CaMKII-dependent phosphorylation of the GTPase Rem2 is required to restrict dendritic complexity

Abstract

The morphogenesis of the dendritic arbor is a critical aspect of neuronal development, ensuring that proper neural networks are formed. However, the molecular mechanisms that underlie this dendritic remodeling remain obscure. We previously established the activity-regulated GTPase Rem2 as a negative regulator of dendritic complexity. In this study, we identify a signaling pathway whereby Rem2 regulates dendritic arborization through interactions with Ca(2+)/calmodulin-dependent kinases (CaMKs) in rat hippocampal neurons. Specifically, we demonstrate that Rem2 functions downstream of CaMKII but upstream of CaMKIV in a pathway that restricts dendritic complexity. Furthermore, we show that Rem2 is a novel substrate of CaMKII and that phosphorylation of Rem2 by CaMKII regulates Rem2 function and subcellular localization. Overall, our results describe a unique signal transduction network through which Rem2 and CaMKs function to restrict dendritic complexity.

Figure 1.

Rem2 and CaMK signaling pathways interact to regulate dendritic morphology. A, Representative images (top) and quantification of staining intensity (bottom) of hippocampal neurons transfected at 2 DIV with GFP and an empty vector (Mock), the Rem2 shRNA used in subsequent morphology experiments (RNAi 1), or a second Rem2 shRNA (RNAi 2). At 5 DIV, the neurons were fixed and stained for Rem2 and MAP2. *p < 0.05, two-way ANOVA with Scheffé's post hoc test. n > 50 neurons per condition. Scale bar, 5 μm. B, Representative images of 5 DIV neurons transfected with a GFP-expressing plasmid (Mock) or a GFP-expressing plasmid plus a Rem2 shRNA (RNAi) and then untreated or treated for 24 h with 20 μm W-7 or 20 μm KN-93. Unless otherwise indicated, in this and all subsequent figures, scale bars represent 50 μm. C, Dendritic complexity of 5 DIV mock and Rem2 RNAi neurons with or without 20 μm W-7 treatment for 24 h. In this and all subsequent figures, complexity is quantified via Sholl analysis, and a bar above or below the curve indicates significance at those particular radii compared with mock. Bars are colored and/or dashed to match the style/color of the data curve to which they refer. D, Dendritic complexity of 5 DIV mock and Rem2 RNAi neurons with or without 20 μm KN-93 treatment for 24 h. Here, mock and RNAi data from C are replotted for ease of comparison with other conditions. *p < 0.05; **p < 0.01, multivariate ANOVA with Scheffé's post hoc test. n = 21–46 neurons per condition.

Figure 2.

Rem2 is upstream of CaMKIV in a signaling pathway that regulates dendritic complexity. A, Quantification (left) and representative images (right) of dendritic complexity at 5 DIV of mock transfected neurons compared with Rem2 RNAi-, CA CaMKIV-, and DN CaMKIV-transfected neurons. B, Representative images (left) and quantification (right) of dendritic complexity at 5 DIV of mock and Rem2 RNAi-transfected neurons compared with neurons cotransfected with both Rem2 RNAi and either CA or DN CaMKIV. Mock and RNAi data from A are replotted for ease of comparison with other conditions. *p < 0.05, multivariate ANOVA with Scheffé's post hoc test. n = 42–52 neurons per condition.

Figure 3.

Rem2 signals in parallel to a CaMKIα pathway to mediate dendritic complexity. A, Quantification (left) and representative images (right) of dendritic complexity at 5 DIV of mock transfected neurons compared with Rem2 RNAi-, CA CaMKIα-, and DN CaMKIα-transfected neurons. B, Representative images (left) and quantification (right) of dendritic complexity at 5 DIV of mock and Rem2 RNAi-transfected neurons compared with neurons cotransfected with both Rem2 RNAi and either CA or DN CaMKIα. Mock and RNAi data from A are replotted for ease of comparison with other conditions. *p < 0.05, **p < 0.01, multivariate ANOVA with Scheffé's post hoc test. n = 35–57 neurons per condition.

Figure 4.

CaMKII signals through Rem2 to inhibit dendritic complexity. A, Quantification (left) and representative images (right) of dendritic complexity at 5 DIV of mock transfected neurons compared with Rem2 RNAi-, CA CaMKIIα-, and DN CaMKIIα-transfected neurons. B, Representative images (left) and quantification (right) of dendritic complexity at 5 DIV of mock and Rem2 RNAi-transfected neurons compared with neurons cotransfected with both Rem2 RNAi and either CA or DN CaMKIIα. Mock and RNAi data from A are replotted for ease of comparison with other conditions. C, Quantification (left) and representative images (right) of dendritic complexity at 5 DIV of mock transfected neurons compared with Rem2 RNAi and CaMKIIα RNAi, alone or in combination as indicated, transfected neurons. D, Representative images (left) and quantification (right) of dendritic complexity at 5 DIV of mock transfected neurons compared with neurons expressing the CaMKII-specific peptide inhibitor CaMKIIN. Mock data from A are replotted for ease of comparison with other conditions. *p < 0.05, multivariate ANOVA with Scheffé's post hoc test. n = 30–49 neurons per condition.

Figure 5.

