BIOLOGICAL OVERVIEW

Cadherin-N (CadN and also referred to as N-cadherin) is the fourth cadherin to be discovered in Drosophila. Shotgun, to date the best characterized Drosophila cadherin, is predominantly expressed in epithelial tissues in embryos and is critically required for dynamic epithelial rearrangements during embryogenesis. Two other proteins of the Drosophila cadherin superfamily have been characterized: Fat, a tumor suppressor, and Dachsous. None of these proteins are expressed primarily in the nervous system. While Shotgun is expressed in the embryonic nervous system, its expression is limited to midline glial cells of the central nervous system and to adherens junctions between sensory dendrites and their accessory cells in the periphery (Iwai, 1997).

Cadherin-N was discovered in an intensive search for novel Drosophila cadherin genes. CadN differs from Shotgun and from classic vertebrate cadherins by the presence of 15 cadherin repeats, versus the 6 found in Shotgun and the 4 found in all typical vertebrate cadherins. Drosophila Shotgun and Cadherin-N also share a similar complex region adjacent to the transmembrane domain. This region is not found in vertebrate cadherins. Nevertheless, the intracellular domains of both Cadherin-N and Shotgun bears a strong resemblence to the intracellular domains of vertebrate cadherins. It is apparent that the extracellular domains have evolved at a different (more conservative) rate than the extracellular domains (Iwai, 1997).

Fasciclin II-expressing axonal tracts are deformed in mutant flies. Anti-Fas II antibody labels four neurons that pioneer the first two longitudinal axon pathways: vMP2 and MP1. Loss of Cadherin-N does not appear to perturb the pathfinding of the pioneers at stage 12 and 13, since the pioneer growth cones normally navigate through the vMP2 pathway, and the pathway is generated as in the wild type. At stage 14, when follower neurons begin Fas II expression and join the pioneer tracts, three classes of pattern alteration are recognized. (1) In normal embryos, the vMP2 tract is constricted toward the midline in a segmentally repeated fashion, whereas the mutant route weaves to a lesser extent. Moreover, local associations between the vMP2 and MP1 fascicles are diminished in the mutants. (2) One commissural axon fascicle periodically detected in each segment stains more intensely in mutants than in wild type, implying that more axons join the mutant fascicle. (3) The MP1 pathway sometimes looks discontinuous. Irregular patterns become more conspicous at late stage 16 when Fas II-positive axons are assembling into three longitudinal bundles. Occasionally, the two more laterally located pathways are interupted, and the disconnected terminals are often swollen and/or have turned laterally. The bundles bifurcate locally, which represents either defasciculation or abnormal fusion (Iwai, 1997).

Apterous can be used as a marker for neuron growth directed by CadN. The LIM homeodomain transcription factor Apterous is expressed in only three interneurons per abdominal hemisegment; the fascicles of their ipsilaterally (same side) projecting axons can be monitored using DNA to encode an enzyme marker driven by the apterous promoter. The Apterous axons first extend medially, then their growth cones make right-angled turns and grow in an anterior direction. Subsequently, the transverse portion of each Apterous axon starts a medial shift, and its turning point reaches the most medial surface of the longitudinal axon tracts. In Cadherin-N mutants, the inital outgrowth is normal, however after the turn, mutant fascicles take oblique trajectories in contrast to the parallel pathways in normal embryos. Most of the mutant growth cones of the Apterous neurons reach adjacent anterior segments and navigate along the most medial path. However, they cannot immediately fasciculate with the misrouted axons of counterpart neurons in more anterior segments. Without fascicle formation, mutant growth cones appear to continue navigation, indicating that bundling is not an absolute prerequisite for further axon growth. Because of the broad neuronal distribution of Cadherin-N, it is supposed that abnormal patterning is not restricted to Apterous neurons (Iwai, 1997).

sequoia-induced axon targeting is mediated through the ubiquitously expressed Cadherin-N cell adhesion molecule

