Abstract
The Hedgehog signaling pathway plays a crucial role in development and disease. Its putative origins in an ancient system involved in regulating bacterial lipid transport and homeostasis offers clues about how the pathway might work today.
Complex body plans require sophisticated cell–cell signaling pathways. How these pathways evolved is often not very well understood. Here, we argue that the Hedgehog (Hh) signaling pathway may have arisen from systems that were originally designed for the transport and homeostasis of certain bacterial sterol analogs—the hopanoids. We propose a possible scenario for the evolution of Hh signaling and discuss how evolutionary considerations can shed light on the mysterious communication between the membrane-bound Hh transducers Patched (Ptc) and Smoothened (Smo).
The importance of the Hh signaling pathway has long been documented, even in Greek mythology; the tale of the one-eyed Cyclops was likely inspired by rare birth defects related to reduced Hh signaling [1],[2]. Whereas reduced Hh signaling can cause these and other developmental defects, inappropriate activation of Hh signaling contributes to certain forms of cancer, including basal cell carcinoma, the most commonly occurring form of skin cancer [3]. Both effects reflect the essential role of Hh in the control of patterning and growth during development and in the adult animal (for more detailed reviews, see [4]–[6]).
The current model for the production, transport, and transmission of the Hh signal is summarized in Figure 1. Two important components acting at the cell membrane are Ptc, which is the likely receptor for Hh, and the seven-pass transmembrane protein Smo, which acts downstream of Ptc. In the absence of Hh, Ptc blocks Smo activity. How this repression is achieved, and subsequently overcome by the Hh ligand, remains a mystery. One key to solving it may lie in understanding the multiple connections of the Hh pathway to lipid metabolism (reviewed in [7]). In addition, we argue that evolutionary considerations can identify a possible scenario for the origin of Hh signaling. To begin addressing both aspects, we start with a detailed look at Ptc and Smo.
Evidence for Ptc-Like Proteins in Bacteria
Ptc proteins are members of a large superfamily that includes bacterial and archaeal resistance-nodulation division (RND) transporters [8] (Figure 2). RND transporters are proton antiporters that catalyze the active transmembrane efflux of numerous substrates from the cell ([9]; proton antiporters, or “counter-transporters,” use the physiological proton gradient at the membrane to pump out their substrates in exchange for protons that are allowed to flow inside). Ptc-family proteins have 12 transmembrane segments and have been described as hybrids of an RND-derived domain and a second domain, the so-called “sterol-sensing domain” (SSD) [10],[11]. However, our re-analysis of this family suggests that this classification of two distinct protein domains is somewhat arbitrary. Most eukaryotic and prokaryotic members of the family can be aligned over their entire length (Figures S1 and S2), with two well-conserved central blocks—of five transmembrane regions each—clearly visible. While these blocks roughly correspond to the suggested SSD and RND domains, we propose that they most likely stem from an ancient, internal duplication within the gene (this is also evident from the internal symmetry in the three-dimensional RND protein structure [12]). Some eukaryotic proteins in the superfamily have lost one of these two halves (the C-terminal half), and the remaining N-terminal half has been shown to have a role in sensing sterols [10]. For this reason, the first half of the protein can indeed be defined as a SSD, but it is as much of RND origin as is the second half. Other full-length members of the Ptc superfamily in eukaryotes (Figure 2) include Niemann-Pick C1 protein (NPC1) and Dispatched (Disp) [10]. The former is thought to be involved in cholesterol homeostasis, whereas the latter acts in Hh signaling to facilitate the release of the cholesterol-modified Hh protein from Hh-secreting cells (Figure 1). Taken together, a parsimonious interpretation of the available sequence data is that all current Ptc-family proteins are indeed of ancient origin over their entire length, and that they represent the oldest traceable components of Hh signaling.
