The Drosophila Ndr serine/threonine kinase Tricornered (Trc) is required for the normal morphogenesis of epidermal hairs, bristles, laterals, and dendrites. Mutations in trc and its interacting factor furry (fry) cause cells to form ectopic wing hairs, implying that these genes function temporally before hair initiation (Geng, 2000; Cong, 2001). In vivo evidence has been obtained that Trc function is regulated by phosphorylation and that mutations in key regulatory sites result in dominant negative alleles. Wild-type, but not mutant Trc, is found in growing hairs, and no Trc has been detected in pupal wing nuclei, implying that in this developmental context Trc functions in the cytoplasm. The furry gene and its homologues in yeast and Caenorhabditis elegans have been implicated as being essential for the function of the Ndr kinase family. Drosophila furry is also found in growing hairs, its subcellular localization is dependent on Trc function, and it can be coimmunoprecipitated with Trc. These data suggest a feedback mechanism involving Trc activity that regulates the accumulation of Fry in developing hairs (He, 2005).

Tricornered is involved in a repulsion mechanism regulating the distribution of dendrites in dendritic fields. To cover neuronal receptive fields completely but without redundancy, neurons of certain functional groups exhibit tiling of their dendrites via dendritic repulsion. Two evolutionarily conserved proteins, the Tricornered and Furry, are essential for tiling and branching control of Drosophila sensory neuron dendrites. Dendrites of fry and trc mutants display excessive terminal branching and fail to avoid homologous dendritic branches, resulting in significant overlap of the dendritic fields. Trc control of dendritic branching involves regulation of RacGTPase, a pathway distinct from the action of Trc in tiling. Time-lapse analysis further reveals a specific loss of the ability of growing dendrites to turn away from nearby dendritic branches in fry mutants, suggestive of a defect in like-repels-like avoidance. Thus, the Trc/Fry signaling pathway plays a key role in patterning dendritic fields by promoting avoidance between homologous dendrites as well as by limiting dendritic branching (Emoto, 2004).

Dendritic tiling refers to the complete but nonredundant coverage of a receptive field by dendrites of functionally homologous neurons, like tiles covering a floor. Tiling has been well characterized in the mammalian retina. Retinal ganglion cells (RGCs) in the rabbit retina, for example, can be grouped into at least 11 distinct physiological classes. Dendrites of some RGCs of the same subtype typically cover the whole retina with minimal overlap, whereas dendrites of different subtypes overlap extensively. Similarly, amacrine cells in the rabbit retina are classified into at least 22 subclasses based on the branching pattern of their dendrites, and several subtypes appear to tile the retina. Tiling thus ensures efficient and unambiguous representation of the entire visual field and is likely to be of general importance. Indeed, dendritic tiling among sensory neurons of moths and flies suggest that tiling is an evolutionarily conserved mechanism for dendritic field organization (Emoto, 2004 and references therein).

Drosophila dendrite arborization (da) sensory neurons provide a suitable system to study cellular and molecular mechanisms underlying dendritic development. The 15 da neurons in each abdominal hemisegment are classified into four subtypes based on their unique dendritic arborization profiles (Grueber, 2002). In addition, class III and class IV da neurons exhibit tiling in a subtype-specific manner (Grueber, 2002; Sugimura, 2003). Several lines of evidence suggest that the dendritic tiling of class IV neurons arises from class-specific competition between dendrites of neighboring neurons. (1) Dendrites of class IV neurons often appear to stop growing, make a turn, or retract as they come within a short distance of each other, whereas there is extensive overlap between dendrites of neurons belonging to different classes (Grueber, 2002, 2003a; Sugimura, 2003). (2) Laser ablation of certain class IV neurons during late embryonic stages causes dendrites of the surrounding class IV neurons to grow into the territories of the ablated neurons (Grueber, 2003a; Sugimura, 2003). Conversely, duplication of a class IV neuron results in a reduction rather than any overlap of their respective fields (Grueber, 2003a). (3) Despite the dendritic overgrowth and/or overbranching in mutants, such as flamingo and sequoia, there is still tiling between the extra branches within the same neuron as well as between different neurons (Grueber, 2002). These observations suggest that a like-repels-like mechanism is responsible for the dendritic tiling of class IV neurons; however, the underlying molecular mechanisms remain unknown (Emoto, 2004 and references therein).

