BIOLOGICAL OVERVIEW

Mutations in double-time effect circadian rhythms in Drosophila. dbt contributes to circadian rhythmicity by determining the stability of Period (Per) protein. It has been proposed that Dbt promotes molecular cycles of per and timeless (tim) expression by ensuring the instability of monomeric Per proteins, thus allowing Per accumulation only in conjunction with high titers of Tim (Price, 1998). Cloning of dbt has revealed that Dbt protein is closely related to a family of protein kinases (Kloss, 1998). The most closely related kinases are the delta and epsilon isoforms of human casein kinase I (Fish, 1995).

Double-time turns out to be identical to the discs overgrown (dco) cell growth regulating gene of Drosophila. In contrast to the weak double-time alleles, which appear to affect only the circadian rhythm, discs overgrown alleles show strong effects on cell survival and growth control in imaginal discs. The Discs overgrown protein is a crucial component in the mechanism that links cell survival during proliferation to growth arrest in imaginal discs. Since the amino acid sequences and three-dimensional structures of Casein kinase I delta/epsilon enzymes are highly conserved, the results suggest that these proteins may also function in controlling cell growth and survival in other organisms. Analysis of phenotypes of various discs overgrown mutants, including heteroallelic combinations of strong and null alleles, suggests that this gene is required for inhibition of apoptosis during cell proliferation as well as for growth arrest in imaginal discs (Zilian, 1999).

This overview will treat the involvement of Double-time in the photoperiod response. Information about the role of dbt/discs overgrown in the regulation of imaginal disc survival, proliferation and growth arrest can be found in the Effects of mutation section. The current understanding of the molecular regulation of circadian rhythmicity in Drosophila comes from integrating genetics and molecular biology. Null mutations in either of two genes, per and tim, abolish behavioral rhythmicity, while alleles encoding proteins with missense mutations have been recovered at both loci and show either short- or long-period behavioral rhythms. The RNA and protein products of these genes oscillate with a circadian rhythm in wild-type flies. These molecular rhythms are abolished by null mutations of either gene, and the periods of all molecular rhythms are correspondingly altered in each long- and short-period mutant, indicating a regulatory interaction between these genes. Production of these molecular cycles appears to depend on the rhythmic formation and nuclear localization of a complex containing the Per and Tim proteins. A physical interaction of Per and Tim is required for nuclear localization of either protein, and nuclear activity of these proteins coordinately regulates per and tim transcription through a negative feedback loop. A complex of two transcription factors, Clock and Cycle, positively regulates both per and tim transcription, and this positive regulation is suppressed by nuclear Per/Tim proteins. A model for the Drosophila clock has been proposed in which delayed formation of Per/Tim complexes ensures separate phases of per/tim transcription and nuclear function of the encoded proteins. Entrainment of this oscillator is regulated through the Tim protein, which is rapidly eliminated from the nucleus and cytoplasm of pacemaker cells when Drosophila are exposed to daylight (Price, 1998 and references).

Although some key features of the Drosophila clock have been identified, the involvement of additional, essential factors is suspected from prior work. Per fails to accumulate in the absence of Tim even in the presence of high per RNA levels, pointing to the existence of an activity that destabilizes cytoplasmic Per monomers. Both Per and Tim (Edery, 1994 and Zeng, 1996) as well as Clock (Lee, 1998), are phosphorylated with a circadian rhythm, which is an indication of the presence of unidentified kinases. Per, in particular, becomes progressively phosphorylated over many hours, and the timing of its phosphorylation is changed in period-altering mutants (Edery, 1994 ), suggesting either circadian regulation of Per phosphorylation or a role in establishing rhythmicity. double-time alleles have been found that either shorten or lengthen the periods of behavioral and molecular rhythms. A strongly hypomorphic dbt allele has been isolated that is associated with pupal lethality and blocks circadian oscillations of per and tim gene products in larvae. Dbt is therefore a central clock component alongside Per and Tim. dbt period-altering alleles alter the kinetics of Per phosphorylation and degradation. The hypomorphic allele constitutively produces unusually high levels of Per proteins that are hypophosphorylated. Thus, a normal function of Dbt appears to be regulation of Per accumulation. Evidence is presented that dbt contributes to circadian rhythmicity by determining the stability of Per. It is proposed that DBT activity in wild-type flies promotes molecular cycles of per and tim expression by ensuring the instability of monomeric Per proteins, thus allowing Per accumulation only in conjunction with high titers of Tim (Price, 1998 and references).

What is the effect of dbt mutation on Per and Tim levels in the brain? It was reasoned that the strongly hypomorphic P-element insertion allele dbt^P might show the most dramatic effects on clock gene cycling. Although dbt^P embryos take longer to develop into third instar larvae than do their heterozygous siblings, the foraging motility of these larvae and their touch sensitivity appear normal. Behavioral studies have demonstrated that a circadian clock is active in Drosophila larvae. It has recently been shown that a specific group of central brain cells is likely to compose the larval pacemaker (Kaneko, 1997). In each hemisphere of the third instar larval brain, four to five cells coexpress Per and Tim with circadian oscillations that are in phase with the oscillations of these proteins in adult pacemaker cells. The only detectable staining in larval brain hemispheres for pigment-dispersing hormone (PDH), a marker for adult lateral neurons (Helfrich-Forster, 1995), is found in the cell bodies and axons of these Per-Tim expressing cells (Kaneko, 1997). Therefore, the Per-Tim-PDH coexpressing larval brain cells can be considered larval lateral neurons (lvLNs). In per^o larvae, Tim is constitutively cytoplasmic in lvLNs; in tim⁰¹ larvae, Per is undetectable in these cells by immunocytochemistry. Both of these mutant phenotypes are characteristic of adult clock cells. Per and Tim oscillations can also be seen in two groups of cells at the anterior of the third instar larval brain, but in one of these groups, the oscillations are out of phase with the lvLNs and may be regulated by activity of the lvLNs. Since there are no good markers to distinguish clearly between these two anterior cell types, for analysis, focus was placed on the lvLNs (Price, 1998).

