NG2 proteoglycan is expressed exclusively by mural cells during vascular morphogenesis.

Immunofluorescence mapping demonstrates that the NG2 proteoglycan is invariably expressed by the mural cell component of mouse neovascular structures. This pattern is independent of the developmental mechanism responsible for formation of the vasculature (vasculogenesis or angiogenesis). Thus, NG2 is expressed in the embryonic heart by cardiomyocytes, in developing macrovasculature by smooth muscle cells, and in nascent microvessels by vascular pericytes. Due to the scarcity of proven markers for developing pericytes, NG2 is especially useful for identification of this cell type. The utility of NG2 as a pericyte marker is illustrated by two observations. First, pericytes are associated with endothelial tubes at an early point in microvessel development. This early interaction between pericytes and endothelial cells has important implications for the role of pericytes in the development and stabilization of microvascular tubes. Second, the pericyte to endothelial cell ratio in developing capillaries varies from tissue to tissue. Because the extent of pericyte investment is likely to affect the physical properties of the vessel in question, it is important to understand the mechanisms that control this process. Additional insight into these and other aspects of vascular morphogenesis should be possible through use of NG2 as a mural cell marker.

Chondroitin sulfate and cytoplasmic domain-dependent membrane targeting of the NG2 proteoglycan promotes retraction fiber formation and cell polarization.

Targeting of the NG2 proteoglycan to cellular retraction fibers was studied by expressing mutant NG2 molecules lacking specific structural elements of the proteoglycan. Both the cytoplasmic domain and the chondroitin sulfate chain of NG2 appear to have roles in sorting NG2 to subcellular microdomains destined to become retraction fibers. Neither of these structural features alone is sufficient to allow optimal targeting of NG2 to retraction fibers, but together they promote efficient localization of the proteoglycan to these sites. This pattern of NG2 sorting seems to be necessary for optimal retraction fiber formation, as cells expressing poorly targeted NG2 mutants are noticeably deficient in their ability to extend retraction fibers. Furthermore, retraction fiber formation correlates strongly with the tendency of cells to assume a polarized morphology with NG2-positive retraction fibers at one pole of the cell and actin-rich lamellipodia at the other. This polarization can be triggered either through engagement of NG2 by the substratum or by exposure to lysophosphatidic acid, a potent activator of the rho GTPase. These results suggest a possible role for NG2 in regulating rho-dependent mechanisms in the trailing processes of motile cells.

Cytoskeletal reorganization induced by engagement of the NG2 proteoglycan leads to cell spreading and migration.

Cells expressing the NG2 proteoglycan can attach, spread, and migrate on surfaces coated with NG2 mAbs, demonstrating that engagement of NG2 can trigger the cytoskeletal rearrangements necessary for changes in cell morphology and motility. Engagement of different epitopes of the proteoglycan results in distinct forms of actin reorganization. On mAb D120, the cells contain radial actin spikes characteristic of filopodial extension, whereas on mAb N143, the cells contain cortical actin bundles characteristic of lamellipodia. Cells that express NG2 variants lacking the transmembrane and cytoplasmic domains are unable to spread or migrate on NG2 mAb-coated surfaces, indicating that these portions of the molecule are essential for NG2-mediated signal transduction. Cells expressing an NG2 variant lacking the C-terminal half of the cytoplasmic domain can still spread normally on mAbs D120 and N143, suggesting that the membrane-proximal cytoplasmic segment is responsible for this process. In contrast, this variant migrates poorly on mAb D120 and exhibits abnormal arrays of radial actin filaments decorated with fascin during spreading on this mAb. The C-terminal portion of the NG2 cytoplasmic domain, therefore, may be involved in regulating molecular events that are crucial for cell motility.

Tyrosine phosphorylation of L1 family adhesion molecules: implication of the Eph kinase Cek5.

The L1 family comprises transmembrane cell adhesion molecules of the immunoglobulin superfamily that play an important role in neuronal migration and axon outgrowth, fasciculation, and myelination. Consistent with a crucial role in developmental processes, mutations in L1 cause severe brain malformations. Although L1 activates intracellular signaling pathways, little is known about the membrane proximal events of L1 signaling. The cytoplasmic domains of L1 family proteins contain several conserved tyrosine residues that are potential targets for receptor tyrosine kinases. Here, we report that the L1 family protein Ng-CAM is phosphorylated on tyrosine in embryonic day 13 chicken retina. This is the first demonstration of in vivo tyrosine phosphorylation of an L1-like molecule. Because chicken embryo kinase 5 (Cek5) is a receptor tyrosine kinase expressed in neuronal processes and activated in the chicken embryonic retina, we have analyzed the possible role of Cek5 in L1 phosphorylation. The rat glioblastoma cell line B28 was stably transfected with human L1. Additional transient transfection with Cek5 cDNA led to expression of Cek5 in its tyrosine-phosphorylated, activated form. Biochemical analysis revealed that L1 is phosphorylated on tyrosine in Cek5-transfected cells but not in control transfectants. Furthermore, direct phosphorylation of the L1 cytoplasmic domain by Cek5 was demonstrated in an in vitro kinase assay. Tyrosine phosphorylation may represent a novel mechanism of signal cascade initiation through L1.

Mutational analysis of the L1 neuronal cell adhesion molecule identifies membrane-proximal amino acids of the cytoplasmic domain that are required for cytoskeletal anchorage.

The preferential localization of the L1 cell adhesion molecule in the axons and growth cones of differentiating neurons suggests the existence of a mechanism for targeting or anchoring the molecule to these locations. We have used B28 glioma cells, which have an extremely flattened morphology, as a model system to study the organization of L1 on the cell structure. Transfection of L1 cDNA into B28 cells results in expression of the L1 protein in organized linear cell surface arrays which are codistributed with cytoskeletal stress fibers, but not with microtubles or intermediate filaments. Transfection studies with L1 deletion mutants identify the juxtamembrane segment of the cytoplasmic domain as the critical entity for arrangement of L1 into ordered cell surface arrays. The seventh cytoplasmic amino acid of L1, lysine 1150, and to a lesser extent the fourth cytoplasmic amino acid, lysine 1147, appear to be critical residues for maintaining normal L1 anchorage and distribution.

Interaction of the NG2 proteoglycan with the actin cytoskeleton.

The NG2 chondroitin sulfate proteoglycan is a membrane-spanning molecule expressed by immature precursor cells in a variety of developing tissues. In tightly adherent cell lines with a flattened morphology, NG2 is organized on the cell surface in linear arrays that are highly co-localized with actin and myosin-containing stress fibers in the cytoskeleton. In contrast, microtubules and intermediate filaments in the cytoskeleton exhibit completely different patterns of organization, suggesting that NG2 may use microfilamentous stress fibers as a means of cytoskeletal anchorage. Consistent with this is the observation that cytochalasin D disrupts the organization of both stress fibers in the cytoskeleton and NG2 on the cell surface. Very similar linear cell surface arrays are also seen with three other cell surface molecules thought to interact with the actin cytoskeleton: the alpha 5 beta 1 integrin, the CD44 proteoglycan, and the L1 neuronal cell adhesion molecule. Since the cytoplasmic domains of these four molecules are dissimilar, it seems possible that cytoskeletal anchorage in each case may occur via different mechanisms. One indication of such differences can be seen in colchicine-treated cells which have lost their flattened morphology but still retain long actin-positive tendrils as remnants of the actin cytoskeleton. NG2 and alpha 5 beta 1 are associated with these tendrils while CD44 and L1 are not, suggesting that at least two subclasses of cell surface molecules exist which can interact with different subdomains of the actin cytoskeleton.