MINK and TNIK differentially act on Rap2-mediated signal transduction to regulate neuronal structure and AMPA receptor function

Abstract

Misshapen/NIKs (Nck-interacting kinases)-related kinase (MINK) and closely related TRAF2/Nck-interacting kinase (TNIK) are proteins that specifically bind to activated Rap2 and are thus hypothesized to relay its downstream signal transduction. Activated Rap2 has been found to stimulate dendritic pruning, reduce synaptic density and cause removal of synaptic AMPA receptors (AMPA-Rs) (Zhu et al., 2005; Fu et al., 2007). Here we report that MINK and TNIK are postsynaptically enriched proteins whose clustering within dendrites is bidirectionally regulated by the activation state of Rap2. Expression of MINK and TNIK in neurons is required for normal dendritic arborization and surface expression of AMPA receptors. Overexpression of a truncated MINK mutant unable to interact with Rap2 leads to reduced dendritic branching and this MINK-mediated effect on neuronal morphology is dependent upon Rap2 activation. While similarly truncated TNIK also reduces neuronal complexity, its effect does not require Rap2 activity. Furthermore, Rap2-mediated removal of surface AMPA-Rs from spines is entirely abrogated by coexpression of MINK, but not TNIK. Thus, although both MINK and TNIK bind GTP-bound Rap2, these kinases employ distinct mechanisms to modulate Rap2-mediated signaling. MINK appears to antagonize Rap2 signal transduction by binding to activated Rap2. We suggest that MINK interaction with Rap2 plays a critical role in maintaining the morphological integrity of dendrites and synaptic transmission.

Figure 1.

Tissue and developmental distribution of MINK/TNIK proteins. A, Western blot analysis of postnuclear supernatants isolated from adult rat tissues. Molecular mass is indicated in kilodaltons. B, Developmental regulation of MINK and TNIK. Postnuclear supernatant Western blot analyses of hippocampus and cortex isolated from embryonic, postnatal and adult rat tissues. E18, Embryonic day 18; P3, P5, P8, postnatal day 3, 5, and 8, respectively; Ad, adult. C, MINK and TNIK are enriched in PSD. Western blot analyses of PSD subcellular fractionation from rat brain. S1, P1, Whole-brain postnuclear supernatant and membrane fractions, respectively; S2, cytosolic proteins; P2, crude synaptosomal fraction (20 μg each); PSD, Triton X-100-extracted PSD (5 μg of protein). Antibodies for immunoblotting are indicated to the right of panels. D, Subcellular distribution of MINK/TNIK in hippocampal neurons. Confocal microscopy of 23-d-old hippocampal neurons stained for MINK/TNIK, phalloidin, and bassoon. Arrows indicate colocalized puncta. Box in “merge” panel indicates region of higher magnification. Scale bars, 20 μm.

Figure 2.

MINK- and TNIK-RNAi diminish neuronal complexity. A, Mean cell body MINK/TNIK immunostaining intensity for hippocampal neurons transfected with pSuper or pSuper-based RNAi constructs targeting MINK (MINK-RNAi), TNIK (TNIK-RNAi), or luciferase (Luc-RNAi), as indicated. Mean integrated intensity values were normalized to pSuper-transfected neurons. Error bars indicate ±SEM. ***p < 0.001 relative to pSuper control, ANOVA. n ≥ 12 neurons for each. B, Representative images of hippocampal neurons cotransfected with pSuper or RNAi constructs as indicated, along with DsRed to visualize transfected cell morphology. Immunostaining was performed with antibodies to endogenously expressed MINK/TNIK. Arrows indicate transfected neurons. Dendritic segments at top are 40 μm long. Scale bar, 50 μm. C, Sholl analysis, transfections as indicated (n ≥ 12 neurons for each). A significant difference between transfected groups was confirmed by a 5 × 5 mixed-effect ANOVA, with transfected group as the five-level between-group effect and distance from the soma as the five-level within-group effect (F_(4,332) = 15.625; p < 0.0001). Post hoc tests (PLSD) revealed that, relative to pSuper control, a loss of either MINK, TNIK, or simultaneous knockdown of MINK and TNIK significantly reduced dendritic complexity (****p < 0.0001).

Figure 3.

