Novel phospho-switch function of delta-catenin in dendrite development

Abstract

In neurons, dendrites form the major sites of information receipt and integration. It is thus vital that, during development, the dendritic arbor is adequately formed to enable proper neural circuit formation and function. While several known processes shape the arbor, little is known of those that govern dendrite branching versus extension. Here, we report a new mechanism instructing dendrites to branch versus extend. In it, glutamate signaling activates mGluR5 receptors to promote Ckd5-mediated phosphorylation of the C-terminal PDZ-binding motif of delta-catenin. The phosphorylation state of this motif determines delta-catenin's ability to bind either Pdlim5 or Magi1. Whereas the delta:Pdlim5 complex enhances dendrite branching at the expense of elongation, the delta:Magi1 complex instead promotes lengthening. Our data suggest that these complexes affect dendrite development by differentially regulating the small-GTPase RhoA and actin-associated protein Cortactin. We thus reveal a "phospho-switch" within delta-catenin, subject to a glutamate-mediated signaling pathway, that assists in balancing the branching versus extension of dendrites during neural development.

Figure 1.

Illustration of domains within delta-catenin, Magi1, and Pdlim5. (A) Schematics of the proteins delta-catenin, p120-catenin, Pdlim5, and Magi1. (B) Conservation of the C-terminus of delta-catenin across species.

Figure 2.

Delta-catenin plays a major role in the establishment of hippocampal neuron dendrite morphology. (A) Representative images of 7-DIV rat hippocampal neurons expressing GFP, delta-catenin cDNA, and delta-catenin shRNA. (B) Quantification of average dendrite length of neurons expressing GFP (35.96 ± 1.56 µm), delta-catenin cDNA (49.14 ± 1.72 µm; P < 0.0001), and delta-catenin shRNA (24.98 ± 1.13 µm; P < 0.0001). (C) Average dendrite density of neurons expressing GFP (16.69 ± 1.26), delta-catenin cDNA (30.90 ± 2.03; P < 0.0001), and delta-catenin shRNA (14.56 ± 1.43; P < 0.0001). (D) Sholl analysis of dendrite morphology for control, delta-catenin overexpressing, and delta-catenin shRNA-expressing neurons. Relative to GFP-expressing cells, neurons overexpressing delta-catenin possessed significantly more dendrites at 25–70 µm from the soma (P ≤ 0.0001) and from 70 to 110 µm from the soma (P < 0.05). ***, P ≤ 0.0001. Error bars indicate SEM. For B and C, n ≥ 12 neurons; for D, n = 6. For B and C, significance was determined using a one-way ANOVA followed by Tukey’s test. For D, a two-way ANOVA with Bonferroni post-hoc analysis was used. Scale bars, 20 µm.

Figure 3.

Point mutants that mimic phosphorylation (versus lack thereof) within the PDZ-binding motif of delta-catenin suggest a role of this modification in directing dendritic morphology. (A) Representative images of 7-DIV rat hippocampal neurons transfected with GFP (control), delta-catenin cDNA, delta-catenin-EE (phospho-mimic) cDNA, and delta-catenin-AA (phospho-null/P.Null) cDNA. OE, overexpression. (B) Quantification of average dendrite length of neurons expressing GFP (35.96 ± 1.56 µm), delta-catenin cDNA (49.14 ± 1.72 µm; P < 0.0001), delta-catenin-EE (phospho-mimic) cDNA (33.59 ± 0.90 µm; P = 0.601), and delta-catenin-AA (phospho-null) cDNA (61.50 ± 2.286 µm; P < 0.0001). (C) Average dendrite density of neurons expressing GFP (16.69 ± 1.26), delta-catenin cDNA (30.90 ± 2.03; P < 0.0001), delta-catenin-EE (phospho-mimic) cDNA (34.00 ± 2.65; P < 0.0001), and delta-catenin-AA (phospho-null) cDNA (20.00 ± 1.96; P = 0.601). (D) Sholl analysis of dendrite morphology for control, delta-catenin-overexpressing, phospho-mimic delta-catenin–overexpressing, and phospho-null delta-catenin–overexpressing neurons. Relative to GFP-expressing cells, neurons overexpressing phospho-mimic delta-catenin possessed significantly more dendrites at 15–35 µm from the soma (P < 0.05). Conversely, neurons overexpressing phospho-null delta-catenin possessed significantly more dendrites at 60–80 µm (P < 0.05) relative to GFP controls. ***, P ≤ 0.0001. Error bars indicate SEM. For all conditions in A–C, n ≥ 12 neurons; in D, n = 6. For B and C, significance was determined using a one-way ANOVA followed by Tukey’s test. For D, a two-way ANOVA with Bonferroni post-hoc analysis was used. Scale bars, 20 µm.

