Cholesterol-free and cholesterol-bound Hedgehog

Abstract

Hedgehog (Hh) is a signaling ligand conserved from flies to humans that is covalently bound to both palmitate and cholesterol moieties. These lipid modifications are crucial for Hh signaling. A recent article reports that in both flies and human-cultured cells a cholesterol-free form of Hh (SHh-N*/Hh-N*) is produced and secreted. In the Drosophila wing disc, Hh associated with Lipoproteins-lipophorin complexes (Lpp) would lead to the accumulation of Cubitus interruptus (Ci), the transcription factor in the Hh pathway but this would be insufficient to activate Hh target genes. On the other hand, Lpp-free Hh-N* would act in synergy with Lpp-associated Hh to eventually activate target gene expression. This suggests that Hh can be secreted in 2 different forms that would have distinct and synergic functions.

The Hedgehog (Hh) gene family encodes secreted ligands that regulate patterning in both vertebrates and invertebrates. Hh graded distribution is tightly controlled to ensure correct patterning. Hh is a double lipid-modified molecule with addition of palmitate at its N-terminus and cholesterol at its C-terminus (for review, see ref. 1). In both flies and vertebrates, Hh palmitoylation is required for Hh secretion.^2-4Cholesterol addition at its C-terminus^5,6 is crucial for Hh signaling range, but its role remains controversial: it has been reported in Drosophila to either increase or decrease Hh range of action (for review, see ref. 7). In the Drosophila wing disc, lipid modifications would enable Hh to reach target cells at a distance. The Eaton laboratory proposed that Hh would be boarded on lipoprotein-lipophorin complex (Lpp) that originate from the fat body. Hh-Lpp would move across the tissue, thus allowing Hh long-range signaling.^8,9 These Lpp also contains lipids that would repress the Hh pathway in the absence of Hh.¹⁰ Palm et al. raised the possibility that in the wing disc Hh can be secreted in 2 forms that would have complementary functions and would synergize to activate Hh target genes.¹¹ First, cholesterol-modified Hh would be secreted in a Lpp-associated manner and would lead to the accumulation of the transcription factor Ci in a full-length, but inactive form. Second, cholesterol-free Hh (Hh-N*) would be secreted in a Lpp-free manner and would not be able to activate Hh target genes by itself either. However, when both Hh and Hh-N* are present together, they could act in synergy to eventually trigger Hh target gene expression.

Palm et al. first showed that Hh proteins can be secreted in a lipoparticle-associatedmanner but also in a lipoprotein-free manner when overexpressed in both human cultured cells and in the Drosophila hemolymph. The lipoparticle-associated SHh/Hh secretion would require either palmitate or cholesterol since cholesterol-free and palmitate-free SHh/Hh can still associate with lipoparticles. Conversely, a form of SHh/Hh that lacks both palmitate and cholesterol cannot associate with lipoparticles. Therefore, any lipid moiety is sufficient to promote Hh association to a broad diversity of lipoparticles.

The authors next investigated the molecular characteristics of lipoparticle-free Hh. They showed that Lpp-free Hh is cholesterol-free. Indeed, when Drosophila S2 cells are transfected with ³H-cholesterol, the soluble pool of Hh that is Lpp-free is also free from cholesterol. It is worth noting that reducing Lpp levels leads to an accumulation of Lpp-free Hh. Since the authors suggest that Lpp-free Hh is cholesterol-free, it implies that one of the pleiotropic effects of Lpp knockdown is to somehow promote the formation of cholesterol-free Hh, which should be further investigated.

Next, Palm et al. analyzed whether cholesterol-free Hh is produced in vivo. First, reducing Lpp levels in the disc leads to a slight decrease of Patched expression range. Palm et al. hypothesized that the secretion of a cholesterol-free form of Hh would account for the remaining Patched expression. Using Triton X-114 phase separation, they showed that the soluble Hh pool is detected exclusively in the aqueous phase, suggesting that this pool of Hh is cholesterol-free (Hh-N*). This contradicts pioneer studies that used the Triton X-114 protocol and showed that cholesterol-free Hh is detected both in the aqueous and the detergent phase.⁵ Therefore further biochemical evidence is required to definitely prove the in vivo existence of Hh-N*.

