Article Figures & Data
Figures
- Figure 1.
A687T and A687S substitutions increase Nrxn3α-mediated postsynaptic recruitment of PSD-95 and gephyrin, forming more heterologous synapses in vitro. A, Diagram indicating the location of the A687 substitutions in the EGF-B-LNS4 linker of Nrxn3α. A mutation in this linker in Nrxn1α are highlighted in magenta and imposed onto the Nrxn3α primary amino acid sequence. The conserved alanine (bold) is indicated. The human A687 (purple), the patient mutation (A687T, green), the substitution to mimic side-chain bulk (A687V, orange), and the substitution to mimic polar/hydrophilic side-chain properties (A687S, blue) are indicated. A Nrxn3α mutation (T678I) linked with intellectual disability is highlighted in yellow and a mutation in human Nrxn1α (E715K) associated with ASD is highlighted in magenta. B, Diagrams depicting the different A687 amino acid side-chain substitutions in Nrxn3α. C, D, Schematic depicting the experimental workflow of the heterologous synapse formation assay (C) and a representation of the recruitment of postsynaptic specializations to HEK cells expressing Nrxn3α variants (D). E, Representative images of immunolabeled HEK293T expressing GFP only, WT, or A687 substitutions and DIV14 primary hippocampal neuron cocultures. F, G, Summary graphs of the intensities of the excitatory postsynaptic scaffold, PSD-95 (F), and of the inhibitory postsynaptic scaffold, gephyrin (G), reveal that A687S selectively increased the amount of PSD-95 recruited without changes in gephyrin recruitment. Data represented are mean ± SEM; mean values are quantified from the average of the number of cells in all experiments. Numbers in bar graphs describe cells from each individual experiment. PSD-95 intensity: GFP: 0.1723 ± 0.033 A.U.; WT: 0.7925 ± 0.095 A.U.; p = 0.0007; A687T: 0.64 ± 0.090 A.U.; p = 0.8700; A687V: 1.354 ± 0.07607 A.U.; p = 0.7423; A687S: 1.35 ± 0.209 A.U.; p = 0.0039; one-way ANOVA, multiple comparisons relative to WT. Gephyrin intensity: GFP: 0.64 ± 0.012 A.U.; WT: 1.178 ± 0.2188 A.U.; p = 0.0003; A687T: 1.04 ± 0.1585 A.U.; p = 0.9859; A687V: 0.9086 ± 0.1585 A.U.; p = 0.8349; A687S: 1.28 ± 0.2692 A.U.; p = 0.9651; one-way ANOVA, multiple comparisons relative to WT. **p < 0.01. Number of cells analyzed from the number of independent experiments indicated in each bar graph.
- Figure 2.
The A687 substitutions do not alter Nrxn3α surface expression in hippocampal cultured neurons. A, Schematic illustrating the molecular replacement approach (top) and representative image of lentivirus infected primary hippocampal cultures (bottom). B, qRT-PCR of Nrxn3α mRNA after shRNA-mediated knockdown of endogenous Nrxn3α after lentiviral infection (left), normalized mRNA levels after molecular replacement approach with A687 substitutions (right). Nrxn3α relative to shRNA knockdown: WT: 9.149 ± 0.6306, n = 3, p = <0.0001; A687T: 10.44 ± 1.908, n = 3, p = <0.0001; A687V: 10.99 ± 2.590, n = 4, p = <0.0001; A687S: 9.793 ± 0.4335, n = 3, p = <0.0001; one-way ANOVA, multiple comparisons. C, Diagram highlighting the locations of the A687 substitutions and the corresponding codons, and the relative location of the HA-epitope tag. D, Representative images of DIV14 hippocampal neurons live surfaced immunolabeled HA-tagged isoforms of Mock, WT, and A687 substitutions. E, Quantification of the surface levels of HA-Nrxn3α WT and A687 variants following molecular replacement. Intensity, GFP: 0.0 ± 0.0, n = 44/3, p = <0.0001; WT: 60.41 ± 3.187 A.U., n = 58/3; A687T: 68.75 ± 3.395 A.U., n = 65/3, p = 0.3341; A687V: 63.82 ± 3.281 A.U., n = 79/3, p = 0.9301; A687S: 57.46 ± 2.859 A.U., n = 73/3, p = 0.9608; density per μm2, GFP: 0.000 ± 0.000, p = <0.0001; WT: 4.750 ± 0.3142; A687T: 4.659 ± 0.2757, p = 0.9989; A687V: 4.994 ± 0.2100, p = 0.9475; A687S: 4.314 ± 0.20181, p = 0.6607; GFP: 0.000 ± 0.000 μm2, p = <0.0001; WT: 1.124 ± 0.1114 μm2; A687T: 1.425 ± 0.1224 μm2, p = 0.2228; A687V: 1.285 ± 0.0932 μm2, p = 0.7677; A687S: 1.301 ± 0.0884 μm2, p = 0.7127; one-way ANOVA, multiple comparisons. Number of independent experiments identified in bar graphs in B. Number of cells analyzed from the number of independent experiments indicated in each bar graph in E.
