Author contributions: N.M.: conception and design, collection and assembly of data, data analysis and interpretation, financial support, and manuscript writing; G.G.: conception and design, collection of data, data analysis and interpretation, and financial support; M.R.: collection of data and data analysis and interpretation; O.T.-C.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing; R.E.R.: conception and design, collection of data, data analysis and interpretation, financial support, and manuscript writing.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS July 20, 2012.
During spinal cord development, progenitors in the neural tube are arranged within spatial domains that generate specific cell types. The ependyma of the postnatal spinal cord seems to retain cells with properties of the primitive neural stem cells, some of which are able to react to injury with active proliferation. However, the functional complexity and organization of this stem cell niche in mammals remains poorly understood. Here, we combined immunohistochemistry for cell-specific markers with patch-clamp recordings to test the hypothesis that the ependyma of the neonatal rat spinal cord contains progenitor-like cells functionally segregated within specific domains. Cells on the lateral aspects of the ependyma combined morphological and molecular traits of ependymocytes and radial glia (RG) expressing S100β and vimentin, displayed passive membrane properties and were electrically coupled via Cx43. Cells contacting the ventral and dorsal poles expressed the neural stem cell markers nestin and/or vimentin, had the typical morphology of RG, and appeared uncoupled displaying various combinations of K+ and Ca2+ voltage-gated currents. Although progenitor-like cells were mitotically active around the entire ependyma, the proliferative capacity seemed higher on lateral domains. Our findings represent the first evidence that the ependyma of the rat harbors progenitor-like cells with heterogeneous electrophysiological phenotypes organized in spatial domains. The manipulation of specific functional properties in the heterogeneous population of progenitor-like cells contacting the ependyma may in future help to regulate their behavior and lineage potential, providing the cell types required for the endogenous repair of the injured spinal cord. Stem Cells2012;30:2020–2031
Neurogenesis in discrete niches of the adult mammalian brain represents a remarkable form of plasticity [1, 2]. Although the mature spinal cord in mammals is regarded as a non-neurogenic structure, the region surrounding the central canal (CC) shares key features with stem cell niches of the brain . For example, cells lining the CC express markers of neural stem cells [4, 5] and keep the ability to proliferate [6, 7]. In addition, we recently reported that some CC-contacting cells in neonatal rats display molecular and functional features of immature neurons like those of adult neurogenic niches . In response to spinal cord injury, ependymal cells proliferate and migrate to the lesion site where they differentiate into scar-forming astrocytes and myelinating oligodendrocytes [4, 9]. However, some studies in the normal and the diseased mammalian spinal cord suggest that ependymal cells may support neurogenesis [10–12].
Neural stem cell niches in the adult brain are formed by progenitors with heterogeneous properties that interact within a complex three-dimensional organization . During spinal cord development, progenitors in the neural tube are arranged within domains that generate specific cell types [14, 15]. Some of these progenitors are retained postnatally in low vertebrates. In turtles, for example, neurogenic progenitors are functionally clustered on the lateral aspects of the CC . Little is known about the functional complexity and organization of the spinal cord ependymal region as a stem cell niche in mammals.
We speculated that as in low vertebrates, the ependyma of the mammalian spinal cord during early postnatal life—when developmental refinement of spinal circuits still occurs (e.g., myelination)—may contain progenitor-like cells functionally grouped within specific domains around the CC. To test this idea, we combined immunohistochemistry for molecular markers of neural stem cells and patch-clamp recordings of CC-contacting cells in neonatal rats. Here we show that the cells lining the lateral aspects and those on the dorsal and ventral poles of the CC constitute a heterogenous population of progenitor-like cells with characteristic molecular and functional properties. Most of the cells on the lateral aspects of the ependyma had the morphology of typical ependymocytes, exhibited passive membrane properties, and were electrically coupled via connexin 43 (Cx43). In contrast, the cells contacting the poles of the CC had the morphology of radial glia (RG) with a single cilium, were uncoupled, and displayed various combinations of K+ and Ca2+ voltage-gated currents. Although, progenitor-like cells were mitotically active around the ependyma, the proliferative capacity seemed higher on cells contacting the lateral aspects of the CC. Our findings show an unforeseen functional and molecular diversity of progenitors within the ependyma of the rat spinal cord segregated within specific spatial domains. It is tempting to speculate that like in the embryo, these domains represent a reservoir of progenitors with different functional roles and lineage potential.
