Distinct mechanisms control the specific synaptic functions of Neuroligin 1 and Neuroligin 2

Abstract

Neuroligins are postsynaptic cell-adhesion molecules that regulate synaptic function with a remarkable isoform specificity. Although Nlgn1 and Nlgn2 are highly homologous and biochemically interact with the same extra- and intracellular proteins, Nlgn1 selectively functions in excitatory synapses whereas Nlgn2 functions in inhibitory synapses. How this excitatory/inhibitory (E/I) specificity arises is unknown. Using a comprehensive structure-function analysis, we here expressed wild-type and mutant neuroligins in functional rescue experiments in cultured hippocampal neurons lacking all endogenous neuroligins. Electrophysiology confirmed that Nlgn1 and Nlgn2 selectively restored excitatory and inhibitory synaptic transmission, respectively, in neuroligin-deficient neurons, aligned with their synaptic localizations. Chimeric Nlgn1-Nlgn2 constructs reveal that the extracellular neuroligin domains confer synapse specificity, whereas their intracellular sequences are exchangeable. However, the cytoplasmic sequences of Nlgn2, including its Gephyrin-binding motif that is identically present in the Nlgn1, is essential for its synaptic function whereas they are dispensable for Nlgn1. These results demonstrate that although the excitatory vs. inhibitory synapse specificity of Nlgn1 and Nlgn2 are both determined by their extracellular sequences, these neuroligins enable normal synaptic connections via distinct intracellular mechanisms.

Keywords: Distinct Mechanisms; Extracellular Domain; Intracellular Domain; Neuroligin; Synapse Specific.

Figure 1. Nlgn1 and Nlgn2 are specifically localized on excitatory and inhibitory synapses.

(A) Schematic overview of experimental timeline and set-up. (B) Schematic of Nlgn1-WT and Nlgn2-WT constructs. (C) Representative image from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons, infected with Cre (blue) and either Nlgn1-WT (top) or Nlgn2-WT (bottom). The neurons were labeled with antibodies to Homer1 (red), HA (green), and MAP2 (blue). Scale bar: 20 μm. The right panels show an enlarged box area (arrowheads indicate HA puncta overlapped with Homer1 puncta). Scale bar: 5 μm. (D) Summary graph of the HA-Homer1 overlap percentage in Nlgn1-WT and Nlgn2-WT conditions. (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 13/3 and 11/3 for each column, left to right. Statistical significance was assessed by unpaired t test, ****p < 0.0001). (E) Representative image of hippocampal neurons from DIV14-16 cultured Nlgn1234 conditional knockout mice, infected with Cre (blue) and either Nlgn1-WT (top) or Nlgn2-WT (bottom). The neurons were labeled with antibodies to Gephyrin (red), HA (green), and MAP2 (blue). Scale bar: 20 μm. The right panels show an enlarged box area (arrowheads indicate HA puncta overlapped with Gephyrin puncta). Scale bar: 5 μm. (F) Summary graph of the HA-Gephyrin overlap percentage in Nlgn1-WT and Nlgn2-WT conditions. (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 15/3 and 14/3 for each column, left to right. Statistical significance was assessed by unpaired t test, ****p < 0.0001). Source data are available online for this figure.

Figure 2. The intracellular sequence of Nlgn2 is necessary for inhibitory synapse function but can be fully replaced with the intracellular sequence of Nlgn1.

(A) Schematic of Nlgn2-WT, Nlgn2-GPI (Nlgn2 chimeric construct only has extracellular domain), Nlgn2-Nlgn1 (Nlgn2 extracellular domain with Nlgn1 intercellular domain) constructs. (B) Representative traces of evoked GABAR-IPSC recorded from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons in the five conditions (ΔCre, Cre, Cre+Nlgn2-WT, Cre+Nlgn2-GPI, and Cre+Nlgn2-Nlgn1). (C) Summary graph of evoked GABAR-IPSC amplitude in all conditions (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 16/3, 16/3, 16/3, 14/3, and 14/3 for each column, left to right). Nonsignificant p > 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001, one-way ANOVA with post hoc Dunnett’s Multiple comparisons. Source data are available online for this figure.

