Successful inhibition of gene expression by double-stranded RNA interfer- ence has already been reported for two honey bee genes expressed during
embryonic stages [17, 18]. Here, this technique was applied to silence the ex- pression of csp5, and gain insight into the function of this gene. Eggs, 12±2
hours POV, at the pre-blastodermal or blastodermal stage, were injected with a double stranded csp5 RNA construct, and their phenotype inspected at regular intervals. Around 44 hours, at the germ band stage, the buds of the head appendages and the protocerebral lobes, and the deutocerebrum were visible in both the control and injected embryos (gures 5.7.c, 5.7.d). The phenotype triggered by the injection of double-stranded RNA was most dramatic around 67-74 hours POV, just before the control eggs hatched. At this stage, all normal embryos have the characteristics of the rst instar larva, as shown in gure 5.8.b.
(a) 12 hours treated
1 mm
(b) 12 hours control
(c) 44 hours treated (d) 44 hours control
Figure 5.7: RNAi phenotypes at 12 and 44 hours. Controls were injected with DNA containing the same sequence as the double stranded RNA.
All dsRNA injected embryos show developmental abnormalities. Figure 5.8 illustrates a few selected examples. The head, if visible, has a blob-like shape with blurry edges and lacks well developed cerebral lobes, mouthparts and antennae (gure 5.8.c-1). The other type of deformation is a complete lack
of the head (gure 5.8.c-2, 5.8.c-3, 5.8.c-4, 5.8.c-5). Instead, this part of em- bryo has a smooth uniform appearance. In all surviving embryos, the thorax lacks segmentation. There is a noticeable dierence in embryo size, as most of the treated embryos occupy the whole egg volume. Approximately 30% of dsRNA-treated embryos show mosaic phenotypes characterised by visibly dierentiated last 5-7 abdominal segments, some with distinctly developed spiracles (gure 5.8.c). In 74 hours-old treated embryos, the chorion surface on the posterior side of the egg is no longer smooth (gure 5.8.d-1, 5.8.d-2), and appears to be torn. This may be related to the rst stage of normal hatching during which the chorion membrane begins to dissolve followed by gentle movements by the head end of the egg [42]. However, no movement was observed in this type of `larvae' and none of the treated individulas were able to hatch (gure 5.8.b). When the chorion was removed manually from 10 mosaic embryos (74-76hrs-old) that possessed visible last 5-7 abdominal segments the anterior portion disintegrated rapidly (gure 5.8.d-3), whereas the posterior part of this embryo with visible segments remained intact, sug- gesting that the cuticle was partially synthesised.
As noted in [17] the mosaic phenotypes might be caused by incomplete diusion of the dsRNA construct before initiation of the cellular blastoderm stage at 12 hrs.
1 mm
1mm
(a) 66 hours control (b) 74 hours control (hatching larva)
1
2
3
4
5
1
2
3
(c) 66 hours treated (d) 74 hours treated
Figure 5.8: RNAi phenotypes at 66 and 74 hours. Controls were injected with DNA containing the same sequence as the double stranded RNA. Rep- resentative embryos out of 1960 injected eggs are shown.
5.5 Discussion
This chapter presents a detailed characterisation of the honey bee uth gene, encoding a member of the arthropod CSP protein family. It is shown that uth is expressed in a distinct temporal pattern characteristic of maternal- zygotic genes. Following the rst wave of expression in the queen's ovary the uth message is transmitted to newly-laid eggs, but disappears completely at 24 hours POV. The maternal transcript is replaced at 40-44 hours by a de novo produced embryonic mRNA, that again disappears before hatching. In situ hybridization shows a restricted spatial expression of uth with the high- est level of the message found in the areas that are sclerotised. Consistent with this expressional pattern, the inhibition of uth by RNAi results in a diuse morphology in injected embryos that also are unable to hatch and consequently die before the rst instar larva. Taken together, these data demonstrate that uth belongs to the embryonic lethal category, and encodes an essential protein that is involved in embryonic integument synthesis. This nding suggests a potential link between the role of some CSPs in embry- onic development and chemosensation. With regard to a specic function it is reasonable to assume that csp5 encodes a carrier protein transport- ing lipophilic compounds used for embryonic integument synthesis. Because the embryonic `molts' in the honey bee have received very little attention the synthesis of embryonic covers or integuments in this species is not well understood. In other insects, the integument undergoes pronounced subcel- lular alterations, correlated with changing functions. So far, Manduca sexta is the only holometabolous insect in which the deposition of embryonic mem-
branes and cuticles has been examined by electron microscopy [84, 192]. As in hemimetabolous insects, the embryo of M. sexta secretes three covers at approximately the same developmental stage. One dierence is the second embryonic cover, which in Hemimetabola exhibits a cuticular organization, but M. sexta has a membranous, cuticulin-like structure [192].
The honey bee draft genome assembly revealed a predicted `cuticular' pro- teome that appears to be smaller than those in other insects suggesting, that social life may reduce the pressure on protecting the development of pre-imaginal forms [160]. However, the last integument deposited by the embryo before hatching must represent the cuticle of the rst instar larva, an important developmental stage that determines the capacity of a newly hatched organism to function in the outside world. For most animals, the transition from a relatively predictable egg shell environment to the unpre- dictable environment of the outside world is critical [165]. Thus, further studies on uth (csp5 ) and its RNAi-induced mutant phenotype will provide important insights into the molecular mechanism underlying this process. In addition, an analysis of the RNAi phenotype by microarrays should iden- tify other genes involved in this transition, and thus allow rapid progress to be made in unravelling the entire network of interactions during embryonic cuticle synthesis.
