Control of Excitatory and Inhibitory Synapse Formation by Neuroligins

See allHide authors and affiliations

Science  25 Feb 2005:
Vol. 307, Issue 5713, pp. 1324-1328
DOI: 10.1126/science.1107470

This article has a correction. Please see:


The normal function of neural networks depends on a delicate balance between excitatory and inhibitory synaptic inputs. Synapse formation is thought to be regulated by bidirectional signaling between pre- and postsynaptic cells. We demonstrate that members of the Neuroligin family promote postsynaptic differentiation in cultured rat hippocampal neurons. Down-regulation of neuroligin isoform expression by RNA interference results in a loss of excitatory and inhibitory synapses. Electrophysiological analysis revealed a predominant reduction of inhibitory synaptic function. Thus, neuroligins control the formation and functional balance of excitatory and inhibitory synapses in hippocampal neurons.

Adhesion molecules bridge the pre- and postsynaptic compartments of synapses in the central nervous system. Neuroligin-1 (NL-1), a member of the Neuroligin family of postsynaptic adhesion molecules, can trigger formation of functional presynaptic terminals in axons through interaction with its axonal receptor β-neurexin [(13), reviewed in (46)]. To explore whether the β-neurexin–neuroligin complex acts bidirectionally and controls postsynaptic differentiation, we overexpressed NL-1 in cultured hippocampal neurons (7). Analysis of dendritic morphology, postsynaptic scaffolding molecules, and postsynaptic glutamate receptor distribution revealed that NL-1 promotes assembly of the postsynaptic apparatus (Fig. 1). NL-1–overexpressing neurons showed a 68 ± 7% increase in the density of dendritic spine–like protrusions. Spines in NL-1–expressing cells frequently exhibited irregular, handshaped heads with multiple presynaptic terminals labeled for the vesicular glutamate transporter 1 (vGlut1), a marker of excitatory synapses (Fig. 1A; fig. S1). The density of synaptic puncta containing the scaffolding proteins PSD-95 and Homer was increased significantly (Fig. 1B). Moreover, staining for the NR1 subunit of N-methyl-d-aspartate (NMDA) receptors revealed that NL-1 strongly promotes NMDA receptor recruitment (Fig. 1D). We also observed recruitment of AMPA-type glutamate receptors, as indicated by clustering of GluR2/3 subunits in some NL-1–expressing cells. However, high NL-1 levels led to dispersion of GluR2/3, likely due to the depletion of cytoplasmic binding partners for NL-1 (8). In summary, these experiments show that NL-1 is a potent inducer of excitatory postsynaptic differentiation.

Fig. 1.

NL-1 promotes postsynaptic differentiation. Hippocampal neurons were cotransfected with expression vectors for hemagglutinin (HA)–tagged NL-1 and EGFP or with EGFP vectors only. (A) Immunostaining for vGlut1 and EGFP in control cells expressing EGFP (left column) and cells coexpressing EGFP and NL-1 (right column). NL-1–induced spine structures contacting multiple presynaptic terminals (right). (B) Immunostaining for PSD-95 and EGFP (left) or HA epitope to detect NL-1 (right). (C) Immunostaining for Homer and EGFP (left) or HA epitope to detect NL-1 (right). (D) Immunostaining for NMDA-receptor subunit 1 (NR1) and EGFP (left) or HA epitope to detect NL-1 (right). Scale bar, 5 μm. (E) Quantification of postsynaptic protein recruitment, dendritic spine induction, and synapse formation in cells expressing NL-1 and EGFP-transfected control cells. vGlut1/PSD-95 shows density of puncta with colocalizing pre- and postsynaptic markers (SEM, n = 10, ***P < 0.001).

NL-1–induced postsynaptic differentiation may be entirely mediated through scaffolding proteins interacting with the intracellular tail of NL-1 or may require extracellular interactions with the β-neurexin–NL-1 complex. To investigate the respective contributions of the NL-1 extracellular and intracellular domains to postsynaptic differentiation, we analyzed two NL-1 mutants. In the mutant NL-swap, the extracellular cholinesterase domain of NL-1 was exchanged with the homologous sequence from acetylcholinesterase to yield a mutant NL protein in which the β-neurexin–binding site was inactivated (1). In the intracellular mutant NLΔC, the cytoplasmic tail of NL-1 was truncated, which removed the PDZ-binding motif that interacts with postsynaptic scaffolding molecules such as PSD-95 (9). When overexpressed in hippocampal neurons, the extracellular mutant NL-swap could still stimulate the recruitment of PSD-95 into clusters at the cell membrane. However, these PSD-95 clusters were largely extrasynaptic and did not align with the presynaptic marker vGlut1 (Fig. 2A). This result suggests that this NL swap acts as a dominant-negative mutant that uncouples nucleation of the postsynaptic scaffold from the presynaptic terminal. Expression of the intracellular mutant NLΔC did not stimulate PSD-95 clustering, which confirmed that the cytoplasmic tail of NL-1 is essential for PSD-95 recruitment (Fig. 2A). Despite the inability to recruit PSD-95, NLΔC increased NMDA-receptor cluster density at synapses, albeit less efficiently than did the wild-type protein (Fig. 2, B and C). These findings suggest that NMDA receptors are primarily recruited to NL-1–induced synapses independently of PSD-95 and that synaptic recruitment requires the ability of NL-1 to interact with β-neurexin or additional extracellular ligands. Therefore, the β-neurexin–neuroligin complex provides a nucleation site for the assembly of postsynaptic scaffolding molecules and NMDA receptors opposite the presynaptic terminal through intracellular and extracellular interactions.

