Interneurite affinity is regulated by heterophilic nectin interactions in concert with the cadherin machinery

Abstract

Neurites recognize their specific partners during the formation of interneuronal connections. In hippocampal pyramidal neurons, axons attach to dendrites for their synaptogenesis, but the dendrites do not form stable contacts with each other, suggesting the presence of a mechanism to allow their selective associations. Nectin-1 (N1), an immunoglobulin domain adhesive protein, is preferentially localized in axons, and its heterophilic partner, N3, is present in both axons and dendrites; we tested their potential roles in interneurite recognition. The overexpression of N1, causing its mislocalization to dendrites, induced atypical dendrodendritic as well as excessive axodendritic associations. On the contrary, the genetic deletion of N1 loosened the contacts between axons and dendritic spines. Those actions of nectins required cadherin-catenin activities, but the overexpression of cadherin itself could not accelerate neurite attachment. These results suggest that the axon-biased localization of N1 and its trans-interaction with N3 in cooperation with the cadherin machinery is critical for the ordered association of axons and dendrites.

Figure 1.

Differential distribution of nectins in neurites of the hippocampal neuron. (a and b) Double immunostaining for nectins and MAP2 (dendrite marker) or tau (axon marker) in isolated neurons at 5 DIV. Nectin-1 (N1) is more abundant in the axon than in dendrites (a). N3 is detectable throughout the neurites (b). Arrows point to branches of the axon. (c and d) Close-up views of early contacts between a dendritic filopodium and axon (indicated by arrowheads) that were triple stained for F-actin, β-catenin, and N1 (c) or N3 (d) at 6 DIV. Each nectin is sharply concentrated at the contact sites, colocalizing with β-catenin, and is absent from the free surfaces of the axon. Axons are indicated by the dotted lines, as the original actin stain was faint. Asterisks indicate dendrites. (e and f) N1 is not localized at the dendrodendritic crossing points (indicated by arrows). N3 is detectable around the dendrodendritic crossing points but is not particularly concentrated there, although this molecule is highly concentrated at early synapses present on the same dendrite (indicated by arrowheads). β-Catenin is localized at some of the dendrodendritic crossing points. Neurons were examined at 10 DIV. (g and h) Distribution of exogenous N1 (exN1) or exN3 in mature neurons at 21 DIV. In close-up views of their dendrites, exN1 is detected on fibers running on the dendritic shaft (g); the majority of spines on the same dendrite, detected by F-actin staining, do not have exN1, some of which are indicated by arrowheads. On the contrary, exN3 is evenly detected on all spines as well as on the shaft portion of the dendrite (h). See Fig. S1 for lower magnification views (available at http://www.jcb.org/cgi/content/full/jcb.200601089/DC1). (i) Relative intensity of immunofluorescence signals emanating from endogenous (endo) or exogenous (ex) nectins on the axons (ax) or dendrites (dd; mean ± SEM [error bars]). **, P < 0.001 versus the axon; n = 16 for dendrites, and n = 6 for axons. AU, arbitrary unit. Bars (a and b), 10 μm; (c and d) 2 μm; (e–h) 5 μm.

Figure 2.

Effects of the overexpression of nectins on neurite patterning. (a–c) Neurons transfected with N1 (a) or N3 (b) and nontransfected (c) at 6 DIV. Cultures were triple stained for nectin, MAP2, and tau. In N1 transfectants (a), the relative level of N1 in dendrites is increased, MAP2-positive dendrites aberrantly touch each other (indicated by arrowheads), and tau-positive axons have become irregularly entangled around their own dendrites. In N3 transfectants (b), dendrites extend radially, as seen in the nontransfected control (c), and axon extension is not disturbed by the dendrites of the same neuron. Arrows point to axon branches in contact with dendrites. The reactivity of anti-MAP2 antibodies toward axons tended to increase in nectin transfectants for some unknown reason. (d and e) Neurons transfected with N1 or N3 and double stained for nectin and MAP2 at 14 DIV. Dendrodendritic attachment inducing their looping appearance occurs extensively in N1 transfectants (d, arrowheads) but much less in N3 ones (e). (f and g) Close-up views of dendrodendritic contacts under nectin overexpression. Cells were quadruple stained for MAP2, exogenous (ex) and endogenous (en) nectins, and F-actin at 5 DIV. In N1 transfectants (f), exN1 and enN3 are concentrated together at dendrodendritic contact sites, as indicated by arrowheads. In N3 transfectants (g), such concentration does not occur even when dendrodendritic contacts are formed (arrowheads). In these dendrites, exN3 and endoN1 colocalize at noncontact portions. Bars (a–e), 20 μm; (f and g) 5 μm.

