coracle

Heterotrimeric G proteins regulate a noncanonical function of septate junction proteins to maintain cardiac integrity in Drosophila

The gene networks regulating heart morphology and cardiac integrity are largely unknown. The heterotrimeric G protein gamma subunit 1 (Ggamma1) has been shown to mediate cardial-pericardial cell adhesion in Drosophila. This study shows that G-oalpha47A and Gβ13F cooperate with Ggamma1 to maintain cardiac integrity. Cardial-pericardial cell (CC-PC) adhesion also relies on the septate junction (SJ) proteins Neurexin-IV (Nrx-IV), Sinuous, Coracle, and Nervana 2, which together function in a common pathway with Ggamma1. Furthermore, Ggamma1 signaling is required for proper SJ protein localization, and loss of at least one SJ protein, Nrx-IV, induces cardiac lumen collapse. These results are surprising because the embryonic heart lacks SJs and suggest that SJ proteins perform noncanonical functions to maintain cardiac integrity in Drosophila. These findings unveil the components of a previously unrecognized network of genes that couple G protein signaling with structural constituents of the heart (Yi, 2008).

The results of this study show that the heterotrimeric G proteins G-oα47A, Gβ13F, and Gγ1 function together to maintain CC-PC adhesion during the late stage of heart formation in Drosophila. By mapping a new broken hearted (bro) mutant (Nrx-IV) and characterizing additional candidate genes, a noncanonical role was discovered for SJ proteins in mediating CC-PC and CC-CC adhesion outside SJs. Four SJ proteins, Nrx-IV, Sinu, Cora, and Nrv2, operate in a common pathway with Gγ1 to maintain cardiac integrity; these proteins require Gγ1 for proper subcellular localization in the heart. Mechanistically, the presence of SJ proteins in both CCs and PCs suggests that these proteins act in trans to maintain cell-cell adhesion in the dorsal vessel. A model is favored in which the extracellular domain of Nrx-IV engages in heterophilic interactions with SJ-proteins such as Neuroglian or Contactin, and that these interactions would be stabilized by ECM proteins such as Pericardin (Prc). Alternatively, the SJ proteins may directly interact with ECM proteins to provide a structural basis for cardiac integrity (Yi, 2008).

Heterotrimeric G proteins G-oα47A/G-iα65A, Gβ13F, and Gγ1 function with the GPCR moody and the RGS protein loco to regulate SJ formation in the Drosophila brain-blood barrier (Schwabe, 2005). Although loco mutant embryos show the bro heart phenotype, moody mutations do not induce a heart phenotype. A search of the Drosophila protein interaction map reveals that the GPCR CG32447 interacts with both the SJ protein Sinu and the RGS Kermit. Kermit also interacts with Loco, suggesting that the CG32447 GPCR participates in the control of cardiac integrity. However, a deficiency uncovering CG32447 does not induce the bro phenotype. Since the screen for bro mutants, visualized as a perturbation in the ordered expression pattern of Hand-GFP in cardial and pericardial cells, did not identify a GPCR that maintains cardiac integrity, it is concluded that the GPCR regulating cardiac integrity is either pleiotropic, with an early embryonic function that precludes its identification as a regulator of cardiac integrity, or is redundant to a second GPCR in the dorsal vessel (Yi, 2008).

Alternatively, cardiac integrity may be regulated by a GPCR-independent mechanism. In neuroblasts, G-iα65A, Gβ13F, Gγ1, and loco regulate mitotic spindle orientation, protein localization, and ultimately asymmetric cell division via a GPCR-independent signaling pathway. During neuroblast cell division, heterotrimeric G proteins are activated by the GTPase exchange factor (GEF) Ric-8, but not by GPCRs (see David, 2005). However, the lethal mutation ric-8^G0397 does not induce the bro phenotype (Yi, 2008).

During blood-brain barrier formation, sequestering Gβγ or hyperactivating G-oα47A signaling in glial cells leads to SJ defects, whereas hyperactivating G-iα65A signaling does not affect SJ function. A similar relationship exists among heterotrimeric G proteins during asymmetric cell division in neuroblasts. In contrast, sequestering Gβγ in the dorsal vessel has no effect on cardiac integrity, while hyperactivating G-oα47A in the embryonic heart induces the bro phenotype. It is concluded that the bro phenotype in Gβ13F or Gγ1 mutants is caused by misregulation of G-oα47A signaling. This is in sharp contrast to the G proteins regulating blood-brain barrier formation and asymmetric cell division where Gβγ dimers activate a set of downstream effectors distinct from that of G-oα47A signals (Yi, 2008).

G protein signaling regulates SJ formation in Drosophila and tight junction formation in mammalian cells. Even though SJs are analogous to vertebrate tight junctions, it is striking that G protein signaling components colocalize with both SJ and tight junction proteins. In addition, Gα_s interacts with the tight junction protein ZO-1 throughout junction formation, suggesting that Gα subunits physically regulate tight junction assembly. Thus, septate/tight junction proteins appear to be direct targets of G proteins in both flies and vertebrates (Yi, 2008 and references therein).

