Presenilin-dependent receptor processing is required for axon guidance

Abstract

The Alzheimer's disease-linked gene presenilin is required for intramembrane proteolysis of amyloid-β precursor protein, contributing to the pathogenesis of neurodegeneration that is characterized by loss of neuronal connections, but the role of Presenilin in establishing neuronal connections is less clear. Through a forward genetic screen in mice for recessive genes affecting motor neurons, we identified the Columbus allele, which disrupts motor axon projections from the spinal cord. We mapped this mutation to the Presenilin-1 gene. Motor neurons and commissural interneurons in Columbus mutants lacking Presenilin-1 acquire an inappropriate attraction to Netrin produced by the floor plate because of an accumulation of DCC receptor fragments within the membrane that are insensitive to Slit/Robo silencing. Our findings reveal that Presenilin-dependent DCC receptor processing coordinates the interplay between Netrin/DCC and Slit/Robo signaling. Thus, Presenilin is a key neural circuit builder that gates the spatiotemporal pattern of guidance signaling, thereby ensuring neural projections occur with high fidelity.

Figure 1

Columbus Mutants Display Midline Motor Axon Guidance Defects (A–F) Motor axons in transverse sections of E12.5 mouse embryos at the brachial level labeled with ISL^MN:GFP-F transgenic reporter. Boxed regions in A–C are enlarged in D–F respectively (FP, floor plate). n>8 embryos for each genotype. (G–L) Flat-mount images of E13 mouse spinal cords at lumbar levels with anterior on top. Boxed regions in G–I are enlarged in J–L respectively. Dotted line marks medial edge of MN cell bodies. Note that motor columns are slightly disorganized at the lumbar level of PS1 mutants, leading to an increased distance between the motor columns. n>10 embryos for each genotype. (M) Schematic of Columbus mutation in PS1 gene. (N) RT-PCR analysis of PS1 mRNA using primers flanking intron 11 (arrowheads in (M)). (O) Western blot analysis of PS1 protein in Columbus mutants. Full-length PS1 protein is proteolytically processed in vivo to a 30-kDa N-terminal fragment (NTF) and a 20-kDa C-terminal fragment (CTF). Antibody against PS1 N-terminus recognized the full-length holoprotein (FL) and NTF in wild type but not in the Columbus mutant. Likewise, an antibody against the C-terminus of PS1 recognized the 20-kDa CTF in wild type but not Columbus mutants. GAPDH was used as an internal control. Scale bar = 100 μm for A–C, 40 μm for D–F, 75 μm for G–I, 30 μm for J–L. See also Figure S1

Figure 2

PS1 Deficient Motor Neurons are Attracted to Netrin (A) Schematic of motor explant repulsion assay. MN explants (MN) are cocultured with floor plate (FP) or Cos cell aggregates in 3D Collagen/Matrigel matrices for 24 hours. Motor axons are visualized with the transgenic ISL^MN:GFP-F reporter. (B) Schematic of motor explant attraction assay. MN explants (MN) are cocultured with Cos cell aggregates (Cell Agg) in 3D Collagen/Matrigel matrices for 15 hours, which leads to minimal motor axon outgrowth when cocultured with control Cos cell aggregates. (C–F) Motor explant repulsion assay. GFP-labeled mouse motor explants were cocultured with FP (FP was to the left of explants). FPs and motor explants from PS1 heterozygous or wild type littermates were used as controls. (G) Histogram showing quantification (proximal-distal (P:D) ratio) of outgrowth from explants in culture with FPs. n=8 (Ctrl FP+Ctrl MN), n=7 (Ctrl FP+KO MN), n=7 (KO FP+Ctrl MN), n=5 explants (KO FP+KO MN). Data are presented as the mean±SEM (* p<0.05). (H–K) Motor explant repulsion assay. GFP-labeled mouse motor explants were cocultured with Cos cell aggregates transfected as indicated (cell aggregates were to the left of explants). (L) P:D ratio of outgrowth from explants in the presence of cell aggregates. For control (WT/Het) motor explants, n=13 (control), n=10 (Sema3A), n=17 (Slit2); for PS1 KO motor explants, n=9 (control), n=11 (Sema3A), n=9 (Slit2). Data are presented as the mean±SEM (* p<0.05). (M–R) Motor explant attraction assay. GFP-labeled mouse motor explants were cocultured with Cos cell aggregates transfected as indicated (cell aggregates were to the left of explants). (S) P:D ratio of outgrowth from explants in the presence of Cos cell aggregates. For control (WT/Het) motor explants, n=11 (control), n=10 (Shh), n=12 (Netrin); for PS1 KO motor explants, n=15 (control), n=8 (Shh), n=10 (Netrin). Data are presented as the mean±SEM (* p<0.05). (T) Histogram showing quantification (neurite numbers relative to control) of outgrowth from motor explants in the presence of recombinant Netrin-1. For control (WT/Het) motor explants, n=21 (0 ng/ml), n=25 (50 ng/ml), n=19 (250 ng/ml); for PS1 KO motor explants, n=13 (0 ng/ml), n=15 (50 ng/ml), n=16 (250 ng/ml). Data are presented as the mean±SEM (* p<0.05). See also Figure S2

