Introduction

Synaptic transmission between neurons in a circuit is the fundamental organizing principle of brain function. It is generally assumed, therefore, that disruptions in the formation, patterning, or maintenance of synapses underlie a wide range of debilitating neurological and psychiatric disorders. As tautologous as this assumption may be, its usefulness for understanding and, eventually, treating such disorders will require identification of the molecules that control synapse development in the central nervous system (CNS). Much progress has been made in the past decade, with many of the candidates thus far identified being synaptic adhesion molecules, including the "classical" cadherins (see Chapter 5); immunoglobulin superfamily members such as the sidekicks1, NCAM (see Chapter 6), SynCAM (see Chapter 8), and the SYGs2,3; and the neuroligins/ neurexins (see Chapter 7; for recent reviews, see refs. 4-6). The critical roles identified for such a diverse group of synaptic adhesion molecules provide experimental support for the

♦Department of Biological Sciences and Neuroscience Graduate Program, University of Iowa, 143 Biology Building, Iowa City, IA 52242, USA; [email protected]

classic "chemoaffinity" hypothesis of Sperry7, which suggested that the specificity of synaptic patterning could be mediated by cell-surface tags interacting in a "lock-and-key" manner to join pre- and postsynaptic partners.

The ~60 protocadherin adhesion molecules encoded by the Pcdh-a, -pP, and -y gene families have, since their discovery8, been widely discussed as leading candidates for mediating the molecular recognition of synaptic partners due to their diversity, combinatorial neuronal expression, and partial synaptic localization. In this chapter, I review what is known about the organization, expression, and function of the clustered protocadherin genes. Particular attention is paid to the y-protocadherins, for which our recent work has identified critical functions in synapse development and neuronal survival9,10.

3. CLONING OF PROTOCADHERIN GENES AND ORGANIZATION OF THE Pcdh-a, p AND -yGENE CLUSTERS

The term "protocadherin" was coined in 1993 by Shintaro Suzuki and co-workers11 to refer to two novel cDNAs isolated in a degenerate oligonucleotide PCR screen for sequences homologous to the "classical" cadherins. Since then, nearly 70 protocadherin genes have been identified, all of which exhibit varying numbers of ectodomains similar to those of the classical cadherins that mediate Ca2+-dependent cell-cell adhesion. This makes the protocadherins the largest and most diverse group in the cadherin superfamily, which also includes the type I and II classical cadherins (e.g., ^-cadherin, £-cadherin, cadherin-7) that form adherens junctions, the desmocollins and desmogleins that form desmosomes, the 27- or 34-ectodomain proteins Dachsous and Fat, and the 7 transmembrane-domain Flamingo cadherins (reviewed in refs. 12,13; see Chapter 5).

The genes isolated by Sano et al.11, initially termed protocadherin-42/Pcdh1 and protocadherin-43/Pcdh2, encoded transmembrane proteins with six or seven cadherin ectodomains. The cytoplasmic domains of these protocadherins, however, differed significantly from those of the classical cadherins, apparently lacking their characteristic catenin-binding sites. These and several other protocadherin genes subsequently isolated by Suzuki's group14-18 and by that led by Takeshi Yagi19 were found to be widely expressed in the brain and, in some cases, localized to synapses, providing the first hint of the importance of protocadherins for nervous system function.

A major advance in the field came in 1999, when Wu and Maniatis discovered over 50 genes encoding three families of protocadherins (termed Pcdh-a, -P, and -f) clustered in a tandem array spanning ~750 kilobase on human chromosome 5q318. An orthologous array with a similar arrangement was subsequently found on mouse chromosome 1820 and, with some variation, in chimpanzee21, rat21,22, chicken23, zebrafish21,24,25, and even coelacanth26. These observations extended, and gave coherence to, a number of previous reports of novel protocadherin genes that are now known to be members of the Pcdh-a, -P, or -y loci: for example, protocadherin-43/Pcdh2n,15 and protocadherin2Cn reside in the Pcdh-y locus, Pcdh3lA is a member of the Pcdh-P family, and the eight genes reported as cadherin-related neuronal receptors (CNRs ; ref. 19) make up more than half of the Pcdh-a family.

The organization of the Pcdh-a, -P, and -ygene clusters is reminiscent of the immunoglobulin and T-cell receptor genomic loci, in that multiple "variable" (V) exons encoding variant isoforms are arrayed in tandem and, for the Pcdh-a and -y clusters, are spliced to three short, invariant "constant" (C) exons (Figure 9.1; the Pcdh-fi locus is similar but lacks any constant exons; refs. 8,20,21). Each large V exon encodes the six cadherin ectodomains, the transmembrane domain, and the proximal cytoplasmic domain of a single Pcdh isoform, with the C-terminal having 125-150 amino acids of the a- and y-Pcdhs encoded by their respective C exons (Figure 9.1). The resulting a- and y-Pcdh proteins are thus exactly the kind of molecules that one might predict would underlie a Sperry-like "chemoaffinity" mechanism: their divergent extracellular domains could engage in a variety of specific homo- or heterophilic interactions at the cell surface, while their shared cytoplasmic domain could be involved in conserved signaling mechanisms leading to the assembly or stabilization of the synaptic apparatus. In mice, there are 14 Pcdh-a, 22 Pcdh-P, and 22 Pcdh-yV exons (20; see Figure 9.1). The Pcdh-ygenes can be further grouped into three subfamilies, termed a (12 genes), b (7 genes), and c (3 genes), based on sequence similarity (Figure 9.1). Two of the Pcdh-agenes are referred to as c1 and c2 because their similarity to the Pcdh-yc subfamily (c3, c4, and c5) is greater than their similarity to the other 12 Pcdh-a genes8,21.