MATERIALS AND METHODS
Neonatal rats (Sprague Dawley, P0–P5) were used. For some electrophysiological and immunohistochemical experiments, P15–P21 and P40 rats were also used. All experimental procedures were performed in accordance with the ethical guidelines established by our local Committee for Animal Care.
Animals were anesthetized (50 mg/kg, i.p.; Pentobarbital) and fixed by intracardiac perfusion with 4% paraformaldehyde in 0.1 M phosphate buffer (PB, pH 7.4). To identify the molecular phenotype of cells around the CC, the following primary antibodies were used (supporting information Table 1): anti-S100β, anti-vimentin, anti-nestin, anti-3CB2, anti-brain lipid binding protein (BLBP), anti-glial fibrillary acidic protein (GFAP), anti-Cx43, anti-platelet derived growth factor receptor α (PDGFRα), anti-NG2, anti-pericentrin, anti-proliferating cell nuclear antigen (PCNA), and anti-phosphohistone H3 (pH3). Tissues were sectioned with a vibrating microtome (60–80 μm thick) and placed in PB with 0.5% bovine serum albumin (BSA) for 30 minutes and then incubated with primary antibodies in PB with 0.3% Triton X-100 (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com). Tissues were then incubated in secondary antibodies conjugated with different fluorophores or HRP. The HRP was revealed with 3,3-diaminobenzidine (DAB, Sigma-Aldrich). Nuclei were stained with Syto 64 (Invitrogen, Carlsbad, CA, http://www.invitrogen.com). For double-labeling using anti-3CB2 and anti-Nestin antibodies raised in mice, we applied a sequential immunofluorescence procedure using anti-mouse IgG1 Alexa 488 and anti-mouse IgM Cy3 (Jackson Immunoresearch Laboratories, West Grove, PA, http://www.jacksonimmuno.com). The number of PCNA and pH3 positive nuclei was counted in 60-μm-thick sections chosen randomly (30 sections from three animals). Values are expressed as the mean ± SEM. Statistical significance was set at p < .05 and evaluated using the independent t test. Control experiments were performed suppressing the primary antibodies. The confocal images were acquired with Fluoview 5 (Olympus VF300).
Slice Preparation and Electrophysiology
Rats anesthetized with isoflurane (Forane, Abbot Laboratories, Berkshire, UK, http://www.abbott.com) were decapitated and the cervical enlargement was dissected out. Transverse 300-μm-thick slices were cut, placed in a chamber, and superfused (1 ml min−1) with Ringer's solution (in mM): NaCl, 124; KCl, 2.4; NaHCO3, 26; CaCl2, 2.4; MgSO4·6H2O, 1.3; HEPES, 1.25; KH2PO4, 1.2; and glucose, 10; saturated with 5% CO2 and 95% O2 to keep pH 7.4. In low Ca2+ Ringer's solution, CaCl2 was lowered to 0.2 mM whereas MgSO4 was increased to 4 mM. All experiments were performed at room temperature (22°C–24°C). Cells were visualized with differential interference contrast optics (Leica DM LFS, Leica Microsystems GmbH Wetzlar, Germany, http://www.leica.com) and patch-clamp whole-cell recordings obtained with electrodes filled with (in mM): K-gluconate, 122; Na2-ATP, 5; MgCl2, 2.5; CaCl2, 0.003; ethylene glycol-bis(β-aminoethylether)-N,N,N′,N′-tetraacetic acid, 1; Mg-gluconate, 5.6; K-HEPES, 5; H-HEPES, 5; and biocytin, 10; pH 7.3 (5–10 MΩ). In some cases, Alexa 488 or 594 hydrazide (250–500 μM, Invitrogen) was added to the pipette solution. Current and voltage-clamp recordings were performed with a Multiclamp 700B and pClamp10 (Molecular Devices, Union City, CA, http://www.moleculardevices.com). Seal resistances were between 4 and 18 GΩ. The series resistance and whole-cell capacitance were not compensated. In voltage clamp mode, cells were held at −70 mV, and the resting membrane potential was estimated from the current–voltage relationship (at I = 0). To subtract leak currents, we used a P4 protocol by Clampex10 (Molecular Devices). Liquid junction potentials were determined and corrected off-line . Values are expressed as the mean ± SEM. Statistical significance was set at p < .05 and evaluated using the independent t test or the Wilcoxon matched-paired test. The activation and inactivation curves for K+ currents were determined as described elsewhere .
