LRRTM2 controls presynapse nano-organization and AMPA receptor sub-positioning through Neurexin-binding interface

Abstract

Synapses are organized into nanocolumns that control synaptic transmission efficacy through precise alignment of postsynaptic neurotransmitter receptors and presynaptic release sites. Recent evidence show that Leucine-Rich Repeat Transmembrane protein LRRTM2, highly enriched and confined at synapses, interacts with Neurexins through its C-terminal cap, but the role of this binding interface has not been explored in synapse formation and function. Here, we develop a conditional knock-out mouse model (cKO) to address the molecular mechanisms of LRRTM2 regulation, and its role in synapse organization and function. We show that LRRTM2 cKO specifically impairs excitatory synapse formation and function in mice. Surface expression, synaptic clustering, and membrane dynamics of LRRTM2 are tightly controlled by selective motifs in the C-terminal domain. Conversely, the N-terminal domain controls presynapse nano-organization and postsynapse AMPAR sub-positioning and stabilization through the recently identified Neurexin-binding interface. Thus, we identify LRRTM2 as a central organizer of pre- and post- excitatory synapse nanostructure through interaction with presynaptic Neurexins.

Fig. 1. The C-terminal, but not the extracellular LRR domain, clusters LRRTM2 at excitatory synapses through the non-canonical PDZ-binding motif ECEV.

a Representation of LRRTM2 (AphaFold2 model using ColabFold v1.5.2) at the plasma membrane carrying a biotin Acceptor Peptide (AP) tag (red) in its N-terminal domain and labeled with monomeric streptavidin (mSA, PDB4JNJ). b Schematics of LRRTM2- WT- and mutants lacking the extracellular LRR domain (∆LRR-), the intracellular domain (-ΔC), the PDZ-binding domain ECEV (-ΔECEV), or mutated at the YxxC motif (-YACA). AP acceptor peptide, LRR leucine-rich repeat, TM transmembrane. Asterisks indicate the amino acids in the YxxC motif replaced with Alanines (LRRTM2-YACA). c DIV15 hippocampal neurons expressing Cre-mCherry, BirA^ER, biotinylated AP-LRRTM2 (WT, ∆C or ΔLRR) immunostained for endogenous PSD-95 as a postsynaptic marker. Cre-mCherry (blue) is overlaid with PSD-95 (green) and AP-LRRTM2 (red). d Quantification of AP-LRRTM2 (WT, ∆C, or ΔLRR) cluster density and percentage of PSD-95 clusters colocalized with AP-LRRTM2 clusters, showing decreased synaptic localization for the ΔC. Data acquired from three experiments were presented as mean values ± SEM (WT: n = 18, ∆C: n = 17, ∆LRR = 10 cells) **p < 0.01, ****p < 0.0001. Data were compared by one-way analysis of variance test, followed by post hoc Dunn’s test. e DIV15 hippocampal neurons expressing Cre-mCherry, BirA^ER, biotinylated AP-LRRTM2 (WT, ΔECEV or YACA) immunostained for endogenous PSD-95. Cre-mCherry (blue) is overlaid with PSD-95 (green) and AP-LRRTM2 (red). f Quantification of AP-LRRTM2 (WT, ΔECEV or YACA) cluster density and PSD-95 cluster density showing decreased density for the ΔECEV, but not for the YACA. Data acquired from three experiments, presented as mean values ± SEM (WT: n = 11, ΔECEV: n = 10, YACA: n = 10 cells), *p < 0.05, **p < 0.001. Data were compared by one-way analysis of variance test, followed by post hoc Dunn’s test.