LvLNs are indeed present in dbt^P larvae. PDH staining is detected in the cytoplasm of four cells in each hemisphere in all dbt^P larval brains examined at different times in alternating light-dark (LD) and constant darkness (DD) cycles. The axons of these dbt^P lvLNs fasciculate and head to the anterior of the brain as in wild type. However, dbt^P lvLNs are found slightly more peripherally than in wild type, as seen in the staining patterns of Per and Tim. This probably indicates a subtle developmental effect of dbt on the architecture of the brain, which might be expected given that the dbt^P mutation causes lethality before completion of pupal development (Price, 1998).

Although regulation of Tim's light sensitivity and nuclear localization are not affected by dbt^P mutation, oscillations of Tim protein and TIM mRNA cease in dbt^P larval brains in constant darkness. When wild-type larvae are transferred to constant darkness, Tim continues to oscillate robustly with only night time Tim accumulation in the lvLNs. In contrast, in dbt^P larvae transferred to constant darkness (DD), Tim is weakly detected in lvLNs in the first subjective morning at CT5 (CT, circadian time, indicates time in DD), disappears by CT10, and is undetectable thereafter in the lvLNs. Tim is always detected in dbt^P in the anterior larval brain cells; these cells serve as a positive control for the procedure. For the lvLNs, the differential effects of dbt^P on Tim in LD and DD are not due to selecting larvae from slightly different developmental stages since identical results were derived from larvae that had been synchronized developmentally. In wild-type larvae, TIM mRNA shows robust oscillations in the lvLNs in both LD and DD. TIM mRNA levels oscillate in the lvLNs of dbt^P mutants in an LD cycle, indicating that Per/Tim complexes can still negatively regulate tim gene expression in dbt^P larvae and that this regulation can be blocked by light-dependent degradation of Tim. When dbt^P larvae are transferred to DD, TIM mRNA is weakly detected in lvLNs on the first subjective morning (CT2) but is undetectable thereafter in these cells. Thus, the effects of the dbt^P mutation on levels of Tim protein probably reflect more direct effects of dbt^P on TIM mRNA levels (Price, 1998).

Per protein levels oscillate in wild-type lvLNs in an LD cycle, reaching peak levels at ZT23. In DD, Per continues to cycle in wild-type lvLNs and is detected at CT1 but not CT13. Per proteins produced by dbt^P larvae show three significant differences from wild type. (1) Per is constitutively expressed in lvLNs in LD and DD cycles; (2) the intensity of staining in the lvLNs is stronger in dbt^P than in wild-type larvae; (3) the pattern of expression in dbt^P is widened to include regions of the brain not significantly stained in a wild-type background. The elevated level of Per in dbt^P was confirmed by Western blotting using extracts from dissected larval brains collected in LD. The latter results show that Per protein accumulation is dramatically increased by the dbt^P mutation, and the high levels of accumulated Per protein do not show significant differences between ZT12 and ZT24 in LD cycles in dbt^P. In addition, the electrophoretic mobility of Per proteins is relatively high and uniform in dbt^P larvae, in contrast to the broad spectrum of lower Per protein mobilities observed in wild-type larvae and adult heads. As the spectrum of protein mobilities in wild-type Drosophila reflects Per protein phosphorylation (Edery, 1994), the results suggest that Per is hypophosphorylated in dbt^P mutants (Price, 1998).

To determine whether the high levels of Per protein found by Western blotting of dissected dbt^P larval brains reflects altered PER mRNA levels, RNase protection was used to detect per RNA in these tissues at ZT14-16 (time of expected peak PER mRNA accumulation in wild-type Drosophila). PER mRNA is expressed at similar levels in wild-type and dbt^P larval brains. Thus, the aberrant accumulation of Per proteins in dbt^P mutants does not reflect increased per transcription or per RNA stability but must be downstream of these events (Price, 1998).

The pattern of Per expression in dbt^P is similar to a Per beta-galactosidase fusion protein, Per-SG, expressed from the per promoter. In adults, per-SG RNA oscillates, but Per-SG protein does not. In fact, the Per-SG protein accumulates over progressive cycles, suggesting that it is a stable protein. Per-SG, detected with an antibody against beta-galactosidase, is expressed in larvae in the lvLNs and other cell clusters in the brain hemispheres, as well as cells adjacent and close to the ventral ganglion midline. The presence of the noncycling Per-SG fusion protein therefore marks cells in which the per promoter is active, or has been active, during development. Comparable patterns were seen with two independent SG lines. Per in wild-type larvae has also been detected at very low levels in these cells. Thus, the pattern of staining for Per seen in dbt^P reflects the normal spatial expression of per, but this pattern is only easily visible with a stable fusion protein or in a dbt^P background (Price, 1998).

Per is detected at high levels in dbt^P larval brains in DD as in LD. The persistence of Per proteins in lvLNs in DD is presumably occurring in the presence of very low levels of the Tim protein, which fall below immunocytochemical detection in these cells in DD. In dbt⁺ larvae and adults, Per accumulation is suppressed in the absence of Tim. It therefore seems likely that Per in dbt^P has become less dependent on Tim for its accumulation than in wild type, especially since Per is detectable in brain cells where Tim is not detected in wild-type or dbt^P larvae. Ideally this possibility would have been tested using tim⁰¹; dbt^P larvae. However, it has not been possible to obtain third instar larvae from two different tim⁰¹; dbt^P/TM6 lines. This may indicate a shift in the lethal phase of dbt^P by the tim⁰¹ mutation. An alternative to tim⁰¹ was possible: constant light (LL) in a wild-type background produces a tim⁰¹ phenocopy through light-dependent degradation of Tim. In wild-type larvae, Per accumulation is suppressed in response to LL as previously seen for adults. In contrast, in dbt^P larvae raised in LL, Per continues to be strongly detected in the lvLNs and the other Per-expressing cells. The persistence of Per in DD and LL indicates that Per proteins can accumulate in dbt^P mutants even with very low levels of Tim (Price, 1998).