Figure S1.

Confocal images of dendrites of rat hippocampal neurons expressing HA-epitope–tagged phospho-mimic delta-catenin (delta-EE; duplicate left panels) versus HA-epitope–tagged phospho-null delta-catenin (delta-AA; duplicate right panels). To facilitate determination of their subcellular localization, neurons were intentionally selected that exhibited low-level (faint) expression of either construct. At this resolution, fluorescent imaging of either mutant revealed no consistent/quantifiable difference in their localization to dendrite tips (white arrows) versus branch points (blue arrows). This is consistent with our hypothesis that predistributed delta-catenin along dendrites becomes transiently phosphorylated in response to local mGluR5 activation, promoting complex formation with Pdlim5 and branching (e.g., involving greater inhibition of RhoA), rather than delta-catenin becoming localized/translocated to the growing or future branch points only after the phosphorylation of its C-terminal PDZ-binding motif. Scale bars, 10 µm.

Figure 4.

Glutamate signaling is sufficient to drive phosphorylation of delta-catenin and influence its role in GTPase modulation and dendrite development. (A) Validation of specificity of detection method for phosphorylated delta-catenin. Immunoblot images of HEK293 cells transfected with WT delta-catenin and delta-catenin-AA (phospho-null) cDNA. (B) Immunoblots of total delta-catenin and immunoprecipitated phospho-delta-catenin from 7-DIV rat hippocampal cells. We estimate that phospho-delta represents ∼3% of total delta-catenin in neurons. (C) Immunoblots of immunoprecipitated phospho-delta-catenin from 7-DIV rat hippocampal cells following treatment with DHPG (20 µM; 10 min) versus no treatment. (D) Immunoblots of immunoprecipitated phospho-delta-catenin from 7-DIV rat hippocampal cells following treatment with MTEP (25 µM; 10 min) versus no treatment. (E) Quantification of fold change in delta-catenin phosphorylation following DHPG treatment in 7-DIV rat hippocampal cells. Treatment of 7-DIV cultures with 20 µM DHPG for 10 min produced an 8.202-fold increase in relative levels of phosphorylated delta-catenin compared with controls (P = 0.0373); graph represents cultures from four independent experiments. (F) Quantification of fold change in delta-catenin phosphorylation following MTEP treatment in 7-DIV rat hippocampal cells. Treatment of 7-DIV cultures with 25 µM MTEP for 10 min resulted in a 42% reduction in relative levels of phosphorylated delta-catenin compared with controls (P = 0.0069); graph represents cultures from two independent experiments. (G) Representative images of 7-DIV rat hippocampal neurons transfected with GFP and subjected to no treatment or 20 µM DHPG for 6 h. For each condition image, representative close-ups of dendritic branches are shown. (H) Quantification of dendritic density following DHPG treatment (23.92 ± 1.06; P = 0.0002). n ≥ 12 neurons. (I) Sholl analysis of dendrite morphology for control and DHPG-treated neurons. Neurons treated with DHPG developed significantly more dendrites than GFP controls at 20–35 µm from the soma (P < 0.05); n = 6. (J) Representative images of 7-DIV rat hippocampal neurons expressing delta-catenin shRNA and subjected to no treatment or 20 µM DHPG for 6 h. For each condition image, representative close-ups of dendritic branches are shown. (K) Quantification of dendritic density in neurons expressing delta-catenin shRNA following DHPG treatment (15.75 ± 1.18; P = 0.6422). n ≥ 9 neurons. (L) Sholl analysis of dendrite morphology for control and DHPG-treated neurons expressing delta-catenin shRNA. DHPG treatment did not significantly alter dendritic morphology in neurons lacking delta-catenin; n = 6. For E, F, H, and K, significance was determined with a two-tailed t test. For I and L, a two-way ANOVA with Bonferroni post-hoc analysis was used. *, P < 0.05; ***, P ≤ 0.0001. Error bars indicate SEM. Scale bars, 20 µm in G and J and 5 µm in their respective magnified insets. IP, immunoprecipitation. WCE, whole cell extract.

