Membrane-tethered monomeric neurexin LNS-domain triggers synapse formation

Abstract

Neurexins are presynaptic cell-adhesion molecules that bind to postsynaptic cell-adhesion molecules such as neuroligins and leucine-rich repeat transmembrane proteins (LRRTMs). When neuroligins or LRRTMs are expressed in a nonneuronal cell, cocultured neurons avidly form heterologous synapses onto that cell. Here we show that knockdown of all neurexins in cultured hippocampal mouse neurons did not impair synapse formation between neurons, but blocked heterologous synapse formation induced by neuroligin-1 or LRRTM2. Rescue experiments demonstrated that all neurexins tested restored heterologous synapse formation in neurexin-deficient neurons. Neurexin-deficient neurons exhibited a decrease in the levels of the PDZ-domain protein CASK (a calcium/calmodulin-activated serine/threonine kinase), which binds to neurexins, and mutation of the PDZ-domain binding sequence of neurexin-3β blocked its transport to the neuronal surface and impaired heterologous synapse formation. However, replacement of the C-terminal neurexin sequence with an unrelated PDZ-domain binding sequence that does not bind to CASK fully restored surface transport and heterologous synapse formation in neurexin-deficient neurons, suggesting that no particular PDZ-domain protein is essential for neurexin surface transport or heterologous synapse formation. Further mutagenesis revealed, moreover, that the entire neurexin cytoplasmic tail was dispensable for heterologous synapse formation in neurexin-deficient neurons, as long as the neurexin protein was transported to the neuronal cell surface. Furthermore, the single LNS-domain (for laminin/neurexin/sex hormone-binding globulin-domain) of neurexin-1β or neurexin-3β, when tethered to the presynaptic plasma membrane by a glycosylinositolphosphate anchor, was sufficient for rescuing heterologous synapse formation in neurexin-deficient neurons. Our data suggest that neurexins mediate heterologous synapse formation via an extracellular interaction with presynaptic and postsynaptic ligands without the need for signal transduction by the neurexin cytoplasmic tail.

Figure 1.

Generation and characterization of Nrx TKD neurons. A, Design of the lentiviral Nrx TKD vector (top) and quantifications of Nrx1, Nrx2, and Nrx3 mRNAs in Nrx TKD neurons (bottom). H1, Human H1 promoter; U6, human U6 promoter; Ub, ubiquitin promoter. For mRNA quantifications, β-actin was used as endogenous control in quantitative RT-PCR measurements of neurons infected with control or Nrx TKD lentiviruses; levels are normalized to those of β-actin and the control. B, C, Nrx TKD decreases neurexin protein levels; this decrease is reversed by reexpression of Nrx3β. B depicts a representative immunoblot stained with a pan-neurexin antibody (A473), with β-actin immunoblot as a loading control (bottom). C depicts quantifications of α- and β-neurexins using fluorescently labeled secondary antibodies (normalized for β-actin). In control neurons, α-neurexins are more abundant than β-neurexins; Nrx3β rescue protein is present in two forms (Nrx3β′ and Nrx3β″) that may correspond to different glycosylation states. D, Quantification of mRNA levels in control infected hippocampal neurons or hippocampal neurons infected with the Nrx TKD lentivirus without or with expression of Nrx3β. mRNAs were measured by quantitative RT-PCR using Fluidigm dynamic arrays. Hippocampal neurons were infected at DIV3, and mRNA levels were quantified at DIV14. Rows represent the evaluated genes and columns individual cultures of neurons. The heat map (blue to red) represents the relative expression of a gene normalized to its average expression in the control cultures. mRNA levels of endogenous neurexins are shown on the top three rows. The Nrx3 RT-PCR assay is specific to mouse and does not detect the human Nrx3β rescue mRNA, validating the continued effectiveness of the Nrx TKD even after rescue overexpression. NL, Neuroligin; Slitrks, Slit- and Trk-like; CL/Lphn1–2, G-protein-coupled receptor CIRL/latrophilin; NeuN/Rbfox3, neuronal nuclei; Gpr6, G-protein-coupled receptor 6; Glt1, glial high-affinity glutamate transporter; Eya1, eyes absent 1 homolog. E, F, Effect of the Nrx TKD on the levels of selected synaptic proteins. Syt1, Synaptotagmin-1; Syb2, synaptobrevin-2. Protein levels were determined by quantitative immunoblotting with fluorescently labeled secondary antibodies (E, representative immunoblots; F, summary graphs of protein levels determined with fluorescent secondary antibodies). Actin and VCP were used as loading controls. VCP levels relative to actin showed no significant difference between three groups. G, H, Nrx TKD does not impair neuronal viability, as determined by measurements of the density of NeuN-positive neurons (G, representative images; H, summary graph). I, J, Nrx TKD does not decrease synapse density, as determined by synapsin staining (I, representative images; J, summary graph). Data are means ± SEM (A, n = 3; C, n = 5; D, n = 4; F, n = 3–4; H, n = 3; J, n = 4 independent cultures). Statistical analyses were performed by Student's t test comparing test samples to the control. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 2.

