4 Neuroligins: Genes and Proteins Structure

17



Neurexins and Neuroligins



353



identity to their human orthologs, while the fourth rodent neuroligin

(NLGN4*) gene is significantly different and has diverged during evolution

from its corresponding human neuroligin 4 gene (Bolliger et al. 2008). Human

neuroligin genes are localized to 3q26 (NLGN1), 17p13 (NLGN2), Xq13

(NLGN3), Xp22.3 (NLGN4) and Yq11.2 (NLGN5 or NLGN4Y). All neuroligins are expressed in the central nervous system, where neuroligin 1 is localized

to the excitatory glutamatergic synapses, while neuroligin 2 is localized to the

inhibitory GABAergic synapses (Song et al. 1999, Varoqueaux et al. 2004).

Although recent findings suggest that neuroligin 3 might also be present in some

GABAergic synapses, it is mostly localized to glutamatergic synapses (Budreck

and Scheiffele 2007).

The extracellular structure of neuroligin is similar to acetylcholinesterase.

However, the amino acids, corresponding to the catalytic triad of acetylcholinesterase, are not conserved in neuroligins resulting in a loss of its enzymatic

activity. In addition, the substrate entrance site is completely inaccessible for

the substrate in neuroligins (Arac et al. 2007, Koehnke et al. 2008). Like

acetylcholinesterase, the extracellular domains of neuroligins interact and

form constitutive dimers (Comoletti et al. 2003). Two helices from each monomer form a four-helix bundle at the binding interface and the dimerization

process is primarily driven by hydrophobic interactions. Dimerization of postsynaptic neuroligins is probably important for the stable association with

presynaptic neurexins and/or for the activation of presynaptic signaling via

neurexin and consequential neurexin clustering. The significance of neuroligin

dimerization is underscored by the observation, that mutations of the putative

dimerization sequences in the four-helix bundle at the binding interface of the

neuroligin 1 monomer abolishes its ability to form synapses (Dean et al. 2003).

Although the EF hands within the neuroligin structure were hypothesized to be

Ca2+-binding sites, so far the experimental data have not been able to support

this hypothesis (Arac et al. 2007).



17.5 Splicing of Both Neurexins and Neuroligins Determines

Affinity and Specificity of Their Interaction

Neurexins and neuroligins bind to each other and form heterotypic intercellular

junctions, which are regulated by the alternative splicing of the primary transcripts encoding both molecules (Ichtchenko et al. 1995, 1996, Comoletti et al.

2003). This interaction is Ca2+ dependent, highly hydrophilic, and requires

water for stabilization of the neurexin–neuroligin complex. Therefore, it is

highly dependent on the Ca2+ and the ionic concentrations of the environment

(Chen et al. 2008). Binding to neuroligin is mediated by the last LNS domain of

a-neurexins or by the sole LNS domain of b-neurexins. Structurally, an LNS

domain is composed of two seven-stranded b-sheets forming a jelly roll with

structural similarity to lectins (Rudenko et al. 1999). Recent studies suggest that



354



A.A. Chubykin



the structural basis of the neurexin–neuroligin interaction specificity is determined by the selection of differentially spliced exons in the neurexin and

neuroligin transcript, respectively. In a neurexin molecule, SS#2, SS#3, and

SS#4 splice sites in different LNS domains all map to the loops that surround

the Ca2+-binding site and together form a so-called ‘hypervariable surface’ of

the LNS domains (Rudenko et al. 2001, Arac et al. 2007, Fabrichny et al. 2007,

Chen et al. 2008, Shen et al. 2008). This ‘hypervariable surface’ directly interacts

with the corresponding binding sites of neuroligins. Mutations of the amino

acids composing these hypervariable surfaces disrupt the interaction of neurexin with neuroligin and completely abolish synaptogenic activity (Graf et al.

