Functional specificity of liquid-liquid phase separation at the synapse

Abstract

The mechanisms that enable synapses to achieve temporally and spatially precise signaling at nano-scale while being fluid with the cytosol are poorly understood. Liquid-liquid phase separation (LLPS) is emerging as a key principle governing subcellular organization; however, the impact of synaptic LLPS on neurotransmission is unclear. Here, using rat primary hippocampal cultures, we show that robust disruption of neuronal LLPS with aliphatic alcohols severely dysregulates action potential-dependent neurotransmission, while spontaneous neurotransmission persists. Synaptic LLPS maintains synaptic vesicle pool clustering and recycling as well as the precise organization of active zone RIM1/2 and Munc13 nanoclusters, thus supporting fast evoked Ca²⁺-dependent release. These results indicate although LLPS is necessary within the nanodomain of the synapse, the disruption of this nano-organization largely spares spontaneous neurotransmission. Therefore, properties of in vitro micron sized liquid condensates translate to the nano-structure of the synapse in a functionally specific manner regulating action potential-evoked release.

Fig. 1. Action potential-dependent neurotransmission is disrupted following LLPS manipulation.

a Nuclear speckle staining (images are maximum intensity projections from z-stacks). b, Number of nuclear speckles per 10 µm² of DAPI, c nuclear speckle area and (d) nuclear speckle fluorescence intensity (control n = 15, 1 min n = 21, 10 min n = 14 ROIs). e Representative traces of eEPSCs during 3% 1,6-HD wash on, first trace in black. f eEPSC amplitude during treatment (n = 10 cells). g Representative traces of eIPSCs during 3% 1,6-HD wash on, first trace in black. h eIPSC amplitude during treatment (n = 10 cells). i Representative traces of eEPSCs during 3% 1,5-PD wash on, first trace in black. j eEPSC amplitude during treatment (n = 10 cells). k Representative traces of eIPSCs during 3% 1,5-PD wash on, first trace in black. l eIPSC amplitude during treatment (n = 12 cells). m Representative traces of eEPSCs during 3% 2,5-HD wash on, first trace in black. n eEPSC amplitude during treatment (n = 10 cells). o Example traces of eIPSCs during 3% 2,5-HD wash on, first trace in black. p eIPSC amplitude during treatment (n = 11 cells). q Sample traces of eIPSC kinetics following 3% aliphatic alcohol treatment, first stimulation in black. r Analysis of eIPSC rise times (20-80%) with 3% aliphatic alcohol treatment during wash-on for ten stimulations, followed by a 20 stimulation wash-off period (stimulation events not included are in Supplementary Fig. 1j–l, to compare same treatment length) (1,6-HD n = 10, 1,5-PD n = 12, 2,5-HD n = 11 cells). s Comparison of eIPSC rise times for stimulation #6-10 (period when see robust amplitude effect) during 3% aliphatic alcohol treatment (1,6-HD n = 50, 1,5-PD n = 60, 2,5-HD n = 55 stimulation events). Values are mean ± SEM. Significance reported as: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Statistical analyzes conducted were one-way Kruskal-Wallis test with Dunn’s multiple comparisons test (b, s), one-way ANOVA with Dunnett’s multiple comparison test (c, d), one-way Friedman test with Dunn’s multiple comparisons test comparing amplitude to the 1^st stimulation every ten stimulations (f, h, j, l, n, p). Source data are provided as a source data file. Exact p-values reported in Supplementary Data 1.

Fig. 2. Spontaneous neurotransmission persists despite LLPS disruption.

a Example traces of mEPSCs at baseline, 2 min 3% 1,6-HD treatment, and during wash off. b Analysis of mEPSC frequency normalized to baseline, c mEPSC amplitude normalized to baseline, d mESPC event rise times (10–90), and e mEPSC event decay times (n = 11 cells). f Example traces of mIPSCs at baseline, 2 min 3% 1,6-HD treatment, and during wash off. g Analysis of mIPSC frequency normalized to baseline, h mIPSC amplitude normalized to baseline, i mIPSC event rise times (10–90), and j mIPSC event decay times (n = 8 cells). k Example traces of mEPSCs at baseline, 2 min 3% 1,5-PD treatment, and during wash off. l Analysis of mEPSC frequency normalized to baseline, m mEPSC amplitude normalized to baseline, n, mESPC event rise times (10–90), and o, mEPSC event decay times (n = 11 cells). Values are mean ± SEM. Significance reported as: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Statistical analyzes conducted were one-way Friedman test with Dunn’s multiple comparisons test (b, c, g, h, j, l–n) and one-way ANOVA with Tukey’s multiple comparison test (d, e, i, o). Source data are provided as a source data file. Exact p-values reported in Supplementary Data 1.

