BIOLOGICAL OVERVIEW

The recent confirmation of frizzled2 as the long sought Wingless receptor is illustrative of the detective work flavor of developmental biology. The discovery of a large family of transmembrane receptors homologous to the Drosophila frizzled gene left a large hole in receptor research, just waiting to be filled. In fact, the only family of ligands known to be as large as the frizzled family, for which no receptors had been identified, is the Wnt family (Wang, 1996), for which the Drosophila signaling protein Wingless is the prototype. This was circumstantial evidence, but only that, for assigning a frizzled family member a role in Wingless reception.

The second piece of circumstantial evidence that frizzled family members serve as Wingless receptors was the discovery that dishevelled is required for the proper functioning of both the wingless and frizzled pathways (Krasnow, 1995). Dishevelled, a phosphoprotein, is synthesized uniformly throughout the embryo, but its level of phosphorylation varies with the level of wingless activity (Yanagawa, 1995). Genetics experiments had already discerned a biochemical hierarchy of protein action in response to wingless signals. porcupine and dishevelled act upstream, prior to the action of shaggy/zeste white 3, while armadillo is required downstream of shaggy. In fact Porcupine acts in cells secreting Wingless, while Dishevelled acts in cells that receive wingless signals (Siegfried, 1994).

Wingless signals function to inactivate Shaggy, the glycogen synthase kinase 3 also known as Zeste white 3 that phosphorylates Armadillo protein. In response to WG protein, the cytoplasmic pool of Armadillo protein increases, whereas the membrane-bound form of ARM is less affected. ARM protein is present in two different forms, with two different patterns of migration, a result of differences in phosphorylation. Both forms are associated almost exclusively with the membrane fraction. In cells that have been exposed to WG protein, a dramatic increase in the faster migrating (underphosphorylated) form of ARM is seen in the soluble cytoplasmic fraction and, to a lesser extent, in the membrane fraction. The slower migrating (phosphorylated) form of ARM, detectable only in the membrane fraction, is not affected by WG (Yanagawa, 1995).

With all the evidence pointing to frizzled family members as Wingless receptors, the search began in earnest for Frizzled homologs in Drosophila, and indirectly, for the putative Wingless receptor. fz2 was cloned on the basis of its homology to the frizzled family. Experiments then ensued to find out whether it could in fact act biologically as a Wingless receptor.

Schneider cultured cells (also known as S2 cells) fail respond to Wingless protein, that is they fail to stabilize ARM in response to added WG. It is possible that S2 cells don't respond because they do not express fz2. To test this possibility, a plasmid carrying the fz2 gene was transfected into S2 cells. When fz2-transfected cells are incubated in the presence of WG, the level of faster migrating (hypophosphorylated) ARM increases. In addition, transfected S2 cells show strong surface staining when incubated with WG and anti-WG antibodies, indicating a direct effect of fz2-transfection on WG binding (Bhanot, 1996).

How do wingless signals get to the nucleus? The pathway is beginning to be understood in vertebrates. A component of the wingless pathway has been identified in Xenopus. A maternally expressed Xenopus homolog of mammalian HMG box factors (XTcf-3) binds to ß-catenin, the vertebrate homolog of Armadillo. XTcf-3 is a transcription factor that mediates ß-Catenin-induced axis formation in Xenopus embryos. It has been proposed that the ßcatenin-XTcf-3 complex is responsible for activation of targets genes in response to upstream Wnt (Wnt is the vertebrate homolog of Wingless) signals. By an as yet unknown mechanism, Wnt signals allow cytoplasmic ß-catenin to interact with XTcf-3 (Molenaar, 1996). The discovery of the involvement of an HMG-domain protein in vertebrate Wnt signaling encouraged a search for a similar HMG-domain protein in Drosophila. Isolated as a homolog of the vertebrate HMG-domain proteins described above, Drosophila Pangolin is found to be able to physically interact with vertebrate beta-catenin (Brunner, 1997).