2006). The binding of Ca2+ in the center of the hypervariable surface is essential

for establishing the complex with neuroligin and plays a key structural role in

stabilizing the complex. The presence of SS#4 insert results in major structural

rearrangements of the LNS domain. Specifically, part of the extended loop

close to the Ca2+-binding site attains a helical conformation and provides

additional protein contact points to the Ca2+ ion increasing the affinity of

Ca2+ ion binding. In addition, SS#4 is located in close proximity to one of the

salt bridges formed by the interaction between neurexin and neuroligin (Chen et

al. 2008). Thus, the insertion at splice site SS#4 probably changes the Ca2+binding affinity, as well as rearranges the topology of the ‘hypervariable surface’. In contrast, the presence of insert at SS#B in the neuroligin protein could

impede these rearrangements and disrupt the neighboring salt bridge, which is

necessary for the formation of the neurexin 1 b-SS#4:neuroligin 1+SS#B

complex (Shen et al. 2008).

Recent surface plasmon resonance (SPR) experiments have revealed that

neurexin 1b-SS#4 binds all neuroligin 1 isoforms, but has the highest affinity for

neuroligin 1-SS#B (Comoletti et al. 2006). Its interaction with neuroligin 2

(regardless whether the SS#A insert is included) is two orders of magnitude

weaker than that with neuroligin 1 (so far, the SS#B insert has only been

identified in neuroligin 1). Neurexin 1b-SS#4 also binds neuroligin 3 and

neuroligin 4. The preference for its interaction with these molecules is

NL1>NL4>>NL3>NL2 (Fig. 17.1b). On the other hand, the presence of the

SS#4 insert favors binding of neurexin 1b to any neuroligins without the SS#B

insert, and particularly to neuroligin 2.

Similar to neurexin splice site SS#4, neuroligin splice site SS#B determines

specificity of neuroligin binding to presynaptic neurexins. It is a master switch

between a more ‘promiscuous’ form that binds both a- and b-neurexins when

the insert at the splice site is absent and a more selective form that only binds to

b-neurexins when this insert is present (Boucard et al. 2005) (Fig. 17.1b,c).

Revisiting the expression profile data for neuroligins and neurexins, neuroligin

1 and neuroligin 3 are localized mostly to excitatory synapses, and neurexin

1b-SS#4 preferentially interacts with these ‘excitatory’ neuroligins. Neuroligin 2

is localized to inhibitory synapses and neurexin 1b+SS#4 preferentially interacts

with this ‘inhibitory’ neuroligin (Song et al. 1999, Budreck and Scheiffele 2007).

The expression pattern of neurexins +SS#4 and –SS#4 is consistent with these



17



Neurexins and Neuroligins



355



findings. In situ hybridizations have shown that neurexins including the SS#4

sequence have higher levels of expression in the striatum, substantia nigra, and

cerebellar nuclei, which predominantly have inhibitory neurons (Oertel and

Mugnaini 1984), while neurexins without the SS#4 sequence are enriched in the

pyramidal cell layers of the hippocampus which contains mostly excitatory

neurons (neurexins 1 and 2-SS#4 in CA3, neurexin 3-SS#4 in CA1–CA4) (Ichtchenko et al. 1995).



17.6 The Role of Neurexins and Neuroligins in Synapse

Formation and Stabilization

17.6.1 In Vitro Synapse Formation Assays

The function of neurexins and neuroligins has long been unclear until it was

demonstrated in an artificial synapse formation assay. These results demonstrated that neuroligin 1, when expressed in non-neuronal HEK293 cells, can

induce presynaptic differentiation in co-cultured pontine explants (Scheiffele

et al. 2000). Neuroligin expression alone was sufficient to induce presynaptic

differentiation of neurons. The presynaptic terminals formed by the neurons in

these co-cultures were morphologically indistinguishable from regular neuronal

synapses. They contain synaptic vesicles, which are filled with neurotransmitters,

and can be released by hypertonic solutions or by high potassium buffers. In cocultures with neuronal cells, the co-expression of neuroligins and glutamate

receptors in non-neuronal cells formed ‘minimal synapse’. Lipophilic FM dye

stainings and direct whole-cell patch clamp recordings of glutamate currents in

these non-neuronal cells have shown that these ‘minimal synapses’ are fully

functional. Such preparations facilitated the characterization of different glutamate receptors subunit kinetics, as well as some of the presynaptic properties

exhibited by these ‘minimal synapses’ (Fu et al. 2003, Sara et al. 2005).

