The glia limitans, or glial limiting membrane, is the outermost layer of nervous tissue of the brain and spinal cord, lying directly under the pia mater. It is composed of a dense multilayered meshwork of astrocyte processes, covered by an outer basal lamina that makes intimate contact with cells of the pia mater. The astrocyte processes that cover the surface (end-feet) are firmly attached to the basal lamina (Peters et al., 1991; Nolte, 2002; Rao and Jacobson, 2005; Squire et al., 2008). This arrangement for the structure of the outermost surface of the brain has been described in many different textbooks. In this study, the glia limitans on different area surfaces of the murine brain and spinal cord, and monkey cerebral cortex surface were studied with immunohistochemistry using the astrocyte markers glial fibrillary acidic protein (GFAP) and glutamine synthase (GS) and it was found that most surface areas of murine brain and spinal cord were covered by a layer of astrocyte somas. In contrast, the glia limitans was composed of a layer of foot-plates from the marginal astrocytes in the cerebral cortex of Cynomolgus monkeys, in keeping with the classical concept.
Using the double immunostaining method, the glia limitans on the surfaces of mouse and monkey brain and spinal cord were studied systematically. The results showed that the superficial glia limitans of mouse brain and spinal cord comprise a layer of surface astrocytes, while the glia limitans of monkey comprise a layer of foot-plates from marginal regions as described in histology and neuroscience textbooks. These surface astrocytes first appear at embryonic day (E)16 in spinal cord and at E17 on the ventral surface of the brain. At postnatal day (P)1, a layer of astrocytes covered the outermost regions of the spinal cord. At P10, the layer of astrocytes also covered the brain surface. The highest proliferation rates of surface astrocytes were found at E17 in the spinal cord and at E18 in the forebrain. Anat Rec, 296:995–1007, 2013. © 2013 Wiley Periodicals, Inc.
MATERIALS AND METHODS
Animals and Tissue Preparation
All experimental procedures were approved by the Institutional Animal Care and Use Committee at the Second Military Medical University. Kuming mice at prenatal stages of development (embryonic day (E)14, 15, 16, 17, 18), postnatal stages (postnatal day (P)1, P5, P10, P20, P30), and 6 adult Kunming mice were used. The prenatal animals were killed by immerging in ice-water. Postnatal and adult animals were killed by an intraperitoneal injection with chloral hydrate. After the animals were killed, they were perfused through the aorta with 0.9% NaCl solution and 4% paraformaldehyde in 0.1 mol/L phosphate buffer, pH 7.4. The brains and spinal cords were dissected out and then refixed in 4% paraformaldehyde in 0.1 mol/L phosphate buffer pH 7.4 for 4–6 hr, then transferred to 25% sucrose in phosphate-buffered saline (PBS) and kept in the solution until they sank to the bottom. Three blocks of frontal cortex from three Cynomolgus monkeys (Macaca fascicularis) obtained from the Animal centers of the Second Military Medical University were immersed in a fixative of 4% paraformaldehyde in 0.1 mol/L phosphate buffer, pH 7.4 for 24 hr and then transferred to 25% sucrose in PBS and kept in the solution until they sank to the bottom. Thereafter, the tissue blocks were rapidly frozen and coronal and horizontal sections (20 μm in thickness) were cut with a Leica cryostat and floated in PBS.
The following protocol was used for double-labeling immunostaining. The preparations were washed 3 × 5 min in PBS, and then preincubated in antiserum solution 1 for 30 min, followed by incubation with different combinations of GFAP (mouse anti-rat; Boster, Wuhan, China) diluted 1:400, laminin antibody (rabbit anti-rat; Boster, Wuhan, China) diluted 1:600, GS (goat anti-rat; Santa Cruz, USA) diluted 1:100, Ki-67 antibody (rabbit anti-rat; Boster, Wuhan, China) diluted 1:200 in antiserum solution, at room temperature. Subsequently, the preparations were incubated with Cy3-conjugated donkey anti-rabbit IgG (CN:711-165-152, Immunoreasearch Lab, USA) diluted 1:400 for laminin and Ki-67 antibodies, FITC conjugated donkey anti-mouse IgG (CN:715-095-151, Immunoresearch Lab, USA) diluted 1:200 for GFAP antibody, and FITC-conjugated donkey anti-goat IgG (CN:713-095-003, Immunoresearch Lab, USA) diluted 1:200 for GS antibody in antiserum solution 2 for 2 hr at room temperature. The secondary antibodies used here are specialized for multiple immunostaining from Jackson Immunoresearch Lab. All the incubations and reactions were separated by 3 × 10 min washes in PBS.
