(a) Postnatal day 7 (P7) cerebellar granule neurons (CGNs) from WT, NgR1^−/−, NgR12^−/−, and NgR3^−/− pups are strongly inhibited when plated on crude CNS myelin substrate (40μg/ml). In marked contrast, CGNs from NgR triple mutant (NgR123^−/−) mice grow longer neurites on CNS myelin. On CNS myelin isolated from NogoABC;MAG;OMgp triple null (NMO^−/−) mice (40μg/ml), CGNs from WT, NgR1^−/−, NgR12^−/−, and NgR3^−/− pups show enhanced neurite outgrowth. A further release of inhibition is observed when NgR123^−/− neurons are plated on NMO^−/− CNS myelin. On a BSA control substrate, neurite length of all five genotypes is comparable. (b) Quantification of neurite length. At least 300 neurites of TuJ1–labeled cells were counted per condition (n=9 independent experiments). Light gray bars (BSA); black bars (WT myelin); dark gray bars (NMO myelin). Results are presented as mean ±SEMs. ** P< 0.001 (one way ANOVA, Tukey’s post–hoc). Scale bar, 50μm.

(a) Coronal sections of E18 rat brain showing the binding pattern of AP–NgR1 and AP–NgR3. No binding is observed for AP–NgR2. (b–c) Binding of (b) NgR1–Fc and (c) NgR3–Fc to E18 brain sections is abolished upon deletion of a cluster of basic residues (motif m2) in the stalk region. (d) Schematic of receptor deletion constructs and their relative binding to E18 rat brain tissue compared to soluble NgR1 [NgR1(ΔGPI)]. Soluble NgR1 (red) and NgR3 (yellow), but not NgR2 (green), bind strongly to brain tissue sections. The LRRs of NgR1, previously shown to participate in myelin inhibitor binding, are dispensable for binding to neural tissue. Deletion of a cluster of basic amino acid residues in the C–terminal region of the NgR1 and NgR3 stalk (motif m2), or replacement of these residues with alanines [NgR1(7ala)], completely abolishes binding. (e) Sequence alignment of binding motifs m1 and m2 of NgR1 and NgR3. In the LRR–CT domain, residues F278 and R279 in NgR1 and residues F273 and R274 in NgR3 (motif m1) are important for GAG binding. Motif m2 is comprised of a cluster of basic amino acids, including residues 414–426 in NgR1 and residues 403–415 in NgR3. The conserved residues are bolded and the basic residues are in blue. Scale bar, 40μm.

(a) Binding of soluble AP–NgR1 to P1 rat brain tissue sections is sensitive to heparinase (Hep’ase) and chondroitinase–ABC (Ch’aseABC) treatment (1unit/ml). (b) Binding of soluble AP–NgR1 and AP–NgR3 to E18 rat brain tissue sections in the presence of PBS (vehicle control), 50μg/ml monosulfated CS–GAGs (CS–A, CS–B, or CS–C), or 50μg/ml disulfated CS–GAGs (CS–D or CS–E). Only CS–B, CS–D, and CS–E compete effectively with NgR1 and NgR3 for binding to brain tissue sections. (c) ELISA binding studies revealed a direct association of soluble NgR1 and RPTPσ(1–3) with specific GAGs (insets show Scatchard plot analysis). The calculated Kds for CS–B, CS–D, CS–E, and heparin are 3.2, 1.3, 1.0, and 1.2nM (NgR1) and 1.4, 3.0, 0.1, and 0.1nM [RPTPσ(1–3)], respectively. (d) WT P7 rat CGNs are strongly inhibited when plated on substrate–bound CSPGs. Long neurites are seen on a BSA control substrate. In the presence of NgR1–Fc or RPTPσ(1–3)–Fc, but not NgR1(7ala)–Fc or IgG1 control, CSPG inhibition is abolished. On a BSA substrate, soluble receptors do not influence neurite outgrowth. Quantification of neurite length is shown as a percentage of the IgG1–BSA control (100%). At least 300 neurites of TuJ1–labeled cells were counted per condition (n=4 independent experiments). Gray bars (BSA); black bars (CSPGs). Results are presented as mean ±SEMs. ** P< 0.001 (one way ANOVA, Tukey’s post–hoc), n.s.= not significant. Scale bar: a, 20μm; b, 40μm; d, 50μm.

(a) In transiently transfected COS–7 cells, wildtype NgR1 and the GAG–binding–deficient mutant [NgR1(7ala)] show very similar binding of the MAIs AP–Nogo66, MAG–Fc, and AP–OMgp. The NgR1 fragment F278–E445 (AP–NgR1^Δ15CT+stalk) is sufficient for high–affinity binding to neural GAGs. When residues 278–445 in NgR1 are replaced with the corresponding sequences of NgR2, resulting in construct NgR1^C27–K277/NgR2^V281–L420, binding of MAIs is not diminished compared to wildtype NgR1. Ligand binding to GFP–transfected–COS 7 cells is shown as a negative control. (b) Schematic of NgR1, showing the regions necessary for binding of MAIs and GAGs. The GAG–binding motifs m1 and m2 of NgR1 are distinct and dissociable from the Nogo–, MAG–, and OMgp–binding sequences. (c) In the presence of bath applied CS–B, CS–D, or CS–E GAGs (1mg/ml), binding of AP–Nogo66 to NgR1–expressing COS–7 cells is not altered. Scale bar, 20μm.