Figure 5.

Delta-catenin interacts with Magi1 and Pdlim5 in a phospho-dependent manner. (A) Schematic of a protein-domain microarray. (B) Phosphorylation of the PDZ-binding motif of delta-catenin modulates the protein’s ability to interact with several partner proteins. The PDZ5 domain of Magi1 is bound by the unphosphorylated motif but not by the phosphorylated motif (blue). The sole PDZ domain within Pdlim5 is bound by the phosphorylated but not unphosphorylated delta-catenin PDZ-binding motif (red). (C) Quantification of microarray data represented in B. (D) Phosphorylation at either serine in the PDZ-binding motif of delta-catenin is sufficient to allow interaction with the PDZ5 domain of Pdlim5. Phosphorylation at both sites results in a more efficacious binding of Pdlim5 by delta-catenin. (E and F) Test of the phospho-dependency of the Magi1:delta-catenin versus Pdlim5:delta-catenin interactions, using a GRA we developed. Mutant delta-catenin constructs (red) with phospho-null properties (S1242A, S1245A), or conversely with phospho-mimic properties (S1242E, S1245E), were expressed in HEK293 cells and relocalized to the Golgi body. (E) Magi1 (green) co-relocalizes to the Golgi in cells expressing phospho-null delta-catenin but not phospho-mimic delta-catenin (red). (F) Pdlim5 (green) co-relocalizes to the Golgi in cells expressing phospho-mimic delta-catenin but not with phospho-null delta-catenin (red). (G) Quantification of delta-catenin localization in GRA. (H) Quantification of Magi1/Pdlim5 localization in GRA. Graphs represent cultures from at least two independent experiments. ***, P ≤ 0.0001. Box plots indicate upper and lower quartiles (top/bottom of box), median value (center line of box), and the maximum/minimum values (whiskers). For G and H, significance was determined using a one-way ANOVA followed by Tukey’s test. Scale bars, 10 µm.

Figure S2.

Delta-catenin interacts with specific mapped domains of Magi1 and Pdlim5. Through the application of our GRA in HEK293 cells, we validated our protein-domain microarray via domain mapping of Pdlim5 and Magi1. (A–D) Mapping of PDZ domain(s) of Magi1 required for interaction with delta-catenin. (A) Full-length Magi1 (green) co-relocalizes with delta-catenin (red) to Golgi body. (B) Deletion of PDZ4 and PDZ5 of Magi1 renders the protein unable to interact with delta-catenin, thereby preventing co-relocalization of Magi1 with delta-catenin (red) to the Golgi body. (C) Deletion of only PDZ5 of Magi1 also renders Magi1 (green) unable to co-relocalize to the Golgi, suggesting that PDZ5 is required for Magi1’s interaction with delta-catenin. (D) Expression of only PDZ4 and PDZ5 of Magi1 (green) successfully co-relocalizes to the Golgi with delta-catenin (red). (E and F) Validation and mapping of Pdlim5:delta-catenin interaction using a GRA in HEK293 cells. (E) Full-length Pdlim5 co-relocalizes with delta-catenin to Golgi body. (F) Deletion of the PDZ domain in Pdlim5 renders the protein unable to interact and co-relocalize with delta-catenin. (G) Quantification of delta-catenin localization in GRA. (H) Quantification of Magi1/Pdlim5 localization in GRA. Graphs represent cultures from at least two independent experiments. ***, P ≤ 0.0001. Box plots indicate upper and lower quartiles (top/bottom of box), median value (center line of box), and the maximum/minimum values (whiskers). For G and H, significance was determined using a one-way ANOVA followed by Tukey’s test. Scale bars, 10 µm.