Although the embryonic heart lacks SJs, the current results are consistent with the idea that SJ proteins are direct targets of G proteins in the dorsal vessel. G protein mutants phenocopy SJ-protein mutants and G proteins operate in a common pathway with SJ proteins to maintain cardiac integrity. In addition, proper localization of SJ proteins in the embryonic heart requires G protein signaling, and G proteins regulate at least one SJ protein at the posttranscriptional level. Finally, loss of G-oα47A signaling (G-oα47A mutants) and hyperactivation of G-oα47A signaling (overexpressing G-oα47A) both result in the bro phenotype; thus Gα signaling is localized to specific foci in cells of the dorsal vessel. It is proposed that an appropriate level of Gα signaling mediates SJ-protein localization, whereas loss or hyperactivation of Gα signaling mislocalizes SJ proteins leading to a loss in cardiac integrity (Yi, 2008).

Cell-cell adhesion plays an essential role during organ morphogenesis. In the Drosophila heart, cell-cell adhesion along three distinct CC membrane domains is required to maintain cardiac integrity. Medioni (2008) provide a detailed description of two CC domains participating in cell-cell adhesion: the adherent domain, positioned immediately dorsal and ventral to the cardiac lumen, promotes cell-cell adhesion between CCs on opposing sides of the heart, and the basal-lateral adherent domain, positioned along the lateral CC membrane, promotes cell-cell adhesion between neighboring CCs on one side of the heart. These studies suggest that a third CC membrane domain, referred to as the pericardial adherent domain, is positioned opposite to the luminal domain and promotes PC-CC adhesion. The loss of cell-cell adhesion along each of the three CC domains gives rise to a unique phenotype: luminal collapse (referred to hereafter as type-1), breaks between neighboring cardial cells (type-2), and loss of PC-CC adhesion (type-3), respectively. The unique nature of these three phenotypes can provide insight into the molecular pathways regulating cardiac integrity (Yi, 2008).

Loss of heterotrimeric G proteins or SJ proteins induces the type-3 (bro) phenotype, and mutations in at least one SJ-protein gene, Nrx-IV, leads to the type-1 phenotype. In addition, Sinu, Cora, and Nrv2 localize to the luminal and perhaps the adherent domains, suggesting that loss of these proteins will also cause the type 1 phenotype. The type 2 phenotype is observed in a subset of Gγ1 embryos, but not in any other heterotrimeric G protein or SJ-protein mutants. Thus, the pathways regulating cell-cell adhesion along the CC basal-lateral membrane may be distinct from those identified in this study (Yi, 2008).

The guidance ligand Slit has been shown to regulate multiple aspects of cardiogenesis in Drosophila, and mutations in slit induce luminal collapse (referred to hereafter as type-1), breaks between neighboring cardial cells (type-2), and likely loss of PC-CC adhesion (type-3) phenotypes. In addition, slit mutant embryos show mesoderm migration and CC polarity defects, however these defects are genetically separable from cardiac integrity defects. Slit signals through the Robo receptors and mutations in genes encoding downstream components of the Robo signaling pathway do not dominantly enhance slit mutations. In contrast, mutations in genes encoding integrins or integrin ligands, such as scab, mys, and Lan-A, dominantly enhance slit mutations and transheterozygous embryos show the type-2 phenotype. This study suggests that Slit activates two pathways during cardiogenesis: one pathway utilizes typical Robo signaling to regulate mesoderm migration and CC polarity while a second pathway uses atypical, or Robo-independent, signaling to regulate cell adhesion between neighboring CCs and likely between opposing CCs to promote lumen formation. Although the role of Slit in regulating PC-CC adhesion has not been studied in detail, one possibility is that Slit signals through G-oα47A/Gβ13F/Gγ1 to regulate CC-CC and even PC-CC adhesion (Yi, 2008).

SJ proteins are functionally interdependent and localization of Sinu to SJs requires Nrx-IV, Cora, and Nrv2 (Wu, 2004), while Nrx-IV, Cora, Cont, and Nrg are equally interdependent for localization to SJs. In addition, both Nrv2 and Nrx-IV are transmembrane proteins, and the extracellular domain of Nrv2 at least is required for SJ function. Since every SJ-protein mutant examined showed PC-CC adhesion defects, SJ proteins likely form interdependent complexes in PCs and CCs. The extracellular domains of SJ proteins may act in trans, either through direct interactions with SJ proteins along opposing membranes or through indirect interactions with ECM proteins such as Pericardin, to maintain cardiac integrity. A search of the Drosophila protein interaction map reveals an interaction between Pericardin and Sinu, supporting the latter possibility. Alternatively, SJ proteins could be required for the formation or function of adherens junctions in the dorsal vessel (Yi, 2008).

All of the bro genes have close vertebrate orthologs. Since the function of mevalonate pathway genes in heart development is conserved from Drosophila to vertebrates (D'Amico, 2007; Edison, 2005; Yi, 2006), it is speculated that G protein-mediated regulation of SJ proteins is also evolutionarily conserved. To date, the role of heterotrimeric G proteins in regulating vertebrate heart development has not been identified, but heterotrimeric G proteins do play a role in heart disease. In contrast, Sinu is a member of the Claudin protein family and even though this protein family is rather divergent (Wu, 2004), vertebrate Claudin-1 is required for normal heart looping in the chick. In addition, Claudin-5 localizes to the lateral membrane of cardiomyocytes and is associated with human cardiomyopathy. Lastly, mutations in the prc ortholog, collagen alpha-1(IV), cause vascular defects in mice and humans. Taken together, these studies raise the possibility that heterotrimeric G proteins and tight junction proteins ensure proper vertebrate cardiovascular morphogenesis (Yi, 2008).