The authors then investigated the signaling properties of Lpp-associated Hh and Hh-N* in the wing disc. They found that by overexpressing Hh in the hemolymph, ectopic Hh isdetected in the anterior compartment of wing discs. In these discs, Ci accumulates in its full-length form, which is the one able to activate Hh target genes. Surprisingly, the Hh target genes are not expressed, which suggests that Lpp-associated Hh promotes the stabilization of full-length Ci in an inactive form. To investigate whether Hh-N* could have additional effects on Hh target gene expression when co-expressed with Hh, Palm et al. overexpressed Hh in the fat body in Lpp-RNAi larvae. With these settings, Hh-N* is produced at moderate levels with some Lpp-associated Hh in the hemolymph. Palm et al. suggest that Hh-N* can activate target genes when full-length Ci is stabilized by remaining Lpp-Hh as an anterior overgrowth is observed in these discs. However, one can notice that the expression range of the Hh target Dpp-lacZ appears to be unaffected, although an extra Dpp-lacZ stripe perpendicular to the endogenous one is detected. The origin of this extra-stripe is not discussed but still can account for the anterior compartment overgrowth observed in these discs. Besides, the analysis of the other Hh targets is required to confirm that Hh-N* with remaining Hh-Lpp can activate the transcription of Hh target genes. Still, to confirm that Hh-N* can synergize with Hh-Lpp to activate Hh target genes, Palm et al. expressed in the fat body either Hh, Hh-N (cholesterol-free Hh variant genetically engineered) or both and subsequently analyzed Hh target gene expression in the wing disc. Strikingly, combination of low levels of Hh-N and Hh-WT in the fat body leads to a broad expression of Collier and Engrailed in the anterior compartment, suggesting that Hh and Hh-N can synergize to activate Hh targets. However, moderate levels of Hh-N in the fat body leads to a similar broad expression of Collier and Engrailed. Thus, Hh-N by itself is capable to induce target genes at a distance, which contradicts the model of Palm et al. in which Hh-N* alone is not able to induce Hh target genes. Still, the authors propose that the Hh-N genetically engineered would have the feature of both Lpp-associated Hh and Hh-N*, thus explaining why Hh-N is able to induce Hh target genes. However, a much simpler hypothesis is that cholesterol-free Hh can signal by itself at a long-range independently of endogenous Hh, as reported in several instances.^12-14 Besides, Palm et al. proposed that Hh-N* and Hh would have complementary functions as Hh-N overexpression decreases the amount of cleaved-Ci without changing the amount of full-length Ci. However, since the amount of cleaved-Ci is not documented for Hh-N* and since the authors report that Hh-N may have different features from Hh-N*, it is hard to state that Hh-N* and Hh have complementary functions.

Altogether, Palm et al. brought evidence that both mammals and flies can release Lpp-associated Hh and Lpp-free Hh-N*. Particularly, Palm et al. raised the possibility that wing imaginal discs produce Hh-N*. These data support a model in which Hh-N* and Hh could synergize to activate Hh target genes in the wing disc (Fig. 1). However, further investigations are required to confirm the consistency of this model. Most of all, it is crucial to understand how two distinct Hh variants that bind to the same receptor Patched could trigger differential responses. Also, biochemical and genetic experiments are needed to understand how cholesterol-free Hh is produced. Indeed, since cholesterol is required for the correct processing of Hh, it implies that a putative esterase should remove the cholesterol moiety to generate a pool of Hh-N*.

Figure 1. Proposed model of Hh and Hh-N* secretions and actions in the Drosophila wing imaginal disc. The left cell is producing Hh, the right one is receiving Hh through its receptor Patched (purple). Hh is first covalently bound to cholesterol and palmitate moieties (1). Processed Hh would be secreted in a Lpp-associated manner (Lpp-Hh, red lines) (2). A putative unknown esterase would lead to the formation of a cholesterol-free pool of Hh that is secreted in a Lpp-free manner (Hh-N*, blue lines) (3). Lpp-Hh would stabilize the inactive form of full-length Ci (4). Hh-N* would decrease the amount of cleaved Ci (5), but importantly would promote the switch from inactive to active full-length Ci (6). Thus, when present together, Hh and Hh-N* would act in synergy to eventually trigger the transcription of the Hh target genes (7).

Still, a crux of the different current models is that cholesterol would enable a long-range signaling of Hh, although Hh activates its targets at a shorter distance compared with other ligands such as Decapentaplegic or Wingless. Based on this assumption a number of mechanisms have been proposed. A first model is that Hh could board on cytonemes originating from receiving cells.¹⁵ Alternatively, Hh would be secreted apically and then released basolaterally with a complex choreography to board cytonemes that originate from the Hhproducing cells.^16,17 A third model proposed by Palm et al. is that both cholesterol-bound Hh and cholesterol-free Hh could be secreted, and act in synergy although the putative mechanism that generate cholesterol-free Hh is unknown.¹¹ Thus, how can we reconcile all these mutually exclusive mechanisms to explain how Hh gradient is generated in the wing disc? A clue may come from a simple but still instructive experiment: overexpressing cholesterol-free Hh leads to ectopic activation of Hh targets at a long-range that is never encountered in vivo.^12,14 Furthermore, it has been shown that cholesterol-free Hh induces a long-range plateau of Hh targets, suggesting that the cholesterol adduct is required for the establishment of a short-range Hh gradient.¹⁴ Therefore, the role of the cholesterol moiety is to ensure a short-range Hh spread, rather than enabling a “long-range” signaling. Most importantly, one should conclude that the mechanism that truly accounts for Hh short-range gradient formation should specifically involve the cholesterol moiety.

Palm W, Swierczynska MM, Kumari V, Ehrhart-Bornstein M, Bornstein SR, Eaton S. Secretion and signaling activities of lipoprotein-associated hedgehog and non-sterol-modified hedgehog in flies and mammals. PLoS Biol. 2013;11:e1001505. doi: 10.1371/journal.pbio.1001505.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.