The results that were obtained from these artificial synapse formation assays

suggest that the exposure of the neurons to large enough amounts of neuroligin

may be sufficient to induce formation of the presynaptic terminals. This conclusion was supported by the finding that substrate-immobilized recombinant

neuroligin 1 promotes the clustering of synaptic vesicle antigens in cultured

neurons (Dean et al. 2003). Similarly, antibody cross-linking of epitope-tagged

neurexins, which were transfected into neurons, brings the neurexin molecules

together independently of neuroligins and causes the same effect (Dean et al.,

2003). Together with the requirement of neuroligin dimerization for synaptogenesis, these findings raised a hypothesis, that presynaptic clustering of neurexins is

involved in either the formation or the stabilization of presynaptic terminals. This

raised the question whether the neurexin–neuroligin signaling is bidirectional. As

neurexins could similarly induce postsynaptic differentiation in the dendrites of

the neurons, which contact the non-neuronal neurexin-expressing cells (Graf



356



A.A. Chubykin



et al. 2004), trans-synaptic neurexin–neuroligin complexes appear to function

bidirectionally.

Artificial synapse formation assays revealed how the neurexin–neuroligin

splice code regulates the specificity of synaptogenesis. Culturing dissociated

hippocampal neurons with non-neuronal cells, which were transfected with

neurexin 1b-SS#4 constructs, results in the clustering of both the PSD-95

excitatory and the gephyrin inhibitory synaptic marker. This suggests an induction of both excitatory and inhibitory postsynaptic differentiations by neurexin

1b-SS#4 (Graf et al. 2006). On the other hand, neurexin 1b+SS#4 preferentially promotes the clustering of the gephyrin and does not induce PSD-95

clustering (Boucard et al. 2005, Chih et al. 2006, Comoletti et al. 2006, Graf et

al. 2006). It is important to note though that the lack of preference for excitatory postsynaptic clustering by the neurexin 1b-SS#4 can be explained by the

low sensitivity of the artificial synapse formation assays. The quantity of the

recombinant protein that is expressed on the surface of a non-neuronal cell is

many times higher than in normal neurons. Therefore, even a low-affinity

interaction between particular neurexin and neuroligin isoforms is potentially

able to induce the clustering of corresponding synaptic markers. The extent of

synaptogenesis is correlated with the amounts of neuroligin exposed on the

surface of a dendrite to the axons. This suggests that the surface levels of

neuroligin in vivo must be tightly controlled (Chubykin et al. 2005).



17.6.2 Intracellular Signaling of Neurexins and Neuroligins

The bidirectional neurexin–neuroligin signaling involves intracellular interactions between the C-terminal regions of neurexins and neuroligins with synaptic

scaffolding molecules. On the presynaptic side, the PDZ-binding motif of the

C-terminal neurexin region binds Ca2+/calmodulin-activated Ser-Thr kinase

(CASK). CASK is an unusual protein kinase and a member of the membraneassociated guanylate kinases (MAGUK) family of adaptor proteins (Hata et al.

1996, Mukherjee et al. 2008). Several different interacting partners of CASK

have been identified. First, it interacts stoichiometrically with Velis and Mint-1

(Butz et al. 1998). In addition, it also binds protein 4.1. Together with neurexins

this complex was shown to be a potent nucleator center for actin polymerization

(Biederer and Suădhof 2001). CASK can also bind Ca2+ and K+ channels and

potentially clusters these channels at the presynaptic terminal (Maximov,

Suădhof and Bezprozvanny 1999, Leonoudakis et al. 2004). K+ channels can

be phosphorylated by CASK and thereby induce an enhancement of K+

currents (Marble et al. 2005). CASK can also phosphorylate the neurexin

C-terminal region, a process that is dependent on neuronal activity. After an

inhibition of N-methyl-D-aspartic acid (NMDA) receptors and Na+ channels

the CASK-dependent phosphorylation of b-neurexin is increased more than

twofold (Mukherjee et al. 2008).



17



Neurexins and Neuroligins



357



The importance of CASK for clustering Ca2+ channels at the presynaptic

terminal is suggested from the findings that a-neurexin knockout mice have

impaired presynaptic Ca2+ channel functions. While the total number of the

surface Ca2+ channels is unchanged in these animals, Ca2+ currents are

severely suppressed. In addition, the deletion of a-neurexin genes severely

impaired the evoked synaptic inhibitory transmission in the neocortex and the

excitatory transmission in the brainstem. Furthermore, the frequencies of both

inhibitory and excitatory miniature postsynaptic currents (mPSCs or minis) are

drastically decreased in a-neurexin knockout animals. These changes in synaptic physiology are associated with a twofold decrease in the density of inhibitory

GABAergic terminals in both the brainstem and the neocortex (Missler et al.

