InteractiveFly: GeneBrief

Melted: Biological Overview | Regulation | Developmental Biology | Effects of Mutation | Evolutionary Homologs | References

BIOLOGICAL OVERVIEW

The insulin/PI3K signaling pathway controls both tissue growth and metabolism. Melted has been identified as a new modulator of this pathway in Drosophila. Melted interacts with both Tsc1 and Foxo and can recruit these proteins to the cell membrane. Evidence is provided that in the melted mutant, Tor activity is reduced and Foxo is activated. The melted mutant condition mimics the effects of nutrient deprivation in a normal animal, producing an animal with 40% less fat than normal (Teleman, 2005).

The insulin/PI3K signaling pathway acts via a series of protein relocalization and phosphorylation events to relay information from the cell's environment -- including growth-factor levels and nutrient availability -- into the cell where it controls gene expression (via Foxo) and protein translation (via Tor). Stimulation of the insulin receptor activates PI3K, which increases the level of phosphoinositol (3, 4, 5) triphosphate (PIP3) at the plasma membrane. PIP3 binding recruits the protein kinase Akt to the membrane, where it is activated through phosphorylation by PDK1. Insulin-regulated Akt acts through two effector pathways: winged helix-Forkhead family transcription factors and the protein kinase Tor. Tor regulates the translational machinery of the cell via two well-characterized effectors: S6K and 4E binding protein (4E-BP). S6K controls ribosome biogenesis and, thus, the biosynthetic capacity of the cell. 4E-BP binds to the translation factor eIF4E and prevents assembly of a protein complex that facilitates recruitment of the ribosome (Teleman, 2005).

In addition to mediating insulin-responsiveness, the Tor pathway also integrates information on cellular nutritional status and stress from the heterodimeric Tsc1/2 complex. Tsc2 serves as a GTPase-activating protein that inactivates Rheb and, thereby, reduces Tor activity. Tsc1/2 mutations result in Tor hyperactivation and tissue overgrowth. Mutants in mouse Tsc1 or Tsc2 have benign overgrowths called harmatomas and show increased susceptibility to tumor formation. Cellular AMP levels are sensed by AMP-activated protein kinase (AMPK) which phosphorylates and activates Tsc2 to inhibit Tor. Tor activity is also regulated by oxygen, through the hypoxia-induced transcription factor, HIF. HIF controls expression of REDD1/Scylla/Charybdis, which reduce Tsc1/2 activity (Teleman, 2005).

Mutations in the core components of the pathway -- the insulin receptor (InR), phosphoinositide-3-kinase (PI3K), Pdk1, Akt/PKB, tuberous sclerosis complex (Tsc1/Tsc2), Rheb, the target of rapamycin (Tor), and ribosomal protein S6 kinase (S6K) -- all cause tissue growth abnormalities or lethality. A considerable body of evidence indicates that the PI3K pathway controls metabolism as well as tissue growth. Flies and mice with reduced InR/IGF receptor and PI3K activity are small and have elevated fat levels. Mice mutant for Akt2 become insulin resistant and develop lipoatrophy as they age. Humans mutant for Akt2 also have abnormal metabolism and are 35% leaner than average. One output branch of the core pathway is via Foxo transcription factors. In response to insulin, Akt phosphorylates Foxo proteins: this promotes their interaction with 14-3-3 proteins and leads to cytoplasmic retention and inactivation. Foxo transcription factors have been implicated in the control of fat metabolism and lifespan in C. elegans, flies, and mice. The Tor branch of the pathway is also beginning to be implicated in fat metabolism. Tor phosphorylates and regulates S6K and 4E-BP. S6K mutant mice are resistant to diet induced obesity. 4E-BP1 mutant mice have a defect in fat metabolism. 4E-BP also controls fat metabolism in the fly. Recent studies have identified the eIF4E kinase LK6 as a modulator of growth and fat metabolism (Teleman, 2005).

A novel modulator of the insulin/PI3K pathway, Melted, has been identified. The melted gene encodes a PH domain protein that interacts with both Tsc1 and Foxo. Melted protein can recruit the Tsc1/2 complex to the cell membrane and thereby modulate its output via the Tor pathway. Melted can also recruit Foxo to the membrane in an insulin-regulated manner and thereby influence expression of Foxo targets. By reducing Tor activity and at the same time increasing Foxo activity, the melted mutant mimics the effects of nutrient deprivation in a normal animal, producing a lean phenotype (Teleman, 2005).

Melted contains two functional domains, identified as regions of high conservation between the fly and human proteins: an N-terminal protein interaction domain and a C-terminal PH domain. The PH domain targets Melted to the cell membrane. PH-GFP fusion proteins show sharp membrane localization, and fractionation assays show that in steady state, the majority of endogenous Melted protein is membrane associated. Rescue assays show that unlike the full-length protein, Melted protein missing the PH domain cannot rescue the melt^Δ1 mutant phenotypes. Thus, Melted protein requires its PH domain, and presumably its membrane localization, to have biological activity in vivo (Teleman, 2005).

On this basis, it is proposed that Melted may function as an adaptor, facilitating association of the TSC complex and Foxo with their upstream-signaling inputs. Foxo and Tsc2 are phosphorylated in vivo by Akt/PKB, which becomes activated at the cell membrane when it binds PIP3 and is phosphorylated by Pdk1. Although the phosphorylation of Tsc2 by PKB is not strictly necessary for viability, it is possible this phosphorylation is modulatory in function. Tsc2 is also regulated via phosphorylation by AMPK. AMPK is membrane associated through its myristoylated β subunit. Therefore, it is plausible that recruiting the Tsc complex and Foxo to the cell membrane might alter their state of activation. It is considered unlikely that Melted could sequester the TSC complex and Foxo proteins at the membrane. Instead, an intriguing possibility is that by transiently binding these proteins, Melted could facilitate their phosphorylation (Teleman, 2005).

Reduced Foxo phosphorylation is expected to increase Foxo activity, and indeed the following was found: (1) upregulation of the Foxo target 4E-BP in the melted mutant and (2) that reduced Foxo levels could suppress the fat-accumulation defect in the melted mutant. These observations indicate that Foxo activity is increased in the melted mutant. Similarly, reduced Tsc2 phosphorylation is expected to lead to increased Tsc2 activity and, thus, reduced Tor activity. Indeed, it was found that Tor activity becomes limiting for 4E-BP phosphorylation in the melted mutant (Teleman, 2005).

melt^Δ1 mutant animals are lean because of lower lipid levels in adipose tissue. This phenotype is autonomous to the fat body because it can be rescued by tissue-specific expression of Melted. Evidence is presented that elevated Foxo activity in the mutant is important in this context. Adipose tissue undergoes a dramatic transcriptional change upon loss of melted function. At the 99% confidence level, 249 genes were misregulated by more than 1.5-fold in the fat body. 63% of the misregulated genes for which a function is annotated are metabolism related. Although many of these are involved in lipid or carbohydrate metabolism, it was surprising to find a significant number of genes involved in proteolysis and amino acid metabolism. Recently, Foxo has been shown to promote protein turnover during fasting (Teleman, 2005).

Members of the Tor signaling pathway have been implicated in the control of fat metabolism in flies and mice. Both S6K and 4E-BP mutant mice are either lean or resistant to diet-induced obesity. Likewise, 4E-BP mutant flies show increased sensitivity to nutrient deprivation, leading to abnormal fat loss [Teleman, A.A., Chen, Y.-W., and Cohen, S.M. (2005). 4E-BP functions as a metabolic brake used under stress conditions but not during normal growth. Genes Dev., in press]. In the fly, activation of the Tor pathway in the fat body promotes fat accumulation. Because Tor activity is reduced in the melted mutant fat body, it is believed that this reduction contributes to the observed leanness. This idea is supported by the fact that inactivating Foxo in the melted mutant leads to a 90% rescue, whereas simultaneously inactivating Foxo and restoring Tor function (via PI3K) has a more pronounced effect, elevating fat levels to above normal. When the animal is under conditions of limited nutrition, in which Foxo activity is high, the effects of Foxo-dependent transcription appear to dominate and shift the animal to a mode of net fat consumption. When the animal is under normal feeding conditions, the ability of Tor to serve as a sensor of cellular nutritional status may be important in controlling fat accumulation (Teleman, 2005).