Loss of MINK or TNIK reduces synapse number and AMPA-R function. A, Mean density of dendritic protrusions analyzed per 10 μm of dendritic length in neurons transfected with pSuper or RNAi constructs as indicated, along with DsRed to visualize transfected cell morphology. Error bars indicate ±SEM. **p < 0.01 relative to pSuper control, ANOVA. n ≥ 12 neurons for each. B, Hippocampal neurons transfected with various constructs, along with DsRed to visualize transfected cell morphology were immunostained for sGluR1 expression. Quantification of mean integrated sGluR1 cluster intensity in dendritic segments normalized to pSuper. Error bars indicate ±SEM. ***p < 0.001 relative to pSuper control, ANOVA. n ≥ 12 neurons for each. C, Representative images of MINK- and TNIK-RNAi effect on GluR1 surface expression in dendritic segments (40 μm). D, E, Effects of MINK downregulation on excitatory synaptic transmission in organotypic hippocampal slice culture. Top, Sample-evoked AMPA-EPSC traces are shown from pairs of nontransfected and neighboring transfected CA1 hippocampal pyramidal cells. Bottom, Pairwise analysis between transfected cells and neighboring nontransfected cells shows that Luc-RNAi has no significant effect on AMPAR-EPSC amplitude (D) (p > 0.3, ANOVA; 15 pairs), while MINK-RNAi causes a significant reduction (E) (**p < 0.01, ANOVA; 13 pairs). Filled circles represent a single pair of recordings; open red circle and error bars represent mean ±SEM. F, Summary of Luc-RNAi and MINK-RNAi expression effects on AMPA-R-EPSCs. Each bar graph represents average of ratios obtained from multiple pairs of transfected and nontransfected neighboring neurons. Error bars indicate mean ±SEM. *p < 0.05 for MINK-RNAi relative to Luc-RNAi control, ANOVA.

Figure 4.

Activated Rap2 binds MINK in hippocampal neurons. A, Representative images of hippocampal neurons transfected with GFP and constitutively active Rap2 (HA-Rap2(ca)) and/or Flag-tagged MINK constructs as indicated (left). Scale bars, 50 μm. Representative 40 μm dendritic segments shown below each low-magnification image. B, Schematic diagram of wild-type MINK, kinase dead MINK (MINK KD) and MINK lacking CNH domain (MINK ΔCNH) constructs used for overexpression in neurons. C, Representative images of Rap2 mutant construct effects on dendritic clustering of endogenous staining for MINK/TNIK. Neurons were transfected with GFP and Rap2 dominant-negative [Rap2(dn)], Rap2(ca), or Rap2(ca) lacking its CAAX motif (Rap2(ca) ΔCAAX), and stained for endogenous MINK/TNIK. D, Bar graph of mean integrated intensity of dendritic MINK/TNIK clusters normalized to empty vector-transfected hippocampal neurons. Error bars indicate ±SEM. ***p < 0.001, relative to empty vector-transfected neurons, ANOVA. n ≥ 12 neurons for each. E, Representative images characterizing MINK/TNIK clusters induced by active Rap2. Neurons were transfected with HA-Rap2(ca) and stained for endogenous MINK/TNIK and syntaxin 6 as indicated. Scale bar, 50 μm. Representative 40 μm dendritic segments shown below each low-magnification image.

Figure 5.

MINKΔCNH-mediated reduction of dendritic complexity requires Rap2 activation. A, Representative images of hippocampal neurons transfected with GFP to visualize transfected cell morphology along with Flag-tagged constructs (indicated left). Scale bar, 50 μm. B, MINK ΔCNH compromises dendritic complexity. Neurons were transfected as indicated (right) and subjected to Sholl analysis (n ≥ 12 neurons for each). A significant difference between transfected groups was confirmed by a 7 × 5 mixed-effect ANOVA, with transfected group as the seven-level between-group effect and distance from the soma as the five-level within-group effect (F_(6,574) = 54.051; p < 0.0001). Post hoc tests (PLSD) determined MINK ΔCNH alone, and Rap2(ca) either alone or with MINK or MINK KD each displayed significantly fewer dendritic branch point crossings relative to GFP control (****p < 0.0001). C, Rap2 activation is required for MINK ΔCNH to reduce dendritic arborization. Neurons were transfected as indicated (right) and subjected to Sholl analysis (n ≥ 12 neurons for each). A significant difference between transfected groups was confirmed by a 6 × 5 mixed-effect ANOVA, with transfected group as the six-level between-group effect and distance from the soma as the five-level within-group effect (F_(5,504) = 31.438; p < 0.0001). Post hoc tests (PLSD) revealed expression of MINK ΔCNH alone, or MINK ΔCNH coexpressed with HA-Rap1(dn) both significantly reduced dendritic complexity relative to empty vector control (****p < 0.0001 for each relative to control). Dominant-negative Rap2 subtly decreased complexity relative to control (**p < 0.01 for each relative to control). However, when MINK ΔCNH was coexpressed with HA-Rap2(dn) the number of dendritic branch point crossings failed to decrease below the level of HA-Rap2(dn) expressed alone [p = 0.5673 for HA-Rap2(dn) expressed with MINK ΔCNH relative to HA-Rap2(dn) alone].

Figure 6.

MINK disrupts Rap2-mediated regulation of surface GluR1 clusters. A, Representative images of MINK expression effects on Rap2-mediated reduction in surface AMPA receptor clusters. B, Quantification of mean sGluR1 cluster intensity in dendritic segments cotransfected with DsRed and pGW1 empty vector, or combinations of transfections as indicated (below graph). Mean cluster integrated intensities were normalized to empty vector control. Error bars indicate ±SEM, where **p < 0.01, ***p < 0.001 is relative to empty vector, and ^#p < 0.05, ^##p < 0.01, ^###p < 0.001 is relative to MINK cotransfected with Rap2(ca), ANOVA. n ≥ 12 neurons for each.