Ca2+-Triggered (de)ubiquitination Events in Synapses

Abstract

Neuronal communication relies on neurotransmitter release from synaptic vesicles (SVs), whose dynamics are controlled by Ca²⁺-dependent pathways, as many thoroughly studied phosphorylation cascades. However, little is known about other post-translational modifications, such as ubiquitination. To address this, we analyzed resting and stimulated synaptosomes (isolated synapses) by quantitative mass spectrometry. We identified more than 5000 ubiquitination sites on ∼2000 proteins, the majority of which participate in SV recycling processes. Several proteins showed significant changes in ubiquitination in response to Ca²⁺ influx, with the most pronounced changes in CaMKIIα and the clathrin adaptor protein AP180. To validate this finding, we generated a CaMKIIα mutant lacking the ubiquitination target site (K291) and analyzed it both in neurons and non-neuronal cells. K291 ubiquitination, close to an important site for CaMKIIα autophosphorylation (T286), influences the synaptic function of this kinase. We suggest that ubiquitination in response to synaptic activity is an important regulator of synaptic function.

Keywords: CAMKI; calcium; post-translational modification; synapse; ubiquitin.

Graphical abstract

Fig. 1

Workflow for the quantitative analysis of ubiquitinated proteins in depolarized synaptosomes under different conditions. A, synaptosomes were isolated from the brains of 5–6-week-old Wistar rats by homogenization of brain tissue followed by differential centrifugation and discontinuous Ficoll gradient centrifugation. B, synaptosome depolarization was induced by KCl in the presence of Ca²⁺ or the Ca²⁺-chelator EGTA and was quenched after 2 min by the addition of lysis buffer. Three independent stimulations were performed for each condition (EGTA versus Ca²⁺). C, equal amounts of proteins were subsequently precipitated by methanol/chloroform protein precipitation and sequentially digested with LysC and trypsin, followed by ubiquitin remnant-containing (K-ε-GG) peptide enrichment and chemical labeling with isobaric TMT6 reagents. Differently labeled peptides were combined and analyzed by LC-MS/MS. Two independent TMT6 experiments were performed. Peptide identification and quantification was performed in MaxQuant and the extracted reporter ion intensities were further processed in R.

Fig. 2

Pathway enrichment analysis of ubiquitinated proteins identified in synaptosomes and comparison of our data set with the literature. A, rank order of protein signals depicting the number of identified ubiquitination sites per protein in our samples, which include both Ca²⁺- and EGTA-treated syntaptosomes. B, sunburst diagram depicting significantly enriched biological process terms based on the SynGo database (47). C, detailed list of enriched biological processes based on the ShinyGO (54). D, comparison of our ubiquitination data set derived from rat synaptosomes with a previous ubiquitination data set derived from mouse brain (35) based on sequence similarity of the six amino acids flanking N- and C-terminal the modified lysine residue.

Fig. 3

Ubiquitination changes in depolarized synaptosomes under different stimuli. A, volcano plot showing log₂(intensity fold change) of ubiquitination sites quantified under Ca²⁺ vs. EGTA conditions against –log₁₀(q-value). The color encodes the significance of changes, highlighting with red and orange the ubiquitination sites that change significantly at FDRs of 1% and 5%, respectively. B, sunburst diagram depicting enriched biological process terms of proteins possessing regulated ubiquitination sites based on the SynGO database (47). C, synaptosome depolarization was induced by KCl in the presence of Ca²⁺ or the Ca²⁺-chelator EGTA and was quenched after 2 min by the addition of lysis buffer. Six independent stimulations were performed for each condition (EGTA versus Ca²⁺). Equal amounts of proteins were subsequently precipitated by methanol/chloroform protein precipitation and sequentially digested with LysC and trypsin. Standard/heavy peptides were spiked in the mixture of endogenous/light peptides prior to ubiquitin-remnant (K-ε-GG) peptide enrichment. Eluted (K-ε-GG) peptides were analyzed by LC-MS/MS. Peptide identification and quantification were performed in Skyline and the extracted peak areas were further processed in R. D, representative extracted fragment ion chromatograms for endogenous/”light” and standard/”heavy” CaMKIIα peptide (QETVDCLKK) ubiquitinated at K291, under calcium-deprived (EGTA) and calcium-free conditions. The different colors represent distinct fragments b and y ions of the indicated peptide. E, Log₂(light-to-heavy peptide intensity fold change) of ubiquitination sites quantified under Ca²⁺ vs. EGTA conditions against the average log₂ (light-to-heavy peptide intensity ratio) for the Ca²⁺ and EGTA conditions. The color shows the statistical significance (FDR) of log₂(light-to-heavy peptide intensity fold change).