Fig. 2. Membrane diffusion and confinement of LRRTM2 are regulated by the C-terminal domain.

a Schematics of mSA-labeled AP-LRRTM2 diffusing at the plasma membrane. b Representative images from DIV15 neurons co-expressing Cre-EGFP, Homer1c-DsRed, BirA^ER, and AP-LRRTM2 mutants (WT, ΔC, ΔECEV or YACA) labeled with mSA-STAR935P to track individual molecules by universal Point Accumulation for Imaging in Nanoscale Topography (uPAINT). Homer1c-DsRed (gray) overlaid with AP-LRRTM2 trajectories (immobile: D < 0.005 μm² s⁻¹, red; diffusive: D > 0.005 μm² s⁻¹, cyan; see the “Methods” section). c Mean square displacement over time, showing increased overall diffusion upon deletion of CTD, or mutation of YxxC. Data presented as mean values ± SEM. d Diffusion coefficients of free-diffusing proteins (D_diff, see the “Methods” section) shown as violin plots for AP-LRRTM2-WT, -∆C, -∆ECEV and -YACA (trajectories, WT: 1779, ΔC:2671, ΔECEV: 2070, YACA: 2918; ***p < 0.001, ****p < 0.0001). Data was compared by one-way analysis of variance test, followed by post-hoc Dunn’s test. e Percentage of non-synaptic tracks, showing an increase with ΔC and YACA-LRRTM2 (*p < 0.05, ***p < 0.001). Data presented as mean values ± SEM. Data were compared by one-way analysis of variance test, followed by post-hoc Dunn’s test. f Fractions of immobile, confined, and diffusive trajectories of AP-LRRTM2 mutants, showing decreased immobile fractions for the ΔC and the YACA mutants (**p < 0.01, ***p < 0.001, ****p < 0.0001). Data presented as mean values ± SEM. Data were compared by two-way analysis of variance test, followed by Tukey’s multiple comparison test. g Individual synaptic trajectories overlaid with Homer1c-DsRed (gray). h Mean synaptic dwell time of AP-LRRTM2, showing a specific decrease of time spent at synapses in the absence of the PDZ-like binding motif (ΔC, *p < 0.05 and ΔECEV, **p < 0.01), but not in YACA condition. Data presented as mean values ± SEM. Data were compared by one-way analysis of variance test, followed by post-hoc Dunn’s test. i Fractions of immobile, confined, and diffusive trajectories for synaptic AP-LRRTM2 showing a decrease in immobile fraction in ΔC condition (***p < 0.001, ****p < 0.0001). Data presented as mean values ± SEM. Data were acquired from three independent experiments (WT: n = 16, ΔC: n = 18; ΔECEV: n = 18, YACA: n = 16 cells). Data were compared by two-way analysis of variance test, followed by Tukey’s multiple comparison test.

Fig. 3. The YxxC motif regulates LRRTM2 membrane turnover.

a Representative images of DIV15 hippocampal neurons expressing Cre-mCherry, Homer1c-BFP and either SEP-LRRTM2-WT or SEP-LRRTM2-YACA. On the right, Cre-mCherry (blue) is overlaid with Homer (red) and LRRTM2 (green). b Pseudocolor images of SEP fluorescence recovery after photobleaching in spines from neurons expressing Cre-mCherry and SEP-LRRTM2-WT or -YACA. c Corresponding normalized fluorescence recovery curves in spines and d slow pool fraction in both conditions. Data are presented as mean values ± SEM. Data acquired from three independent experiments (SEP-LRRTM2-WT: n = 37 regions, SEP-LRRTM2-YACA: n = 32 regions; ****p < 0.0001)). Data were compared by non-parametric Mann–Whitney test. e Representative images of DIV15 hippocampal neurons expressing Cre-mCherry (blue), Homer1c-BFP (red), and SEP-LRRTM2-WT or -YACA (green). Dashed squares indicate the positions of panel i. f Schematics of the pH change protocol used to visualize intracellular and extracellular protein pools. At pH 7.4, the SEP-tag is fluorescent, whereas at acidic pH (pH 5.5), its fluorescence is quenched. Bath application of pH 5.5 renders surface SEP-tagged proteins non-fluorescent, and NH₄Cl de-acidifies intracellular vesicles, rendering intracellular SEP-tagged proteins fluorescent. g Normalized mean average intensity of SEP-LRRTM2-WT and SEP-LRRTM2-YACA overtime during the pH change protocol illustrated in f. Data are presented as mean values ± SEM. h Percentage of SEP-LRRTM2-WT and SEP-LRRTM2-YACA protein fluorescence signal localized in the intracellular compartments of dendrites. Data are presented as mean values ± SEM. Data were compared by non-parametric Mann–Whitney test. i Pseudocolor images of insets from (e) during the pH change protocol, revealing intracellular protein pools by subtraction of the “baseline” signal (extracellular) from the “NH₄Cl” signal (total) in the shaft and spines. j Percentage of SEP-LRRTM2-WT and SEP-LRRTM2-YACA intracellular pool distribution in spines and dendritic shaft. g–i Data acquired from three independent experiments (SEP-WT: n = 9 and SEP-YACA: n = 11 cells).