Morphological Identification of the Recorded Cells
During whole-cell patch-clamp recordings, cells were filled with a fluorophore and/or biocytin. In most cases, cells were first imaged in living slices with an FG7 frame grabber (Scion Instruments) using ImageJ (NIH). Then, slices were fixed by immersion in 4% paraformaldehyde in 0.1 M PB for 12–24 hours. Following PB rinsing, the slices were blocked with 0.5% BSA in PB (1 hour) and then incubated in PB containing 0.3% Triton X-100 with the streptavidin-fluorophore complex.
Transmission Electron Microscopy (TEM)
Anesthetized animals were fixed by intracardiac perfusion with 4% paraformaldehyde and 1% glutaraldehyde in 0.1 M PB, pH 7.4. Slices obtained with a vibrating microtome (100–200 μm thick) were washed in PB and postfixed in 1% OsO4 in PB 0.1 M, dehydrated, and epoxy-resin embedded.
Series of semithin sections were cut and stained with boraxic methylene blue. Ultrathin sections were obtained from trimmed blocks and mounted on one hole grids. To obtain frontal views of the apical processes of cells contacting the CC poles, we made alternating series of sections of different thickness. After reaching the level of the ependymal cells nuclei, several series of semithin and ultrathin sections were cut, mounted on one-hole grids, and examined with a Jeol X 100 transmission electron microscope.
Sections were processed for revealing nestin using a secondary antibody conjugated with horseradish peroxidase (HRP) and DAB as chromogen. The selected sections were postfixed for 1 hour (1% OsO4 in PB 0.1 M), washed, dehydrated, and epoxy-embedded in flat molds. Series of sections were processed as already described.
Progenitor-Like Cells in the CC: Molecular Clues
Some ependymal cells in adult mice express markers of neural precursors . To identify potential progenitor-like cells and their spatial distribution within the ependymal region of the neonatal rat spinal cord, we analyzed the expression of molecules expressed in neural stem cells . Most cells surrounding the CC expressed the ependymal cell marker S100β (Fig. 1A) but a narrow strip of tissue on the dorsal pole was unstained (Fig. 1A, arrow). Both the dorsal and ventral poles of the CC were contacted by closely packed nestin+ processes (Fig. 1B-1) that projected within the midline to reach the pia (Fig. 1B-2, 3, arrowheads). Unlike the ventral pole, nestin+ fibers in the dorsal pole of the CC did not coexpress S100β (Fig. 1C). As revealed by Syto 64, most cell bodies corresponding to nestin+ fibers located far from the CC (Fig. 1D, 1E, arrowheads) with only a few lying close to it (Fig. 1F, arrowhead). We also found a faint expression of nestin in few short radial processes arising from the lateral aspects of the ependyma (data not shown). Vimentin and 3CB2 also expressed in cells that projected their distal processes to the midline. However, unlike nestin+ cells, vimentin+/3CB2+ cells were also abundant on the lateral aspects of the ependyma, giving rise to short processes projecting away from the CC (Fig. 1G, 1H, arrowheads). Although a subset of progenitors in the embryo and the adult brain characteristically expresses GFAP and/or BLBP [19, 20], we did not find GFAP or BLBP expression in ependymal cells but outside the ependymal region (Fig. 1I, 1J). In the spinal cord of neonatal rats, BLBP+ cells had a morphology similar to migrating cells (Fig. 1I, arrowheads), whereas cells expressing GFAP showed the typical morphology of astrocytes (Fig. 1J, arrowheads). Most 3CB2+ cells on the lateral aspects of the CC also expressed S100β (supporting information Fig. 1, arrowheads).
