Wnt and Hedgehog family proteins are secreted signalling molecules (morphogens) that act at both long and short range to control growth and patterning during development. Both proteins are covalently modified by lipid, and the mechanism by which such hydrophobic molecules might spread over long distances is unknown. The Drosophila lipoprotein particle, Lipophorin, bears lipid-linked morphogens on its surface and is required for long-range signaling activity of Wingless and Hedgehog. Wingless, Hedgehog and glycophosphatidylinositol-linked proteins copurify with lipoprotein particles marked by lipophorin, and co-localize with these particles in the developing wing epithelium of Drosophila. In larvae with reduced lipoprotein levels, Hedgehog accumulates near its site of production, and fails to signal over its normal range. Similarly, the range of Wingless signalling is narrowed. A novel function is proposed for lipoprotein particles, in which they act as vehicles for the movement of lipid-linked morphogens and glycophosphatidylinositol-linked proteins (Panakova, 2005).

In the developing wing of Drosophila, Hedgehog activates short-range target gene expression up to five cells away from its source of production, and longer-range targets over more than twelve cell diameters. Wingless can signal through a range of over 30 cell diameters. These morphogens are anchored to the membrane via covalent lipid modification. The mechanisms that allow long-range movement of molecules with such strong membrane affinity are unclear (Panakova, 2005).

Like Wingless and Hedgehog, glycophosphatidylinositol (gpi)-linked proteins transfer between cells with their lipid anchor intact. Gpi-linked green fluorescent protein (GFP) expressed in Wingless-producing cells spreads into receiving tissue at the same rate as Wingless, where it co-localizes with Wingless in endosomes. Thus, it is proposed that these proteins travel together on a membranous particle, which has been called an argosome (Greco, 2001). How might argosomes form? One possibility is that argosomes are membranous exovesicles. Such particles could be generated by plasma membrane vesiculation, or by an exosome-related mechanism. Alternatively, argosomes might resemble lipoprotein particles like low-density lipoprotein (LDL). Vertebrate lipoprotein particles are scaffolded by apolipoproteins and comprise a phospholipid monolayer surrounding a core of esterified cholesterol and triglyceride. Insects construct similar particles called lipophorins (Arrese, 2001; van der Horst, 2002). Lipid-modified proteins of the exoplasmic face of the membrane (such as GFPgpi, Wingless or Hedgehog) might insert into the outer phospholipid monolayer of such a particle via their attached lipid moieties. This study use biochemical fractionation to determine the sort of particle with which lipid-linked proteins associate, and genetic means to address its function (Panakova, 2005).

Lipid-linked proteins copurify with lipophorin: Sedimentation of Wingless, Hedgehog and gpi-linked proteins were compared to that of transmembrane proteins, exosomes and lipophorin particles. To mark exosomes, flies were used expressing a vertebrate CD63:GFP fusion construct. CD63 is a tetraspanin that localizes to internal vesicles of multivesicular endosomes, and is released on exosomes (Escola, 1998; Wubbolts, 2003). In Drosophila imaginal discs, CD63:GFP localizes to late endosomes in producing cells, consistent with vertebrate studies. It is released and endocytosed by neighbouring cells between one and three cell diameters away, indicating that it is present on exosomes (Panakova, 2005).

To mark lipoprotein particles, antibodies were made to Drosophila apolipophorins I and II (ApoLI and ApoLII); these proteins are generated by cleavage of the precursor pro-Apolipophorin (Sundermeyer, 1996; Kutty, 1996). Lipophorin is produced in the fat body (Kutty, 1996); consistent with this, apolipophorin transcripts cannot be detected in imaginal discs. Nevertheless, the ApoLI and ApoLII proteins are as abundant in discs as in the fat body (Panakova, 2005).

Plasma membrane and exosomal markers are completely pelleted after centrifugation for 3 h at 120,000g, whereas most ApoLII remains in the supernatant. Most Wingless:GFP and Hedgehog is present in the pellet, as are the gpi-linked proteins Fasciclin, Connectin, Klingon and Acetylcholineasterase; this is not unexpected, because these proteins localize to the plasma membrane and internal membrane compartments. Surprisingly, however, some Wingless:GFP (6%), Hedgehog (2%) and gpi-linked proteins (14%-22%) remain in the supernatant (Panakova, 2005).