Figure 3.

Effects of the expression of chimeric nectin molecules. (a) Diagram of nectin constructs used. TM, trans-membrane region. The N1-derived regions are light blue, and N3-derived regions are pink. + indicates where the two regions are fused. (b and c) Neurons transfected with N13 or N31 and triple stained for the chimeric nectin MAP2 and tau at 5 DIV. In N13 transfectants (b), N13 molecules are clustered in various regions, and dendrites and axons are strongly entangled. In N31 transfectants (c), N31 molecules are localized most abundantly in axons, and the atypical association of neurites is less extensive than in the case of N13 transfectants. (d and e) Neurons transfected with N13 or N31 and double stained for the chimeric nectin and MAP2 at 14 DIV. N13 strongly induces dendrodendritic attachments (d), and N31 only induces these weakly (e). (f) Statistical analysis of dendritic arbor pattern. Number of dendrites, dendrite length, and dendrite branch number in neurons nontransfected (Ctrl) or transfected with N1, N13, N3, or N31 were measured at 7 DIV. Histogram shows the mean plus SEM (error bars) for each sample (n = 20 for dendrite and branch number; n = 40 for dendrite length). No significant difference was found between these samples. For the right histogram, n = 20. **, P < 0.001 versus control, N3, and N31. The circle-crossing index represents the mean number of dendrites that cross the circle (40 μm in diameter) superimposed on the soma of each neuron. This index is expected to increase when dendrites turn and form loops as a result of dendrodendritic attachments. Bars, 20 μm.

Figure 4.

Nectin interactions at neurite contact sites. (a) Dendrodendritic associations observed between different neurons and induced by nectin overexpression. Neurons were independently transfected with N1 or N3, and these cells were mixed at a 1:1 ratio, cultured for 14 d, and triple stained for exN1, exN3, and MAP2. In the pair of neurons situated next to each other (one overexpressing exN1 [single asterisk] and the other overexpressing exN3 [double asterisks]), their dendrites have almost entirely intermingled. The nontransfected neuron located at the bottom left is less extensively associated with them. Neurons with exN1 or exN3 were identified by the abundance of the respective molecules at the cell body regions. (b) Heterophilic nectin interaction at axodendritic interfaces. An axon extending from a remote neuron with exN1, located outside at the top left corner, has attached to dendrites of a neuron with exN3, which is visualized by MAP2 staining, in the culture prepared as in panel a. At their contact sites, exN3 is exclusively concentrated along the exN1-positive axon. Faint fluorescence on the neuronal body in the exN1 panel is likely a result of the nonspecific reaction of the antibodies. (c–f) Interaction of nectins at the interfaces between neurites and 293 cells. Neurons were plated onto mixed cultures of 293 cells nontransfected or transfected with N1 (c and d) or N3 (e and f), incubated for 5 d, and double stained for exN1 or exN3 and MAP2 or tau. Dendrites have strongly recruited N1 molecules derived from 293 transfectants to their contact sites (c) but have N3 ones only weakly (e). Axons recruited these nectins indiscriminately (d and f). Bars (a and b), 20 μm; (c–f) 10 μm.

Figure 5.