Drosophila muscleblind targets coracle

Muscleblind-like proteins, muscleblind (Mbl) in Drosophila and MBNL1-3 in vertebrates, are regulators of alternative splicing. Human MBNL1 is a key factor in the etiology of myotonic dystrophy (DM), a muscle wasting disease caused by the occurrence of toxic RNA molecules containing CUG/CCUG repeats. MBNL1 binds to these RNAs and is sequestered in nuclear foci preventing it from exerting its normal function, which ultimately leads to mis-spliced mRNAs, a major cause of the disease. Muscleblind-proteins bind to RNAs via N-terminal zinc fingers of the Cys(3)-His type. These zinc fingers are arranged in one (invertebrates) or two (vertebrates) tandem zinc finger (TZF) motifs with both fingers targeting GC steps in the RNA molecule. This study shows that mbl genes in Drosophila and in other insects also encode proteins with two TZF motifs, highly similar to vertebrate MBNL proteins. In Drosophila the different protein isoforms have overlapping but possibly divergent functions in vivo, evident by their unequal capacities to rescue the splicing defects observed in mbl mutant embryos. In addition, using whole transcriptome analysis, several new splicing targets were identified for Mbl in Drosophila embryos. Two of these novel targets, kkv (krotzkopf-verkehrt, coding for Chitin Synthase 1) and coracle (coding for the Drosophila homolog of Protein 4.1), are not muscle-specific but expressed mainly in epidermal cells, indicating a function for mbl not only in muscles and the nervous system (Irion, 2012). Drosophila mbl is known to code for several proteins with one N-terminal tandem zinc finger motif; this study found that the genes in Drosophila and in other insects (honey bee, wasp, mosquito) also encode protein isoforms with two TZF motifs, highly similar to the Mbl orthologs in vertebrates. The zinc fingers show a very high degree of conservation between insect Mbl and vertebrate MBNL proteins with almost 80% amino acid similarity. The only significant difference is the spacing between two Cys residues in the second zinc finger of the first TZF motif, with two additional amino acids in insect proteins (Irion, 2012).

Also, the genomic organization of the muscleblind genes is very similar, not only between insects, but also between insects and vertebrates. The intron positions are highly conserved and so is the large size of intron 2. This intron, which splits the coding sequence for the first TZF motif, spans 75 kb in Drosophila and, according to the latest genome assembly, more than 700 kb in honeybees. Nevertheless the spliced product could be easily detected in RNA from pupae (Irion, 2012).

In Drosophila transcripts for a large number of mbl isoforms coding for proteins with one or two TZF motifs are present during all stages of the life cycle. It proved to be difficult to estimate the relative abundance of the different types of transcripts. PCR primers designed to detect both types lead almost exclusively to the amplification of transcripts coding for only one TZF motif. This could be due to much higher abundance of these transcripts, however, it might also reflect a bias in the PCR amplification. Especially because primers designed to detect isoforms with two TZF motifs work very efficiently (Irion, 2012).

There is no indication that vertebrates express MBNL proteins with only one TZF motif. However, for human MBNL1 and MBNL3 it has been shown that truncated versions of the proteins, lacking either one of the two TZF motifs, are still able to bind to RNA and to regulate splicing in a cell culture assay (Grammatikakis, 2011; Irion, 2012 and references therein).

To find additional targets for Mbl in Drosophila the complete transcriptome of stage 16-17 embryos hemizygous for a null-allele was analyzed by Illumina sequencing. Given the fact that a lower accuracy in splice site selection for the α-actinin and ZASP52 transcripts was detected in mutant embryos compared to wild type, the analysis of the RNAseq data focused on the number of total and of unique intron-exon reads as a measure of splicing precision. The known Mbl targets in Drosophila are amongst the highest-ranking transcripts with more than 85 unique intron-exon reads. In the complete list of 81 candidates there is a clear enrichment for the GO annotation terms 'actin binding', ‘cytoskeletal protein binding', 'protein binding' and ‘tropomyosin binding', indicating the prominent function of mbl in muscle development and differentiation. However, many genes with a large number of intron-exon reads represent loci with multiple, often interspersed transcripts, which makes analysis very difficult, e.g. Pde1c, Ect4 or l(3)82Fd. Some genes with many intron-exon reads also harbor transposons, which does account for some of the reads, e.g. Cda5 (Irion, 2012).

Taking these limitations into consideration, the analysis revealed several new potential candidates as targets for Mbl in Drosophila. Ten of these candidates were tested by RT-PCR and splicing defects were found in five of them. The observed defects are either a shift in the ratio of different isoforms (CG33205, cora), the incorrect selection of mutually exclusive exons (wupA, kkv) or general defects in splicing accuracy (Mf) (Irion, 2012).

An important question is whether the different proteins, with one or two TZF motifs and with different C-termini, have different functions in vivo. One possibility could be that all the different Mbl isoforms in Drosophila, which are generated from one gene by extensive alternative splicing, have similar and redundant functions regulating the alternative splicing of the same pre-mRNAs. Another possibility could be that the different proteins have different pre-mRNA targets. As a third alternative some isoforms could be involved in functions altogether different from the regulation of splicing, such as RNA localization. In vertebrates, MBNL proteins are encoded by three separate genes, and for MBNL2 a function in localizing the integrin α₃ transcript has been demonstrated (Adereth 2005). Whether any other Mbl protein besides MBNL2 is involved in RNA localization remains to be seen (Irion, 2012).