- Figure 3.
Molecular replacement with Nrxn3αA687S selectively increases the size and numbers of PSD-95 puncta and surface AMPAR puncta in vitro. A, Representative images of DIV14 primary hippocampal neurons immunolabeled for vGluT1 and PSD-95. B, C, Summary graphs of presynaptic vGluT1 (B) and postsynaptic PSD-95 (C) intensity (left), density (middle), and size (right) reveals that A687T selectively increases the size of vGluT1, and A687S significantly increases PSD-95 size and density. vGlut1 intensity: GFP: 89.47 ± 2.079 A.U. n = 102/4; shRNA: 90.54 ± 2.656 A.U., n = 105/4, p = 0.9998; WT: 97.88 ± 3.294 A.U., n = 74/4, p = 0.3276; A687T: 95.82 ± 2.419 A.U., n = 106/4, p = 0.6507; A687V: 95.21 ± 2.719 A.U., n = 105/4, p = 0.6507; A687S: 86.62 ± 2.814 A.U., n = 96/4, p = 0.9773; vGlut1 density per 10 μm: GFP: 4.028 ± 0.1683; shRNA: 4.314 ± 0.1910, p = 0.8744; WT: 3.818 ± 0.1780, p = 0.9760; A687T: 4.355 ± 0.1563, p = 0.8833; A687V: 4.080 ± 0.1987, p > 0.9999; A687S: 4.512 ± 0.2110, p = 0.4335; vGlut1 size: GFP: 0.8980 ± 0.06289 μm2; shRNA: 1.193 ± 0.07321 μm2, p = 0.2745; WT: 1.194 ± 0.08530 μm2, p = 0.3747; A687T: 2.065 ± 0.1666 μm2; A687V: 1.170 ± 0.05989 μm2, p = 0.3650; A687S: 1.277 ± 0.1005 μm2, p = 0.0825; PSD-95 intensity: GFP: 130.5 ± 6.2224 A.U.; shRNA: 141.6 ± 6.395 A.U., p = 0.8390; WT: 128 ± 7.289 A.U., p = 0.9999; A687T: 138.4 ± 6.524 A.U., p = 0.9562; A687V: 132.9 ± 6.487, p = 0.9998; A687S: 134.5 ± 7.378 A.U., n = 96/4, p = 0.9982; PSD-95 density per 10 μm: GFP: 3.771 ± 0.2545; shRNA: 3.351 ± 0.2175, p = 0.7890; WT: 2.817 ± 0.1914, p = 0.0764; A687T: 4.336 ± 0.0204, p = 0.4937; A687V: 3.847 ± 0.2459, p = 0.9999; A687S: 4.833 ± 0.2584, p = 0.0100; PSD-95 size: GFP: 0.6241 ± 0.03119 μm2; shRNA: 0.6087 ± 0.03365 μm2, p = 0.9996; WT: 0.5588 ± 0.03339 μm2, p = 0.8485; A687T: 0.7021 ± 0.03544 μm2, p > 0.6371; A687V: 0.6185 ± 0.02780 μm2, p > 0.9999; A687S: 0.8623 ± 0.05375 μm2, p < 0.0001; one-way ANOVA, multiple comparisons. D, Representative images of DIV14 primary hippocampal neurons live surface labeled for GluA1. E, Summary graphs of surface GluA1 intensity (top left), density (top right), and size (bottom left) reveals that A687S significantly increases the size and density of surface localized GluA1. There was no change in AMPAR colocalization PSD-95 clusters (bottom right). Intensity, GFP: 64.65 ± 2.346 A.U., n = 106/3; shRNA: 65.01 ± 2.377 A.U., n = 110/3, p = 0.9998; WT: 62.90 ± 1.618 A.U., n = 116/3, p = 0.9659; A687T: 60.92 ± 1.745 A.U., n = 125/3, p = 0.35769; A687V: 61.20 ± 2.146 A.U., n = 112/3, p = 0.6677; A687S: 58.71 ± 2.230 A.U., n = 122/3, p = 0.1658; density per 10 μm, GFP: 5.811 ± 0.1643; shRNA: 6.165 ± 0.1444, p = 0.3394; WT: 5.808 ± 0.1670, p > 0.9999; A687T: 5.669 ± 0.1281, p = 0.9518; A687V: 6.059 ± 0.1661, p = 0.7164; A687S: 6.468 ± 0.1792, p = 0.0169; size, GFP: 1.047 ± 0.07031 μm2; shRNA: 0.8941 ± 0.06724 μm2, p = 0.2191; WT: 0.9147 ± 0.07112 μm2, p = 0.3306; A687T: 1.012 ± 0.05476 μm2, p = 0.9899; A687V: 0.9884 ± 0.06772 μm2, p = 0.8995; A687S: 1.568 ± 0.1120 μm2, p < 0.0001; one-way ANOVA, multiple comparisons. Data represented are mean ± SEM; mean values are quantified from the average of the number of dendrites in all experiments. Numbers in bar graphs represents the number of dendrites analyzed from the number of independent experiments. **p < 0.01; ****p < 0.0001.