Where does Dbt act in the cell? In tim⁰¹ mutants, which block Per nuclear translocation, Per is unstable indicating the presence of a cytoplasmic activity that destabilizes Per monomers. In wild-type adults, constant light suppresses Tim, which subsequently results in very low levels of Per, and the same holds true for wild-type third instar larvae when raised in constant light. However, in dbt^P mutants, similar high levels of Per accumulate under LD, DD, and LL conditions. Since the dbt^P allele allows comparable Per accumulation with either high or low levels of Tim, it is concluded that dbt^P allows Tim-independent Per accumulation and that DBT is a component of the cytoplasmic activity that destabilizes Per monomers in wild-type and tim⁰¹ flies. Consistent with this conclusion, predominantly cytoplasmic accumulation for the expanded Per pattern is seen in dbt^P larval brains. tim does not appear to be expressed in this expanded pattern in wild-type larvae, so expanded accumulation of Per monomers in dbt^P, but not wild-type, larval brains indicates novel cytoplasmic stability for Per in the mutant. There is also evidence that Dbt influences stability of nuclear Per proteins. Per is detected immunocytochemically in nuclei of mutant dbt lvLNs prior to Per's disappearance, suggesting that the differing kinetics of Per degradation in wild type and dbt mutants reflect different rates of Per elimination from the nucleus. Increased nuclear stability of Per monomers is apparent from the analyses of dbt^P lvLNs: Per is lost from wild type but persists in dbt^P nuclei after exposure to light has removed most Tim proteins. Thus, dbt may affect the stability of both cytoplasmic and nuclear Per monomers. This does not mean that Dbt must function in both subcellular compartments, since a posttranslational modification generated in the cytoplasm could have delayed effects in the nucleus. Thus it is proposed that Dbt activity in wild-type flies promotes molecular cycles of per and tim expression by ensuring the instability of monomeric Per proteins, thus allowing Per accumulation only in conjunction with high titers of Tim (Price, 1998).

A model is presented for the functioning of Dbt in photoperiod response. Dbt function promotes phosphorylation of cytoplasmic Per monomers. Once modified, Per proteins turn over rapidly. Physical association of Per and Tim stabilizes Per either because residual, unphosphorylated Per proteins are incorporated into Per/Tim dimers and these are no longer subject to modification by Dbt, or phosphorylated Per proteins are stabilized by association with Tim. Because monomeric Per proteins also stably accumulate in nuclei in dbt^P mutants, but not in wild-type Drosophila (Price, 1998), the latter alternative is favored. The model indicates that instability of phosphorylated Per monomers delays Per/Tim heterodimerization until PER and TIM mRNA levels are high. Thus, phosphorylation promotes a delay between phases of per /tim transcription and Per/Tim complex function, which establishes molecular oscillations of RNA and protein. Edery (1994) has shown that Per is phosphorylated over many hours and that the most highly phosphorylated forms of the protein occur at times of night when Per proteins are predominantly nuclear. This suggests that Dbt may act in both the cytoplasm and the nucleus or that additional kinases continue to phosphorylate Per in the nucleus. The finding that dbt^P mutants accumulate high levels of nuclear Per and that dbt^L mutants show delayed degradation of hyperphosphorylated, monomeric forms of Per (Price, 1998) suggests that Dbt activity can influence stability of nuclear Per proteins after they have dissociated from TIM. However, this could be due to the production of a phosphorylated form of Per by cytoplasmic Dbt, with subsequent transfer of phosphorylated Per to nuclei as a Per/Tim complex. It is not yet known whether all monomeric Per proteins are phosphorylated in response to Dbt, so that Per/Tim complexes stabilize previously phosphorylated Per proteins, or whether only Per proteins that have escaped phosphorylation as a monomer contribute to Per/Tim complexes. Future immunocytochemical studies of DBT will allow subcellular localization of the protein and resolution of some of these issues (Kloss, 1998).

Noncanonical FK506-binding protein BDBT binds Dbt to enhance its circadian function and forms foci at night

The kinase Doubletime is a master regulator of the Drosophila circadian clock, yet the mechanisms regulating its activity remain unclear. A proteomic analysis of Doubletime interactors led to the identification of an unstudied protein designated CG17282. RNAi-mediated knockdown of CG17282 produced behavioral arrhythmicity and long periods and high levels of hypophosphorylated nuclear Period and phosphorylated Doubletime. Overexpression of Doubletime in flies suppresses these phenotypes and overexpression of CG17282 in S2 cells enhances Doubletime-dependent Period degradation, indicating that CG17282 stimulates Doubletime's circadian function. In photoreceptors, CG17282 accumulates rhythmically in Period- and Doubletime-dependent cytosolic foci. Finally, structural analyses demonstrated CG17282 is a noncanonical FK506-binding protein with an inactive peptide prolyl-isomerase domain that binds Doubletime and tetratricopeptide repeats that may promote assembly of larger protein complexes. CG17282 was names Bride Of Doubletime and was established as a mediator of Doubletime's effects on Period, most likely in cytosolic foci that regulate Period nuclear accumulation (Fan, 2013).

While FKBPs were originally identified as mediators of the immunosuppressive effects of FK506 on calcineurin and rapamycin on the Target of Rapamycin (TOR), subsequent work has suggested their involvement in a wide range of signaling processes, including ones involved in neurodegeneration and cancer. In many cases, their function derives from their catalysis of cis-trans conversions of peptide bonds involving prolines. However, BDBT lacks the necessary catalytic residues, as do several other noncanonical FKBPs. One of these noncanonical FKBPs (FKBP38) has been proposed to interact with TOR to suppress its activity, while interactions between FKBP38 and the small GTP-binding protein RHEB relieve this repression and activate TOR. Intriguingly, TOR and RHEB have recently been show to modulate the circadian clock of Drosophila (Fan, 2013).

