Reconstitution of the central and peripheral nervous system during salamander tail regeneration

Abstract

We show that after tail amputation in Ambystoma mexicanum (Axolotl) the correct number and spacing of dorsal root ganglia are regenerated. By transplantation of spinal cord tissue and nonclonal neurospheres, we show that the central spinal cord represents a source of peripheral nervous system cells. Interestingly, melanophores migrate from preexisting precursors in the skin. Finally, we demonstrate that implantation of a clonally derived spinal cord neurosphere can result in reconstitution of all examined cell types in the regenerating central spinal cord, suggesting derivation of a cell with spinal cord stem cell properties.

Fig. 1.

DRG number is restored by 35 d, but the correct inter-DRG distance is restored by 49 d of regeneration. Axolotl tail tissue was immunostained simultaneously using mouse anti–βIII-tubulin and mouse anti-NeuN antibody and then was detected in a single channel using a Cy5-coupled anti-mouse antibody. (A) Merged DIC/fluorescence image The arrow shows the amputation plane. (Scale bar: 1 mm.) (B–D) Bright cell clusters adjacent to the spinal cord can be observed at DRG 11–13 (B, B′, arrowheads) and 17–19 (C, C′, arrowheads). Postcloaca DRG 24–26 in the posterior tail region can identified by the axonal route emerging from the spinal cord region (D, D′, arrowheads). (Scale bars: B′–D′ and B′′–D′′, 0.7 mm.) (E) Inter-DRG distance was measured in all DRG posterior to the cut plane in amputated animals and compared with comparable DRG in control animals at 35 dpa. The inter-DRG distance is smaller in regenerated tails than in controls when assessed by the Student’s t test (P < 0.0001). (F) At 49 dpa the inter-DRG distances are comparable in regenerated tails and control animals (P = 0.39 for difference, Student’s t test).

Fig. 2.

Transplanted GFP⁺ spinal cord participated in regeneration of spinal cord and DRG. A 2- to 4-mm gap in the spinal cord was made in the axolotl tail and was replaced with a comparable length of spinal cord from the germline GFP⁺ transgenic animal. (A) Image immediately after the transplanted tail was amputated. (B) At 29 dpa the tail has regenerated much of its length, and the regenerated spinal cord is largely GFP⁺. (C) GFP⁺ spinal cord regenerates 105 dpa. Cross-sections from spinal cord-transplanted animal immunostained for βIII-tubulin. (D–F) Three examples of DRG adjacent to transplanted, regenerated GFP⁺ spinal cord. (D) An example in which the whole DRG was GFP⁻. (E) An example of a DRG with a mixture of GFP⁺ and GFP⁻ cells. (F) An example in which the whole DRG was GFP⁺. In D–F, GFP is shown as green, βIII-tubulin as red, and Hoechst staining as blue. (Scale bars: A–C, 1 mm; D, F, 100 μm; E 50 μm.)

Fig. 3.

Spinal cord neurospheres can be derived in FGF-2 and express neural stem cell markers. (A) Overview of procedure. Spinal cord tissue was isolated from GFP⁺ axolotl tail tissue, dissociated in papain, and expanded in serum-free medium containing growth factors and supplements (B27). (B) Low-magnification view of neurosphere culture after 2 wk. (Scale bar: 200 μm.) (C) Higher-magnification view of neurosphere. (Scale bar: 100 μm.). (D) Growth-factor dependence of neurosphere derivation. After dissociation, axolotl spinal cord cells were counted and expanded in serum-free medium in the presence of various growth factors; 50,000 cells were plated for each condition. Among the different factors and combinations of factors [e.g., FGF-2, EGF, FGF-2/EGF, PDGF, and leukemia-inhibitory factor (LIF)], FGF-2 yielded the highest number of neurospheres. The y-axis represents the yield of neurospheres per 50,000 plated cells. n = 3. (E) Confocal imaging of anti-Sox2 of a cross-section of an immunostained axolotl spinal cord shows neural progenitor cells. (F) Anti–Musashi-1 immunostaining of a spinal cord cross-section corroborates luminal location of neural progenitor cells. (G) Anti-GFAP immunostaining of a spinal cord cross-section reveals the presence of radial glia in the axolotl spinal cord. (H) Anti–βIII-tubulin immunostaining of a spinal cord cross-section shows neuronal cells in the axolotl spinal cord. (I) Confocal imaging of an anti-Sox2–immunostained neurosphere cross-section shows that the majority of cells are neural progenitors. (J) Anti–Musashi-1 immunostaining of a neurosphere cross-section shows that many cells are neural progenitors. (K) Anti-GFAP immunostaining of a neurosphere cross-section indicates that the majority of cells are apparently GFAP⁺. (L) Anti–βIII-tubulin immunostaining of a neurosphere indicates that some cells are or have differentiated into neuronal cells. Red indicates the antigen-specific immunostaining signal; blue indicates Hoechst staining. (Scale bars: E–L, 20 μm.) (M–O) To induce neuron outgrowth, neurospheres were plated onto poly-l-lysine/laminin–coated Petri dishes. After 7 d in culture, neurosphere-derived cells were positive for βIII-tubulin immunostaining (M) and were faintly positive for GFAP (N) but were negative for MBP immunostaining (O, Inset). Data shown in M, N, and O Inset are representative of three independent experiments. (O) In the presence of Shh agonist (50 ng/mL) and PDGF (25 ng/mL), neurosphere cells differentiated into MBP⁺ glia. Data shown are the results of two independent experiments. Inset shows MBP staining of a neurosphere grown in the absence of Shh agonist and PDGF. (Scale bars: M, 50 μm; N and O, 100 μm; O Inset, 50 μm.)

