CHAPTER I INTRODUCTION AND LITERATURE REVIEW pt 3

Neuroligin and neurexin structure and function:

Neurexins are type I single-pass transmembrane proteins with a complex extracellular domain structure similar to laminin A, slit and agrin, which are adhesion molecules implicated in axon guidance and synaptogenesis (reviewed in (Craig and Kang, 2007, Dean and Dresbach, 2006, Südhof, 2008)). Therefore, neurexins were hypothesized to act as cell-recognition molecules. Drosophila melanogaster and Caenorhabditis elegans each have a single neurexin gene. Humans have three neurexin genes, each of which has two alternative promoters. The upstream promoter allows transcription of longer forms, the α-neurexin transcripts. All three genes also have an internal promoter that leads to β-neurexin transcripts.

Neurexins are expressed predominantly in neurons. Mammalian neurexins have an unusually long and non-canonical signal sequence. α-Neurexins contain six LNS (laminin, nectin, sex-hormone binding globulin) domains, a transmembrane region and a short intracellular region that has a type I PDZ interaction site (Figure 5). The β-neurexins are essentially truncated α-neurexins, containing only a single LNS domain (Figure 6).

Neuroligins are type I single pass transmembrane proteins with a cleaved signal peptide followed by an extracellular cholinesterase-like domain. This domain has no cholinesterase activity; there is a Gly instead of the Ser residue of the Ser-His-Asp (Glu) catalytic triad. There are four neuroligin genes in humans, and NLGN4 has an X and Y chromosome form. Crystal structures for the extracellular domains of NLGN1 and NLGN4 have been solved both in isolation and in complex with βNrxn (Arac et al., 2007, Fabrichny et al., 2007). These studies show that the neuroligins exist as constitutive dimers, as previously suggested by functional studies (Dean and Dresbach, 2006). The specific arrangement of the neuroligin/neurexin complex consists of two neurexin protomers bound to opposite surfaces of the neuroligin dimer. This organization is consistent with these molecules bridging the 20 nm synaptic cleft. Data obtained from the crystal structures validates the crucial roles for alternative splice insertions in neuroligins and neurexins in modulating interactions between these proteins, and thereby regulating synapse specificity. Although the structure of the binding interface is well conserved within the neuroligin family, some variations exist, which could be responsible for differing binding affinities between various neuroligin and neurexin proteins. One unexpected conclusion from the structural studies is that the EF-hand motif of the neuroligins does not bind Ca⁺⁺ as previously thought. Instead, it was determined that the Ca⁺⁺ binding required for neurexin-neuroligin interaction is accomplished by two Ca⁺⁺-binding sites located at the neurexin/neuroligin interface (Arac et al., 2007, Fabrichny et al., 2007).

The cholinesterase-like domain is followed by a carbohydrate attachment region, a single transmembrane domain and a short C-terminal terminus containing a type I PDZ-binding motif (Figure 5 and Figure 6). Through this PDZ-binding motif, neuroligins interact with PSD95 (Post Synaptic Density protein-95kD) in mammalian glutamatergic neurons. PSD95 is a multi-domain scaffolding protein, which was previously thought to recruit neuroligin to the synapse. Dresbach and colleagues (Dresbach et al., 2004), however, used GFP-tagged deletion constructs of neuroligin-1 in cultured hippocampal neurons to show that neuroligin localization to synapses is not dependent on binding PSD95, nor on neuroligin-1’s extracellular domain. Instead, an eleven amino acid intracellular sequence distinct from its PDZ-binding motif targeted neuroligin to synapses. A neuroligin mutant lacking this intracellular sequence failed to localize to synapses.