The segregation of axon and dendrite projections into distinct synaptic layers is a fundamental principle of nervous system organization and the structural basis for information processing in the brain. Layer-specific recognition molecules that allow projecting neurons to stabilize transient contacts and initiate synaptogenesis have been identified. However, most of the neuronal cell-surface molecules critical for layer organization are expressed broadly in the developing nervous system, raising the question of how these so-called permissive adhesion molecules support synaptic specificity. This study showed that the temporal expression dynamics of the zinc-finger protein Sequoia is the major determinant of Drosophila photoreceptor connectivity into distinct synaptic layers. Neighbouring R8 and R7 photoreceptors show consecutive peaks of elevated sequoia expression, which correspond to their sequential target-layer innervation. Loss of sequoia in R7 leads to a projection switch into the R8 recipient layer, whereas a prolonged expression in R8 induces a redirection of their axons into the R7 layer. The sequoia-induced axon targeting is mediated through the ubiquitously expressed Cadherin-N cell adhesion molecule. The data support a model in which recognition specificity during synaptic layer formation is generated through a temporally restricted axonal competence to respond to broadly expressed adhesion molecules. Because developing neurons innervating the same target area often project in a distinct, birth-order-dependent sequence, temporal identity seems to contain crucial information in generating not only cell type diversity during neuronal division but also connection diversity of projecting neurons (Petrovic, 2008).

In the compound eye, each ommatidium contains eight photoreceptor neurons (R1-R8) that form synapses in distinct optic lobe ganglia, the peripheral lamina and the deeper medulla. Axons of the colour-sensitive R8 and R7 cells project into the medulla and segregate into two out of ten synaptic layers, M3 and M6 respectively. In a mosaic screen for genes that control R8/R7 target layer selection, mutants of sequoia (seq⁵ and seq⁶) were identified, with a frequent loss of R-cell innervation in M3 and M6. Using the MARCM (mosaic analysis with a repressible cell marker) technique for the selective labelling of homozygous mutant R8 and R7 cell axons, it was found that more than 90% of seq mutant R7 cells terminate their axon projections in the outer M1-M3 layers, correlating with an innervation gap in the M6 layer of the same medulla column. Similarly, seq mutant R8 axons frequently (87%) stop above their prospective M3 target layer. The analysis of different R8/R7 genetic mosaics showed that wild-type R8/R7 axons are not influenced by a seq mutant neighbour axon of the same medulla column, suggesting independent target recognition by these two R cells (Petrovic, 2008).

sequoia encodes a nuclear protein with two putative zinc-finger domains related to the DNA-binding domain of Tramtrack. In the visual system, the four independently isolated sequoia alleles seq⁵, seq⁶, seq²² and seq^SH1898 lead to indistinguishable R8/R7 axon-targeting phenotypes. No expression could be detected in homozygous seq⁵ and seq⁶ mutant R cells with an anti-sequoia monoclonal antibody and the R7 targeting defects could be rescued after the expression of a sequoia transgene in the mutant cells. To determine whether sequoia functions cell-autonomously in R cells to control axon target selection, R7 cell-type-specific MARCM clones were induced. Almost all of the single homozygous mutant R7 terminals surrounded by heterozygous R8 and R7 axons stop prematurely in M3. Because the loss of sequoia does not lead to changes in early R8 and R7 specification or their subsequent projection towards their optic lobe ganglion, it is concluded that sequoia is required for R8/R7 axons to connect to their synaptic target layer (Petrovic, 2008).

The sequoia targeting phenotype becomes obvious during early pupal development when the R8 and R7 axons terminate sequentially into two separate medulla layers. In contrast to the adult brain, the medulla target field during initial R8/R7 ingrowth appears more homogeneous and distinct layer organization becomes visible in the subsequent segregation of lamina and medulla neuron processes. seq mutant R8 and R7 axons reach the medulla target area at the appropriate time and in the correct topographic order, but gaps in the initial R7 layer can be detected; this is due to the termination of R7 axons in the R8 layer. Similarly to the R7 projection defect in the adult visual system, the mis-targeting of R7 into the R8 layer during the pupal stage seems to be due to the disruption of seq function directly in R7 and is not influenced by defects in R8 (Petrovic, 2008).

