Jellyfish lack true Hox genes!

I'm going to briefly summarize an interesting new article on cnidarian Hox genes…unfortunately, it requires a bit of background to put it in context, so bear with me for a moment.

First you need to understand what Hox genes are. They are transcription factors that use a particular DNA binding motif (called a homeobox), and they are found in clusters and expressed colinearly. What that means is that you find the Hox genes that are essential for specifying positional information along the length of the body in a group on a chromosome, and they are organized in order on the chromosome in the same order that they are turned on from front to back along the body axis. Hox genes are not the only genes that are important in this process, of course; animals also use another class of regulatory genes, the Wnt genes, to regulate development, for instance.

A gene can only be called a Hox gene sensu stricto if it has a homeobox sequence, is homologous to other known Hox genes, and is organized in a colinear cluster. If such a gene is not in a cluster, it is demoted and called simply a Hox-like gene.

Hox genes originated early in animal evolution. Genes containing a homeobox are older still, and are found in plants and animals, but the particular genes of the Hox system are unique to multicellular animals, and that key organization arrangement of the set of Hox genes in a cluster is more unique still. The question is exactly when the clusters arose, shortly after or sometime before the diversification of animals.

If you take a look at animal phylogeny, an important group are the diploblastic phyla, the cnidarians and ctenophores. They branched off early from the metazoan lineage, and they possess some sophisticated patterns of differentiation along the body axis. We know they have homeobox containing genes that are related to the ones used in patterning the bodies of us vertebrates, but are they organized in the same way? Did the cnidaria have Hox clusters, suggesting that the clustered Hox genes were a very early event in evolution, or do they lack them and therefore evolved an independent set of mechanisms for specifying positional information along the body axis?

Once you understand all that, the premise of this paper by Kamm et al. is straightforward, as is their conclusion. They are examining the arrangement of Hox genes in the genomes of two cnidarian species, Nematostella and Eleutheria. They've identified homeobox-containing genes homologous to those in triploblasts, and are asking if their chromosomal organization is also homologous. If it is, that would suggest that the Hox clusters evolved before the cnidarian split, and that their morphological complexity is generated in a way similar to ours. If it isn't, that means the Hox clusters evolved after the split in the triploblast lineages, and most interestingly, that the cnidaria are using their Hox-like genes in novel ways to generate their forms.

Now here's the answer, in visual form:

The Hox-like Genes of Nematostella and Eleutheria Are Organized Differently and Do Not Reflect the Clustered Patterns Characteristic
of True Hox Genes
The figure summarizes schematically the genomic organization of Hox-like genes in Nematostella vectensis and Eleutheria dichotoma. In the
case of Nematostella, contigs were assembled from GenBank whole-genome shotgun trace files, whereas Eleutheria contigs represent genomic
fosmid clones. Rectangles show the position of the genes in the genomic context but only approximately represent the sizes of the genes. Details
of intron-exon structure have been omitted except in the cases of Cnox-1ed and Cnox-3ed, both of which contain large introns. Arrows show
transcriptional orientation of the genes. Hox-like genes are in dark blue with probable orthologs between Nematostella and Eleutheria framed
with the same color. Hox-related genes are in light blue and non-hox genes are in red and numbered: 1, retrotransposon; 2, Rho/Rsa-related
gene; 3, putative metalloproteinase inhibitor; 4, putative ANF-receptor; 5, Phosphatidylinositol phosphate kinase; 6, putative lam_G domain;
7, 8, 10-12, fragmented reverse transcriptases; 9, homolog to Danio rerio putative protein; 13, reverse transcriptase; 14, resembles metabotropic
glutamate receptor; 15, two or more ORFs resembling Homo sapiens put. protein; 16, last exon of RRN3; 17, Dfp domain gene; 18, POU gene; 19,
1.2 Kb ORF resembling Plasmodium MAEBL (interrupted by vector).

That caption is full of jargon, but the title, at least, is clear. No clustered Hox genes. The Hox-like genes are there, but they're scattered all over, with miscellaneous other genes interleaved between them, and they also show sequence differences that set them apart from the canonical Hox genes. The arrangement isn't even the same between Nematostella and Eleutheria!

The differences go deeper. Colinearity means that a Hox-like gene that is homologous to one of the anterior Hox genes ought to have anterior function, and be expressed consistently at the anterior end of the animal. This is also not true. Expression patterns are complex and shift over the course of development, as we can see in this series illustrating the location of Cnox (an anterior Hox-like gene) expression at different stages.

(A-F) Cnox-5ed in 3-day-old planula larvae (A, B), 4- to 5-day-old planula larvae (signal moves toward the center as an ectodermal ring; [C, D]),
and in young primary polyps (oral and aboral; [E, F]).
(G and H) Aboral expression of Cnox-4ed in a 10-day-old primary polyp. Scale bar equals 50 µm.

Comparing different species also shows differences in where Cnox is turned on. It's highly variable, unlike what we see in triploblasts.

Expression Patterns of Related Hox-like Genes Are Heterogenous among Cnidarians and Do Not Conform to Colinearity Rules
The "anterior Hox-like" genes (= Cnox-5ed in Eleutheria, anthox6 in Nematostella, and Cnox-1pc in Podocoryne) provide a striking example of
heterogeneity in expression patterns. In planulae, these orthologs are expressed ectodermally and aborally (Eleutheria), entodermally
and orally (Nematostella), or aborally in both ecto- and endoderm (Podocoryne). In polyps, the corresponding genes are not
expressed in Podocoryne, but are expressed both orally and aborally in Eleutheria and at the oral end only in Nematostella.

I think the answer is clear. The last common ancestor of humans and sea anemones had Hox-like genes, but they weren't organized in clusters. Our ancestors evolved a patterning mechanism that used linked Hox genes in a colinear cluster, while the Cnidaria went in a different direction, evolving complex forms by different rules that used the Hox-like genes in a more arbitrary way.

Whereas the consensus view has been that a Hox clus-
ter was present in the ancestral cnidarian, our
analyses of sequence relationships, gene organization,
and expression data indicate that definitive Hox clusters
are not present in cnidarians and are therefore a synapomorphy of the Bilateria. The situation in cnidarians is
therefore very different to that even in very derived members of the Bilateria. For example, whereas in urochordates the ancestral Hox cluster has fragmented, the individual genes show high levels of sequence identity
and similar (A/P-restricted) patterns of expression to
their orthologs in other bilaterians. In cnidarians,
not only are the genes dispersed, but also there are no
clear relationships in terms of expression patterns or sequence identity. Cnidarians have genes related to anterior and posterior Hox/Cdx genes, but most of the Hox-like genes present are likely to postdate divergence with
the bilaterian line, accounting for their unclear relationships to true Hox classes.

The Cnidaria are an eminently successful group of animals, with a large number of diverse species and remarkable morphological diversity. Obviously, Hox clusters are not the only way to build a complex animal, and the Hox system is a relatively late addition to the metazoan toolbox. One new question to ponder is the nature of the primordial positional information system in the Metazoa—was it perhaps the Wnt genes? What other innovations will we find in the diploblasts?

Kamm K, Shierwater B, Jakob W, Dellaporta SL, Miller DJ (2006) Axial patterning and diversification in the Cnidaria predate the Hox system. Curr Biol 16:1-7.