However, FKBP proteins have also been implicated in regulation of nuclear localization and protein stability. For instance, the noncanonical FKBP-like protein (FKBPL) has been implicated in the nuclear import of steroid hormone receptors in complexes with HSP90 proteins. An interesting possibility is that BDBT is involved in regulating the import of Per/Dbt complexes to the nucleus, and that at least some of this regulation is negative, as Per exhibits increased nuclear accumulation in BDBT knockdown flies. This hypothesis is consistent with structural work, which uncovered a resemblance between BDBT and the HSP90-binding protein FKBP51. The HSP90-binding site in FKBP51 localizes to its TPR domain and all but one of the residues that account for HSP90 binding are conserved in BDBT in spite of the low sequence homology with BDBT. Since the N-terminal, PPIase-like domain of BDBT binds to Dbt in HEK293 cells, it is possible that BDBT assembles a Dbt/Per/HSP90 complex, with Dbt bound to the PPIase-like domain, HSP90 to the TPR domain, and Per bound to Dbt (Fan, 2013).

FKBPs have also been implicated in the regulation of the stabilities of proteins with which they form a complex. A role in enhancement of Per's phosphorylation-dependent proteolysis is particularly attractive for BDBT, as it would explain the RNAi knockdown phenotype in head extracts (elevated levels of hypophosphorylated Per) and the enhancement of Dbt-dependent degradation of Per in S2 cells. The cytosolic BDBT foci in Drosophila photoreceptors accumulate at a time (ZT13-19) when Per transitions from a destabilized cytosolic form to a stabilized nuclear form, and the data supporting the involvement of BDBT in enhancement of Per proteolysis suggest that BDBT may be a negative regulator of this transition (i.e., BDBT antagonizes Per accumulation and nuclear localization). The BDBT foci are intriguing in light of the finding by Neyer of Per/Tim cytosolic foci, which form prior to accumulation of Per and Tim in S2 cell nuclei (Meyer, 2006). It was proposed that processes in these foci trigger the nuclear accumulation of both Per and Tim. Since the suggestion from this work is that BDBT foci antagonize nuclear accumulation of Per and no obvious Per foci were observed that colocalize with the BDBT foci, it is possible that BDBT antagonizes focal accumulation of Per or immediately triggers the degradation of Per in these foci. In this scenario, in contrast to the situation in S2 cells, Per might accumulate in foci in vivo only when BDBT is not present or active in the foci, and Per's presence in foci in vivo may therefore be difficult to detect because it rapidly accumulates and localizes to nuclei when BDBT is not active in the foci. It is also possible that focal accumulation of BDBT negatively regulates BDBT activity toward Dbt, since highest levels of BDBT foci are detected at ZT19, when Per is rapidly accumulating in nuclei and therefore any BDBT inhibition of Per nuclear accumulation might be inhibited (Fan, 2013).

Because the mammalian orthologs of Dbt (CKIepsilon and CKIdelta) are also essential for the molecular mechanism of the mammalian circadian clock, it is possible that the mechanism in which bdbt participates is conserved in mammals. While this manuscript was in preparation, FKBP and FKBP-like proteins were reported to form complexes with mammalian CKIepsilon and CKIdelta (Kategaya, 2012), although neither these proteins nor any other protein in the mammalian genome is an ortholog of BDBT. The lack of homology between the PPIase-like region in BDBT, which mediates binding to DBT, and the ones found in vertebrates, was initially surprising. However, the binding experiments indicate that the BDBT binding site in DBT spans its well-conserved N-terminal kinase domain and the poorly conserved C-terminal tail. Thus, it seems likely that the binding modes between BDBT and Dbt on one hand and the ones between the vertebrate homologs of BDBT and CKIepsilon and CKIdelta differ substantially. While it is not known if this interaction has any role in the mammalian circadian clock, these results offer the tantalizing prospect that this class of proteins and their roles are conserved in the mechanisms of the mammalian and Drosophila clocks (Fan, 2013).

Reciprocal action of Casein Kinase Iepsilon on core planar polarity proteins regulates clustering and asymmetric localisation

The conserved core planar polarity pathway is essential for coordinating polarised cell behaviours and the formation of polarised structures such as cilia and hairs. Core planar polarity proteins localise asymmetrically to opposite cell ends and form intercellular complexes that link the polarity of neighbouring cells. This asymmetric segregation is regulated by phosphorylation through poorly understood mechanisms. This study shows that loss of phosphorylation of the core protein Strabismus in the Drosophila pupal wing increases its stability and promotes its clustering at intercellular junctions, and that Prickle negatively regulates Strabismus phosphorylation. Additionally, loss of phosphorylation of Dishevelled - which normally localises to opposite cell edges to Strabismus - reduces its stability at junctions. Moreover, both phosphorylation events are independently mediated by Casein Kinase Iepsilon. It is concluded that Casein Kinase Iepsilon phosphorylation acts as a switch, promoting Strabismus mobility and Dishevelled immobility, thus enhancing sorting of these proteins to opposite cell edges (Strutt, 2019).

Phosphorylation is a widespread means of controlling protein activity, regulating protein-protein interactions, protein stability and conformation. The activity of most signalling pathways is regulated by phosphorylation of pathway components. This includes the 'core' planar polarity pathway: however, compared to other signalling pathways, the molecular mechanisms are poorly understood (Strutt, 2019).

The core planar polarity proteins (hereafter, the 'core proteins') regulate the production of polarised structures or polarised cell behaviours in the plane of a tissue. This includes polarised production of cilia and of stereocilia bundles in the inner ear, and the coordinated polarisation of tissue movements necessary for convergence and extension of the body axis. In Drosophila, the core pathway controls the production of polarised hairs and bristles on many adult tissues, for example the trichomes that emerge from the distal edge of each cell in the adult wing (Strutt, 2019).

The core pathway specifies polarised structures via the asymmetric localisation of pathway components. In the Drosophila pupal wing, the seven-pass transmembrane protein Frizzled (Fz), and the cytoplasmic proteins Dishevelled (Dsh) and Diego (Dgo) localise to distal cell ends, where the trichome will emerge. The four-pass transmembrane protein Strabismus (Stbm, also known as Van Gogh [Vang]) and Prickle (Pk) localise to proximal cell ends, and the atypical cadherin Flamingo (Fmi, also known as Starry Night [Stan]) localises to both proximal and distal cell ends (see Planar polarity and the cloud model of core protein localisation). Fmi mediates homophilic adhesion that is important for coupling polarity between cells (Strutt, 2019).

