Figure 1.
Genetic Labeling of Spinal Cord Ependymal Cells
Transgenic mice with tamoxifen-inducible Cre recombinase (CreER) under the control of the FoxJ1 promoter (A and B) or the Nestin second intron enhancer (C and D) drive expression and induce recombination after 5 daily tamoxifen injections (resulting in β-gal expression) in cells lining the central canal in the adult spinal cord.
(A and C) Overviews of coronal sections from the thoracic spinal cord and (B and D) higher magnification of the central canal region demonstrating recombination in the majority of cells and cytoplasmic CreER protein 6 days after the last tamoxifen administration. Cell nuclei are visualized with DAPI in (A and C). Scale bars indicate 100 μm in (A) and (C), and 25 μm in (B) and (D).
Figure 2.
Characterization of Spinal Cord Ependymal Cells
(A–C) Immunoelectron microscopy of the central canal in a FoxJ1-CreER mouse. Pseudocoloring in (A) illustrates the localization of CreER-immunoreactive radial ependymal cells (B), cuboidal ependymal cells, and tanycytes (C).
(D) Table with color code for (A–C), describing the characteristics of the three cell types that line the central canal.
(E) Venn diagram illustrating the percentage of cuboidal ependymal cells, tanycytes, radial ependymal cells, and intermediate morphologies (see Figures S4 and S5, and Table S1 for details on the ultrastructural analysis). Scale bars indicate 10 μm in (A) and 3 μm in (B) and (C).
Figure 3.
Ependymal Cells Display Neural Stem Cell Properties In Vitro
(A) Schematic depiction of tamoxifen administration paradigm and (B) analysis of neural stem cell properties and in vivo and in vitro recombination frequency.
(C) The high proportion of recombined (rec.) neurospheres (mean + SD, n = 6 mice for each transgenic mouse line) demonstrates that the majority derives from ependymal cells.
(D) Estimate of the proportion of neurospheres that derive from ependymal cells by normalization to the recombination rate in tissue sections of the spinal cords that were used to initiate the cultures (mean ± SD).
(E) Recombined primary neurospheres from FoxJ1-CreER mice on R26R background visualized by X-gal staining (arrow points to one unrecombined neurosphere).
(F) Differentiation of a clonally derived recombined neurosphere into neurons (βIIItub), astrocytes (GFAP), and oligodendrocytes (O4).
(G) Flow cytometric isolation of GFP+ cells from the spinal cord of adult FoxJ1-CreER mice based on the IRES-GFP signal (GFP gate). 7-AAD labels dead cells, which were excluded.
(H–K) Brightfield (BF) and fluorescent images (I and K) of a single GFP+ sorted cell and a neurosphere (J) formed from such a cell, which is GFP− due to the lack of FoxJ1 expression (K).
Scale bars indicate 400 μm in (E), 20 μm in (F), 100 μm in (H) and (I), and 50 μm in (J) and (K).
Figure 4.
Ependymal Cells Self-Renew In Vivo
(A and B) BrdU incorporation in ependymal cells after 4-wk administration in the drinking water. Many of the labeled cells are found in pairs (arrowheads).
(A) shows a coronal and (B) a sagittal section.
(C–E) The recombination (rec.) rate of ependymal cells remains at the same level from 2 d to 10 mo after tamoxifen administration (mean and SD from 3–4 mice at each time point).
The scale bar indicates 25 μm in (A–D).
Figure 5.
Ependymal Cells Are Activated by Injury
Uninjured and adjacent injured segments from mice 4 d after a dorsal funiculus incision. Recombined cells leave the ependymal layer in the injured segments (arrowheads). (A and B) Ki67 immunoreactivity indicates ependymal proliferation in the injured, but not in the uninjured segment. Migrating recombined cells lose Sox3 expression (D), but most are Sox9 immunoreactive, and a smaller population is GFAP immunoreactive (F–I). Some ependymal cells within the ependymal layer (outlined by hatched line in [I]) become GFAP immunoreactive (arrows). (J and K) Electron micrographs of an extended ependymal cell with a dense filamentous matrix (f) in the cytoplasm 4 wk after injury in a FoxJ1-CreER mouse. n, nucleus.