Fig. 4. LRRTM2 cKO impairs synaptic AMPAR stabilization.

a DIV15 hippocampal neurons expressing soluble mCherry and SEP-GluA1 subunit of AMPARs (control), Cre-mCherry and SEP-GluA1 (Cre) or mCherry, SEP-GluA1, BirA^ER and AP-LRRTM2-WT (Cre + LRRTM2-WT). On the right, mCherry (red) is overlaid with SEP-GluA1 (green). b Pseudocolor images of FRAP experiments performed on SEP-GluA1 localized in spines for the different conditions described in (a). c Corresponding normalized fluorescence recovery curves, showing faster recovery of SEP-GluA1 fluorescence in the absence of LRRTM2, and d slow pool fraction, showing a decrease of this fraction in the absence of LRRTM2. Data are presented as mean values ± SEM. Data obtained from three independent experiments (control: n = 51; Cre: n = 34; Cre + LRRTM2-WT: 39 regions, ****p < 0.0001). Data were compared by one-way analysis of variance test, followed by post hoc Dunn’s test. e DIV15 hippocampal neurons expressing soluble mCherry and SEP-GluA2 subunit of AMPARs (control), Cre-mCherry and SEP-GluA2 (Cre) or mCherry, SEP-GluA2, BirA^ER and AP-LRRTM2-WT (Cre + LRRTM2-WT). On the right, mCherry (red) is overlaid with SEP-GluA2 (green). f FRAP of SEP-GluA2 localized in spines for the different conditions described in (e), showing faster recovery in the absence of LRRTM2. g Corresponding normalized fluorescence recovery curves and h slow pool fraction show that GluA2 is less stabilized at synapses in the absence of LRRTM2. Data are presented as mean values ± SEM. Data obtained from three independent experiments (control: n = 58; Cre: n = 55; Cre + LRRTM2-WT: n = 16 regions, *p < 0.05). Data were compared by one-way analysis of variance test, followed by post hoc Dunn’s test.

Fig. 5. LRRTM2 controls synaptic AMPAR stabilization through neurexin-binding site.

a Crystal structure of hLRRTM2 (gray) in complex with Neurexin-1β (green) (PDB 5Z8Y), showing interaction site E348 recently identified (dark blue) and calcium ion (green), and D259 and D261 (light blue). b Schematics of EQ- (E348 mutated to Glutamine (Q)) and DT/AA- LRRTM2 (D259/T261 mutated to Alanines (A)). c COS-7 cells expressing EGFP and biotinylated WT-, EQ-, or DT/AA- LRRTM2, labeled with mSA-ATTO565, incubated with Nrxn1β-Fc and antiFc-A647. No Nrxn1β-Fc at EQ-LRRTM2 cell surface, showing disrupted binding. d Normalized average surface intensity of LRRTM2. e Average surface intensity of Nrxn1β-Fc normalized to expression levels of AP-LRRTM2, showing total disruption of Nrxn1β-binding exclusively with EQ. Data from three independent experiments (cells, WT: n = 60, EQ: n = 77, DT/AA: n = 62) ****p < 0.0001. f Co-culture showing recruitment of endogenous presynaptic synapsin1 (green) onto COS-7 cells expressing WT-LRRTM2 or DT/AA-LRRTM2, but not ΔLRR- or EQ-LRRTM2 (magenta) and g corresponding quantifications showing loss of synapsin1 recruitment in absence of LRR domain or mutation of E348, but not mutation of D259/T261. Data from three independent experiments (cells, WT: n = 29, ΔLRR: n = 35, EQ: n = 31, DT/AA: n = 27, ****p < 0.0001). h DIV15 neurons expressing Cre-EGFP, BirA^ER, and WT-, EQ- or DT/AA-LRRTM2 labeled for endogenous GluA1/2 and VGluT1 and i corresponding quantifications of synapse density (GluA1/2/VGluT1 apposition), showing specific decreased density upon EQ mutation. Data from three experiments (cells, WT: n = 14, EQ: n = 20, DT/AA: n = 13) *p < 0.05. j DIV15 neurons expressing Cre-mCherry, SEP-GluA1, BirA^ER and WT-, EQ- or DT/AA-AP-LRRTM2. k Normalized average intensity of spine SEP-GluA1, showing decreased intensity with EQ-LRRTM2 (regions, WT: n = 60, EQ: n = 45, DT/AA: n = 57), ***p < 0.001. l SEP-GluA1-containing spines before and after FRAP, m normalized FRAP curves in spines and n corresponding slow pool fraction, showing a selective reduction in EQ-LRRTM2. Data from three independent experiments (regions, WT: n = 50, EQ: n = 31, DT/AA: n = 59), **p < 0.01. d, e, g, i, k, n Data presented as mean values ± SEM, compared by one-way analysis of variance test, followed by post hoc Dunn’s test.

