To examine neuron-glia interactions of hippocampal cells, including glial-guided neuronal migration, glial organization of neuronal positioning and neuronal regulation of astroglial differentiation, rat hippocampal tissue, harvested between embryonic day 16 (E16) and postnatal day 3 (P3), was dissociated into a single cell suspension and plated in glass coverslip microcultures (Hatten and Liem, 1981; Hatten et al., 1984). Immunostaining the cells with antibodies against the glial filament protein (AbGFP) revealed developmental stage-specific changes in the number and extent of morphological differentiation of hippocampal astroglial cells. At E16-E18, fewer than 5% of the cells were AbGFP-positive; stained cells were immature, bearing very short processes. By E19-E20, the number of stained cells increased to 15% of the total cell population. Three forms of differentiated glial cells predominated, a bipolar form bearing processes 30-50 microns, an elongated form which resembled the radial glia of hippocampus, bearing processes 120 microns in length, and a stellate form with 3 or more processes 30-50 microns in length. At P0-P3, glial morphological differentiation varied with the culture substratum; differentiated forms resembling those seen at E20 occurred on Matrigel, but not on polylysine. Quantitation of the distribution of neurons relative to AbGFP-stained glial processes revealed developmental stage-specific changes in glial organization of neuronal positioning in the cultures. In cultures of E16-E18 hippocampal cells, the neurons did not preferentially associate with astroglial cells. By E19-E20, extensive neuron-glia interactions occurred, with 80-90% of the neurons being located within 5-10 microns of a glial process. In addition to their organization of neuronal positioning, E20 hippocampal astroglial cells supported extensive neuronal migration. Migrating hippocampal neurons displayed a cytology and neuron-glia cell apposition identical to that described for migrating cerebellar granule cells in vitro (Edmondson and Hatten, 1987), closely apposing their cell soma against the hippocampal glial process and moving along the glial arm by extending a thickened, leading process. Migration was seen only along highly elongated glial profiles resembling radial glial seen in vivo. The morphological differentiation of hippocampal glial cells in vitro was dependent on cell-cell interactions with neurons. In the absence of neurons, purified hippocampal astroglia had flat, undifferentiated profiles and proliferated rapidly. The addition of hippocampal neurons rapidly arrested glial growth and induced glial process extension.

To identify glial receptor systems in CNS migration, cerebellar granule neuron migration was assayed on glass fibers coated with polylysine, astroglial membranes (AM fibers), or the extracellular matrix proteins collagen (COLL fibers), fibronectin (FN fibers), and laminin (LAM fibers). By video microscopy, granule cells migrated along AM fibers with the cytology, neuron-fiber apposition, and dynamics seen on living glia. The demonstration that immobilized astroglial membranes support neural migration suggests that astroglial receptor systems, in combination with glial fiber geometry, promote CNS neural migration. Moreover, granule neurons migrated rapidly on LAM fibers, moved relatively slowly on FN fibers, and not at all on COLL fibers. Antibody perturbation analyses suggested that, whereas astrotactin provides the neural receptor for migration on astroglial membranes, integrin beta 1 provides the neural receptor for migration on LAM fibers. These results suggest that multiple receptor systems support CNS neural migration.

The contributions of cell-cell interactions to the establishment of specific patterns of innervation within target brain regions are not known. To provide an experimental analysis of the regulation of afferent axonal growth, we have developed an in vitro assay system, based on the developing mouse cerebellum, in which afferent axons from a brainstem source of mossy fiber afferents, the basilar pontine nuclei, were cocultured with astroglia or granule neurons purified from the cerebellum. In the absence of cells from the cerebellum, pontine explants produced axons that fasciculated and extended rapidly on a culture surface treated with poly-lysine or laminin. When pontine neurites grew onto cerebellar astroglial cells, outgrowth was more abundant than on substrates alone, suggesting that glial cells provide a positive signal for axon extension. Time-lapse video microscopy indicated that the rate of neurite extension increased from less than 50 microns/hr to more than 100 microns/hr when axonal growth cones moved from the culture substratum onto an astroglial-cell surface. Acceleration of neurite extension was also observed as pontine neurites grew onto other pontine neurites. By contrast, when pontine neurites grew on granule neurons, the appropriate targets of mossy fibers, the length of pontine neurites was greatly reduced. As growing axons terminated on granule neurons, the target cells appeared to provide a “stop-growing signal” for axon extension. The length of pontine neurites decreased with increasing granule neuron density. Two lines of evidence suggested that the stop signal was contact mediated. First, video microscopy showed that pontine growth cones stopped extending after contacting a granule neuron. Second, the length of afferent axons was not reduced when pontine neurites grew at a distance from granule neurons. Competition experiments where both astroglia and granule neurons were plated together suggested that the growth arrest signal provided by granule neurons could override the growth-promoting signal provided by astroglial cells. These results suggest that specific cell- cell interactions regulate the growth of pontine afferent axons within their cerebellar target, with axoaxonal and axoglial interactions promoting axon extension and axon-target cell interactions interrupting axon extension.