Electrophysiological Properties of Cells Lining the CC
Progenitors in the embryo  and the adult brain [22–24] have characteristic electrophysiological signatures. To check whether progenitor-like cells in different regions of the ependyma have specialized functional properties, we made patch-clamp recordings of cells on the lateral aspects of the ependyma (Fig. 2A). We found cells (n = 36) with linear voltage–current relationships (Fig. 2B), relatively low input resistances (123.63 ± 24.44 MΩ, n = 34; Fig. 2B, 2C), and hyperpolarized resting membrane potentials (−84.35 ± 2.13 mV, n = 36; Fig. 2C). Morphological analysis revealed that recorded cells belonged to clusters of dye-coupled ependymocytes lining substantial portions of the CC (Fig. 2D, 2F). The low input resistance and dye coupling suggested coupling via gap junctions. In line with this interpretation, the gap junction decoupler carbenoxolone (100 μM) significantly increased their input resistance (from 103.8 ± 26.4 to 388.5 ± 189.6 MΩ, p < .05, Wilcoxon's matched-pairs test, Fig. 2E). The size of the clusters ranged from large groups of cells covering the entire lateral aspect of the CC (Fig. 2F-1) to rather small-cell conglomerates close to the ventral or dorsal poles (Fig. 2F-2–4). Interestingly, some of these clustered cells had processes entering the dorsal or ventral midline (Fig. 2F-2, 3, arrowheads) and projecting toward the pia (Fig. 2F-4, arrow). Occasionally, we recorded uncoupled cells on the lateral aspects of the CC with a process projecting toward the parenchyma (n = 4, data not shown).
Because the electrophysiological data suggested the existence of gap junctions between clustered cells, we studied the molecular basis of electrical coupling. Although several connexin subtypes are expressed in the developing brain, Cx43 is the most abundant in the subventricular zone (SVZ) of the developing forebrain . Immunohistochemistry for Cx43 revealed a high density of punctae on the lateral aspects of the CC (Fig. 2G, arrow). Notice however, the lack of Cx43 punctae on the dorsal and ventral poles of the ependyma. We conclude that on the lateral aspects of the CC, the non-neuronal cells have morphological and electrophysiological characteristics of ependymocytes and tanycytes  functionally associated by means of gap junctions.
Electrophysiological Signature of Midline Progenitor-Like Cells
Because our data indicate that the cells contacting the poles of the CC have molecular phenotypes different from those on the lateral aspects, we speculated that they may also have different electrophysiological properties. Patch-clamp recordings of CC-contacting cells projecting to the raphae revealed that these cells had hyperpolarized resting membrane potentials (−81.95 ± 2.31 mV, n = 47; supporting information Fig. 2A) and high input resistances compared with cells in the lateral aspects of the CC (361.22 ± 56.29, n = 48, p < .05, independent t test; supporting information Fig. 2A). On the dorsal and ventral poles of the CC, cells had the typical morphology of RG (Figs. 3A, 3F, and 4A) and appeared uncoupled (n = 71). Some cells had a relatively thick apical process (supporting information Fig. 2B-1, arrow) with many finger-like protrusions (supporting information Fig. 2B-2, arrowheads) and a thinner distal fiber projecting to the pia (supporting information Fig. 2B-1, arrowhead). However, other cells had smooth apical and distal processes (supporting information Fig. 2C, arrows). RG contacting the dorsal or ventral aspects of the CC had their cell bodies located at different distances from the CC lumen (supporting information Fig. 2D), resembling the morphology of RG during interkinetic nuclear migration 
RG lying within the midline had a complex repertoire of active properties with different types of outward and inward currents. In some RG (16 of 65, supporting information Table 2), depolarizing voltage steps produced an outward current (Fig. 3A, 3B-1) with minimal inactivation in response to sustained depolarization (Fig. 3B-1, 3). This current had an activation threshold close to −40 mV with a Vh = 5.37 ± 1.77 mV (Fig. 3B-2, 3C, 3D) and was sensitive to 10 mM tetraethylammonium (TEA) (Fig. 3E; three out of three cells) suggesting the involvement of delayed rectifier K+ currents (IKD).