The 120,000g supernatant (S120) contains both free soluble proteins and lipoprotein particles. To separate them, isopycnic density centrifugation was performed. In these gradients, lipophorin moves to the top low-density fraction whereas soluble proteins are present in higher-density fractions. Gpi-linked proteins are found almost entirely in the top fraction with lipophorin. Treating the S120 with Phosphatidylinositol-specific phospholipase C (PI-PLC) before density centrifugation shifts their migration to higher-density fractions. This suggests that gpi-linked proteins associate with low-density particles via their gpi anchor (Panakova, 2005).

Similarly, when S120s from larvae that express Wingless:GFP or Hedgehog:HA in imaginal discs are subjected to isopycnic density centrifugation, these proteins are found in the lowest-density fraction with ApoLII, as is endogenous Hedgehog. Antibodies to endogenous Wingless detect a doublet in the top fraction and a band of somewhat higher mobility in high-density fractions. These data indicate that non-membrane-bound Wingless and Hedgehog associate with low-density particles in imaginal discs in vivo; other larval tissues may secrete Wingless in a non-lipophorin-associated form (Panakova, 2005).

To ask whether lipid-linked proteins associate with lipophorin, or with some other low-density particle, ApoLII was immunoprecipitated from larval S120s and precipitates were probed for Wingless, Hedgehog or GFPgpi. These proteins are immunoprecipitated by anti-ApoLII. Hedgehog and Fas-1 also immunoprecipitate with ApoLII from the more purified top fraction of KBr gradients. Thus, lipid-linked morphogens and gpi-linked proteins associate directly with lipophorin particles (Panakova, 2005).

Lipophorin-RNAi perturbs lipid transport: To assess the role of lipophorin in larval growth and development, the levels of ApoLI and II were reduced by RNA interference directed against two different regions of the apolipophorin messenger RNA. Similar phenotypes were produced by each construct. To express double-stranded (ds)RNA, a modified GAL4:UAS system was used in which expression of inverted repeats can be temporally controlled by heat-shock-dependent excision of an intervening HcRed cassette by the flippase (FLP) recombinase. Extracts were tested from wild-type larvae or larvae harbouring hs-flp, GAL4 driver and UAS dsRNA constructs at various times after heat shock to see how fast lipophorin levels were reduced. Larvae of the latter genotype made only 50% of the wild-type level of ApoLII, even in the absence of heat shock; basal activity of the heat-shock promoter in the fat body causes HcRed excision in approximately 50% of fat-body cells, although excision strictly depends on heat shock in other larval tissues. Although they survive less frequently, these flies have no obvious phenotype (Panakova, 2005).

After heat shock, all fat-body cells excise the HcRed cassette and ApoLII levels decrease further. After four days, ApoLII is reduced to 5% of wild-type levels. ApoLI levels are reduced with similar kinetics. These animals prolong the third larval instar and rarely pupariate. All the experiments described below were performed on third-instar larvae 4-6 days after heat shock (Panakova, 2005).

To investigate the requirement for lipophorin in lipid transport, the accumulation of neutral lipids in larval tissues was assessed by staining them with Nile Red. Cells of the posterior midgut normally contain many small lipid droplets. Lipophorin reduction causes a dramatic expansion of these droplets, suggesting that lipophorin is required for the efficient extraction of lipid from the midgut (Panakova, 2005).

The wild-type fat body contains both small and large lipid droplets. Fat bodies of lipophorin-RNAi larvae are reduced in size and have fewer small lipid droplets, although larger droplets appeared normal. These data suggest that lipophorin delivers lipid to the fat body (Panakova, 2005).

Lipid droplets in discs from lipophorin-RNAi larvae are fewer and smaller than in the wild type. Their discs are also reduced in size, particularly in the wing pouch. Thus, discs require lipophorin for accumulation of lipid droplets and for growth. Neither Caspase3 activation nor membrane phosphatidylinositol 3,4,5-phosphate (PIP₃) accumulation is altered in lipophorin- RNAi discs , suggesting that their small size is not due to cell death or reduced insulin signalling (Panakova, 2005).