Might eukaryotic Ptc-family members still act as proton antiporters today (we think so) and, in that case, what is their substrate? In bacteria, the actual substrates of RND transporters are often not known, and they are generally assumed to be quite diverse. However, one particular RND subfamily, having unusually high sequence similarity to eukaryotic Ptc-family proteins, has been tentatively linked to an intriguing substrate: hopanoids (Figure 2A–2C).
Hopanoids are the structural and functional analogs of sterols and, like sterols, are synthesized from squalene precursors. Accordingly, members of this particular subfamily of RND transporters have been termed “hopanoid biosynthesis-associated RND transporters,” or “hpnN” [13]. Further strengthening this link is the observation that hpnN genes can sometimes be found immediately adjacent to hopanoid biosynthesis genes, presumably forming co-transcribed transcription units (operons; Figure 2D and 2E). Remarkably, most of these hpnN-type RND genes have a reciprocal–best-match relationship with Ptc-family proteins in eukaryotes (Figure S3), i.e., they show up with the highest sequence similarity in homology searches extending across the eukaryote/prokaryote division. This makes the hpnN-type transporters the best candidates for functional counterparts of Ptc-like proteins outside eukaryotes. While this observation of course does not establish any current or past substrate(s) for the eukaryotic proteins, it does imply that structural analogs of sterols can be substrates. From this, a prediction might be that transporting/sensing such molecules (or rather, cyclic terpenoids in general) might still be the conserved function of eukaryotic Ptc-family proteins.
The Origins of Smo
In contrast to Ptc, Smo-like proteins cannot be traced beyond the eukaryotes. Smo is a member of the eukaryote-specific superfamily G-protein–coupled receptors (GPCRs) [14]; ligands for this superfamily are typically small molecules. Given that Ptc might act by transporting/extruding a lipophilic molecule, could Smo's ligand be a lipophilic molecule as well? The majority of described GPCRs do not have strongly lipophilic ligands, but some examples exist—for example, the opsin/retinal pair in the eye [15] and the leukotriene receptors [16]. Of note with respect to Hh signaling is that some GPCRs can even sense steroids, for example, the putative estradiol receptor GPR30 [17],[18].
As Smo seems to be a much more recent innovation than Ptc, what conceivable chain of events might have connected Smo to Ptc activity and led to Hh signaling as it appears today? Below, we present a possible, minimal evolutionary path that could have generated today's setup, including an explanation for the observed unusual double-inhibition in the pathway, in which Hh acts by inhibiting Ptc, and Ptc in turn functions by inhibiting Smo.
A Possible Evolutionary Scenario
We assume that the original function of Ptc was simply to transport an unwanted lipid molecule out of the cell. Smo, on the other hand, derives from a protein family whose main function is to sense and to transduce extracellular signals (i.e., the GPCR family). Therefore, we propose the following scenario: let us imagine that, in primitive eukaryotes, Smo was initially a receptor sensing lipid molecules and was acting upstream of the primitive Ptc transporter (Figure 3). The two molecules would have formed a simple homeostasis system; Smo would sense the abundance of a certain lipid and would transcriptionally induce Ptc whenever this lipid was in excess and needed to be removed from the membrane (i.e., pumped away). We propose that when multicellular organisms arose, this system was available and was recruited for a new purpose: cell-to-cell signaling.
There is evidence indicating that this repurposing could have been made possible via the fortuitous action of an intein. Inteins are a class of protein-coding genes that can insert themselves into other genes, and “splice” themselves out again after translation, joining the two fragments of the host protein together to restore its function. In the case of the Hh protein (whose C-terminal half encodes such an intein [19]), let us suppose that one of the two fragments to be joined happened to be a lipid moiety instead of a protein. The result would have been a lipid with a nonspecific, bulky protein domain (in this case, the future Hh N-terminal domain referred to as HhN) attached to it, and while that lipid would still have been a substrate for Ptc, the protein domain might now have blocked the transport mechanism, sterically, and prevented Ptc from pumping its normal substrate.
