Discoidin domain receptor 1 functions in axon extension of cerebellar granule neurons

Abstract

In the developing cerebellum, granule neuron axon outgrowth is a key step toward establishing proper connections with Purkinje neurons, the principal output neuron of the cerebellum. During a search for genes that function in this process, we identified a receptor tyrosine kinase discoidin domain receptor 1 (DDR1) expressed in granule cells throughout their development. Overexpression of a dominant-negative form of DDR1 in immature granule cells results in severe reduction of neurite outgrowth in vitro, in dissociated primary culture, and in vivo, in organotypic slices of neonatal cerebellum. Granule cells that fail to extend axons are positive for differentiation markers such as TAG-1 and the neuron-specific class III beta-tubulin, suggesting that development is affected after granule cells commit to terminal differentiation. DDR1 activation appears to be mediated by its ligand, collagen, which is localized to the pial layer of the developing cerebellum, thereby leading to granule cell parallel fiber extension. Our results therefore indicate that collagen-DDR1 signaling is essential for granule neuron axon formation and further suggest a unique role of pia in cerebellar cortex histogenesis.

Figure 1

DDR1 is expressed in the developing CNS. DDR1 mRNA is detected in parasagittal sections of P6 (A,B) and adult (C,D) mouse brain by in situ hybridization. Labeled cells are seen in the cerebellum, dentate gyrus of the hippocampal formation, rostral migratory stream, olfactory bulb, and weakly in the cortex at P6 (A) and in the cerebellum and dentate gyrus of the hippocampal formation in the adult (C). At higher magnification of a cerebellar folium, expression is seen in the EGL, the ML, and the IGL at P6 (B) and in the IGL as well as in scattered cells in the ML in the adult (D). (E) Northern analysis detects DDR1 in total RNAs (10 μg/lane) from various tissues at P6. (Lane 1) Olfactory bulb; (lane 2) hippocampus; (lane 3) cerebral cortex; (lane 4) cerebellum; (lane 5) purified cerebellar granule cells; (lane 6) lung; (lane 7) spleen; (lane 8) liver; (lane 9) kidney; (lane 10) heart. GAPDH is used as a control for RNA levels. (F) Steady levels of DDR1 signal are detected in whole brain at embryonic stages: (lane 1) E10.5; (lane 2) E12.5; and also in cerebellum throughout its developmental stages: (lane 3) E14; (lane 4) E17; (lane 5) P0; (lane 6) P6; (lane 7) P10.

Figure 2

The ligand of DDR1 and collagen IV are colocalized to the cerebellar pial surface. Immunofluorescent staining using DDR1–Fc (A–C) or an anti-collagen IV antibody (D–F) on parasagittal sections of mouse cerebellum identified positive signals at the pial surface of the cerebellum. Serially adjacent sections were stained with DDR1–Fc or an anti-collagen IV antibody at P0 (A,D), P6 (B,E), and adult (C,F). In an adult stage, positive signal was also present in the blood vessels of the cerebellum. (G) Graded localization of collagen IV from the pial surface detected on a P6 cerebellar sagittal section (30 μm).

Figure 3

Coexpression of DDR1Δ or DDR1–Fc with wild-type DDR1 down-regulates tyrosine phosphorylation of DDR1 kinase. 293T cells were transiently transfected with the FLAG-tagged full-length wild-type DDR1 together with various expression constructs indicated. Eighteen hours after transfection, cells were serum-starved overnight and then stimulated with collagen I (10 μg/ml) for 2 hr. Immunoblotting was performed using antibodies indicated. Lane 2 shows that DDR1 was phosphorylated in the presence of collagen I. This phosphorylation was inhibited by the coexpression of DDR1Δ (lane 3). Lanes 4 and 5 show that DDR1–Fc could also reduce phosphorylation of DDR1 as opposed to the negative control (ssFc; lane 4). Lane 6 shows that phosphorylation persisted after collagens were removed from the culture supernatant, and the cells were further incubated for 3 hr before harvesting.

Figure 4

A soluble form of DDR1 functions dominant negatively in neurite extension of granule cells cultured on pial cells. Granule cells were placed at a low density on a pial monolayer expressing either DDR1–Fc (A,C) or ssFc (B,D) and cultured for 36 hr. Pial cells infected with GFP viruses expressing recombinant Fc-fusion proteins were visualized by GFP epifluorescence, confirming that almost all pial cells express a uniform level of fusion proteins. After fixation, granule cells (red) were visualized by antibody staining using either Tuj-1 (A,B) or anti-TAG-1 antibody (C,D) with a Cy-3 conjugated secondary antibody. Fewer granule cells extended neurites in the presence of DDR1–Fc, as opposed to in the presence of ssFc. Arrowheads indicate granule cells positive for antibody staining. Expression of recombinant Fc-fusion proteins in the pial culture supernatant was confirmed by immunoblotting using an anti-human IgG1–Fc antibody (E). DDR1–Fc migrated at 98 kD, and the ssFc migrated at 27 kD. (F) The percentage of Tuj-1-positive granule cells with neurites >20 μm in length, placed on the pial monolayer expressing either DDR1–Fc (5 ± 2%) or ssFc (70 ± 6%), was scored. Scale bars, 10 μm in A,B and 20 μm in C,D.

