Constructing a phylogenetic tree for classical MHC genes from different organisms

Question

I am interested in constructing a phylogenetic tree containing the classical HLA α1 and α2 domains and potential classical MHC α1 and α2 domains of a different organism. I am using the amino acid sequences for the tree.

However, I am unsure of what bioinformatic tools and software to use and what parameters to fill in.

Lastly any advice on how to read a phylogenetic tree and how to relate the sequence similarity between two sequences by looking at the tree would be very helpful.

My aim is to investigate the classical MHC genes of a non human organism.

Any help and advice would be greatly appreciated!

Answer 1

This is a very unusual situation as a main effort is finding the outgroup where these two domains aren't duplicated.

General situation

In animals a gene tree nearly always follows a species tree, with the exception of gene duplication events, which frequently stem from genome duplication events. Making a tree with lots of genes gets hard to follow and actually becomes pointlessly problematic when you get a branch that does not quite follow the species phylogram —say, you have rabbit, mouse and human genes and the inferred tree, due to poor signal, comes out with mouse+human together as opposed to rabbit+mouse. As a result a nice tree needs to present the species before the genome duplication, often sea lamprey —I wrote a blog post recently were I discuss nice species to pick for a multiple sequence alignment.

In this simple case, one can get the protein sequence of interest and blastp it against human, sea lamprey and fruit fly, see how divergent are the paralogues in human and if the split is more recent that the latter two, if so, the latter will be an outgroup and after getting a few representative species in between, one can make a MSA with Muscle, which contains also a tool to make a neighbour joining tree.

Domain duplication

In your case you have something more complicated in that you have a series of paralogues, each gene encoding a protein (e.g. HLA-A) containing two globular domains, which form the MHC-1 fold, and you want make a tree of this... which means you want find out when the two domains duplicated (i.e. outgroup)? If the domains work together (correct?) and have the same fold (as you alluded to, correct?), this would mean that before the duplication event of the domains, the ancestral protein may have worked as a homodimer.

I am not familiar with the literature, so do not know if this has already been looked at —but I suspect it has.

If not, clicking on the sequence of one of the two domains in the Uniprot entry of one of the paralogues and doing a few iterations of psi-blast against human and a few other distant species (such as zebra fish, sea urchins, which are basal to vertebrates, and fruit fly, which I do not believe have a MHC etc.) most of these will have two hits per protein... but one might not. This is the outgroup!