(a) In vitro, WT P7 CGNs are strongly inhibited when plated on substrate–bound CSPGs (10μg/ml). Studies with CGNs from NgR1^−/−, NgR12^−/−, and NgR3^−/− mutants revealed no significant release of CSPG inhibition. Loss of all three NgRs (NgR123^−/−) or RPTPσ alone (RPTPσ^−/−) leads to a significant, yet incomplete release of CSPG inhibition. On a control substrate (BSA), neurite length of all six genotypes is comparable. (b) Quantification of neurite length. Gray bars (BSA); black bars (CSPGs). At least 300 neurites of TuJ1–labeled cells were counted per condition (n=8 independent experiments). Results are presented as mean ±SEMs. ** P< 0.001 (one way ANOVA, Tukey’s post–hoc), n.s.= not significant. Scale bar, 70μm.

(a) Longitudinal sections of uninjured and injured adult mouse optic nerve (7 days following retro–orbital nerve crush) were incubated with NgR1–Fc, NgR3–Fc, NgR1(7ala)–Fc, and NgR3Δm2–Fc. Soluble NgR1–Fc and NgR1(7ala)–Fc, but not NgR3–Fc or NgR3Δm2–Fc, bind weakly to uninjured optic nerve sections. Following injury, binding of NgR1–Fc and NgR3–Fc is strongly increased and depends on the presence of the proteoglycan binding motif m2, as no increase was observed for NgR1(7ala)–Fc and NgR3 m2–Fc. Following crush injury to the optic nerve, GFAP immunoreactivity is upregulated along the entire nerve. Treatment of injured optic nerve sections with 1unit/ml chondroitinase ABC–(Ch’ase) strongly reduces NgR1–Fc and NgR3–Fc binding, suggesting that endogenous CSPGs participate in the binding of NgR1 and NgR3. (b) Quantification of binding to optic nerve sections. All binding is shown as a fold increase relative to AP–Fc. At least 20 sections were counted per condition (n=4 independent experiments). Light gray bars (Uninjured); black bars (Injured); dark gray bars (Injured + Ch’ase). Results are presented as mean ±SEMs. ** P< 0.001 (one way ANOVA, Tukey’s post–hoc), n.s.= not significant. Scale bar, 30μm.

(a) Sections of adult wildtype (WT) and Nogo receptor triple mutant (NgR123^−/−) mouse retina were subjected to in situ hybridization with digoxigenin–labeled cRNA probes specific for NgR1, NgR2, and NgR3 transcripts. All three receptors are strongly expressed in the ganglion cell layer (arrow) and the inner nuclear layer, but are absent from the outer nuclear layer of the retina. No signal was detected on parallel–processed sections of NgR123^−/− retina. (b) Hoechst 33342 nuclear staining, as well as anti–calbindin and anti–calretinin immunolabeling, of adult WT and NgR123^−/− retina did not reveal any noticeable differences in retinal organization among the two genotypes. (c) Toluidine blue labeling of epon–embedded adult WT and NgR123^−/− optic nerve cross sections reveals a comparable number of axons and degree of myelinated fibers. (d–f) The fidelity of RGC central projections in six–week–old WT and NgR123^−/− mice was assessed by anterograde fiber tracing. Five days after injection of Alexa 594–conjugated Cholera Toxin β into the right eye and Alexa 488–conjugated Cholera Toxin β into the left eye, mice were sacrificed, perfused, and brain sections analyzed by fluorescence microscopy. Right eye (red) and left eye (green) RGC projections to the (d) superior colliculus, (e) suprachiasmatic nucleus and (f) lateral geniculate nucleus in NgR123^−/− mice are indistinguishable from age–matched WT controls. Scale bar: a, b, 80μm; c, 5μm; d, 100μm; e, f, 60μm.

Two weeks following injury, regenerating axons in optic nerve sections were visualized by anti–GAP43 immunolabeling. The injury site is marked with an asterisk. (a) WT mice show very limited regenerative axonal growth following injury. (b) In NgR123^−/− mice, many GAP43⁺ fibers grow beyond the lesion site. (c) In NgR13/RPTPσ^−/− (NgR13/σ^−/−) mice, a further increase of GAP43⁺ fiber growth is observed. (a′), (b′), and (c′) depict higher magnification images of the region 0.5 to 0.75mm distal to the lesion site [dotted line in images (a), (b), and (c), respectively]. (g) Quantification of the number of GAP43⁺ axons at 0.2 to 1.2mm distal to the lesion site. Light gray bars (WT, n=6); black bars (NgR1^−/−, n=7); purple bars (RPTPσ^−/−, n=5); dark gray bars (NgR123^−/−, n=8); blue bars (NgR13/σ^−/−, n=4). (d) Intraocular injection of Zymosan enhances regenerative axonal growth in WT mice. A further increase is observed in (e) NgR123^−/− mice, which is further enhanced in (f) NgR13/σ^−/−, mice. (d′), (e′), and (f′) depict higher magnification images of the region 0.5 to 0.75mm distal to the lesion site [dotted line in images (d), (e), and (f), respectively]. (h) Quantification of the number of GAP43⁺ axons at 0.2 to 1.6mm distal to the lesion site in Zymosan–injected mice. Light gray bars (WT + Zymosan, n=6); black bars (NgR1^−/− + Zymosan, n=6); purple bars (RPTPσ^−/− + Zymosan, n=4); dark gray bars (NgR123^−/− + Zymosan, n=8); blue bars (NgR13/σ^−/− + Zymosan, n=3). Results are presented as mean ±SEMs. ** P< 0.05 (one way ANOVA, Tukey’s post–hoc). Scale bar, 200μm.