Frizzled is involved in tissue polarity in Drosophila. A method was devised to induce a gradient of fz expression with the highest levels near the distal wing tip. The result is a large area of proximally pointing hairs in this region. This reversal of polarity is seen when fz expression is induced just before the start of hair morphogenesis, at the time polarity is established, suggesting that the gradient of Fz protein acts fairly directly to reverse hair polarity. A similar induction of the dishevelled (dsh) gene, which acts cell autonomously and functions downstream of fz in the generation of tissue polarity, results in a distinct tissue polarity phenotype, but indicates no reversal of polarity; this argues that fz signaling is required for polarity reversal. The finding that functional dsh is required for the reversal of polarity argues that the reversal requires normal fz signal transduction. The data suggest that cells sense the level of Fz protein on neighboring cells and use this information in order to polarize themselves. A polarizing signal is transmitted from cells with higher Fz levels to cells with lower levels. These observations enable the proposal of a general mechanism to explain how Wnts polarize target cells (Adler, 1997).

Recent evidence points to a role for lipid signaling downstream of Frizzled proteins. Stimulation of a G-protein-linked receptor initiates the hydrolysis of a membrane-bound inositol lipid, generating at least two second messengers: diacylglycerol and inositol-1,4,5-trisphosphate (InsP3). Diacylglycerol stimulates protein kinase C while InsP3 promotes the release of intracellular calcium (see Drosophila InsP3 receptor). The rat protein Frizzled-2 causes an increase in the release of intracellular calcium, which is enhanced by Xwnt-5a, a member of the Wnt family. Pertussis toxin (PTX) (which is a specific inhibitor of G alpha0 and G alphai subunits of G proteins that act by preventing the catalysis of GDP-GTP exchange stimulated by receptors) inhibits rat protein Frizzled-2 modulation of calcium flux. A nonhydrolysable GDP analog that irreversibly inactives G-protein-coupled events, inhibits rat FZ-2 induced Ca2+ transients. The release of intracellular calcium is suppressed by an inhibitor of the enzyme inositol monophosphatase, and hence of the phosphatidylinositol signaling pathway. This suppression can be rescued by injection of the compound myo-inositol, which overcomes the decrease in this intermediate caused by the inhibitor. These results indicate that some Wnt proteins work through specific Frizzled homologs to stimulate the phosphatidylinositol signalling pathway via heterotrimeric G-protein subunits, and that FZ-2 stimulates the phosphatidylinositol cycle through the betagamma subunits of pertussis-toxin-sensitive G proteins, leading to release of intracellular Ca2+ and diverse cellular responses. Since Gbetagamma subunits also activate protein kinase C, which may be involved in Wnt signaling, the responses by cells and embryos to signaling through Frizzled homologs could involve the stimulation of multiple cytoplasmic pathways. In early vertebrate embryos, regulation of the phosphatidylinositol pathway may be important for establishing the embryonic mesoderm and in other processes (Slusarski, 1997).

How is signaling specificity achieved by two closely related receptors? Both Frizzled and Fz2 signaling pathways use Dishevelled (Dsh) as a transduction component, raising the intriguing question of how two structurally related receptors signal through a common protein into distinct effector pathways. Fz2 has a higher affinity for Wg than Fz, and removal of either Fz or Fz2 has subtle, but different, effects on the patterning of the embryonic nervous system. Moreover, only Fz is specifically required for epithelial planar polarity by signaling through a Wg-Arm-independent pathway (Boutros, 2000).