Fig. 6. The Neurexin-binding site E348 is required for the nanoscale organization of presynaptic RIM scaffolds and postsynaptic AMPA receptors.

a Schematics of trans-synaptic nanocolumns, where postsynaptic AMPARs anchored by PSD-95 scaffolds are aligned in front of release sites organized by presynaptic scaffolds such as RIM. b Example of reconstructed images from dual-color dSTORM of endogenous RIM1/2 (blue) and GluA1/2 (red) in the control condition (WT), showing apposition between RIM and AMPAR nanoclusters. Below are representative examples of RIM1/2-GluA1/2 apposition in different conditions showing disruption of GluA1/2 nanoscale organization in the presence of EQ and DT/AA compared to WT, and selective disruption of presynaptic RIM nano-organization in EQ condition. c Quantifications of RIM1/2 localizations in subsynaptic densities (SSDs) and RIM1/2 cluster surface, showing a selective decrease in EQ condition, but not in DT/AA condition. d Quantifications of GluA1/2 localizations in SSDs, and cluster surface, showing a decrease with both EQ and DT/AA mutants. e Quantifications of RIM1/2 and GluA1/2 SSD number per synapse, depicting a specific decrease of RIM1/2 SSD only with the EQ mutant. Data presented as mean values ± SEM. f Quantifications of the mean distances between RIM1/2 and GluA1/2 SSD centroids and relative frequency distribution of these distances, showing the increased distance between presynaptic scaffolds and postsynaptic AMPARs in the mutant conditions. Data obtained from three independent experiments (WT: n = 8; EQ: n = 10; DT/AA: n = 8 cells), *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data were compared by one-way analysis of variance test, followed by post-hoc Dunn’s test. g Representative mEPSC traces recorded from DIV15 hippocampal neurons expressing Cre-GFP, BirA^ER, and WT-, EQ-, or DT/AA- LRRTM2. h Cumulative graph and plot of mean mEPSC amplitudes, showing a significant shift of mEPSC amplitudes in the mutant conditions compared to WT. Data presented as mean values ± SEM. i cumulative graph of inter-event interval and plot of mean mEPSC frequency. Data presented as mean values ± SEM. Data acquired from three independent experiments (WT, n = 20; EQ, n = 17; DT/AA, n = 14) ****p < 0.0001. Data were compared by one-way analysis of variance test, followed by post-hoc Dunn’s test.

Fig. 7. Summary model.

a Crystal structure of LRRTM2 (gray) in complex with Neurexin-1β (green) (PDB 5Z8Y) showing the interaction site with glutamic acid E348 and a calcium ion (dark blue and dark green, respectively, boxed region). b Model of LRRTM2-containing nanocolumn showing interaction with Nrxn1β through E348, the interaction of Nrxn1β with scaffold molecule CASK, and of LRRTM2 with PSD-95 through PDZ-binding motifs (black), and the concave interface of LRRTM2 which could stabilize GluA2-containing AMPARs at synapses. c Working model: disruption of LRRTM2-Nrxn1β binding upon E348Q mutation induces a loss of synapse density. At the synaptic level, LRRTM2-dependent AMPAR and RIM SSDs are lost upon EQ mutation.