Fig. 4

PTMs on the regulatory domain of Ca²⁺/calmodulin dependent kinase II α (CaMKIIα) and their quantification in depolarized synaptosomes under different conditions. A, a horizontal bar represents the CaMKIIα sequence, with colored regions showing the domains annotated according to Chao et al., 2011 (76). The regulatory ubiquitination site (K291) resides in the regulatory domain of CaMKIIα very close to the autophosphorylation site (T286). B, dodecameric structure of CaMKIIα and the conformational states of a CaMKIIα subunit; in the closed conformation (PDB code 2VN9 (78)) the regulatory domain folds back to the kinase domain, blocking access to the active site of the enzyme. The binding of Ca²⁺/calmodulin to the regulatory segment releases the active site of the enzyme, rendering the enzyme catalytically active and T286 accessible for phosphorylation (PDB code 2WEL (78)). Part of the K291 structure represented by a sphere is missing in the PDB codes. The PDB codes correspond to the human CaMKIIδ subunit 0 to 310 aa, which shares 92.58% sequence identity with human CaMKIIα and therefore we can safely assume that these domains have the same structure. C, theoretical example where all ubiquitinated molecules of CaMKIIα are phosphorylated at T286 during stimulation, leading to a 100% decrease in the levels of the ubiquitinated peptide. Phosphatase (pptase) treatment removes the confounding phosphorylation allowing the accurate quantification of CaMKIIα ubiquitination at K291, independent of its phosphorylation status. D, representative extracted fragment ion chromatograms for endogenous/”light” and standard/”heavy” CaMKIIα peptide (QETVDCLKK) ubiquitinated at K291, and its doubly modified variant bearing ubiquitination at K291 and phosphorylation at T286, before and after QuickCIP treatment. The different colors represent distinct fragments b and y ions of the indicated peptide. E, summary barplot showing the mean light-to-heavy peak area ratios. Limma statistical testing was performed to determine significant differences and account for the synaptosome preparation batch effect (N = six independent stimulation experiments, with 2 MS measurement replicates for each experiment). ∗p < 0.05, ∗∗∗p < 0.001. We note that for the sake of simplicity, we show here only one synaptosome preparation batch with three independent stimulation experiments. For a detailed view of both synaptosome batches refer to supplemental Fig. S5.

Fig. 5

Functional assay to monitor the effects of CaMKIIα expression in HeLa cells and neurons. A, generation of CaMKIIα K291R mutant that cannot be ubiquitinated at K291. B, generation of HeLa Kyoto cell lines stably expressing either CaMKIIα-WT or the mutant variant K291R. C and D, Violin plots illustrating the endogenous/”light”-to-standard/”heavy” peptide intensities in HeLa cells under different conditions (Ionomycin/Ca²⁺-vs-DMSO). We note that we used the same standard/”heavy” peptide with the sequence QETpVDCLK to normalize the endogenous/light peptides QETpVDCLK and QETpVDCLR. A two-sample t test was performed to determine significant differences (N = three independent stimulation experiments, with 2 MS measurement replicates for each experiment). ∗p < 0.05, ∗∗∗p < 0.001. E, neurons were transfected with either the wild type (WT) or the K291R variants of CaMKIIα and were analyzed by fluorescence microscopy 6 to 8 days later. The green channel indicates the CaMKIIα expression, while the magenta channel shows anti-synaptotagmin 1 antibodies (directly conjugated to the fluorophore Atto647N), which are taken up by recycling synaptic vesicles, during a 60-min incubation. After washing with Tyrode's solution, the cells were fixed with PFA and imaged. F, synapses were identified based on the synaptotagmin 1 signal, which was correlated with the CaMKIIα expression signal within the area of each synapse, using a Pearson correlation analysis. Subsequently, the fluorescence intensity of the synaptotagmin 1 label was quantified in the boutons in which the two signals were well correlated (meaning true presynaptic boutons, and not presynapses of non-transfected neurons that overlapped with CaMKIIα-expressing dendrites). A paired t test between the wild type and the mutant was performed to determine significant differences (p = 0.03, N = three independent experiments, with hundreds of synapses analysed for each experiment).