tricornered (trc) and furry (fry) are evolutionally conserved genes implicated in regulating cell morphology (Adler, 2002). The trc and fry genes encode a serine/threonine kinase of the ACG family (Tamaskovic, 2003b) and a large (380 kDa) protein with no known functional domain (Cong, 2001), respectively. In budding yeast, the Trc homolog Cbk1p promotes nuclear translocation of the Ace2p transcription factor, which controls the daughter cell-specific expression of cell separation genes (Colman-Lerner, 2001; Weiss, 2002). Cbk1p also controls polarized cell growth through an Ace2p-independent mechanism (Colman-Lerner, 2001; Weiss, 2002). The Fry homolog Tao3p (also named Pag1p) is required for both Cbk1p functions (Du, 2002; Nelson, 2003). Mutation of a fungal homolog of trc, cot1, causes a drastic increase of the hyphal branching in Neurospora (Yarden, 1992). Interestingly, mutations of the trc homolog sax-1 in C. elegans cause sensory neurons to have ectopic neurites (Zallen, 2000). In Drosophila, mutations of either trc or fry result in branched bristles and multiple wing hair phenotypes (Geng, 2000; Cong, 2001). However, the roles of Trc and Fry in neurons are obscure (Emoto, 2004 and references therein).

Trc and Fry are shown to function cell autonomously in Drosophila da neurons to regulate dendritic tiling and branching. Trc-kinase activity is required for the dendritic branching and tiling in vivo and is positively regulated by Fry. The control of dendritic branching but not tiling involves negative regulation of the RacGTPase signaling pathway by Trc. These findings suggest that Trc/Fry utilizes two distinct signaling pathways to shape the dendritic fields: one pathway to limit dendritic branching and a separate pathway to promote like-repels-like response of dendritic processes (Emoto, 2004).

Thus, Trc/Fry signaling pathway plays an essential role in regulation of dendritic branching. trc and fry mutants typically displayed excessive terminal branches, whereas the major dendrite architecture appeared normal in all four classes of da neurons. For example, the primary, secondary, and tertiary branches of class IV neurons did not show overbranching in trc and fry mutants whereas the terminal branches were increased by a factor of two. It is thus likely that Trc and Fry specifically regulate the branching of fine structures, but not overall architecture, of dendrites. Consistent with this idea, no obvious defect was observed in neurons with simple dendrites, such as bipolar neurons, external sensory neurons, and chordotonal neurons (Emoto, 2004).

Trc and Fry homologs have been identified in a variety of species including S. cerevisiae (Colman-Lerner, 2001; Weiss, 2002; Du, 2002), S. pombe (Verde, 1998; Hirata, 2002), N. Crassa (Yarden, 1992), C. elegans (Zallen, 2000 ; Gallegos, 2004), Drosophila (Geng, 2000 and Cong, 2001), and mammals (Tamaskovic, 2003b). Moreover, Fry homologs show a strong genetic interaction with Trc homologs. In S. cerevisiae and S. pombe, mutations in either of these genes cause severe cell growth defect as well as cell morphology defect. In Drosophila, however, other than the dendritic branching phenotype in neurons, no obvious cell growth defects were found in any tissues in the trc and fry mutant third instar larvae. Zallen reported that mutations in the trc homolog sax-1 gene cause extra neurite formation in sensory neurons of C. elegans while the overall structure of the neuron appears normal. Thus, the Trc/Fry signaling appears to mediate similar functions in sensory neurons of flies and worms. In addition, mutations of trc or fry result in branched bristles and multiple wing hair phenotypes in Drosophila (Geng, 2000; Cong, 2001), and mutation of a fungal homolog of trc, cot1, causes a drastic increase of the hyphal branching in Neurospora. Thus, the Trc/Fry signaling is likely to be a general mechanism to regulate branching of cellular processes (Emoto, 2004).