Figure 3. The cytoplasmic gephyrin binding motif is required for Nlgn2 function whereas tyrosine phosphorylation is not.

(A) Alignment of Nlgn1 and Nlgn2 gephyrin binding sequence. (B) Schematic of Nlgn2-Nlgn1Y770A (Nlgn2 extracellular domain with Nlgn1 intercellular domain with a gephyrin binding point mutation), Nlgn2-Nlgn1DelGeph (Nlgn2 extracellular domain with Nlgn1 intercellular domain truncation of the entire gephyrin binding sequence) constructs. (C) Representative traces of evoked GABAR-IPSC recorded from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons in the four conditions (ΔCre, Cre, Cre+Nlgn2-Nlgn1Y770A, Cre+Nlgn2-Nlgn1DelGeph). (D) Summary graph of evoked GABAR-IPSC amplitude in all conditions (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 15/3,16/3,15/3, and 15/3 for each column, left to right). Nonsignificant p > 0.05; ***p < 0.001, one-way ANOVA with post hoc Dunnett’s Multiple comparisons. Nonsignificant relations are indicated as ns. Source data are available online for this figure.

Figure 4. A conserved cytoplasmic 21 residue sequence is necessary for Nlgn2 function, whereas the PDZ-domain binding sequence is dispensable.

(A) Alignment of Nlgn1 and Nlgn2 intracellular domain. Colored boxes show the different Nlgn2 intracellular domain truncations. (B) Schematic of Nlgn2-mt1, Nlgn2-mt2, Nlgn2-mt3, Nlgn2-mt3, Nlgn2-mt4 and Nlgn2-mt5 constructs. (C) Representative traces of evoked GABAR-IPSC recorded from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons in the seven conditions (ΔCre, Cre, Cre+Nlgn2-mt1, Cre+Nlgn2-mt2, Cre+Nlgn2-mt3, Cre+Nlgn2-mt4, Cre+Nlgn2-mt5). (D) Summary graph of evoked GABAR-IPSC amplitude in all conditions (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 19/4,19/4,19/4,19/4,17/4,13/3, and 13/3 for each column, left to right). Nonsignificant p > 0.05; **p < 0.01; ***p < 0.001, one-way ANOVA with post hoc Dunnett’s Multiple comparisons. Nonsignificant relations are indicated as ns. Source data are available online for this figure.

Figure 5. Different from the functional promiscuity of the Nlgn1 and Nlgn2 cytoplasmic sequences, their extracellular sequences encode synapse type-specificity that is independent of MDGA binding.

(A) Schematic of Nlgn1-Nlgn2 (Nlgn1 extracellular domain with Nlgn2 intercellular domain) constructs. (B) Representative traces of evoked GABAR-IPSC recorded from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons in the three conditions (ΔCre, Cre, Cre+Nlgn1-Nlgn2). Note that we rescued Nlgn1-Nlgn2 here from the identical three batches of Fig. 2, so the ΔCre and Cre traces here are the same as in Fig. 2. (C) Summary graph of evoked GABAR-IPSC amplitude in all conditions (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 16/3,16/3, and 12/3 for each column, left to right). ****p < 0.0001, one-way ANOVA with post hoc Dunnett’s Multiple comparisons. Nonsignificant relations are indicated as ns. Note that we rescued Nlgn1-Nlgn2 here from the identical three batches of Fig. 2, so the ΔCre and Cre data here are the same as in Fig. 2. (D) Schematic of Nlgn2-MDGA1mt (Nlgn2 with MDGA1-binding point mutation) constructs. (E) Representative traces of evoked GABAR-IPSC recorded from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons in the three conditions (ΔCre, Cre, Cre+Nlgn2-MDGA1mt). (F) Summary graph of evoked GABAR-IPSC amplitude in all conditions (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 16/3,15/3, and 18/3 for each column, left to right). Nonsignificant p > 0.05; **p < 0.01, one-way ANOVA with post hoc Dunnett’s Multiple comparisons. Nonsignificant relations are indicated as ns. Source data are available online for this figure.