This study highlights the diculties of assigning homologies, sensu strictu, to a multi-gene family encoding proteins with a moderate level of conserva- tion. It also puts into perspective the emerging multifunctionality of CSPs
that have so far been considered as predominantly olfactory proteins ([122]). While the ongoing genome projects will reveal the highly conserved core of proteins in dierent insect species, it is probable that the crystallographic analysis of orphan proteins will be essential to understanding the origins of evolutionary novelties in dierent lineages.
This work, together with two previous studies [17, 18] shows that embry- onic development in the honey bee can be successfully manipulated by RNA silencing. When applied to the whole genome, this approach should iden- tify the majority of genes required for early developmental processes in the honey bee, paving the way for a signicantly improved understanding of these processes and their regulation at the molecular level.
Conclusions
The sequencing and annotation of the honey bee genome has been widely applauded as an important scientic achievement [45, 139]. It is not only the second non-dipteran to be sequenced, but also the rst eusocial in- sect. It is therefore expected that the lessons that are emerging from the honey bee genome will invigorate what might be called `comparative so- ciogenomics'. One area that is central to social behaviour, and in which honey bees seem poised to make a major contribution, is chemosensation. By searching through the honey bee genome dataset, two gene families encoding OBPs and CSPs, that are both of great interest to the eld of chemosen- sory research, have been annotated and characterised. Comparisons with other genomes reveal that both families are relatively small in the honey bee conrming that there is little relationship between total gene number and the morphology and behavioural capacities of diverse organisms in dierent phyla. The evidence presented here broadens the understanding of these gene families, and oers means for designing experiments that will answer a variety
of questions about both chemoreception and chemosensory transduction. The greatest strength of the genomics approach is its ability to uncover new molecules without making any prior assumptions. One of the most in- teresting achievements of this work is a functional description of a chemosen- sory gene that is essential for embryonic development. This gene, designated unable-to-hatch, encodes a protein involved in the process of making embry- onic integuments in the honey bee about which almost nothing is known. Because this gene is amenable to RNAi silencing it oers a convenient exper- imental model that should lead to a better understanding of how embryonic integuments in insects are made.
6.1 Roles of OBPs and CSPs in Olfaction
The insect genomes encode fewer olfactory receptors and more putative ol- factory carriers than the genomes of vertebrates and nematodes (table 6.1). Although the honey bee follows this trend, the genome of this species encodes fewer OBPs/CSPs, but more ORs than those of other insects. Thus, if one considers gene numbers alone, the ndings presented here lend more support to the combinatorial model of olfaction in accordance with which OBP/CSPs and ORs act together to generate odour specicity [63]. In this model, the smaller number of OBP/CSPs in the honey bee would be compensated by the expanded receptor family, resulting in a discriminative capacity similar to that of the combined OBPs/ORs gene families in other insects. However, an extensive expressional proling of the honey bee OBPs and CSPs calls
this view into question. The fact that many honey bee OBPs and CSPs are expressed outside the olfactory system suggests other non-olfactory func- tions for these genes. The small subset of olfactory-specic OBPs and CSPs may perform `general duties' as broad specicity carriers, with little or no contribution to odour recognition.
Species OBPs CSPs ORs
Anopheles gambiae 66 7 79
Apis mellifera 21 6 160
Bombyx mori >30 >16 ?
Drosophila melanogaster 51 4 62
Tribolium castaneum 46 19 ?
Table 6.1: Number of CSP, OBP and OR genes in various insects.
One way to experimentally test the hypothesis of olfactory discrimination by a combinatorial use of ORs and odorant carriers would be to examine the assortment of odorant carriers associated with a type of olfactory sensillum expressing a given subset of ORs. If this assortment of olfactory carriers does not change from a sensillum of this type to another sensillum of the same type, the combinatorial hypothesis would be in doubt. One possible way to characterise the populations of ORs and olfactory carriers expressed in a given sensillum would be to extract the RNA from an individual sensil- lum, amplify its RNA and look at the expression of the genes of interest by hybridisation to a custom microarray containing those genes.
Clearly, the next step in the characterisation of both OBPs and CSPs and their roles in olfaction is to determine which genes are actually expressed in chemosensory sensilla, and which proteins are localized to the olfactory lymph. The honey bee is an excellent model organism to address this issue, because of the relatively small size of both gene families and the few members specic to olfactory organs.
It is believed that typical olfactory sensory neurons express a single OR [107], plus the conserved or83b [15]. Some D. melanogaster olfactory neurons express two ORs [55]. Olfactory neurons expressing the same OR converge to the same glomerulus [53, 178]. In the honey bee, olfactory neurons are also usually uniglomerular [26]. It can thus be hypothesised that most honey bee sensory olfactory neuron also express a single olfactory receptor. This is further supported by the fact that the number of ORs in D. melanogaster and in A. mellifera are close to the number of olfactory glomeruli in each of these species. It has also been shown that the sensory neurons of a given poreplate each project to a distinct glomerulus [26]. This suggests that each poreplate expresses a combination of 15-30 dierent ORs. The number of combinations of 20 ORs taken from a pool of 160 ORs amounts to a staggering 1.4×1025
possibilities, and could complicate the identication of populations of sensilla expressing the same subset of ORs. Since this number of combinations is far above the number of poreplates on honey bee antennae (334,629 in drones and 46,818 in workers [44]), it may even be that each poreplate expresses a distinct subset of ORs.
Depending on the ORs pattern of expression in dierent olfactory sensilla, a test of the combinatorial hypothesis in the honey bee may not even be possible. These and other related questions await further studies on the expression of the honey bee ORs in dierent olfactory sensilla.