The effect of Melted on both tissue growth and metabolism is modulatory. In the absence of Melted protein, both tissue growth and lipid metabolism function, although with modified characteristics. The magnitude of the effect caused by Melted overexpression is smaller than that caused by Tsc1/2 loss of function. This phenotype is similar to those of other components of the PI3K/Tor pathways that have recently been studied in flies. For instance, Foxo mutant flies have no detectable growth abnormalities and are impaired in their response to oxidative stress. Whether melt^Δ1 mutant flies are also sensitive to oxidative stress was tested by feeding them food with H₂O₂, and indeed they are. Scylla and Charybdis are two genes upstream of the Tsc1/2 complex that regulate S6K activity. Scylla and Charybdis double-mutant flies show very mild growth defects but are impaired in their response to hypoxia and have abnormal lipid levels. Lk6 kinase, like 4E-BP, regulates eIF4E activity. Lk6 mutant flies are 20% smaller than controls and contain elevated lipid levels. Thus, the Tor pathway appears to be closely regulated by several modulators, Melted being a new member of this group (Teleman, 2005).

Melted is highly conserved between flies and mammals, and human Melted expression rescues the mutant fly phenotype. In view of the overall similarity in the biochemistry and biological functions of the insulin/PI3K pathway in flies and mammals, the possibility that Melted might have a comparable role in mammalian metabolism merits consideration (Teleman, 2005).

Opposite feedbacks in the Hippo pathway for growth control and neural fate

Signaling pathways are reused for multiple purposes in plant and animal development. The Hippo pathway in mammals and Drosophila coordinates proliferation and apoptosis via the coactivator and oncogene, YAP/Yorkie (Yki), which is homeostatically regulated through negative feedback. In the Drosophila eye, cross-repression between the Hippo pathway kinase, LATS/Warts (Wts), and growth regulator, Melted, generates mutually exclusive photoreceptor subtypes. This study show that this all-or-nothing neuronal differentiation results from Hippo pathway positive feedback: Yki both represses its negative regulator, warts, and promotes its positive regulator, melted. This postmitotic Hippo network behavior relies on a tissue-restricted transcription factor network-including a conserved Otx/Orthodenticle-Nrl/Traffic Jam feedforward module-that allows Warts-Yki-Melted to operate as a bistable switch. Altering feedback architecture provides an efficient mechanism to co-opt conserved signaling networks for diverse purposes in development and evolution (Jukam, 2013).

A fundamental strategy in animal development is to re-purpose the same signaling pathways for a diversity of functions. This study identified a tissue-specific transcription factor network that enables the otherwise homeostatic Hippo growth control pathway to act as a bistable switch for terminal cell fate. This alteration in network level properties -- such as positive vs. negative feedback -- within biochemically conserved pathways is an efficient means to re-use a signaling network in contexts as distinct as proliferation and terminal differentiation (Jukam, 2013).

How is the R8-specific Hippo regulatory circuit achieved? The two interlinked positive feedback loops (one with wts, one with melt) provide the R8 Hippo pathway with multiple points of potential regulation. Context-specific expression of wts and melt is defined by overlapping expression of four transcription factors: Otd, Tj, Pph13, and Sens. Otd and Pph13 are expressed in all photoreceptors and generate a permissive context that endows the initially equipotent R8s with the competence to become either subtype: Otd promotes melt/Rh5 whereas Pph13 promotes wts/Rh6 expression. This competence is further restricted by expression of Tj in R7 and R8, and Sens in R8s, which ensures that melt and wts cross-regulation is restricted to R8s. Importantly, it is the status of Yki activity and resulting feedback that assures the outcome of p vs. y fate: in pR8s, Yki functions with Otd and Tj to promote melt and Rh5; in yR8s, wts inhibits Yki, preventing melt and Rh5 expressionand allowing 'default' wts and Rh6 expression by Pph13 and Sens. Each of these four transcription factors regulates a partially overlapping subset of R8 subtype fate genes, and together, the network cooperates at multiple regulatory nodes to provide the specific context for repurposing the Hippo pathway (Jukam, 2013).

While other instances of pathways with both positive and negative feedback exist, these are conceptually different from R8 Hippo regulation. For example, in Sprouty (hSpry) regulation of Ras/MAPK-mediated EGFR signaling, EGFR induces hSpry2 expression but hSpry2 inhibits EGFR function (negative feedback); however, hSpry2 also promotes EGFR activity by preventing Cbl-dependent EGFR inhibition (positive feedback). hSpry2 positive feedback is likely coupled to its negative feedback to fine-tune the length and amplitude of receptor activation. In contrast, the opposite Hippo pathway feedbacks occur in vastly different cell types (mitotic epithelial cells vs. post-mitotic neurons), and both forms of feedback cannot co-exist in R8 since Yki's repression of wts expression (positive feedback) would make Yki up-regulation of Hippo regulators (negative feedback) inconsequential (Jukam, 2013).

Gaining positive feedback or losing negative feedback within Hippo signaling could permit oncogenesis. Indeed, the Yki ortholog, YAP, is an oncogene and is amplified in multiple tumors, and LATS1/2 (Wts) down-regulation is associated with non-small cell lung carcinomas, soft tissue sarcoma, metastatic prostate cancers, retinoblastoma, and acute lymphoblastic leukemia. Otx and MAF factors are also oncogenic in a number of tissues. Thus, understanding the regulatory networks identified here in other contexts will be crucial for deciphering how normal signaling pathways can go awry (Jukam, 2013).

The current findings also reveal that a Crx/Otd-Nrl/Tj feedforward module plays a conserved role in post-mitotic photoreceptor fate specification in both flies and mammals. Both induce one photoreceptor fate at the expense of another, and both regulate opsins with a feedforward loop wherein Crx/Otd activates Nrl/Tj expression and Crx-Nrl or Otd-Tj synergistically activate downstream targets. Given such deep evolutionary conservation, this module may be critical for generating photoreceptor diversity in other complex visual systems (Jukam, 2013).

This work has two main implications. First, although positive feedback is well documented in other switch-like, irreversible cell fate decisions such as in Xenopus oocyte maturation or cell cycle entry, this work suggests that positive feedback could have a broad role in terminal neuronal differentiation, which often requires permanent fate decisions to maintain a neuron's functional identity. Second, the changes in network topology in R8 photoreceptors allows a finely tuned growth control pathway to be used as a switch in a permanent binary cell fate decision. Context-specific regulation allows the feedback architecture to change in an otherwise conserved signaling module. This may be a general mechanism to endow signaling networks with new systems properties and diversify cell fates in development and evolution (Jukam, 2013).

Opposing transcriptional and post-transcriptional roles for Scalloped in binary Hippo-dependent neural fate decisions

The Hippo tumor suppressor pathway plays many fundamental cell biological roles during animal development. Two central players in controlling Hippo-dependent gene expression are the TEAD transcription factor Scalloped (Sd) and its transcriptional co-activator Yorkie (Yki). Hippo signaling phosphorylates Yki, thereby blocking Yki-dependent transcriptional control. In post-mitotic Drosophila photoreceptors, a bistable negative feedback loop forms between the Hippo-dependent kinase Warts/Lats and Yki to lock in green vs blue-sensitive neuronal subtype choices, respectively. Previous experiments indicate that sd and yki mutants phenocopy each other's functions, both being required for promoting the expression of the blue photoreceptor fate determinant melted and the blue-sensitive opsin Rh5. This study demonstrates that Sd ensures the robustness of this neuronal fate decision via multiple antagonistic gene regulatory roles. In Hippo-positive (green) photoreceptors, Sd directly represses both melt and Rh5 gene expression through defined TEAD binding sites, a mechanism that is antagonized by Yki in Hippo-negative (blue) cells. Additionally, in blue photoreceptors, Sd is required to promote the translation of the Rh5 protein through a 3'UTR-dependent and microRNA-mediated process. Together, these studies reveal that Sd can drive context-dependent cell fate decisions through opposing transcriptional and post-transcriptional mechanisms (Xie, 2019).

Ensuring that the correct complement of genes remains on or off in any given cell type is an essential feature of multicellular organisms. This is particularly critical in the peripheral nervous system, where exclusive sensory receptor expression is necessary for selective and specific activation of a given sensory neuron. Such exclusion is well-established in the visual system of most animals, where individual photoreceptors (PRs) express a single opsin photopigment and repress the expression of others to prevent sensory overlap. The gene regulatory mechanisms underlying this mutual exclusion, however, are still under investigation (Xie, 2019).

