L1-mediated branching is regulated by two ezrin-radixin-moesin (ERM)-binding sites, the RSLE region and a novel juxtamembrane ERM-binding region

Abstract

We investigated how the neural cell adhesion molecule L1 mediates neurite outgrowth through L1-L1 homophilic interactions. Wild-type L1 and L1 with mutations in the cytoplasmic domain (CD) were introduced into L1 knock-out neurons, and transfected neurons were grown on an L1 substrate. Neurite length and branching were compared between wild-type L1 and L1CD mutations. Surprisingly, the L1CD is not required for L1-mediated neurite outgrowth but plays a critical role in neurite branching, through both the juxtamembrane region and the RSLE region. We demonstrate that both regions serve as ezrin-moesin-radixin-binding sites. A truncation mutant that deletes 110 of 114 amino acids of the L1CD still supports neurite outgrowth on an L1 substrate, suggesting that a coreceptor binds to L1 in cis and mediates neurite outgrowth and that L1-ankyrin interactions are not essential for neurite initiation or outgrowth. These data are consistent with a model in which L1 can influence L1-mediated neurite outgrowth and branching through both the L1CD and a coreceptor.

Figure 1.

Cerebellar neurons growing on L1 are not highly polarized. Double labeling of wild-type neurons growing on L1 (2 DIV) with the axonal marker anti-tau1 (A) and the dendritic marker anti-MAP2 (B) reveals colocalization of tau1 and MAP2 (C). Scale bar, 100 μm.

Figure 2.

Schematic demonstration of the L1 intracellular mutations. Number of the amino acids in the L1CD of wild-type L1 (aa 1144-1257) is numbered by their position in the open reading frame of the human L1 gene. The juxtamembrane region, the YRSLE sequence, and the ankyrin-binding sites are highlighted. The numbers of the key residues are indicated on top or at the bottom of the corresponding residue. The underlined residues at the juxtamembrane region are predicted to be the ERM binding site by homology alignment to ICAM-2. The L1-1176 construct is truncated after the Y1176 residue. The L1-1147 construct is truncated after the K1147 residue. The L1-1180 residue is truncated after the E1180 residue. The L1ΔRSLE construct has an internal deletion from the R1177 to E1180 residue. The juxtamembrane mutants L1-4A and L1-1151Y>A change critical residues in the juxtamembrane region to alanine. The mutated residues are underlined.

Figure 3.

RSLE mutations support L1-mediated neurite outgrowth. L1KO neurons were transfected with wild-type L1 (A), L1ΔRSLE (B), L1-1176 (C), or L1-1180 (D) and plated on an L1 substrate. After 48 hr, neurons were fixed and stained with a monoclonal anti-human L1 antibody (7B5). Scale bar, 100 μm.

Figure 4.

The effects of L1 cytoplasmic domain mutations on L1-mediated neurite outgrowth. L1KO neurons were transfected with missense mutations and grown on an L1 substrate. Total neurite length, longest neurite length, and branching number were quantified. The mean ± SEM values of the mutant-transfected neurons were always normalized by the mean values of neurons transfected with WT hL1 in the same experiment. Values shown are the average of the mean ± SEM percentage values from three experiments. ANOVA (Fisher's PLSD) was done using Statview 4.5. Statistical significance is shown. ^***p < 0.001 in all three experiments; ^**p < 0.05 in all three experiments; ^*p < 0.05 in two of the three experiments.

Figure 5.

The effects of L1 cytoplasmic domain mutations on L1-mediated branching, primary neurites (branches extending from the soma), and nodes (branches arising from a neurite, not the soma). Primary neurites and nodes were quantified. The mean ± SEM values of the mutant-transfected neurons were always normalized by the mean values of neurons transfected with WT hL1 in the same experiment. Values shown are the average of the mean ± SEM percentage values from three experiments. ANOVA (Fisher's PLSD) was done using Statview 4.5. Statistical significance is shown. ^***p < 0.001 in all three experiments; ^**p < 0.05 in all three experiments.

Figure 6.

The L1CD is not necessary for L1-mediated neurite outgrowth, but the L1CD regulates branching through the juxtamembrane region. L1KO neurons were transfected with WTL1 (A), L1-1147 (B), L1-4A (C), or L1-1151Y>A (D) and plated on an L1 substrate. After 48 hr, neurons were fixed and stained with a monoclonal anti-human L1 antibody (7B5). Scale bar, 100 μm.

Figure 7.

Both the juxtamembrane region and the RSLE region are critical for the L1-ERM interaction. The pAS2 bait vectors containing wild-type and mutant forms of the L1CD fused to the Gal4 DNA-binding domain are cotransformed with the prey vector pACT2 containing ezrin (A), Drosophila ankyrin (B), or μ2 subunit of AP-2 (C) fused to the Gal4 activation domain and tested for interactions on histidine-deficient plates containing 10 mm 3-AT (A, B) or 5 mm 3-AT (C). Each plate includes one positive control p53+SV40, which is a diploid of AH109 [pGBKT7-53] (p53 was fused to GAL4 DNA-binding domain), and Y187 [pTD1-1] (SV40 large T antigen was fused to GAL4 activation domain), because the p53 and the SV40 large T antigen was known to interact in the yeast two-hybrid assay. A negative control was included on each plate, which is a diploid of AH109 [pAS2-L1CD] and Y187 [pTD1-1].