High-affinity detection of biotinylated endogenous neuroligin-1 at excitatory and inhibitory synapses using a tagged knock-in mouse

Abstract

Neuroligins (NLGNs) are important cell adhesion molecules mediating trans-synaptic contacts between neurons. However, the high-yield biochemical isolation and visualization of endogenous NLGNs is hampered by the lack of efficient antibodies. Thus, to reveal their subcellular distribution, binding partners, and synaptic function, NLGNs were extensively manipulated using knock-down, knock-out, or overexpression approaches, leading to controversial results. As an alternative to the manipulation of NLGN expression level, we describe here the generation of a knock-in (KI) mouse strain in which native NLGN1 was N-terminally tagged with a small biotin acceptor peptide (bAP) that can be enzymatically biotinylated by the exogenous delivery of biotin ligase. After showing that KI mice exhibit normal behavior as well as similar synaptic number, ultrastructure, transmission properties, and protein expression levels when compared to wild type counterparts, we exploited the fact that biotinylated bAP-NLGN1 can be selectively isolated or visualized using high-affinity streptavidin conjugates. Using immunoblotting and immunofluorescence, we show that bAP-NLGN1 binds PSD-95 and gephyrin and populates both excitatory and inhibitory synapses, challenging the historical view that NLGN1 is exclusively localized at excitatory synapses. Using superresolution optical and electron microscopy, we further highlight that bAP-NLGN1 forms in the synaptic cleft a subset of nanodomains, which contain each a few NLGN1 dimers and whose number positively scales with the postsynapse size. Overall, our study not only provides an extensively characterized KI mouse model which will be available to the scientific community but also an unprecedented view of the nanoscale organization of endogenous NLGN1.

Keywords: biotinylation; cell adhesion molecule; knock-in mouse; neuroligin-1; synapse.

Fig. 1.

Behavioral characterization of KI mice. (A) Design of the bAP-NLGN1 KI mouse strain. (B) Body weight of WT and KI mice over time. (C–E) Behavioral assessment of WT and KI mice at different ages. Grouped analysis on WT and KI mice was made for each parameter (general aspect, sensory properties, locomotion, and anxiety). Data were compared by a two-way ANOVA, showing no significant (n.s.) difference between genotypes. (F) Schematics of the elevated plus maze test. (G) Representative heat maps showing the cumulated positions of representative WT and KI mice exploring their environment for 10 min. (H) Time spent in closed-arms. (I) Total distance traveled in apparatus. All values were obtained from 9 WT and 10 KI mice. Data are presented as mean ± SEM and were compared by Mann–Whitney tests (n.s.).

Fig. 2.

Brain development and synapse ultrastructure in KI mice. (A) Diagram of cortical slices ranging from Bregma −1.5 to −1.9 mm, used for immunohistochemistry. (B) Representative Nissl staining in sections from WT and KI mice. (C) Cell density in slices from either genotype (n.s.). (D and E) Immunohistochemical staining of VGLUT1 (red) and VGAT (magenta), respectively, performed on cortical slices from WT and KI mice. Fluorescence signal intensity (F and G) and surface area (H and I), of VGluT1 and VGAT staining, respectively, normalized to WT levels (in %). Data represent mean ± SEM of n = 8 hemispheres from 4 mice for each genotype. Statistical comparison was made by two-way ANOVA, showing no significant difference between WT and KI on all measurements. (J and K) Electron micrographs showing excitatory and inhibitory synapses from the cortex of WT and KI mice. Pre- and postsynaptic elements appear in yellow and blue, respectively, and the synaptic cleft is shown with arrowheads. (L–N) Surface area, number of neurotransmitter vesicles, and number of mitochondria in presynapses, respectively. (O) PSD size of excitatory synapses. Data represent mean ± SEM of 59 synapses from 2 WT mice, and 70 synapses from 2 KI mice (no significant difference between WT and KI for any of the parameters).

Fig. 3.