To address this question the ability was tested of Drosophila Mbl proteins with one or two TZF motifs to rescue the splicing defects occurring in homozygous mutant embryos. It has previously published that Drosophila Mbl regulates alternative splicing of two transcripts in embryos, α-actinin and ZASP52 (MachucaTzili, 2006). In homozygous mutant embryos the tissue specific and developmental timing dependent splicing of α-actinin is mis-regulated, e.g., with the premature occurrence of adult isoforms already during embryogenesis. Also in mutant embryos aberrant ZASP52 transcripts can be detected, where a cryptic splice site in exon 15 is used. In both cases there was a clear rescue of the splicing defects by transgene 1, which encodes a protein with only one TZF motif. Transgene 2, which codes for a protein with both TZF motifs, only rescues the α-actinin splicing but not the splicing of ZASP52 (Irion, 2012).

A third published target for Mbl in Drosophila is troponinT (VicenteCrespo, 2008). It has been shown that in mutant pupae alternative splicing of transcripts from this gene is mis-regulated. Also in embryos there is a shift in the ratio of two different isoforms of troponinT transcripts in mbl mutants. This defect is not rescued by either of the two transgenes tested (Irion, 2012).

Having identified additional transcripts mis-spliced in mbl mutant embryos, the two different transgenes were tested for their ability to rescue these defects. There is generally good rescue with transgene 1 whereas transgene 2 only rescues the splicing of cora and kkv to some extent (Irion, 2012).

The observed differences between the two transgenes indicate that there might be functional differences between Mbl proteins with one and two TZF motifs. Both are clearly regulators of alternative splicing, as they efficiently rescue the splicing of α-Actinin, which also demonstrates that expression occurs from both transgenes and that the proteins function in the nucleus. The different rescue capacities for the other mbl targets could indicate different specificities of the protein isoforms, where the protein with two TZF motifs might have other, not yet identified, targets. An alternative explanation, which cannot be ruled out at this stage, is that the proteins are present in the cells in very different amounts or that they localize mostly to different sub-cellular compartments. Because both transgenes are inserted at the same genomic location on the X-chromosome and the same GAL4 driver-line was used it seems unlikely that significantly different RNA levels should be transcribed. Therefore one would have to assume that the translation efficiencies of the RNAs are different, or that the extra 121 amino acid residues affect the stability or localization of the protein encoded by transgene 2. This hypothesis will only be testable after generation of an antibody, which allows the detection of Mbl proteins at endogenous levels on western blots (Irion, 2012).

Apart from CUG repeats, the five nucleotide sequence, 5'-AGUCU-3', has been identified as a consensus binding motif for Drosophila Mbl by in vitro-studies using a protein with the first TZF motif (Goers, 2008). This sequence motif occurs, however, only rarely in intronic sequences flanking those exons that are mis-spliced in mbl mutants. It has also been suggested that Mbl recognizes complex RNA secondary structures, which would not be easy to predict (Irion, 2012).

In contrast, it was found that expression of the human gene, MBNL1, can rescue embryonic lethality in Drosophila, suggesting that both proteins recognize the same target RNAs. In vitro selection has lead to the definition of a binding motif for the human protein, which is a GC motif embedded in pyrimidines, 5'-YGCY-3', with a preference for at least one pyrimidine-base (Y) being U . In efficient high-affinity binding sites this motif often occurs in several copies in the RNA. In a simple model for the regulation of alternative splicing, exon exclusion is promoted when MBNL binds upstream of the exon, whereas binding downstream of the exon enhances its inclusion (Irion, 2012).

Given the high degree of similarity between the zinc finger motifs in Drosophila Mbl and human MBNL1 it seems likely that the Drosophila protein recognizes the same binding sites in pre-mRNAs. Analysis of the complete Drosophila genome showed that the binding motif occurs on average 1.72 times in intronic sequences close to exons (<200 bp) (Vipin T. Sreedharan and Gunnar Rätsch, personal communication to Irion, 2012). In the intronic sequences close to the exons 5 and 6 in Actn there are indeed clusters of YGCY motifs. In this case there is also good correlation with the model that Mbl binding downstream enhances inclusion (exon 5a) whereas binding upstream of the exon predominantly leads to exclusion (exon6). For the other exons where Mbl regulates splicing, mostly identified in this study, the correlation is less clear. For example, in the case of wupA there is no clustering of YGCY motifs detectable, although the presence of Mbl protein clearly leads to the inclusion of exon 6a and the exclusion of exons 6b-6d. In several other cases it is not simple the inclusion or exclusion of an exon that is regulated by Mbl, but the selection between alternative splice sites for one exon (Irion, 2012).

Drosophila is an important model organism to study the function of Mbl proteins in vivo during development and their role in the pathogenesis of myotonic dystrophy. This study has shown that the fly gene codes for proteins that are more similar to the human proteins than previously recognized. The different protein isoforms seem to have distinct functions in the regulation of alternative splicing. Further analysis will be needed to understand the in vivo contributions of the different protein isoforms towards the regulation of alternative splicing and possibly in other processes (Irion, 2012).

Protein Interactions

Neurexin is localized apicolaterally, adjacent to Crumbs, which delimits the zonula adherens. These two proteins are not coexpressed, placing NRX apicolaterally. Both Fasciclin3 and NRX colocalize at salivary gland synaptic junctions. NRX precisely colocalized with Coracle, the Drosophila homolog of mammalian protein 4.1, except in the PNS and CNS where Coracle is only expressed in a few cells. No defects in the localization of Discs large protein is detected in Nrx mutants. However, Coracle is not restricted to septate junctions in Nrx mutants. These results suggest that the short cytoplasmic portion of NRX that shows homology to glycophorin C is required to localize Coracle to septate junctions, creating a parallel with red blood cell cytoskeletal anchoring proteins (Baumgartner, 1996).