The Drosophila eye has long served as a powerful model to understand the functions and architecture of gene regulatory networks underlying PR subtype cell fate specification. Each of the approximately 750 individual eye units (ommatidia) in the Drosophila compound eye contains 8 PRs. Based on the specific opsin that is expressed in the R8 photoreceptor, two major ommatidial subtypes, pale (p) and yellow (y), are present in the adult eye. Pale ommatidia are primarily defined based on the expression of the blue-sensitive opsin, Rhodopsin 5 (Rh5), while yellow ommatidia express the green-sensitive opsin, Rh6. These ommatidial subtypes are randomly distributed through the eye in a 30:70 blue:green ratio, and are established and maintained through a bistable negative feedback loop between two signaling molecules: the pleckstrin homology-containing protein Melted (Melt) and the Hippo signaling kinase Warts (Wts, aka Lats) (Xie, 2019).

Wts is a core component of the Hippo kinase complex that phosphorylates and inactivates the transcriptional co-activator Yorkie (Yki). Hippo signaling is best understood in the context of growth regulation, where Wts and Yki function in a homeostatic feedback loop: Wts blocks Yki function and Yki initiates its own inactivation by promoting Hippo pathway gene expression. In contrast, in post-mitotic PR fate decisions, Yki promotes the expression of the wts repressor, melt, generating a double-negative 'on/off' feedback loop between wts and Yki that ensures two stably maintained fate choices. In green PRs, Hippo signaling promotes the expression of green fate determinants (wts and Rh6), and prevents the expression of Yki-dependent blue fate determinants (melt and Rh5). In blue PRs, Yki promotes melt, thereby repressing wts and inhibiting Hippo signaling, further promoting Yki-dependent activation of blue fate effectors and suppression of green fate effectors. Thus, Wts-positive (Yki-inactive) cells adopt the default green/wts/Rh6 fate, while Wts-negative (Yki-active) cells acquire the blue/melt/Rh5 fate (Xie, 2019).

Yki, a YES-associated protein (YAP), is a transcriptional co-activator that does not bind DNA itself, but instead requires a DNA-binding partner. The primary binding partners for Yki/Yap factors are members of the TEAD family of transcription factors. In Drosophila, the single TEAD family member is encoded by Scalloped (Sd). Sd/TEAD and Yki/YAP can physically interact and together activate TEAD-site-containing reporter expression in vitro. Furthermore, in ectopic yki conditions, sd/TEAD is essential for yki/YAP to induce tissue overgrowth and activate target gene expression. However, in vivo, sd mutants do not phenocopy yki growth phenotypes and sd mutants do not show changes in yki target gene expression. These data suggest that Sd and Yki use distinct mechanisms to control tissue size. Studies aimed at addressing this conundrum have shown that in developing wing, eye, and follicle cells, Sd functions as a transcriptional repressor under 'Hippo-on' conditions to inhibit cell growth, and that in 'Hippo-off' cells, Yki antagonizes Sd repression to promote growth regulatory genes. This suggests that Sd and Yki can play opposite roles during growth (Xie, 2019).

In post-mitotic PRs, it has been previously shown that sd mutants phenocopy yki's knockdown phenotype in PR subtype fate specification: both sd and yki are necessary to promote blue PR fate and inhibit green PR fate. Combined, these findings suggest that sd and yki function together in this cell fate specification event. This study investigated the molecular basis underlying this interaction. Sd was found to play roles at both the transcriptional and post-transcriptional level to ensure blue vs green PR subtype fate decisions. At the transcriptional level, Sd directly represses blue fate effector gene expression in Hippo (Wts)-positive green PRs, and Yki antagonizes this repression in Hippo (Wts)-negative blue PRs. This is consistent with previously reported antagonism between Sd and Yki. In addition to this function, it was found that Sd promotes blue fate through a post-transcriptional, microRNA (miRNA)-dependent process in Wts-negative blue PRs, revealing a cooperative interaction with Yki in promoting blue PR fate. Together, these new findings elucidate a multi-tiered regulatory network involving the Drosophila TEAD transcription factor that functions at both the transcriptional and post-transcriptional level to precisely specify neuronal subtype fate (Xie, 2019).

The mutually exclusive expression of sensory receptor genes in sense organs is essential to prevent sensory input overlap in the mature organism. This study shows that, in the fly retina, the TEAD factor Sd achieves this in blue and green PRs using two different mechanisms: direct transcriptional repression of the blue fate determinant melt and blue Rh5 opsin genes in green photoreceptors, and relief of post-transcriptional control of the Rh5 mRNA in blue photoreceptors. In addition, Yki, a major Sd cofactor, antagonizes Hippo-specific and Sd-dependent repression of melt and Rh5 to promote blue PR fate. Thus, Sd and Yki play multiple roles to ensure a robust bistable cell fate decision in post-mitotic sensory neurons (Xie, 2019).

The antagonistic relationship between Sd repression and Yki de-repression is similar to the model previously proposed in cell cycle control. Nevertheless, the mechanisms by which Sd represses gene expression in green PRs remains unknown. In cell growth, for instance, repression is mediated in part through Tgi, a Tondu domain containing protein, which Yki competes with to alleviate repression. However, no significant change was detected in Rh5 protein or reporter expression with knockdown of Tgi in PRs, suggesting the existence of another Sd co-repressor in this system. Indeed, a zinc finger protein Nerfin-1 was recently identified as a Tgi-independent Sd co-repressor that participates in Hippo-dependent cell growth and competition during Drosophila eye development (Guo, 2019). Preliminary studies showed that knockdown of nerfin-1 led to an expansion of Rh5-expressing blue PRs at the expense of green PRs, comparable to the expanded expression of Sd site mutants in the melt and Rh5 reporters. Therefore, Nerfin-1 is very likely to be at least one Sd co-repressor during blue- and green PR fate specification in the Drosophila eye. Combined, these findings suggest Sd repression activity is a general mechanism in controlling the output of the Hippo pathway (Xie, 2019).

If the role of Sd in green PRs were solely to repress Rh5 transcription, then Rh5 mRNA levels might be expected to be elevated in sd mutants relative to controls. Instead, a ~50% reduction was observed. This observation could reflect two possibilities, which are not mutually exclusive. First, based on previous and unpublished findings that Otd cooperates with Yki to activate Rh5 in Hippo-negative blue PRs, it is expected that in sd mutants, where all R8s switch to Hippo-positive (and hence Yki-inactive) green PRs, Rh5 activation in green PRs would be reduced. Second, since the current studies suggest a new role for miRNAs in the post-transcriptional control of Rh5, it is possible that Rh5 mRNA stability is affected in sdmutants (Xie, 2019).

In terms of the post-transcriptional control of Rh5, it was demonstrated that the Rh5 3' UTR was required to prevent its co-expression with Rh6 in sd knockdown green PRs. In addition, the simultaneous knockdown of sd and miRNA processing machinery genes led to Rh5 protein de-repression (and co-expression with Rh6) in a substantial subset of green R8 cells. Together, these data suggest miRNA-dependent regulation of Rh5 depends on Sd, either directly or indirectly. It is posited that, as a transcription factor, Sd prevents the transcription of Rh5-directed miRNA genes. However, follow-up studies will be important for defining the complete repertoire of miRNA-dependent events involved in this Hippo-directed cell fate decision. For example, possible differences in an pRh5 reporter and endogenous Rh5 protein were reported in retinas mutant for the transcription factor PvuII-PstI homology 13 (pph13). While this disparity could be due to the rhabdomere defects observed in pph13 mutants, there is potential for a role for Pph13 in Rh5 post-transcriptional regulation. Finally, it is possible that the Rh5 3'UTR recruits other non-coding RNAs or proteins to regulate its expression (Xie, 2019).

Combined, the bimodal functions of Sd in Yki-vs Wts-positive cells form a feedforward regulatory module in post-mitotic PR fate decisions, robustly preventing sensory receptor overlap. Feedforward modules between transcription factors and miRNAs have been previously reported in neuronal differentiation and other biological processes. For example, the proto-transcription factor c-Myc can directly activate E2F1 transcription, but also limit E2F1 translation by activating miR-175p and miR-20a. In contrast to the c-Myc-miRNAs-E2F1 activation module, which fine-tunes a proliferative signal in dividing cells, however, the Sd-miRNA-Rh5 repression module ensures a robust ON-OFF switch in the terminal PR differentiation process. If similar mechanisms take place during Hippo-dependent cell growth remains to be determined (Xie, 2019).