Dendritic tiling has been proposed to play a key role in patterning the dendritic fields of particular neurons, including Drosophila class IV sensory neurons (Jan, 2003; Grueber, 2004). Previous studies suggest that tiling in class IV neurons arises from repulsion between homologous dendrites (Grueber, 2003a; Sugimura, 2003). Consistent with this idea, time-lapse study reveals no overlap of the dendrites in late embryonic, first, second, or third instar larval stages, suggesting that tiling in class IV neurons is established when dendrites first meet and is maintained throughout the larval stage (Emoto, 2004).

Genetic and molecular evidence has been provided that Trc and Fry play a crucial role in establishing and maintaining the dendritic tiling of class IV neurons. In trc or fry mutants, terminal branches of class IV dendrites fail to avoid crossing each other, not only within the same neuron, but also between different neurons, leading to a significant overlap of dendritic fields. Since trc and fry null mutants display both overbranching and tiling phenotypes, one obvious possibility is that the tiling phenotype simply results from the overbranching phenotype. The following lines of evidence support the notion that the dendritic tiling defect in trc and fry mutants is not secondary to the overbranching phenotype. (1) When normalized by the total branch number or the total branch length, the number of dendritic crossings in trc and fry null mutants remains significantly greater than that in wild-type controls. (2) In the absence of dendritic overbranching, a robust tiling phenotype is still observed in the fry⁶ hypomorphic allele, as well as larvae transheterozygous for fry¹and fry⁶ or for fry¹ and trc¹. (3) trc and fry mutants display not only iso-neuronal tiling defects but also hetero-neuronal tiling defects. Such a significant overlap of the dendritic fields between neighboring class IV neurons is unlikely to result from a simple increase of terminal dendritic branches if each dendritic branch retains a like-repels-like activity. Indeed, a series of mutants, such as flamingo and sequoia, with dendritic overgrowth and/or overbranching have been isolated in previous studies; however, they appear not to have tiling defects in class IV dendrites (Grueber, 2002). The independence of tiling and branching phenotypes is further supported by the observation that the dominant-negative RacN17 suppresses the dendritic branching but not the tiling phenotype due to expression of the dominant-negative TrcK112A mutant (Emoto, 2004).

Time-lapse observations show that in wild-type larvae, ~70% of terminal branches appear to make a dramatic turn before they cross nearby branches, again supporting the idea that the like-repels-like mechanism plays a central role in class IV dendritic tiling. Compared to wild-type, fry mutant dendrites fail to turn away from nearby branches, but they showed normal net growth and retraction. Taken together, these data strongly suggest a role for Trc and Fry in the like-repels-like behavior of the class IV dendrites. It remains possible, however, that other mechanisms function in parallel with the Trc/Fry signaling pathway to establish tiling, since some dendritic branches of class IV neurons still appear to tile in trc and fry mutants (Emoto, 2004).

Sax-1 and Sax-2 (the worm homologs of Trc and Fry) also have an essential role in mechanosensory neurite tiling (Gallegos, 2004). This finding, together with the current study, strongly suggest an evolutionarily conserved role for the Trc/Fry signaling pathway in dendritic tiling. Indeed, of the two mammalian trc homologs ndr1 and ndr2, ndr2 is highly expressed in various tissues including the brain, while ndr1 displays a relatively specific localization in muscles (Devroe, 2004). Although it has not been established whether a tiling mechanism contributes to dendritic field specification in the central nervous system outside of retina, cerebral cortical neurons are known to exhibit contact-mediated growth inhibition of neurites (Sestan, 1999). It will be of interest to examine potential roles of Trc and Fry homologs in regulation of dendritic tiling as well as branching in vertebrate nervous systems. Additionally, considering a close correlation between abnormal dendrite patterning and mental retardation, it might be intriguing to examine the relationship between ndr1/2 genes and mental retardation diseases (Emoto, 2004).