A consequence of this would have been that cells were now provided with a way to control the activity of the Ptc/Smo system in a neighboring cell. Simply by expressing and secreting Hh, they could sterically block Ptc in the neighboring cell. In other words, Ptc would sense and bind the cholesterol moiety of the secreted Hh (mistaking it for its ligand), but this binding would inhibit its exporter function due to the bulky protein attached to the cholesterol. This would in turn lead to the accumulation of the non-modified, normal ligand for Smo, because Ptc would no longer be available for depleting it. Importantly, this does not require any specific protein interface between Hh and Ptc (at least not initially). In the receiving cell, the only change that would have been necessary is the gradual addition of new target genes to the existing Smo pathway, in order to exploit this fortuitous new intercell “communication channel.” It should be noted that, from the outset, the new system would have been capable of graded (quantitative) responses; the original homeostasis system already had been capable of sensing different levels of the lipid, and the new ligand Hh would also be able to interfere to various extents, depending on the amounts produced and secreted.
Predictions of This Evolving Model and How To Test Them
Our model is parsimonious, since only a few simple steps are required, and at each step a selectable advantage for the organism is conceivable. However, the model is also speculative—how can it be tested?
First, we propose that Ptc-family proteins are not only derived from RND transporters, but actually still work through the RND molecular mechanism. RND transporters are thought to function as trimers (three identical proteins are needed to form a functional unit in the membrane), and this arrangement is actually essential for the proposed molecular mechanism of pumping [20]. This is consistent with the proposition that catalytically active forms of Ptc are trimeric as well [21] and suggests they also require a proton gradient. Our model would predict that any prevention of trimerization should completely abolish repression of Smo by Ptc, while Hh binding might still be possible.
Second, we propose that at least some RNDs should be sterically inhibited by the covalent addition of bulky protein domains to their substrates; we think this is the case, because the substrate is generally thought to move “through” the RND and is extruded via an inner channel. This could be tested by adding bulky domains to RND substrates in bacteria or by replacing much of Hh with an unrelated but similarly sized protein domain which might still show (partial) activity in Hh signaling.
Third, we predict that the binding mode of Hh to Ptc was originally such that the cholesterol moiety would sterically fit into the original RND substrate pocket, and this might still be the case today. This could be tested by co-crystallizing an RND/hopanoid pair, as well as a Hh/Ptc pair, and comparing their binding modes. However, it is worth noting that, today, the cholesterol modification of Hh is no longer essential, so we assume that Ptc and Hh have evolved a specific protein interface in the meantime.
Fourth, our model makes predictions with respect to the surprising observation that Caenorhabditis elegans, while still having a clear ortholog of Ptc (Figure 2), lacks a homolog of Smo. As such, nematodes seem to represent an incomplete state of the pathway, and it might be tempting to try to use them to reconstitute aspects of signaling in organisms such as Drosophila. Would the C. elegans Ptc molecule, when introduced into Drosophila, show any effect on Drosophila Smo? From our model, we would predict that it might indeed be able to repress Smo (since we propose that it transports a ubiquitous lipid), but importantly, it should not specifically bind Hh—at least not when Hh is devoid of cholesterol.
Fifth, our model posits that Ptc works by transporting an agonist away from Smo (as opposed to transporting an “antagonist” towards Smo). This predicts that the action of Ptc on Smo is strictly cell-autonomous, and that Ptc and Smo need to co-localize for Ptc to function, at least in all those subcellular structures where Smo would otherwise be competent to signal.
Lastly, the communication between Ptc and Smo would happen via locally restricted lipid/sterol level fluctuations; we are essentially proposing that Ptc creates a local depletion of a specific membrane component. Perhaps this could be tested using lipidomics approaches on fractionated membrane samples, searching for changes in lipid abundances in response to Hh signaling or in response to Ptc overexpression. For this, interesting subcellular compartments would include lipid rafts, vertebrate cilia [22], or the late endosomes where Ptc and Smo co-localize [23]. It seems generally imperative that the location, movement, and site of action of Ptc and Smo are mapped with as much detail as possible, using the endogenous proteins at normal levels.
