences, along with the Volvox sequence, beta catenins, and alpha importins and aligned them with a program known as CLUSTALW (using the default settings). I then plugged this alignment into another program known as PHYLIP to make a tree. I made the trees using the default settings except that I excluded positions with gaps and corrected for multiple substitutions. Keep in mind that I am a novice at this, so don’t mistake my results for some type of rigorous scientific analysis. I’m just going to show you why I am encouraged.
So let’s look at the trees.
For this tree, I retrieved the first three listed proteins for Cryptophyta, Alveolata, stramenopiles, and Euglenozoa as the pool of ARM repeat proteins. Here is the tree:
For this tree, I retrieved the bottom three listed proteins for Cryptophyta, Alveolata, stramenopiles, and Euglenozoa as the pool of ARM repeat proteins. Here is the tree:
For this tree, I retrieved the first three listed proteins for Parabasalia (parabasalids), Heterolobosea, Fornicata, and Amoebozoa as the pool of ARM repeat proteins. Here is the tree:
For this tree, I retrieved the last three listed proteins for Parabasalia (parabasalids), Heterolobosea, Fornicata, and Amoebozoa as the pool of ARM repeat proteins. Here is the tree:
Summary: Note that in all four trees, the Volvox sequence nests with the known beta catenins and does not find itself as an outlier or spread among the other ARM repeat proteins. What’s more, all the trees support my earlier contention that the human beta catenin is more similar to the hypothetical Volvox protein than it is to Aardvark, a known beta catenin homolog, since Volvox consistently groups closer to the human and sponge beta cantenins. Thus, I see even more evidence that this Volvox protein is a homolog of beta catenin.