Whether yki is also involved in Sd's post-transcriptional control in blue PRs remains unresolved, as yki itself is essential for blue PR fate, and hence, Rh5-expressing cells. Previous studies have demonstrated that Yki is important for the activation of at least one miRNA to promote cell growth (i.e. bantam). However, in the case of Rh5 regulation, the miRNA must be repressed in Yki-expressing cells, rather than activated. In this context, it is worth noting that the Yki ortholog YAP has been shown to mediate widespread miRNA suppression in tumor cells (Hippo-negative) by sequestering an RNA helicase p72/DDX-17, a regulatory component of microRNA-processing machinery. Comparably, the results suggest that the miRNA(s) is/are inactive in Yki-positive blue PRs in order to allow Rh5 protein expression. These findings raise the possibility that YAP/Yki- and TEAD/Sd-dependent regulation of miRNA biogenesis is a universal mechanism in control of the Hippo signaling pathway in tissue growth and neuronal cell fate decisions (Xie, 2019).

REGULATION

Transcriptional Regulation

Opposite Feedbacks in the Hippo Pathway for Growth Control and Neural Fate

Signaling pathways are reused for multiple purposes in plant and animal development. The Hippo pathway in mammals and Drosophila coordinates proliferation and apoptosis via the coactivator and oncogene, YAP/Yorkie (Yki), which is homeostatically regulated through negative feedback. In the Drosophila eye, cross-repression between the Hippo pathway kinase, LATS/Warts (Wts), and growth regulator, Melted, generates mutually exclusive photoreceptor subtypes. This study shows that this all-or-nothing neuronal differentiation results from Hippo pathway positive feedback: Yki both represses its negative regulator, warts, and promotes its positive regulator, melted. This postmitotic Hippo network behavior relies on a tissue-restricted transcription factor network - including a conserved Otx/Orthodenticle-Nrl/Traffic Jam feedforward module - that allows Warts-Yki-Melted to operate as a bistable switch. Altering feedback architecture provides an efficient mechanism to co-opt conserved signaling networks for diverse purposes in development and evolution (Jukam, 2013).

A fundamental strategy in animal development is to re-purpose the same signaling pathways for a diversity of functions. This study identified a tissue-specific transcription factor network that enables the otherwise homeostatic Hippo growth control pathway to act as a bistable switch for terminal cell fate. This alteration in network level properties—such as positive versus negative feedback—within biochemically conserved pathways is an efficient means to re-use a signaling network in contexts as distinct as proliferation and terminal differentiation (Jukam, 2013).

How is the R8-specific Hippo regulatory circuit achieved? The two interlinked positive feedback loops (one with wts, one with melt) provide the R8 Hippo pathway with multiple points of potential regulation. Context-specific expression of wts and melt is defined by overlapping expression of four transcription factors: Otd, Tj, Pph13, and Sens. Otd and Pph13 are expressed in all photoreceptors and generate a permissive context that endows the initially equipotent R8s with the competence to become either subtype: Otd promotes melt/Rh5 whereas Pph13 promotes wts/Rh6 expression. This competence is further restricted by expression of Tj in R7 and R8, and Sens in R8s, which ensures that melt and wts cross-regulation is restricted to R8s. Importantly, it is the status of Yki activity and resulting feedback that assures the outcome of pR8/Rh5 vs. yR8/Rh6 (p vs. y) fate: in pR8s, Yki functions with Otd and Tj to promote melt and Rh5; in yR8s, wts inhibits Yki, preventing melt and Rh5 expression and allowing 'default' wts and Rh6 expression by Pph13 and Sens. Each of these four transcription factors regulates a partially overlapping subset of R8 subtype fate genes, and together, the network cooperates at multiple regulatory nodes to provide the specific context for repurposing the Hippo pathway (Jukam, 2013).

While other instances of pathways with both positive and negative feedback exist, these are conceptually different from R8 Hippo regulation. For example, in Sprouty (hSpry) regulation of Ras/MAPK-mediated EGFR signaling, EGFR induces hSpry2 expression but hSpry2 inhibits EGFR function (negative feedback); however, hSpry2 also promotes EGFR activity by preventing Cbl-dependent EGFR inhibition (positive feedback). hSpry2 positive feedback is likely coupled to its negative feedback to fine-tune the length and amplitude of receptor activation. In contrast, the opposite Hippo pathway feedbacks occur in vastly different cell types (mitotic epithelial cells versus post-mitotic neurons), and both forms of feedback cannot co-exist in R8 since Yki’s repression of wts expression (positive feedback) would make Yki up-regulation of Hippo regulators (negative feedback) inconsequential (Jukam, 2013).

Gaining positive feedback or losing negative feedback within Hippo signaling could permit oncogenesis. Indeed, the Yki ortholog, YAP, is an oncogene and is amplified in multiple tumors, and LATS1/2 (Wts) down-regulation is associated with non-small cell lung carcinomas, soft tissue sarcoma, metastatic prostate cancers, retinoblastoma, and acute lymphoblastic leukemia. Otx and MAF factors are also oncogenic in a number of tissues. Thus, understanding the regulatory networks identified here in other contexts will be crucial for deciphering how normal signaling pathways can go awry (Jukam, 2013).

The current findings also reveal that a Crx/Otd-Nrl/Tj feedforward module plays a conserved role in post-mitotic photoreceptor fate specification in both flies and mammals. Both induce one photoreceptor fate at the expense of another, and both regulate opsins with a feedforward loop wherein Crx/Otd activates Nrl/Tj expression and Crx-Nrl or Otd-Tj synergistically activate downstream targets (Hao, 2012). Given such deep evolutionary conservation, this module may be critical for generating photoreceptor diversity in other complex visual systems (Jukam, 2013).

This work has two main implications. First, although positive feedback is well documented in other switch-like, irreversible cell fate decisions such as in Xenopus oocyte maturation or cell cycle entry, this work suggests that positive feedback could have a broad role in terminal neuronal differentiation, which often requires permanent fate decisions to maintain a neuron’s functional identity. Second, the changes in network topology in R8 photoreceptors allows a finely tuned growth control pathway to be used as a switch in a permanent binary cell fate decision. Context-specific regulation allows the feedback architecture to change in an otherwise conserved signaling module. This may be a general mechanism to endow signaling networks with new systems properties and diversify cell fates in development and evolution (Jukam, 2013).

Protein Interactions

Melted binds to phosphoinositides

Tests were performed to see if bacterially expressed Melted protein binds to immobilized phosphoinositides. At higher concentration, Melted bound many phosphoinositides but at lower concentrations, showed preferential binding to PI(5)P. The PH domain of Melted expressed as a GST fusion exhibits a similar binding pattern, although with reduced selectivity. Although Melted does not show strong binding to PIP3 in vitro, a GFP fusion to the PH domain (MeltPH-GFP) was generated and tested in S2 cells for insulin-induced relocalization. MeltPH-GFP is detectable predominantly at the cell membrane when cells are serum-starved to deprive them of insulin and after insulin-stimulation without apparent difference. The MeltPH-GFP fusion is also predominantly cortical in wing imaginal disc cells, whereas a Melted-GFP fusion protein lacking the PH domain is cytoplasmic, suggesting that the majority of Melted protein is localized to the membrane under normal conditions. This was confirmed by subcellular fractionation of wing-disc cells. The majority of the endogenous Melted protein was recovered in the membrane fraction, with a lesser amount in the cytoplasm. No signal was detected in corresponding fractions prepared from melted mutant tissue, confirming the specificity of the antibody. Control proteins fractionated predominantly as cytoplasmic (Y14) or membrane-associated (PVR). These results suggest that the endogenous Melted protein is membrane localized via its PH domain and that its localization is not regulated by insulin-dependent changes in PIP3 levels. The PH domain is required for Melted function in vivo (Teleman, 2005).

