|Home | About | Journals | Submit | Contact Us | Français|
Activity-independent and activity-dependent mechanisms work in concert to regulate neuronal growth, ensuring the formation of accurate synaptic connections. CPG15, a protein regulated by synaptic activity, functions as a cell-surface growth-promoting molecule in vivo. In Xenopus laevis, CPG15 enhanced dendritic arbor growth in projection neurons, with no effect on interneurons. CPG15 controlled growth of neighboring neurons through an intercellular signaling mechanism that requires its glycosylphosphatidylinositol link. CPG15 may represent a new class of activity-regulated, membrane-bound, growth-promoting proteins that permit exquisite spatial and temporal control of neuronal structure.
The cpg15 gene was identified in a forward genetic approach designed to isolate activity-regulated genes that mediate synaptic plasticity (1). In the adult rat, cpg15 is induced in the brain by kainate (KA) and in visual cortex by light (2). During development, cpg15 expression is correlated with times of afferent in-growth, dendritic elaboration, and synaptogenesis (2). Sequence analysis predicts a small, secreted protein (2) that is membrane-bound by a glycosylphosphatidylinositol (GPI) linkage (3).
Antiserum generated against bacterially expressed rat CPG15 recognizes a protein from rat brain dentate gyrus extracts (Fig. 1A) (4) of the size predicted by sequence analysis. A protein of similar size is induced in Xenopus laevis after KA injections into the brain ventricle (Fig. 1A) (5). In situ hybridizations using a partial clone of Xenopus cpg15 indicate that the CPG15 mRNA is expressed in retinal ganglion cells and in differentiated neurons throughout the central nervous system (CNS) of stage-47 tadpoles (6). Xenopus CPG15 protein is present in neurons and axons throughout the CNS (7, 8). In the optic tectum, differentiated neurons label in a honeycomb pattern similar to N-CAM (neural cell adhesion molecule) and other cell-surface antigens, while cells in the proliferative zone have undetectable levels of CPG15 (Fig. 1C).
To investigate the cellular function of CPG15, we used a recombinant vaccinia virus (VV) to express CPG15 in optic tectal cells of albino Xenopus tadpoles (9, 10). Tadpoles were infected by ventricular injection with VV carrying rat cpg15 and β-galactosidase (β-gal) cDNAs in a dual promoter vector, or with a control virus containing only the β-gal cDNA (11). Two days after viral infection and approximately 24 hours after the beginning of expression of foreign protein (9), single tectal cells were labeled with DiI (10, 12). Confocal images through the entire structure of each neuron were collected at 24-hour intervals over a period of 3 days, and three-dimensional (3D) images were reconstructed from this (13).
The most prominent effect of CPG15 on the morphology of tectal projection neurons was that the dendritic arbors of neurons from CPG15VV-infected animals increased their total dendritic branch length (TDBL) and became more complex than arbors of neurons from β-gal–infected or uninfected animals (Fig. 2) (14). This effect was quantified as an increase in averaged TDBL (Fig. 3A) and by Sholl analysis (Fig. 3B).
We measured the distribution of dendritic arbor sizes, expressed as TDBL, within the population of neurons from CPG15VV-infected animals and from control animals (Fig. 3C). All three populations of neurons showed a gradual shift toward larger TDBLs as their dendritic arbors grow. The shift toward larger TDBLs was greatest in neurons from CPG15VV-infected animals. This analysis also demonstrates the presence of a subpopulation of cells in CPG15VV-infected animals that have a greater TDBL than any control cell. These “outliers” can be detected on all three days of imaging and are the most distinctive manifestation of the CPG15 overexpression phenotype.
Interneurons were unaffected by CPG15VV infection (Fig. 4) (14, 15). It is therefore likely that the regulation of interneuron dendritic development is controlled differently from that of projection neurons, perhaps by different molecular participants.