2003). The splice site code theory may provide an explanation why only the

density of GABAergic synapses is decreased. a-neurexins only interact with

neuroligins-SS#B and dystroglycans (specific for GABAergic synapses), while

b-neurexins bind both neuroligins-SS#B and neuroligins+SS#B. Insert at the

splice site SS#B has been identified only in neuroligin 1, therefore, the neuroligin

1 isoform with the SS#B sequence is unable to bind a-neurexins. Consequently,

a subpopulation of inhibitory synapses that expresses neuroligin 2-SS#B and

interacts with a-neurexins would be eliminated, while the subpopulation of

excitatory synapses that expresses neuroligin 1+SS#B and interacts only with

b-neurexins would be spared.

Considering the involvement of CASK, it is tempting to hypothesize that aneurexin may provide a nucleation core for the formation of the presynaptic

machinery and that CASK binding is necessary for the successive clustering of

the Ca2+ channels close to the proper synaptic vesicle fusion sites. Once this

recruitment is accomplished, CASK might modulate Ca2+ channels function

by phosphorylation of the channel protein (Atlas 2001). The deletion of aneurexins may then result in a mislocalization of Ca2+ channels at the presynaptic terminals (Missler et al. 2003).

The C-terminal region of all neuroligins contains a PSD-95-Dlg-ZO homology (PDZ)-binding motif, which can interact with the third PDZ domain

of PSD-95 and the second PDZ domain of synaptic scaffolding molecule

(S-SCAM). S-SCAM is also known as membrane-associated guanylate kinase

with inverted domain organization (MAGI) (Fig. 17.1c). Interestingly, S-SCAM

also binds the proline-rich region of neuroligins, which is located upstream of the

PDZ-binding motif and this binding might be involved in the proper localization

and trafficking of neuroligins to synapses (Iida et al. 2004). In addition to PSD-95

and S-SCAM, neuroligins also interact with the Shank protein in yeast-twohybrid assays. Shanks represent another family of scaffolding proteins that is

involved in spine formation and metabotropic glutamate receptor recruitment

(Sheng and Kim 2000, Meyer et al. 2004). PDZ domains of PSD-95 and S-SCAM

also bind the C-terminal sequences of K+ channels and NMDA receptors, and

thus may potentially mediate clustering of K+ channels and NMDA receptors in

the postsynaptic density (Irie et al. 1997, Hirao et al. 1998).



358



A.A. Chubykin



17.6.3 The Link Between Cell Adhesion and Synaptic Plasticity

Specific localization of neuroligin 1, 3 and neuroligin 2 to excitatory and

inhibitory synapses, respectively, suggests that their selective expression may

dictate the type of synapse formed. Indeed, overexpression of neuroligin 1 and 2

in primary dissociated neurons selectively increases either the density of glutamatergic or of GABAergic synapses, respectively. This in turn raises the frequency of corresponding miniature postsynaptic currents (mPSCs) and the

evoked response, either excitatory for neuroligin 1 or inhibitory for neuroligin

2 expression. Interestingly, neuroligin 1 overexpression has stronger effect on

the NMDA receptor-mediated excitatory postsynaptic currents (EPSCs) than

the alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)

receptor-mediated EPSCs and results in an increase of the NMDA/AMPA

ratio (Prange et al. 2004, Chubykin et al. 2007). This suggests that neuroligin

1 may induce the formation of so-called ‘silent synapses’ that lack AMPA

receptors, which may become stabilized by synaptic activity and ‘activated’ by

AMPA receptor recruitment. The observation that neurexin 1b, which is

expressed in PC12 cells, induces the clustering of PSD-95 and of NMDA, but

not of AMPA receptors in contacted dendrites of the co-cultured hippocampal

neurons, supports this hypothesis. AMPA receptors may be later recruited into

these contacts by glutamate application or by the neuronal overexpression of a

constitutively active form of calmodulin kinase II (Nam and Chen 2005).

