Pair-Rule Gene Orthologues Have Unexpected Maternal Roles in the Honeybee (Apis mellifera)

Pair-rule genes are a class of segmentation genes first identified in Drosophila melanogaster. In Drosophila, these genes act to translate non-periodic information produced by the overlapping patterns of gap gene expression into patterns of gene expression in every other segment. While pair-rule genes are, for the most part, conserved in metazoans, their function in pair-rule patterning is not. Many of these genes do, however, regulate segmentation in arthropods and do so with dual-segment periodicity. Here we examine the expression and function of honeybee orthologues of Drosophila pair-rule genes. Knockdown of the expression of these genes leads to extensive patterning defects, implying that they act in early patterning, as well as segmentation in honeybee embryos. We show that these pair-rule gene orthologues indeed regulate the expression of honeybee maternal and gap genes implying roles in maternal patterning of the honeybee embryo.


Introduction
Pair-rule genes are a set of genes identified in Drosophila melanogaster that act in segmentation [1]. Pair-rule genes translate non-segmental information, from the overlapping gradient expression domains of gap and maternal coordinate genes, into segmental information (Reviewed in [2]). The overlapping periodic expression domains of pair-rule genes lead to the segmental expression of segment polarity genes, which act to initiate and maintain the parasegment boundary, effectively completing segmentation. Mutations in pair-rule genes cause different defects in adjacent segments, for example even-skipped (eve) causes loss of each even-numbered segment [1]. Pair-rule genes are often expressed in a pattern consistent with their function, eve, for example, is expressed in even-numbered segments [3].
In other arthropods the expression patterns and function of pair-rule genes have been difficult to interpret [4][5][6][7][8][9][10][11][12][13][14][15]. The orthologs of Drosophila pair-rule genes are often not expressed in a classical pair-rule pattern, either not being expressed in any pattern similar to pair-rule [4,9], or being expressed with 'dual segment periodicity' [12,15]. In these latter cases the genes are often expressed in broad domains covering two segments, with the expression patterns splitting to form segmental stripes. The function of these genes in non-drosophilid insects has often been difficult to determine due to the lack of tools available to manipulate gene expression.
The most extensive analysis of pair-rule gene orthologue function outside Drosophila has been carried out in the beetle Tribolium castaneum [11,16]. In this species the orthologues of the primary pair-rule genes eve, runt (run) and odd-skipped (odd) act as pair-rule genes, but produce asegmental embryos when knocked down because they activate each other's expression. They also show that orthologues of paired (prd) and sloppy-paired (slp) act as secondary pair-rule genes whereas hairy (h), fushi-taratzu (ftz), oddpaired and Tenacin-major do not act as pair-rule genes at all.
Expression patterns suggesting pair-rule or dual segment periodicity patterning, especially in orthologues of Drosophila pair-rule genes, have also been found in Schistocerca gregaria locusts [12], the cricket Gryllus bimaculatus [7], the spider mite Tetranychus urticae [14] and in the myriapod Strigamia maritima [17]. This is not to say that these genes are expressed in these patterns, or have pair-rule functions, in all arthropods. In locusts eve and ftz are expressed in a posterior domain [4,9], in the milk-weed bug Oncopeltus fasciatus eve has no pair-rule function, but does act as a gap gene [6]. In the crustacean Sacculina carcini, ftz is expressed only in the nervous system [18], and in the myriapod Lithobious forficatus, eve is expressed in a posterior domain and a few segmental stripes of cells during segmentation [19].
It seems that while pair-rule patterning is conserved in arthropods, many orthologues of Drosophila pair-rule genes often do not act in segmentation, or have other roles in development.
Here we describe the expression patterns and functions of four honeybee pair-rule gene orthologues, fushi-taratzu, even-skipped, runt and hairy. In honeybee three pair-rule orthologous gene expression patterns have been previously described, prd [15], eve [20,21] and ftz [22] but no functional analysis has taken place. Honeybees develop in a long-germ band mode, similar to Drosophila [23]. The currently accepted phylogeny of holometabolous insects, however, implies that long germ development in these two species evolved independently [24,25]. This gives us the opportunity to examine the function of pair-rule genes in a distantly related, but morphologically similar, embryo.
Here we report that each of these genes is expressed with dual segment periodicity during honeybee segmentation but that, remarkably, three of these genes are also expressed maternally and have functions that affect the expression of maternal coordinate and gap genes in honeybees.