To determine whether CPG15 could mediate intercellular signaling, we established experimental conditions in which imaged cells were not likely to be infected with the CPG15VV. Animals were infected with CPG15VV at a lower titer than in previous experiments, imaged in vivo over a 2-day period, then fixed and processed for β-gal immunohistochemistry (16). Because the VV contains a dual promoter vector with the CPG15 cDNA cloned downstream of a strong promoter and the β-gal cDNA driven by the weaker p7.5 vaccinia promoter, β-gal expression served as our marker for infected cells that were also expressing the CPG15 protein. β-gal immunohistochemistry showed that none of the 14 DiI-labeled neurons imaged in this experiment were infected with CPG15VV, probably due to the low infection levels. Three of 14 neurons had TDBLs greater than any control neuron and were therefore outliers. We found that outliers do not need to express CPG15 in order to exhibit an enhanced growth response; however, each one was in close proximity to infected neurons or radial glia (Fig. 5) (14). This demonstrates that the CPG15 growth promoting effect is mediated by a signaling molecule capable of communicating between neighboring cells, possibly CPG15 itself.
Database searches do not reveal CPG15-related proteins. The CPG15 amino acid sequence lacks the immunoglobulin G domains common to adhesion molecules (17). CPG15 shows borderline similarity to ligands of the eph subfamily of receptor tyrosine kinases (ephrins) (18). Two of the ephrins—ephrin-A2 (ELF-1; Cek7-L; LERK6) and ephrin-A5 (AL1; RAGS; LERK7)—are anchored to the membrane by a GPI linkage (18, 19) and have been implicated as guidance molecules for topographic mapping of retinotectal projections (20). It appears that these cell-surface–bound ligands are unable to function as soluble factors and require presentation in a membrane-bound form to activate their receptors by direct cell-to-cell contact (21). We noted that CPG15-expressing neurons and glia were located near the uninfected neurons with the CPG15 “outlier” phenotype. To test whether CPG15 requires the GPI linkage for its growth promoting function, we infected tectal cells with a VV containing a truncated version of CPG15 lacking the GPI consensus sequence (CPG15t3VV) (22). CPG15t3VV was expressed in infected cells (Fig. 1, F and G), but failed to promote dendritic arbor growth (Fig. 3D). Indeed, neurons in CPG15t3VV-infected animals have a significantly slower growth rate than neurons from β-galVV–infected animals (*P < 0.02; Fig. 3D). This indicates that CPG15 functions to promote dendritic growth only in its membrane-bound form. Furthermore, the truncated CPG15 inhibits normal dendritic growth in developing neurons. A truncated human homolog of CPG15 (neuritin) was previously reported to promote neuritogenesis in cultured rat hippocampal neurons (3). These qualitative observations may reflect characteristics of the in vitro assay (for example, CPG15t3 immobilization through binding to the polylysine substrate or different response properties of the cultured neurons).
During nervous system development, synaptic activity influences the formation of neuronal connections, in part by controlling axonal and dendritic structure (23). Our studies are consistent with a model in which visual activity induces CPG15, which in turn promotes neuronal growth and structural plasticity. We show that CPG15, an activity-regulated molecule, promotes dendritic growth through intercellular signaling, suggesting that it may function as a ligand. It is intriguing to consider that GPI-linked ligands may function not only as passive positional labels, as in the case of ephrins, but also as activity-regulated growth-promoting signals. As a membrane-bound protein, CPG15 might confer spatial specificity to a growth promoting signal, which may be lacking in secreted factors. As more proteins are discovered that influence neuronal outgrowth in the brain, studies of their functions will enhance the understanding of wiring of neuronal connections and their modification by activity.
We thank B. Burbach, K. Bronson, I. Miloslavskaya, and N. Dawkins for excellent technical assistance, Z. Li for making the pSC65-CPG15t3 construct, and R. Malinow, K. Svoboda, and J. Yin for critical reading of the manuscript. Supported by NIH (H.T.C. and E.N.), the National Down Syndrome Society (H.T.C.), and the Marie Robertson Fund (E.N.).