Effect of Nrx TKD on heterologous synapse formation. A, B, Nrx TKD, but not single neurexin KDs, impair heterologous synapse formation. COS-7 cells expressing a NL1-mVenus fusion protein were cocultured with hippocampal neurons infected with lentiviruses mediating KDs of individual neurexins or the Nrx TKD; the latter was analyzed without or with expression of Nrx3β rescue protein. After cell fixation and permeabilization, transfected COS cells were visualized by mVenus contained in NL1, while synapses were stained for synapsin (A, representative images; B, summary graphs of synapse formation quantified as the ratio of presynaptic synapsin to postsynaptic mVenus). C, D, Specificity of the Nrx TKD effect on heterologous synapse formation. COS cells expressing mVenus, NL1-mVenus fusion protein, or myc-tagged SlitTrk6 were cocultured with neurons that were infected with control lentiviruses, or lentiviruses expressing the Nrx TKD, without or with coexpression of various Nrx1 and Nrx3 splice variants as indicated. For analyses, coverslips containing live neurons were first incubated with HA antibody (to label surface-exposed Nrx1β and Nrx3β containing HA epitopes) and then fixed, permeabilized, and labeled for synapsin. Heterologous synapse formation was quantified as the ratio of presynaptic synapsin in neurons to postsynaptic mVenus in COS cells, and surface transport of neurexins was assessed as the ratio of HA to synapsin in regions devoid of COS cells. Note that Nrx1βSS4− was not HA-tagged, and is thus not surface stainable (C, representative images; D, summary graphs). Data in B and D are means ± SEM (B, n = 4; D, n = 4 independent cultures). Statistical analyses were performed by Student's t test comparing test samples to the positive control. ***p < 0.001.

Figure 3.

Analysis of point mutations in the Nrx3β cytoplasmic tail. A, Alignment of the human Nrx1, Nrx2, and Nrx3 intracellular sequences shows a high level of conservation. The six blocks of conserved three-residue sequences targeted for alanine (A) substitution mutagenesis are highlighted in different shades of red and numbered on top. The C-terminal PDZ-domain binding sequences are highlighted in green. TMR, Transmembrane region. B, C, Analysis of six alanine substitution mutants in the C-terminal cytoplasmic sequence of Nrx3β (A). COS cells expressing NL1-mVenus were cocultured with control or Nrx TKD hippocampal neurons; the latter were also infected with control lentiviruses or lentiviruses expressing wild-type or mutant HA-tagged Nrx3β (B, representative images of heterologous synapses; C, summary graphs of synapse formation and surface expression of various Nrx3β proteins). For analysis approaches, see legend to Figures 2, C and D. D, Fluorescence images of hippocampal neurons cotransfected with the endoplasmic reticulum marker pEYFP-ER and with wild-type Nrx3β or mutant Nrx3β-KEK containing alanine substitutions in the KEK sequence (A). E, F, Same as B and C, except that two additional mutations of the cytoplasmic KEK sequence of Nrx3β were analyzed: a mutation changing KEK to CQC (Nrx3β-KEK2), and a deletion of the entire TLMKEK sequence (Nrx3β-KEK3). Data shown in C and F are means ± SEM (n = 3 independent cultures). Statistical analyses were performed by Student's t test comparing test samples to the positive control. ***p < 0.001. n.d., Not determined.

Figure 4.