Materials and Methods
Cloning of eve, hairy, ftz and run from A. mellifera The cloning of Am-eve, Am-ftz, Am-run, Am-prd, Am-gt, Am-cad, Amkr, Am-otd-1 and Am-hb has been reported previously [20,22,26,27], A fragment of Am-h coding sequence was amplified using the following oligonucleotide primers TCCCCCGCGCCGACCTC, Amh59 & TTTCTCCTCCACCTCCCGCACCAC, Amh39. Details of all genes discussed in this work can be found in Table S1.
Whole-mount in situ hybridization to honeybee embryos and queen ovaries In situ hybridization on honeybee embryos or queen ovarioles was carried out as described in [28]. Embryos were counterstained with DAPI and mounted in 70% glycerol. Images were captured on an Olympus BX61 microscope with a DP71 camera. Embryos were staged as per [29]. In all cases control embryos and ovaries, stained with sense probes for each gene, were examined and found to have no specific staining.

Expression and RNAi phenotypes of honeybee pair-rule gene orthologues
To determine the domains of expression and function of honeybee pair-rule gene orthologues we examined the RNA expression and RNA interference knockdown phenotypes of Ameve, Am-runt, Am-h and Am-ftz (Figures 1, 2, 3, 4).
Am-eve RNA is present in the ovarioles of the queen ovary, in maturing oocytes and in the posterior nurse cells ( Figure 1A). In about-to-be laid oocytes, Am-eve RNA becomes enriched at the anterior pole ( Figure 1B). In early embryos (stage 1-4) Am-eve mRNA is distributed through the embryo, and enriched around energids at early stages ( Figure 1C) and generally throughout the embryo as it cellularises ( Figure 1D). As stage 4 progresses Am-eve RNA is lost from both anterior and posterior poles of the embryo leaving a broad domain of expression, modulated with stronger expression at the anterior and posterior ends of the domain, in central regions of the embryo ( Figure 1E). In late stage 4 and the beginning of stage 5, broad stripes of cells begin to express Am-eve RNA in trunk regions of the embryo and then split, with new broad stripes appearing in anterior-posterior sequence, and then splitting as the next broad stripe appears ( Figure 1F and [20]).
To determine the role of Am-eve in honeybee embryogenesis we injected double stranded RNA (dsRNA) targeting Am-eve into justlaid embryos and then incubated these embryos until hatching (72 hours later). Compared to control RNAi injections ( Figure 1G), Am-eve RNAi produces phenotypes ranging from individuals with fused posterior segments and an absent terminal segment (marked by bar in Figure 1H) to asegmental larvae with a distinct head with mouthparts as the only identifiable morphology ( Figure 1I).
Am-run RNA is detected in the ovary in mid-stages, where RNA is expressed by the nurse cells and present in an anterior domain, and a stripe along one side of the oocyte (Figure 2A). In just-aboutto-be-laid oocytes ( Figure 2B) Am-run RNA is present only in anterior regions. In early embryos, Am-run RNA is present in a very faint domain in abdominal regions ( Figure 2C). By stage 5, broad stripes of cells begin to express Am-run RNA in anteriorposterior sequence along the embryo ( Figure 2D), which then split as development proceeds ( Figure 2E and [27]).
Knockdown of Am-run expression resulted in larvae with defective segmentation ( Figure 2F). The majority of Am-run RNAi larvae have only slight indications of segmentation, particularly reflected in the organization of trachea. Segmentally-organised trachea are less densely spaced, perhaps indicating a loss of alternating segments, or an expansion of remaining segments. The labrum is enlarged (lb) and gnathal appendages are absent ( Figure 2F).
Am-hairy mRNA is detected in the ovary in mid-stage oocytes, with faint RNA expression in the nurse cells and strong staining for RNA throughout the oocyte ( Figure 3A). This RNA staining is greatly decreased in just-about-to-be-laid oocytes ( Figure 3B) and is absent from early embryos up to stage 4 ( Figure 3C). Zygotic expression of Am-h RNA is first detected late in stage 5 as a broad thoracic stripe, quickly joined by thinner stripes in anterior to posterior sequence (arrows in Figure 3D). During stage 6, eight stripes of Am-hairy form in anterior to posterior sequence in abdominal regions ( Figure 3E-F).
Am-h RNAi injected embryos produce larvae with fused thoracic and anterior abdominal segments, with many larvae showing fusion of all segments ( Figure 3G).