Melted protein binds to Tsc1 and modulates TOR activity

To study the function of the Melted N-terminal domain, a yeast two-hybrid screen was performed with MeltDPH as bait to screen 2.7 million clones from a larval cDNA library. Six interacting genes were isolated: Tsc1 (four independent clones), 14-3-3e (three independent clones), CG16719, CG5171, CG6767, and CG8242 (one each). When expressed in S2 cells by cotransfection, Melted coimmunoprecipitated with Myc-tagged Tsc1. Interaction of Melted and Tsc1 in S2 cells was also visualized by immunofluorescence microscopy. Tsc1myc was uniformly distributed in the cytoplasm when expressed alone. Coexpression with Melted causes Tsc1-myc to shift predominantly to the cell membrane. When expressed in S2 cells, Tsc2 is uniformly distributed in the cytoplasm. Coexpression of Tsc2 with Melted does not cause a relocalization of Tsc2 to the membrane, suggesting that Melted and Tsc2 do not interact directly. However, when coexpressed with Melted and Tsc1, Tsc2 also relocates, indicating that Melted can recruit the Tsc1/2 complex to the cell membrane (Teleman, 2005).

Does this interaction affect Tor pathway activity? Previous reports have shown that increased PI3K or Tor signaling in the adipose tissue increases fat accumulation or reduces fat consumption (autophagy). The observation that melted mutants are lean, could be explained if Tor activity is reduced in the mutant. To test this more directly, the level of Tor-dependent phosphorylation of 4E-BP was assayed in control and melt^Δ1 mutant fat bodies. Phosphorylation of Drosophila 4E-BP at positions 36/47 depends on Tor activity and can be visualized with an antibody specific to the corresponding phosphorylated peptide from human 4E-BP. Although the total level of 4E-BP protein is higher in melt^Δ1 mutant tissue, the level of phosphorylation is not correspondingly elevated (perhaps even mildly reduced). The increase in total 4E-BP level reflects a 2- to 3-fold increase in transcript levels in the mutant fat body. 4E-BP levels were increased by a comparable amount in wild-type fat body, as a control, with ppl-Gal4 UAS-4E-BP and the level of 4E-BP phosphorylation was observed to be correspondingly elevated. This indicates that Tor activity is not limiting in wild-type fat body. Thus, the observation that 4E-BP phosphorylation does not increase, despite increased total 4E-BP, provides evidence that Tor activity is reduced in the melted mutant adipose tissue (Teleman, 2005).

Melted binds to Foxo and modulates Foxo activity

As a second means to identify possible functions of Melted, the Eukaryotic Linear Motif server) was used to look for functional motifs conserved between fly and human Melted. The only conserved motifs found in the N-terminal region of these proteins were two Forkhead-associated domain ligand domains (LIG_FHA_1). Forkhead transcription factors FoxA2, FoxA3, FoxC2, and FoxO1 are involved in glucose and fat metabolism. Insulin signaling activates Akt, which phosphorylates Foxo and leads to its retention in the cytoplasm. It was therefore asked if Melted affects the subcellular localization of a Foxo-GFP fusion protein. Foxo-GFP is predominantly nuclear in the absence of insulin stimulation in serum-starved S2 cells and increases in the cytoplasm after insulin stimulation. In serum-starved cells cotransfected to express Melted, Foxo-GFP is still primarily nuclear, but much of the nonnuclear protein appears at the membrane colocalized with Melted. Upon insulin stimulation, a robust increase in the level of Foxo-GFP was observed at the cell membrane. The interaction was confirmed by coimmunoprecipitation of Melted with Foxo in insulin-stimulated S2 cells (Teleman, 2005).

The observation that insulin stimulation induces a shift toward membrane localization of Foxo in the presence of Melted in S2 cells raised the possibility that melted regulates Foxo activity in vivo. To address this, expression of the Foxo target 4E-BP was examined in wild-type and melted mutant animals. Under fed conditions, insulin signaling is active and 4E-BP transcript levels are relatively low. In wild-type flies that were starved for 24 hr to reduce insulin levels and thereby activate Foxo, 4E-BP transcript increased ~4-fold. In starved flies lacking Melted, 4E-BP transcript increased over 25-fold. This increase in 4EBP transcription was absent in the starved melted/Foxo double mutant, confirming that it is Foxo dependent. Thus, in the absence of Melted, Foxo activity is higher than normal, suggesting that Melted limits Foxo activity in vivo (Teleman, 2005).

Melted acts via Tor and Foxo

To determine whether the elevated Foxo activity observed in melted mutants contributes to the lean phenotype of these animals, the normalized triglyceride levels of melted mutant and melted foxo double-mutant flies were compared. Reducing Foxo activity suppresses the leanness of the melted mutant to a considerable degree, reaching near normal fat levels. The rescue was highly statistically significant. foxo mutants did not show higher-than-normal fat levels compared to wild-type. These observations suggest that Melted acts by regulating Foxo activity to control expression of genes important in fat metabolism (Teleman, 2005).

melted mutants also exhibit reduced Tor activity. To determine whether Tor activity affects fat accumulation, the effects were tested of increasing Tor activity in wild-type and melted mutant adipose tissue. Use was made of a UAS-Tor transgene that can provide Tor activity in vivo when expressed at appropriate levels. It was confirmed that expression of UAS-Tor under ppl-Gal4 control in adipose tissue leads to increased total body fat, as does increasing PI3K activity. In contrast, a comparable elevation of Tor expression in melted mutant flies has no effect on fat levels. Both this result and the significant rescue caused by removal of Foxo indicate that in the melted mutant, the Foxo branch of the pathway becomes limiting for fat accumulation. In view of this finding, it was next asked whether elevated Tor pathway activity could increase fat levels in the melted mutant if Foxo activity was simultaneously reduced. To do so, use was made of the catalytic subunit of PI3K (Dp110) to inactivate Foxo and simultaneously activate Tor. The fat body driver lsp2-Gal4 or the UAS-Dp110 transgenes have little effect on their own in the melted mutant background, but when combined, the elevated PI3K activity in the fat body increases fat levels of the melted mutant. The effect is stronger than that of removing Foxo only, increasing fat levels to above normal. Taken together, these observations suggest that the Tor branch of the pathway contributes to the control of fat levels under conditions in which Foxo activity levels are low. This is normally the case in feeding animals in which insulin levels are relatively high (Foxo activity is elevated under starvation conditions: as seen by comparing 4E-BP levels in fed versus starved wild-type and foxo mutant flies). Under conditions in which insulin levels are low or in the melted mutant, in which Foxo activity is elevated, the effects of Foxo appear to dominate (Teleman, 2005).

DEVELOPMENTAL BIOLOGY

Embryonic

The melted (melt) gene was identified by a single revertible P-element insertion (melt^S144114) that results in abnormal morphology and mild loss of peripheral neurons (Salzberg, 1997). Use was made of plasmid-rescued fragments to clone two partially overlapping melt cDNAs. In addition, through database searches the HL03627 clone was identified as a candidate melt cDNA. The full-length sequence of the three clones was used to assemble a 2.903-kb melt cDNA that is incomplete at the 3'-end. BLAST searches with the sequence of Melt gave no significant results, except a limited homology to a predicted protein from C. elegans. In addition, Melt does not contain any functional domains or motifs. Early in embryogenesis (stage 5), melt RNA is expressed in 8 or 9 stripes and in the invaginating ventral furrow. During germ band extension, melt is expressed in discrete domains in each segment of the embryo. Later, this pattern is refined to several rows of ectodermal cells in the anterior of each segment. There are also low levels of expression in the brain and the gut. In conclusion, Melt is a novel protein of unknown function, which, on the basis of its expression pattern, may be required for ectodermal patterning (Prokopenko, 2000).

The suggestion that Melted is involved in ectodermal patterning should be re-explored in light of Teleman's demonstration (Teleman, 2005) of the involvement of Melted in the Tor pathway.

EFFECTS OF MUTATION

Melted was identified in a gain-of-function screen for genes affecting tissue growth during Drosophila development. EP31685 causes overgrowth when expressed in the posterior half of the wing with en-GAL4. When expressed ubiquitously with tubulin-GAL4, EP31685 caused a small but statistically significant increase in total body weight. EP31685 is located ~50 bp upstream of the annotated gene melted (melt, CG8624). Using a UAS-melt transgene, it was verified that the melted transcription unit is responsible for the tissue overgrowth phenotype (Teleman, 2005).

