Dlp: the second HSPG in Slit/Robo signalling

The strengthening of phenotype in the sdc²³,ttv^00681b double mutant indicated that one or more HSPGs are required in addition to Sdc in Slit/Robo signalling. As the zygotic

Discussion axons and muscles, double mutants with sdc²³ were generated. The analysis of sdc²³;dlp^A187 homozygous double mutant embryos revealed an increase in the penetrance of the CNS and muscle phenotype (Fig. 19a and b) though the increase in muscle phenotype was not statistically significant (p>0.001). Phenotypic analysis showed that in sdc²³;dlp^A187 homozygous double mutant embryos 50% of counted segments showed ventral midline crossover of CNS ipsilateral axons in contrast to sdc²³ homozygous mutant embryos where only 19% of counted segments showed ventral midline crossovers (Fig. 19a). The muscle phenotype also showed an increase in penetrance from 1.1 muscle crossover per embryo in sdc²³ homozygous mutant embryos to 1.5 muscle crossover per embryo in sdc²³;dlp^A187 homozygous double mutant embryos (Fig. 19b). These results suggest Dlp to be a part of Slit/Robo signalling. The fact that both sdc²³,ttv^00681b and sdc²³;dlp^A187 double mutants showed an increase in CNS and muscle phenotype strongly suggests that Dlp is the specific additional HSPG that functions in Slit/Robo signalling.

In further support is the fact that Dlp is expressed on CNS axons and in stripes that overlap with muscle attachment sites in the embryo.

However, at this time it is premature to specify the mechanism of Dlp action. In analogy

Figure 21: A model for combinatorial mode of action of Sdc and Dlp in Slit/Robo signalling.

Dlp serves to transport and/or concentrate Slit on the target tissue and Sdc acts as a coreceptor to stabilise Slit/Robo interaction for efficient transduction of the repellent signal into the cell.

Discussion to the mechanism that has been proposed for the transfer of Dpp and Hh (Belenkaya et al., 2004; Han et al., 2004), it can be speculated that Dlp is involved in the extracellular transport of Slit. Alternatively, Dlp might be increasing the concentration of Slit outside the target cells by trapping it with its HSGAG chains the same way it traps Wg outside cells (Baeg et al., 2001). Absence of a cytoplasmic domain excludes any direct intracellular signalling role of Dlp. However the possibility of the GPI anchor of Dlp, recruiting Robo receptor and other essential signalling components into lipid rafts to facilitate signalling cannot be excluded. Nevertheless, the observation that zygotic dlp null mutants do not have a slit like phenotype strongly suggests that Sdc is the primary HSPG that is required for Slit/Robo signalling with Sdc and Dlp having a combinatorial mode of action. This combinatorial mode of action as proposed in Figure 21 can be verified by tissue-specific expression of sdc and dlp in sdc²³;dlp^A187 double mutants. If Dlp indeed functions to transport Slit then expression in sdc²³;dlp^A187 double mutants of dlp in the intermediate tissue through which Slit has to be transported should rescue the phenotype to that of a sdc²³ single mutant.

In conclusion, this work has shed some light on the mechanism of action of Sdc in Slit/Robo signalling. The data proves that Sdc function is required on Slit target tissues:

the CNS ipsilateral axons and muscles that express the Slit receptor Robo. Sdc plays no essential role in the secretion of Slit or its transport. Sdc acts via its extracellular domain on the surface of the target tissue and its mode of anchorage causes no disparity in its function. In all probability, Sdc does not only function to trap Slit and increase its concentration on the target tissue thereby enhancing the probability of Slit and Robo interaction but as an active coreceptor to stabilise Slit and Robo interaction. This is supported by the fact that a secreted form of Sdc does not rescue the sdc mutant phenotype. In the target tissue Sdc does not participate in intracellular signalling to transduce the repellent Slit signal. Additionally, Sdc does not direct the reorganisation of actin cytoskeleton in response to the repellent Slit signal.

Indirect evidence suggests that shed ectodomain of Drosophila Sdc is inactive in Slit/Robo signalling. Therefore, shedding of Sdc by the proteolytic activity of an

Discussion regulation of Slit/Robo signalling whereby shedding of Sdc terminates Slit and Robo interaction. This model would be in accordance with the described function of Sdc for the recycling of FGF receptor in Drosophila cells (Zimmermann et al., 2005). Therefore the idea that Sdc functions as a coreceptor to stabilise Slit and Robo interaction and shedding of the extracellular domain of Sdc by a serine protease activity results in the recycling of the Robo receptor together with the transmembrane and cytoplasmic domain of Sdc is feasible. This model proposes a function of the intracellular domain of Sdc in intracellular trafficking through the endosomal compartments, which fit to the results of Zimmermann et. al. (Zimmermann et al., 2005). The open question is whether this internalisation of the Sdc/Robo complex is required for the recycling of Robo or whether this process results in the downregulation of Robo activity. A detailed analysis of the rescue experiments with the sdc deletion transgene, sdcΔC-GFP, which lacks the cytoplasmic domain will be used to distinguish between these models.

Analysis of double mutants also suggests that Dlp plays a role in transducing the Slit repellent signal. A probable model is that Sdc acts in Slit recipient cells like a coreceptor to facilitate/stabilise Slit and Robo interaction while Dlp functions to transport and/or concentrate Slit on the surface of the target tissue. Therefore, when only Sdc is absent there is still a high concentration of Slit at the surface of the target tissue due to Dlp activity, so that the Slit repellent signal is transduced but at a low efficiency resulting in a weak slit like phenotype. In contrast, loss of Dlp results in a lower concentration of Slit on the axon growth cone surface due to impaired transport and/or lack of concentration.

Despite this, sufficient amount of Slit is present on the growth cone surface and aided by the coreceptor Sdc, the Slit repellent signal is transduced efficiently into the growth cone.

Therefore, no slit like phenotype is observed in the case of dlp homozygous zygotic mutants. However, in the absence of both Dlp and Sdc, Slit is neither transported and/or concentrated on the target surface nor is the coreceptor present to facilitate interaction between the low amounts of Slit present and its receptor Robo. This results in a dramatic reduction in the transduction of the Slit repellent signal into the target tissue, the consequence of which is an increase in the penetrance of slit like phenotype as compared to sdc²³ homozygous mutants. This model of a combinatorial mode of Sdc and Dlp activity required for two independent steps of signal transduction: ligand transport and

Discussion ligand reception, is worthy of investigation in other signal transduction pathways as well.

It has been shown that Dlp is required for Dpp and Wg transport in the wing disc (Belenkaya et al., 2004; Han et al., 2004) but the role of Sdc is not tested yet. Therefore, an interesting point to investigate in other signalling systems is if Dlp always functions for transport and Sdc for reception or if the mechanistic function can change depending on the signalling system and tissue.