The overall direction of polarisation is determined by tissue-specific global cues. Polarity is then thought to be refined and amplified by feedback interactions between the core proteins. Mathematical modelling has suggested that feedback may involve destabilisation of complexes of opposite orientation and/or stabilisation of complexes in the same orientation. This can lead to sorting of complexes such that they all align in the same direction (Strutt, 2019).

With regard to possible stabilising mechanisms, core protein asymmetry is associated with clustering of proteins into punctate membrane subdomains and reduced core protein turnover. Based on a detailed study of core protein organisation in puncta, it was recently proposed that core proteins form a non-stoichiometric 'cloud' around a Fmi-Fz nucleus. Feedback interactions lead to sorting of complexes, and multiple protein-protein interactions are thought to promote a phase transition into higher order 'signalosome-like' structures, where arrays of complexes of the same orientation are stabilised. Interestingly, Stbm stoichiometry was found to be much higher than that of the other core proteins. The reasons for this are unclear, but could relate to a role for Stbm in promoting higher order structures. Furthermore, Pk may stabilise Stbm by promoting complex clustering (Strutt, 2019).

Mechanisms of destabilisation may include competitive binding between core proteins. More specifically, Pk (a 'proximal' complex component) is known to destabilise Fz and/or Dsh ('distal' components) in the same cell. In addition, Pk has been suggested to destabilise complexes containing Stbm and Fmi. However, knowledge of additional molecular mechanisms by which core proteins might become destabilised or clustered together is very poor, and post-translational modifications such as phosphorylation are likely to be a key element (Strutt, 2019).

Indeed, core protein phosphorylation is essential for feedback amplification of asymmetry. In particular, reduced activity of Casein Kinase Iε (CKIε, also known as Discs Overgrown [Dco] or Doubletime [Dbt] in flies) causes planar polarity defects and a reduction in core protein asymmetry. Interestingly, CKIε has been implicated in phosphorylation of both Stbm and Dsh. CKIε was first found to bind to and phosphorylate the vertebrate Dsh homologue (Dvl) in canonical Wnt signalling. In planar polarity in flies, Dsh phosphorylation correlates with its recruitment to cellular junctions by Fz, where it is incorporated into stable complexes, and decreased Dsh phosphorylation is seen in &dco; mutants (Strutt, 2019).

The exact phosphorylation sites for CKIε in Dsh/Dvl are not well defined, but a mutation of a serine/threonine-rich region upstream of the PDZ domain affects Dvl recruitment to membranes in Xenopus. Moreover, mutation of one of these residues (S236 in fly Dsh) blocks phosphorylation of Dsh by Dco in vitro. However, a transgene in which these residues were mutated largely rescued the planar polarity defects of dsh mutants in the adult fly wing (Strutt, 2019).

More recently, CKIε has been implicated in phosphorylating Stbm and its vertebrate homologue Vangl2. In particular, Wnt gradients were proposed to lead to a gradient of Vangl2 phosphorylation and asymmetry in the vertebrate limb. CKIε promotes Stbm/Vangl2 phosphorylation in cell culture. Two clusters of conserved serine and threonine residues were identified as CKIε phosphorylation sites. Mutation of some or all of these residues leads to a loss of Stbm/Vangl2 phosphorylation in cell culture, and defects in planar polarisation (Strutt, 2019).

The fact that CKIε has been implicated in phosphorylating both Stbm/Vangl2 and Dsh/Dvl in cell culture leads to the question of whether both proteins are bona fide targets in vivo. For instance, both Fz and Dsh/Dvl have been proposed to promote Stbm/Vangl2 phosphorylation by CKIε. Thus, it is possible that only Stbm/Vangl2 are direct targets of CKIε and that Stbm/Vangl2 phosphorylation has a secondary effect on Fz-Dsh/Dvl behaviour. Moreover, mechanistic insight into how these phosphorylation events affect core protein sorting and asymmetry is lacking (Strutt, 2019).

This study demonstrates that CKIε has independent and reciprocal actions on Dsh and Stbm during planar polarity signalling in Drosophila. This study used phosphorylation site mutations in Stbm to show that lack of Stbm phosphorylation leads to its clustering in 'mixed' puncta that contain complexes in both orientations. CKIε-dependent phosphorylation increases Stbm turnover at junctions, and thus promotes complex sorting, while phosphorylation of Dsh decreases its turnover. Pk negatively regulates Stbm phosphorylation and increases Stbm stability. These results support a direct role for Dco in phosphorylating both Stbm and Dsh in vivo in planar polarity signalling (Strutt, 2019).

This paper describes a dual role for CKIε/Dco kinase in regulating planar polarity in the fly pupal wing. In the first case, Dco promotes phosphorylation of Stbm. Stbm phosphorylation acts as a switch, changing Stbm from a stable immobile form that can enter junctional complexes, to an unstable mobile form that can redistribute within cells. Inhibiting Stbm phosphorylation causes an increase in Stbm stability at junctions that prevents sorting of complexes: thus complexes are 'locked' in an unsorted state. In contrast, hyperphosphorylation of Stbm destabilises Stbm, allowing it to leave junctions, hence permitting complex sorting. A second role for Dco is to mediate Dsh phosphorylation, which increases Dsh localisation at junctions. Significantly, the effects of Dco on Dsh are independent of Stbm and vice versa (Strutt, 2019).

In the 'cloud model', it is envisaged that multiple binding interactions drive a phase transition from a loosely packed, disordered association of core proteins in non-puncta, towards a highly cross-linked array of complexes within puncta that are all aligned in the same orientation. Stbm is well-placed to be a key component driving such a clustering mechanism, as not only can it multimerise with itself, but it also has a high stoichiometry within junctions. Also consistent with a role for Stbm in complex clustering is the observation that Stbm phosphorylation site mutants act as dominant negatives, recruiting wild-type Stbm into non-polarised puncta. Phosphorylation may inhibit a clustering mechanism, due to an increase in negative charge (Strutt, 2019).