A negative control of omitting the primary antibody was carried out. No staining was also observed in those preparations (data not shown).
Quantitative analysis for the percentages of GFAP-immunoreactive (ir) and Ki-67-ir astrocytes of the total number of astrocytes on the superficial surfaces of the forebrain and spinal cord during different developmental stages was performed as follows: 30 sections for each mouse were chosen. The number of double labeled astrocytes with both GFAP and Ki-67 antibodies and the number of single labeled astrocytes with GFAP antibody were countered under a 20× object lens. The total from each of the 30 sections from each mouse was calculated then the percentages of GFAP-ir and Ki-67-ir astrocytes were calculated from the total number of GFAP-ir astrocytes. The mean percentages was calculated and the data expressed as the mean ± SE of the mean (n = number of mice).
Statistical analysis was performed using one-way ANOVA followed by unpaired t-test, and P < 0.05 was considered significant.
Images were taken with a Nikon digital camera DXM1200 (Nikon, Japan) attached to a Nikon Eclipse E600 microscope (Nikon). Images were imported into a graphics package (Adobe Photoshop).
An astrocyte layer with GFAP-ir was detected in the coronal section of the forebrain dorsal glia limitans of mouse (Fig. 1A,D). Almost all the surface of this part of the brain was covered by astrocyte-like cells. The cell bodies of these astrocyte-like cells contacted tightly with the basal lamina, as demonstrated by laminin antibody (Fig. 1D). In the horizontal sections of the forebrain dorsal glia limitans, an excellent picture was obtained by the immunostaining with GS and GFAP antibodies (Fig. 2A,B,D,E). In these sections, two totally different shape astrocytes were detected: a protoplasmic astrocyte, located in the parenchyma (Fig. 2F), the other a surface astrocyte, covering the surface of the brain (Fig. 2E). In the horizontal section, the cell body of the surface astrocyte was oval or polygonal with no processes (Fig. 2A,B,D,E and Supporting Information Fig. 1). The cell body was 20–30 μm in length. The nucleus was circular with a diameter of 10–15 μm. In the section immunostained by the GFAP antibody, the glial intermediate filaments were observed to tangle in the plasma of the surface astrocyte (Figs. 2B,C and 3A,D). The arrangement of glial intermediate filaments of the surface astrocytes was totally different from that of the astrocytes in the parenchyma (Fig. 2C,F). For the surface astrocytes, the glial intermediate filaments were arranged in an orderly fashion in the oval or polygonal bodies. While in the parenchymal astrocytes the glial intermediate filaments were arranged mainly to one side of the body and the main processes (Figs. 2F and 3A). Similar results were observed in the lateral brain surfaces. The ventral surface of the forebrain was also covered by the surface astrocytes (Fig. 4A). The surface astrocyte layer was thicker in the irregular part than that in the smooth part (Fig. 4B). There were no obvious surface astrocytes with GFAP and GS-ir detected in the olfactory bulbs (Fig. 4D).
In the midbrain, the highest density of surface astrocytes was detected in the superior and inferior colliculus (Fig. 4E,F). In other parts of the midbrain and brain stem, similar immunostaining of surface astrocytes was obtained. The density of surface astrocyte bodies was lower than that in the surface of superior and inferior colliculus.
In the spinal cord, a moderate density of surface astrocytes was detected. In the ventral and lateral part of the spinal cord, the surface astrocyte bodies were in contact with the basal lamina, as shown by the laminin antibody. The morphological character of these surface astrocytes was a small triangular cell body located on the surface and a very long process protruding to the central part of the spinal cord (Fig. 5C,D). The surface astrocyte density was higher in the dorsal part than in the ventral and lateral part of the spinal cord, but their processes were much shorter (Fig. 5D).
On the cerebellar surface, classic glia limitans were detected with GFAP immunostaining. The long process terminals (foot-plates) of Bergmann glial cells were in tight contact with the basal lamina, as shown by laminin antibody immunostaining (Fig. 5A,B). The foot-plates from the Bergmann glial cells were arranged in an orderly fashion and almost filled up the outermost surface of the cerebellar cortex. The foot-plate and terminal process formed a “T” shape. The foot-plate was about 20–30 μm in length (Fig. 5B). Very few surface astrocytes were detected, although in some of the cerebellar cortex sections, one or two surface astrocytes were observed.
