Schizophrenia Research Forum - A Catalyst for Creative Thinking

Neuroligin 1—A New Substrate for γ-Secretase

Adapted from a story that originally appeared on the Alzheimer Research Forum.

31 October 2012. Add another notch in γ-secretase’s belt. In the October 18 Neuron, two independent research teams report that neuronal firing boosts γ-secretase cleavage of neuroligin 1, a postsynaptic membrane protein linked to autism and other brain disorders. The proteolysis slows spine formation and presynaptic transmitter release. As with processing of amyloid-β precursor protein (APP) and Notch, the extracellular domain of neuroligin sheds first. Here, the two papers implicate different enzymes. Work led by Taisuke Tomita and Takeshi Iwatsubo at the University of Tokyo points toward ADAM10, the APP/Notch sheddase. The other study—led by Michael Ehlers at Duke University Medical Center, Durham, North Carolina—fingers matrix metalloproteinase 9 (MMP9). Based on the new data, the authors think both sheddases cleave neuroligin 1, with potentially different functions depending on developmental stage or cellular context. The work also hints at a possible role for neuroligin in synaptic dysfunction associated with Alzheimer’s disease (AD), and a potential role for these enzymes in autism. The new findings will be presented at an Alzforum Webinar on November 1.

A third paper in the same issue of Neuron adds another angle to the neuroligin 1 (NLGN1) story. Roger Nicoll and Seth Shipman of the University of California, San Francisco, report that the synaptic protein mediates hippocampal long-term potentiation (LTP), a prerequisite for learning and memory. The LTP effect appears to be exclusive to NLGN1, not NLGN3, the other neuroligin at excitatory synapses. It requires an extracellular region of neuroligin 1 that governs its association with neurexins on the presynapse. Considering all three studies, it seems reasonable that “a molecule involved in the creation and strengthening of synapses, when depleted by proteolysis, might lead to the weakening or loss of synapses,” Shipman said of neuroligin 1.

Mammals express four subtypes of neuroligins, which interact with neurexins to build and sustain synapses. Recent work highlights a synaptic role for γ-secretase, too. Last year, scientists reported that the proteolytic complex controls neurotransmission by cleaving proteins that enhance synaptic activity (see Restituito et al., 2011). Other research identified neurexin as a substrate for presenilin (PS), γ-secretase’s catalytic component, and showed that familial AD-linked PS1 mutations differentially affect neurexin processing (Bot et al., 2011; Saura et al., 2011). Moreover, mutations in neuroligin genes have been linked to cognitive deficits (see Südhof, 2008 review) including autism spectrum disorder (ASD) (Yanagi et al., 2012), schizophrenia (Sun et al., 2011), and mental retardation (Qi et al., 2009). In mice, synapses and social behavior deteriorate when they either overexpress (Dahlhaus et al., 2010) or lack (Varoqueaux et al., 2006) neuroligin 1. However, scientists understood little about the mechanisms regulating expression of this protein.

In the first study, lead author Kunimichi Suzuki and colleagues studied neuroligin 1 processing in adult rat cortex. In addition to full-length neuroligin 1 (NLGN1), they detected C-terminal fragments and even smaller NLGN1-derived peptides on immunoblots of membrane fractions. The smallest fragments did not appear in extracts treated with a γ-secretase inhibitor, indicating that γ-secretase was critical for NLGN processing. To confirm this, the researchers overexpressed neuroligins in embryonic fibroblasts from PS1/PS2 double knockout mice that completely lack γ-secretase activity. They saw a buildup of C-terminal fragments, an effect alleviated by overexpressing human PS1.

ADAM10 partners with γ-secretase to cleave various membrane proteins including APP, Notch, and cadherins—leading the scientists to wonder if ADAM10 cuts neuroligins as well. They overexpressed NLGN1 in embryonic fibroblasts from normal mice or ADAM10 knockouts, and found less soluble NLGN1 ectodomain in the ADAM10-deficient cells. The data finger ADAM10 as a sheddase for NLGN1. However, Suzuki and colleagues detected small amounts of the ectodomain even in the absence of ADAM10, leaving the possibility that additional proteases may make NLGN1’s extracellular cut, Tomita told Alzforum.