Figure 6.

Delta-catenin interacts with Magi1 and Pdlim5 in developing hippocampal neurons. (A) Expression of endogenous delta-catenin (green), Magi1 (red, top), and Pdlim5 (red, bottom) in developing primary rat hippocampal neurons at 1 DIV, 7 DIV, and 28 DIV. Scale bars, 20 µm. (B and C) Detection of direct associations between endogenous delta-catenin and the proteins Magi1 and Pdlim5 in developing hippocampal neurons. A PLA (Duolink), in combination with delta-catenin, Magi1, and Pdlim5 antibodies, was used to detect direct endogenous protein–protein interactions in 7-DIV hippocampal neurons. (B) Visualization of Magi1:delta-catenin (left) and Pdlim5:delta-catenin (center) interactions, as detected by the presence of red puncta. Insets show presence of puncta on/near small protrusions from the cell in several cases. Antibodies specific to delta-catenin and c-Jun were used in a negative control assay (right). Scale bars, 20 µm in main panels and 5 µm in zoomed insets. (C) Quantification of average puncta per neuron in each condition (delta-catenin+ Magi1: 16.01 [P = 0.0159]; Pdlim5: 37.73 [P = 0.0007]; c-Jun/Control: 2.393); n = 3 independent experiments. (D) Quantification of puncta/cell detected by a PLA (Duolink) in 7-DIV rat hippocampal neurons treated with DHPG. Cells were fixed immediately following treatment (no treatment versus 20 µM DHPG; 10 min) and subjected to the assay. DHPG treatment produced an increase in Pdlim5:delta-catenin puncta relative to no treatment (54.20 ± 3.24 vs. 20.46 ± 1.41; P = 0.0007), while simultaneously reducing the amount of Magi1:delta-catenin complex detected per cell relative to controls (8.95 ± 1.59 versus 21.15 ± 7.24; P = 0.0473); n = 3 independent experiments. (E) Super-resolution confocal image of delta-catenin (red) and Magi1 (green) expression in 4-DIV primary rat hippocampal neurons. White/yellow arrows indicate Magi1 puncta near dendritic tips. Both right-hand panels are three-dimensional projection (x-y-z) views of the region outlined in the left-hand panels. (F) Super-resolution confocal image of delta-catenin (red) and Pdlim5 (green) expression in 4-DIV primary rat hippocampal neurons. Right-hand panels are x-y-z views of the region outlined in the left-hand panels. For E and F, top left panels are epi-fluorescence images of the cell/region subjected to STORM imaging. For C, significance was determined using a one-way ANOVA followed by Tukey’s test. For D, a two-tailed t test was used. *, P < 0.05; ***, P ≤ 0.0001. Error bars represent SEM.