NLGN1 and NLGN2 isoforms exhibit synapse-specific localization and function. NLGN1B is concentrated at glutamatergic synapses (Song et al., 1999, Banovic et al., 2010) and central cholinergic synapses, whereas NLGN2A is enriched at GABAergic and glycinergic synapses(Varoqueaux et al., 2004). NLGN3 has been detected at both glutamatergic and GABAergic synapses in vitro (Gutierrez et al., 2009), but localization of NLGN3 and NLGN4 in vivo is currently unclear. At glutamatergic synapses, NLGN1 binds to the scaffolding protein PSD95, which can then interact with NMDA-type glutamate receptors (Kornau et al., 1995, Tejedor et al., 1997). NLGN1 also may interact with the extracellular matrix protein thrombospondin (TSP) (Xu et al., 2010) which is secreted from astroglia (Christopherson et al., 2005). At GABAergic synapses, NLGN2 binds via the cytoplasmic tail to collybistin, or gephyrin (Graf et al., 2004, Poulopoulos et al., 2009) and is important for the postsynaptic localization of both glycine (Kirsch et al., 1993, Feng et al., 1998) and GABA_A receptors at many synapses (Essrich et al., 1998, Kneussel et al., 1999a, Kneussel et al., 1999b).

Vertebrate neurexin genes exhibit complex patterns of alternative splicing. This splicing is accomplished by usage of alternative exons, splice donor sites and splice acceptor sites. The alternative sequences are inserted at five sites (designated 1, 2, 3, 4, and 5) (Figure 4, Figure 5 and Table 1) in α-neurexins, two of which (sites 4 and 5) are also present in β-neurexins (Figure 4, Figure 6 and Table 1). This splicing potentially generates 3,524 alternative-splice variants (Ullrich et al., 1995, Chih et al., 2006). In the fruit fly D. melanogaster, the NRXN gene is less complex and apparently encodes fewer variants (Li et al., 2007).

Vertebrate neuroligin transcripts are alternatively spliced, although there is less complexity than with the neurexins. Alternative splicing occurs at one site in NLGN2 and NLGN 3, and two sites in NLGN1 (designated A and B). Different alternative exons (a 20 amino acids A1 (NLGN1 and NLGN3), a 17 amino-acid A2 (NLGN2 only) and a 20 amino-acid A2 (NLGN3 only)) can be included at position A. The alternative exon at position B inserts nine amino acids that form an N-glycosylation site. All of the insertions are in the extracellular domain (Ichtchenko et al., 1996)( Figure 4 and Figure 6). In total, alternative splicing potentially generates twelve variant isoforms of neuroligin, such as NLGN1(−) which lacks any insertions, and NLGN1B, which contains an insertion at site B (Table 2). NLGN4X and NLGN4Y do not appear to be alternatively spliced (Table 2 and Figure 4).

Splicing of β-neurexin confers specificity of binding to isoforms of neuroligin (Ichtchenko et al., 1995, Boucard et al., 2005, Chih et al., 2006, Comoletti et al., 2006, Ichtchenko et al., 1996). Ichtchenko and colleagues found that an insertion at an alternatively spliced LNS domain abrogated binding to Nlgn1. Subsequently, a group of studies by Boucard et al. (Boucard et al., 2005), Graf et al. (Graf et al., 2006) and Chih et al. (Chih et al., 2006) indicated that both the splicing at site 4 in β-neurexins, and splicing at site B in neuroligin-1 regulate binding selectivity and synapse function. Using Biacore (surface plasmon resonance), affinity chromatography and the artificial synapse induction assay, researchers were able to derive a “code” for binding affinities and activities for alternatively spliced neurexin and neuroligin. In essence, the presence of a 30-residue insertion at site 4 in β-neurexin reduces the affinity for neuroligin-1 proteins that contain an insertion at site B (+B). The β-neurexin site 4 insertion does not affect the high-affinity interaction with neuroligin-1 proteins that lack an insertion at site B (-B) or with neuroligin-2 proteins. Neuroligin-2 lacks any B splice site and exists only in the insertion-less form (Boucard et al., 2005, Graf et al., 2006, Chih et al., 2006).

The presence of an insertion at site 4 in α-neurexins does not affect their binding. All α-neurexins bind selectively to -B neuroligins (Boucard et al., 2005, Chih et al., 2006). Ability to bind β–neurexin was shown in these studies to be important for the artificial synapse induction, but not for growth of a synapse. Interestingly, the nine amino acid B-insertion itself is not what regulates the interaction of neuroligin-1 with specific neurexins, but instead, it is the N-linked glycosylation of this insert (Suckow et al., 2008, Chih et al., 2006). +B neuroligin-1 is the most abundant form of neuroligin-1 in rat brains (Chih et al., 2006).