The protein 4.1 superfamily comprises a diverse group of cytoplasmic proteins, many of which have been shown to associate with the plasma membrane via binding to specific transmembrane proteins. Coracle, a Drosophila protein 4.1 homolog, is required during embryogenesis and is localized to the cytoplasmic face of the septate junction in epithelial cells. Using in vitro mutagenesis, it has been demonstrated that the amino-terminal 383 amino acids of Coracle define a functional domain that is both necessary and sufficient for proper septate junction localization in transgenic embryos. Genetic mutations within this domain disrupt the subcellular localization of Coracle and severely affect its genetic function, indicating that correct subcellular localization is essential for Coracle function. The localization of both Coracle and the transmembrane protein Neurexin to the septate junction displays an interdependent relationship, suggesting that Coracle and Neurexin interact with one another at the cytoplasmic face of the septate junction. Consistent with this notion, immunoprecipitation and in vitro binding studies demonstrate that the amino-terminal 383 amino acids of Coracle and the cytoplasmic domain of Neurexin interact directly. Together these results indicate that Coracle provides essential membrane-organizing functions at the septate junction, and that these functions are carried out by an amino-terminal domain that is conserved in all protein 4.1 superfamily members (Ward, 1998).

The interdependence between Coracle and NRX for proper localization suggests that at least one other protein in the presumptive septate junction serves as the initial target for both proteins to be properly localized. Based on the protein 4.1 paradigm of a ternary complex consisting of protein 4.1, glycophorin C, and p55 , it is predicted that a PDZ repeat-containing protein is a part of the complex containing Coracle and NRX. The most likely candidate for this additional protein is DLG, based on its extensive sequence similarity to p55. DLG is expressed maternally, and initially is uniformly distributed along the lateral membrane (and to a lesser extent throughout the cytoplasm). Coincident with the expression of Coracle and NRX, this subcellular localization is refined to the presumptive septate junction. This expression pattern might be expected of a protein that serves to "prepattern" the septate junction. However, attempts to detect any interaction between Coracle and DLG by immunoprecipitation have failed; no genetic interaction between coracle and dlg mutant alleles have been detected. The embryonic defects associated with dlg mutants are different from those of coracle and Nrx. These results suggest that DLG is not involved in a ternary complex together with Coracle and NRX, despite its structural similarity with p55. The question of whether there is another PDZ repeat-containing protein that functions to stabilize Coracle-NRX interactions remains to be answered, although the structural similarities between the respective Drosophila and human proteins strongly suggest that one exists. The recent identification of EBP50 as a PDZ repeat-containing protein that associates with ERM proteins suggests that an interaction with a PDZ repeat-containing protein may be a ubiquitous feature of protein 4.1 members. Regardless, the results described here strongly suggest that at least one other component is involved in Coracle/NRX localization and function (Ward. 1998).

Drosophila Discs large 1 interacts with protein 4.1 homologs. All members of the protein 4.1 superfamily share a highly conserved N-terminal 30-kDa domain whose biological function is poorly understood. It is believed that the attachment of the cytoskeleton to the membrane may be mediated via this 30-kDa domain, a function that requires formation of multiprotein complexes at the plasma membrane. Synthetically tagged peptides and bacterially expressed proteins were used to map the protein 4.1 binding site on human erythroid glycophorin C, a transmembrane glycoprotein, and on human erythroid p55, a palmitoylated peripheral membrane phosphoprotein. The 30-kDa domain of protein 4.1 binds to a 12-amino acid segment within the cytoplasmic domain of glycophorin C and to a positively charged, 39-amino acid motif in p55. Sequences similar to this charged motif are conserved in other members of the p55 superfamily, including the Drosophila Discs-large tumor suppressor protein. Thus protein 4.1, known to interact with the cytoskeleton, also interacts with Discs large family members (Marfatia, 1995).

Distribution of two family 4.1 proteins, Expanded and Coracle, are disrupted in dlg mutants. Loss of Discs large also affects the distribution of Fasciclin III and neuroglian, two transmembrane proteins thought to be involved in cell adhesion (see in particular Discs large: Biological overview). These results suggest that Dlg serves as a binding protein linking cell surface receptors with the cytoskeleton via family 4.1 proteins (Woods, 1996).

One essential function of epithelia is to form a barrier between the apical and basolateral surfaces of the epithelium. In vertebrate epithelia, the tight junction is the primary barrier to paracellular flow across epithelia, whereas in invertebrate epithelia, the septate junction (SJ) provides this function. New proteins have been identified that are required for a functional paracellular barrier in Drosophila. In addition to the previously known components Coracle (Cora) and Neurexin (Nrx), four other proteins [Gliotactin, Neuroglian (Nrg), and both the alpha and ß subunits of the Na+/K+ ATPase] are required for formation of the paracellular barrier. In contrast to previous reports, it is demonstrated that the Na pump is not localized basolaterally in epithelial cells, but instead is concentrated at the SJ. Data from immunoprecipitation and somatic mosaic studies suggest that Cora, Nrx, Nrg, and the Na+/K+ ATPase form an interdependent complex. Furthermore, the observation that Nrg, a Drosophila homolog of vertebrate neurofascin, is an SJ component and is consistent with the notion that the invertebrate SJ is homologous to the vertebrate paranodal SJ. These findings have implications not only for invertebrate epithelia and barrier functions, but also for understanding of neuron-glial interactions in the mammalian nervous system (Genova, 2003).