Rem2 is a phosphoprotein with six putative CaMKII phosphorylation sites. A, Western blot of lysates from E18 rat cortical neuron lysates that were (from left) untreated (Control), mock treated (Mock), or treated with CIAP for 30 min. The blot was probed with a polyclonal antibody against rat Rem2 (Ghiretti and Paradis, 2011) and an antibody against β-actin as a loading control. B, Alignment of the CaMKII consensus phosphorylation sites in the rat Rem2 protein with other vertebrate Rem2 homologs, demonstrating that putative CaMKII phosphorylation sites 2, 4, 5, and 6 are highly conserved in vertebrates. C, Coomassie-stained SDS-PAGE gel depicting the peptides present in an in vitro kinase assay containing ATPγP³² and (left to right) activated CaMKII alone or activated CaMKII plus purified GST, purified GST-Rem2, purified GST-Rem2 S241A/S308A, or purified GST-Rem2 S69A/S241A/S308A/S334A. Note that the intensity of the 65 kDa band corresponding to GST-Rem2 constructs is comparable across all lanes, demonstrating equal protein loading. D, Autoradiograph of the gel in C identifying proteins phosphorylated by CaMKII. GST-Rem2 runs at 65 kDa, and decreasing phosphorylation is observed with S241A/S308A and S69A/S241A/S308A/S334A mutants. Note that autophosphorylated CaMKII is detected in all lanes (30 kDa) as is phosphorylated GST (26 kDa). E, Western blot demonstrating that the peptides between 26 and 65 kDa detected by Coomassie stain in C and by autoradiography in D represent GST-Rem2 truncation products, as they are detected by both anti-GST and anti-Rem2 antibodies, along with full-length GST-Rem2 (65 kDa). Note that GST (26 kDa) is also detected by both anti-GST and anti-Rem2 antibodies as the Rem2 antibody was raised against GST-Rem2 in rabbits (Ghiretti and Paradis, 2011). The high-molecular-weight band (70 kDa) is DnaK, a bacterial contaminant from the GST or GST-Rem2 protein purification process.

Figure 6.

Serine 241 and 308 are required for Rem2 to mediate dendritic complexity. A, Representative images of 5 DIV mock and Rem2 RNAi-transfected neurons compared with neurons transfected with either an RNAi-resistant Rem2 cDNA alone [overexpression (OE)] or cotransfected with a Rem2 shRNA and an RNAi-resistant Rem2 cDNA (Rescue), as indicated. Note that overexpression of the “wild-type” RNAi-resistant Rem2 cDNA is shown here for comparison purposes only, as it has been extensively documented that overexpression of Rem2 is similar to wild type under these conditions (Ghiretti and Paradis, 2011), and thus this condition is not requantified in B–D. B, Quantification of dendritic complexity of 5 DIV mock transfected neurons compared with neurons transfected with either a Rem2 shRNA alone (RNAi) or a Rem2 shRNA and an RNAi-resistant Rem2 cDNA (Wild Type Rescue). C, Quantification of dendritic complexity of 5 DIV mock transfected neurons compared with neurons transfected with a Rem2 shRNA alone (RNAi), an RNAi-resistant Rem2 cDNA with serine to alanine mutation at S69 or S241 (OE), or a Rem2 shRNA and an RNAi-resistant Rem2 cDNA with serine-to-alanine mutation at S69 or S241 (Rescue). Mock and RNAi data from B are replotted for ease of comparison with other conditions. Note that S69A OE and S69A Rescue are not significant from mock or wild-type rescue. D, Quantification of dendritic complexity of 5 DIV mock transfected neurons compared with neurons transfected with a Rem2 shRNA alone (RNAi), an RNAi-resistant Rem2 cDNA with serine-to-alanine mutation at S308 or S334 (OE), or a Rem2 shRNA and an RNAi-resistant Rem2 cDNA with serine-to-alanine mutation at S308 or S334 (Rescue). Mock and RNAi data from B) are replotted for ease of comparison with other conditions. Note that S308A OE, S334A OE, and S334A Rescue are not significant from mock or wild-type rescue. *p < 0.05, multivariate ANOVA with Scheffé's post hoc test. n = 41–58 neurons per condition.

Figure 7.

CaMKII phosphorylation leads to enhanced nuclear localization of Rem2. A, Representative images of the somas of 5 DIV mock transfected neurons compared with neurons transfected with CA CaMKII. Neurons were labeled with DAPI (to label nuclei) and immunostained with anti-Rem2 (to label endogenous Rem2). Scale bar, 10 μm. B, Representative images of the somas of 5 DIV neurons transfected with Rem2 S241A/S308A compared with those transfected with Rem2 S241A/S308A and CA CaMKII. Neurons were labeled with DAPI and immunostained with anti-myc (to label Rem2 S241A/S308A). Scale bar, 10 μm. C, Quantification of the ratio of nuclear-to-cytoplasmic endogenous Rem2 expression. *p < 0.05, pairwise ANOVA. n > 30 neurons per condition. D, Quantification of the ratio of nuclear-to-cytoplasmic myc-tagged Rem2 S241A/S308A expression. There was no significance by pairwise ANOVA (p = 0.099). n > 30 neurons per condition.

Figure 8.

Model of CaMK-Rem2 interactions in mediating changes in dendritic complexity. A, After phosphorylation by CaMKII at S241 and/or S308, Rem2 translocates to the nucleus where it either directly or indirectly inhibits CaMKIV, which normally stimulates the transcription of genes that promote dendritic complexity. The activation of the CaMKII/Rem2 cascade leads to inhibition of dendritic complexity by preventing CaMKIV from activating these genes. B, Critical serine residues are conserved in all members of the RGK protein family (Rad, Gem, Rem, and Rem2) in mouse, as shown by alignment of sequences surrounding Rem2 S241 (top, asterisk) and Rem2 S308 (bottom, asterisk) and the NLS (bottom, gray box). The NLS was identified by http://nls-mapper.iab.keio.ac.jp/.