Fz overexpression during Drosophila eye development causes a gain-of-function (GOF) planar polarity phenotype. Overexpression of Fz2 in the developing wing activates Wg-Arm targets. To compare the functional equivalence of Fz and Fz2 (Fz will be referred to as Fz1) for activating either the planar polarity or Wg-Arm pathways, Fz1 and Fz2 were expressed with tissue-specific enhancers in imaginal discs during Drosophila development. Whereas Fz1 overexpression in eye and wing discs results in planar polarity phenotypes, Fz2 expression leads to planar polarity defects with only very low penetrance. Conversely, overexpression of Fz2 in wing imaginal discs leads to formation of ectopic bristles (a wg GOF phenotype), whereas Fz1 overexpression does not affect bristle formation. Thus, Fz receptors have distinct signaling abilities in imaginal discs, despite their redundant role for Wg-Arm signaling in loss-of-function (LOF) analysis (Boutros, 2000).

To assess Wnt-beta-cat signaling in a quantifiable in vitro assay, both receptors were injected into Xenopus embryos and Wnt target induction in animal cap explants was analyzed. In this heterologous assay, Fz2 induces strong activation of the Wnt-beta-cat targets Xnr-3 and Siamois (Sia), whereas Fz1 induces a much weaker response. Thus, taken together with the imaginal disc phenotypes, Fz2 is a strong activator of Wnt-beta-cat signaling, and Fz1 is a potent activator of the planar polarity pathway. However, both receptors retain a low intrinsic potential to cross-activate either pathway (Boutros, 2000).

Differential Dsh localization may determine Fz signaling specificity, whereby Fz1, but not Fz2, can induce recruitment of Dsh to the membrane in Xenopus. At normalized protein levels for Fz1 and Fz2, however, no differences were observed in their ability to recruit Dsh. Titration experiments with Fz1 and Fz2 RNA concentrations show very similar threshold levels for either receptor in Dsh membrane localization. Thus, differential Dsh recruitment is unlikely to be the mechanism by which specificity between these Fz receptors is generated (Boutros, 2000).

Fz receptors are serpentine transmembrane proteins composed of an extracellular ligand-sequestering domain (CRD), a seven-pass transmembrane segment, and a COOH-terminal cytosolic tail. To determine which domains in Fz1 and Fz2 are required for directing signaling into either pathway, chimeric and truncated receptors were constructed. These chimeric proteins were tested for their signaling potential in Drosophila imaginal disc development in wings and nota, eyes, and legs for their ability to induce either GOF Wg-Arm signaling or planar polarity phenotypes (Boutros, 2000).

Both Fz1-2 and Fz1-1-2 chimeric proteins activate a Wg-Arm target (Achaete) in the wing imaginal disc, induce ectopic marginal bristles, and show wg-associated effects in the leg. However, they have no significant effect on planar polarity signaling in the eye, the wing, or the notum. Thus, the Wg-Arm signaling outcome corresponds with the presence of the Fz2 cytoplasmic tail. In contrast, GOF planar polarity phenotypes are observed with chimeric Fz2-1 receptors in the wing, the notum, and the eye that are indistinguishable from those caused by Fz1. Both Fz2-1 and Fz2-2-1 show a (mild) dominant-negative phenotype for Wg-Arm signaling, as judged by a reduction in Ac expression, (partial) absence of marginal bristles and notches in the wing margin, and the presence of stunted legs. These data suggest that the chimeric receptors containing the high-affinity Fz2 CRD, but lacking intracellular Fz2 sequences, can sequester Wg efficiently without eliciting an efficient signal transduction response. To test this possibility directly, stabilization of Wg was analyzed in wing discs. Whereas Fz1 or Fz1-2 type chimeras have no significant effect on Wg stability, all chimeric receptors with the Fz2 CRD strongly sequester Wg. The importance of cytoplasmic sequences for efficient Wnt target activation was confirmed in Xenopus animal cap assays (exchanging cytoplasmic domains switches the signaling efficiency). All chimeric receptors are able to recruit Dsh indistinguishably from wild-type Fz1 and Fz2 (Boutros, 2000).