- Figure 4.
Molecular replacement with Nrxn3αA687S selectively increases mEPSC amplitude and evoked AMPAR-mediated transmission at excitatory synapses. A, Summary graphs of the passive membrane properties of neurons infected with lentiviruses encoding only GFP or Nrxn3α shRNA and following molecular replacement with Nrxn3α WT or A687. There were no changes in membrane capacitance (Cm) and membrane resistance (Rm) between conditions. Cm: GFP Control: 173.3 ± 9,875 pF, n = 18; shRNA: 164.7 ± 11.01 pF, n = 21, p = 0.9797; WT: 180.9 ± 6.328 pF, n = 16, p = 0.9912; A687T: 167.3 ± 6.328 pF, n = 20, p = 0.9962; A687V: 176.0 ± 7.296 pF, n = 17, p > 0.9999; A687S: 178.2 ± 7.338 pF, n = 17, p = 0.9989; Rm: GFP Control: 190.1 ± 14.23 mΩ; shRNA: 242.1 ± 21.43 mΩ, p = 0.1132; WT: 224.8 ± 11.46 mΩ; A687T: 237.5 ± 10.53 mΩ, p = 0.5386; A687V: 221.6 ± 8.653 mΩ, p = 0.5929; A687S: 207.8 ± 15.64 mΩ, p = 0.9564; one-way ANOVA, multiple comparisons versus GFP. B, Representative traces from DIV14 to DIV15 primary hippocampal neurons infected with the lentiviruses indicated in the panel. C, D, Summary graphs of the cumulative distribution of mEPSC frequency (C) and average frequency per condition (D). E, F, Summary graphs of the cumulative distribution of mEPSC amplitude (E) and average amplitude per condition (F). The A687S substitution selectively increased mEPSC amplitude without altering mEPSC frequency. Frequency: GFP Control: 1.293 ± 0.1362 Hz; shRNA: 1.289 ± 0.1278 Hz, p > 0.9999; WT: 1.180 ± 0.1577 Hz, p = 0.0853; A687T: 1.984 ± 0.1984 Hz, p = 0.0105; A687V: 1.292 ± 0.1279 Hz, p > 0.99999; A687S:1.089 ± 0.1905 Hz, p = 0.8429; amplitude: GFP Control: 12.03 ± 0.5846 pA; shRNA: 12.98 ± 0.4583 pA, p = 0.6454; WT: 11.73 ± 0.4320 pA, p = 0.9963; A687T: 11.90 ± 0.5683 pA, p = 0.9998; A687V: 12.64 ± 0.6503 pA, p = 0.9228; A687S: 15.47 ± 0.7739 pA, p = 0.0004; one-way ANOVA, multiple comparisons. G, Representative traces (left) and summary graph (right) of evoked AMPAR EPSC amplitudes reveal A687S significantly increased AMPAR-mediated synaptic transmission. GFP: 732 ± 90 pA; shRNA: 805 ± 110 pA, p = 0.98; WT: 714 ± 100 pA, p = 0.999; A687S: 1323 ± 160 pA, p = 0.008; one-way ANOVA multiple comparisons to GFP. H, Representative traces (left) and summary graph (right) of evoked NMDAR EPSC amplitudes. GFP: 828 ± 222 pA, p = 0.931; shRNA: 649 ± 195 pA, p = 0.969; WT: 699 ± 228 pA; A687S: 655 ± 168 pA, p = 0.924; one-way ANOVA multiple comparisons to GFP. Data represented are mean ± SEM; mean values are quantified from the average of the number of cells in all experiments. Numbers in bar graphs indicate the number of cells analyzed from the number of independent experiments.