Am-ftz RNA is expressed maternally in oocytes and nurse cells at mid-stages of oogenesis ( Figure 4A). Am-ftz mRNA comes to be enriched at the anterior pole of mature oocytes ( Figure 4B). Am-ftz RNA is associated with energids as they populate the egg surface after laying, causing redistribution of the anterior maternal RNA (stage 2, Figure 4C and D). By late stage 4, Am-ftz RNA is expressed in a broad abdominal domain of cells and is absent from cells at the anterior and posterior poles ( Figure 4E). Expression in this broad domain first becomes modulated and later splits (stage 5) to form seven broad stripes of cells expressing Am-ftz RNA ( Figure 4F, G and [22]).
Injection of dsRNA targeting Am-ftz results in larvae with absent anterior segmentation and head patterning but with clear thoracic and abdominal segments ( Figure 4H and 4I). These phenotypes are similar to those obtained with weak knockdown of the anteriorpatterning genes, Am-otd1 and Am-hb [32].
The larval phenotypes of honeybee RNAi knockdown experiments are often difficult to interpret because the cuticle is weak and has few landmarks that allow segments to be distinguished. To better interpret the phenotypes seen in our RNAi knockdown experiments we examined their effects on segmentation gene expression.

Segment polarity gene expression in pair-rule gene knockdown embryos
Pair-rule genes in Drosophila feed patterning information forward into the segment polarity network. By examining the expression of a marker segment polarity gene, engrailed (named e30 in honeybee [33]), we aimed to determine if segment polarity gene expression is affected by pair-rule gene knockdown. We stained the nuclei of stage 9 embryos, at the end of the segmentation process, with DAPI and examined the expression of e30 RNA, which marks the anterior compartment of each parasegment ( Figure 5A and B).
In all our RNAi experiments, except those targeting Am-ftz, a range of phenotypes was produced. Figure    In severely affected Am-eve RNAi stage 9 embryos ( Figure 5C and D) no expression of e30 appears ( Figure 5D) and the embryos have considerably fewer cells as determined by DAPI staining ( Figure 5C). More mild phenotypes have clear stripes of cells expressing e30 RNA, with loss or weakened expression only in the stripes marking posterior segments ( Figure 5E).
Severely affected Am-run RNAi embryos ( Figure 5F and G) also have no expression of e30 RNA ( Figure 5F). In DAPI stained embryos the germ band is distinguishable but is not visibly segmented ( Figure 5F). DAPI staining also reveals defects in the extra-embryonic membranes of Am-run RNAi embryos. In wild-type stage 9 embryos, the amnion is visible over the dorsal surface of the yolk ( Figure 5A). In Am-run RNAi embryos, the amnion is expanded, particularly in the posterior ( Figure 5F). The serosa, normally removed before staining, is located at the anterior (distinguishable from embryonic tissue due to its large, widely spaced nuclei), and fails to expand to envelope the embryo ( Figure 5F). In more weakly affected embryos ( Figure 5H), the serosa envelops the embryo, as in control embryos, and stripes of e30 RNA expressing cells are visible in the germ-band, but are less ordered than in control injected embryos.
In severely Am-h RNAi embryos ( Figure 5I and J) anterior stripes of e30 RNA are absent, but disorganized expression is seen in central regions ( Figure 5J) where the abdominal stripes would normally be present. Staining with DAPI confirms loss of anterior and abdominal segments ( Figure 5I). As in Am-run RNAi treated embryos, the serosa has failed to expand in Am-h RNAi embryos, but the amnion is present and slightly expanded in the posterior ( Figure 5I). In more mildly affected embryos, disorganised stripes of e30 RNA are present throughout the germband ( Figure 5K). These experiments demonstrate that knockdown of these pairrule gene orthologues have profound affects on e30 RNA staining. Am-h RNAi embryos have a phenotype consistent with roles in segmentation for these genes. The phenotypes of the other pairrule orthologues are also consistent with patterning roles earlier in development, as for Am-ftz, obscuring somewhat their later roles in segmentation, some of which may produce pair-rule-like phenotypes (Am-run, Figure 2F), potential pair-rule like modulation of e30 RNA stripes ( Figure 5H and K)), or more generally in segmentation as implied by the lack of e30 RNA staining in Ameve RNAi and Am-run RNAi embryos.