Synaptic transmission and protein expression levels in KI mice. (A) Representative voltage-clamp recordings performed on CA1 neurons in acute slices from WT or KI mice, upon stimulation of Schaffer’s collaterals. EPSCs were recorded sequentially in the same neuron at two holding potentials (−70 mV and +40 mV) to isolate AMPA- and NMDA-receptor mediated synaptic transmission, respectively. (B and C) Amplutides of NMDA-receptor and AMPA-receptor mediated EPSCs, respectively, versus electrical stimulation intensity. (D) Paired ratio between NMDA- and AMPA-receptor mediated EPSCs measured at a 40 V stimulation. (E) AMPA-receptor mediated EPSCs in response to two stimulations separated by 50 ms in slices from WT and KI mice. (F) Corresponding paired pulse ratio. (G) Traces of AMPA-receptor mediated spontaneous EPSCs in CA1 neurons. (H and I) Amplitude and frequency of those events in the two genotypes. Data are presented as mean ± SEM (n = 10 to 12 neurons from 3 independent slices) and analyzed by unpaired t test (n.s.). (J) Immunoblots of hippocampal cell lysates from WT and KI mice, β-actin being used as loading control. The molecular weight markers in kDa indicated on the Left. (K) Relative levels of various synaptic proteins, analyzed by semiquantitative immunoblotting (n = 4 to 5 independent cultures for both WT and KI conditions). Data represent mean ± SEM and were compared by individual Mann–Whitney tests for each protein (**P < 0.01, all other differences being n.s.) (L) Western blot of proteins extracted from cortical tissue of WT or KI mice at different developmental ages, and probed with NLGN1 antibody. Total protein staining was used as a loading control. (M) Corresponding NLGN1 expression level at each age. Dots in the bars represent individual animals (n = 4 from 2 independent experiments). Data represent the mean value and were compared by multiple Mann–Whitney tests (*P < 0.05).

Fig. 4.

Isolation of biotinylated bAP-NLGN1 and binding partners by streptavidin pull-down. (A) Schematics of NLGN1 containing the 15 aa N-terminal bAP tag that can be enzymatically biotinylated by BirA^ER, then labeled with streptavidin conjugates. (B) Hippocampal cultures from WT or KI mice were infected at DIV3-5 with BirA^ER-HA-IRES-GFP AAV1 (BirA+), or control IRES-GFP AAV1 (BirA-). (C and D) Cultures were lysed at DIV14, and protein extracts were either loaded as starting material or precipitated with streptavidin beads. Separated proteins were identified by Western blot with antibodies to HA, GFP, or NLGN1 (molecular weight markers in kDa indicated on the Left). (E) Cultures from KI mice infected with BirA+ or BirA- AAV1 were lysed at DIV14 and protein extracts were immunoblotted for NLGN2, NLGN3, PSD-95, and gephyrin. (F) Band intensity for the 3 NLGN isoforms in the BirA+ condition, normalized to the signal in the BirA- condition (value of 1). Dots in the bars represent individual cultures.

Fig. 5.

Detection of biotinylated bAP-NLGN1 with fluorescent streptavidin. (A) Dissociated hippocampal neurons from WT or KI mice infected with BirA+ or BirA- AAV1 at DIV3-5 (GFP reporter in green) were live labeled at DIV14 with SA-AF647 (magenta) and visualized by epifluorescence microscopy. (B) SA-AF647 intensity on GFP-expressing neurons in each condition. Dots show > 20 cells per condition from 3 independent experiments. Data represent mean ± SEM and were compared by a two-way ANOVA followed by Sidak’s multiple comparison test (****P < 0.0001). (C) Hippocampal neurons from KI mice electroporated at DIV0 with BirA^ER and shRNAs to either p53 (shCTRL) or NLGN1 (shNLGN1), both containing a GFP reporter (green), were labeled at DIV14 with SA-AF647 (magenta). (D) SA-AF647 signal on GFP-expressing neurons for each condition. Dots show 52 cells per condition from 2 independent experiments, and data represent mean ± SEM compared by a Mann–Whitney test (****P < 0.0001). (E) COS-7 cells electroporated with GFP-GPI or GFP-NRXN1β were added to DIV8 hippocampal cultures from KI mice previously infected with BirA+ AAV1, and after 48 h cocultures were labeled with SA-AF647 (magenta). (F) Fractional area occupied by SA-AF647 clusters on COS-7 cells, multiplied by the SA-AF647 intensity ratio between COS-7 cells and dendrites, in the two conditions. Dots show 22 to 38 cells per condition from 3 separate experiments, and data represent mean ± SEM compared by a Mann–Whitney test (****P < 0.0001).