Interestingly, excess clustering of unphosphorylated Stbm in unsorted complexes is also expected to lead to destabilising feedback interactions with the other core components. When Stbm is unphosphorylated, the increase in Stbm stability is sufficient for Stbm to 'win' over Fmi and Fz. Thus, there is an overall increase in Stbm stability in phosphomutant Stbm puncta, that is accompanied by decreased stability of Fmi and Fz (Strutt, 2019).

Pk both promotes Stbm stability and reduces its phosphorylation. A role for Pk in increasing Stbm stability is not surprising, as overexpression of Pk is known to cause excess clustering of the core proteins. A number of mechanisms can be envisioned by which Pk could affect Stbm phosphorylation. A previous study provided evidence that Pk has two roles: firstly, it acts via Dsh to destabilise Fz in the same cell (see Model for how Pk and phosphorylation of Stbm regulate complex sorting and clustering); secondly, it acts via Stbm to stabilise Fz in adjacent cells. In the first case, Pk would promote sorting of complexes, and one possibility is that Stbm is inaccessible to the kinase in sorted complexes, and thus Pk is indirectly reducing Stbm phosphorylation by promoting sorting. Arguing against this, loss of fz or dsh also abolishes core protein asymmetry, but no hyperphosphorylation is seen. The boundary FRAP experiments instead support Pk acting directly in the same cell to stabilise Stbm. A mechanism is therefore proposed whereby direct binding of Pk to Stbm protects Stbm from phosphorylation (Strutt, 2019).

Interestingly, Stbm has a significantly higher stoichiometry within junctions than Pk. One possibility is that Stbm forms multimers, and that association of Pk with these multimers causes a conformational change that reduces accessibility to kinase-binding sites. Alternatively, Pk might recruit a phosphatase (albeit no candidates for such a phosphatase are known). The reduced negative charge might then allow Stbm to form higher order structures, which promotes clustering of the entire core protein complex into puncta (Strutt, 2019).

Puncta formation in both wild-type and phosphomutants is also dependent on Dsh. Dsh is another a good candidate for promoting clustering as it too can multimerise, and thus puncta formation may be dependent on clustering on both sides of the complex. Moreover, direct interactions between Stbm and Dsh may promote clustering of unsorted complexes in the absence of phosphorylation (Strutt, 2019).

Feedback models for core protein asymmetry suggest that particular components of the core pathway signal to other components to either stabilise or destabilise them. An attractive model would be that Fz or Dsh recruits a kinase which phosphorylates Stbm and destabilises complexes of the opposite orientation. Consistent with this, a proportion of Dco localises to apicolateral junctions in pupal wings. However, no change was seen in Stbm phosphorylation in fz or dsh mutants, nor are Fz and Dsh required for the hyperphosphorylation of Stbm seen in pk mutants. Therefore, it is concluded that Stbm phosphorylation is more likely to be constitutive. Such constitutive phosphorylation would be sufficient to keep Stbm mobile and allow complex sorting; and Pk would then counterbalance this and promote complex stability. The balance between Stbm phosphorylation/complex mobility and Pk binding (leading to reduced Stbm phosphorylation) would resolve over time towards a more stable state as complexes segregate to opposite cell edges (Strutt, 2019).

It is noted that in normal development, Stbm downregulates Pk levels. This suggests Pk levels are finely tuned, in order to prevent unrestrained clustering (as seen when Pk is overexpressed) (Strutt, 2019).

Evidence is also provided that Dco regulates Dsh phosphorylation and junctional levels independently of Stbm. These findings are consistent with previous observations that Dsh phosphorylation correlates with its recruitment by Fz into junctional complexes. The mechanism by which Dsh phosphorylation acts in planar polarity remains to be elucidated, but the data show that &dco; overexpression phenotypes are suppressed by reduced dsh gene dosage, and that Dsh phosphomutants have reduced core protein asymmetry in pupal wings. Furthermore, a small but significant decrease in Dsh stability at junctions is observed in Dsh phosphomutants. Overall, these data are consistent with a model in which phosphorylation of Dsh promotes its stable association at junctions (Strutt, 2019).

In summary, it is proposed that Dco regulates the asymmetric localisation of the core proteins by reciprocal actions on Stbm and Dsh. Dco regulates Stbm phosphorylation and turnover and causes it to leave junctions, while phosphorylation of Dsh by Dco promotes its junctional association (Strutt, 2019).

Fan, J. Y., Agyekum, B., Venkatesan, A., Hall, D. R., Keightley, A., Bjes, E. S., Bouyain, S. and Price, J. L. (2013). Noncanonical FK506-binding protein BDBT binds DBT to enhance its circadian function and forms foci at night. Neuron 80(4): 984-996. PubMed ID: 24210908

Strutt, H. and Strutt, D. (2020). DAnkrd49 and Bdbt act via Casein kinase Iepsilon to regulate planar polarity in Drosophila. PLoS Genet 16(8): e1008820. PubMed ID: 32750048

DAnkrd49 and Bdbt act via Casein kinase Iepsilon to regulate planar polarity in Drosophila

The core planar polarity proteins are essential mediators of tissue morphogenesis, controlling both the polarised production of cellular structures and polarised tissue movements. During development the core proteins promote planar polarisation by becoming asymmetrically localised to opposite cell edges within epithelial tissues, forming intercellular protein complexes that coordinate polarity between adjacent cells. This study describes a novel protein complex that regulates the asymmetric localisation of the core proteins in the Drosophila pupal wing. DAnkrd49 (an ankyrin repeat protein) and Bride of Doubletime (Bdbt, a non-canonical FK506 binding protein family member) physically interact, and regulate each other's levels in vivo. Loss of either protein results in a reduction in core protein asymmetry and disruption of the placement of trichomes at the distal edge of pupal wing cells. Post-translational modifications are thought to be important for the regulation of core protein behaviour and their sorting to opposite cell edges. Consistent with this, it was found that loss of DAnkrd49 or Bdbt leads to reduced phosphorylation of the core protein Dishevelled and to decreased Dishevelled levels both at cell junctions and in the cytoplasm. Bdbt has previously been shown to regulate activity of the kinase Discs Overgrown (Dco, also known as Doubletime or Casein Kinase Iε), and Dco itself has been implicated in regulating planar polarity by phosphorylating Dsh as well as the core protein Strabismus. This study demonstrates that DAnkrd49 and Bdbt act as dominant suppressors of Dco activity. These findings support a model whereby Bdbt and DAnkrd49 act together to modulate the activity of Dco during planar polarity establishment (Strutt, 2020).