To identify additional components of the Drosophila SJ, a collection of P element insertion mutations was screened for a phenotype attributable to a loss of the paracellular barrier. Two genes, Na pump alpha subunit (Atpalpha) and Nervana 2 (Nrv2), which encodes the ß subunit of the Na+/K+ ATPase) were identified as essential for the barrier function of the SJ. In addition, Neuroglian (Nrg), which is homologous to known components of the PSJ, and Gli, which is necessary for the blood-brain barrier, were tested and found to be necessary for the paracellular barrier. Direct immunostaining, epitope-tagged expression constructs, and GFP-tagged proteins indicate that Nrv2, ATPalpha, and Nrg localize to the SJ, and that they are interdependent for this localization. In keeping with this finding, the existence of a protein complex containing Cora, Nrx, Nrg, and Nrv is demonstrated. Taken together, these results suggest a novel complex involving the Na+/K+ ATPase that is necessary for establishing and maintaining the primary paracellular barrier in invertebrate epithelia, the SJs. Thus these studies provide new insights into the structure and function of SJs in both invertebrate epithelial cells and in the homologous PSJ of the vertebrate nervous system (Genova, 2003).

Cora has been shown to bind to the cytoplasmic tail of Nrx in the SJ. Studies of the PSJ have shown that the mammalian homologs of Nrx and Nrg interact via their extracellular domains. Together, these observations suggest the existence of a multiprotein complex at the SJ in which Cora binds to Nrx, which in turn binds to Nrg. The finding that Nrx and Nrg coimmunoprecipitate when either anti-Cora or anti-Nrg antibodies are used to immunoprecipitate is consistent with this model. Because Drosophila epithelial cells express all three proteins, it is not possible to rigorously distinguish whether this interaction occurs within the same cell or between adjacent cells. However, the observation that wild-type cells are unable to efficiently assemble Cora and Nrx at the boundary with cora^- cells suggests that intercellular interaction with the same complex on adjacent cells is required for SJ assembly. In addition, Nrv is found to coimmunoprecipitates with both Cora and Nrx. Nrg has not been detected in this complex, suggesting that the interaction between NRV2 and the Cora-Nrx complex occurs independently of Nrg, perhaps on the cytoplasmic side of the membrane. Although these results imply the possibility of an interaction between Cora and the cytoplasmic tail of NRV2, this seems unlikely in light of observations that NRV1, 2.1, and 2.2 all localize to the SJ, despite having different cytoplasmic tails. Thus, it is more likely that the interaction between Cora and the ATPase occurs either through Nrx or the alpha subunit (Genova, 2003).

Somatic mosaic analysis has demonstrated that this complex of Cora, Nrx, Nrv, ATPalpha, and Nrg can be disrupted without affecting overall polarity, or other components of the SJ. No component essential for the paracellular barrier has been identified that is unaffected in mutant cells, suggesting that the substrate upon which this complex assembles has yet to be found. Previous studies have demonstrated that Ankyrin binds both the cytoplasmic domain of Nrg and, as has been described in mammalian cells, the alpha subunit of NA⁺/K⁺ ATPase. In addition, Ankyrin colocalizes with Nrg at points of Nrg-induced S2 cell adhesion complexes. Thus, one candidate for a substrate upon which this complex assembles is Ankyrin, a well-known member of the membrane skeleton (Genova, 2003).

Muscle dystroglycan organizes the postsynapse and regulates presynaptic neurotransmitter release at the Drosophila neuromuscular junction: Dg is the principal component involved in Cora localization

The Dystrophin-glycoprotein complex (DGC) comprises dystrophin, dystroglycan, sarcoglycan, dystrobrevin and syntrophin subunits. In muscle fibers, it is thought to provide an essential mechanical link between the intracellular cytoskeleton and the extracellular matrix and to protect the sarcolemma during muscle contraction. Mutations affecting the DGC cause muscular dystrophies. Most members of the DGC are also concentrated at the neuromuscular junction (NMJ), where their deficiency is often associated with NMJ structural defects. Hence, synaptic dysfunction may also intervene in the pathology of dystrophic muscles. Dystroglycan is a central component of the DGC because it establishes a link between the extracellular matrix and Dystrophin. This study focused on the synaptic role of Dystroglycan (Dg) in Drosophila. Dg is concentrated postsynaptically at the glutamatergic NMJ, where, like in vertebrates, it controls the concentration of synaptic Laminin and Dystrophin homologues. Synaptic Dg controls the amount of postsynaptic 4.1 protein Coracle and alpha-Spectrin, as well as the relative subunit composition of glutamate receptors. In addition, both Dystrophin and Coracle and required for normal Dg concentration at the synapse. In electrophysiological recordings, loss of postsynaptic Dg did not affect postsynaptic response, but, surprisingly, led to a decrease in glutamate release from the presynaptic site. Altogether, this study illustrates a conservation of DGC composition and interactions between Drosophila and vertebrates at the synapse, highlights new proteins associated with this complex and suggests an unsuspected trans-synaptic function of Dg (Bogdanik, 2008).