Fig. 6.

Subcellular localization of biotinylated bAP-NLGN1 at excitatory and inhibitory synapses. (A) Dissociated hippocampal neurons from KI mice were coelectroporated at DIV0 with BirA^ER and intrabodies to PSD-95 and gephyrin (Xph20-mRuby2 and GPHN.FingR-GFP, respectively). (B) At DIV14, neurons were live-labeled with SA-AF647 and visualized by epifluorescence microscopy. Image of a dendritic arbor showing bAP-NLGN1 labeled with SA-AF647 (magenta), PSD-95 (cyan), and gephyrin (green) positive synapses. A specific segment (rectangle) is zoomed on the Right, and insets highlight individual synapses, with SA-AF647 clusters overlapping with PSD-95 (yellow arrows), or gephyrin (orange arrows) puncta. (C) Line scans across synapses showing NLGN1 clusters either in extrasynaptic locations (ExtraSyn), overlapping with PSD-95 (PSD-95+), with gephyrin (GPHN+), or with both PSD-95 and gephyrin (PSD-95+/GPHN+). (D) Distribution of bAP-NLGN1 clusters in the different compartments. (E and F) Enrichment of bAP-NLGN1 at PSD-95 or gephyrin puncta, and fraction of these puncta overlapping with bAP-NLGN1, respectively. Dots in the bars represent n = 39 cells for each condition, from 2 independent experiments. Data represent mean ± SEM and were compared by a Mann–Whitney test (*P < 0.05, ***P < 0.001). (G) Surface area of bAP-NLGN1 clusters as a function of their localization (n = 436 to 2224 clusters in 39 cells, from 2 independent experiments). Data were compared by a Kruskal–Wallis test followed by Dunn’s multiple comparison test (****P < 0.0001).

Fig. 7.

Superresolved localization of biotinylated bAP-NLGN1. (A–L) Dissociated hippocampal neurons from KI mice were electroporated at DIV0 with BirA^ER and either Xph20-GFP or GPHN.FingR-GFP. dSTORM experiments were performed at DIV14 on fixed cultures, after live-labeling neurons with SA-AF647. (A and B) Images of dendritic segments showing PSD-95 or gephyrin positive postsynapses (green) and the superresolved map of single bAP-NLGN1 molecule detections (gold). On the Right, individual examples of excitatory and inhibitory postsynapses. (C) Enrichment of bAP-NLGN1 localization density in PSD-95 or gephyrin puncta, or control extrasynaptic puncta, normalized to the average value on the dendritic shaft. Dots in the bars represent individual cells. Data represent mean ± SEM and were compared by a Kruskal–Wallis test followed by Dunn’s multiple comparison tests (****P < 0.0001). (D and E) Images of bAP-NLGN1 localizations constructed with SR-Tesseler, showing several nanodomains (gold) within PSD-95 and gephyrin puncta, respectively (white contours). The corresponding surface area of postsynaptic puncta is indicated. (F–H) Frequency distribution of the number, length, and distance from puncta center respectively, of bAP-NLGN1 nanodomains within PSD-95 and gephyrin puncta. (I and K) Average surface density of bAP-NLGN1 detections throughout the dendritic shaft, in postsynapses or in synaptic nanodomains, for either PSD-95 or gephyrin puncta. Dots represent individual cells, for each type of compartment. Data were compared by a Kruskal–Wallis test followed by Dunn’s multiple comparison test (****P < 0.0001). (J and L) Correlation between the bAP-NLGN1 nanodomain number and the PSD-95 or gephyrin surface area, respectively. Dots represent individual synaptic puncta. r, Pearson’s correlation coefficient; black line, linear regression with 95% CI (gray). All data related to PSD-95+ puncta were obtained from 4 independent experiments (1390 domains/1065 synaptic puncta/43 cells). Data related to GPHN+ puncta were obtained from 3 independent experiments (984 domains/631 synaptic puncta/36 cells). (M) Hippocampal neurons from KI mice electroporated with BirA^ER were cocultured with neurons from WT mice electroporated with EGFP + mEos3.2-NRXN1β (green), resulting in the formation of trans-synaptic contacts between axonal NRXN1 and dendritic NLGN1. At DIV14, cultures were live labeled with SA-AF647 (magenta), then fixed and processed for sequential dSTORM and PALM. (N) Hippocampal neurons from KI mice infected with BirA+ AAV1 were labeled at DIV14 with SA-AF647 (magenta), then fixed and immunostained for RIM1/2 using a secondary antibody conjugated to AF532 (cyan), and processed for 2-color dSTORM. (O and P) Histograms of the nearest distance between the centroid of bAP-NLGN1 nanodomains and the outline of either presynaptic mEos3.2-NRXN1β or RIM1/2 densities, respectively. Each dataset was obtained from 2 independent experiments. The number of bAP-NLGN1 synaptic puncta analyzed is indicated on each graph.