Conclusions
The intriguing homology between components of lipid homeostasis pathways and components of the Hh signaling pathway leads to the hypothesis that the central membrane–players of the Hh signaling cascade—Smo and Ptc—evolved from a pre-existing lipid-sensing/homeostasis pathway. We propose a model of simple evolutionary steps, which posits that Ptc acts by pumping an activator of Smo, rather than an inhibitor. This scenario is compatible with most experimental data so far. The step-wise construction of pathways from older, pre-existing modules is turning out to be a general theme in developmental biology [24]. The simple evolutionary model we propose here may be a good starting point, but of course the evolution of the pathway could easily have been much less straightforward, taking leaps and bounds that we have not envisioned. To generally think more along evolutionary lines may nevertheless help to explain some of the more “exotic” findings—here and elsewhere in developmental biology.
Supporting Information
Abbreviations
- Disp
Dispatched
- GPCR
G-protein–coupled receptor
- Hh
Hedgehog
- hpnN
hopanoid biosynthesis-associated RND transporters
- Ptc
Patched
- RND
resistance-nodulation division
- Smo
Smoothened
- SSD
sterol-sensing domain
References
- 1.Cooper MK, Porter JA, Young KE, Beachy PA. Teratogen-mediated inhibition of target tissue response to Shh signaling. Science. 1998;280:1603–1607. doi: 10.1126/science.280.5369.1603. [DOI] [PubMed] [Google Scholar]
- 2.Chiang C, Litingtung Y, Lee E, Young KE, Corden JL, et al. Cyclopia and defective axial patterning in mice lacking Sonic hedgehog gene function. Nature. 1996;383:407–413. doi: 10.1038/383407a0. [DOI] [PubMed] [Google Scholar]
- 3.Pasca di Magliano M, Hebrok M. Hedgehog signalling in cancer formation and maintenance. Nat Rev Cancer. 2003;3:903–911. doi: 10.1038/nrc1229. [DOI] [PubMed] [Google Scholar]
- 4.McMahon AP, Ingham PW, Tabin CJ. Developmental roles and clinical significance of hedgehog signaling. Curr Top Dev Biol. 2003;53:1–114. doi: 10.1016/s0070-2153(03)53002-2. [DOI] [PubMed] [Google Scholar]
- 5.Varjosalo M, Taipale J. Hedgehog: functions and mechanisms. Genes Dev. 2008;22:2454–2472. doi: 10.1101/gad.1693608. [DOI] [PubMed] [Google Scholar]
- 6.Jiang J, Hui CC. Hedgehog signaling in development and cancer. Dev Cell. 2008;15:801–812. doi: 10.1016/j.devcel.2008.11.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Eaton S. Multiple roles for lipids in the Hedgehog signalling pathway. Nat Rev Mol Cell Biol. 2008;9:437–445. doi: 10.1038/nrm2414. [DOI] [PubMed] [Google Scholar]
- 8.Taipale J, Cooper MK, Maiti T, Beachy PA. Patched acts catalytically to suppress the activity of Smoothened. Nature. 2002;418:892–897. doi: 10.1038/nature00989. [DOI] [PubMed] [Google Scholar]
- 9.Tseng TT, Gratwick KS, Kollman J, Park D, Nies DH, et al. The RND permease superfamily: an ancient, ubiquitous and diverse family that includes human disease and development proteins. J Mol Microbiol Biotechnol. 1999;1:107–125. [PubMed] [Google Scholar]
- 10.Kuwabara PE, Labouesse M. The sterol-sensing domain: multiple families, a unique role? Trends Genet. 2002;18:193–201. doi: 10.1016/s0168-9525(02)02640-9. [DOI] [PubMed] [Google Scholar]