To analyze how astroglial cells attain the complex shapes that support neuronal migration and positioning in vitro (Hatten et al., 1984; Hatten 1985), early postnatal mouse cerebellar cells were plated in microcultures, and glial process outgrowth was monitored by high-resolution time-lapse video microscopy combined with immunocytochemical localization of antisera to glial filament protein (GFP), and by electron microscopy. The 2 principal astroglial forms seen in these cultures, stellate and Bergmann-like (Hatten et al., 1984), begin to develop their distinctive shapes by the outgrowth of processes in the first 8 hr after the cells are plated. Glial process extension is most vigorous in this period, resulting predominantly in stellate forms. A second population of glial cells, having fewer, longer processes reminiscent of Bergmann glia in vivo, first appears about 5 hr after plating. During the next 16-24 hr, while the stellate cells only slightly increase their process length, the bipolar cells double their length. The most striking feature of the elongating glial process is its highly motile tip, which rapidly extends microspikes and lamellopodia. Unlike the neuronal growth cone, which is the expanded terminal of a thin neurite shaft, the glial growing tip forms the end of a wide, paddle-like process that is filled with motile mitochondria and masses of glial filaments, and is bordered by an undulating lamella fringed by microspikes. Soon after the emergence of glial processes, cell-cell interactions between the growing glial process tip and granule neurons occur. Within minutes of an initial encounter between the glial process and the neuron, contact relationships that are stable during the observation period form between the cells. Subsequently, many neurons extend a small neurite onto the glial process, and astroglial process extension continues by the movement of the glial growing tip out beyond the neuron. Thus, cerebellar astroglia in vitro develop complex shapes in the same fashion as do neurons: the outgrowth of processes tipped by a motile ending. The growing tips of astroglial processes interact with neurons, resulting in the stable association of neurons and glia.

Models of astrocyte differentiation stress a lineage program that involves a progressive loss of astroglial support of neuronal differentiation. These models predict that astroglial promotion of neurite extension declines with the “age” of the astrocyte. An alternative view is that astroglial support of neurite growth is regulated by epigenetic factors that induce the cells either to differentiate and support neuronal functions or to undergo cell proliferation and fail to support neurons. To compare the contribution of astroglial cell “age” to astroglial support of neurite extension, mouse cerebellar astroglia were maintained in vitro for 3–90 d, and assayed for their ability to support neurite formation. When cultured in isolation, astroglial support of neurite extension declined with time in vitro, as assayed by quantifying outgrowth from explants of pontine nuclei, falling from a robust level just after the astroglia were harvested to negligible levels 21–90 d later. Since previous studies have shown that neurons can change the state of astroglial cells (Hatten, 1985), we tested the neurite promoting activity of astroglia that were cultured for 21–90 d in vitro and subsequently induced to differentiate by the addition of neurons. When granule neurons were added to aged astroglia and pontine explants plated 2 d later, neurite growth from the explants was exuberant, regardless of the time astroglia spent in vitro prior to the addition of neurons. The state of astroglia that were growth promoting or growth inhibiting was examined by bromodeoxyuridine staining and with antisera to glial filament protein. Aged astroglia cultured alone and thus inhibitory to axon growth, proliferated at high rates and had polygonal shapes. In contrast, aged astroglia to which neurons had been added, proliferated at low rates and developed process-bearing stellate shapes. To test further whether proliferation levels related to the growth-supporting properties of astroglia, astroglia were plated alone in medium without serum, or with the addition of transforming growth factor-beta 1, each treatment known to arrest proliferation. In both cases, promotion of neurite growth was restored in aged astroglia, but the morphology of astroglia did not correlate with the ability to support neurite growth. Finally, the growth-inhibiting properties of aged astroglia do not appear to be mediated by diffusible factors, and require close apposition with living astroglial cells. We conclude that astroglial support of neurite extension depends on the state of differentiation of astroglial cells, and that these properties can be modified by coculture with neurons or conditions that arrest of astroglial proliferation, irrespective of astroglial “age”.