Deletions were prepared with P-element-mediated male recombination starting from EP31685 to generate melted mutants. melt^Δ1 is a 22-kb deletion that removes the entire melted gene and two adjacent genes. The second allele, melt^Δ2 is a 2.5-kb deletion that removes the first exon of CG8624. Flies homozygous mutant for melt^Δ1 are semiviable (70% viable to adult) and fertile but are ~10% smaller than control flies. Though small in magnitude, the reduction in body size is statistically significant. Flies heterozygous mutant for melt^Δ1/melt^Δ2 are also significantly smaller than control flies. Two additional lines of evidence indicate that the growth defect is due to loss of melted and not the other genes deleted in melt^Δ1: (1) the growth defect of the homozygous melt^Δ1 deletion flies is fully rescued by ubiquitous expression of the UAS-melt transgene; (2) flies expressing UAS-RNAi constructs directed against two different regions of the melt transcript cause tissue undergrowth, resembling the melt^Δ1 phenotype (Teleman, 2005).

It was next asked whether Melted might also affect fat metabolism. Drosophila store fat mainly as triglycerides. Total body triglyceride was measured for melt^Δ1 mutant and control flies reared under identical controlled conditions. When normalized to total protein to take into account the 10% reduced body size of melted mutants, melt mutant flies had only 60% as much triglyceride as control flies. This reduction in fat content was statistically significant and was rescued by ubiquitous expression of a UAS-melt transgene. Further confirmation that the leanness of the mutant is due to reduced melted expression was obtained by ubiquitous expression of a melted UAS-RNAi construct in flies (25% leaner). Total body triglycerides of wandering third instar melt^Δ1 mutant larvae were also 20% lower than controls, indicating that Melted regulates fat levels throughout development as well as in the adult (Teleman, 2005).

Fat levels can be controlled by humoral factors, including adipokinetic hormone (AKH), which induces mobilization of fat reserves. Insulin-like peptides (ILPs) also control fat metabolism in the fly. To determine whether elevated AKH or ILP-2, -3, and -5 expression could explain the leanness of melted mutants, transcript levels were tested by quantitative RT-PCR. Transcript levels were not significantly elevated. Because altered expression of the known humoral regulators did not provide an explanation for the mutant phenotype, it was asked if there was a defect in adipose tissue. The 'fat body' is the main fat-storage organ of the fly, containing over 80% of total body triglycerides. melt^Δ1 mutant and control fat body tissue was isolated from larvae; 25% lower triglyceride levels were found in the mutant tissue. The total body leanness of melt^Δ1 mutants was rescued by expressing melted specifically in the fat body with ppl-GAL4. In contrast, expression of melted in the nervous system with elav-GAL4 or in brain neurosecretory cells with dILP3-GAL4 does not rescue the mutant. This indicates that Melted activity is required in adipose tissue (Teleman, 2005).

Transcriptional profiling of melted mutant adipose tissue: the leanness observed in the mutant is likely due to reduced triglyceride accumulation in adipose tissue

To better understand the metabolic impact of the loss of Melted function in the adipose tissue, microarray expression profiling of melted mutant versus control fat body was performed. Use was made of microarrays containing 11,445 cDNA clones (DGC1 and DGC2 collection). Allowing for a 1% false-positive rate, 315 genes were identified that were upregulated and 405 genes that were downregulated in the melted mutant. This represents 6% of all genes sampled and reflects substantial reprogramming of the transcriptional profile of the adipose tissue. Within this set, genes altered by at least 1.5-fold (249 genes) were considered, and grouped according to their functional annotation. 46% of the regulated genes are involved in metabolism. Interestingly, many of these are involved in lipid metabolism. Ten are cytochrome P450 enzymes involved in the oxidation of lipophilic molecules. The transcriptional regulation of some of these genes was confirmed by semiquantitative RT-PCR. In addition to the genes involved in fat metabolism, a significant proportion of the misregulated metabolic genes are involved in protein degradation (Teleman, 2005).

Several of the downregulated genes are involved in the accumulation of triglycerides. One of the most highly downregulated genes in the melt^Δ1 mutant adipose tissue is the transcription factor sugarbabe (2.4-fold). sugarbabe was previously identified as a gene controlling the conversion of sugars to fats. It is the second most highly upregulated gene when flies were fed sugar but deprived of lipids, and it becomes expressed in the adipose tissue, gut, and Malphighian tubules. Acetyl-CoA sythase (AcCoAS) was also identified in the same screen as a gene upregulated on a sugar diet (7.3-fold), whereas in the melt^Δ1 mutant adipose tissue it was downregulated (1.8-fold). The downregulation of both sugarbabe and AcCoAS suggests that melt^Δ1 mutant adipose tissue might not be able to accumulate enough lipid. This is corroborated by the finding that glycerol kinase (Gyk) and phosphoenolpyruvate carboxykinase (PEPCK) are among the genes most downregulated in melted mutant adipose tissue. In order to generate triglycerides, both free fatty acids and 3-phosphoglycerol are required by the cell. In vertebrate brown adipose tissue, 3-phosphoglycerol is made by Gyk, whereas in white adipose tissue, it is made by PEPCK. PEPCK is rate limiting in that loss of function in the mouse leads to lipodystrophy, whereas overexpression in mouse adipose tissue leads to obesity. Therefore, the finding that both Gyk and PEPCK are downregulated in the melt^Δ1 mutant fat body suggests that these animals are lean because they do not accumulate enough lipid in the adipose tissue. Consistent with what is known in vertebrates, increased PEPCK levels were detected by quantitative RT-PCR in other nonadipose tissues under starvation conditions and in the melted mutant; thus, the transcriptional changes in adipose and nonadipose tissues do not always correlate. Recently, adipose triglyceride lipase has been reported to catalyze the initial step in triglyceride hydrolysis in mice and inhibition of this enzyme markedly decreases total adipose acyl-hydrolase activity. BLAST searches have identified two Drosophila homologs of adipose triglyceride lipase: CG5295 and CG5560. Interestingly, CG5295 expression is markedly reduced in the melt^Δ1 mutant fat body. This suggests that lipid hydrolysis might be downregulated in the mutant adipose tissue (Teleman, 2005).

To test experimentally these predictions from expression data, measurements were made of circulating lipids, which, in the fly, are mobilized from the fat body and delivered to peripheral tissues as diacylglycerides (DAG) in the hemolymph. Hemolymph DAG was low in melt^Δ1 mutant larvae compared to controls. Thus, it is not likely that the reduced triglycerides in the adipose tissue can be explained by increased mobilization of fat in the form of circulating DAG. The level of circulating blood sugar (trehalose + glucose) was also measured and it was found to be not elevated in the melt^Δ1 mutant. These experiments, together with the expression data, suggest that the leanness observed in the mutant is likely due to reduced triglyceride accumulation in adipose tissue (Teleman, 2005).

The growth regulators warts/lats and melted interact in a bistable loop to specify opposite fates in Drosophila R8 photoreceptors

Color vision in Drosophila relies on the comparison between two color-sensitive photoreceptors, R7 and R8. Two types of ommatidia in which R7 and R8 contain different rhodopsins are distributed stochastically in the retina and appear to discriminate short (p-subset) or long wavelengths (y-subset). The choice between p and y fates is made in R7, which then instructs R8 to follow the corresponding fate, thus leading to a tight coupling between rhodopsins expressed in R7 and R8. warts, encoding large tumor suppressor (Lats) and melted, encoding a PH-domain protein, play opposite roles in defining the yR8 or pR8 fates. By interacting antagonistically at the transcriptional level, they form a bistable loop that insures a robust commitment of R8 to a single fate, without allowing ambiguity. This represents an unexpected postmitotic role for genes controlling cell proliferation (warts and its partner hippo and salvador) and cell growth (melted) (Mikeladze-Dvali, 2005b).

The fly eye provides a powerful system to study cell-fate decisions: it develops from a flat epithelium into a complex three-dimensional structure of multiple cell types in less than a week. The adult eye allows the fly to perform various visual tasks, ranging from motion detection and the discrimination of colors to measuring the orientation of polarized light for navigation (Mikeladze-Dvali, 2005b).

In the fly compound eye, each of the 800 ommatidia is a single optical unit that contains 8 photoreceptor cells (PRs). The 8 PRs form widely expanded membrane structures, the rhabdomeres, which contain the photosensitive Rhodopsins (Rh). The rhabdomeres of the six outer PRs (R1-R6) form a trapezoid. R1-R6 all express the broad spectrum rhodopsin1 (rh1 or ninaE) and are morphologically and functionally invariant in all ~800 ommatidia (Mikeladze-Dvali, 2005b).