The observation that the Fz1-2 chimera does not dominantly interfere with planar polarity signaling in the eye and because as yet no "planar polarity" ligand has been identified raises the question of whether planar polarity signaling depends on the presumed ligand binding Fz1 CRD. This hypothesis was tested by rescue of the fz-/- polarity mutant with sev-Fz1 and chimeric transgenes. In contrast to Fz1, however, neither Fz1-2 and Fz2-1 chimeras nor Fz2 rescue the fz mutant, indicating that the Fz1 CRD, coupled to its signaling unit, is required for correct levels of activation of planar polarity signaling in the eye in vivo. Thus, although overexpression of Fz2-1 induces a GOF planar polarity phenotype, identical to that induced by Fz1, it cannot replace Fz1 in a LOF background. Although both receptor isoforms, Fz1 and Fz2-1, when overexpressed, are capable of activating planar polarity signaling and perturbing correct polarity determination, the pathway is activated to the correct level only by the Fz1 CRD (and ligand)-dependent regulation of the receptor. Because the establishment of correct ommatidial polarity results from small differences in Fz signaling levels between neighboring R3 and R4 cells, the ability to precisely respond to the ligand in a spatially and temporally controlled manner is essential. Presumably, only Fz1 is appropriately regulated through its CRD to instruct correct ommatidial polarity (Boutros, 2000).

Fz1 and Fz2 appear to have different abilities to activate Wg-Arm and planar polarity signaling in the GOF assays in imaginal discs. Whereas Fz2 induces a Wg-Arm GOF phenotype, Fz1 causes GOF planar polarity phenotypes. The chimeric receptors define the respective cytoplasmic tail (Fz2) or the cytoplasmic domains (Fz1) as largely, but not solely, responsible for mediating these differences in the GOF assays. The Wg-Arm GOF phenotype is ligand- and CRD domain-dependent, because it can only be observed close to the source of Wg. Also, Fz2 has a stronger effect than Fz1-2 or Fz1-1-2. The planar polarity ligand is, possibly, a member of the Wnt family with a different CRD binding affinity from that of Wg. The mechanism by which the ligand-CRD interaction regulates Fz signaling is unclear. The present data cannot distinguish between an activating (conformational) change, or alternatively, a constitutive signaling capacity by Fz's that is inhibited by another factor and needs to be antagonized by the ligand (similar to Smoothened/Patched signaling with Hedgehog) (Boutros, 2000).

How can one reconcile the Fz1 and Fz2 redundancy for Wg signaling in LOF analysis and the dominant-negative behavior of the Fz2-1 chimeras? Fz2 is a high-affinity Wg receptor, and fz2 transcription is down-regulated by Wg, whereas Fz1 (a low-affinity receptor) is expressed fairly uniformly. Thus, Fz2 might be the primary Wg receptor, and Fz1 substitutes only in its absence. Moreover, another Drosophila Fz family member, Fz3, acts as a negative attenuator of Wg signaling and is positively regulated by Wg, suggesting that the expression patterns of Fz2 and Fz3 shape the Wg response, whereas Fz1 does not contribute to this effect. In this context, overexpression of Fz2-1, consisting of a high-affinity Wg-binding CRD fused to a low-efficiency signaling unit, adversely affects the signaling outcome and causes a dominant-negative phenotype (Boutros, 2000).

These experiments provide a model for how signaling specificity can be achieved by closely related receptors, and they demonstrate that LOF studies, like GOF experiments, might only provide a partial answer in case of redundancies. Quantitative differences in ligand affinity and signal transduction efficiency of Fz receptors could provide overlapping and nonoverlapping functions in different cells, depending on the threshold needed to induce targets and expression levels of the various members of the receptor family. Thus, the relative ratio of the different Fz receptors on the cell surface and their degree of occupancy could be an important factor determining the signaling outcome. Additional factors such as coreceptors could influence the signaling outcome: for example, the heparan sulfate proteoglycan Dally has been identified as a coreceptor in Wg signaling. Fz1 and Fz2 signaling preferences provide an example of how quantitative differences in signaling levels can lead to redundant and specific roles for these receptors during development and evolution (Boutros, 2000).