Fig. 8.

Visualization of bAP-NLGN1 in organotypic hippocampal slices by confocal microscopy and TEM. (A) Molecular strategy to biotinylate bAP-NLGN1 and stain it with fluorescent or nanogold streptavidin conjugates. (B and C) Two-color confocal images of CA3 neurons in organotypic slices from WT or KI mice, infected with BirA+ AAV9 (cyan) at DIV3, then labeled with SA-AF647 (magenta) at DIV14. (D) Higher magnification images of synapses (yellow arrowheads) located on proximal (Top) and distal (Bottom) dendrites. (E–H) Single CA3 pyramidal neurons in organotypic hippocampal slices from KI mice were coelectroporated at DIV7 with BirA^ER, tdTomato, and either Xph20-GFP or GPHN.FingR-GFP, live labeled with SA-AF647 at DIV14, then fixed and observed in confocal microscopy. (E and F) Images of PSD-95 and gephyrin (green), respectively, together with bAP-NLGN1 (magenta), and tdTomato (blue). Arrowheads show bAP-NLGN1 accumulated at postsynapses. (G and H) Proportion and surface density of PSD-95 or gephyrin puncta containing bAP-NLGN1 (7 fields of view, 5 neurons, 2 independent experiments). Data are presented as mean ± SEM and were compared by unpaired t test (G, ****P < 0.0001) or Mann–Whitney test (H, *** P = 0.0006). (I–L) TEM images of synapses in the CA3 region of organotypic slices from WT and KI mice, respectively. Slices were infected with BirA+ AAV9, then labeled at DIV14 with SA-Nanogold, fixed, and silver enhanced. Pre- and postsynapses are highlighted in yellow and blue, respectively. (I and J) Canonical synapses formed by CA3 axons onto distal dendrites, with zooms on the synaptic cleft (Right insets). (K and L) Giant synapses formed between mossy fibers and thorny excrescences on proximal dendrites. Red and blue arrows show synaptic and extrasynaptic bAP-NLGN1, respectively. (M) Subcellular distribution of SA-Nanogold particles in the synaptic cleft and in pre- or postsynaptic compartments. Data represent mean ± SEM and were compared by a Kruskal–Wallis test followed by Dunn’s multiple comparison tests (****P < 0.0001). (N) Distribution of SA-Nanogold particles across the synaptic cleft (343 particles, 22 synapses). (O) Number of bAP-NLGN1 nanodomains in the synaptic cleft (n = 40 synapses from 3 experiments). (P) Number of SA-Nanogold particles per synapse versus the length of the synaptic cleft.