- 11.Incardona JP. From sensing cellular sterols to assembling sensory structures. Dev Cell. 2005;8:798–799. doi: 10.1016/j.devcel.2005.05.011. [DOI] [PubMed] [Google Scholar]
- 12.Murakami S, Nakashima R, Yamashita E, Yamaguchi A. Crystal structure of bacterial multidrug efflux transporter AcrB. Nature. 2002;419:587–593. doi: 10.1038/nature01050. [DOI] [PubMed] [Google Scholar]
- 13.Selengut JD, Haft DH, Davidsen T, Ganapathy A, Gwinn-Giglio M, et al. TIGRFAMs and Genome Properties: tools for the assignment of molecular function and biological process in prokaryotic genomes. Nucleic Acids Res. 2007;35:D260–264. doi: 10.1093/nar/gkl1043. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ruiz-Gomez A, Molnar C, Holguin H, Mayor F, Jr, de Celis JF. The cell biology of Smo signalling and its relationships with GPCRs. Biochim Biophys Acta. 2007;1768:901–912. doi: 10.1016/j.bbamem.2006.09.020. [DOI] [PubMed] [Google Scholar]
- 15.Filipek S, Stenkamp RE, Teller DC, Palczewski K. G protein-coupled receptor rhodopsin: a prospectus. Annu Rev Physiol. 2003;65:851–879. doi: 10.1146/annurev.physiol.65.092101.142611. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Yokomizo T, Izumi T, Chang K, Takuwa Y, Shimizu T. A G-protein-coupled receptor for leukotriene B4 that mediates chemotaxis. Nature. 1997;387:620–624. doi: 10.1038/42506. [DOI] [PubMed] [Google Scholar]
- 17.Revankar CM, Cimino DF, Sklar LA, Arterburn JB, Prossnitz ER. A transmembrane intracellular estrogen receptor mediates rapid cell signaling. Science. 2005;307:1625–1630. doi: 10.1126/science.1106943. [DOI] [PubMed] [Google Scholar]
- 18.Martensson UE, Salehi SA, Windahl S, Gomez MF, Sward K, et al. Deletion of the G protein-coupled receptor GPR30 impairs glucose tolerance, reduces bone growth, increases blood pressure, and eliminates estradiol-stimulated insulin release in female mice. Endocrinology. 2008;150:687–698. doi: 10.1210/en.2008-0623. [DOI] [PubMed] [Google Scholar]
- 19.Hall TM, Porter JA, Young KE, Koonin EV, Beachy PA, et al. Crystal structure of a Hedgehog autoprocessing domain: homology between Hedgehog and self-splicing proteins. Cell. 1997;91:85–97. doi: 10.1016/s0092-8674(01)80011-8. [DOI] [PubMed] [Google Scholar]
- 20.Murakami S. Multidrug efflux transporter, AcrB—the pumping mechanism. Curr Opin Struct Biol. 2008;18:459–465. doi: 10.1016/j.sbi.2008.06.007. [DOI] [PubMed] [Google Scholar]
- 21.Lu X, Liu S, Kornberg TB. The C-terminal tail of the Hedgehog receptor Patched regulates both localization and turnover. Genes Dev. 2006;20:2539–2551. doi: 10.1101/gad.1461306. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Rohatgi R, Milenkovic L, Scott MP. Patched1 regulates hedgehog signaling at the primary cilium. Science. 2007;317:372–376. doi: 10.1126/science.1139740. [DOI] [PubMed] [Google Scholar]
- 23.Incardona JP, Gruenberg J, Roelink H. Sonic hedgehog induces the segregation of patched and smoothened in endosomes. Curr Biol. 2002;12:983–995. doi: 10.1016/s0960-9822(02)00895-3. [DOI] [PubMed] [Google Scholar]
- 24.Pires-daSilva A, Sommer RJ. The evolution of signalling pathways in animal development. Nat Rev Genet. 2003;4:39–49. doi: 10.1038/nrg977. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.