Subcellular localization of frizzled receptors, mediated by their cytoplasmic tails, regulates signaling pathway specificity

The Frizzled (Fz; called here Fz1) and Fz2 receptors have distinct signaling specificities activating either the canonical Wnt/beta-catenin pathway or Fz/planar cell polarity (PCP) signaling in Drosophila. The regulation of signaling specificity remains largely obscure. This study shows that Fz1 and Fz2 have different subcellular localizations in imaginal disc epithelia, with Fz1 localizing preferentially to apical junctional complexes, and Fz2 being evenly distributed basolaterally. The subcellular localization difference directly contributes to the signaling specificity outcome. Whereas apical localization favors Fz/PCP signaling, it interferes with canonical Wnt/beta-catenin signaling. Receptor localization is mediated by sequences in the cytoplasmic tail of Fz2 that appear to block apical accumulation. Based on these data, it is proposed that subcellular Fz localization, through the association with other membrane proteins, is a critical aspect in regulating the signaling specificity within the Wnt/Fz signaling pathways (Wu, 2004).

Frizzled chimeras described below are denoted by three numbers, the first being the source of the N-terminal Wnt-interacting cysteine-rich domain (CRD), the second being the source of the remaining proximal extracellular domain and 7 transmembrane region and loop region (collectively referred to as 7-TM), and the third refering to the source of the intracellular C-tail (Wu, 2004).

It was asked whether the apical localization of Fz is required for PCP signaling? The Fz1-1-2 chimera, which is distributed ubiquitously within the apical-basolateral membrane, only partially rescues the fz^- eye phenotype, and it can also cause defects related to canonical Wg/Arm signaling. In contrast, apically localized Fz1-1-2S ('S' for 'short') fully rescues the fz^- phenotype and has no additional effects. The Fz1-1-2 chimera also shows much weaker PCP phenotypes in the GOF assay. Taken together, these results suggest that a reduction in the apical localization of Fz leads to a reduction in PCP signaling activity. However, about 80% of the chirality defects in fz^- eyes are rescued by tub-fz1-1-2, and in the wing tub-fz1-1-2 rescues the fz^- mutant to a similar extent as tub-fz1-1-1 and tub-fz1-1-2S, suggesting that Fz1-1-2 contains substantial PCP signaling activity (Wu, 2004).

Because both GOF and loss-of-function studies indicate that the Fz1 7-TM region is critical for Fz1 function, Fz1-1-2 is expected to have Fz/PCP signaling activity, although with altered subcellular distribution. Thus, the remaining PCP signaling activity of Fz1-1-2 seen is probably due to the presence of some of this protein in apical regions. It is difficult to determine how much of Fz1-1-2 is actually localized to this membrane region. Since the immunohistochemical staining indicates that it is not excluded apically, it is assumed that Fz1-1-2 has enough apical localization to participate when PCP signaling is initiated. It has been suggested that wing cell orientation does not depend on absolute Fz levels, but instead depends on relative Fz/PCP activity differences in a Fz activity gradient across a field. Thus, although the absolute activity of Fz1-1-2 is reduced (based on weaker GOF phenotypes and weaker rescue of fz^- in the eye), the relative difference might be sufficient for the partial rescue (Wu, 2004).

In this context, it is worth noting that tub-fz1-1-2 rescues the fz^- phenotype better in the wing than in the eye, whereas there is no apparent difference in rescue activity between the eye and the wing for tub-fz1-1-1 or tub-fz1-1-2S. The difference could be due to the observed nonautonomous PCP signaling effects in the wing, where neighboring cells affect each other's planar polarization. Fz1-1-2 may allow some wing cells to adopt the correct orientation, which then in turn influences many of the remaining wing cells to also orient themselves correctly through nonautonomous interactions (Wu, 2004).