One such candidate—MMP9—emerged in the second study, which began with a different focus. The main thrust, said first author Rui Peixoto, now at Harvard Medical School in Boston, was to “look at how neuroligins are regulated by activity.” His team focused on mature neurons, given that NLGN1/2/3 triple knockout mice develop normal numbers of structurally sound synapses, yet have severe problems with synaptic transmission (Varoqueaux et al., 2006).

In mouse cortical neurons cultured 21 days in vitro (DIV21), Peixoto and colleagues saw NLGN1 levels plummet soon after they depolarized the cells, suggesting that neuronal activity regulates NLGN1 expression. Next, they explored possible mechanisms. After early experiments ruled out NLGN1 internalization and lysosomal degradation, they turned their attention to proteolytic processing. By treating the cultures with drugs that dampen or enhance neuronal activity, the scientists found that NLGN1 cleavage requires NMDA receptor activation and Ca2+/calmodulin-dependent kinase (CaMK) signaling. However, the NLGN1 fragments failed to show up in cells treated with a broad-spectrum matrix metalloproteinase (MMP) inhibitor. More specific drugs pinpointed MMP9 as the protease responsible for NLGN1 cleavage. Additional studies implicated γ-secretase in NLGN1 processing as well.

The functional impact of NLGN1 proteolysis seems to extend across the synapse—stunting growth of new spines on the postsynaptic cell and suppressing neurotransmitter release from the presynapse. The Japanese group saw more spines in primary rat neurons overexpressing an ADAM10-resistant form of NLGN1, compared to cells overexpressing the normal protein. In the second study, Peixoto and colleagues generated two NLGN1 mutants—one MMP9 resistant, the other with a thrombin site insert, making it cleavable on demand. DIV15-DIV25 neurons expressing non-cleavable NLGN1 churned out more transmitter, whereas presynaptic release was down in cells undergoing thrombin-induced NLGN1 proteolysis.

While each study identifies a different principal protease for NLGN1 shedding, both agree on further, activity-dependent processing of the postsynaptic protein by γ-secretase. The sheddase discrepancy could stem from developmental differences, Peixoto suggested. He noted that Suzuki and colleagues used younger neurons with still-developing synapses, whereas his team used DIV21 neurons with mature synapses. “It could well be that ADAM10 plays more of a role in basal and constitutive shedding events, whereas MMP9 is needed after synaptic-activity-dependent remodeling,” suggested Suzuki’s coauthor Paul Saftig, from Christian-Albrechts University in Kiel, Germany. Furthermore, Tomita said, the data do not “exclude the possibility that some other event is activating either or both proteases.”

Given the growing body of work linking NLGN1 mutations to cognitive dysfunction, Tomita’s group plans to examine whether NLGN1 processing by α- and γ-secretase is relevant to AD. These ideas will be discussed in the Alzforum Webinar.—Esther Landhuis.

Suzuki K, Hayashi Y, Nakahara S, Kumazaki H, Prox J, Horiuchi K, Zeng M, Tanimura S, Nishiyama Y, Osawa S, Sehara-Fujisawa A, Saftig P, Yokoshima S, Fukuyama T, Matsuki N, Koyama R, Tomita T, Iwatsubo T. Activity-Dependent Proteolytic Cleavage of Neuroligin-1. 18 Oct 2012;76:410-22. Abstract

Peixoto R, Kunz PA, Kwon H, Mabb AM, Sabatini BL, Philpot BD, Ehlers MD. Trans-Synaptic Signaling by Activity-Dependent Cleavage of Neuroligin 1. 18 Oct 2012;76:396-409. Abstract

Shipman SL and Nicoll RA. A Subtype-Specific Function for the Extracellular Domain of Neuroligin 1 in Hippocampal LTP. 18 Oct 2012;76:309-316. Abstract