Figure S3.

shRNA-mediated KD of delta-catenin, Magi1, and Pdlim5. (A) Immunoblots of delta-catenin, Magi1, and Pdlim5 from hippocampal neuron culture cell lysates from 1 DIV to 28 DIV. (B) Immunoblots of cell lysates from HEK293 cells coexpressing HA-delta-catenin, FLAG-Magi1, or Myc-Pdlim5 with either a control vector or the appropriate shRNA. We find that delta-catenin shRNA results in a near-complete loss of delta-catenin (∼100% KD), while the Magi1 (∼68% KD) and Pdlim5 (∼60% KD) shRNAs produce slightly less complete KDs of their respective proteins. (C–E) Analysis of neurons expressing Magi1 shRNA compared with rescues (neurons expressing shRNA-KD and cDNA-overexpression constructs). (C) Expression of Flag-tagged Magi1 in Magi1 shRNA-expressing neurons resulted in a rescue of the shRNA dendrite length phenotype (control: 35.96 ± 1.56 µm; shRNA: 21.72 ± 0.98 µm; rescue: 33.94 ± 1.44 µm, P < 0.0001). (D) Expression of Magi1 cDNA in combination with Magi1 shRNA did not produce a significant impact on dendrite density (control: 16.69 ± 1.26; shRNA: 17.92 ± 2.01; rescue: 18.22 ± 0.85 µm, P = 0.7911). (E) Sholl analysis of Magi1 shRNA and rescue neurons reveals that rescue of the shRNA via expression of Magi1 results in a dendritic arborization pattern similar to that of controls. (F–H) Analysis of neurons expressing Pdlim5 shRNA compared with rescues (neurons expressing shRNA-KD and cDNA-overexpression constructs). (F) Expression of Myc-tagged Pdlim5 in Pdlim5 shRNA–expressing neurons resulted in a rescue of the shRNA dendrite length phenotype (control: 35.96 ± 1.56 µm; shRNA: 53.25 ± 2.99 µm; rescue: 32.83 ± 1.57 µm, P < 0.0001). (G) Expression of Pdlim5 cDNA in combination with Pdlim5 shRNA did not produce a significant impact on dendrite density (control: 16.69 ± 1.26; shRNA: 16.33 ± 1.61; rescue: 19.00 ± 1.41, P = 0.5931). (H) Sholl analysis of Pdlim5 shRNA and rescue neurons reveals that rescue of the shRNA via expression of Pdlim5 results in a dendritic arborization pattern similar to that of controls. For C, D, F, and G, n ≥ 9 neurons. For E and H, n = 6. ***, P ≤ 0.0001. Error bars indicate SEM. For C, D, F, and G, significance was determined using a one-way ANOVA followed by Tukey’s test. For E and H, a two-way ANOVA with Bonferroni post-hoc analysis was used.

Figure 7.

Magi1 and Pdlim5 have complementary roles in dendritic branching and elongation during neuronal development. (A) Representative images of 7-DIV rat hippocampal neurons transfected with GFP (control), Magi1 cDNA, or Magi1 shRNA. (B) Quantification of average dendrite length of neurons expressing GFP (35.96 ± 1.56 µm), Magi1 cDNA (60.98 ± 2.69 µm; P < 0.0001), and Magi1 shRNA (21.72 ± 0.98; P < 0.0001). (C) Average dendrite density of neurons expressing GFP (16.69 ± 1.26), Magi1 cDNA (20.5 ± 1.04; P = 0.132), and Magi1 shRNA (17.92 ± 2.01; P = 0.785). (D) Sholl analysis of dendrite morphology for control, Magi1-overexpressing, and Magi1 shRNA-expressing neurons. Relative to GFP-expressing cells, neurons overexpressing Magi1 possessed significantly more dendrites at 35–130 µm from the soma (P < 0.05), indicating the presence of longer dendrites. Conversely, neurons subjected to shRNA-mediated KD of Magi1 developed significantly fewer dendrites at 15 µm and 30–40 µm (P < 0.05). (E) Representative images of 7-DIV rat hippocampal neurons transfected with GFP (control), Pdlim5 cDNA, or Pdlim5 shRNA. (F) Quantification of average dendrite length of neurons expressing GFP (35.96 ± 1.56 µm), Pdlim5 cDNA (35.69 ± 1.01 µm; P = 0.9916), and Pdlim5 shRNA (53.25 ± 2.99 µm; P < 0.0001). (G) Average dendrite density of neurons expressing GFP (16.69 ± 1.26), Pdlim5 cDNA (38.13 ± 3.06; P < 0.0001), and Pdlim5 shRNA (16.33 ± 1.61; P < 0.0001). (H) Sholl analysis of dendrite morphology for control, Pdlim5-overexpressing, and Pdlim5 shRNA–expressing neurons. Relative to GFP-expressing cells, neurons overexpressing Pdlim5 possessed significantly more dendrites at 15–40 µm from the soma (P ≤ 0.0001). Conversely, neurons subjected to shRNA-mediated KD of Pdlim5 developed significantly more dendrites at 35–95 µm (P < 0.05), indicating the presence of longer dendrites. ***, P ≤ 0.0001. Error bars indicate SEM. For all conditions in A–C and E–G, n ≥ 12 neurons; for D and H, n = 6. For B, C, F, and G, significance was determined using a one-way ANOVA followed by Tukey’s test. For D and H, a two-way ANOVA with Bonferroni post-hoc analysis was used. Scale bars, 20 µm. OE, overexpression.