Fig. 3. Genetic augmentation of spontaneous neurotransmission during LLPS disruption.

a Example mEPSC traces at baseline, immediately following 3% 1,6-HD application, after 20 minutes, and 40 minutes of 3% 1,6-HD exposure to the same cell. b Analysis of mEPSC frequency over time with continued exposure to 3% 1,6-HD (1,6-HD n = 5, control n = 5 cells). c Schematic of fusing synaptic vesicle with V-ATPase, Synaptotagmin 1 (Syt1), and Synaptosomal-Associated Protein 25 kDa (SNAP25) variants L50S and D166Y (Created in BioRender. Guzikowski, N. (2024)). d Example mEPSC traces of four groups including empty vector (control), Synaptotagmin 1 knockdown (Syt1 KD), SNAP25 L50S variant, and SNAP25 D166Y variant. e Analysis of baseline mEPSC frequency of Syt1 KD and two SNAP25 genetic variants (Empty Vector n = 9, Syt1 KD n = 8, L50S n = 7, D166Y n = 12 cells, same data used for baseline in 1,6-HD treatment experiments). f–n Across the three different genetic backgrounds representative traces and quantification of 2 min 3% 1,6-HD treatment on mEPSC frequency and amplitude (Syt1 KD n = 8, L50S n = 7, D166Y n = 12 cells). o Example mEPSC traces of control (DMSO), 10 minute folimycin pre-treatment (also included in the bath), and 10 minute 1,6-HD treatment in the D166Y SNAP25 genetic background. p Analysis of mEPSC frequency and q mEPSC amplitude (control n = 9, folimycin n = 8, 1,6-HD n = 9 cells). Values are mean ± SEM. Significance reported as: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Statistical analyzes conducted were unpaired two tailed t-test or two tailed t-test Mann-Whitney test performed between groups at baseline, 4 min, 20 min, and 40 min intervals (b), one-way Kruskal-Wallis test with Dunn’s multiple comparisons test (e, p), one-way ANOVA with Dunnett’s/Tukey’s multiple comparison test (g, h, m, n, q), one-way Friedman test with Dunn’s multiple comparisons test (j, k). Source data are provided as a source data file. Exact p-values reported in Supplementary Data 1.

Fig. 4. Synaptic vesicle pool clustering dynamics altered with liquid condensate melting.

a Diagram of vesicle pools within the pre-synapse (Created in BioRender. Guzikowski, N. (2023)). b Representative traces of excitatory RRP upon 0.5 M sucrose perfusion. c Peak amplitude of 0.5 M sucrose response (n = 9 cells). d Representative traces of inhibitory RRP upon 0.5 M sucrose perfusion. e Peak amplitude of 0.5 M sucrose response (n = 10 cells). f Representative image of dSTORM glutamate SV pool clusters along an axon. g Single cluster of dSTORM VGluT single molecule localizations representing the glutamate containing SV pool at an individual bouton. h Cumulative frequency histogram of VGluT cluster volume for all synapses with inlaid bar graph of ROI averages (control n = 12, 1,6-HD n = 10 ROIs). i Cumulative frequency histogram of VGluT cluster sphericity of all synapses with inlaid bar graph of ROI averages (control n = 12, 1,6-HD n = 10 ROIs). j eEPSC Paired Pulse Ratio (PPR) and (k) eEPSC amplitude during repetitive stimulation (only evoked currents large enough to differentiate from stimulation artifact and baseline activity included in analysis, 1^st and 2^nd stim data from Fig. 4k used for PPR) (control n = 6, 1,6-HD n = 4 cells). l Single cluster of dSTORM VGAT single molecule localizations representing the GABA containing SV pool at an individual bouton. m Cumulative frequency histogram of VGAT cluster volume of all synapses with an inlaid bar graph of ROI averages (control n = 12, 1,6-HD n = 11 ROIs). n Cumulative frequency histogram of VGAT cluster sphericity of all synapses with an inlaid bar graph of ROI averages (control n = 12, 1,6-HD n = 11 ROIs). o eIPSC PPR and (p) eIPSC amplitude during repetitive stimulation (only evoked currents large enough to differentiate from stimulation artifact and baseline activity included in analysis, 1^st and 2^nd stim data from Fig. 4p used for PPR) (control n = 8, 1,6-HD n = 7 cells). Values are mean ± SEM. Significance reported as: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Statistical analyzes conducted were two-tailed Mann-Whitney test (c), two-tailed unpaired t test (e, h, i, j, m–o), and Kolmogorov-Smirnov test (h, i, k, m, n, p). Source data are provided as a source data file. Exact p-values reported in Supplementary Data 1.