Planar polarity describes the phenomenon whereby cells coordinate their polarity in the plane of a tissue: for example the hairs on the skin point in the same direction, cilia coordinate their beating, and cells coordinate their movements during tissue morphogenesis. Understanding the mechanisms by which this coordinated polarisation occurs is of prime importance, as disruption of polarity can have diverse consequences, including neural tube closure defects, hydrocephalus and defects in neuronal migration (Strutt, 2020).

The fly wing is a well-characterised model system in which to study planar polarity. Each cell within the adult wing produces a single hair, or trichome, which points towards the distal end of the wing. Furthermore, viable mutations that cause characteristic swirling of the trichomes have been identified, and the genes associated with these mutations were subsequently found to be highly conserved, and to regulate planar polarity throughout the animal kingdom (Strutt, 2020).

The core planar polarity proteins (hereafter known as the core proteins) are the best characterised group of proteins that regulate planar polarity. In the pupal wing, the core proteins adopt asymmetric subcellular localisations prior to trichome emergence, and in their absence trichomes emerge from the centre of the cell rather than at the distal cell edge. The core proteins comprise the atypical cadherin Flamingo (Fmi, also known as Starry Night [Stan]), the transmembrane proteins Frizzled (Fz) and Strabismus (Stbm, also known as Van Gogh [Vang]), and three cytoplasmic proteins Dishevelled (Dsh), Prickle (Pk) and Diego (Dgo). Fmi localises to proximal and distal cell edges in the pupal wing, but is excluded from lateral cell edges, while Fz, Dsh and Dgo localise to distal cell edges and Stbm and Pk to proximal cell edges. These proteins form intercellular complexes at cell junctions, that couple neighbouring cells and allow them to coordinate their polarity (Strutt, 2020).

The mechanisms by which the core proteins become asymmetrically localised are poorly understood. The overall direction of polarity is thought to be determined by tissue-specific 'global' cues: these may include gradients of morphogens or Fat/Dachsous cadherin activity, or other cues such as mechanical tension. In the wing these global cues may directly regulate core protein localisation or act indirectly via effects on growth and tissue morphogenesis. Global cues are thought to lead to subtle biases in core protein localisation within cells that are subsequently amplified by feedback between the core proteins, in which positive (stabilising) interactions between complexes of the same orientation are coupled with negative (destabilising) interactions between complexes of opposite orientation. In mathematical models, such feedback interactions have been demonstrated to be sufficient to amplify weak biases in protein localisation, leading to sorting of complexes and robust asymmetry (Strutt, 2020).

Experimental evidence for feedback is only beginning to be elucidated. Cell culture experiments have demonstrated competitive binding between several of the core proteins, which may be important for feedback. Furthermore, it has recently been shown that the core protein Pk acts through Dsh to destabilise Fz in the same cell, while stabilising Fz across cell junctions via Stbm (Strutt, 2020).

In order for feedback to operate, cells must utilise the general cellular machinery: for example active endocytosis is necessary for Pk to destabilise Fz. Furthermore, post-translational modifications of the core proteins are likely to be key mediators of feedback. For example loss of ubiquitination pathway components and some protein kinases have been shown to disrupt planar polarity. In flies, a Cullin-3/Diablo/Kelch ubiquitin ligase complex regulates Dsh levels at cell junctions, while the de-ubiquitinase Fat Facets regulates Fmi levels. Stbm also negatively regulates Pk levels, and ubiquitination of Pk by Cullin-1/SkpA/Slimb promotes internalisation of Fmi-Stbm-Pk complexes. Similarly, in vertebrates, the Stbm homologue Vangl2 may promote local degradation of Pk via ubiquitination by Smurf E3 ubiquitin ligases. Furthermore, Drosophila Fz phosphorylation is mediated by atypical Protein Kinase C, and Dsh is a target of phosphorylation by the Discs Overgrown (Dco, also known as Doubletime [Dbt] or Casein Kinase Iε [CKIε]) and Abelson kinases. Dco/CKIε has also been implicated in phosphorylation of Stbm in both flies and vertebrates (Strutt, 2020).

This study describes the identification of two new regulators of planar polarity in the Drosophila wing. Bride of Doubletime (Bdbt) and DAnkrd49 were shown to be binding partners that regulate each other's levels. Loss of either protein disrupts asymmetric localisation of the core proteins. Furthermore, they regulate overall levels of Dsh and Dsh phosphorylation in the pupal wing, and evidence is provided that they act by modulating the activity of the kinase Dco (Strutt, 2020).

This study present evidence for a novel protein complex regulating planar polarity, consisting of the ankyrin repeat protein DAnkrd49 and the non-canonical FKBP family member Bdbt. DAnkrd49 and Bdbt physically interact in vitro and regulate each other's levels in vivo. This suggests that they may act in a complex in which each is required to stabilise the other, a model supported by the phenotypic similarity in mutant clones. Their activity is required for core protein asymmetry in the pupal wing, and for normal levels and phosphorylation of the cytoplasmic core protein Dsh (Strutt, 2020).