Figure 8.

Influence of Magi1 and Pdlim5 on dendrite development weakens in absence of interaction with delta-catenin. (A) Amino protein sequences of human Magi1 PDZ5 and mouse Pdlim5 PDZ. Residues mutated to prevent interaction with delta-catenin are highlighted in red (Magi1: H62A/I66Y; Pdlim5: H63A/Q67Y). (B) Amino acid schematic of a general PDZ-binding motif and delta-catenin’s PDZ binding motif, showing serines at −3 and −6 locations. (C and D) Models of the Magi1:delta-catenin (C) and Pdlim5:delta-catenin (D) interactions. (E and F) Validation of lack of interaction between mutant Pdlim5 and Magi1 with delta-catenin via a GRA in HEK293 cells. (G) Quantification of delta-catenin localization in GRA. (H) Quantification of Magi1/Pdlim5 localization in GRA. Graphs represent cultures from at least two independent experiments. (I) Representative images of 7-DIV primary rat hippocampal neurons expressing Pdlim5 cDNA, Pdlim5-H63A/Q67Y cDNA, Magi1 cDNA, and Magi1-H62A/I66Y cDNA. (J) Quantification of average dendrite length of neurons expressing GFP (35.96 ± 1.56 µm), Pdlim5 cDNA (35.69 ± 1.01 µm; P = 0.9916), Pdlim5-H63A/Q67Y cDNA (38.63 ± 1.85 µm; P = 0.873), Magi1 cDNA (60.98 ± 2.69 µm; P < 0.0001), and Magi1-H62A/I66Y cDNA (36.97 ± 1.86 µm; P = 0.997). Overexpression of Magi1-H62A/I66Y failed to induce the dendrite extension that results from native Magi1 overexpression (P ≤ 0.0001). (K) Average dendrite density of neurons expressing GFP (16.69 ± 1.26), Pdlim5 cDNA (38.13 ± 3.06; P < 0.0001), Pdlim5-H63A/Q67Y cDNA (25.70 ± 2.32; P = 0.0491), Magi1 cDNA (20.50 ± 1.04; P = 0.0.690), and Magi1-H62A/I66Y cDNA (19.4 ± 2.06; P = 0.914). Neurons overexpressing Pdlim5-H63A/Q67Y failed to increase dendrite density to the extent of native Pdlim5 overexpression (P = 0.0016). (L) Sholl analysis of dendrite morphology for GFP control, Pdlim5-overexpressing, and Pdlim5 mutant (H63A/Q67Y)–overexpressing neurons. Relative to GFP-expressing cells, neurons overexpressing Pdlim5-H63A/Q67Y possessed significantly more dendrites at 40 µm (P = 0.0315) and 50 µm from the soma (P = 0.0333). Relative to neurons overexpressing native Pdlim5, neurons overexpressing Pdlim5-H63A/Q67Y developed significantly fewer dendrites at 15–25 µm from the soma (P < 0.05). (M) Sholl analysis of dendrite morphology for GFP control, Magi1-overexpressing, and Magi1 mutant (H62A/I66Y)–overexpressing neurons. Relative to overexpression of native Magi1, neurons overexpressing Magi1-H62A/I66Y possessed significantly fewer dendrites at 40 µm (P = 0.0090) and 60 µm from the soma (P = 0.0357). Overexpression of Magi1-H62A/I66Y produced no changes in dendrite length/density relative to GFP controls. *, P < 0.05; ***, P ≤ 0.0001. For G and H, box plots indicate upper and lower quartiles (top/bottom of box), median value (center line of box), and the maximum/minimum values (whiskers). For J–M, error bars indicate SEM. For all conditions in I–K, n ≥ 12 neurons; in L and M, n = 6. For G, H, J, and K, significance was determined using a one-way ANOVA followed by Tukey’s test. For L and M, a two-way ANOVA with Bonferroni post-hoc analysis was used. For E and F, scale bars, 10 µm. For I, scale bars, 20 µm. OE, overexpression.