The expression pattern of sequoia is highly dynamic in early differentiating R8/R7 cells during the phase of axonal growth and medulla innervation. The onset of sequoia expression in R cells reflects their sequential birth and differentiation order. As a result of the temporal gap in neuronal differentiation between R8 and R7 from the same ommatidium, together with a rapid downregulation of sequoia, the periods of elevated sequoia expression in these two R cells are non-overlapping. By the time the R7 cell initiates axonal projection accompanied by a high expression level of sequoia, the axon of the neighbouring R8 cell has already reached the medulla, and sequoia expression has been turned off. Therefore a tight correlation is observed between the R-cell-specific sequoia expression profile and the sequential R8/R7 medulla target layer innervation (Petrovic, 2008).

The fact that sequoia is expressed in both R8 and R7, but the loss of sequoia leads to specific R8/R7 axon-targeting defects, raises the possibility that connection specificity is mediated through the difference in the temporal pattern of sequoia expression. To test the importance of precise sequoia expression timing, synchronized the R8/R7 sequoia expression profiles were syncronized. Similarly to the seq loss of function situation, constitutive seq expression in all R cells does not lead to changes in the cell fate of R8 or R7 or their initial axonal projection towards the medulla. However, whereas in wild-type individuals R8/R7 growth cones transform their morphology over subsequent hours from an 'expanded' to a 'condensed' appearance, growth cones of R8/R7 cells with a prolonged sequoia expression remain in the 'expanded' state and R8 axons extend towards and terminate precisely in the R7 layer. The co-innervation of R8 and R7 axons is maintained during the subsequent steps of medulla reorganization, because a single R8/R7 layer in M6 can be detected in the medulla of adult flies. These data indicate that the downregulation of sequoia expression in R8 cells is critical for their growth cones to become stabilized in their initial target layer. Prolonged sequoia expression specifically in R8 or R7 and the sequoia overexpression in R8 in an R7 deficient-background support the model in which the endogenous level of sequoia determines the competence of ingrowing R-cell axons to connect to their appropriate target layer (Petrovic, 2008).

To obtain insights into the molecular mechanisms through which R8/R7 axon targeting competence is mediated, whether the loss of sequoia leads to changes in the expression of neuronal cell adhesion/receptor molecules was tested. A significant decrease in the expression levels of the receptor protein tyrosine phosphatase Lar (leukocyte-antigen-related-like) and the homophilic cell adhesion molecule Cadherin-N (CadN) was found in sequoia mutant R cells. Although CadN is the only factor known so far to be essential for the initial R8/R7 axon targeting, its widespread expression in R cells and in the target region has made it unlikely that homophilic CadN interactions alone are sufficient to specify R8/R7 target choice. Therefore, whether the sequoia targeting mechanism would allow CadN to provide spatio-temporal specificity in initial axon-target adhesion was examined. Loss of CadN and sequoia leads to an identical early termination of R7 axons in the outer medulla layer and later co-innervation of R7 and R8 in the adult M3 layer. After the removal of CadN from sequoia-overexpressing R8/R7 cells, a complete suppression of the M3-->M6 layer switch can be observed because all brains show the characteristic M6-->M3 layer switch of CadN mutant R7 axons. This result indicates that the sequoia-mediated axon targeting functions through CadN. However, the sequoia-induced M3-->M6 layer switch is not caused simply by an increase in CadN adhesion, because overexpression of CadN does not affect the R8/R7 projections into the medulla and the overexpression of sequoia does not lead to a significant increase in the level of CadN in R cells. From these data it is conclude that homophilic CadN interactions are sequentially used in R8 and R7 to mediate the sequoia-regulated axon-target interaction (Petrovic, 2008).

On the basis of these results it is proposed that initial afferent lamination in the developing visual system of Drosophila is not controlled by the layer-specific expression of recognition molecules but is mediated through temporally restricted competence of ingrowing axons to interact with the target region. Here, permissive cell adhesion molecules provide spatio-temporal recognition specificity in axon-target interaction. Interestingly, Cadherin-N has been shown to function in subsequent steps of medulla innervation by lamina neurons, supporting the idea that the same type of adhesion molecule could be used in a repetitive fashion to support sequential layer formation. A critical aspect of the specificity of temporal recognition is the rapid downregulation of the targeting competence after the initial axon-target contact to prevent the reactivation of early projecting axons during the subsequent steps of innervation. The sequoia-mediated axon targeting mechanism controls an essential step in the coordinated development between sensory neurons and their central nervous system (CNS) target field in the Drosophila visual system. Starting with afferent-derived signals inducing early target neuron differentiation, the sequoia-controlled patterning of initial axon innervation ensures the correct R8/R7 positioning for subsequent afferent-induced target field organization and layer-specific molecule expression to stabilize the initial innervation (Petrovic, 2008).

