Recombinant optogenetic and chemogenetic proteins that manipulate neuronal activity are potent tools for activating and inhibiting neuronal circuit function. However, there are few analogous tools for manipulating the structure of neural circuits. Here, we introduce three rationally designed genetically encoded tools that use E3 ligase-dependent mechanisms to trigger the degradation of synaptic scaffolding proteins, leading to functional ablation of synapses. First, we developed a constitutive excitatory synapse ablator, PFE3, analogous to the inhibitory synapse ablator GFE3. PFE3 targets the RING domain of the E3 ligase Mdm2 and the proteasome-interacting region of Protocadherin 10 to the scaffolding protein PSD-95, leading to efficient ablation of excitatory synapses. In addition, we developed a light-inducible version of GFE3, paGFE3, using a novel photoactivatable complex based on the photocleavable protein PhoCl2c. paGFE3 degrades Gephyrin and ablates inhibitory synapses in response to 400 nm light. Finally, we developed a chemically inducible version of GFE3, chGFE3, which degrades inhibitory synapses when combined with the bio-orthogonal dimerizer, HaloTag ligand-trimethoprim. Each tool is specific, reversible, and capable of breaking neural circuits at precise locations.
Neural circuits, groups of neurons connected by synapses, are the basic units of computation in the brain and are fundamental to understanding its function and pathology. Although there are many genetically encoded tools for modulating neuronal function, existing tools for eliminating synapses are inadequate for providing a clear understanding of neural circuits. Previously, we developed GFE3, a protein that can ablate inhibitory synapses efficiently, rapidly, and without toxicity by targeting an E3 ligase to the scaffolding protein Gephyrin[1]. GFE3 consists of GPHN.FingR (Fibronectin intrabody generated by mRNA display), a recombinant, antibody-like protein, which binds Gephyrin at inhibitory synapses[2],[3] fused to the RING domain of the E3 ligase XIAP (RING XIAP) [4]. GFE3 mediates the ubiquitination of Gephyrin and disassembly of inhibitory synapses.
Importantly, it specifically targets postsynaptic sites containing GABA A receptors and can be expressed in genetically determined cell types. Thus, GFE3 can manipulate circuits in a manner that is not possible with optogenetics[5], DREADDs[6], and modulators of neurotransmitter release[7], which don’t target specific postsynaptic receptors, or with traditional pharmacological approaches that target specific receptors but do not allow for cell-type specificity. GFE3 has been used to probe the contribution of inhibitory inputs to LTP in hippocampal circuits[8], recurrent loops in the oscillator that drives rhythmic whisking[9],[10], and the temporal coordination of vocalization and inspiration[11]. However, there currently exists no tool analogous to GFE3 that ablates excitatory synapses. Furthermore, because GFE3 is constitutively active, its temporal and spatial resolution is limited.
In this study, we generated three tools for ablating synapses based on GFE3: 1. PFE3, an excitatory synapse ablator; 2. paGFE3, a photoactivatable version of GFE3; 3. chGFE3, a chemically activated version of GFE3. We show that the expression of PFE3 in cultured neurons causes the loss of PSD-95 puncta and excitatory synapses, which is reversible. In vivo, its expression leads to the functional loss of excitatory synaptic transmission. We generated paGFE3 by incorporating GPHN.FingR and the RING XIAP into a novel photoactivatable complex based on the photocleavable protein PhoCl2c[12],[13]. paGE3 is activated by 400 nm light, causing ablation of inhibitory synapses within 5 hr of exposure that is subsequently reversible. In the absence of 400 nm light, paGFE3 has no background activity. In addition, the expression of paGFE3 labels inhibitory synapses, allowing their size and location to be monitored before and after ablation. chGFE3, a chemogenetic version of GFE3 analogous to paGFE3, mediates efficient, reversible degradation of labeled synapses when a cell-permeant chemical is added.
Initially, to generate an excitatory synapse ablator, we fused RING XIAP to PSD-95.FingR, which binds at high affinity and specificity to the scaffolding protein PSD-95, which is found at excitatory synapses[14]. However, PSD-95.FingR-RING XIAP did not efficiently ablate PSD-95 (data not shown). As an alternative, we fused PSD-95.FingR to the RING domain of the E3 ligase Mdm2 (RING Mdm2), which is necessary for PSD-95 degradation and excitatory synapse ablation during homeostatic plasticity[15]. To determine whether PSD-95.FingR-RING Mdm2 degrades PSD-95 we examined how co-expression with PSD-95.FingR-RING Mdm2 affected the expression of exogenous PSD-95-myc in COS7 cells. Cells transfected with PSD-95 alone had a PSD-95-myc expression level of 203 ±32 au, as measured by western blot, while the expression level of PSD-95 in cells co-expressing PSD-95-myc and PSD-95.FingR-RING Mdm2 reduced to 107 ± 12 au, a ∼50% decrease (n = 5, p < 0.05, ANOVA with multiple comparisons). To determine whether this reduction in PSD-95 was due to ubiquitination, we tried a third condition, where the ubiquitination inhibitor TAK243[16] was added to cells co-expressing PSD-95-myc and PSD-95.FingR-RING Mdm2. In this case, PSD-95 expression levels increased to 239 ± 31 au, a ∼20% increase relative to cells expressing PSD-95-myc alone, which was not significant (n = 5, p > 0.6, ANOVA with multiple comparisons). Furthermore, cells co-expressing PSD-95-myc and PSD-95.FingR-RING Mdm2 without TAK243 had ∼55% less PSD-95 vs. those with TAK243, a significant difference (p < 0.05, ANOVA with multiple comparisons, Fig. 1A, B, S1B).