The center of the trapezoid is occupied by the two inner PRs, R7 and R8. The rhabdomeres of R7 are positioned on top of R8, so that they share the same optic path. Inner PRs are involved in color vision and can be viewed as equivalent to vertebrate cones. Each R7 and R8 expresses only one of the four rhodopsins, rh3, rh4, rh5, or rh6 in a highly regulated manner, defining three different subtypes of ommatidia: 'yellow' (y), 'pale' (p) (for their appearance under UV illumination), and the 'dorsal rim area' (DRA). Ommatidia in the DRA express rh3 in both R7 and R8 and are specified in a very restricted region by the gene homothorax. They are believed to function as polarized light detectors (Mikeladze-Dvali, 2005b).

In contrast, color vision depends on the y and p ommatidial subtypes that are randomly distributed through the main part of the retina, with a bias of y (~70%) over p subtype (~30%). In the p subtype, R7 expresses the UV-sensitive Rh3 and R8 the blue-sensitive Rh5. In the y subtype, R7 expresses a distinct UV-sensitive Rh4 while R8 expresses the green-sensitive Rh6. As in many other sensory systems, expression of a given Rhodopsin excludes all others to prevent sensory overlap. While the p subtype is better suited to discriminate among shorter wavelengths, the y subtype should discriminate amongst longer wavelengths (Mikeladze-Dvali, 2005b).

The choice between the p and y fate is first made in R7: once an R7 commits to the p fate and expresses rh3, it sends an instructive signal to the underlying R8, which then also commits to the p fate and expresses rh5. In the absence of the R7 signal (i.e., when R7 expresses rh4 or in a sevenless mutant), R8 commits to the y fate and expresses rh6. The stochastic choice appears to be made by each R7 independently of its neighbors, resulting in the biased random distribution of p and y ommatidia throughout the main part of the retina (for review see Mikeladze-Dvali, 2005a).

Four genes required in R8 cells for ensuring the correct choice of y versus p cell fate have been identified. The warts (wts) gene, which encodes the Drosophila large tumor suppressor (also known as lats) and melted (melt) play a critical role in the specification of p and y R8 cells, without affecting the R7 choice. wts encodes a Ser/Thr kinase, while melt encodes a Pleckstrin Homology (PH) domain protein. wts is necessary and sufficient for R8 to adopt the y fate, while melt plays the opposite role and specifically induces the p fate in R8. wts and melt are expressed in a complementary manner in the yR8 and pR8 subsets, respectively. Evidence is presented that the two genes repress each other's transcription to form a bistable loop. melt seems to respond to the R7 signal, while wts appears to regulate the output of the loop. The tumor-suppressor genes hippo (hpo) and salvador (sav), which encode the two molecular partners of Wts/Lats, have phenotypes identical to wts. Interestingly, melt has been reported to regulate growth and fat metabolism in Drosophila. Thus, genes known to regulate both cell growth (melt) and proliferation (wts, sav, hpo) interact antagonistically during retinal patterning (Mikeladze-Dvali, 2005b).

To identify genes involved in the differentiation of p or y PR subsets, a Gal4 (pGawB) enhancer trap screen was performed in adult flies using GFP expression as a reporter. One insertion produced a strong GFP signal in inner PRs. Staining of sectioned adult eyes for the UAS-lacZ reporter gene revealed Gal4 expression in a large subset of R8 cells. Additional expression was found in DRA R7 and R8, as well as in outer PRs in the ventral half of the eye. Occasionally, weak expression was also found in some R7 cells, but not in any PR subset-specific pattern. Staining of the same enhancer trap (driving UAS-lacZnuc expression) with antibodies against β-Gal, Rh6 (α-Rh6), and Rh5 (α-Rh5) in whole-mounted retinas revealed that the reporter was specific to Rh6-positive R8 and was excluded from the Rh5-positive R8, indicating that the targeted gene is expressed in the yR8 subtype (Mikeladze-Dvali, 2005b).

The genomic DNA flanking the pGawB transposon, which is inserted upstream of the third exon of the gene warts (wts), was identified. An existing wts nuclear lacZ enhancer trap line P[lacZ,w⁺] was stained. lacZ expression in this line (wtsZn) was also specific to the y subset of R8 cells as well as the DRA and some ventral outer PRs, confirming the restricted expression pattern of wts (Mikeladze-Dvali, 2005b).

wts-Gal4 appears to be activated by a late eye-specific enhancer of wts, which first directs expression long after R8 has exited the cell cycle. wts therefore appears to play two distinct roles: a ubiquitous role in proliferating cells and a more restricted role in terminally differentiated PR (Mikeladze-Dvali, 2005b).

Flies with wts-Gal4 insertion were homozygous viable and did not exhibit any visible growth phenotype. However, it was noticed that heterozygous wts-Gal4 flies always exhibited a strong rh phenotype when present in combination with one specific UAS-lacZ reporter construct (P w[+mC] = UAS-lacZ.B Bg4-2-4b, FlyBase #1777). The y/p R8 ratio was dramatically affected: most R8 expressed rh5, while rh6 expression was almost completely lost, with wts-Gal4 expression reduced to the remaining rh6 expressing R8. However, specification of R7 and of outer PRs was unaffected. This phenotype was only observed with this specific UAS-lacZ transgene, and not with UAS-GFP or other UAS-lacZ transgenes. When homozygous (in the absence of wts-Gal4), this UAS-lacZ line manifested an even more severe R8 opsin phenotype: about 90% of R8 expressed rh5 at the expense of rh6. This suggested that this particular insertion disrupted a gene affecting the p/y choice in R8 (Mikeladze-Dvali, 2005b).

This UAS-lacZ P element was found to be inserted 21 bp upstream of the transcriptional start site of the gene melted (melt). The Melt protein has a C-terminal PH domain and is conserved from C. elegans to humans. Insertions in melt were initially identified in a screen for genes affecting peripheral nervous system development. Thus the role of melt in R8 subtype specification and its interaction with wts was examined (Mikeladze-Dvali, 2005b).

Since R7 and R8 in a given ommatidium share the same optic path, their fates must be tightly regulated. The decision of a given ommatidium to become y or p is initially made by R7. Once R7 has chosen its fate, it imposes it onto the underlying R8. To coordinate opsin expression between R7 and R8, R8 has to respond to the R7 signal with high fidelity (Mikeladze-Dvali, 2005b).

This study shows that wts and melt act in R8 to prevent an ambiguous response to the instructive R7 signal. wts and melt play opposite roles in the specification of R8 subtypes. In the absence of wts, the yR8 subtype is completely misspecified into pR8. By contrast, in melt mutants, the pR8 subtype is lost with expansion of yR8. Overexpression of wts or melt leads to the transformation of all R8 into the y or p fate, respectively. The complementary expression patterns of the two genes in y or p R8 subtypes are set up in response to the pR7 signal. Therefore, wts and melt appear to interpret the signal from R7, and mutations in wts and melt render R8 insensitive to this signal without influencing R7 or outer PR (Mikeladze-Dvali, 2005b).

The decision to express wts or melt in R8 is determined by R7, but the two genes repress each other's transcription. Thus, wts and melt act in a loop of negative crossregulation. However, if R7 imposes its fate upon R8, what then is the role of this crossregulation? It is suggested that the bistable loop allows only an unambiguous readout while R7 provides an asymmetric bias of this choice (Mikeladze-Dvali, 2005b).

In a negative bistable crossregulatory loop, the input signal biasing cell-fate choice might act at any level. Similarly, any member of the loop can serve as the output. For instance, wts could positively regulate rh6 expression (yR8 fate), while melt could activate rh5 (pR8 fate). Double misexpression and double loss-of-function experiments suggest that wts is the output regulator of the loop. When both wts and melt are ectopically expressed, all R8 acquire the y fate, i.e., the fate imposed by wts. In melt, wts double mutants, all R8 acquire the p fate. These phenotypes resemble the single gain- or loss-of-function phenotypes of wts, which appears to be necessary and sufficient for rh6 expression. In contrast, while melt is sufficient to induce rh5 in yR8, rh5 remains expressed in the absence of melt in the double mutant. This argues that melt is not necessary for the pR8 fate (rh5). In melt, wts double-mutant eyes, rh5 does not depend on instruction from pR7, which confirms that rh5 expression is a consequence of the absence of wts (a derepression rather than activation by the pR7 signal) (Mikeladze-Dvali, 2005b).