These results described above indicate that the glia limitans are mainly composed of a layer of surface astrocytes in the mouse brain and spinal cord, while in the cerebellar cortex the glia limitans is composed of a layer of foot-plates from Bergmann glial processes.
The glia limitans of forebrain from rat, rabbit, guinea pig, and chicken were further studied. The glia limitans of these four animals are also mainly composed of a layer of surface astrocytes (Supporting Information Fig. 2).
Examination of different developmental stages of mouse brain revealed that scattered surface astrocytes with GFAP-ir were first detected on the ventral surface of the forebrain at E17 (Fig. 6A). From E17 onward, the density of the surface astrocytes increased. At E18 and P1, surface astrocytes were detected on the lateral and dorsal surfaces of the forebrain, respectively (Fig. 6B–D). From P10 onward, all surfaces of the forebrain were covered by a layer of surface astrocytes (Fig. 6E,F).
Examination of different developmental stages of mouse spinal cord revealed that scattered surface astrocytes were first detected in the spinal cord at E16 (Fig. 7A). From E16–E18, the density of surface astrocytes in the spinal cord increased and at E18 a layer of continuous GFAP-ir astrocytes was detected on the spinal cord surfaces (Fig. 7B–E). During embryonic development, GFAP-ir astrocytes and their processes were detected only in the marginal layer or white matter. The cell bodies of GFAP-ir astrocytes were detected mainly on the outermost surfaces of the spinal cord. The processes of GFAP-ir astrocytes generally terminated on the mantle layer or gray matter. The branches and complexity of GFAP-ir astrocyte processes increased gradually from E16 to adult (Fig. 5). From P20 onward, the density and morphology of the surface astrocytes were similar to that of the adult spinal cord (Figs. 5C,D and 7F).
The proliferation rate of the surface astrocytes was studied with Ki-67 and GFAP double immunostaining. Surface astrocytes labelled with both Ki-67 and GFAP–ir were detected when the surface astrocytes appeared in the surface layers of brain and spinal cord. The highest proliferation rate of surface astrocytes was at E17 in the spinal cord and at E18 in the forebrain. After these stages, the proliferation rate decreased gradually. At P20, the proliferation rate was similar with that of the adult in both the forebrain and spinal cord (Figs. 8A–F and 9A,B).
On the outermost surface of the monkey cerebral cortex, very few astrocytes were detected. In coronal sections of the cerebral cortex marginal areas, many GFAP-ir astrocytes were detected. These marginal astrocytes were found to have many long processes stretching around the cell body, many of which were directed to and were in contact with the basal lamina, as demonstrated by the laminin antibody. Those marginal astrocyte processes stretching to the surface formed the foot-plate-like structures, which contacted tightly with basal lamina (Fig. 10C). In the horizontal section of the monkey cerebral cortex, there were no GFAP-ir astrocytes with a typical surface astrocyte characterization, as seen in the mouse forebrain (Figs. 1D–F and 2A–E). These results indicate that the glia limitans of the monkey cerebral cortex is mainly composed of foot-plates.
The superficial glia limitans in the mouse central nervous system (CNS) were examined with double labeling immunofluorescence and were found in most areas of the brain and spinal cord, the outer layers of both of which were covered by a layer of astrocytes. These surface astrocytes appeared on the forebrain surfaces at E17 and on the spinal cord at E16. At P1, a layer of astrocytes covered the outermost surface of the mouse spinal cord. At P10, the astrocytes also covered the forebrain surface. However, there were two regions that were not covered by a layer of astrocytes. One is the outermost layer of the olfactory bulb, where no obvious astrocytes or foot-plates of the marginal astrocytes were detected by GFAP and GS antibodies and the other region is the outermost layer of the cerebellar cortex, where a layer of foot-plates from Bergmann glia cells comprised the glial limitans.