Figure 3

Inhibition of Netrin/DCC Signaling Rescues the Midline-crossing Phenotype (A and B) Motor explant repulsion assay. GFP-labeled PS1 KO motor explants were cocultured with control (WT/Het) FPs in the presence of (A) vehicle control or (B) DCC function blocking antibody. (C) P:D ratio of outgrowth from explants in the presence of FPs. n=9 (control), n=7 (DCC Ab). Data are presented as the mean±SEM (* p<0.05). (D–O) Flat-mount images of GFP-positive MNs in E13 lumbar mouse spinal cords. Boxed regions in D–F and J–L are enlarged in G–I and M–O respectively. Anterior is to the top for all the images. Note that motor columns are slightly disorganized in PS1/Netrin-1 and PS1/DCC double mutants, similar to those observed in PS1 single mutants. At least three embryos were assayed for each genotype. Scale bar = 50 μm for D–F, 30 μm for G–I, 50 μm for J–L, 30 μm for M–O.

Figure 4

Inhibition of γ-secretase Activity Switches MNs to a Netrin-responsive State (A and B) Motor explant repulsion assay. GFP-labeled motor explants were cocultured with FPs in the presence of (A) DMSO vehicle or (B) γ-secretase inhibitor L-685458. (C) P:D ratio of outgrowth from explants in the presence of FPs. n=7 (DMSO), n=9 (γ-sec inh). Data are presented as the mean±SEM (* P<0.05). (D) Side view and (E) top view schematic of an assembled Dunn chamber. ISL^MN:GFP-labeled chick MNs cultured on a coverslip have been inverted over the chamber. The inner well is filled with control media whereas the outer well is filled with media containing 50 ng/ml of Netrin-1. A gradient forms across the annular bridge due to diffusion of Netrin-1 from the outer to the inner well. (F) In the presence of DMSO, growing chick MNs did not change their trajectories in the Netrin gradient. (G) In the presence of γ-secretase inhibitor, motor axons turned toward increasing concentrations of Netrin. Note that all images have been rotated such that the gradient increases along the y-axis. Scale bar = 10 μm. (H) Definition of the initial angle, α, the angle between the initial axon position and the gradient; and angle turned, β, the angle between the vectors representing the initial and final position of the axon. (I) Scatter plot of the angle turned versus the initial angle for motor axons in the presence of DMSO or γ-secretase inhibitor L-685458. n = 41 (DMSO), n = 31 (γ-sec inh). (J) Histogram showing the mean angle turned (±SEM) for axons in the presence of DMSO or γ-secretase inhibitor L-685458. (* p < 0.05; Kolmogorov-Smirnov test).

Figure 5

DCC Stubs Cause Motor Neuron Chemoattraction to Netrin (A) In WT embryos, DCC is first cleavaged by a metalloprotease that leads to “shedding” of the ectodomain segment, generating membrane-tethered DCC stubs. DCC stub is subsequently processed by γ-secretase to generate DCC-intracellular domain (ICD). (B) In PS1 KO embryos, the production of DCC-ICD is disrupted, and DCC stubs accumulate to high levels on the cell membrane. (C) Western-blot analysis of DCC protein in mouse spinal cords. Protein extracts from the spinal cords of Columbus mutants (Col, lane 2), PS1 knockouts (KO, lane 4), or their control littermates (Ctrl, lane 1 and 3) were analyzed by immunoblotting with the DCC intracellular domain-specific antibody. High levels of DCC stub were apparent in Columbus and PS1 KO embryos. GAPDH was used as an internal standard. (D–G) Motor explant attraction assay. GFP-labeled mouse motor explants were cocultured with Netrin-1 cell aggregates in the presence of (D) DMSO vehicle, (E) metalloprotease inhibitor GM6001, (F) γ-secretase inhibitor L-685458, or (G) both GM6001 and L-685458. (H) P:D ratio of outgrowth from mouse motor explants in the presence of Netrin-1 cell aggregates. n=16 (DMSO), n=8 (MP inh), n=20 (γ-sec inh); n=19 explants (MP inh+γ-sec inh). Data are presented as the mean±SEM (* P<0.05). (I–M) Flat-mount images of chick spinal cords that have been electroporated with plasmids encoding myristylated DCC intracellular domain (DCC stub), DCC full-length (DCC-FL), DCC extracellular domain (DCC-ECD) or DCC intracellular domain (DCC-ICD) as indicated. Hb9-DsRed reporter (red) was co-electroporated to label MNs. n>8 embryos for each plasmid. Scale bar = 100 μm. (N–Q) Chick motor explant attraction assay. Chick spinal cords were electroporated with plasmids encoding myristylated DCC intracellular domain (DCC stub), DCC intracellular domain (DCC-ICD) or control plasmids as indicated. Hb9:DsRed reporter (red) was co-electroporated to label MNs. Motor explants were dissected from electroporated spinal cords and cocultured with the Netrin-1 cell aggregates in 3D Collagen/Matrigel matrices for 15 hours. Minimal axon outgrowth was observed from MNs in control conditions. (R) Histogram showing quantification (neurite numbers relative to control) of outgrowth from chick motor explants in the presence of Netrin-1 cell aggregates. n=16 (control), n=13 (DCC stub), n=9 (control+L-685458), n=14 (DCC-ICD+L-685458). Data are presented as the mean±SEM (* p<0.05).