Neural progenitors often generate distinct subtypes of neurons in an invariant temporal sequence during development. Similarly to its role in generating cell type diversity during neuronal division, temporal identity is used subsequently to generate connection diversity in projecting neurons. Mis-expression of sequoia in more mature R cells during the later steps of differentiation has no effect on axonal projection, indicating that R cells lose their competence to respond to the sequoia-induced targeting identity. A narrow developmental window of competence has also been described for the temporal identity factors in the embryonic and postembryonic nervous system. Similarly, dynamic expression of sequoia can be observed throughout the development of the nervous system in Drosophila, and seq mutants show a severe disruption of the connectivity pattern in various brain regions. In vertebrates, retina development is also characterized by the sequential generation of distinct cell types followed by their assembly into a highly laminated CNS structure. Recent in vivo imaging studies in zebrafish have shown that axon-dendrite interactions occur in a sequential manner and illustrate a transient requirement for some of the cell types during the assembly of laminated retinal circuits. Thus, a sequoia-related control mechanism might be more broadly applicable to the development of circuit specificity in the CNS (Petrovic, 2008).

PROTEIN STRUCTURE

Amino Acids - 3097

Structural Domains

The enormous size of Cadherin-N is primarily due to the presence of 15 cadherin repeats in the extracellular region, presenting a contrast to the 4 repeats in all vertebrate classic cadherins and the 6 repeats in Shotgun, the Drosophila E-cadherin. Both Drosophila cadherins have insertions of similar sequences between the last extracellular cadherin repeat and the membrane-spanning segment. The insert contains a series of subdomains: an Fcc box (fly classic cadherin box), a cysteine-rich segment (C-rich 1), a laminin A (See Drosophila Laminin A) globular segment (LmA-G), and another cysteine-rich segment (C-rich 2). The Fcc box, comprising 170 amino acids is defined as such because database searches with its sequences identify only the comparable regions of Shotgun as a relative. Similar sequences are not found in vertebrate cadherins. The whole Cadherin-N LmA-G displays 25% sequence identity to mouse laminin A and to the presynaptic transmembrane protein Neurexin. The Cadherin-N cytoplasmic domain is much more similar to those domains of Shotgun and vertebrate classic cadherins with respect to both size and sequence. The intracellular domains of the two Drosophila cadherins and mouse N-cadherin range between 157 and 160 amino acids in length, and have 37% and 46% sequence identity in any combination among them, with the higher figure representing the Cadherin-N identity to murine N-cadherin, and the lower representing Shotgun identity to murine N-cadherin. The degree of sequence conservation between the two Drosophila cadherins (41% identity) is lower than the 63% identity between N- and E-cadherin in the same vertebrate species, as for example, in mice (Iwai, 1997).

Antibodies were raised to the extracellular domain and to the cytoplasmic region of Cadherin-N and were designated as DN-Ex and DN-In, respectively. Both antisera recognize a pair of bands of roughly 300 kDA, of which the upper band is always faint. When alpha-catenin is immunoprecipitated, the smaller band is easily detected, while the large is not. It is assumed that the larger band is a precursor, and that the smaller represents the mature form generated by proteolytic digestion. In addition to these high molecular weight forms, DN-Ex detects a major 200 kDa molecule, and DN-ln labels an intense 120 kDa band. The production of these two fragments can be explained by postulating that the mature form is cleaved into an N-terminal 200 kDa and a C-terminal 120 kDa fragment. It is likely that the 200 Da form is derived only from the extracellular domain: the 200 kDa band is still present in a Cadherin-N mutant strain at a proximal position in the extracellular domain. Both 120 and 200 kDa forms coprecipitate with alpha-catenin (Iwai, 1997).

date revised: 10 February 2013

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.