Parkin Deficiency Reduces Hippocampal Glutamatergic Neurotransmission by Impairing AMPA Receptor Endocytosis

Abstract

Mutations in the gene encoding Parkin, an E3 ubiquitin ligase, lead to juvenile-onset Parkinson's disease by inducing the selective death of midbrain dopaminergic neurons. Accumulating evidence indicates that Parkin also has an important role in excitatory glutamatergic neurotransmission, although its precise mechanism of action remains unclear. Here, we investigate Parkin's role at glutamatergic synapses of rat hippocampal neurons. We find that Parkin-deficient neurons exhibit significantly reduced AMPA receptor (AMPAR)-mediated currents and cell-surface expression, and that these phenotypes result from decreased postsynaptic expression of the adaptor protein Homer1, which is necessary for coupling AMPAR endocytic zones with the postsynaptic density. Accordingly, Parkin loss of function leads to the reduced density of postsynaptic endocytic zones and to impaired AMPAR internalization. These findings demonstrate a novel and essential role for Parkin in glutamatergic neurotransmission, as a stabilizer of postsynaptic Homer1 and the Homer1-linked endocytic machinery necessary for maintaining normal cell-surface AMPAR levels.

Significance statement: Mutations in Parkin, a ubiquitinating enzyme, lead to the selective loss of midbrain dopaminergic neurons and juvenile-onset Parkinson's disease (PD). Parkin loss of function has also been shown to alter hippocampal glutamatergic neurotransmission, providing a potential explanation for PD-associated cognitive impairment. However, very little is known about Parkin's specific sites or mechanisms of action at glutamatergic synapses. Here, we show that Parkin deficiency leads to decreased AMPA receptor-mediated activity due to disruption of the postsynaptic endocytic zones required for maintaining proper cell-surface AMPA receptor levels. These findings demonstrate a novel role for Parkin in synaptic AMPA receptor internalization and suggest a Parkin-dependent mechanism for hippocampal dysfunction that may explain cognitive deficits associated with some forms of PD.

Keywords: AMPA receptor; Homer; Parkin; endocytic zone; synapse.

Figure 1.

Loss of Parkin decreases spontaneous excitatory neurotransmission. A, Immunoblot of primary hippocampal neurons lentivirally transduced on 2–3 DIV with control, GFP/shParkin (shParkin), or shParkin+GFP-hu-Parkin (rescue) constructs; collected on 14 DIV; and probed with Parkin, GFP, and tubulin antibodies. B, Quantification of shParkin knockdown efficacy, expressed as percentage of endogenous Parkin. Expression of shParkin for ∼11 d led to a >60% knockdown of endogenous Parkin (n = 3 experiments; **p < 0.01, unpaired t test). C, Representative traces of spontaneous mEPSCs from hippocampal neurons transduced on 2–3 DIV with control, GFP-hu-Parkin (Parkin), shParkin, or rescue constructs. D, E, Quantification of mEPSC frequency (D) and amplitude (E) for the four conditions (n = 16 for control, 22 for shParkin, 21 for Parkin, 20 for rescue; *p = 0.03, one-way ANOVA). Error bars represent SEM. F, Representative traces of mEPSCs from hippocampal neurons cultured from wild-type and Parkin knock-out rats. G, H, Quantification of mEPSC frequency (G) and amplitude (H) for the two conditions (n = 15 for WT, 31 for Parkin KO; **p = 0.004, ***p < 0.0001, one-way ANOVA). Error bars represent SEM.

Figure 2.

Synapse density is unaltered by Parkin knockdown. A, Images of 15 DIV hippocampal neurons expressing control, shParkin, or rescue constructs and immunostained for VAMP2 and Homer to label excitatory synapses. Scale bar, 10 μm. B, Images of 15 DIV hippocampal neurons expressing SAP102-GFP+/− shParkin (green) and immunostained for VAMP2 (red). Scale bar, 10 μm. C, Quantification of synapse density, measured as VAMP2/Homer-immunopositive puncta/unit length primary dendrite and expressed as a fraction of control condition (n ≥ 5 fields of view per condition with >100 VAMP2/Homer puncta per field, results confirmed in 3 independent experiments; no significant differences, unpaired t test). Error bars represent SEM. D, Quantification of synapse density, measured as in C (n ≥ 5 fields of view per condition with >100 VAMP2/SAP102 puncta per field, results confirmed in 3 independent experiments; no significant differences, unpaired t test). Error bars represent SEM.

Figure 3.