The widely accepted hypothesis about the function of the DGC complex is its protective role in the sarcolemma against muscle contraction induced size changes. This study analyzed the synaptic function of a core member of the DGC, Dystroglycan. Drosophila Dg is concentrated at the NMJ, and most Dg immunoreactivity at the NMJ is postsynaptic. A proportion of synaptic Dg contained the mucin-like domain (MLD), which is the most heavily glycosylated domain in vertebrate Dg. Haines (2007) has shown that the MLD containing Drosophila Dg isoform is indeed glycosylated. Thus, like the vertebrate cholinergic NMJ, the Drosophila NMJ is enriched in Dg, and notably in glycosylated forms of this protein. These data are in accordance with concentration of Dystrophin at the Drosophila NMJ, suggesting the presence of all DGC members at the postsynapse (Bogdanik, 2008).

It is possible that the NMJ defects observed in the dg mutants used in this study are a consequence of a general muscle dysfunction, due to the loss of Dg at extrasynaptic sites. Indeed, muscle dysfunction has been observed in dg null mutants that are lethal at the embryonic and first instar larval stage. However, the mutants analyzed in this study are hypomorphs and the allelic combination used, dg^e01554/dg³²³, is viable. The larvae crawl, pupate and give rise to fertile adults, which do not show any wing position phenotype corresponding to flight muscle degeneration. Although it cannot be ruled out that there are some subtle muscle defects at extrasynaptic sites, the data illustrate that synaptic electrophysiological and morphological defects are already present in these mild loss of function conditions (Bogdanik, 2008).

The lanA gene, encoding a Laminin A subunit, stabilizes the initial motoneuron/muscle contact during synaptogenesis. This study shows that Laminin is still present during late larval stages, and that it is concentrated around synapses in varicosities. The data indicate that, like in mice where Dg is required for synaptic Utrophin, Laminin alpha5 and Laminin alpha1 concentration, Drosophila Dg controls synaptic Laminin and Dystrophin concentration. In addition, Dystrophin is required for Dg sarcolemmal localization in vertebrate muscles, and both Dystrophin and Utrophin account for part of the clustering of Dg at the NMJ. This study shows that Dystrophin also controls synaptic Dg concentration. Thus the interdependence between Laminin, Dg and Dystrophin at the NMJ seems to be conserved phylogenetically. Importantly, in dystrophin/utrophin double mutants, a significant amount of Dg remains at the synapse, indicating that other proteins control, in parallel, its synaptic localization. The current observations indicate that, similarly, the Utrophin-Dystrophin homologue in flies does not account for the whole synaptic localization of DG, and Coracle was identified as a new, additional synaptic anchor for Dg (Bogdanik, 2008).

Looking for any new potential partners of Dg, Cora localization was studied in late larval stages at the NMJ. Cora has a function in early larval stages, but no clear synaptic localization of Cora was seen in late larval stages, as seen using a monoclonal antibody recognizing all Cora isoforms. Instead, a strong immunoreactivity in NMJ associated glial cells has been reported. A polyclonal antibody was used that recognized only the large Cora isoform. With this antibody, no immunoreactivity was detected in any NMJ associated glial cell, but a postsynaptic concentration of Cora, which partially disappeared in a cora hypomorph mutant, was easily detected and increased when Cora was overexpressed in the muscle. These data indicated that the observed staining was indeed Cora. The Localization of protein 4.1 members in vertebrate muscle fibers is not well documented. It has been shown that protein 4.1R isoforms were indeed present in the muscle cells, notably at the cell periphery (probably the sarcolemma). Interestingly, in DMD patients, the peripheral localization of protein 4.1R isoforms is lost, although the sub-sarcolemmal spectrin cytoskeleton is still present. This set of data already indicates that protein 4.1 sarcolemmal localization is dependent on the DGC complex. The current data show that this is the case at the NMJ, and that Dg is the principal component involved in Cora localization, since loss of postsynaptic Dys gives much weaker phenotypes compared to loss of postsynaptic Dg. In addition, Cora was shown to co-immunoprecipitate with Dg, indicating the presence of the two proteins in the same complex, although further biochemical analysis will be required to assess whether they interact directly or indirectly (Bogdanik, 2008).

Unexpectedly, it was observed that Cora was required for the normal postsynaptic localization of Dg and, to a lesser extent, of Dys. This result was observed using a hypomorph cora mutant in which the C-terminal domain is partially deleted. In this mutant, synaptic amount of Cora was strongly reduced. Further structure-function studies will be required to understand 1) which domain of Cora is required for its synaptic localization and for its interaction with Dg, 2) which part of Dg C-terminal tail is involved in Cora interaction. Previous studies have shown that the juxtamembrane region of the C-terminal Dg tail interacts with Ezrin, a protein containing a FERM domain, like Cora. It is possible that the same Dg domain interacts with Cora (Bogdanik, 2008).

Since Cora controls synaptic GluRIIA abundance, an expected consequence of the loss of synaptic Cora in dg mutant NMJ was a reduction in the amount of GluRIIA subunit at the NMJ. Such a reduction was found, but to a mild degree. This small effect may be due to the fact that dg-induced reduction of synaptic Cora is not as strong as a complete cora loss of function, which was the situation analyzed originally. The small effect observed on DGluRIIA probably explains why there was no change in the amplitude of mEJCs in dg loss of functions. Indeed, DGluRIIA is the dominant subunit compared to DGluRIIB and a significant loss of DGluRIIA should lead to a decrease in mini amplitude (Bogdanik, 2008).

