All ten arthropod Hox genes are present in the spider Cupiennius salei
The combined data from other arthropods [e.g. [8,19,20]], summarized in Hughes and Kaufman [12], imply that the Hox complex of the last common ancestor of all arthropods contained ten Hox genes. The present data of the Cs-pb and Cs-Scr genes combined with our previous work [15,17-19] show that at least one copy of each of the ten arthropod Hox genes is present in the spider Cupiennius (Fig. 6). At least three of the Hox genes are even present as two copies (see below). The expression data on Cs-pb and Cs-Scr make Cupiennius the first chelicerate for which expression data are known for all ten different arthropod Hox genes; previously the chelicerate data were an assemblage from three different species [see also Ref. [12]].
At least three duplicated Hox genes in the spider Cupiennius salei
Our data show that at least three Hox genes are present as two copies in Cupiennius [combined data from this paper and Ref. [15]]. There are previous reports on duplication of Hox genes in chelicerates. Cartwright et al. [13] could identify one to four representatives per Hox gene class in the horseshoe crab Limulus polyphemus. However, there is no expression data for the Limulus Hox genes. In the spiders Achaearanea [14] and Cupiennius [15], previous one duplicated Hox gene each had been described. For mite and pycnogonids no duplicated Hox genes have been described [e.g. [20,21]].
In all three cases in Cupiennius (Dfd, Scr, and Ubx), the two paralogs are expressed in comparable but not identical domains. They are expressed in the same segments with differences in the intrasegmental patterns. This shows a striking similarity with what is seen for the duplicated Hox genes of the various paralogous groups in vertebrates that are also expressed in similar but not completely identical expression domains [24].
Gene duplications offer several possible outcomes [25]. One option is that one of the copies gets silenced or lost again during evolution. A second option is that one copy retains the ancestral function, freeing the other copy to diverge and evolve new functions (neofunctionalization). A third possibility is that each of the two copies performs a different subset of the ancestral functions (subfunctionalization). The differences in the intrasegmental expression patterns of the two copies in Cupiennius suggest that each of the two copies performs different functions that must be the result of either a neofunctionalization or a subfuntionalization event. As there is no genomic information available yet, it remains unclear whether there are additional duplicated Hox genes in Cupiennius (see also next section).
An important question is why duplicated Hox genes are present in the spider and why they are retained? Are they remnants of a large duplication event that are maintained because of neofunctionalization or subfunctionalization events? Or is there another evolutionary advantage for the spider to have multiple copies of some Hox genes? Presently it is difficult to answer these questions. In chelicerates there seems to be a tendency towards having more Hox genes [this paper, [13-15]], this in contrast to insects where there is a reduction of true Hox genes as two Hox genes -Hox3 and ftz- lost their homeotic function and obtained new functions in the insect embryo, which is associated with a divergence of the sequence of the gene [26-28].
Hox gene duplications have been proposed to be one of the genetic mechanisms behind the diversification of vertebrates [e.g. [29]]. However it remains difficult to draw a direct link between Hox gene duplications and morphological evolution. Recent results from Lynch et al [30] suggest an important role for the action of positive Darwinian selection in the divergence of vertebrate Hox genes after cluster duplications. The locations in the homeodomain of the sites that are under positive selection suggest that they are involved in protein-protein interactions. This suggests that adaptive evolution actively contributed to Hox gene function [30]. Indeed, in the Cs-Ubx-2 homeodomain there are two amino acid exchanges compared to the homeodomain of Cs-Ubx-1 or of Ubx of most other arthropods (Fig 1). Only in the honey-bee and the crustaceans Moina and Artemia there is one amino acid exchange in the homeodomain, in all three cases an A to S exchange at position 37 of the homeodomain (not shown). Also one of the two exchanges in Cs-Ubx-2 is an A to S on position 37. The sequence divergence in the homeodomain of Cs-Ubx-2 thus might be associated with a functional divergence. However the mechanism of the divergence is unknown, leaving open the role of Hox gene duplication in morphological evolution of chelicerates.
Duplicated genes in the spider: a whole genome duplication?
The most important question that comes up now is on the origin of these three duplicated Hox genes in Cupiennius. There are two options. First, they result from a duplication of the complete cluster. This would imply that either additional Hox genes are present as two copies that have not be found so far, or that one copy has been lost for the other Hox genes, as has happened to some of the Hox genes in the duplicated vertebrate clusters. Mammals for instance possess four Hox clusters, but most of the paralogs are not present as four copies as some of them have been lost in some of the clusters [11]. All data for Cupiennius Hox genes were obtained via either PCR approaches or cDNA library screening [15]. As there is no genome project for the spider yet, this means that it is presently unclear whether additional Hox genes are present as duplicated copies in the spider. The second possible explanation for the three duplicated Cupiennius Hox genes could be three independent tandem duplications of the individual genes. Additional analyses are required to identify the genomic organization of the spider Hox genes, and to find out whether these genes are indeed organized in two clusters, or whether the duplicated genes are serial duplications within a single Hox cluster.