Figure 6

Inhibition of Slit/Robo Signaling Induces Motor Axon Growth Toward Netrin-1. (A) Spinal cord diagram showing Slit/Robo-mediated silencing mechanism. (Step 1) Pre-crossing commissural axons (red) are attracted to the FP (grey triangle) by Netrin (+) receptor DCC; (Step 2) the midline Slit (−) activates Robo which blocks Netrin/DCC attraction. Slit is expressed in the motor column (green circle), leading to the possibility of a “self-silencing” mechanism that inhibits Netrin responsiveness in MNs. (B–E) Chick motor explant attraction assay. Chick spinal cords have been electroporated with plasmids encoding Robo1 extracellular domain (DN-Robo1), myristylated Robo1 intracellular domain (Myr-Robo1), or control plasmids as indicated. Hb9:DsRed (red) reporter was co-electroporated with above plasmids to label MNs. (F) Histogram showing quantification (neurite numbers relative to control) of outgrowth from chick motor explants in the presence of Netrin-1 cell aggregates. n=13 (control), n=8 (DN-Robo1), n=10 (control+Robo1-Fc), n=10 (Myr-Robo1+Robo1-Fc). Data are presented as the mean±SEM (* p<0.05). (G and H) Motor explant attraction assay. Mouse motor explants were cocultured with Netrin-1 cell aggregates in the presence of vehicle control or Robo1-Fc. (I) P:D ratio of outgrowth from motor explants in the presence of Netrin-1 cell aggregates. n=16 (BSA); n=19 (Robo1-Fc). Data are presented as the mean±SEM (* p<0.05). (J and K) Flat-mount images of E13.5 mouse spinal cords at lumbar levels with anterior on the left. Dotted line marks medial edge of MN cell bodies. At least five embryos were assayed for each genotype. See also Figure S3

Figure 7

DCC Stubs Do Not Interact with Robo (A) Interaction of Robo and DCC in 293-T cells cotransfected with plasmids encoding Robo1-Myc and DCC. Sixteen hours after transfection, cells were treated with vehicle DMSO or γ-secretase inhibitor for an additional ten hours, then incubated for 20 min with recombinant Netrin-1 and Slit2 and subjected to immunoprecipitation (IP) with antibodies to Myc or DCC extracellular domain (N-ter). Immunoprecipitates were analyzed by immunoblot using antibodies to Myc or DCC intracellular domain (C-ter). (B) Diagram showing the interaction of DCC and Robo receptor in the presence of DCC stubs. Activation of Robo by Slit leads to interaction of Robo with full-length DCC, DCC stubs are excluded from the hetero-receptor complex formed between DCC-FL and Robo, but the DCC stubs can associate with DCC-FL complex lacking Robo. (C) Immunostaining of TAG-1-positive commissural axons in transverse sections of E12.5 mouse spinal cords. TAG-1 staining in PS1 KO embryos was thicker at the midline (red bracket), and the ventral funiculus (yellow bracket) was absent compared to controls (WT or Het). At least 6 embryos were assayed for each genotype. Scale bar = 100 μm. (D) Model for PS1 function in axon navigation. (Box 1) In wild type spinal cords commissural axons (red) are initially attracted to the FP by Netrin until they reach the midline and encounter repulsive Slit ligands. (Box 2) Robo becomes activated triggering repulsion from the midline and silencing the attractive response toward Netrin through interaction with DCC. Thus, commissural neurons are first attracted to the FP, but then grow through the midline and enter the contralateral ventral funiculus. MNs (green) differ from commissural neurons in that they also express Slits. (Box 2) Slit/Robo interactions in MNs prevent these cells from acquiring responsiveness to FP derived Netrin. (Box 1 and 3) MNs may acquire responsiveness to Netrin later in development when their axons reach the periphery and Slit levels decline. Concomitantly, peripheral sources of Netrin could induce DCC stub production to overcome residual Slit/Robo silencing, thus triggering a response to Netrin. In PS1 mutants, commissural axons (red) fail to exit the FP and motor axons (green) misproject toward the midline due to abnormal attraction to Netrin. (Box 3) In the absence of PS1, the sequential cleavage of DCC is disrupted leading to the accumulation of DCC stubs on the membrane that are resistant to Slit/Robo silencing, thus triggering attraction to Netrin. See also Figure S4