- Figure 5.
Molecular replacement with Nrxn3αA687S selectively increases synapse density, independent of changes in GABAAR surface expression in vitro. A, Representative images of DIV14 primary hippocampal neurons immunostained for VGAT and gephyrin. B, C, Summary graphs of presynaptic vGAT (B) and postsynaptic gephyrin (C) intensity (left), density (middle), and size (right) reveals that A687S significantly increases VGAT and gephyrin density. vGAT intensity: GFP: 78.62 ± 2.072 A.U., n = 111/3; shRNA: 93.66 ± 2.11 A.U., n = 96/3, p < 0.00001; WT: 87.11 ± 1.90 A.U., n = 111/3, p = 0.0072; A687T: 85.76 ± 2.513 A.U., n = 104/3, p = 0.2925; A687S: 78.62 ± 1.997 A.U., n = 109/3, p > 0.9999; 83.66 ± 2.453 A.U., n = 104/3, p = 0.5568; vGAT density per 10 μm: GFP: 2.223 ± 0.095; shRNA:1.969 ± 0.093, p = 0.4868; WT: 2.418 ± 0.109, p = 0.7216; A687T: 1.848 ± 0.084, p = 0.0831; A687S: 2.857 ± 0.105, p < 0.00001; A687V: 2.268 ± 0.105; vGAT size: GFP: 1.717 ± 0.067 μm2; shRNA: 2.281 ± 0.108 μm2, p < 0.0001; WT: 1.827 ± 0.088 μm2, p = 0.9347; A687T: 1.8143 ± 0.087, p = 0.8975; A687V: 1.811 ± 0.086 μm2, p = 0.9661; A687S: 1.615 ± 0.068 μm2, p = 0.9547; one-way ANOVA, multiple comparisons. Gephyrin intensity: GFP: 95.54 ± 2.889 A.U.; shRNA: 95.93 ± 1.813 A.U., p > 0.9999; WT: 104.7 ± 2.458 A.U., p = 0.0613; A687T: 99.68 ± 2.273 A.U., p = 0.8167; A687V: 99.67 ± 2.338 A.U., p = 0.8044; A687S: 95.13 ± 2.087 A.U., p > 0.9999; gephyrin density per 10 μm: GFP: 3.565 ± 0.135; shRNA: 3.277 ± 0.133, p = 0.6404; WT: 3.369 ± 0.128, p = 0.8875; A687T: 3.130 ± 0.121, p = 0.1687; A687V: 3.1116 ± 0.130, p = 0.1276; A687S: 4.106 ± 0.132, p = 0.0356; gephyrin size: GFP: 1.258 ± 0.068 μm2; shRNA: 1.346 ± 0.097 μm2, p = 0.9378; WT: 0.9884 ± 0.054 μm2, p = 0.0387; A687T: 1.091 ± 0.059 μm2, p = 0.4671; A687V: 0.9859 ± 0.054 μm2, p = 0.034; A687S: 1.121 ± 0.061 μm2, p = 0.6742; one-way ANOVA, multiple comparisons. D, Representative images of DIV14 primary hippocampal neurons live surface labeled for the GABAAR subunit, GABAγ2. E, Summary graphs of surface GABAA intensity (top left), density (top right), and size (bottom left) reveals that A687S significantly decreased the density of surface localized GABAA. Intensity: GFP: 60.65 ± 1.321 A.U., n = 176/5; shRNA: 56.51 ± 1.425 A.U., n = 196/5, p = 0.1426; WT: 55.53 ± 1.335 A.U., n = 186/5, p = 0.0474; A687T: 53.62 ± 1.516 A.U., n = 174/5, p = 0.0030; A687S: 54.24 ± 1.364 A.U., n = 191/5, p = 0.0067; A687V: 54.64 ± 1.502A.U., n = 188/5, p = 0.0132; density per 10 μm2: GFP: 5.344 ± 0.123; shRNA: 5.204 ± 0.125, p = 0.9589; WT: 5.060 ± 0.125, p = 0.5364; A687T: 4.952 ± 0.108, p = 0.1949; A687S: 4.744 ± 0.011, p = 0.044; A687V: 4.894 ± 0.108, p = 0.0767; size: GFP: 1.157 ± 0.050 μm2; shRNA: 1.236 ± 0.053 μm2, p = 0.9069; WT: 1.282 ± 0.058 μm2, p = 0.5907; A687T: 1.363 ± 0.061 μm2, p = 0.0964; A687V: 1.157 ± 0.044 μm2, p > 0.9999; A687S: 1.280 ± 0.057 μm2, p = 0.6000; one-way ANOVA, multiple comparisons. A687S induced a decrease in GABAAR colocalization with gephyrin clusters (bottom right). GFP: 0.8083 ± 0.016 n = 91/3; shRNA: 0.7601 ± 0.016, n = 114/3 p = 0.2149; WT: 0.8199 ± 0.014, n = 109/3 p = 0.9947; A687T: 0.8121 ± 0.014, n = 102/3, p > 0.9999; A687V: 0.8430 ± 0.012, n = 110/3, p = 0.5925; A687S: 0.7301 ± 0.016, n = 110/3, p = 0.0042; one-way ANOVA, multiple comparisons. Data represented are mean ± SEM; mean values are quantified from the average of the number of dendrites in all experiments. Numbers in bar graphs describe number of dendrites from the number of independent experiments.
- Figure 6.
Molecular replacement with Nrxn3αA687S does not alter inhibitory synaptic transmission in vitro. A, Summary graph of passive membrane properties of neurons infected with lentivirus encoding GFP alone, Nrxn3α shRNA along or following molecular replacement with Nrxn3α WT or A687. There were no changes in membrane capacitance (Cm) and membrane resistance (Rm) between conditions. Cm: GFP Control: 167.7 ± 8.618 pF, n = 16/4; shRNA: 158.2 ± 7.560 pF, n = 20/4, p = 0.9616; WT: 155.3 ± 6.069 pF, n = 19/4, p = 0.8909; A687T: 156.2 ± 6.734 pF, n = 20/4, p = 0.9169; A687V: 161.7 ± 8.118, pF, n = 22/4, p = 0.9946; A687S: 150.1 ± 8.610 pF, n = 20/4, p = 0.6396; one-way ANOVA, multiple comparisons; Rm: GFP Control: 207.2 ± 26.83 mΩ; shRNA: 270.5 ± 21.20 mΩ, p = 0.3276; WT: 262 ± 21.15 mΩ, p = 0.4797; A687T: 253.4 ± 17.66 mΩ, p = 0.6730; A687V: 272.1 ± 1731 mΩ, p = 0.2546; A687S: 269.5 ± 20.20 mΩ, p = 0.3196; one-way ANOVA, multiple comparisons. B, Representative traces from DIV14 to DIV15 primary hippocampal neurons infected with lentivirus indicated above each trace. C, D, Summary graphs of the cumulative distribution of mIPSC frequency (C) and average frequency per condition. Frequency: GFP Control: 2.859 ± 0.3755 Hz; shRNA: 2.04 ± 0.2907 Hz, p = 0.4166; WT: 1.892 ± 0.2414 Hz, p = 0.2462; A687T: 2.348 ± 0.2787 Hz, p = 0.8475; A687V: 2.674 ± 0,2767 Hz, p > 0.9990; A687S: 2.283 ± 0.3062 Hz, p = 0.7700; one-way ANOVA, multiple comparisons; amplitude: GFP Control: 15.87 ± 0.7440 pA; shRNA: 12.81 ± 0.6756 pA, p = 0.0297; WT: 14.14 ± 0.8320 pA, p = 0.5231; A687T: 13.51 ± 0.6736 pA, p = 0.1738; A687V: 15.33 ± 0.5504 pA, p = 0.9940; A687S: 14.51 ± 0.5440 pA, p = 0.7472; one-way ANOVA, multiple comparisons (D). E, F, Summary graphs of the cumulative distribution of mIPSC amplitude (E) and average amplitude per condition (F). Data represented are mean ± SEM; mean values are quantified from the average of the number of cells in all experiments. Statistical significance was determined by one-way ANOVA, multiple comparisons (to GFP) **p < 0.01. The number of cells monitors from independent experiments indicated in the bar graphs in A.