In other RG (25 of 65, supporting information Table 2), depolarizing voltage steps (from a holding potential of −90 mV) evoked outward currents that had both noninactivating and inactivating components (Fig. 3F–33H). To separate these components, we applied the same stimulation protocol but from a holding potential of −30 mV (Fig. 3G-2). Under these conditions, we observed an outward current with a slower onset and no inactivation, suggesting the presence of IKD channels. By subtracting the delayed noninactivating current (Fig. 3G-2) from the total current (Fig. 3G-1), we were able to separate an outward current with a fast onset and a prominent time-dependent inactivation (Fig. 3G-1, 2), suggesting an A-type K+ current (IA, ). In line with this interpretation, TEA (10 mM) blocked the noninactivating component of the outward current (Fig. 3H-1, 2; 10 out of 10 cells) but spared the inactivating current which was blocked by the selective A-type K+ channel blocker 4-aminophyridine (4-AP, 2 mM; Fig. 3H-3; 10 of 10 cells). IA activated transiently at membrane potentials of approximately −40 mV with a Vh = −5.79 ± 1.2 mV (Fig. 3I, 3J).
Besides displaying IKD and IA, another subgroup of RG characterized by generating voltage-gated inward currents (6 of 65; supporting information Table 2). The slow transient inward current required relatively modest depolarizations (threshold approximately −55 mV; Fig. 4B-1) and remained in the presence of both tetrodotoxin (TTX, 1 μM; data not shown) and K+ channel antagonists (Fig. 4B-2). However, the inward current was abolished by 3 mM Mn2+ or in low Ca2+ Ringer's solution (Fig. 4B-3, n = 7) suggesting the involvement of low voltage-activated Ca2+ currents (ICa). In current clamp mode, this inward current generated a slow low threshold spike (LTS, Fig. 4C-1) that disappeared in low Ca2+ Ringer's solution (Fig. 4C-2).
We also found RG that displayed IKD plus ICa without IA (10 of 65, data not show) and others that only had ICa (6 of 65, data not shown). Finally, we recorded few cells (2 of 65) displaying passive membrane responses similar to those of lateral ependymocytes. The electrophysiological phenotypes described above were equally found in the ventral or dorsal poles and in animals within the range of explored ages (P0–P5).
To identify the molecular phenotypes of recorded cells, we combined the labeling of recorded cells with immunohistochemistry for specific markers of progenitors. Some cells recorded on the poles of the CC expressed nestin (supporting information Fig. 3A-1–4), but others (15 of 18) did not react with nestin antibodies (supporting information Fig. 3B). Because the presence of IKD and IA is characteristic of oligodendrocyte progenitor cells , we speculated that nestin− cells may be early oligodendrocyte progenitors. To test this idea, we performed immunocytochemistry for oligodendrocyte progenitor markers such as the PDGFRα and the chondroitin sulfate proteoglycan NG2. Although we found abundant PDGFRα and NG2 reactive cells both in the gray and white matter, CC-contacting cells were always negative (supporting information Fig. 3C, 3D).
Fine Structure of Midline CC-Contacting Cells
Progenitors in the embryo and postnatal neurogenic niches have characteristic fine structure features with pronounced apical–basal polarity . A key component of this polarity is the presence in the apical pole of a centrosome, a structure that regulates the cell cycle and microtubule organization . We found that pericentrin—a key component of the pericentriolar material—was expressed in the apical poles of nestin+ cells contacting the midline but not in their processes (Fig. 5A, arrow). The selective location of pericentrin in the apical poles of midline RG may reflect the presence of cilia. Because progenitors typically possess a single primary cilium , we made a careful analysis of the apical process of cells contacting the poles of the ependyma. Serial section studies revealed that the apical process of these cells bear a single cilium (Fig. 5B–5D). Interestingly, some of the cilia exhibited a 9 + 0 microtubule organization (Fig. 5C, arrowhead in inset) whereas others had a 9 + 2 organization (Fig. 5C, arrow in inset; 5D, arrowhead). Some of the apical process of the midline RG exhibited irregular profiles with lateral projections and lamella (Fig. 5D). These images probably correspond to the abundant finger-like processes arising from the apical process of some midline RG stained intracellularly (supporting information Fig. 2B).