Am-ftz
The defects in extra-embryonic membranes in these specimens imply that our RNAi experiments are disrupting the formation and/or patterning of these membranes. We thus examined the expression of a marker of extra-embryonic fate, Am-zen [22,32], at stage 5 in control and injected embryos. Knockdown of any one of our honeybee pair-rule gene orthologues causes significant changes in Am-zen expression and morphology of the extraembryonic membranes at early stages ( Figure S1). Pair-rule gene expression in pair-rule orthologue knockdown embryos In Tribolium, RNAi knockdown of either Tc-eve or Tc-run produces larvae lacking abdominal segments, due to a pair-rule regulatory circuit in which pair-rule genes activate each other's expression [16]. Since knockdown of Am-eve and Am-run also produces larvae with loss of segments and segment polarity gene expression, we examined the effect of knockdown on the interplay of gene regulation between these genes ( Figure 6) to determine if a similar circuit is present.
Am-eve ( Figure 6A and [20]), Am-run ( Figure 6B and [27]) and Am-h ( Figure 6C) are expressed during segmentation as dual segment periodicity stripes, which form in anterior posterior sequence and then split to form segmental stripes during stage 5 and 6).
In Am-eve RNAi embryos, no striped expression of Am-run or Am-h ( Figure 6D-E) occurs. In Am-run RNAi embryos, Am-eve expression collapses to a single abdominal stripe of cells ( Figure 6F), stripes of Am-h are disrupted ( Figure 6I). In Am-h RNAi embryos Am-eve stripes are reduced in central regions ( Figure 6H), Am-run becomes ubiquitously expressed, with striped variation in expression levels, in anterior and central regions ( Figure 6I). In Am-ftz RNAi embryos, Am-eve stripes are reduced in the anterior and fused in central regions ( Figure 6J). Am-run stripes become poorly defined and Amrun RNA is present as low levels throughout the embryo ( Figure 6K). Ubiquitous expression of Am-h, with striped variation in expression levels, occurs in Am-ftz RNAi embryos ( Figure 6L).
These experiments indicate considerable cross-talk between pair-rule genes but also show that many of the defects observed in RNAi phenotypes of pair-rule gene orthologues are more extensive than might be expected if they were acting only as pair-rule genes. This makes it very difficult to determine if the effect of gene knock-down in each experiment is direct, through regulation of that pair-rule gene, or indirect, through regulation of some early part of the segmentation process.
Do honeybee pair-rule gene orthologues regulate gap gene expression?
The early maternal expression of Am-eve, Am-ftz and Am-run, and the severe effects on both anterior/posterior and dorso/ventral patterning, imply that these genes may have significant patterning roles early in development. To test this possibility we examined the effect of RNAi knockdown of these genes on the expression of previously identified honeybee gap genes [20,30,32].
Am-gt is required for patterning the anterior of honeybee embryos [20]. At stage 4, Am-gt RNA is detected in a thoracic domain and a posterior stripe ( Figure 7A and [20]). In Am-eve RNAi ( Figure 7B) embryos, the Am-gt expression domain in the anterior of the embryo is absent, while the posterior domain expands towards the anterior, especially in ventral regions of the embryo. In Am-run RNAi ( Figure 7C) embryos, the anterior domain of Am-gt expression does not extend as dorsally as in control embryos, but the domain extends toward the posterior, with RNA detected in abdominal regions of the embryo. The anterior domain of Am-gt also does not extend as far to the anterior as in control embryos. The posterior stripe of Am-gt is absent in Am-run RNAi embryos (asterisk, Figure 7C).
In Am-h RNAi stage 4 embryos, Am-gt expression is similar to wildtype, except the anterior domain forms a slightly different shape and perhaps doesn't extend as far anterior as in control embryos ( Figure 7D arrowheads).
In Am-ftz RNAi embryos, the anterior domain of Am-gt RNA expression is absent, while the posterior stripe is still present, though often disrupted or slightly expanded ( Figure 7E).
Am-kr is expressed in a central domain in stage 4 embryos where it acts to pattern thoracic and abdominal segments ( Figure 7F and [20]). Knockdown of Am-eve leads to weak over-expression of Am-kr RNA throughout the embryo excepting the posterior pole ( Figure 7G).
Knockdown of Am-run expression ( Figure 7H) results in a slight reduction in the extent of the Am-kr expression domain.
Knockdown of Am-h has little affect on the expression of Am-kr, though in some embryos (as in that pictured), the domain is shifted slightly to the anterior (Figure, 7I).
Knockdown of Am-ftz appears to reduce the both intensity and the width of the expression domain of Am-kr ( Figure 7J).