In current histology and neuroscience textbooks, the definition of the glia limitans is the outermost layer of the brain and spinal cord, lying directly under the pia mater, being composed of a dense meshwork of astrocyte processes, which are firmly attached to the basal lamina (Peters et al., 1991; Nolte, 2002; Rao and Jacobson, 2005; Squire et al., 2008). This definition needs to be reconsidered, because this study has shown that the glia limitans in the forebrain and spinal cord of the mouse is composed of a layer of astrocytes, although the glia limitans in the cerebellar cortex is composed of foot-plates of Bergmann glial cells. In the monkey cerebral cortex, we have also shown that the glia limitans is composed of a layer of foot-plates of marginal astrocytes.
Astrocytes are the most numerous and diverse glial cells in the CNS. The astrocytes are comprised of protoplasmic astrocytes and fibrous astrocytes of the gray and white matter, respectively. Protoplasmic astrocytes are endowed with many fine processes, some of which stretch to the pial surface, where they form “subpial” end feet. The processes of fibrous astrocytes also establish subpial end feet (Verkhratsky and Butt, 2007). The morphology of the surface astrocytes seen in the present study is totally different from that of the two main astrocytes in the CNS. In the forebrain, the squamous somas of surface astrocytes contact directly with the basal lamina. In the spinal cord, surface astrocytes with pyramidal somas have processes which extend into the boundary of the spinal cord gray.
The glia limitans is a barrier of marginal astrocyte foot processes in primates. Its likely role is to prevent over migration of nerve cells into the meninges (Peters et al., 1991). The glial limitans are divided into two different parts according to their location: the glia limitans perivascularis and the glia limitans superficialis (Engelhardt and Coisne, 2011). In this study, we found that the glia limitans superficialis mainly comprised a layer of squamous somas of surface astrocytes in mouse rather than the foot-plates of marginal astrocytes in primates (Peters et al., 1991; Nolte, 2002; Rao and Jacobson, 2005; Squire et al., 2008). We believe that this surface astrocyte layer could also act as a physical barrier.
This study showed that the peak in the proliferation rate of the surface astrocytes was at E17 and E18 in the spinal cord and forebrain, respectively. Up to 60% of surface astrocytes in the spinal cord at E17 and up to 35% in the forebrain at E18 were double-labeled with Ki-67 and GFAP antibodies. Antigen Ki-67 is a nuclear protein and cellular marker for proliferation (Peter et al., 1986; Brown and Gatter, 1990; Scholzen and Gerdes, 2000). This data implies that about 35% and 60% of the surface astrocytes in the forebrain and spinal cord are in the active phases of the cell cycle. Our study has also shown that GFAP-ir astrocytes first appeared on the surface of the brain and spinal cord, and then on their parenchyma. Thus, it is possible that at least some of the GFAP-ir astrocytes could proliferate on the outer surfaces of brain and spinal cord and then migrate into the parenchyma. This hypothesis needs to be further investigated.
GFAP-ir astrocytes appear first on the ventral surface of the forebrain at E17, then on the lateral surface at E18, and finally on the dorsal surface at P1. This data implies that GFAP-ir surface astrocytes might originate on the ventral surface and then migrate from the ventral to the lateral surface and finally to the dorsal surface. This hypothesis is supported by previous reports, which show that the basal lamina is a transplant-derived astrocyte migration route (Goldberg and Bernstein, 1988, 1989). As the basal lamina presents on the outermost layer, surface astrocytes from the ventral surface could migrate to the other region surfaces along the superficial basal lamina. This hypothesis also requires further experimental investigation.
The glia limitans serves an important structural and physiological function in human beings. A comparable structure of the glia limitans has been found in other animals. For example, insects have a sheath of perineurial glial cells that envelops the nervous system (Oland and Tolbert, 1987). In cephalopod molluscs, glial cells form a seamless sheath completely around the blood space (Brightman, 1991). Monkeys and other primates have been found to have a glia limitans similar to that of humans (Pannese, 1994; Nolte, 2002) and this was further confirmed in this study. However, in mouse, the superficial glia limitans is comprised of a continuous sheath of GFAP-ir astrocytes, as for insects (Oland and Tolbert, 1987).
In summary, this study has shown that the superficial glia limitans in the mouse CNS is a layer of astrocytes. These surface astrocytes appear first on the ventral surface at E17 and on the spinal cord at E16. At P1, a layer of astrocytes covered the outermost regions of the spinal cord. At P10, the astrocytes also covered the brain surface. The proliferation rate peaks of the surface astrocytes were at E17 and E18 in the spinal cord and forebrain, respectively.
The authors thank Dr. Gillian E. Knight for her excellent editorial assistance.