Alternatively, NMDA currents could be regulated by neuroligin 1 through its

interaction with PSD-95 or S-SCAM and the subsequent recruitment and

stabilization of NMDA receptors at the postsynaptic density.

Consistent with the overexpression studies in cultured neurons, transgenic

mice, which overexpress neuroligin 2, have an increased frequency of mIPSCs

and a decreased balance of excitatory to inhibitory neurotransmission (E/I)

(Hines et al. 2008). Experiments with acute cortical and hippocampal slice

preparations from neuroligin knockout mice have also shown results consistent

with culture experiments. In neuroligin 1 knockout mice the NMDA receptormediated neurotransmission is decreased, while inhibitory neurotransmission

remains unaffected. At the same time, in neuroligin 2 knockout mice the

amplitude of IPSCs is decreased, while there is no change in excitatory neurotransmission (Chubykin et al. 2007). In addition, acute suppression of neuroligin 1 expression in lateral amygdala using lentiviral-mediated delivery of small

hairpin RNA (shRNA) against neuroligin 1 resulted in the specific decrease of

NMDA receptor-mediated EPSCs and the subsequent decrease of NMDA/

AMPA ratio in thalamo-amygdala synapses of principal neurons. There is no

change in the intrinsic electrophysiological properties or the input–output

curves. Furthermore, acute suppression of neuroligin 1 expression abolishes

the induction of NMDA receptor-dependent form of long-term potentiation

(LTP) in the amygdala. This in turn resulted in deficits of both contextual- and

cued-fear memory. These results suggest that neuroligin 1 expression is



17



Neurexins and Neuroligins



359



necessary for NMDA receptor-mediated neurotransmission and for synaptic

plasticity, which is involved in the storage of long-term memory in the amygdala (Kim et al. 2008b).

Interestingly, recent reports further support the role of the neurexin–neuroligin

trans-synaptic complex in bidirectional signaling. The neurexin–neuroligin

complex alters presynaptic short-term plasticity by increasing the sensitivity of

the presynaptic release machinery to the extracellular Ca2+ concentration (Futai

et al. 2007). These findings are consistent with the previous observations that in

a-neurexins knockout mice the presynaptic Ca2+ channel function is impaired

(Missler et al. 2003). This signaling process could be important for synchronizing

synapse stabilization and elimination between pre- and postsynaptic sites.

Furthermore, it might also provide a potential feedback mechanism for homeostatic changes.

Such homeostatic changes might account for the observation that neuroligin

and a-neurexin knockout mice still form synapses. Thus, the neurexin–neuroligin

trans-synaptic complexes do not appear to play a role in the initial formation of

synapses, but rather in their activity-dependent stabilization (Varoqueaux et al.

2006). This activity-dependent stabilization involves an intricate interplay

between the Ca2+ channels, CASK, and neurexins on the presynaptic side. On

the postsynaptic side, NMDA receptors, PSD-95, or S-SCAM cooperate with

excitatory neuroligins (Fig. 17.1c) and GABA receptors and dystroglycans with

inhibitory neuroligins (Fig. 17.1d).



17.7 Neurexin and Neuroligin Gene Polymorphisms in Autism

Spectrum Disorders and Mental Retardation

Autism is a severe disorder of the nervous system, which is characterized by

difficulties in social interactions and is often accompanied by learning disabilities. Autistic patients also have perceptual processing abnormalities, which are

expressed in a hypersensitivity to auditory and tactile stimuli (Kootz et al.

1981). They also exhibit impairments in executive functions and motor control,

procedural, emotional and social memory (Squire and Zola 1996). The mechanisms underlying the development and progression of autism are still unknown.

There is a significant evidence of a genetic etiology for autism. The rate of

occurrence of autism is about 1/500 and twin studies show that the concordance

rate is 70–90% in monozygotic twins and 0–10% for dizygotic twins (Steffenburg et al. 1989, Blasi et al. 2006, Szatmari et al. 2007). The incidence of disease

for male versus female affected is 4:1 for autism and 8:1 for Asperger syndrome.

Male predisposition to these disorders may be partially explained by the growing number of identified X-linked mutations. This includes mutations in two Xlinked genes encoding neuroligins NLGN3 and NLGN4 in siblings with autism

spectrum disorders (Jamain et al. 2003). A number of mutations in the NLGN3

and the NLGN4 gene cause pathological phenotypes. A missense mutation in



360



A.A. Chubykin



human neuroligin 3 (R451C) and two frameshift mutations in neuroligin 4 have

been identified in patients with familial forms of autism, mental retardation,

and/or Asperger syndrome (Jamain et al. 2003, Zoghbi 2003, Laumonnier et al.

