PSD95 nanoclusters are postsynaptic building blocks in hippocampus circuits

The molecular features of synapses in the hippocampus underpin current models of learning and cognition. Although synapse ultra-structural diversity has been described in the canonical hippocampal circuitry, our knowledge of sub-synaptic organisation of synaptic molecules remains largely unknown. To address this, mice were engineered to express Post Synaptic Density 95 protein (PSD95) fused to either eGFP or mEos2 and imaged with two orthogonal super-resolution methods: gated stimulated emission depletion (g-STED) microscopy and photoactivated localisation microscopy (PALM). Large-scale analysis of ~100,000 synapses in 7 hippocampal sub-regions revealed they comprised discrete PSD95 nanoclusters that were spatially organised into single and multi-nanocluster PSDs. Synapses in different sub-regions, cell-types and locations along the dendritic tree of CA1 pyramidal neurons, showed diversity characterised by the number of nanoclusters per synapse. Multi-nanocluster synapses were frequently found in the CA3 and dentate gyrus sub-regions, corresponding to large thorny excrescence synapses. Although the structure of individual nanoclusters remained relatively conserved across all sub-regions, PSD95 packing into nanoclusters also varied between sub-regions determined from nanocluster fluorescence intensity. These data identify PSD95 nanoclusters as a basic structural unit, or building block, of excitatory synapses and their number characterizes synapse size and structural diversity.

a. Schematic diagram of the gene targeting strategy. Wild type gene structure is shown including terminal exon (red), targeting vector and targeted allele. The neomycinresistance cassette was excised by Cre recombination. Dark boxes, exons; green box, mEos2; red triangles, loxP sites; DTA, diphtheria toxin cassette; neo, neomycinresistance cassette. b. Low magnification (20×) scanning of hemi-coronal brain sections from (left to right) a WT brain section (negative) and homozygous PSD95-eGFP mouse, heterozygous PSD95-mEos2 mouse and a WT section immunostained with an anti-PSD95 antibody. WT and PSD95-eGFP sections captured at the same illumination intensity and exposure time for comparison; PSD95-mEos2 and PSD95 antibody contrast enhanced for expression pattern comparison. Scale bar 2 mm. c. High magnification (100×) confocal imaging of PSD95-eGFP, PSD95-mEos2 and PSD95 antibody in three hippocampal subregions Scale bar 2 µm. d. High magnification in the CA1 SR of PSD95-mEos2 co-stained for synaptophysin revealing pre and postsynaptic puncta, with example inset images of synapses. Scale bar 1 µm, inset scale bars 200 nm. e. High magnification in the CA1 SR of PSD95-eGFP co-stained for synaptophysin revealing pre and postsynaptic puncta, with example inset images of synapses. Scale bar 1 µm, inset scale bars 200 nm. f. Immunoblot for PSD95 with tubulin loading control from crude synaptosome extracts of WT, heterozygous and homozygous PSD95-mEos2 mice. Schematic diagrams of the Dlg4 gene encoding the PSD95 protein. mEos2 is targeted to the C-terminal locus of PSD95. Neomycin cassette is removed from the genome following crerecombination. g. Baseline synaptic transmission was significantly affected by genotype (F(2, 22.078) = 16.21; P = 0.00005). Input-output relationships illustrate averaged peak fEPSP amplitudes in slices from PSD-95 mEos2/mEos2 (n = 24 slices; N = 7 mice), PSD95 +/mEos2 (n = 34; N = 11) and WT mice (n = 34; N = 9) in response to stimulation of Schäffer collaterals by biphasic voltage pulses of 0. f. FWHMs of capsids were measured to assess resolution by the size of capsids imaged with different depletion laser powers. Note that 0% STED laser power condition was a confocal image without gating, while other images included gating 2-8 ns. n = number of capsids analysed.
g. Gating delay was tested on different capsids to assess the impact of resolution.
Scale bar 500 nm h. FWHMs were measured to assess the impact on resolution.
Supplementary Figure 3. Assessing effective resolution of PALM on PSD95-mEos2 brain sections.
a-c. Localisation precision error plotted in histograms from images from CA1 SO in 3 different mice. N = number of localisation events within a whole image sequence. Small differences were noted in the mean average localisation precision averaged across 3 images from within each mice (mean ± standard deviation). Taking into account the precision fitting and the mean average NN value for each detected NC, the effective lateral resolution for PSD95-mEos2 imaging can vary between brain samples from 50 to 60 nm.  a. Example of a dendrite from a dye filled pyramidal neuron from cortical layer II/III, CA1 pyramidal neuron dendrite in the CA1 SR , CA3 pyramidal neuron dendrite in the CA3 SL and a granule cell dendrite in the DG ML . Scale bars 2 µm. b.

Supplementary Figure 4. FWHM analysis of PSD95 structures, and validation of Imaris
Spine head diameter positively correlated with the number of PSD95-eGFP NCs expressed within the spine. Quantifications pooled from across all spines of neurons. Kruskall-Wallis test with pairwise comparison test confirmed statistically significant differences in spine diameters. Figure 9. Mean fluorescence intensities and localisation densities.

Supplementary
a. Bar chart of the mean fluorescence intensity of PSD95-eGFP PSDs from each subregion (colour coded as per Fig.2a) b. Bar chart of the mean fluorescence intensity of PSD95-eGFP NCs from each subregion (colour coded) and each mouse brain section (1-3). Relative differences between sub-regions were observed in each of the three mouse brain sections.
c. Bar chart of the mean density of PSD localisations from each sub-region in PSD95-mEos2 brain sections.
d. Bar chart of the mean density of NC localisations from each sub-region in PSD95-mEos2 brain sections.
e. Bar chart of the mean density of PSD localisations from each sub-region in PSD95-mEos2 mouse brain section (1-3). Variability between the samples masks any significant trends of sub-regional differences in localisations per NC.