Figure 4

A soluble form of DDR1 functions dominant negatively in neurite extension of granule cells cultured on pial cells. Granule cells were placed at a low density on a pial monolayer expressing either DDR1–Fc (A,C) or ssFc (B,D) and cultured for 36 hr. Pial cells infected with GFP viruses expressing recombinant Fc-fusion proteins were visualized by GFP epifluorescence, confirming that almost all pial cells express a uniform level of fusion proteins. After fixation, granule cells (red) were visualized by antibody staining using either Tuj-1 (A,B) or anti-TAG-1 antibody (C,D) with a Cy-3 conjugated secondary antibody. Fewer granule cells extended neurites in the presence of DDR1–Fc, as opposed to in the presence of ssFc. Arrowheads indicate granule cells positive for antibody staining. Expression of recombinant Fc-fusion proteins in the pial culture supernatant was confirmed by immunoblotting using an anti-human IgG1–Fc antibody (E). DDR1–Fc migrated at 98 kD, and the ssFc migrated at 27 kD. (F) The percentage of Tuj-1-positive granule cells with neurites >20 μm in length, placed on the pial monolayer expressing either DDR1–Fc (5 ± 2%) or ssFc (70 ± 6%), was scored. Scale bars, 10 μm in A,B and 20 μm in C,D.

Figure 5

Collagen does not affect the proliferative status of granule cells. (A,B) Granule cells were purified and cultured in a 96-well microtiter plate with indicated amounts of collagen I or IV for 48 hr. The culture was pulsed with [³H]thymidine for the last 12 hr, and incorporated counts were measured by a scintillation counter. (C,D) Granule cells were also cultured on a microtiter plate precoated with indicated amounts of collagen I or collagen IV, and thymidine incorporation was measured. Increasing amounts of collagen I and IV, either soluble or immobilized, did not affect the proliferative status of the granule cells.

Figure 5

Collagen does not affect the proliferative status of granule cells. (A,B) Granule cells were purified and cultured in a 96-well microtiter plate with indicated amounts of collagen I or IV for 48 hr. The culture was pulsed with [³H]thymidine for the last 12 hr, and incorporated counts were measured by a scintillation counter. (C,D) Granule cells were also cultured on a microtiter plate precoated with indicated amounts of collagen I or collagen IV, and thymidine incorporation was measured. Increasing amounts of collagen I and IV, either soluble or immobilized, did not affect the proliferative status of the granule cells.

Figure 6

Expression of dominant-negative DDR1 inhibits neurite extension of granule cells in culture. P6 cerebellar granule cells extend neurites when infected with pLIA (A) or with wild-type DDR1 (B). When retrovirus expressing DDR1Δ was used, infected granule cells did not extend neurites (C). Granule cells with processes longer than three cell-body lengths were scored as neurite-harboring cells. The effect was quantitated in the graph (D). The experiment was performed seven times, and the number of cells with versus the number of cells without neurites were counted. The averages were normalized to the pLIA control. As compared with pLIA (100%), DDR1Δ decreased neurite outgrowth to 32 ± 6.7%, and the wild-type virus did not decrease neurite outgrowth (91 ± 9%) (P = 0.0032 in a paired t-test). Scale bar, 50 μm.

Figure 7

Expression of dominant-negative DDR1 inhibits parallel fiber extension of granule cells developing in situ. Coronal slices of P8 cerebellum were infected either with pLIA or with DDR1Δ/LIA. Thirty-six to 72 hr later, parallel fibers were visualized by confocal microscopy after an anti-AP antibody staining. (A) Granule cells (arrowheads) infected with DDR1Δ/LIA are seen with truncated axons. (B) A cell infected with pLIA extends long bipolar axons (p). (C) Another cell (arrowhead) infected with pLIA migrates inward along glial fibers, revealing a T-shaped parallel fiber (p) with a leading process (lp). The percentages of neurons with parallel fibers were scored (D). Infection of DDR1Δ reduced the numbers of granule cells with parallel fibers (pLIA = 78 ± 7%, DDR1Δ = 44 ± 5%, n = 4, P = 0.01 by a t-test). Scale bars, 10 μm in A and 20 μm in B,C.

Figure 7

Expression of dominant-negative DDR1 inhibits parallel fiber extension of granule cells developing in situ. Coronal slices of P8 cerebellum were infected either with pLIA or with DDR1Δ/LIA. Thirty-six to 72 hr later, parallel fibers were visualized by confocal microscopy after an anti-AP antibody staining. (A) Granule cells (arrowheads) infected with DDR1Δ/LIA are seen with truncated axons. (B) A cell infected with pLIA extends long bipolar axons (p). (C) Another cell (arrowhead) infected with pLIA migrates inward along glial fibers, revealing a T-shaped parallel fiber (p) with a leading process (lp). The percentages of neurons with parallel fibers were scored (D). Infection of DDR1Δ reduced the numbers of granule cells with parallel fibers (pLIA = 78 ± 7%, DDR1Δ = 44 ± 5%, n = 4, P = 0.01 by a t-test). Scale bars, 10 μm in A and 20 μm in B,C.