Loss of synaptic Dystroglycan resulted in a clear decrease in postsynaptic spectrin cytoskeleton, as assessed with alpha-Spectrin immunoreactivity. Although the spectrin defect may be a consequence of the loss of synaptic Cora, a more direct interaction between Dg and the spectrin cytoskeleton remains a possibility. Hence, the link between Dg, Cora and spectrin cytoskeleton remains to be further defined. The postsynaptic spectrin cytoskeleton has been shown to play a role in the repartition of postsynaptic receptor fields. Indeed, loss of postsynaptic immunoreactivity for both alpha and beta-Spectrin leads to a disorganization of postsynaptic receptor fields. Such a defect was sught in the dystroglycan loss of function conditions, but but none was found. This is probably due to the fact that the loss of spectrin immunoreactivity in these mutants was not complete (Bogdanik, 2008).

This study demonstrated that Dg plays a functional role in neuromuscular synaptic transmission. Indeed, glutamate release was decreased by approximately 40% in absence of muscle Dg. The main specificity of the insect NMJ, compared to the vertebrate NMJ is the presence of glutamate as a neurotransmitter instead of acetylcholine. Hence, these synapses are not only NMJ models, but also models of glutamatergic synapses, which are by far the most frequent synapses found in the vertebrate brain. Study of Dg function in mammalian brain synapses has illustrated an alteration of LTP in DG^-CNS mice, but no modification of the amplitude of synaptic responses evoked by low frequency stimulation of Schaeffer collaterals, and no changes in paired-pulse facilitation. In the current study, a reduced synaptic response was detected at low frequency, indicating a function of Dg in basal glutamatergic synaptic transmission (Bogdanik, 2008).

One surprising result in the electrophysiology experiments was the fact that defects in quantal content of the dg mutant are also present, with the same intensity in flies expressing a 24B Gal4 driven dg-RNAi. This indicated that loss of postsynaptic Dg leads to a functional change in the other synaptic compartment, the presynapse. Such a presynaptic effect associated with postsynaptic modifications is not new, since the NMJ function displays homeostasis, and decrease in postsynaptic responsiveness is often associated to increase in neurotransmitter release and vice-versa in order to maintain constant EJCs. The molecular mechanisms involved in this homeostatic control are largely unknown. In this study, in dg mutants, homeostatic control is likely absent since mini amplitude (receptor field) is not altered in absence of postsynaptic Dg, but glutamate release is modified. This suggests that Dg-deficient muscles inappropriately signals to the presynaptic release machinery. Previous studies have observed a similar trans-synaptic effect of loss of muscle Dystrophin onto presynaptic quantal content has been observed. What can be the mechanisms involved? One possibility is that postsynaptic Dg directly controls the levels of synaptic ECM molecules such as Laminin. These proteins, by interacting with presynaptic receptors, would affect the structure of the presynapse, e.g. the amount, size or molecular composition of active or periactives zones. This hypothesis is strongly supported by the finding in mouse, that a synaptic Laminin-calcium channel interaction organizes active zones in motor nerve terminals. Another presynaptic Laminin receptor could be the synaptic vesicle protein SV2. Looking for modifications at the presynapse, no obvious change was detected in the number and size of active zones, using Bruchpilot immunoreactivity as a marker and no modification was detected in the immunoreactivity of the periactive zone marker Fas2. Still, the regulation of synaptic Laminin by Dg, together with the observed presynaptic electrophysiological phenotype, make the hypothesis of Laminin bridging postsynaptic Dg and the presynapse, at least in periactive zones, very likely (Bogdanik, 2008).

These findings, i.e. the new components of a Dystroglycan complex, as well as the unexpected trans-synaptic role of Dg pave the way for understanding the role of the DGC in the formation, maintenance and plasticity of glutamatergic synapses (Bogdanik, 2008).

The 4.1 protein coracle mediates subunit-selective anchoring of Drosophila glutamate receptors to the postsynaptic actin cytoskeleton

Glutamatergic Drosophila neuromuscular junctions contain two spatially, biophysically, and pharmacologically distinct subtypes of postsynaptic glutamate receptor (GluR). These receptor subtypes appear to be molecularly identical except that A receptors contain the subunit GluRIIA (but not GluRIIB), and B receptors contain the subunit GluRIIB (but not GluRIIA). A- and B-type receptors are coexpressed in the same cells, in which they form homotypic clusters. During development, A- and B-type receptors can be differentially regulated. The mechanisms that allow differential segregation and regulation of A- and B-type receptors are unknown. Presumably, A- and B-type receptors are differentially anchored to the membrane cytoskeleton, but essentially nothing is known about how Drosophila glutamate receptors are localized or anchored. This study identified Coracle, a homolog of mammalian brain 4.1 proteins, in yeast two-hybrid and genetic screens for proteins that interact with and localize Drosophila glutamate receptors. Coracle interacts with the C terminus of GluRIIA but not GluRIIB. To test whether coracle is required for glutamate receptor localization, receptors were immunocytochemically and electrophysiologically examined in coracle mutants. In coracle mutants, synaptic A-type receptors are lost, but there is no detectable change in B-type receptor function or localization. Pharmacological disruption of postsynaptic actin phenocopies the coracle mutants, suggesting that A-type receptors are anchored to the actin cytoskeleton via Coracle, whereas B-type receptors are anchored at the synapse by another (yet unknown) mechanism (Chen, 2005; full text of article).

Home page: The Interactive Fly © 1997 Thomas B. Brody, Ph.D.

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