- Figure 7.
Molecular replacement with Nrxn3αA687 does not alter peak presynaptic calcium influx or clearance after electrical stimulation in vitro. A, Schematic depicting molecular replacement using primary hippocampal neurons infected with lentivirus encoding mCherry alone or Nrxn3α shRNA and mCherry, lentivirus for molecular replacement with Nrxn3α WT and A687 and GCaMP5G lentivirus. B, Representative images of DIV14 primary hippocampal neurons with GCaMP activation after evoked electrical stimulus of 50 Hz. C, Representative traces of evoked calcium influx after control, Nrxn3α knockdown, and WT or A687 molecular replacement in primary hippocampal neurons. D, Summary graph of peak calcium influx (ΔF / F0) revealed no alterations in peak calcium influx in any conditions. E, Representative traces of evoked calcium influx normalized to the peak after control, Nrxn3α knockdown, and WT or A687 molecular replacement in primary hippocampal neurons. F, Summary graph of the decay time constant representing the calcium clearance from presynaptic terminals (ΔF / F0) revealed no alterations in peak calcium influx in any conditions. Data represented are mean ± SEM; mean values are quantified from the average of the number of cells in all experiments. Peak amplitude+: mCherry Control: 1.696 ± 0.077 ΔF / F0, n = 53/5; shRNA: 1.678 ± 0.111 ΔF / F0, n = 42/5, p > 0.9999; WT: 1.892 ± 0.079 ΔF / F0, n = 50/5, p > 0.9999; A687T: 1.787 ± 0.861 ΔF / F0, n = 50/5, p = 0.9786; A687V: 1.763 ± 0.092 ΔF / F0, n = 61/5, p = 0.9936; A687S: 1.843 ± 0.076 ΔF / F0, n = 64/5, p = 0.8133; one-way ANOVA, multiple comparisons versus GFP; exponential decay: mCherry Control: 14.20 ± 3.046 τ, n = 4; shRNA: 13.04 ± 2.532 τ, n = 4, p = 0.9976; WT: 16.93 ± 1.937 τ, n = 4, p = 0.4979; A687T: 13.51 ± 1.434, n = 4, p = 0.9968; A687V: 15.30 ± 1.463, n = 4, p = 0.9858; A687S: 18.54 ± 2.773 n = 4, p = 0.6122; one-way ANOVA, multiple comparisons (to GFP). Numbers in bar graphs represent the average fluorescence change monitored from a single video recording from four independent experiments (C, D). The data points in F represent the Ca2+ exponential decay time averaged from 100 frames per video and subsequently averaged across independent experiments.
- Figure 8.
The A687S substitution selectively enhances Nrxn3α aggregation with the postsynaptic ligands, NL1 and NL3. A, Representative images of the heterologous cell aggregation assay. Inset images are magnified cell aggregates. B–E, Summary graphs of quantified heterologous cell aggregation assays normalized to WT aggregation. One sample of HEK293T cells are transfected with WT or A687 Nrxn3α and mCherry and coincubated with a sample of HEK293T cells transfected with Nrxn ligands. B, LRRTM2; WT: 1.00 ± 0.0, n = 5, A687T: 1.202 ± 0.038, n = 5, p = 0.0007; A687V: 0.9995 ± 0.073, n = 5, p > 0.9999; A687S: 1.126 ± 0.021, n = 5, p = 0.1043. C, NL1; WT: 1.00 ± 0.0, n = 5, A687T: 0.8051 ± 0.169, n = 5, p = 0.8688; A687V: 1.043 ± 0.1785: 0.8227, n = 5, p > 0.9999; A687S: 1.481 ± 0.0158, n = 5, p = 0.0158. D, NL2; WT: 1.00 ± 0.0, n = 5, A687T: 0.9104 ± 0.05682, n = 5, p = 0.5490; A687V: 0.8227 ± 0.04922, n = 5, p = 0.0063; A687S: 0.9959 ± 0.05292, n = 5, p > 0.9999. E, NL3; NL3: WT: 1.00 ± 0.0, n = 5, A687T: 0.9208 ± 0.169, n = 4, p = 0.9988; A687V: 0.9466 ± 0.1392, n = 5, p > 0.9999; A687S: 1.361 ± 0.104, n = 5, p = 0.0231. The index of aggregation was calculated by quantifying the number of colocalized cells (cells with red + green overlap) divided by the total number of both cell samples. Data represented are mean ± SEM; mean values are quantified from the average experiments analyzed. Numbers in bar graphs describe individual experiments. Statistical significance was determined by one-way ANOVA, multiple comparisons. *p < 0.05; ***p < 0.001; ****p < 0.0001. Number in bars represents the number of independent experiments. Scale bars: 200 μm, inset: 50 μm.