Figure 4. A schematic diagram depicting the NLGN and NRXN mRNA alternative splice patterns

The teal and pink lines represent primary mRNAs. The red and yellow boxes represent alternative exons or exons with alternative splice sights. The blue and green boxes represent shared exons. The numbers 1-5 and letters A and B indicate different splice varients discussed and correspond to numbers 1-5 and letters A and B in figures 5 and 6 and the text.

Figure 5.Neuroligin αNeurexin complexes and variants

Nlgn-Nrxn complexes form through interactions via the ChE and LG domains, differ between αNrxn and βNrxn variants, and are further regulated by alternative splicing. Nlgns and Nrxns interact with multiple binding partners. Finally, α-dystroglycan (α-DG) interacts in splice variant-dependent manner with αNrxn2(−) variants, that is, interactions with LG domain 2 and 4 are inhibited by inclusion of splice insertions. Alterations in ligand binding due to alternative splice insertions range from subtle to large shifts in affinity, which modulate or abolish interactions, respectively.

Autism and Autism Spectrum Disorders (ASDs)

Features and prevalence of ASDs

Autism Spectrum Disorders (ASDs) are a constellation of developmental disorders that include classical idiopathic autism, Rett Syndrome, Asperger Disorder and PDD-NOS (pervasive developmental disorder not otherwise specified). ASDs are characterized by deficits in communication, deficits in social interactions and stereotypic/repetitive behaviors or restricted patterns of interests (Andres, 2002, American Psychiatric Association, 1994, Zwaigenbaum et al., 2005). Affected individuals may present with a delay in the acquisition of spoken language, repetitive use of language and/or motor mannerisms (e.g., hand-flapping), reduced or absent eye contact, a lack of interest in peer relationships, a lack of spontaneous or make-believe play, persistent fixation on parts of objects, hyperactivity, sensory disturbances and self injury. About 70% of individuals with autism have some degree of mental retardation. About 30% have a seizure disorder, irrespective of mental retardation.

The most recent data (Rice, 2009) indicate that the incidence of autism is 1 in 110 children in general, and almost 1 in 70 boys. These same government statistics suggest the prevalence of autism is increasing 10-17 percent annually. There is no generally accepted explanation for this increase, although improved diagnosis and environmental influences are two reasons often considered. The Autism society of America estimates that there are about 1.5 million affected individuals in the United States, and since the cost of proper care over the lifetime of an affected individual ranges between $3.5 million and $5 million, the annual cost of caring for the affected population is currently almost $90 billion annually (Autism Society of America, 2010).

Genetics and environment in the etiology of ASDs

Although it is well established that genetic mutations provide a significant risk factor for the development of autism spectrum disorders (Bailey et al., 1995, Muhle et al., 2004), it is also generally accepted that environmental factors have significant involvement. For example, human twin studies have shown that the concordance among monozygotic twins for a strict diagnosis of autism is 60%, which is approximately 12-fold higher than the concordance among dizygotic twins (Bailey et al., 1995, Muhle et al., 2004). Such data clearly provide strong evidence for a hereditary basis for autism, yet the lack of concordance in 40% of the monozygotic twins also points to the importance of non-hereditary (presumably environmental) factors in the emergence and/or severity of this pervasive developmental disorder.

Identification of "Autism genes"

Neuroligins and ASDs

One of the most striking results to emerge from the intensive international effort to identify “autism-related” genes is the demonstration that a subset of autism cases is associated with mutations in genes encoding neuroligins. It is now accepted that mutations affecting structural components of synapses (e.g., scaffolding molecules and adhesion proteins) provide a significant risk factor for the development of autism spectrum disorders (Bailey et al., 1995, Muhle et al., 2004). Many important studies including those by Jamain et al. (Jamain et al., 2003), Laumonnier et al. (Laumonnier et al., 2004) and others (Yan et al., 2005, Talebizadeh et al., 2006, Yan et al., 2008, Lawson-Yuen et al., 2008, Zhang et al., 2009) have shown that mutations in the genes encoding neuroligin-3 and neuroligin-4 are associated with autism spectrum disorders. Although subsequent studies (Gauthier et al., 2005, Vincent et al., 2004, Ylisaukko-oja et al., 2005) have shown that neuroligin mutations are only associated with autism in a limited number of family pedigrees, these associations provide a critical connection between defects in a specific type of synaptic molecule and a pervasive developmental disorder of the nervous system.