Am-tll is required for terminal patterning and is expressed in a posterior cap of cells and triangular anterior domain at stage 6 ( Figure 7K and [30]). In Am-eve RNAi embryos, the anterior domain of Am-tll RNA expression is absent, while the posterior cap appears unaffected ( Figure 7L). In some Am-eve RNAi specimens (data not shown), the anterior domain is fainter, but not entirely absent.
In Am-run RNAi embryos, the anterior domain of Am-tll expression is absent, and the posterior domain unaffected ( Figure 7M). Am-h RNAi does not have any appreciable affect on Am-tll RNA expression ( Figure 7N). In Am-ftz RNAi embryos, expression of Am-tll is reduced to faint expression in the posterior, and absent from the anterior of the embryo ( Figure 7O).
RNAi knockdown of all of the pair-rule gene orthologues we have examined show some effect on the expression of gap genes in the honeybee. In the case of Am-h, these effects are slight. For Amftz, Am-run and Am-eve, the range and scale of the defects in gap gene expression patterns led us to speculate that these pair-rule gene orthologues may be acting to modify the expression of maternal coordinate genes.

Do honeybee pair-rule gene orthologues regulate maternal coordinate gene expression?
To test this possibility we examined the expression of three previously identified maternal coordinate genes in pair-rule gene knockdown embryos.
In stage 4 embryos the RNA from the anterior patterning gene Am-otd1 is detectable in cells in the anterior third of the embryo (between the arrowheads in Figure 8A) and, weakly, at the posterior terminus ( Figure 8A and [32]). In Am-eve RNAi embryos, Am-otd1 RNA is found throughout the embryo, being absent only in a small domain at the posterior terminus ( Figure 8B arrowheads). Am-run RNAi knockdown results in weak staining for Am-otd1 RNA throughout the embryo ( Figure 8C). Am-otd1 staining appears unchanged in Am-h RNAi embryos with both anterior (arrowheads) and posterior domains showing no effect of knockdown ( Figure 8D). In Am-ftz RNAi embryos, the anterior domain of cells expressing of Am-otd1 mRNA is absent, but the posterior stripe of expression is present as in control embryos ( Figure 8E).
Am-hb, a regulator of anterior and thoracic development [32], is expressed at stage 4 in a distinct thoracic stripe of cells, and in a posterior stripe. (Figure 8F and [32]). Am-eve RNAi embryos show overexpression of Am-hb expression throughout the embryo at stage 4 ( Figure 8G). Am-run RNAi knockdown leads to variable effects on Am-hb expression, producing a disorganized central domain of strong Am-hb expression, with no posterior stripe. The extent of this central domain varies between injected embryos ( Figure 8H). Am-h RNAi embryos show weaker than control staining of the posterior stripe of Am-hb, and slight effects on the thoracic domain, often narrowing it as shown in Figure 8I. Am-ftz RNAi embryos have no expression of Am-hb RNA although occasional specimens show a faint posterior stripe ( Figure 8J).
Am-cad RNA is expressed in cells in a broad abdominal-posterior domain, not including the posterior terminus, where it acts to pattern abdominal and posterior regions ( Figure 8K arrowheads and [20]). In Am-eve RNAi embryos Am-cad RNA spreads more anteriorly than in control embryos, but still with the same posterior boundary ( Figure 8L, arrowheads). Expression also does not extend as far dorsally as in controls. In Am-run RNAi knockdown embryos, Am-cad RNA is present in a much smaller posterior domain than control embryos, not extending as far anterior, or dorsally, but with the same posterior boundary. This smaller expression domain often has different intensities of staining for Amcad RNA ( Figure 8M). In Am-h RNAi embryos, Am-cad RNA extends further into the anterior than in control embryos, while still respecting the posterior boundary ( Figure 8N, arrowheads). In Amftz RNAi embryos, Am-cad expression is reduced to cells in a broad band in the central regions of the embryo of the embryo ( Figure 8O arrowheads), with a faint stripe of cells expressing Am-cad RNA anterior to it (Figure 8O asterisk).
The implication of the phenotypes of pair-rule orthologue gene knockdown on maternal coordinate gene expression is that these genes are having unexpected early patterning roles in honeybee embryos.

Pair-rule gene orthologues and maternal patterning
The expression patterns and RNAi knockdown phenotypes of Am-eve, Am-run and Am-ftz indicate these genes are acting in early patterning of the embryo, as well as in later segmentation. All are expressed maternally, with maternal RNA persisting into early embryonic stages, providing the potential for early patterning roles. Our double-stranded RNA injections in just-laid embryos are likely triggering RNAi-mediated knockdown of this maternally deposited RNA, as well as affecting later striped patterns of zygotic expression. The phenotypes we see are thus reflections of the role of the RNA in the early embryo, not the oocyte.