Figure 6. Nlgn1 extracellular domain provides the specificity of Nlgn1 excitatory function.

(A) Schematic of Nlgn1-WT, Nlgn1-GPI (Nlgn1 chimeric construct only has extracellular domain), Nlgn2-Nlgn1 constructs. (B) Representative traces of evoked AMPAR- and NMDAR-EPSCs recorded from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons. AMPAR-EPSCs and NMDAR-EPSC recorded at −70 mV and +40 mV, respectively, in the five conditions (ΔCre, Cre, Cre+Nlgn1-WT, Cre+Nlgn1-GPI, Cre+Nlgn2-Nlgn1). (C) Summary graph of evoked AMPAR-EPSC amplitude in all conditions (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 22/4, 17/4, 22/4, 23/4, and 16/3 for each column, left to right). Nonsignificant p > 0.05; *p < 0.05, one-way ANOVA with post hoc Dunnett’s Multiple comparisons. Nonsignificant relations are indicated as ns. (D). Summary graph of evoked NMDAR-EPSC amplitude in all conditions (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 22/4, 17/4, 22/4, 23/4, and 16/3 for each column, left to right). Nonsignificant p > 0.05; **p < 0.01; ****p < 0.0001, one-way ANOVA with post hoc Dunnett’s Multiple comparisons. Nonsignificant relations are indicated as ns. Source data are available online for this figure.

Figure 7. Nlgn1 function in excitatory synapses does not require Neurexin-binding, MDGA-binding, or dimerization.

(A) Schematic of Nlgn1-NRXNmt (Nlgn1 with neurexin-binding point mutation), Nlgn1-dimmt (Nlgn1 dimerization point mutation), Nlgn1-MDGA1mt (Nlgn1 with MDGA1-binding point mutation) constructs. (B) Representative traces of evoked AMPAR- and NMDAR-EPSCs recorded from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons. AMPAR-EPSCs and NMDAR-EPSC recorded at −70 mV and +40 mV, respectively, in the five conditions (ΔCre, Cre, Cre+Nlgn1-NRXNmt, Cre+Nlgn1-dimmt, Cre+Nlgn1-MDGA1mt). (C) Summary graph of evoked AMPAR-EPSC amplitude in all conditions (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 17/4, 14/4, 15/4, 18/4, and 15/4 for each column, left to right). Nonsignificant p > 0.05; *p < 0.05, one-way ANOVA with post hoc Dunnett’s Multiple comparisons. Nonsignificant relations are indicated as ns. (D) Summary graph of evoked NMDAR-EPSC amplitude in all conditions (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 17/4, 14/4, 15/4, 18/4, and 15/4 for each column, left to right). Nonsignificant p > 0.05; **p < 0.01, one-way ANOVA with post hoc Dunnett’s Multiple comparisons. Nonsignificant relations are indicated as ns. Source data are available online for this figure.

Figure EV1. Neuroligin protein measurement and co-staining Homer1 with Gephyrin in Nlgn1234 cKO cultured hippocampal neurons (related to Fig. 1).