Fig. 2.

Perturbation of postsynaptic assembly by NL-1 mutants. Hippocampal neurons expressing EGFP, HA-NL-1 [wild type (wt)], extracellular mutant (swap), or a C-terminally deleted mutant (ΔC). (A) Immunostaining for vGlut1, PSD-95, and EGFP or the HA epitope. Boxed area was enlarged in third column. Scale bar, 5 μm. (B) Immunostaining for vGlut1, NR1, and EGFP or the HA epitope. (C and D) Quantification of clustering (C) and synaptic localization (D) of PSD-95 and NR1 in cells expressing wild-type or mutant NL as compared with EGFP-expressing control cells (SEM, n = 10, **P < 0.01).

To evaluate the consequences of reduced neuroligin function, we used RNA interference (10, 11). We generated small-hairpin RNAs (shRNAs) directed against the rodent neuroligin isoforms NL-1, NL-2, and NL-3 and tested their efficiency and specificity by cotransfection with neuroligin expression vectors into HEK293 cells (Fig. 3A). Knockdown of neuroligin expression was strictly isoformspecific, e.g., NL-1–directed shRNAs did not alter NL-2 and NL-3 expression. Introduction of the shRNAs into hippocampal neurons in culture confirmed the suppression of endogenous neuroligins individually and in combination (Fig. 3B; fig. S2). Expression of other neuronal proteins such as PSD-95 or class III β-tubulin was not altered, and introduction of the shRNA vector lacking an insert or containing a control shRNA had no effect on any of the proteins analyzed.

Fig. 3.

Suppression of NL isoforms leads to excitatory synapse loss. (A) HEK293 cells cotransfected with HA-tagged NL-1, -2, or -3 and NL shRNAs (sh-NL1, sh-NL2, sh-NL3). Cell lysates were probed with antibodies against HA and tubulin. The control shRNA (sh-con.) targets p53. (B) Hippocampal neurons were triple-infected with lentiviruses encoding shRNAs that are targeting NL1, NL2, and NL3. Total protein levels of NL-1, -2, and -3, as well as PSD-95 and β3-tubulin, were analyzed by Western blotting. No suppression is observed with a shRNA vector without hairpin insert (sh-vec.) and a control shRNA (sh-con., targeting p53). This experiment underestimates the efficiency of NL protein suppression, because the viruses only infect 90% of all neurons in the culture. (C) Hippocampal neurons were transfected with shRNAs directed against individual NL isoforms or triple-transfected with shRNAs for all three NL isoforms (sh-NL1,2,3) and immunostained with antibodies to vGlut1. Loss of vGlut-1–positive puncta and dendritic spines can be rescued by cotransfection of HA-NL-3 (sh-NL1-1 HA-NL-3) but not by cotransfection of the inactive mutant NL-swap (sh-NL1,2,3 + NL-swap). Scale bar, 5 μm. (D) Quantification of vGlut1 (left) and dendritic + spine density (right) in hippocampal neurons with reduced NL expression (SEM, n = 10, ***P < 0.001). To avoid including cells with NL-3 overexpression, we focused on the 20% of all HA-NL-3–positive cells with lowest HA-NL-3 expression levels. (E) Quantification of vGlut1 puncta that are also positive for the postsynaptic AMPA-receptor subunit GluR1 (GluR1-pos.) in control (sh-con.) and triple-neuroligin-knockdown cells (sh-NL1,2,3) and the effect of triple-neuroligin knockdown on vGlut1 clustering in the presence of 2 μM TTX (SEM, n = 10, ***P < 0.001). (F) Total GluR1 puncta density in hippocampal neurons transfected with control shRNA (sh-con.) or shRNAs targeting NL-1,2,3 (sh-NL1,2,3) (SEM, n = 10, ***P < 0.001).