Am-eve, Am-run and Am-ftz all have early patterning roles, affecting the expression of key maternal genes, as well as gap, pairrule and segment polarity genes. These early patterning defects make a clear interpretation of their interactions during segmentation difficult.
Am-h, despite having a role in limiting Am-cad expression and perhaps affecting hunchback expression, appears to act mainly in segmentation, as these early roles leave little phenotypic effect when knocked down.
Am-eve normally represses Am-otd1 and Am-hb, as both are over expressed in Am-eve RNAi embryos. A change in the expression of these maternal patterning genes probably explains the severe Ameve RNAi phenotypes. Am-ftz has a role in anterior patterning, probably through activation of both Am-otd1 and Am-hb. Am-run appears to be a regulator of posterior development, repressing the posterior domains of both Am-hb and Am-otd1. It is not clear if these interactions are direct or mediated through other factors.
Maternal roles for these genes have not been described in other insects, although maternally expressed genes with a pair-rule mutant phenotype have been described in Drosophila [34][35][36][37][38]. In these cases, maternal expression produces a co-factor for a zygotic pair-rule gene, regulating specific pair-rule genes or acting in cuticle development.
In Drosophila, run has been shown to act with gap gene properties by antagonizing transcriptional activation by Bicoid [39]. The effects, however, are slight, but do suggest that this gene may act more generally in segmentation.
In the cricket Gryllus and in the milkweed bug Oncopeltus, eve orthologues have earlier roles in segmentation, producing gap gene like effects [6,7]. This is thought to be due to the broad expression domain of eve that appears in central regions of these embryos and then splits into stripes. Our results are similar, but Am-eve in the honeybee acts earlier, regulating maternal genes as well as gap genes. Despite these examples, the case of three pair-rule gene orthologues having major patterning roles in early development is unique. Two possible explanations exist. Either these maternal roles are ancestral ones, supported by the early roles for eve in Oncopeltus [6] and Gryllus [7], or these three genes have been coopted into maternal patterning in the lineage leading to honeybees, and these roles are likely to be specific to that lineage. In Oncopeltus and Gryllus, gap gene functions of eve have been suggested as being due to the broad initial domain of eve, which then splits into stripes [6,7]. This expression domain is also present in honeybee Am-eve expression [20], and is not equivalent to the maternal expression of Am-eve, Am-ftz and Am-run we have presented here. We hypothesize, therefore, that the maternal expression of these three pair-rule gene orthologues is due to cooption of these genes into maternal patterning in the honeybee lineage.
It is interesting to note the activity of Am-ftz in this regard. Ftz is a so-called 'rogue' hox gene [10], related to Hox 6, with roles in segmentation in insects. Ftz has changed its expression pattern, implying a change in its function, multiple times in Arthropod evolution [40][41][42]. Am-ftz acts in maternal patterning to regulate anterior development, probably through regulating both Am-hb and Am-otd1. This draws an interesting analogy with Drosophila bicoid, also a rogue Hox gene, though in this case related to Hox 3 rather than Hox 6, which has also taken up a maternal role in Drosophila anterior patterning and regulates both hunchback [43] and otd [44] (ocelliless (oc) in Drosophila). While Am-ftz does not have the Glutamine to Lysine substitution at position 50 of the homeodomain seen in bicoid and related to its evolution from a Hox 3 ancestor [45], it is intriguing that a Hox gene is also found in bees with a maternal, anterior patterning role.

Roles in segmentation and pair-rule patterning
Despite the early patterning roles for these pair-rule gene orthologues in honeybees, it is clear they also function in segmentation. All are expressed with dual segment periodicity, with broad stripes appearing across two segments, which then split to form single stripes, as seen in Gryllus [7], and Schistocerca [12].
In honeybees, the maternal role for many of these genes obscures their activity in segmentation as it is not clear if the defects in pair-rule gene expression reflect a direct regulation event, or if they are the consequences of earlier patterning deficits.

Changing roles for conserved genes in evolution
Our finding of maternal patterning roles for conserved pair-rule genes in the honeybee indicates the propensity of such genes to be co-opted to new functions and new expression domains during evolution. The cooption of these genes into an ancestral process implies that the re-use of conserved genes in novel processes may be a common process that must be taken into account if we are to understand how developmental processes evolve.