- Figure 9.
Domain swapping reveals the sequences of NL1 are necessary to enhance trans interactions with Nrxn3αA687S. A, Diagram of the domain organization (left) and homology (right) of neuroligin-1 and neuroligin-2. The N-termini of NL1 and NL2 only share 27.3% homology. B, Schematic of the NL1/NL2 domain swap. The sequences immediately downstream of the signal peptide for NL1 (aa: 46–85) and NL2 (aa: 15-75) exhibit minimal homology (top). The NL2.1 chimera possesses amino acids 1–75 of NL2 fused in-frame to NL1 and NL1.2 possesses amino acids 1–85 of NL1 fused in-frame to NL2 (bottom). C, D, Representative images. Inset images are magnified cell aggregates (scale bar: 200 μm, inset: 50 μm; top) and summary graphs (bottom) of cellular aggregation between Nrxn3α WT and A687S with NL1 and NL2.1 (C). WT-NL1: 1.00; WT-NL2.1: 0.902 ± 0.085, p = 0.95; A687S-NL1: 1.45 ± 0.09, p < 0.0001; A687S-NL2.1: 1.07 ± 0.1, p = 0.991; one-way ANOVA multiple comparisons to WT-NL1. Cellular aggregation of Nrxn3α WT and A687S with NL2 and NL1.2 N = 13 independent experiments (D). WT-NL2: 1.00; WT-NL1.2: 1.27 ± 0.04, p = 0.65; A687S-NL2: 1.385 ± 0.15, p = 0.171; A687S-NL1.2: 1.79 ± 0.25, p < 0.0001. One-way ANOVA multiple comparisons to WT-NL2. N = 8 independent experiments. E, Coomassie blue-stained Fc proteins, expressed in HEK293T cells and purified by affinity chromatography, for in cell-surface–binding assays. F, Representative images showing HEK293T cells expressing N-terminally HA-tagged Nlgn1, Nlgn2, Nlgn1.2, Nlgn2.1, and LRRTM2 incubated with purified Ig-fused Nrxn3 WT (Ig-Nrxn3 WT), Nrxn3 A687S (Ig-Nrxn3 A687S), or negative control (IgC), analyzed by immunofluorescence imaging for Ig-fusion proteins (red) and HA (blue; left). Quantification of cell surface binding. Data are means ± SEM (n = 13–21 cells/group; *p < 0.05; Mann–Whitney U test). Scale bar, 10 μm (applies to all images; right). G, Representative images showing HEK293T cells expressing N-terminally HA-tagged Nrxn3 WT or Nrxn3 A687S incubated with purified Ig-Nlgn1, Ig-Nlgn1.2, Ig-Nlgn2.1, Ig-LRRTM2, or negative control (IgC), analyzed by immunofluorescence imaging for Ig-fusion proteins (red) and HA (blue; left) and quantification of cell surface binding. Data are means ± SEM (n = 15–20 cells/group). Scale bar, 10 μm (applies to all images; right). Statistical significance was determined by one-way ANOVA, multiple comparisons (Nrxn3α WT to Nrxn3αA687S). *p < 0.05; ***p < 0.001. Each data point in bars represents the average from a single experiment.
- Figure 10.