Figure S4.

Magi1 and Pdlim5 require delta-catenin to significantly affect dendrite morphology. (A) Representative images of 7-DIV primary rat hippocampal neurons expressing delta-catenin shRNA alone and in combination with Magi1 or Pdlim5. (B–D) Analysis of neurons expressing delta-catenin shRNA alone compared with neurons expressing a combination of delta-catenin shRNA and either Magi1 or Pdlim5. (B) Expression of exogenous Magi1 or Pdlim5 in delta-catenin shRNA–expressing neurons fails to generate any impact on dendrite length (delta-cat. shRNA: 24.98 ± 1.13 µm; delta-cat. shRNA + Magi1: 23.01 ± 1.38 µm, P = 0.4753; delta-cat. shRNA + Pdlim5: 25.87 ± 1.03 µm, P = 0.8435). (C) Expression of exogenous Magi1 or Pdlim5 in delta-catenin shRNA–expressing neurons fails to generate any impact on dendrite density (delta-cat. shRNA: 14.56 ± 1.43; delta-cat. shRNA + Magi1: 13.78 ± 1.99, P = 0.9398; delta-cat. shRNA + Pdlim5: 16.00 ± 1.64, P = 0.8127); n ≥ 9 neurons. (D) Sholl analysis of neurons expressing delta-catenin shRNA, delta-cat. shRNA + Magi1, and delta-cat. shRNA + Pdlim5 reveals that expression of either Magi1 or Pdlim5 in delta-catenin KD neurons generates very little impact on the morphology of the dendritic tree; n = 6. Error bars indicate SEM. For B and C, significance was determined using a one-way ANOVA followed by Tukey’s test. For D, a two-way ANOVA with Bonferroni post-hoc analysis was used. Scale bars, 20 µm.

Figure 9.