The following model is proposed: in the absence of an instructive pR7 signal, i.e., in y ommatidia, the loop is biased in favor of wts expression, which represses melt. In p ommatidia, the R7 signal either induces melt expression in R8 or represses expression of wts in R8. In either case, the balance of the loop is shifted, leading to upregulation of melt and complete suppression of wts expression. This system is able to amplify a weak or transient signal to ensure that the cell-fate decision is made unambiguously (Mikeladze-Dvali, 2005b).

There are clearly a number of examples of bistable loop that often reinforce stochastic decisions or transient differentiation stimuli. Bistable systems require positive feedback loops as proposed for the BMP signaling during dorso-ventral patterning in Drosophila or double-negative feedback loops as in the case of the wts-melt loop. The left-right choice by chemosensory ASE neurons in C. elegans is a similar example where a negative bistable loop is involved in making an unambiguous cell-fate decision. This loop includes two transcription factors and two microRNAs. In the left ASE, this loop is strongly biased toward Na⁺-sensitive fate and in the right ASE, toward Cl⁻ sensitivity (Johnston, 2005). This strong bias is likely imposed by a factor outside of the loop. In R8 cells, the wts-melt loop is inherently biased toward y fate. The signal from R7 in p ommatidia biases the choice toward the pR8 fate. The transcription loop described in this study is clearly incomplete since neither Wts nor Melt is a transcription factor. A mutation, daltonien (don), has been identified which genetically interacts with melt, activates the expression of melt (unpublished data cited in Mikeladze-Dvali, 2005b), and appears to encode a component of this loop. Another potential member of the loop is the newly identified transcriptional coactivator Yorkie (Yki), a direct target of the Wts kinase (Mikeladze-Dvali, 2005b).

The bistable loop is specific to those R8 that are involved in color vision: in DRA ommatidia, melt misexpression does not lead to wts downregulation. This is not surprising since R7 and R8 in DRA are specified independently by positional information and do not appear to communicate (Mikeladze-Dvali, 2005b).

The transcriptional regulation of wts and melt expression is surprising, since kinases and PH domain proteins are usually regulated by changes in their activity or subcellular localization. For instance, Wts/Lats kinase activity is regulated through phosphorylation by Hpo in the presence of Sav. However, the nature of the signal that triggers activation of the Wts/Hpo/Sav proliferation control pathway has remained elusive. Thus, identification of the signal from pR7 to R8 could provide important insights into the mechanism by which this tumor-suppressor complex is regulated to control proliferation and cell death (Mikeladze-Dvali, 2005b).

The ability of wts to indirectly regulate transcription of other genes (here melt) is less surprising. wts, sav, and hpo have been reported to negatively regulate the transcription of Cyclin E and DIAP1, leading to a decrease in cell cycle progression and to an increase in cell death. The same (unknown) transcription factor required downstream of wts could therefore also play a role in repressing melt and rh5, and possibly in activating rh6 (Mikeladze-Dvali, 2005b).

Cbk1, the Lats/Wts homolog in S. cerevisiae has been shown to regulate a broad range of daughter specific genes during budding. The asymmetric gene expression between mother and daughter cells is due to Cbk1-dependent activation and nuclear localization of the transcription factor Ace2 in daughter cells. Cbk1 kinase activity requires another gene, Mob2. Recently, a member of the Mob family in Drosophila, Mats, has been shown to bind and synergistically interact with Wts/Lats to control proliferation and apoptosis. Although Melt is not known to regulate the transcription of other target genes, it can affect subcellular localization of FOXO and the TSC1/TSC2 complex to regulate fat metabolism. However, the members of the TOR or InR do not seem to play a role in the specification of R8 subtypes (Mikeladze-Dvali, 2005b).

Wts, together with the Ser/Thr kinase Hpo and the adaptor protein Sav, acts as a potent tumor suppressor. All three genes play a critical role for the establishment of the R8 subtypes. The function described in this study for hpo/sav/wts represents an unexpected new role unrelated to their tumor-suppression function: R8 PRs have exited the cell cycle for at least 4 days when they choose to express a particular rhodopsin, and these cells are not prone to die (PRs are particularly difficult to kill through induction of the cell death pathway). Furthermore, there is no detectable difference in cell size or shape between y and p R8, which specifically express or exclude wts or melt expression. However, it is interesting to note that p and y inner photoreceptors are morphologically distinguishable in Calliphora blowflies. Perhaps Wts and Melt represent an evolutionary remnant of a system in large flies where subtypes required different morphologies. Therefore, specification of the correct R8 fate utilizes two signaling cassettes used for different purposes earlier in development, after these cassettes are no longer in use in these highly differentiated PR cells (Mikeladze-Dvali, 2005b).

Lats1, the human ortholog of Wts, is able to rescue the lethality of wts in flies. Canine Lats1 splice variant is specifically expressed in the retina. Moreover, a gene responsible for an autosomal dominant cone dystrophy (involving impaired color vision, sensitivity to light, and gradual loss of visual activity) has been mapped close to the Lats1 locus. Thus, it might be expected that the hpo/sav/wts pathway functions in the human retina as well. Although, melt knockout mice are viable and fertile, it will be interesting to test whether they are defective in cone differentiation or vision (Mikeladze-Dvali, 2005b).

REFERENCES

Search PubMed for articles about Drosophila Melted

Guo, P., Lee, C. H., Lei, H., Zheng, Y., Pulgar Prieto, K. D. and Pan, D. (2019). Nerfin-1 represses transcriptional output of Hippo signaling in cell competition. Elife 8. PubMed ID: 30901309

Hao, H., Kim, D. S., Klocke, B., Johnson, K. R., Cui, K., Gotoh, N., Zang, C., Gregorski, J., Gieser, L., Peng, W., Fann, Y., Seifert, M., Zhao, K. and Swaroop, A. (2012). Transcriptional regulation of rod photoreceptor homeostasis revealed by in vivo NRL targetome analysis. PLoS Genet 8: e1002649. PubMed ID: 22511886

Johnston, R. J., et al. (2005). MicroRNAs acting in a double-negative feedback loop to control a neuronal cell fate decision, Proc. Natl. Acad. Sci. 102: 12449-12454. 16099833

Jukam, D., Xie, B., Rister, J., Terrell, D., Charlton-Perkins, M., Pistillo, D., Gebelein, B., Desplan, C. and Cook, T. (2013). Opposite Feedbacks in the Hippo Pathway for Growth Control and Neural Fate. Science. PubMed ID: 23989952

Mikeladze-Dvali, T., Desplan, C and Pistillo, D. (2005a). Flipping coins in the fly retina. Curr. Top. Dev. Biol. 69: pp. 1-14. 16243594

Mikeladze-Dvali, T., et al. (2005b). The growth regulators warts/lats and melted interact in a bistable loop to specify opposite fates in Drosophila R8 photoreceptors. Cell 122: 775-787. 16143107

Prokopenko, S. N., He, Y., Lu, Y. and Bellen, H. J. (2000). Mutations affecting the development of the peripheral nervous system in Drosophila: a molecular screen for novel proteins. Genetics 156(4): 1691-715. 11102367

Salzberg, A., et al. (1997). P-element insertion alleles of essential genes on the third chromosome of Drosophila melanogaster: mutations affecting embryonic PNS development. Genetics 147(4): 1723-41. 9409832

Shathasivam, P., Kollara, A., Ringuette, M. J., Virtanen, C., Wrana, J. L. and Brown, T. J. (2015). Human ortholog of Drosophila Melted impedes SMAD2 release from TGF-beta receptor I to inhibit TGF-beta signaling. Proc Natl Acad Sci U S A 112: E3000-3009. PubMed ID: 26039994

Teleman, A. A., Chen, Y. W. and Cohen, S. M. (2005). Drosophila Melted modulates FOXO and TOR activity. Dev Cell 9(2): 271-81. 16054033

Xie, B., Morton, D. B. and Cook, T. A. (2019). Opposing transcriptional and post-transcriptional roles for Scalloped in binary Hippo-dependent neural fate decisions. Dev Biol. PubMed ID: 31265830

date revised: 20 December 2019

Home page: The Interactive Fly © 2017 Thomas Brody, Ph.D.