It has been shown that Fz1 localization is affected in flamingo (fmi) mutant clones at about 30 h APF, leading to the proposal that Fmi recruits Fz1 into apical junctions. However, this study found that Fz1 is localized normally in fmi null mutant clones earlier in the third instar wing disc. What causes the difference between these two observations? PCP signaling in the wing is thought to act in two phases (one 6-24 h APF and the second 24-32 h APF [, and it results in the distal enrichment and maintenance of Fz1. Since Fz1/PCP signaling is modulated by Fmi, Fmi-dependent changes in Fz1 localization likely result from effects on PCP signaling activity. At the same time, Fmi localization is also dependent on Fz1 activity and becomes also less apically localized in fz^- tissue at 30-36 h APF, suggesting that the regulation of apical localization between Fz1 and Fmi is complicated and mutual at these late stages (Wu, 2004).

This study has shown that initial apical localization of Fz1, preceding both stages of PCP signaling, is not fmi dependent. This result suggests that Fmi and Fz1 get recruited to apical junctions independently. During later stages, Fmi and Fz1 then affect each other's localization through PCP signaling. At this point, it remains unclear which molecules initially recruit Fz1 into the apical junctional region (Wu, 2004).

Secreted Wg mainly binds to Fz2 at basolateral membrane regions of the wing epithelium, indirectly suggesting that canonical signaling occurs in the basolateral membrane compartment. The current experiments show that overexpression of Fz1-1-1 or Fz2-1-1 leads to a cell-autonomous loss of wing margin bristles and associated tissue, suggesting that these molecules act like dominant negatives, inhibiting Wnt/β-cat signaling. As these molecules are enriched apically and sequester Dsh there, Fz-Dsh complexes at apical junctions may be largely inactive for canonical Wnt signaling. This result suggests that canonical Wnt signaling and PCP signaling occur in different subcellular compartments. Basolateral Wnt/β-cat signaling is also suggested by the fact that (1) secreted Wg binds to Fz2 at the basolateral membrane and that (2) apical Wg secretion and signaling could lead to mis-specification in disc folds and cells in the peripodial membrane (Wu, 2004).

Both Fz1 and Fz2 are capable of canonical Wnt/β-cat signaling. Consistently, different Fz1/2 chimeras, including related versions of Fz2-1-1 and Fz2-2-1, are capable of rescuing the fz, fz2 double mutant phenotype. However, when Fz1-1-1, Fz2-1-1, or Fz2-2-1 is expressed at high levels, Dsh accumulates at apical junctions, thus decreasing cytosolic Dsh levels. Since the chimeric receptors can rescue the fz, fz2 double mutant when expressed at low levels (under the control of the tub promoter), the relative level of each receptor together with its subcellular localization appear critical for the signaling outcome (Wu, 2004).

In conclusion, this study has shown that subcellular localization contributes to Fz signaling specificity. The data indicate that the localization of Fz1 at apical junctions promotes Fz/PCP signaling, whereas this localization can inhibit canonical Wnt/β-cat signaling. The localization is mediated through sequences in the C-tail (Wu, 2004).

PROTEIN STRUCTURE

Amino Acids - 694

Structural Domains

FZ2 resembles all other members of the frizzled family in having the following structural motifs (beginning at the N terminus): a signal sequence, a domain of 120 amino acids with an invariant pattern of ten cysteine residues, a highly divergent region of 40-100 largely hydrophilic amino acids that is predicted to be flexible, and seven putative transmembrane segments. The C terminus resembles that of most mammalian Frizzled proteins, ending with the sequence S/T-X-V. FZ2 most closely resembles human fz5 and mouse fz8, with which it shares 49% and 45% amino acid identity, respectively. FZ and FZ2 share 33% amino-acid identity (Bhanot, 1996).

date revised: 11 August 97  

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.