Hedgehog function requires lipophorin: To test whether lipophorin association is required for Hedgehog function, Hedgeghog distribution and signalling was examined in lipophorin-RNAi larval discs. In wild-type discs, Hedgehog expressed in the posterior compartment moves across the anterior-posterior (AP) compartment boundary and activates transcription of short and long-range target genes. Cells closest to the source respond by activating the transcription of collier and patched. Further away, Hedgehog activates transcription of decapentaplegic. Levels of Collier and a decapentaplegic reporter construct (dpplacZ) were monitored in wild-type and lipophorin-RNAi discs stained in parallel and imaged under identical conditions. Discs from lipophorin-RNAi larvae activate collier at least as efficiently as those of the wild type. In contrast, the range of activation of dppLacZ is significantly narrowed in lipophorin RNAi discs. dppLacZ is expressed up to 11 cells away from the AP boundary in wild-type discs, but only up to six cells away in lipophorin-RNAi larvae. These data suggest that lipophorin knockdown decreases the range of Hedgehog signalling (Panakova, 2005).

To discover whether Hedgehog trafficking was altered, discs were stained for Hedgehog and Patched. In wild-type discs, Hedgehog moves into the anterior compartment, where it is found in endosomes, often with Patched. Patched-mediated endocytosis is thought to sequester Hedgehog and limit its spread. Hedgehog is most abundant up to five cell rows away from the AP boundary; although Hedgehog signals over a wider range, specific staining there cannot be distinguished from background. In lipophorin-RNAi discs, Hedgehog accumulates to abnormally high levels in the first five rows of anterior cells. 380 Hedgehog spots were found in the most apical 10 µm of the wild-type disc. The lipophorin-RNAi disc contained 1,208 Hedgehog spots in the same region. Most accumulated Hedgehog colocalizes with Patched in endosomes. Furthermore, Patched co-accumulates more extensively with Hedgehog in endosomes than it does in wild-type. These data indicate that lipophorin RNAi either increases the susceptibility of Hedgehog to Patched-mediated endocytosis, or prevents subsequent degradation of the protein (Panakova, 2005).

Drosophila cannot synthesize sterols and relies on dietary sources. To assess whether reduced uptake of sterols or other lipids might cause the changes seen, the effects of lipid deprivation on larval development were explored. Larvae were allowed to hatch and feed on sucrose/agarose plates supplemented with yeast for 2-3 days, then transferred to plates containing chloroform-extracted yeast autolysate, rather than yeast. These larvae are developmentally delayed; after 7 days of lipid deprivation, their discs are much smaller than those of younger late-third-instar larvae. In contrast, yeast-fed siblings pupariate and begin to eclose by this time. Those flies that infrequently eclose after larval lipid depletion are small (35%-60% of normal body weight) but normally patterned. Thus, lipid depletion stalls imaginal growth (Panakova, 2005).

To discover whether lipid starvation affected Hedgehog trafficking or signalling, larvae were deprived of lipid 2 days after hatching and their discs were stained 6 days later. No changes in Hedgehog or Patched distribution are apparent in these discs compared with younger yeast-fed discs of similar size. Furthermore, the range of dpp and collier expression does not differ in lipid-starved and yeast-fed discs. Thus, lipid starvation does not mimic the effects of lipophorin knockdown. It is speculated that lipid-starvation-induced growth arrest prevents membrane sterol from dropping to levels that would interfere with the Hedgehog pathway. Thus, lipophorin does not indirectly affect the Hedgehog pathway via lipid deprivation (Panakova, 2005).

Wingless function requires lipophorin: To discover whether lipophorin RNAi perturbed Wingless trafficking, Wingless distribution was examined. In lipophorin- RNAi discs, extracellular Wingless is less abundant on both the apical and basolateral epithelial surfaces and spreads over shorter distances. However, no consistent alterations were detected in intracellular Wingless. Thus, lipophorin promotes accumulation of extracellular Wingless (Panakova, 2005).

To investigate whether Wingless signalling requires lipophorin, the activation of two target genes was examined. Senseless is produced only in cells near the Wingless source and its expression is unaffected by lipophorin RNAi. Distalless is normally produced in a gradient throughout most of the wing pouch. In lipophorin-RNAi discs, the Distalless gradient is abnormally narrow. This suggests that lipophorin knockdown specifically perturbs long-range Wingless signalling (Panakova, 2005).