However, there is some additional data that point to large-scale duplication of chromosomal fragments or even complete genomes in the spider. So far we also have found in our PCR screens several other genes that are present in two or more copies in the transcriptome of the spider Cupiennius, like extradenticle, homothorax, H15, Wnt5, Wnt7, engrailed, Delta, Suppressor of hairless, Krüppel, runt, pairberry, optomotor blind odd-skipped, apterous, orthodenticle [[19,31-37], our unpublished data]. In contrast, in most other arthropods most of these genes are present as one copy only. The relative high number of duplicated genes may point to a major duplication event in lineage to the spider, which might be caused by a whole genome duplication. A spider genome project would help to verify this.
Two posterior expression boundaries of spider Hox genes
Now data from all ten different arthropod Hox gene classes are known from this spider, another fact becomes obvious, that we already recognized previously based on a smaller data set [38], but which becomes even more prominent by new data on Cs-pb, Cs-Scr, and Cs-ftz [this paper, [19]]. There are two discrete posterior expression boundaries for Hox genes in the Cupiennius (Fig 6). The expression of all anterior Hox genes (lab, pb, Hox3, Dfd, Scr, ftz) ends at the boundary between fourth walking leg (L4) and first opisthosomal segment (O1), which is at the tagma boundary between prosoma and opisthosoma. Also the posterior Hox genes (Antp, Ubx, abdA, AbdB) all have the same posterior expression border: the very posterior end of the embryo. There is only one Hox gene, Cs-Antp, that crosses the tagma boundary (Fig 6). In other arthropods, but also in vertebrates, most of these posterior expression borders are not defined as well as in the spider [12,39].
The reason for the two discrete posterior expression borders remains unclear and we only can speculate on this. Between L4 and O1 is an important morphological boundary, the one between the two tagmata of the spider: the prosoma and the opisthosoma. The Hox genes might play a role in the specification of this boundary. In contrast, several Hox genes cross tagmata borders in other arthropods [e.g. [12,23]]. If the Hox genes play a role in tagma border specification, then this must be a peculiarity of the spider.
Another explanation could be that the anterior Hox genes are required for the specification of the different appendages in the spider. All six anterior Hox genes are expressed in distinct patterns within the appendages suggesting a role of them in appendage specification [[15,17,19], this paper] (see also Fig 3 and 4). It has been shown that Hox gene expression is associated with morphological diversification of leg segments in insects [40]. Indications for interactions between Hox genes in the spider legs come also from the weaker expression of Cs-Dfd-2 in L3 and L4 that coincides with the stronger expression of Cs-Scr-1, Cs-Scr-2 and Cs-ftz in L3 and L4 (Fig. 3F, Fig. 4C, Fig. 4F and reference [19]. Thus there might be a cross regulation between these Hox genes in the legs. Such a role of these Hox genes in the legs may form the reason for shared posterior expression boundaries (Fig. 6). The border between segments with and without appendages coincides with the tagma boundary between prosoma and opisthosoma. Spiders have true appendages on six segments: the cheliceres, the pedipalps, and four pair of walking legs. The more posterior segments do not have true appendages, however the second to fifth opisthosomal segment develop limb buds that give rise to the respiratory organs and the spinnerets [41].
A third possible explanation could be that the discrete Hox gene expression boundary is a result of the segmentation process that acts more upstream in the regulatory cascade and that lays down the segments. It is known from insects that the segmentation gene cascade indeed also controls the expression of Hox genes [42-44]. In insects this is mainly done by orthologs of gap genes. It is not known yet what genes regulate the expression of the spider Hox genes. A number of spider Hox genes obey parasegmental boundaries, as they do in Drosophila [31]. Parasegmental boundaries are important developmental boundaries in the early embryo and are specified by the segmentation gene cascade [45]. Segment-polarity genes like wingless, cubitus interruptus and engrailed maintain the parasegmental boundaries in arthropods. We assume that at least in part the same upstream acting regulatory machinery controls the segment-polarity genes and Hox genes in the spider as their expression boundaries match exactly. The discrete posterior expression border of the spider Hox genes is the result of genes that control them and these therefore may be an output of the upstream segmentation machinery that also control the expression boundaries of the segment-polarity genes. The assumption that the discrete Hox gene expression boundary could be the result of the segmentation process is strengthened by previous work in Cupiennius that suggested that there may be at least partially a difference in the mechanisms that specify the anterior segments and the posterior segments [34]. The posterior segments form sequentially from a posterior growth zone and may be partially regulated in a different way. The discrete boundary of Hox gene expression at the prosoma-opisthosoma boundary therefore could reflect such a difference in the regulation of segmentation between the anterior and the posterior segments.
Answers to all your Biology Questions