Suppression of single or multiple neuroligin isoforms reduced excitatory synapse formation. When the shRNAs were transfected into hippocampal neurons, we observed fewer vGlut1-positive excitatory presynaptic terminals (Fig. 3, C and D). Introduction of an expression vector lacking the shRNA insert or targeting an unrelated mRNA did not alter the density of vGlut1 puncta. Knockdown of each of the neuroligin isoforms also inhibited postsynaptic maturation, as indicated by a significant reduction in the density of dendritic spines, although this was less severe for suppression of NL-3. Simultaneous knockdown of all three rodent neuroligin isoforms resulted in a 70% reduction in the number of morphologically recognizable excitatory synapses as detected by colocalization of vGlut1 and the glutamate receptor subunit GluR1 (Fig. 3E). The number of synapses in triple-neuroligin-knockdown cells was also reduced when cultures were maintained in tetrodotoxin (TTX) to block all sodium channel–dependent action potentials (Fig. 3E). Synapse loss is therefore not a secondary consequence of an essential function of neuroligins in action potential–dependent neurotransmission.

It was interesting that the reduction of excitatory terminal density in triple-knockdown cells was not more severe than in the single-knockdown cells (Fig. 3, C and D). This indicated that the function of all three neuroligin isoforms might be coupled or that a critical level of total neuroligin proteins might be required for normal function. To further investigate this, we tested whether the triple-neuroligin-knockdown phenotype could be rescued by cotransfection of a human NL-3 cDNA, which is resistant to sh-NL3–directed cleavage. Simultaneous transfection of human NL-3 with shRNAs against NL-1, -2, and -3 restored dendritic spine density and vGlut1 clustering, whereas cotransfection of an inactive NL mutant (NL-swap) did not rescue the knockdown phenotype (Fig. 3, C and D). Similarly, defects caused by knockdown of only NL-3 could be suppressed by overexpression of NL-1 (8). Increasing expression of single neuroligin isoforms can, therefore, compensate for defects caused by the loss of other neuroligins. In summary, these experiments demonstrate that neuroligin isoforms are critical for normal excitatory synapse formation and have partially overlapping functions.

Recent studies suggested a potential role for neuroligins at inhibitory synapses (12, 13). Immunostaining for endogenous NL-2 and the vesicular γ-aminobutyric acid transporter (VGAT) confirmed that the NL-2 isoform is concentrated at inhibitory synapses (Fig. 4A). By contrast, NL-1 is concentrated at mature excitatory synapses (14). When transfected into hippocampal neurons, all NL isoforms stimulated the formation of both excitatory and inhibitory terminals (Fig. 4B; fig. S3). NL-2 was more effective than NL-1 or NL-3 with respect to inhibitory terminal induction (Fig. 4, C and D) and might, therefore, preferentially contribute to inhibitory synapse formation. However, each of the three neuroligin isoforms is capable of inducing both excitatory and inhibitory terminals when expressed at a sufficiently high level.

Fig. 4.

NL-2 preferentially promotes inhibitory synapse formation. (A) Hippocampal neurons immunostained with antibodies to NL-2 (green), VGAT (red), and vGlut1 (blue). Right panel shows merge at higher magnification. Scale bar, 5 μm (left) and 10 μm (right). (B) Hippocampal neurons transfected with expression vectors for EGFP, HA-tagged NL-1, NL-2, or NL-3 were immunostained with antibodies to the HA epitope (green), VGAT (red), and vGlut1 (blue). Scale bar, 5 μm. (C) Density of excitatory and inhibitory terminals for cells overexpressing EGFP, NL-1, NL-2, or NL-3, (SEM, n = 10 cells, ***P < 0.001). (D) Ratio of vGlut1:VGAT–positive terminals in NL-overexpressing cells (n = 10 cells, **P < 0.01).

Using shRNA-mediated knockdown revealed an essential function for neuroligins in inhibitory synapse formation. Suppression of any single neuroligin isoform resulted in a reduction in the density of VGAT-positive presynaptic terminals (Fig. 5, A and B). Among the individual isoforms, this effect was most significant for suppression of NL-2. Simultaneous knockdown of all three neuroligin isoforms resulted in the most robust reduction of inhibitory terminals. In a way similar to excitatory synapses, we found that some inhibitory terminals could form, even when all neuroligin isoforms were suppressed. We cannot exclude residual neuroligin levels (<10%) that escaped the shRNA-directed down-regulation as the cause, but we consider it more likely that there are other synapse-inducing proteins that promote neuroligin-independent synapse formation (15). However, acute suppression of neuroligins does result in a substantial reduction of excitatory and inhibitory synapse numbers, demonstrating that they are important regulators of synaptogenesis.

