A glypican shield permits long-range signalling by Wnt

The developmental sculpting of a tissue requires local secretion of morphogens, a small group of highly specialised proteins which spread across the tissue, forming a long-range gradient that determines cell fate in a concentration-dependent manner. Among these morphogens is Wnt, a master regulator of many developmental processes which is also universally required for stem-cell self-renewal.

Despite the importance of Wnt signalling, the way in which the molecule spreads through the aqueous extracellular space to establish a gradient has perplexed the field since the discovery in 2003 that Wnt is lipidated; the lipid modification, palmitoleoylation, is essential for Wnt activity, but renders it insoluble. How, therefore, does an insoluble morphogen get secreted from the cells producing it, make its way through the extracellular space, and then finally engage with its receptor on a different cell?

Jean-Paul Vincent and colleagues have now found the solution to this mystery. In work led by Vincent’s postdoc Ian McGough, they have used a combination of Drosophila genetics, in vivo imaging, biochemistry and structural biology to demonstrate that the glypican Dlp (Dally-like protein) shields the palmitoleate moiety and promotes long-range Wnt signalling (McGough et al 2020).

As its name suggests, Wingless, the main Drosophila Wnt protein, is essential for correct wing development from the wing imaginal discs, small groups of about 30 cells in the embryo, which in the course of larval development give rise to the approximately 50,000 cells that form the adult organs. A multitude of signalling pathways are involved in patterning the wing, making it a valuable model system for studying signalling in a tissue context.

Importantly, the Wnt signalling pathway is well-conserved, allowing extrapolation from Wingless to vertebrate Wnt.

McGough and colleagues began by testing theories put forward previously to resolve the insolubility conundrum; it had been variously suggested that Wnt might travel through the extracellular space in multimeric micelles, with the hydrophobic lipids hidden in the central region, or alternatively that it could be transported in lipoprotein particles or exosomes. There was also evidence that SWIM, a lipocalin known to solubilise extracellular Wnt in tissue culture medium, might be involved.

To visualise potential interactions, the group used a morphotrap assay. A GFP-tagged version of Wingless was expressed from a knock-in allele in a defined stripe in wing imaginal discs, and membrane-associated anti-GFP nanobodies were expressed in a second stripe at right angles to the GFP-Wingless-expressing cells. GFP-Wingless could be detected binding to this transverse stripe of morphotrap nanobodies, showing that it was spreading from its originator cells to form a morphogen gradient. However, there was no co-localisation with markers for either lipoprotein particles or exosomes. Alternatively tagged Wingless molecules expressed from the second allele also did not colocalise, showing that micelles were not forming.

Confirming the imaging data, genetic perturbation of components of lipoprotein particles and exosomes had no effect on the Wingless gradient, and neither did expression of RNAi against Swim, the putative solubilising factor. Therefore, none of these previously advanced theories regarding extracellular movement were correct.

With this result in hand, McGough and colleagues examined a further possibility. Glypicans are a family of proteoglycans which are attached to cell surfaces via a GPI anchor, and are extensively decorated with heparan sulphate chains, previously shown to bind morphogens via electrostatic interactions. It was known from genetic analysis that the Drosophila glypicans Dally and Dally-like protein (Dlp) have a role in ensuring the correct extracellular distribution of Wingless, and confirming this, the group were able to show that loss of either Dally or Dlp led to a reduction in extracellular Wingless, with Dlp deletion having the stronger effect. Removal of the heparan sulphate chains of both glypicans also led to far less extracellular Wingless, and phenocopied the loss of Wingless signalling. However, as interaction with the heparan sulphate chains would not act as a lipid shield, the group decided to follow up a further report that Dlp might have a second Wingless binding site in its core region.

Using an in vivo Wingless binding assay, overexpressed Dlp was shown to trap palmitoleoylated Wingless, but it was unable to bind a normally secreted version of the protein lacking the palmitoleate moiety. To confirm these data, in collaboration with Yvonne Jones’s laboratory in Oxford, the group used biolayer interferometry to measure the binding affinity of purified Dlp^core (the unmodified globular protein) to palmitoleoylated versus unmodified peptides. They found that Dlp^core bound the former but not the latter, albeit with low affinity. In contrast, Dally^core had no such activity; the two proteins appear to have distinct roles.

This distinction between Dlp and Dally persists into the vertebrate kingdom. There are six human glypicans, four of which (GPC 1, 2, 4 and 6) are Dlp-like, with the remaining two (GPC 3 and 5) most related to Dally. These two groups could be distinguished by the in vivo trapping assay; only Dlp-like glypicans could bind and trap palmitoleoylated Wingless. 

If Wingless or Wingless-GFP are expressed in tissue culture, they can be visualised within cells, but are never seen in the medium, due to their insolubility. This gave the group a simple way to test whether Dlp could solubilise palmitoleoylated Wingless in the extracellular space. They co-transfected Wingless-GFP with soluble versions of Dlp or its human homologues GPC4 and 6, and showed that the presence of all three resulted in the appearance of large amounts of soluble Wingless-GFP in the extracellular medium. As expected, Dally-like proteins had no such effect. 

Taken together, these experiments showed that Dlp, but not Dally, binds to the palmitoleate moiety of Wingless, and that this is likely to be how Wingless is solubilised. 

The structure of Dlp had been determined some years previously, and there was no obvious lipid binding pocket. Repetition of the structural analysis indicated that this was indeed the case for Dlp in isolation, but solving the structure of a co-crystal of Dlp with a lipidated peptide gave a different result: two helices in Dlp can move apart to form a tunnel, allowing lipid to be inserted into the protein’s hydrophobic core. Mutation of the tunnel mouth to clamp it shut demonstrated that this opening was essential for lipid binding, and consequently for the ability of Dlp to engage and solubilise palmiteolyated Wnt. 

This elegant study shows that the Dlp subclass of glypicans can shield the lipid moiety of Wingless, and by extension Wnt, to enable the molecule’s transport through aqueous space for presentation to its signalling receptor, Frizzled. The low binding affinity of Dlp for Wnt fits well with its role as a transient custodian; Wnt could move from cell to cell by passing through a lawn of laterally diffusing surface-bound Dlp proteins, with each individual protein binding and then releasing it to the next, all the while keeping its lipid shielded. Such a mechanism would potentially allow for Wnt’s ultimate transfer to Frizzled, which has high affinity for the protein.

There are a number of outstanding questions, which Vincent’s group are currently exploring. How Wnt leaves the cell in which it is expressed to interact with extracellular Dlp proteins is unclear; the role, if any, of cytonemes, specialised filopodia implicated in morphogen movement, is also unknown. Most intriguingly, the mechanism may be of more general importance: Hedgehog, another secreted and lipidated morphogen, faces the same insolubility problems as Wnt, and might also use Dlp to facilitate its long-distance movement.