The specific PDZ-domain binding sequence of Nrx3β is not essential for heterologous synapse formation. A, B, COS cells expressing NL1-mVenus were cocultured with control (Ctrl.) or Nrx TKD hippocampal neurons; the latter were additionally infected with control lentiviruses or lentiviruses expressing wild-type or mutant Nrx3β with either (1) an alanine substitution of the C-terminal three residues (Nrx3β-PDZ/A), (2) a deletion of the C-terminal three residues (Nrx3β-PDZΔ3), or (3) an exchange of the 10 C-terminal residues of Nrx3β for those of NL1 (Nrx3β-PDZ/NL1), which includes a different type of PDZ-domain binding sequence [A, representative images; B, quantifications of synapse formation mediated by the various Nrx3β proteins (top) and their surface expression (bottom)]. Data shown in B are means ± SEM (n = 3 independent cultures). Statistical analyses were performed by Student's t test comparing test samples to the positive control. ***p < 0.001. n.d., Not determined.

Figure 5.

Heterologous synapse formation does not require cytoplasmic neurexin signaling. A, Domain structures of wild-type and mutant Nrx3β that either lacks the entire cytoplasmic tail except for the last 10 residues (Nrx3β-Δ55 + 10), contains the complete NL1 cytoplasmic tail instead of the Nrx3 cytoplasmic tail (Nrx3β-NL1tail), or is anchored on the membrane surface by a GPI moiety (Nrx3β-GPI). B, C, Nrx3β cytoplasmic tail mutants fully sustain heterologous synapse formation (B, representative images; C, synapse formation quantifications). COS cells expressing NL1-mVenus fusion protein were cocultured with control neurons or with Nrx TKD neurons that express the indicated rescue proteins and analyzed as described for Figure 2C. D, E, Heterologous synapse formation induced by LRRTM2 is blocked by the Nrx TKD, but rescued by GPI-anchored Nrx3β. Experiments were performed as in B and C, except that the COS cells expressed LRRTM2 instead of NL1. Data shown in C and E are means ± SEM (C, n = 4; E, n = 3 independent cultures). Statistical analyses were performed by Student's t test comparing test samples to the positive control. **p < 0.01; ***p < 0.001.

Figure 6.

The Nrx3β LNS domain mediates heterologous synapse formation. A, Domain structure of Nrx3β (top) and sequence alignment of human Nrx1, Nrx2, and Nrx3 and of the Nrx3 stalk and cys-loop mutants in the region between the LNS domain and the transmembrane region (bottom; mutated residues or deletions are underlined). B, C, Analysis of the effect of mutations in the O-linked sugar stalk region and the cys-loop of Nrx3β on heterologous synapse formation. NL1-mVenus expressing COS cells were cocultured with hippocampal neurons infected with control or the Nrx TKD lentiviruses without or with rescue with wild-type or mutant Nrx3β [B, representative images; C, summary graphs of synapse formation (top) and Nrx3β surface expression (bottom); for analysis methods, see Figure 3]. Data are means ± SEM (n = 3 independent cultures). Statistical analyses were performed by Student's t test comparing test samples to the positive control. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 7.

Actin localization during heterologous synapse formation. A, Comparative localization of β-actin-GFP coexpressed with Nrx3β (top) or with Nrx3β-NL1tail (bottom) in Nrx TKD neurons that were cocultured with COS cells expressing Flag-tagged NL1. Panels show representative images of a heterologous synapse stained for actin-GFP, vGluT1, NL1, and Nrx3β (via its HA tag) at two magnifications. B, Summary graph of the Manders' coefficients as the degree of colocalization for each fluorescent signal pairs. Data shown are means ± SEM (n = 4 independent cultures, A, B). Statistical analyses were performed by Student's t test comparing test samples to Manders' coefficients of NL1 + Nrx3β interaction. *p < 0.05; **p < 0.01.

Figure 8.

Model of neurexin action in heterologous synapse formation. Comparison of the signal transduction model of neurexins mediated by CASK bound to the neurexin cytoplasmic tail (left) versus the signal transduction model emerging from the current study whereby the LNS domain of neurexins serves as an adaptor that transduces a neuroligin-binding signal to an unidentified neurexin coreceptor that then mediates the signal transduction event required for heterologous synapse formation (right).