(A) Representative images of western blot for Nlgn1, Nlgn2, and Nlgn3 protein expressions from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons, infected with either ΔCre or Cre. Note that the Nlgn3 antibody detects Nlgn3 (blue arrow) and nonspecific band (red arrow). (B) Summary graphs of western blot analysis for Nlgn1, Nlgn2, and Nlgn3 protein expressions. (Bar and line graphs indicate mean ± SEM; samples/experiments = 3/3. 3 technical replicates. Statistical significance was assessed by unpaired t test, ***p < 0.001; ****p < 0.0001). (C) Representative image from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons. The neuron was labeled with antibodies to Homer1 (Green), Gephyrin (red), and MAP2 (blue). Scale bar: 20 μm. The right panels show an enlarged box area (arrowheads indicate Homer1 puncta overlapped with Gephyrin puncta) Scale bar: 5 μm. (D) Summary graph of Homer1 overlap percentage with Gephyrin and Gephyrin overlap percentage with Homer1 (Bar and line graphs indicate mean ± SEM; numbers of cells/experiment = 14/1). Source data are available online for this figure.

Figure EV2. Nlgn2 is specifically localized on inhibitory synapses and is determined by the extracellular sequence of Nlgn2 and cytoplasmic gephyrin binding motif is required whereas tyrosine phosphorylation is not (related to Figs. 2, 3 and 5).

(A) Representative image from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons, infected with Cre (blue) and Nlgn2-GPI, Nlgn1-Nlgn2, Nlgn2-Nlgn1, and Nlgn2-MDGA1mt (from top to bottom). The neurons were labeled with antibodies to Gephyrin (red), HA (green), and MAP2 (blue). Scale bar: 20 μm. The right panels show an enlarged box area (arrowheads indicate HA puncta overlapped with Gephyrin puncta). Scale bar: 5 μm. (B) Summary graph of the HA-Gephyrin overlap percentage in Nlgn2-GPI, Nlgn1-Nlgn2, Nlgn2-Nlgn1, and Nlgn2-MDGA1mt conditions. (C) Summary graph of the surface levels of HA-tagged Nlgn2 forms relative to MAP2 signal. (B, C) (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 10/3, 11/3, 9/3, and 9/3 for each column, left to right. Statistical significance was assessed by one-way ANOVA with post hoc Dunnett’s Multiple comparisons, Nonsignificant p > 0.05; **p < 0.01; ****p < 0.0001). (D) Representative image from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons, infected with Cre (blue) and Nlgn2-Nlgn1Y770A and Nlgn2-Nlgn1DelGeph. The neurons were labeled with antibodies to Gephyrin (red), HA (green), and MAP2 (blue). Scale bar: 20 μm. The right panels show an enlarged box area (arrowheads indicate HA puncta overlapped with Gephyrin puncta). Scale bar: 5 μm. (E) Summary graph of the HA-Gephyrin overlap percentage in Nlgn2-Nlgn1Y770A and Nlgn2-Nlgn1DelGeph conditions. (F) Summary graph of the surface levels of HA-tagged Nlgn2-Nlgn1Y770A and Nlgn2-Nlgn1DelGeph relative to MAP2 signal. (E, F) (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 19/3 and 12/3 for each column, left to right. Statistical significance was assessed by one-way ANOVA with post hoc Dunnett’s Multiple comparisons, Nonsignificant p > 0.05; ****p < 0.0001). Source data are available online for this figure.

Figure EV3. All Nlgn2 intracellular domain truncation constructs can properly traffic to the neuron membrane surface, but Nlgn2-NRXNmt and Nlgn2-dimmt constructs traffic to the neuron membrane surface signification decrease compared to the Nlgn2-WT construct (related to Fig. 4).

(A) Representative image from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons, infected with Cre (blue) and Nlgn2-mt1, Nlgn2-mt2, Nlgn2-mt3, Nlgn2-mt4 and Nlgn2-mt5 (from left to right). The neurons were labeled with antibodies to HA (green) and MAP2 (blue). Scale bar: 20 μm. (B) Representative image from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons, infected with Nlgn2-WT, Nlgn2-NRXNmt, and Nlgn2-dimmt. The neurons were labeled with antibodies to total HA (red) and surface HA (green). Scale bar: 20 μm. (C) Summary graph of the surface HA/total HA in Nlgn2-WT, Nlgn2-NRXNmt, and Nlgn2-dimmt conditions (Bar and line graphs indicate mean ± SEM. numbers of cells/experiments = 10/3, 9/3 and 7/3 for each column, left to right. Statistical significance was assessed by one-way ANOVA with post hoc Dunnett’s Multiple comparisons, ****p < 0.0001). Source data are available online for this figure.