Figure 1:Mdm2.RING ubiquitinates PSD-95.
A) Comparison of cell lysate from COS7 cells transfected with PSD-95-myc alone, PSD-95-myc + Dox-induced TRE-PSD-95.FingR-RING MDM2, and PSD-95-myc + PSD-95.FingR-RING MDM2 + TAK243, a ubiquitination inhibitor. Cells expressing TRE-PSD-95.FingR-RING MDM2 were treated with 1μg/ml Doxycycline for 4 hr to induce expression of PSD-95.FingR-RING and showed a reduction in PSD-95-myc. Cells treated with 20 μM TAK243 showed no significant reduction in PSD-95-myc. * p < 0.05, ns p > 0.6 Quantitation showed a significant reduction in PSD-95 expression when co-expressed with PSD-95.FingR-RING MDM2 in COS7 cells, but not when PSD-95 is expressed alone or when PSD-95 and PSD-95.Fn-RING MDM2 are co-expressed with 20 μM TAK243. * p < 0.05, ANOVA with multiple comparisons. ns, p > 0.05.
B) Cultured cortical neuron expressing transcriptionally regulated PSD-95.FingR-tagRFP before induction of PSD-95.FingR-HA-RING Mdm2 expression with Dox.
C) Same neuron as in C) after induction of PSD-95.FingR-HA-RING MDM2 expression with Dox shows a reduction in PSD-95.FingR-tagRFP labeling.
D) Immunostaining of the neuron in D) for PSD-95.FingR-HA-RING MDM2 (green) and endogenous PSD-95 (red).
E) Quantification of the number of PSD-95 puncta and the total amount of PSD-95 labeled by PSD-95.FingR-tagRFP before and after expression of PSD-95.FingR-HA-RING MDM2. ** p < 0.01
Error bars represent ± sem Scale bars 5 µm.
To test whether TAK243 was blocking ubiquitination, we compared western blot staining of lysates from COS7 cells expressing Ubiquitin-HA with and without TAK243 (Fig. S1A). Staining with anti-HA showed a distinctive laddering pattern in the lane corresponding to cells expressing Ubiquitin-HA without TAK243 consistent with ubiquitination, whereas the lanes corresponding to cells expressing Ubiquitin-HA with TAK243 and a control lane with lysate from untransfected cells showed no staining, confirming that TAK243 blocks ubiquitination. Together, our results are consistent with PSD-95.FingR-RING Mdm2 degrading exogenously expressed PSD-95 through ubiquitination in COS7 cells.
To test whether PSD-95.FingR-RING Mdm2 can degrade endogenous neuronal PSD-95, we co-transfected doxycycline (Dox)-inducible TRE-PSD-95.FingR-HA-RING Mdm2 and transcriptionally regulated PSD-95.FingR-tagRFP in 14 DIV (days in vitro) cultures of rat cortical neurons. Note that PSD-95.FingR-tagRFP efficiently labels endogenous PSD-95, allowing its spatial distribution to be mapped in real-time in living cells[3]. Furthermore, transcriptional regulation matches the expression level of PSD-95.FingR with that of endogenous PSD-95, facilitating labeling with very low background. Following incubation for four days, we imaged the neurons for PSD-95.FingR-tagRFP and subsequently induced the expression of PSD-95.FingR-HA-RING Mdm2 with Dox. After 48 hr, we reimaged the neurons for PSD-95.FingR-tagRFP and then fixed and stained them with anti-PSD-95 and anti-HA. We found that PSD-95.FingR-tagRFP labeling was reduced by 85 ± 2% (p = 0.002, Wilcoxon, n = 9 cells, 3 independent experiments, Fig. 1C, D, F), consistent with efficient ablation of endogenous PSD-95. However, when we counted the number of puncta labeled with PSD-95.FingR at T0 and compared that to the number of puncta labeled with immunostaining of endogenous PSD-95 at the end of the experiment, it showed a reduction of only 52 ± 8% (Fig. 1E, F p = 0.002, Wilcoxon, n = 9 cells, 3 independent experiments). A strategy for improving this relatively low ablation rate might be provided by the results of experiments where the expression of MEF2 was found to cause the elimination of excitatory synapses[17]. In those experiments, efficient synapse elimination was found to require a combination of ubiquitination mediated by Mdm2 and interaction with the proteasome, which was mediated by Protocadherin 10 (PCDH 10). PCDH 10 is a Ca 2+-dependent cell adhesion protein that binds to both PSD-95 and the proteasome via its proteasome interacting region (PIR)[17]. Therefore, we reasoned that adding the PIR domain to the PSD-95.FingR-RING Mdm2 complex might increase the efficiency with which PSD-95 is ablated.
We generated a new protein, PSD-95.FingR-RING Mdm2-PIR, which we called PFE3. To test PFE3, we co-transfected TRE-PFE3-HA with PSD-95.FingR-tagRFP in dissociated cultures of rat cortical neurons. Following four days of incubation, we imaged the PSD-95.FingR-tagRFP and induced the expression of PFE3-HA with Dox. The neurons were imaged 48 hr after induction of TRE-PFE3 and subsequently fixed and immunostained for endogenous PSD-95 and HA (Fig. 2A-C). By comparing images at T0 and 48 hr, we found that the expression of PFE3 reduced the labeling of PSD-95.FingR-tagRFP by 65 ± 4%, a significant difference (Fig. 2A, B, D, n = 14 cells, 3 distinct experiments, p = 0.0001, Wilcoxon). When we checked the