In NLGN3, a single amino acid substitution mutation, R451C, has been identified, and shown to have association with ASDs. In addition to the NLGN3 R451C mutation, several NLGN4 single amino acid substitution mutations have been shown to have association with autism. These NLGN4 mutations include G99S, K378R, V403M, R704C (Yan et al., 2005) and R87W (Zhang et al., 2009). ASD associated frame-shift mutations (Jamain et al., 2003, Laumonnier et al., 2004), copy number variations, deletion mutations (Lawson-Yuen et al., 2008) and regulatory mutations (Daoud et al., 2009) have also been found in NLGN4. Copy number variations in NLGN1 have also been found and shown to be associated with autism (Konstantareas and Homatidis, 1999).

Analysis in heterologous expression cell culture systems showed that the autism-related neuroligin-3 mutation R451C interferes with trafficking of the affected protein to the cell surface (Chih et al., 2004, Comoletti et al., 2004, Chubykin et al., 2005). Although Chubykin et al. (Chubykin et al., 2005) found normal activity of the small amount of protein that got to the cell surface, Chih et al. (Chih et al., 2004) and Comoletti et al. (Comoletti et al., 2004) both reported that the R451C mutation impairs NLGN3’s synaptogenesis-promoting activity.

Neuroligin Mutations in Mice: Phenotypes and Potential Relevance to Autism

In addition to the cell culture systems, mouse mutants have been used for modeling autism through ASD-associated mutations of neuroligin and/or neuroligin disruption. These mouse models led to some intriguing (and potentially autism-relevant) phenotypes. Jamain et al. (Jamain et al., 2008) found that a mouse knockout for neuroligin-4 leads to selective deficits in reciprocal social interactions and communication. Tabuchi et al. (Tabuchi et al., 2007) found that mice carrying a knock-in R451C allele of neuroligin-3 have impaired social interactions, but enhanced spatial learning abilities. A different group knocked the same R451C allele of neuroligin-3 into a different mouse strain, however, and found no substantial behavioral effect (Chadman et al., 2008). A plausible explanation for such a discrepancy rests on strain differences. It is possible that, in humans, the neuroligin mutation alone is not sufficient to trigger autism, and it appears that in mice also, genetic background is important.

Blundell et al. (Blundell et al., 2010) showed that knockout of Nlgn1 in mice results in impaired spatial memory and increased repetitive behavior. This could be due to a developmental defect in glutamatergic synapse maturation, since Wittenmayer (Wittenmayer et al., 2009) showed that synapses in cultured hippocampal neurons derived from Nlgn1 knockout mice remain structurally and functionally immature. Dahlhaus et al. (Dahlhaus et al., 2010) demonstrated that transgenic mice overexpressing Nlgn1 show significant deficits in memory acquisition.

In contrast, Blundell et al. (Blundell et al., 2009) and Jedlicka et al. (Jedlicka et al., 2010) showed that Nlgn2 deficient mice exhibited increased dentate gyrus excitability. Mutant mice lacking Nlgn2 displayed decreased inhibitory synaptic function with a discrete behavioral phenotype that includes a marked increase in anxiety-like behavior, a decrease in pain sensitivity and a slight decrease in motor co-ordination(Blundell et al., 2009). Deletion of Nlgn2 increases hippocampal cortical neuron excitability. This is probably due to impaired GABA_A receptor clustering (Jedlicka et al., 2010). Hines et al. showed that overexpression of Nlgn2 caused stereotyped behaviors, impaired social interactions and anxiety in transgenic mice (Hines et al., 2008).

Nlgn3-deficient mice have reduced ultrasound vocalization and a lack of social novelty preference (Radyushkin et al., 2009). The authors speculate that the latter may be related to an olfactory deficiency observed in the Nlgn3 mutants. Interestingly, some human ASD patients have olfactory deficits (Bennetto et al., 2007, Suzuki et al., 2003). Similarly, Jamain showed that mice lacking Nlgn4 had reduced social interaction and ultrasonic communication (Jamain, 2008).