The MARCM analyses and the rescue studies, together with the expression of Trc and Fry in da neurons, indicate that Trc and Fry function cell autonomously in neurons. In addition, Trc kinase activity is indispensable for the control of dendritic branching and tiling in vivo, and Fry is required for Trc kinase activity, indicative of an important role of intracellular kinase signaling. The kinase domain of Trc is closely related to Rho-kinase (Rok), with 45% amino acid identity and 71% similarity to Drosophila Rok (Drok), but Trc lacks a Rho binding domain and other regulatory domains (Tamaskovic, 2003b). Indeed, C. elegans Sax-1 has a partial genetic interaction with Rho in neuronal cell shape regulation (Zallen, 2000). Whereas no obvious interaction between Trc/Fry and Rho/Drok was observed in Drosophila da neurons, Rac signaling likely plays an important role in Trc/Fry regulation of dendritic branching of class IV neurons. Trc partially suppresses the overbranching phenotype induced by RacWT. Moreover, the dominant-negative RacN17 suppresses the overbranching but not tiling phenotypes in neurons expressing dominant-negative Trc mutant. One possible scenario is that Trc/Fry negatively regulates Rac to control dendritic branching. This notion is further supported by analyses of the active RacGTP protein level in cells expressing wild-type or mutant Trc (Emoto, 2004).

In contrast to the involvement of Rac in dendritic branching, there is no indication that tiling depends on Rac signaling. Rather, two distinct pathways seem to be employed by Trc/Fry to control dendritic branching and tiling. This idea is consistent with genetic data indicating that tiling defects can be separated from branching defects in trc and fry mutants. It is of interest to note that in budding yeast, Cbk1p (Trc homolog) and Tao3p (Fry homolog) play multiple roles involving different downstream pathways (Colman-Lerner, 2001; Weiss, 2002; Nelson, 2003; Emoto, 2004 and references therein).

The repulsion between dendrites likely involves either contact-mediated dendritic interactions or signaling via a short-range diffusible substance -- a signal that is likely to have class-specific components (Lohmann, 2001; Jan, 2003). It is unlikely that Trc and Fry determine the class specificity since Trc and Fry are expressed in all da neurons. Conceivably, Trc may transmit subtype-specific repulsion signals, generated by class-specific factors to downstream components including the cytoskeleton, and induce a like-repels-like behavior of dendrites. The finding that Fry and Trc are involved in the like-repels-like response in class IV neurons has provided an entry point for studying the molecular mechanisms that control dendritic tiling. As shown in this study, phosphorylation of the conserved Ser/Thr of Trc appears critical for the Trc/Fry signaling in vivo, yet little is known about the upstream kinase(s) and the downstream substrate(s) of Trc in any species. A molecular dissection of the Trc kinase signaling pathway in neurons will help elucidate how the Trc/Fry signaling pathway governs dendritic branching and tiling (Emoto, 2004).

GENE STRUCTURE

cDNA clone length - 2126 bp

Bases in 5' UTR - 125

Exons - 5

Bases in 3' UTR - 621

PROTEIN STRUCTURE

Amino Acids - 459 (trc-PA: long form)

Structural Domains

Plasmid rescue of the trc⁷ allele was used to obtain a trc genomic clone. The sequencing of the rescued plasmid shows that it contains part of the Drosophila NDR kinase gene. This DNA fragment was next to the deaf-1 gene, and by sequencing both genomic (accession no. AF247814) and cDNA clones [isolated by the BDGP (clone LD15101, accession no. AF238490)] it was found that the putative trc gene contained four small introns. The P element responsible for the trc^P mutation was inserted into the middle of the first intron. To confirm that the trc gene was identified genomic DNA was sequenced from several trc point mutants and mutations were identified in five EMS-induced trc alleles. Three were missense mutations, one was a nonsense mutation that would produce a truncated protein, and the fifth was a splice junction mutation that should prevent the normal processing of the trc mRNA. The missense mutations argue strongly that the kinase activity of trc is essential for the gene's activity. The conserved regions of kinase superfamily members are typically broken up into 10 subdomains. One missense mutation resulted in the change of gly¹⁰⁰ to glu. This gly is part of the very highly conserved ATP-binding site in kinase subdomain I. A second resulted in the change of gly²³⁶ to glu. This gly is part of the very highly conserved DFG triplet found in kinase subdomain VII. These two sites are among the 12 sites generally recognized as being invariant (or almost invariant) in the kinase superfamily. Thus, these sites are strongly implicated as playing important roles in kinase function and it is very likely that these trc mutations eliminate or substantially reduce kinase activity. This argues that the kinase activity is essential for the function of trc. The third missense mutation resulted in the change of arg³⁹⁵ to pro. This arg is not conserved in all kinases, but it is a site that is conserved in all the NDR proteins. This site is in kinase subdomain X (Geng, 2000).