Parkin deficiency reduces synaptic efficacy. A, DIC (left) and fluorescence images of hippocampal neurons cultured on glial microislands, illustrating the recording paradigm. Presynaptic stimuli were evoked in untransduced neurons, and postsynaptic responses were recorded from neurons transduced with one of the four constructs. B, Representative traces of evoked EPSCs from neurons expressing the construct indicated. C, Quantification of eEPSC amplitude (n = 18 cells for GFP control, 14 for shParkin, 15 for Parkin, 6 for rescue; *p = 0.04, one-way ANOVA). Error bars represent SEM. D, Example of synaptic failures over a 10 s recording trial at 1 Hz. Positive current at action potential threshold produced action potentials in the presynaptic cell (top). A postsynaptic cell held at −70 mV generated eEPSCs or synaptic failures (indicated by red arrow). E, Quantification of EPSC failure rate (n = 18 for control, 14 for shParkin, 15 for Parkin, 6 for rescue; *p = 0.04, one-way ANOVA). Error bars represent SEM.

Figure 4.

Parkin deficiency decreases cell-surface AMPA receptor levels. A, Representative traces of whole-cell currents induced by local application of 100 μm kainate for neurons expressing control, Parkin, shParkin, or rescue constructs. B, Quantification of peak current amplitudes induced by AMPA receptor activation (n = 17 for control, 27 for shParkin, 13 for Parkin, 10 for rescue; *p = 0.03, ***p < 0.0001, one-way ANOVA). Error bars represent SEM. C, Images of 15 DIV neurons expressing control, shParkin, or rescue constructs (all green) immunostained for cell-surface GluA1 or GluA2 (red). Scale bar, 10 μm. D, Quantification of cell-surface GluA1, expressed as a fraction of control (n ≥ 5 fields of view per condition with >50 GluA1 puncta per field, results confirmed in 3 independent experiments; *p < 0.05, unpaired t test). Error bars represent SEM. E, Quantification of cell-surface GluA2, expressed as a fraction of control (n ≥ 4 fields of view per condition with >50 GluA2 puncta per field, results confirmed in 3 independent experiments; *p < 0.05, unpaired t test). Error bars represent SEM. F, Images of 15 DIV neurons expressing SAP102-GFP+/− shParkin immunostained for cell-surface GluA1. Scale bar, 10 μm. G, Quantification of synaptic cell-surface GluA1 based on colocalization with SAP102-GFP, expressed as a fraction of control (n ≥ 6 fields of view per condition with >50 GluA1 puncta per field, results confirmed in 3 independent experiments; ***p < 0.0001, unpaired t test).

Figure 5.

Altered postsynaptic density composition in Parkin KO neurons. A, Left, Immunoblots of purified PSDs from whole brain of WT or Parkin KO rats, probed for the indicated proteins (n = 3 animals per condition). Right, Immunoblots of PSDs from whole brain of WT or PINK1 KO rats, probed for the indicated proteins (n = 2 animals per condition). The black vertical line on the Parkin immunoblot indicates that these lanes are not contiguous. B, Quantification of PSD protein levels for WT and Parkin KO conditions, normalized to GAPDH and expressed as percentage of mean WT protein level (*p < 0.05, unpaired t test). Error bars represent SEM. C, Quantification of PSD protein levels for WT and PINK1 KO conditions, normalized to GAPDH and expressed as percentage of mean WT protein level. Error bars represent SEM. D, Images of 15 DIV hippocampal neurons expressing control, shParkin, or rescue constructs and immunostained for Homer1 and VAMP2. Arrows indicate representative synaptic Homer1 puncta (based on colocalization with VAMP2) for each condition. White rectangular boxes indicate magnified regions for insets. Scale bars, 10 μm for both larger images (white) and insets (black). E, Quantification of Homer1 intensity at synapses, defined as Homer1/VAMP2-immunopositive puncta along dendrites (as in Fig. 2), expressed as a fraction of control (n ≥ 5 fields of view per condition with >100 VAMP2/Homer1 puncta per field, results confirmed in 3 independent experiments; ***p < 0.001, unpaired t test).

Figure 6.