2004). As the protein is truncated in the middle of the acetylcholinesterase

domain, the neuroligin 4 mutation probably acts as a null allele.

The R471C mutation in rat neuroligin 3 has been well characterized. This

mutation (which corresponds to the human R451C substitution) results in an

unpaired cysteine residue and the partial retention of neuroligin 3 protein in the

endoplasmic reticulum (Chih et al. 2004, Comoletti et al. 2004). Since neuroligin

3 can form heterodimers with neuroligins 1 and 2 (Comoletti et al. 2006,

Budreck and Scheiffele 2007), this might explain a dominant-negative gain-offunction effect, which is induced by this mutation. Neuroligin 3 is localized

mostly to glutamatergic synapses. Therefore, the R451C substitution mutation

may result in formation of dimers between the mutant neuroligin 3 form with

wild-type neuroligin 1, thereby disrupting the balance of excitatory versus

inhibitory synapse formation and stabilization. This hypothesis is supported

by the findings that overexpression of the mutant protein in dissociated neurons

results in a massive decrease of synapse density and, consequently, of synaptic

transmission. This mutation specifically impairs the balance of excitatory to

inhibitory neurotransmission (E/I balance) (Chubykin et al. 2007). This disruption of the E/I balance has recently been proposed as a potential mechanism of

autism pathophysiology. The characterization of the R451C mutation in mice

has given further support to this hypothesis. However in contrast to the human

mutation, in the mouse model the shift in the E/I balance was caused by an

increase in inhibitory, rather than a decrease in excitatory neurotransmission.

In addition, these R451C mutant mice also have an increased density of

GABAergic synapses. This is consistent with the observed changes in synaptic

physiology. One of the potential explanations for these discrepancies between

humans and mice could be compensatory homeostatic changes.

The recent identification of various new mutations in the neurexin/neuroligin synapse formation/stabilization pathway suggests that this pathway may

hold significant insights into the pathophysiology of autism spectrum disorders

and mental retardation. In addition, it may provide new candidates for the

mutation screening (also see Chapter 6). Neurexin autism mutations have

recently been discovered (Szatmari et al. 2007, Kim et al. 2008a). In addition,

a gene for contactin-associated protein-like 2 (CNTNAP2), a member of the

superfamily of neurexin-like proteins, has also been identified as an autismsusceptibility gene (Baumgartner et al. 1996, Alarcon et al. 2008, Bakkaloglu et

al. 2008). Mutations in the CASK gene have been reported in patients with the

FG syndrome, which is characterized by developmental abnormalities and

mental retardation (Piluso et al. 2003). Also mutations in the Shank 3 gene

have recently been found in patients with autism spectrum disorders (Durand et

al. 2007). CASK binds to the C-terminal region of neurexins, while Shanks are

putative interacting proteins for neuroligins. Interestingly, Shanks have been

shown to be involved in spine formation and metabotropic glutamate receptor



17



Neurexins and Neuroligins



361



recruitment (Sheng and Kim 2000, Meyer et al. 2004). Metabotropic glutamate

receptors are important for various forms of synaptic plasticity, including longterm depression (LTD). Loss of fragile X mental retardation protein (FMRP)

causes fragile X syndrome in humans and increases protein synthesis-dependent

synaptic plasticity, which is mediated by metabotropic glutamate receptors

(Bear et al. 2004). Thus, there are currently two major theories of autism

grouped by the types of genes mutated in the disease: the first group includes

mutations of the proteins in the neurexin/neuroligin pathway, the second group

includes mutations of the proteins involved in various forms of pre- and postsynaptic plasticity, for example, FMRP (Huber et al. 2002), methyl CpGbinding protein 2 (MECP2) (Moretti et al. 2006), and metabotropic glutamate

receptors (Serajee et al. 2003). Shank proteins may represent a bridge between

the two theories of autism, which involve synapse formation/maintenance and

synaptic plasticity.