Effects of N1 deficiency on axodendritic interactions. (a) Hippocampal neurons obtained from wild-type (WT) and N1-deficient (knockout; KO) brains stained for F-actin at 14 DIV. Note the elongated morphology of the mutant dendritic spines. (b) Close-up views of a wild-type (top) or N1-deficient (bottom) neuron at 14 DIV double stained for F-actin and synaptotagmin (syn). Arrows point to axons, which are identified by their association with spine heads or reactivity to antisynaptotagmin antibodies. In the mutant sample, although some spines are in contact with axons, others (indicated by arrowheads) appear to be free from the axons. (c–f) Synaptotagmin and β-catenin localization in wild-type (c and d) and N1-deficient (e and f) synapses at 21 DIV. These proteins are seen at synaptic sites in both samples, but the β-catenin condensation has decreased, and synaptotagmin puncta have become reduced in size and have even lost from some of the spine heads (the left two arrowheads in panel e) in the mutant samples. Arrowheads point to representative spine heads. (g) Statistical analysis of dendritic spine morphology. Spine length and spine head width were significantly changed in N1 mutants (mean ± SEM [error bars]; n = 50 for each sample). *, P < 0.05 versus N1 ^+/+; **, P < 0.005 versus N1 ^+/+. The data were collected from 10 neurons in two independently prepared cultures at 17 DIV. Bars (a), 20 μm; (b) 5 μm; (c–f) 2 μm.

Figure 6.

Cooperative action of nectins and N-cadherin/catenins. (a) A mixed culture of neurons transfected with N1 or N3 prepared as in Fig. 4 B and triple stained for β-catenin, exN3, and exN1 at 10 DIV. Axons expressing exN3 have migrated on a neuron with exN1, and β-catenin is concentrated together with N3 and N1 along the axons. (b) A mixed culture of 293 cells transfected with N1 or N3 triple stained for exN1, exN3, and β-catenin. Each transfectant was identified by comparing the localization of these two molecules and are marked as 1 (for N1) or 3 (for N3). A single N3 transfectant is surrounded by multiple N1 transfectants. (c and d) Neurons nontransfected (c) or transfected with Flag-tagged N-cadherin cDNA (d; Nakagawa and Takeichi, 1998) were cultured for 7 d and double stained for MAP2 and the Flag tag. (e and f) Hippocampal neurons derived from wild-type (e) or αN-catenin knockout (KO) mice (f) were transfected with N1 and double stained for MAP2 and exN1 at 8 DIV. (g) Statistical analysis of the experiments in panels c and d. n = 20 for dendrite number and dendrite branch number; n = 40 for dendrite length. (h) Immunoblots for N-cadherin expressed in nontransfected (Ctrl) and N-cadherin–transfected (N-cad) cultures at 8 DIV. EGFP-tagged N-cadherin (asterisk; Horikawa and Takeichi, 2001) was used for transfection. The amount of the exogenous N-cadherin is about equal to or slightly lower than that of the endogenous one. As ∼50% of cells were transfected, we can estimate that the level of total N-cadherin per neuron increased two to three times in the transfected neurons. (i) Statistical analysis of the experiments in panels e and f. Error bars represent SEM. **, P < 0.001 versus control. n = 20 for dendrite number and dendrite branch number; n = 40 for dendrite length. 20 neurons were used for this assay. Bars (a–d), 20 μm; (e and f) 10 μm.

Figure 7.

Working models to explain the role of nectins in neurite interactions. (a) In wild-type neurons, N1 abundant in the axon interacts with N3 in the dendrite, and this trans-heterophilic interaction of N1 and N3 promotes homophilic cadherin–cadherin interactions to strengthen synaptic junctions. In the absence of N1, only a basic level of cadherin interactions would take place. Homophilic interactions between N3 and N3 would not be strong enough to sustain normal axodendritic contacts. Dotted arrows indicate possible weaker interactions between nectins. The axonal N3 level appears to decrease with the maturation of neurons. (b) Dendrodendritic interactions are not stable because the N1 level in dendrites is low and N3–N3 interaction is not strong enough. However, when N1 is overexpressed, the misexpressed N1 molecules not only induce atypical dendrodendritic adhesions but also overstabilize axodendritic contacts. KO, knockout.