We extended our TEM analysis to the distal processes of dorsal RG (Fig. 5E, inset). Cross-sections at the level of the dorsal raphe showed a compact population of circular/oval profiles with minor structural differences (Fig. 5E). Despite the similarities in fine structure, our patch-clamp recordings suggested that RG in the midline actually represent a heterogeneous population. TEM immunohistochemistry showed that nestin+ processes running in the raphe (Fig. 5F) were intermingled with others lacking electron-dense precipitate (Fig. 5G, 5H). In line with this, some processes in the midline region only expressed 3CB2 (Fig. 5I, arrow in 5J) whereas some were nestin+/3CB2− (Fig. 5I, asterisks in 5J) and others coexpressed both markers (Fig. 5I, arrowhead in 5J).
Proliferative Potential of CC-Contacting Cells
Because progenitor-like cells located in different portions of the ependyma have heterogenous functional and molecular properties, we explored whether they may have different proliferative activity by means of immunohistochemistry for endogenous cell cycle proteins such as the PCNA (expressed in all phases of the cell cycle) and the phosphohistone H3 (pH3, expressed mostly in M phase). We found abundant cells with PCNA+ nuclei on the lateral aspects of the CC (7.1 ± 0.79 cells per section, 30 sections, n = 3 rats, Fig. 6A; supporting information Fig. 4A, 4B) that coexpressed vimentin (Fig. 6A, arrowhead; 6C, arrowheads) or 3CB2 (Fig. 6D, arrowhead). PCNA+ nuclei were also observed in the dorsal (Fig. 6B, arrow) or ventral (Fig. 6C, arrow) raphae at different distances from the CC lumen, but their number was significantly lower than in the lateral aspects of the CC (3.6 ± 0.68 per section, 30 sections, n = 3 rats, p < .05, independent t test; supporting information Fig. 4A, 4B). Many of the PCNA+ nuclei in the midline corresponded to cells that expressed RG markers such as nestin (Fig. 6B-1–3, arrows), vimentin (Fig. 6C-1, 2, arrows), or 3CB2 (Fig. 6D-1–4, arrows). The analysis of pH3 expression showed that mitotic cells appeared all around the CC but as expected from the PCNA immunostaining were more abundant on the lateral aspects of the ependyma (0.96 ± 0.33 vs. 0.131 ± 0.08 nuclei per section, 29 sections, n = 3 rats, p < .05, independent t test; Fig. 6E, arrows and supporting information Fig. 4B). pH3+ cells on the poles of the ependyma were nestin+ (Fig. 6F). In contrast with PCNA, pH3+/nestin+ nuclei in the midline were found only close to the CC lumen, suggesting that as during embryogenesis, cell division occurred close to the surface in contact with the cerebrospinal fluid. We conclude that there is proliferative activity all around the CC, but most cell divisions take place on the lateral domains of the ependyma.
Collectively, our findings support the idea that the ependyma of neonatal rats contains progenitor-like cells with heterogeneous properties, organized in well-defined spatial domains. It may be possible that as the animal develops, these progenitors disappear or change their properties and organization within spatial domains. To explore this possibility, we made patch-clamp recordings in rats between P15 and P21 (n = 6 cells) because during the first 2 postnatal weeks, spinal circuits mature quickly and rats acquire an adult pattern of locomotion . We found that cells contacting the poles of the ependyma in older rats were still uncoupled with the morphological phenotype of RG and complex electrophysiological properties (supporting information Fig. 5A–5D). Similar to neonatal animals, cells on the lateral aspects were extensively coupled and had passive electrical properties (supporting information Fig. 5E). As in mice , the expression of progenitor cell markers was also retained in the mature spinal cord of rats (P40), with nestin predominating on cells contacting the poles (supporting information Fig. 5F, 5G) whereas 3CB2 expressed on both the poles and lateral aspects of the ependyma (supporting information Fig. 5H). Finally, a substantial number of cells on the lateral aspects were positive for PCNA (supporting information Fig. 5I). Thus, progenitors in the ependyma of the mature rat spinal cord maintain the basic properties and compartmentalization observed in neonates.