Figure EV4. Nlgn1 is specifically localized on excitatory synapses is determined by the extracellular sequence of Nlgn1, and does not require Neurexin-binding, MDGA-binding, or dimerization (related to Figs. 6 and 7).

(A) Representative image from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons, infected with Cre (blue) and Nlgn1-GPI, Nlgn1-Nlgn2, and Nlgn2-Nlgn1. The neurons were labeled with antibodies to Homer1 (red), HA (green), and MAP2 (blue). Scale bar: 20 μm. The right panels show an enlarged box area (arrowheads indicate HA puncta overlapped with Homer1 puncta). Scale bar: 5 μm. (B) Summary graph of the HA-Homer1 overlap percentage in Nlgn1-GPI, Nlgn1-Nlgn2, and Nlgn2-Nlgn1 conditions. (C) Summary graph of the surface levels of HA-tagged Nlgn1 forms relative to MAP2 signal. (B, C) (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 10/3, 9/3, and 15/3 for each column, left to right. Statistical significance was assessed by one-way ANOVA with post hoc Dunnett’s Multiple comparisons, Nonsignificant p > 0.05; **p < 0.01; ****p < 0.0001). (D) Representative image from DIV14-16 cultured Nlgn1234 conditional knockout mice neurons, infected with Cre (blue) and Nlgn1-NRNXmt, Nlgn1-dimmt, and Nlgn1-MDGA1mt. The neurons were labeled with antibodies to Homer1 (red), HA (green), and MAP2 (blue). Scale bar: 20 μm. The right panels show an enlarged box area (arrowheads indicate HA puncta overlapped with Homer1 puncta). Scale bar: 5 μm. (E) Summary graph of the HA-Homer1 overlap percentage in Nlgn1-NRNXmt, Nlgn1-dimmt, and Nlgn1-MDGA1mt conditions. (F) Summary graph of the surface levels of HA-tagged Nlgn1 forms relative to MAP2 signal. (E, F) (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 14/3, 9/3, and 9/3 for each column, left to right. Statistical significance was assessed by one-way ANOVA with post hoc Dunnett’s Multiple comparisons, Nonsignificant p > 0.05). Source data are available online for this figure.

Figure EV5. Nlgn1-Nlgn2 is sufficient for the glutamatergic synaptic transmission function of Nlgn1, but Nlgn2-WT doesn’t (related to Fig. 6).

(A) Summary graph of evoked AMPAR-EPSC amplitude in all conditions (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 22/4, 17/4, 22/4, 7/2, and 11/3 for each column, left to right). Nonsignificant p > 0.05; *p < 0.05, one-way ANOVA with post hoc Dunnett’s Multiple comparisons. Nonsignificant relations are indicated as ns. Note that we rescued Nlgn1-Nlgn2 and Nlgn2-WT here from the same batches of Fig. 6, so the ΔCre and Cre data here are the same as in Fig. 6. (B) Summary graph of evoked NMDAR-EPSC amplitude in all conditions (Bar and line graphs indicate mean ± SEM; numbers of cells/experiments = 22/4, 17/4, 22/4, 7/2, and 11/3 for each column, left to right). Nonsignificant p > 0.05; ***p < 0.001; ****p < 0.0001, one-way ANOVA with post hoc Dunnett’s Multiple comparisons. Nonsignificant relations are indicated as ns. Note that we rescued Nlgn1-Nlgn2 and Nlgn2-WT here from the same batches of Fig. 6, so the ΔCre and Cre data here are the same as in Fig. 6. Source data are available online for this figure.