Previous work on circadian rhythms has established a physical interaction between Bdbt and the kinase Dco, and that Dco activity is regulated by Bdbt (Fan, 2013). Notably, Dco has previously been implicated in promoting both Dsh and Stbm phosphorylation in planar polarity signalling. Thus, the data are consistent with both DAnkrd49 and Bdbt acting to regulate Dco activity in planar polarity. Firstly, loss of function clones of dco are seen, and DAnkrd49/Bdbt share a common phenotype: in their absence reduced overall levels of Dsh and reduced levels of Dsh phosphorylation. Secondly, strong genetic interactions are seen between dco and DAnkrd49/Bdbt. Although a physical interaction has been observed between Dco and Bdbt, no direct interaction was seen between Dco and DAnkrd49. Therefore, a simple model is that the role of DAnkrd49 is to stabilise Bdbt, while Bdbt directly regulates Dco activity. Thus this study defines a regulatory cascade, whereby DAnkrd49 and Bdbt promote Dco activity, which in turn regulates phosphorylation of core proteins and asymmetric localisation (Strutt, 2020).

An alternative model is that DAnkrd49 and Bdbt act to stabilise Dsh, independently of Dco. Proteins of the FKBP family are known to regulate the stability of target proteins. In canonical FKBP family members this stabilisation is a result of PPIase activity, which assists protein folding; whereas other family members stabilise their target proteins by direct binding. The catalytic sites for PPIase are not conserved in Bdbt. A binding interaction between Dsh and either DAnkrd49 or Bdbt was not seen, arguing that Bdbt is unlikely to directly stabilise Dsh (Strutt, 2020).

In addition to defects in planar polarity, other pleiotropic defects are seen in DAnkrd49 clones and Bdbt clones, including cell size defects, reduced proliferation and poor viability. These could plausibly be explained by regulation of Dco activity by DAnkrd49 and Bdbt. Dco acts in multiple signalling pathways. It phosphorylates the tumour suppressor Fat, which regulates cell growth and survival via the Hippo signalling pathway. Interestingly, a hypomorphic mutation in dco causes tissue overgrowth and increased activity of the caspase inhibitor DIAP1, phenocopying loss of Hippo signalling pathway components, while cells completely lacking Dco activity have reduced expression of the caspase inhibitor DIAP1 and reduced proliferation. Dco may also act in additional signalling pathways, and these are thought to include Hedgehog signalling and canonical Wnt signalling. Hence, the multitude of signalling pathways regulated by Dco may explain the complex phenotypes seen in the absence of DAnkrd49 and Bdbt. Alternatively, it is possible that DAnkrd49 and Bdbt may regulate other downstream targets in addition to Dco (Strutt, 2020).

As Dco has also been implicated in phosphorylating Stbm, loss of DAnkrd49 and Bdbt is also expected to reduce Stbm phosphorylation in pupal wings. However this could not be verified directly, as Stbm mobility in SDS-PAGE is only marginally increased when Dco activity is reduced, and is not noticeably altered in extracts from animals with reduced activity of DAnkrd49 or Bdbt. Nevertheless, recent work has shown that phosphorylation of both Stbm and Dsh by Dco is functionally important in establishing correct planar polarity, making the regulation of Dco activity of great interest (Strutt, 2020).

Interestingly, protein interactions studies in human cell lines have identified a STRING network between Ankrd49, CKIδ/CKIε and FKBP family members. Although the FKBP proteins identified in these studies are not the closest orthologues to Drosophila Bdbt, this nevertheless suggests conservation of a regulatory cascade of Ankrd49/FKBP promoting Dco activity, which in turn might phosphorylate core planar polarity proteins. Furthermore, the DISEASES online resource, which integrates results from text mining, manually curated disease-gene associations and genome-wide association studies, has linked Ankrd49 with brachydactyly subtypes. Brachydactyly is a key feature of Robinow syndrome, a disease closely associated with core planar polarity mutations; thus understanding this regulatory cascade may be of importance for human health (Strutt, 2020).

GENE STRUCTURE

Exons - 4

PROTEIN STRUCTURE

Amino Acids - 440

Structural Domains

PROSITE searches have identified an ATP-binding site between amino acids 15-38 and a serine/threonine kinase catalytic domain between amino acids 124 and 136. BLAST searches reveal that DBT is most closely related to human casein kinase I epsilon, being 86% identical at the amino acid level over the length of the kinase domain. Significant homology to other kinases ends with amino acid 292 (Kloss, 1998).

Genomic and cDNA sequence analysis has shown that the transcribed portion of discs overgrown (dco) consists of four exons and three introns. The first intron frequently remains unspliced despite the use of two closely spaced alternative 5' donor splice sites. Surprisingly, all introns reside in the region corresponding to the untranslated leader, whereas the last exon contains the entire coding region. The longest open reading frame encodes a protein of 440 amino acids, of which the 300 N-terminal amino acids have a high level of sequence identity with the catalytic domain of members of the CKI family. The most closely related isoforms are human CKIepsilon and delta, with 86% sequence identity (both isoforms) and 92% or 91% similarity, respectively, through the kinase domain. Even when compared to CKI isoforms of lower eukaryotes, such as HRR25 of budding yeast (78% similarity) or Cki1 and Cki2 of fission yeast (68% similarity), the catalytic domain of Dco is highly conserved. The ultimate evidence for the extreme structural conservation of the catalytic domain of CKI proteins, at least among members of the CKIdelta/epsilon subfamily, comes from a comparison of their 3-D structures at 2 Angstrom resolution, which shows that the yeast and human isoforms are nearly identical. This allows for the interpretation of the effects of dco mutations in the catalytic domain in terms of these known 3-D structures from yeast and humans (Zilian, 1999 and references therein).

The 140 amino-acid C-terminal domain of Dco is not conserved among CKId/e subfamily members. However, these domains do share the feature of consisting predominantly of regions that are abnormally rich in a few amino acids. In addition, they encompass several potential phosphorylation sites, including a site for autophosphorylation and a short arg-rich stretch, possibly serving as a nuclear localization signal. A potential SH3- binding motif in the C-terminal domain of the CKIdelta/epsilon subfamily appears to be conserved in Dco (Zilian, 1999 and references therein).

date revised: 3 December 2023

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.