We show here that the ependyma of the rat spinal cord harbors progenitor-like cells organized in spatially defined domains (Fig. 7). Clusters of coupled cells within lateral domains combined molecular features of ependymocytes and RG . Midline domains contained elements with typical characteristics of neural stem cells  and complex electrophysiological properties that may reflect different functional states or progenitor competence. The different proliferation potential between lateral and midline domains favors the idea they represent a functional compartmentalization of this spinal stem cell niche.
Structural and Molecular Clues of a Spinal Stem Cell Niche
Classic studies conceived the ependyma of the spinal cord as a layer of epithelial cells . Our findings in the rat and those by others in mice [4, 5, 34, 35] suggest that the region around the CC has a complex organization with heterogeneous progenitor-like cells. As in the brain , most cells lining the lateral aspects of the CC expressed the ependymal cell marker S100β, but some coexpressed the RG marker 3CB2 or vimentin and had a basal process projecting away from the CC suggesting a progenitor cell nature . Indeed, many S100β+/3CB2+/vimentin+ cells expressed PCNA—thus being within the cell cycle—with few undergoing division as indicated by pH3 expression .
The expression of nestin—a marker of neuroepithelial cells and RG —defined a second domain of heterogeneous cells contacting the poles of the CC that may also express RG markers. In adult mice, nestin is expressed preferentially on cells contacting the dorsal pole of the CC . The fact that cells contacting the poles of the CC in human neonates are also nestin+  suggests that this is an evolutionary preserved trait of early stages of postnatal development. Progenitors in adult neurogenic niches express GFAP in addition to nestin . However, the ependyma in the rat lacked GFAP immunoreactivity, in contrast with GFAP-green fluorescent protein transgenic mice which bear GFAP+ cells contacting the dorsal pole . The discrepancy of our data with those in mice may be species specific or age related (neonatal vs. adult). Interestingly, cells contacting the ventral but not the dorsal pole expressed the astrocyte and ependymal cell marker S100β, raising the possibility that midline domains may not be identical in their potential . In fact, adult mice progenitors generating neurospheres presumably localize on the dorsal aspect of the CC . Alternatively, nestin+/S100β+ cells on the ventral pole may be a transitional stage between RG and ependymal cells as described during brain development .
Vimentin+ and nestin+ cells contacting the poles of the ependyma had the morphological phenotype of RG with a pronounced apical–basal polarity as embryonic and adult neural stem cells . Although their perikarya laid at various distances from the CC, their centrosomes were always located in apical endfeet. In addition, our electron microscopy study showed that some apical processes had a single cilium with a 9 + 0 organization, a structural specialization thought as a key determinant of neural stem cells . The variety of midline cell morphologies resembled RG undergoing interkinetic nuclear migration during cortical development . Indeed, the fact that pH3 nuclei belonging to nestin+ cells were always close to the CC lumen, whereas PCNA nuclei located at various distances, supports the possibility that as in the embryo , RG nuclei in the postnatal spinal cord move apically to divide.
Functional Diversity: Simple but Working Together or Complex and Working Individually
Spinal ependymocytes in the neonatal rat had electrophysiological properties similar to those of progenitors during cortical development: low input resistances, passive responses, hyperpolarized resting membrane potentials, and extensive gap junction coupling via Cx43 [21, 41]. These properties together with the expression of RG markers within cell clusters may indicate a lineage relationship from RG to ependymal cells as in the SVZ .
In contrast to cells on lateral domains, midline RG were not coupled and thus function as individual units. Unlike neurogenic RG in the developing cortex , RG in the postnatal spinal cord had complex electrophysiological phenotypes displaying various combinations of IKD, IA, and/or ICa. The presence of IKD is a common feature among adult progenitors since it has been reported in hippocampal nestin+ type 2 cells  and GFAP+ cells in the SVZ . Although IA was not found in the adult SVZ , progenitors from the embryonic  and neonatal  SVZ and human stem cells  express IA. The phenotype of midline RG with conspicuous IKD and IA is remarkably similar to that of oligodendrocyte progenitors , raising the possibility they are bipolar precursors committed to the oligodendrocyte lineage still negative for NG2 and PDGFRα .