Mutations affecting other synaptic proteins

Based on the association between neuroligin mutations and autism, a model emerged, first credited to Huda Zoghbi and then later to others (Zoghbi, 2003, Garber, 2007, Bourgeron, 2009), that emphasized the importance of perturbations of synaptic structure and/or synaptic transmission in the etiology of autism. Support for this model increased with reports that mutations in other synaptic proteins are also implicated in autism. A subset of autism cases was linked to mutations affecting neurexin, further strengthening the connection between autism and the neuroligin-neurexin synaptic interaction (Kim et al., 2008, The Autism Genome Project Consortium, 2007). Similarly, mutations in the genes encoding Shank2 and Shank3 (Figure 1), postsynaptic scaffolding proteins associated with neuroligin-containing synapses, are also associated with autism spectrum disorders (Grice and Buxbaum, 2006, Durand et al., 2007, Moessner et al., 2007, Pinto et al., 2010). Mice with haploinsufficiency of the Shank3 gene have deficits in synaptic function, social interaction and social communication (Bozdagi et al., 2010).

Table 3. Neuroligin point mutations in hNLGN4 and hNLGN3 associated with ASDs

*V403M may not be autism associated, or may be incompletely penetrant (Yan et al., 2005)

†R704C is equivalent to R714C in C. elegans, but is only present in worm isoforms that contain exon 13

hNLGN4       AALYYKKDKRRHETHR

NLG-1 (12→13)  IAVRREWGKKRRNEKK

NLG-1 (12→14) IAVRRELSLMPPPPPP

NLG-1 (12→15) IAVRREEPLLSASHKN

 ‡R87F88 is conserved in all neuroligins and all AChEs

The Use of C. elegans in Neurobiology

Since Sidney Brenner first proposed using C. elegans as a model organism in 1974 (Brenner, 1974), it has proven to be extremely useful for studying cell biology in general, and neurobiology, specifically. The strengths of C. elegans as a model organism include technical advantages, informational resources and available reagents. Some of the technical advantages include ease of genetic and genomic analysis (Fay), molecular cloning, transformation (Mello et al., 1991), behavioral analysis (Bargmann, Hart (ed.)) and immunohistochemistry (Duerr). These benefits, together with the wealth of biological information known about its simple nervous system (White et al., 1986), cell lineage (Sulston, 1976, Sulston and Horvitz, 1977), invariant anatomy and wiring, make C. elegans an extremely useful research system. In addition, C. elegans was the first multi-cellular organism to have its entire genome sequenced. 53% of human genes have a homolog in C. elegans (Reinke et al., 2000), and 35% of C. elegans genes are represented in the human genome. There are other Caenorhabditis species, including C. brenneri, C. briggsae, C. japanica and C. remanei, whose genomes have been sequenced. By comparing the genomes of these species, it is possible to identify conserved sequences, including putative regulatory elements in promoter regions. Available reagents include full coverage of cosmids, cDNAs, an RNAi library representing 17,000 genes and mutants (including deletion alleles, conditional alleles and hypomorphs).

It is now well established that C. elegans is an extremely powerful system for analyzing synapse structure and function. Together, the many labs that study C. elegans neurobiology have shown the molecular and functional equivalence of the mammalian and C. elegans genes, pathways and proteins involved in synapse structure and function. Strategies have been successfully employed to identify presynaptic components involved in active zone assembly and stabilization, including RPM-1/highwire, SYD-2/liprin-alpha and SAD-1 (Schaefer et al., 2000, Zhen et al., 2000, Zhen and Jin, 1999, Crump et al., 2001).

Research objectives

The long-term goal of this research is to develop C. elegans as a model for the study of neuroligin, and perhaps other autism-associated proteins, which would nicely complement existing cell culture and mouse-based systems. Therefore, the goal of my research has been to characterize the nlg-1 gene, its protein products and its function(s) in C. elegans. The gene characterization included determining the gene structure, ascertaining the presence or absence of alternative splicing, and documenting the expression pattern. The protein characterization include determining the subcellular localization of the protein, identifying and characterizing the phenotypes of nlg-1 mutant animals, identifying factors affecting localization and undertaking a structure-function analysis of selected domains of the NLG-1 protein. Additionally, we wanted to determine whether mammalian neuroligins could function to replace NLG-1, and the effect(s) of autism-associated point mutations on the localization and function of the C. elegans NLG-1 protein.