Molecular modeling of Nrxn3αA687 substitutions in the Nrxn1α structure. A, The percent sequence identity (78.8%) of the LNS3–EGF-B–LNS4 reelin-like repeat module shared by bovine Nrxn1α and mouse Nrxn3α. B, Sequence alignment of EGF-B–LNS4 from bovine Nrxn1α and mouse Nrxn3α. These sequences share 80.4% sequence identity and have an 89.6% sequence similarity. The EGF-B repeat and LNS4 domain are indicated. Bovine Nrxn1α A723 and mouse Nrxn3α A687 are conserved and highlighted in orange. The conserved tyrosine Y714 (bovine Nrxn1α) and Y678 (mouse Nrxn3α) is indicated by a yellow box. The serine residue that is predicted to form hydrogen bonds with A687T and A687S in bovine Nrxn1α (S878) is conserved in mouse Nrxn3α (S843; indicated by a green box). C, Ribbon diagram of the Nrxn1 LNS3–EGF-B–LNS4 reelin repeat module. A723, which is analogous to the A687 site, is highlighted in red (dashed box). Green, LNS domains; yellow, EGF-B. D, Molecular modeling of A687 substitutions identifies predicted intramolecular interactions following substitutions with threonine, valine, and serine. The threonine substitution was predicted to induce a putative hydrogen bond (3.4 Å; dotted red line) with neighboring serine 878. The valine substitution was predicted to induce two putative steric clashes with the alpha carbon and carboxyl carbon backbone of tyrosine 714 (3.1 Å and 2.9 Å blue dotted line). The serine side chain was predicted to form a hydrogen bond with serine 878 (3.2 Å).
Tables
- Table 1.
Summary table of the hydrogen bonds and steric clashes predicted from molecular modeling
A687T
Prevalence: 0.68
Chi angle: −61.3
A687V
Prevalence: 0.87
Chi angle: 175.5
A687S
Prevalence: 0.47
Chi angle: 175.3
Hydrogen bonds Thr γ-hydroxyl moiety None Ser γ-hydroxyl moiety S878 γ-hydroxyl moiety S878 γ-hydroxyl moiety Distance 3.4 Å 3.2 Å Steric clashes None Val gamma-2 carbon None 1.Y714 α carbon. Overlap: 0.66, 3.1 Å 2. Y714 carboxyl carbon. Overlap: 0.63, 2.9 Å Details of the hydrogen bonds and steric clashes predicted from the molecular model of the LNS3–EGF-B–LNS4 module. The Nrxn3αA687T and Nrxn3αA687S substitutions are predicted to participate in hydrogen bonds with S878. Nrxn3αA687V is not predicted to engage in hydrogen bonds but is predicted to create two steric clashes. Prevalence refers to the frequency of occurrence of a side-chain rotamer observed from a Dunbrack backbone-dependent rotamer library, which is curated from experimentally solved protein structures. Overlap is defined as the sum of the van der Waals radii minus the distance between the two atoms and factors in an allowance for potential hydrogen-bond pairs. To identify clashes, a default value of ≤0.6 Å was used. The chi angle is torsional angle of the side-chain α and β carbons for each substitution of Nrxn3αA687.
Excitatory synapses Inhibitory synapses A687T A687S A687V A687T A687S A687V Hemisynaptic protein recruitment Unchanged Increased PSD95 protein recruitment Unchanged Unchanged Unchanged Unchanged Synaptic morphology Increased vGluT1 size and docked vesicles in AZ Increased PSD95 size and density Unchanged Unchanged Increased Gephyrin and vGAT density Unchanged Postsynaptic receptor surface expression Unchanged Increased GluA1-containing AMPAR size and density Unchanged Unchanged Decreased γ2-containing GABAR density Unchanged Transmission properties Increased glutamate release probability Increased excitatory postsynaptic strength Unchanged Unchanged Unchanged Unchanged Ca2+ signaling Unchanged Unchanged Unchanged Unchanged Unchanged Unchanged Postsynaptic ligand binding (aggregation assay) Increased LRRTM2 interaction Increased NL1 and NL3 interaction Unchanged Unchanged Increased NL3 interaction Decreased NL2 interaction Postsynaptic ligand binding (cell surface binding assay) Not tested Increased LRRTM2 interaction Not tested Not tested Unchanged Not tested Table summarizing the results from all experiments performed in this study. Changes induced by A687S are in bold font.