The complex repertoire of K+ currents may regulate fundamental properties of ependymal progenitor-like cells. IKD channels are major regulators of cell proliferation [46–48], and IA channels are essential for proliferation of multipotent human neural stem cells . Thus, K+ channels in midline RG may be part of epigenetic mechanisms that regulate proliferation. In addition, IA have been implied in the differentiation of oligodendrocyte precursors  and rat spinal cord astrocytes . Thus, another possibility is that K+ currents participate in the transition from RG to postmitotic spinal cells.
A minority of midline RG had ICa strong enough to sustain an LTS, a phenotype described in some floor plate cells . Ca2+ electrogenesis plays a central role during development by regulating events from neural induction  to various aspects of neuronal differentiation . For example, Ca2+ spikes are involved during early steps of differentiation of spinal neurons in Xenopus embryos  and newborn neurons in the adult hippocampus . In rats, a subpopulation of doublecortin+ CC-contacting neurons has a robust Ca2+ LTS . RG displaying ICa could be precursors showing the first signs of differentiation into CC-contacting neurons .
Heterogeneous Progenitors in Two Spatial Domains
Neural progenitors in the developing and adult brain are heterogenous and regionally specified in terms of lineage potential [56, 57]. Based on the expression of various stem cell markers, it has been recently proposed that CC-contacting progenitors in mice are heterogeneous cells with different potentials . We show here a new level of complexity demonstrating CC-contacting progenitors are also functionally heterogeneous and organized in spatial domains. The cells located in midline domains were almost quiescent exhibiting various molecular and structural features of neural stem cells . It is not clear whether the molecular heterogeneity and complex electrophysiological phenotypes within midline domains represent different types of progenitors or various functional/developmental stages of a single precursor. Progenitor-like cells within lateral domains combined features of ependymocytes and RG functionally grouped in multicellular units by Cx43. Gap junction coupling has been shown to promote stem cell proliferation , and this may be a key factor determining the difference in proliferation capabilities between domains. The spatial profile of Cx43 expression and the clusters of coupled progenitors described here resemble those of turtles , suggesting a phylogenetically conserved functional organization. However, unlike their reptilian counterpart, clustered progenitors in the rat did not express BLBP, suggesting a non-neurogenic nature [4, 37].
The fate and potentiality of CC-contacting progenitors as development proceeds remains unclear. Although shortly after birth, RG in the brain  and spinal cord  differentiate to astrocytes, progenitor-like cells remain in the adult spinal cord retaining the ability to react to injury [4, 5]. Ependyma-derived cells migrate away from the CC but unlike their counterparts in the ischemic brain  do not become neurons. Our current findings and the fact that some CC-contacting cells display molecular and functional properties of neuroblasts  supports the idea that the ependyma of the spinal cord has many elements of adult neurogenic niches. Although postnatal neurogenesis around the CC in mice has been reported , several studies suggest that under normal conditions this is just a latent capability of this stem cell niche [8, 35, 62]. The manipulation of specific functional properties in a heterogeneous population of CC-contacting progenitor-like cells may be useful to regulate their behavior and lineage potential, providing the cell types required to repair injured spinal circuits.
We thank G. Fabbiani and M.I. Rehermann for technical assistance; the generous gift by Dr. J. Sáez of the antibody against connexin 43, Dr. W. Stallcup of antibodies against NG2 and PDGFRα, and Dr. S. Doxsey of the antibody against pericentrin. The antibodies 40E-C developed by Dr. A. Álvarez-Buylla, 3CB2 developed by Dr. E.J. De La Rosa, and rat-401 developed by S. Hockfield were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biology, Iowa City, IA 52242. This work was supported by Grant FCE 2369 to N.M. and FCE 2367 to G.G. from ANII; and Grant R01NS048255 from the National Institute of Neurological Disorders and Stroke to R.E.R. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Neurological Disorders and Stroke or the National Institutes of Health. N.M. is a recipient of an ANII fellowship.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTERESTS
The authors indicate no potential conflicts of interests.