Homer1 overexpression rescues cell-surface GluA1 in Parkin-deficient neurons. A, Images of 15 DIV hippocampal neurons cotransfected with mCherry+/− shParkin alone or with GFP-Homer1 and immunostained for surface GluA1. Scale bar, 10 μm. B, Quantification of cell-surface GluA1, expressed as a fraction of mCherry control (n ≥ 20 fields of view per condition with >50 GluA1 puncta per field, results confirmed in 2 independent experiments; ***p < 0.0001, unpaired t test). Error bars represent SEM. C, Immunoblot of lysates from HEK293T cells expressing Myc-Parkin and either soluble GFP or GFP-Homer1, immunoprecipitated (IP) with Myc antibody, and probed with Myc or GFP antibodies. The arrow indicates GFP-Homer1, and the arrowhead indicates Myc-Parkin (just below IgG band).

Figure 7.

Reduced endocytic zone density in Parkin-deficient neurons. A, Images of 15 DIV hippocampal neurons cotransfected with SAP102-GFP+/− shParkin (green) and HA-Dynamin-3 (red). Arrows indicate colocalized puncta. Scale bar, 10 μm. B, Quantification of SAP102/Dynamin-3 colocalization (n ≥ 20 fields of view per condition from 3 independent experiments, 30–100 SAP102 puncta/field; ***p < 0.001, unpaired t test). C, Quantification of Dynamin-3 puncta/unit length dendrite, expressed as a fraction of control (n ≥ 20 fields of view per condition from 3 independent experiments, 20–120 HA-Dynamin puncta per field; *p < 0.05, unpaired t test). D) Images of 15 DIV hippocampal neurons from WT and KO rats, cotransfected with SAP102-GFP and HA-Dynamin-3. Arrows indicate colocalized puncta. Scale bar, 10 μm. E, Quantification of SAP102/Dynamin-3 colocalization (n ≥ 11 fields of view, 30–150 SAP102 puncta per field; ***p < 0.001, unpaired t test). F, Quantification of Dynamin-3 puncta/unit length dendrite, expressed as a fraction of wild-type control (n ≥ 11 fields of view, 15–115 HA-Dynamin puncta per field; *p < 0.05, unpaired t test). G, Quantification of SAP102 puncta/unit length dendrite, expressed as a fraction of wild-type control (n ≥ 11 fields of view, 30–150 SAP102 puncta per field, unpaired t test). H, Immunoblot of primary hippocampal neurons cultured from E18 WT and Parkin KO rats, collected on 14 DIV, and probed with Parkin and tubulin antibodies. Note the absence of Parkin from the KO lysate. I, Immunoblot of lysates from HEK293T cells expressing Myc-Parkin and HA-Dynamin3, immunoprecipitated (IP) with IgG (control) or Myc antibody, and probed with HA or Myc antibodies. The arrowhead indicates Myc-Parkin (just below the IgG band).

Figure 8.

Impaired AMPAR internalization but not recycling in Parkin-deficient neurons. A, Images of 15 DIV hippocampal neurons expressing soluble GFP+/− shParkin, immunostained for cell-surface (surf) and internalized (inter) GluA1 after 0, 15, or 30 min incubation at 37°C. Scale bars, 10 μm. B, Quantification of GluA1 internalization at 15 min, expressed as the ratio of internalized to cell-surface GluA1 and normalized to GFP control condition (n ≥ 10 fields of view per condition, >50 GluA1 puncta per field, results confirmed in 3 independent experiments; ***p < 0.0001, unpaired t test). C, Quantification of GluA1 internalization at 30 min, expressed as the ratio of internalized to cell-surface GluA1 and normalized to the GFP control condition (n ≥ 10 fields of view per condition, >50 GluA1 puncta per field, results confirmed in 3 independent experiments; ***p < 0.0001, unpaired t test). D, Images of 15 DIV hippocampal neurons expressing soluble GFP+/− shParkin and immunostained for internalized (inter) or recycled (rec'd) GluA1 after 0 or 60 min incubation at 37°C. Scale bars, 10 μm. E, Quantification of GluA1 recycling at 60 min, expressed as the ratio of recycled to internalized GluA1 and normalized to the GFP control condition (n ≥ 10 fields of view per condition, >50 GluA1 puncta per field, results confirmed in 3 independent experiments; p < 0.0001, unpaired t test). F, Model of how Parkin deficiency may disrupt postsynaptic AMPAR recycling. In wild-type neurons, Parkin stabilizes Homer1 at the PSD, allowing for PSD/EZ coupling via Homer1/Dynamin-3 interactions and thereby supporting efficient AMPAR capture, internalization, and reinsertion at the cell surface (Lu et al., 2007; Petrini et al., 2009). Parkin-deficient synapses have decreased Homer1 levels, leading to reduced PSD/EZ coupling, impaired AMPAR capture and internalization, and ultimately the diffusion of cell-surface AMPARs away from synapses.