Scale bars indicate 25 μm in (A–H), 10 μm in (I), 1.5 μm in (J), and 0.2 μm in (K).
Figure 6.
Ependymal Cells Contribute to Scar Formation after Spinal Cord Injury
(A) Distribution of β-gal–immunoreactive ependyma-derived cells in coronal sections from an uninjured segment (left) further towards the lesion epicenter (right).
(B and C) Sagittal sections show the distribution of recombined cells 1 mo (B) and 10 mo (C) after a dorsal funiculus incision (indicated by hatched lines).
(D and E) Recombined cells outside the ependymal layer display either (D) the astrocytic marker GFAP or (E) a Sox9/vimentin double-positive profile.
(F and G) Other recombined cells are Olig2 immunoreactive (arrowheads) and have oligodendrocyte morphology at later time points ([F] is at 1 mo and [G] 10 mo).
(H and I) The scar tissue is compartmentalized with patches of ependyma-derived cells.
(J and K) Electron micrographs of β-gal–immunoreactive cells with astrocyte (J) or oligodendrocyte (K) morphology. Boxed areas are shown at a higher magnification in the insets.
(L) Drawing (based on [C]) depicting the distribution of recombined cells of different phenotypes.
(M) Marker profile of recombined ependyma-derived cells (mean and SD from 3 mice at each time point) in the area encompassing the injury indicated by hatched lines in (L).
c, caveolae; f, filaments; ld, lipid droplet; n, nucleus.
Scale bars indicate 25 μm in (D–G) and (H), 50 μm in (I), 2 μm in (J), 1 μm in (K), and 0.15 μm in insets in (J) and (K).
Figure 7.
Relationship between Ependymal Cell-Derived Progeny, Axonal Growth-Modulating Molecules, and Axons in the Scar Tissue
(A–E) Chondroitin sulphate proteoglycans (CSPG), which are axonal growth inhibitory, are abundant in the scar tissue 2 wk after a dorsal funiculus lesion. CSPG are present in a complementary and nonoverlapping pattern to β-gal–expressing ependyma-derived cells. The extracellular matrix proteins fibronectin (A) and laminin (B and F), which are permissive to axonal sprouting, are widely distributed within the scar tissue and overlap with β-gal–expressing ependyma-derived cells. (C–H) Neurofilament (NF)-immunoreactive axons in the scar tissue are rarely present in CSPG+ areas (C–E) but are associated with ependymal cell progeny (C–H) and laminin (F–H) in the core of the forming scar tissue 2 wk after injury. Scale bars indicate 100 μm.
Figure 8.
Ependymal Cells Give Rise to Oligodendrocytes after Injury
(A) Ependymal cell-derived progeny are most abundant within the core of the scar tissue forming at the injury (arrow). Ependyma-derived cells are also found, more sparsely, over a larger area in the intact tissue bordering the lesion (arrowheads), where they are associated with myelin basic protein (MBP)-immunoreactive myelin ensheathing neurofilament (NF)-immunoreactive axons.
(B–D) Ependymal cell-derived progeny harboring an oligodendrocytic morphology are found both in the grey (B) and white matter (C), and some recombined processes wrap around myelinated axon (D).
Scale bars indicate 100 μm in (A), 25 μm in (B) and (C), and 10 μm in (D).
Figure 9.
Ultrastructure of an Ependyma-Derived Oligodendrocyte
Electron micrograph of an ependyma-derived β-gal–expressing cell in the nuclear plane 10 mo after injury (nucleus labeled with asterisk in [A]). (A) The cell displays ultrastructural characteristics of a mature oligodendrocyte such as a denser cytoplasm with fewer intermediate filaments than that of a neighboring astrocyte (As), (B) tight junctions (tj) between the cell body, and an oligodendrocyte process of another cell and (C) granular endoplasmic reticulum (er) cycternae in the perikaryon. Ax, axon; f, filamentous matrix.
(D) A process (p) of the cell adjacent to the myelin sheath on an axon.
Scale bars indicate 1 μm in (A) and 250 nm in (C) and (D).