A comparison of the cDNA sequence, genomic sequence, and the published NDR sequence (Millward, 1995) revealed a couple of changes that were suspected to be due to either sequencing mistakes or naturally occurring polymorphisms. One change that would result in a four-amino-acid difference did not seem likely to have such an explanation. This change was located at an intron/exon border and suggested the possibility of the use of an alternative splice site. To test this, sequencing in this region was carried out in four additional cDNA clones (GH16329, GH24041, LD37189, and LP06419) and examples were found of both splicing patterns. The short form is referred to as trc-S (accession no. AF239171 ) and the long form as trc-L (accession no. AF238490). Three of the cDNA clones were the long form (LD15101, GH16329, and GH24041) and two were the short form (LD37189 and LP06419). Due to the small difference in size the relative abundance of the two forms has not been determined. This four-amino-acid difference is located provocatively in between subdomains VII and VIII, just five amino acids from the sequence identified as a nuclear localization sequence in the human NDR (Millward, 1995). The region between kinase subdomains VII and VIII (and the C- and N-terminal ends of these regions) has been found to play a major role in the recognition of peptide substrates in other kinases and this region is also a site of an activating phosphorylation. Based on the location of this four-amino-acid difference it is suggested that there may be functional differences between the two forms of Trc (Geng, 2000).

The NDR kinases are members of the AGC kinase group (they are closest to the AGC-VII subfamily). Among the well-studied members of the AGC group are the cyclic-AMP-dependent protein kinase (PKA) subfamily and the Ca²⁺-dependent protein kinase C (PKC) subfamily. The NDR kinase was originally identified in the Caenorhabditis elegans genome by sequence analysis and fly and human genes were then cloned on the basis of homology to the worm gene (Millward, 1995). Further progress in genome project sequencing has resulted in the identification of a second human NDR family member. These proteins are quite similar and likely to be orthologs. The 463-amino-acid Trc-L protein is 68% identical and 81% similar to the human NDR protein over a 444-amino-acid stretch. The Trc-L protein is almost as closely related to the C. elgans NDR. Blast E values of e^-150 or less are obtained in comparisons of the individual NDR kinases. Blast searches with Trc find the next most related group of proteins are fungal proteins such as the ORB6 protein of Schizosacchromyces (E values of ~e^-119) and a number of plant kinases identified in sequencing projects. The Trc-L and Orb6 proteins are 49% identical and 66% similar over 451 amino acids. Somewhat more distant members of this kinase subfamily are the COT-1 protein of Neurosopora and the Drosophila WARTS/LATS protein. More distantly related, but still closer to Trc than most kinases are proteins such as the mammalian myotonic dystrophy kinase, the Genghis Khan kinase of flies, the ROCK (Rho-activated kinase of mammals), and the DBF2 protein of yeast. Mutations in the genes that encode many of these kinases result in phenotypes that cause alterations in cell morphology or polarity. For example mutations in orb6 of Schizosacchromyces pombe cause a loss of polarized growth and delayed entry into mitosis. Mutations in cot1 of Neurosopora result in branched hyphae and mutations in warts/lats from flies results in tumors and altered cell morphology. Given the common links to cellular morphology it seems reasonable to suggest that these kinases might have similar targets (e.g., actin cytoskeleton). Hence, it is somewhat surprising that some of these proteins are nuclear [NDR (Millward, 1995), DBF2 (Komarnitsky, 1998)] and others cytoplasmic. Perhaps this group of kinases has both nuclear and cytoplasmic functions and depending on context they are found primarily in one of these two cellular compartments. A sequence in the human NDR was identified by Millward (1995) that acts as a nuclear localization signal and this sequence is conserved in the fly and worm NDRs. Interestingly this sequence is also conserved in Orb6, which was found to be cytoplasmic (Verde, 1998). Trc appears to be primarily nuclear in cultured mammalian cells, although the situation appears more complicated in the fly. This raises the possibility that Trc functions in cellular morphogenesis by regulating the expression of target genes that encode cytoskeleton-interacting proteins and not via the direct modification of the cytoskeleton (Geng, 2000).

date revised: 9 March 2005

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.