Fig. 5. Necessity of liquid condensates in active zone nanostructure for calcium-dependent release.

a Experimental set-up for 1,6-HD included in patch pipette internal solution (Created in BioRender. Guzikowski, N. (2023)). b Example traces of eIPSCs. c Analysis of eIPSC amplitude when aliphatic alcohols are included in the internal solution (control n = 8, 3% n = 5, 10% n = 4 cells, stats performed at minute interval). d Examples traces of whole cell voltage clamp during 45 mM K⁺ perfusion. e Cumulative charge transfer during 45 mM K⁺ perfusion following 1 min control or 3% 1,6-HD pre-treatment (with TTX, APV, Bicuculline) (control n = 8, 1,6-HD n = 9 cells). f Example images of axons transfected with Syb2-GCaMP8s. Arrows denote pre-synaptic boutons identified with 90 mM K⁺. Insets below represent fluorescence intensity peaks at individual boutons. g Peak fluorescence response (n = 687 boutons). h Example fluorescence trace (ROI average of all boutons). i Paired comparison of eIPSC amplitude (n = 12 cells, same recordings used in Fig. 5j, k). j eIPSC amplitude during 3% 1,6-HD wash-on and wash-off (statistical analysis comparing nonlinear fit during drug wash on and average amplitudes compared between groups at every 10^th stim) (wash-off is 2 mM Ca²⁺ for both groups) (2 mM Ca²⁺ n = 12, 8 mM Ca²⁺ n = 12 cells). k Analysis of eIPSC rise times (20-80%) (1^st 10 stim of treatment or baseline per cell included). l dSTORM Munc13 single molecule localizations at an individual bouton color coded based on density heat map. m Munc13 nanocluster diameters (Control n = 129, 1,6-HD n = 137 synapses) and (n) number of munc13 nanoclusters per synapse (Control n = 121, 1, 6-HD n = 132 synapses). o Normalized pair correlation function (g(r)) of an individual synapse. p dSTORM RIM1/2 single molecule localizations at an individual bouton color coded based on density heat map. q RIM1/2 nanocluster diameters (Control n = 146, 1,6-HD n = 134 synapses) and r, number of RIM1/2 nanoclusters per synapse (Control n = 132, 1,6-HD n = 121 synapses). s, Normalized pair correlation function (g(r)) of an individual synapse. Values are mean ± SEM. Significance reported as: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Statistical analyzes conducted were one-way ANOVA with Tukey’s multiple comparison test (c) Simple linear regression (e) one-way Friedman test with Dunn’s multiple comparisons test (g) two-tailed paired t-test (i) two-tailed Mann-Whitney test (j, m, n, q, r) two-tailed unpaired t-test (j), Nonlinear fit (one phase decay) comparison of fit—extra sum-of-squares F test (j) and one-way Kruskal-Wallis test with Dunn’s multiple comparisons test (k). Source data are provided as a source data file. Exact p-values reported in Supplementary Data 1.

Fig. 6. Phase separation of RIM1 regulates the efficacy and integrity of fast evoked calcium-dependent release.

a Example image of RIM1 western blot. b RIM1 protein expression normalized to GDI loading control and scramble (ctrl) (A and B RIM1-KD constructs combined for all further experiments) (ctrl n = 7, RIM1-KD-A n = 4, RIM1-KD-B n = 4 samples, quantitative comparisons were made between samples on different gels/blots.). c Synapse staining (images are maximum intensity projections from z-stacks). d Fluorescence intensity of synaptic puncta (colocalization of RIM1/2 and synapsin1) (ctrl n = 12, RIM1α OE n = 12 ROIs). e Representative traces of basal eIPSCs. f eIPSC amplitude (ctrl n = 32, RIM1 KD n = 25, RIM1α OE n = 10 cells). g Representative traces of eIPSCs during 3% 1,6-HD wash on, first trace in black. h eIPSC amplitude during treatment (comparing nonlinear fit during drug wash on and average amplitudes compared between groups at every 10^th stim, ctrl n = 17, RIM1 KD n = 9, RIM1α OE n = 10 cells). i Sample traces of eIPSC kinetics following 3% 1,6-HD treatment, first stimulation in black. j Analysis of eIPSC rise times (20-80%) (1^st 10 stim of treatment or baseline per cell included) (ctrl n = 170, RIM1 KD n = 90, RIM1α OE = 100 stimulation events). k Representative traces of eIPSCs during treatment, first trace in black. l eIPSC amplitude during treatment (comparing nonlinear fit during drug wash on and average amplitudes compared between groups at every 10^th stim, ctrl n = 9, RIM1 KD n = 9 cells). m Analysis of eIPSC rise times (20–80%) (1^st 10 stim of treatment or baseline per cell included) (ctrl n = 90, RIM1 KD n = 90 stimulation events). n Summary figure of hypothesized pre-synaptic organization; vesicle pool organization relies on LLPS for adequate exocytosis and endocytosis dynamics, evoked release is located within active zone nano-cluster liquid condensates, necessary for VGCC clustering, whereas spontaneous release is located outside liquid RIM1 nano-clusters (Created in BioRender. Guzikowski, N. (2023)). Values are mean ± SEM. Significance reported as: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Statistical analyzes conducted were one-way ANOVA with Dunnett’s/Tukey’s multiple comparison test (b, f), two-tailed Unpaired t test (d, l), one-way Kruskal-Wallis test with Dunn’s multiple comparisons test (m, h, j), Nonlinear fit (one phase decay) comparison of fit – extra sum-of-squares F test (h, l), and two-tailed Mann Whitney test (l). Source data are provided as a source data file. Exact p-values reported in Supplementary Data 1.