Fig. 4.

GFP⁺ cells derived from a passaged neurosphere implant contribute to peripheral nerves and DRG. (A) After the removal of an ∼1-mm section of spinal cord, the neurospheres were implanted in the lesion. (B) After 1–2 wk of wound healing, the implanted neurospheres were integrated in the spinal cord tissue. The tail was amputated close to the integrated GFP⁺ cells, maintaining them within the 500-μm zone. (C) The integrated GFP⁺ cells contribute to spinal cord regeneration upon amputation. (D) MBP immunostaining of a tail cross-section containing regenerated EGFP⁺ spinal cord derived from the passaged neurosphere implant. An MBP⁺/EGFP⁺ nerve is shown in the box. (E) Confocal image of peripheral nerve showing GFP⁺ nuclei and MBP⁺/GFP⁺ myelin. (F) An anti-MBP–stained nerve myelin sheet. (G) GFP channel view of the nerve indicates EGFP⁺ cell nuclei and myelin cytoplasm. Green indicates EGFP⁺ cells, red indicates MBP immunostaining, and blue indicates Hoechst staining. (H) βIII-Tubulin immunostaining of the same cross-section. A DRG containing βIII-tubulin⁺/GFP⁺ cells is shown in the box. (I–K) Confocal image of DRG showing βIII-tubulin⁺/GFP⁺ cell and nerves. Green indicates EGFP⁺ cell bodies and nerves; red indicates βIII-tubulin immunostaining; blue indicates Hoechst staining. (Scale bars: D and H, 50 μm; E, F, and I–K, 10 μm.)

Fig. 5.

Clonally derived neurosphere efficiently contributes to spinal cord regeneration and reconstitutes the whole complement of spinal cord cell types. (A) GFP⁺ cells cultured in microwells of an AggreWell 400 plate at clonal density with GFP⁻ white feeder cells. Arrow indicates a single EGFP⁺ cell in the microwell, imaged on the first day of the culturing procedure. (B) Clonal neurosphere that arose from the single EGFP⁺ cell indicated by the arrow in A after 34 d of culturing. (C) A clonal GFP⁺ neurosphere was engrafted into the injured spinal cord of the nontransgenic host. After 1 wk of healing the tail was amputated close to the implant, maintaining GFP⁺ cells within the 500-μm zone behind the amputation plane. (D–F) Outgrowth of the GFP⁺ spinal cord from the implanted clonal neurosphere monitored at 7, 24, and 54 dpa. (G–J) Cross-sections of the regenerated spinal cord from F at a caudal level. In this portion of the regenerate the full lateral half of the spinal cord was reconstituted from GFP⁺ cells. Cross-sections were analyzed by immunofluorescence for expression of cell type-specific markers. (G) Anti-GFAP staining shows GFP⁺ cells have formed GFAP⁺ radial glia. (H) Anti–βIII-tubulin staining shows GFP⁺/βIII-tubulin⁺ neuronal cells. (I) Cross-section immunostained for Pax6 shows that implanted GFP⁺ cells express Pax6 in the normal, lateral spinal cord domain. (J) Cross-section immunostained for Pax7 shows that implanted GFP⁺ cells in the dorsal domain express Pax7. (K and L) Shh in situ hybridization shows clonally derived Shh⁺/GFP⁺ floor plate cells (arrows). (M and N) Msx1 in situ hybridization shows Msx1⁺/GFP⁺ roof plate cell indicated by arrow. (O) βIII-Tubulin/MBP double-immunostaining of the cross-section. βIII-Tubulin was labeled with Cy5 secondary antibody and is shown in yellow. (P–R) Enlarged views of boxed area in O, which shows the axonal layer of the spinal cord containing MBP⁺/GFP⁺ myelin (arrowheads). In G–J, green indicates GFP⁺ cells; red indicates cell type-specific immunostaining; blue indicates Hoechst staining. In K–M, green indicates GFP, and blue indicates Hoechst staining. In O–R, yellow indicates βIII-tubulin; red indicates MBP; green indicates GFP; and blue indicates Hoechst staining. (Scale bars: A and B, 50 μm; C–F, 1 mm; G–J, 25 μm; K–N, 50 μm; O, 20 μm; P–R, 5 μm.)

Fig. P1.

Neural stems cells expressing a transgene encoding GFP (green) (A) were implanted into the axolotl spinal cord that then was severed distally (B). The implanted cells regenerated both the spinal cord and the peripheral nervous system (C).