One way to unify these two hypotheses to explain the mechanism of autism is

to postulate that mechanisms of synaptic plasticity involve neurexin/neuroligin

signaling in selecting which synapses are stabilized or will be eliminated. Consequently, an impairment of signaling at any step of this pathway may potentially result in mental retardation or autism. In line with this hypothesis, autism

mutants of the proteins involved in different steps of the neurexin/neuroligin

pathway manifest similar impairments in synaptic morphology and physiology.

They also often exhibit similar behavior characteristics of the disorder, such as

impaired social interactions and reduced ultrasound vocalizations (Jamain et al.

2008). Surprisingly, in addition to these impairments, some of the autism

models show improvements in certain types of memory, i.e., an enhanced

spatial learning and memory, which results in an improved performance in

Morris water maze tests (Tabuchi et al. 2007, Hung et al. 2008).

Many of the newly discovered mutations in neurexin and neuroligin genes, as

well as their signaling partners represent de novo mutations. Germ-line mutations increase with age, thus increasing the risk for older parents to have

children with autism. This might contribute to the recent reported increase of

autism cases (Sebat et al. 2007, Zhao et al. 2007).



17.8 Conclusions, the Concept of a Synaptic Code

and Future Directions

The alternative splicing of the various neurexin transcripts may yield more than

1000 protein isoforms. Although alternative splicing has been reported for

other genes, most of those cases result in less than 100 variants. Such a high

degree of receptor diversity is one of the well-established mechanisms to achieve

a large number of unique specificities for molecular interactions. In nervous

system it may allow for an almost unlimited potential for combinatorial



362



A.A. Chubykin



variations of different neurexin isoforms in different cell types and possibly even

individual cells.

Consequently, more detailed characterization of the various neurexin and

neuroligin isoforms and their expression patterns will be essential for our

understanding of the postulated neurexin–neuroligin-dependent synaptic

code. Understanding this neurexin–neuroligin-dependent code might help us

to learn how the brain is wired by specific connections between different cell

types and by individual neurexin–neuroligin pairs, which specify different types

of synapses. More modern, high-throughput, sensitive assays will shed more

light on synaptic complexity in vertebrate brains. One of the predictions would

be that the expression of neurexins–SS#4 is mostly in excitatory neurons, and

neurexins+SS#4 in inhibitory neurons. An individual neuron may express

either neurexin-SS#4 or neurexin+SS#4 isoform, but not both. The SS#4 splice

site could represent one of the potential markers of the neuronal type. Consequently, investigation of how the splicing is regulated becomes very important,

particularly, identification of the cis elements in the neurexin and neuroligin

genes, as well as the trans-acting splicing factors interacting with these elements.

A growing body of evidence suggests that the activity-dependent regulation

of neurexins and neuroligins at both excitatory and inhibitory synapses may be

involved in synaptic plasticity. However, we will need a considerably better and

more detailed knowledge of how synaptic activity is regulated by neurexins and

neuroligins at the molecular level, before we will understand the role of these

molecules in higher cognitive functions, such as memory formation and

behavior.

Acknowledgments The author thanks Thomas C. Suădhof and Dilja Krueger for their helpful

comments on the manuscript.



References

Alarcon M, Abrahams BS, Stone JL et al. (2008) Linkage, association, and gene-expression

analyses identify CNTNAP2 as an autism-susceptibility gene, Am J Hum Genet

82:150–159

Arac D, Boucard AA, Ozkan E et al. (2007) Structures of neuroligin-1 and the neuroligin-1/

neurexin-1 beta complex reveal specific protein-protein and protein-Ca2+ interactions,

Neuron 56:992–1003

Atlas D (2001) Functional and physical coupling of voltage-sensitive calcium channels with

exocytotic proteins: ramifications for the secretion mechanism, J Neurochem 77:972–985

Bakkaloglu B, O’Roak BJ, Louvi A et al. (2008) Molecular cytogenetic analysis and resequencing of contactin associated protein-like 2 in autism spectrum disorders, Am J Hum Genet

82:165–173

Barresi R and Campbell KP (2006) Dystroglycan: from biosynthesis to pathogenesis of

human disease, J Cell Sci 119:199–207

Baumgartner S, Littleton JT, Broadie K et al. (1996) A Drosophila neurexin is required for

septate junction and blood-nerve barrier formation and function, Cell 87:1059–1068