Fig. 5.

Loss of neuroligins leads to imbalance of excitatory and inhibitory transmission. (A) Hippocampal neurons transfected with control shRNAs (sh-con.), shRNAs against individual NL isoforms (sh-NL1, sh-NL2, sh-NL3) or cotransfected with shRNAs for all three NL isoforms (sh-NL1,2,3) were immunostained for EGFP (green) and VGAT (red). Scale bar, 5 μm. (B) Quantification of VGAT puncta density in neuroligin-knockdown cells (SEM, n = 10 cells, *P < 0.05; **P < 0.01). (C) Representative recordings of mIPSCs and mEPSCs from hippocampal neurons transfected with control (sh-con.) or NL-1, -2, and -3 shRNA vectors (sh-NL1,2,3). Examples of mIPSCs (arrowheads) and mEPSCs (arrows) are marked. (D) Amplitudes and frequencies of mEPSCs and mIPSCs in neuroligin-knockdown cells (sh-NL1,2,3; black columns) as compared with control cells (sh-con.; white columns, n ≥ 5, *P < 0.05; **P < 0.01; ***P < 0.001). Neuroligin-knockdown cells show an increased percentage of mEPSCs among the total number of PSCs and a corresponding decrease in the percentage of mIPSCs as compared with control cells.

Because the knockdown of neuroligins affected the numbers of morphologically recognizable excitatory and inhibitory terminals, we investigated whether synaptic transmission was altered. We recorded mIPSCs (miniature inhibitory postsynaptic currents) and mEPSCs (miniature excitatory postsynaptic currents) in triple-neuroligin-knockdown (sh-NL1,2,3) hippocampal neurons (Fig. 5C). In knockdown cells, mIPSC amplitudes and frequency were 52 ± 21% and 93 ± 30% reduced, respectively (Fig. 5D). In comparison, mEPSCs were only slightly affected (Fig. 5D). These selective functional defects resulted in a shift in the balance of excitatory and inhibitory events in the NL-1, -2, and -3 knockdown cells: mEPSCs represented 36.0 ± 3.1% of the total number of synaptic events in control cells and 83.0 ± 3.6% of events in the neuroligin-knockdown cells (Fig. 5D). Combined with the data described above, these results lead to three main conclusions: (i) Neuroligins are potent inducers of postsynaptic differentiation; (ii) The neuroligin isoforms support both excitatory and inhibitory synapse formation; and (iii) Loss of neuroligin isoforms alters the normal excitatory/inhibitory balance in hippocampal neurons.

NL-2 preferentially localizes to inhibitory synapses (13). Our overexpression and knockdown experiments revealed effects of NL-2 on both excitatory and inhibitory synapses, although effects on inhibitory synapses were more prominent. Most likely, initial interactions with excitatory and inhibitory axons are somewhat promiscuous, because mismatched excitatory and inhibitory pre- and postsynaptic components are observed frequently in developing neurons (1618). The preference of individual neuroligin isoforms for excitatory or inhibitory synapses might then be reinforced during synaptic maturation.

Loss of neuroligins had a selective effect on inhibitory synapse function. Despite a 70% decrease in the density of morphologically recognizable excitatory synapses in triple-neuroligin-knockdown cells (Fig. 3E), we did not observe an equivalent reduction in mEPSC frequency. This suggests the preferential loss of inactive and/or silent excitatory synapses or that the remaining excitatory synapses may exhibit increased activity. In either case, the selective reduction of inhibitory synapse function resulted in a significant imbalance of excitatory and inhibitory transmission. This observation is notable, because the excitatory to inhibitory (E/I) input ratio is critical for normal computation of neuronal excitation (19, 20) and is generally kept constant by a homeostatic feedback mechanism (20, 21). A recent study reported a similar alteration of the E/I ratio in cells in which expression of PSD-95 had been suppressed by RNA interference (12). The selective decrease in inhibitory synapse function in neuroligin-knockdown cells may indicate that formation or stabilization of functional inhibitory synapses relies more heavily on neuroligins than does the function of excitatory synapses. An alternative hypothesis is that neuroligins contribute to the homeostatic mechanism that maintains the E/I balance.

Inactivating mutations in human neuroligins are associated with autism spectrum disorders (2225), as are perturbations in the E/I ratio and morphological aberrations in dendritic spines (2628). Our work links neuroligin function with these two phenotypes in hippocampal neurons. Further analysis of the cellular defects caused by reduced neuroligin expression in vitro may, therefore, provide a useful framework for understanding the cellular defects of autism spectrum disorders.

Supporting Online Material

Materials and Methods

Figs. S1 to S3

References and Notes

References and Notes

Stay Connected to Science

Navigate This Article