Conclusions: This study establishes the principle that lipid-linked proteins of the exoplasmic face of the membrane associate with lipoproteins. These include many gpi-linked proteins with diverse functions, as well as the lipid-linked morphogens Wingless and Hedgehog. The mechanism allowing long-range dispersal of lipid-linked proteins is not yet understood. The finding that these proteins exist in both membrane-associated and lipoprotein-associated forms suggests reversible binding to lipoprotein particles as a plausible mechanism for intercellular transfer, and the consequences of lowering lipoprotein levels in Drosophila larvae supports this idea (Panakova, 2005).

Lipophorin knockdown narrows the range of both Wingless and Hedgehog signalling. Hedgehog accumulates to an abnormally high level in cells near the source of production and long-range signalling is inhibited; short-range target genes, however, are expressed normally. These data suggest that Hedgehog does not move as far when lipophorin levels are low. The range over which Hedgehog moves is normally restricted by Patched-mediated endocytosis. In discs from lipophorin RNAi larvae, accumulated Hedgehog co-localizes with Patched in endosomes, suggesting that it is more efficiently sequestered by Patched. How might lipophorin antagonize Patched-mediated sequestration and promote long-range movement (Panakova, 2005)?

The data are consistent with the idea that lipophorin is continuously needed for movement, rather than required only for the release of morphogens. If lipophorin were important only for Hedgehog secretion, lipophorin RNAi would be expected to decrease the amount of Hedgehog found in receiving tissue; this seems not to be the case. Furthermore, altered Hedgehog trafficking in receiving tissue is consistent with a model in which lipophorin is required at each step of intercellular transfer. The idea is favored that reversible association of Hedgehog with lipophorin particles facilitates its transfer from the plasma membrane of one cell to that of the next. This model predicts that lowering lipophorin levels should increase the length of time that Hedgehog spends in the plasma membrane before becoming associated with lipophorin. This would slow its rate of transfer and increase the probability of Patched endocytosing Hedgehog before it moved to the next cell. Hedgehog would then signal efficiently in the short range, but be so efficiently sequestered by Patched that very little protein would travel far enough to activate long-range target genes. These predictions are completely consistent with the current observations (Panakova, 2005).

This model differs significantly from the original concept of argosome function. It was initially speculated that argosomes were exosome-like particles with an intact membrane bilayer, and that lipid-linked morphogens needed to be assembled on these particles to be secreted by producing cells. Instead, it was found that argosomes are exogenously derived lipoproteins that facilitate the movement of morphogens through the epithelium. Many questions remain as to how morphogens become associated with argosomes, and how the spread and cell-interactions of these particles are regulated. Clearly, heparan sulphate proteoglycans are essential for the movement of Hedgehog and Wingless into receiving tissue. Because heparan sulphate binds to vertebrate lipoprotein particles, one might speculate that heparan sulphate proteoglygans (HSPGs) facilitate morphogen movement through lipoprotein binding. Conversely, many gpi-linked proteins, including the HSPG's Dally and Dally-like, are found on lipoprotein particles themselves. These associated proteins have the potential to modulate the cellular affinities or trafficking properties of lipoproteins and the morphogens they carry (Panakova, 2005).

The data suggest that lipophorin particles not only mediate intercellular transfer of Hedgehog, but may also be endocytosed together with the morphogen. Interestingly, LDL-receptor-related proteins Arrow and Megalin have demonstrated roles in Wingless signalling and Hedgehog endocytosis, respectively. It is intriguing to speculate that these receptors might be important for interaction with the lipoprotein-associated form of the morphogen (Panakova, 2005).

Cholesterol has the potential to modulate the activity of the Hedgehog pathway at many different points. Whether changes in the level of cellular cholesterol normally play a role in regulating the activity of the pathway is unclear. This study shows that Hedgehog interacts with the particle that delivers sterol to cells. This observation raises the possibility that internalization of Hedgehog is linked to sterol uptake, and suggests new mechanisms to link nutrition, growth and signalling during development (Panakova, 2005).