Identification of the modulatory kinase and downstream effectors associated with phosphorylation of delta-catenin’s PDZ-binding motif. (A) Representation of fold change in levels of phosphorylated delta-catenin relative to total delta-catenin in response to exogenous coexpression of the indicated kinases in HEK293 cells. Coexpression of constitutively active CDK5 and p25c with delta-catenin resulted in a 15-fold increase in the relative amount of phosphorylated delta-catenin in cell lysates (P = 0.0113); n = 3 independent cultures/experiments. (B) Quantification of fold change in relative levels of phosphorylated delta-catenin in primary rat hippocampal neurons (7 DIV) following DHPG treatment with and without pretreatment with Butyrolactone I. Pretreatment of cultures with Butyrolactone I largely prevented the DHPG-induced increase (P = 0.0404) in delta-catenin phosphorylation levels relative to controls (P = 0.5205); n = 3 independent cultures/experiments. (C) Quantification of levels of active RhoA in HEK293 cells expressing WT delta-catenin (baseline), phospho-null delta-catenin (102% of baseline; P = 0.8814), phospho-mimic delta-catenin (85% of baseline; P = 0.0332), Magi1 + delta-catenin (88% of baseline; P = 0.3152), or Pdlim5 + delta-catenin (69% of baseline; P = 0.0059); n = 4 independent experiments. (D) Representative ratiometric FRET confocal images of 4-DIV primary rat hippocampal neurons expressing a FRET-based (CFP–YFP) RhoA biosensor in combination with phospho-null or phospho-mimic delta-catenin mutants. (E) Quantification of ratio of dendrite-to-soma RhoA activity (Fr) in biosensor-expressing neurons, as observed via ratiometric FRET. Relative to controls (Fr = 2.462 ± 0.142), expression of phospho-null delta-catenin produced no change in dendritic RhoA activity (Fr = 2.713 ± 0.074; P = 0.2098). Expression of phospho-mimic delta-catenin produced a significant decrease in dendritic RhoA activity (Fr = 1.312 ± 0.048) compared with both control (P < 0.0001) and phospho-null delta-catenin -expressing neurons (P < 0.0001). (F) Quantification of RhoA activity in the soma of biosensor-expressing neurons, as observed via ratiometric FRET. Expression of either phospho-null or phospho-mimic delta-catenin in primary rat hippocampal neurons produced no changes in RhoA activity in the soma of neurons (P = 0.8906; P = 0.2903). For D–F, n = 4 neurons from independent cultures. (G) Immunoblot of HA-delta-catenin coimmunoprecipitation with Flag-Cortactin from HEK293 cell lysates. IP, immunoprecipitation. WCE, whole cell extract. (H) Densitometric quantification (normalized to total delta-catenin and antibody) of relative (to phospho-mimic) amounts of delta-catenin mutants pulled down by Cortactin. Phospho-null delta-catenin failed to coimmunoprecipitate with Cortactin, while phospho-mimic delta-catenin successfully immunoprecipitated with Cortactin (P = 0.0077). This association was further stabilized by coexpression of Pdlim5 (P = 0.0301). For H, n = 3 independent cultures/experiments. *, P < 0.05; **, P ≤ 0.001; ***, P ≤ 0.0001. For A–C, E, F, and H, significance was determined using a one-way ANOVA followed by Tukey’s test. Error bars indicate SEM. Scale bars, 10 µm.

Figure 10.

Proposed model of delta-catenin signaling during dendrite development. In the absence of glutamate signaling (left side of schematic), delta-catenin remains unphosphorylated within its PDZ-binding motif (−6 and −3 serines counting back from delta’s C-terminal valine). This allows delta-catenin to bind Magi1, disrupting delta-catenin’s indirect inhibition of RhoA activity. Speculatively in this regard, Magi1 might make delta-catenin less apt to sequester RhoA activators such as p190RhoGEF (which is known to associate with delta-catenin), lessening RhoA inhibition (relative to delta:Pdlim5 association), and ultimately promoting dendrite elongation. Conversely, in response to glutamate signaling and type 1 mGluR activation (right side of schematic), delta-catenin is phosphorylated by CDK5 within its PDZ-binding motif (−6 and likely −3 serines). Such PDZ-binding motif phosphorylation prevents the binding of Magi1. Instead, delta-catenin binds Pdlim5, which enhances delta-catenin’s inhibition of RhoA activity to a greater extent (than delta:Magi1), and which promotes the delta:Cortactin interaction, to further local actin polymerization and dendritic branching. P, phospho/phosphorylated.