Figure 9.1. The Murine Pcdh-a, -p, and -y Gene Clusters. Top: a schematic representation of the three clustered Pcdh loci on ~900 kilobases of mouse chromosome 18. V, variable exon; C, constant exon. Middle: an expanded view of the Pcdh-ylocus, showing the three subfamilies (A, B, and C) of V exons. Bottom: Each V exon, which encodes most of a single y-Pcdh isoform, is expressed from its own promoter and subsequently spliced to the C exons, which encode a shared C-terminal cytoplasmic domain.

Figure 9.1. The Murine Pcdh-a, -p, and -y Gene Clusters. Top: a schematic representation of the three clustered Pcdh loci on ~900 kilobases of mouse chromosome 18. V, variable exon; C, constant exon. Middle: an expanded view of the Pcdh-ylocus, showing the three subfamilies (A, B, and C) of V exons. Bottom: Each V exon, which encodes most of a single y-Pcdh isoform, is expressed from its own promoter and subsequently spliced to the C exons, which encode a shared C-terminal cytoplasmic domain.

This intriguing genomic organization raised the possibility that the Pcdh genes could be substrates for a long-hypothesized somatic DNA recombination mechanism in the CNS, similar to that employed by the immune system to generate the vast repertoire of antibodies and T-cell receptors27-30. Subsequent detailed analyses of Pcdh-a and -y gene expression, however, have failed to provide support for such a mechanism, demonstrating instead that expression is controlled by promoter choice and alternative V exon splicing31,32. The sequence upstream of each Pcdh V exon contains its own promoter region, including an approximately 20-base pair conserved sequence element that is required for expression. Through mechanisms that remain obscure, a given V exon promoter is "chosen" and transcription through the remaining portion of the Pcdh-a or -y cluster proceeds; intervening V exons are then removed when the 5' V exon is cis-spliced to the downstream C exons31,32. Although the Pcdh-P cluster does not contain its own C exons, each Pcdh-ftV exon harbors a consensus 5' splice site near its end8, suggesting the possibility of splicing to the C exons of other clusters. Such intercluster spliced transcripts, while rare, are apparently present in neurons, as are low levels of aV/yC and yV/aC hybrid transcripts produced by trans-splicing between separate pre-mRNA intermediates31-33. Even given that such intercluster splicing is rare, it has the potential to greatly increase the already-significant diversity of the proteins encoded by the Pcdh-a, -ft, and -yclusters.

Although the adhesive capability and specificities of most of the clustered protocadherins remain unknown, the extant studies on individual ft- or y-Pcdh isoforms11,14,15,34 suggest that they mediate calcium-dependent, homophilic adhesion that is weaker than that of the classical cadherins. For one y-Pcdh isoform (yC3/Pcdh2), homophilic adhesion could be strengthened by replacing its cytoplasmic domain with that of £-cadherin15. Interestingly, existing data suggest that the a-Pcdhs may not be homophilic adhesion molecules; instead, an a-Pcdh/CNR isoform (a4/CNR1) was found to mediate heterophilic adhesion to integrins via an RGD site in its first ectodomain35. It has also been reported that several a-Pcdh/CNR proteins can bind reelin, an extracellular matrix-associated protein critical for lamination of the cerebral cortex, also in an RGD-dependent manner36. A function as reelin receptors would, in fact, be consistent with expression studies showing juxtaposition of CNR-positive and reelin-positive cells in the cortex and spinal cord36,37. Other researchers have, however, been unable to replicate the binding of a-Pcdhs to reelin38,39; this, combined with extensive validation of the lipoprotein receptors ApoER2 and VLDLR as reelin receptors40, makes it unclear at present whether a-Pcdhs play any role in reelin signaling.

4. EXPRESSION AND LOCALIZATION OF THE a-, ft-, AND y-PROTOCADHERINS IN THE NERVOUS SYSTEM

The initial studies of individual protocadherin genes now known to reside within the Pcdh-a, -ft, or -yclusters found that their expression was essentially limited to the nervous system11,14,16,18,19. This has been confirmed by more recent studies examining the expression of entire clusters by using riboprobes against the Pcdh-a or -y constant exons; essentially all neurons express Pcdh genes, and outside of the nervous system, only weak lung and muscle expression has been found10,11,34. Expression begins during the embryonic period, and in most cases has been found to peak in the early postnatal period, a time during which many synapses are being formed19,34. In situ hybridization using probes specific for individual V exons has found that they exhibit distinct but apparently overlapping expression patterns within the CNS. Most experiments have relied on side-by-side comparisons of individual expression patterns to reach this conclusion16,18,34, but limited studies employing double-label in situ hybridization with pairs of Pcdh-a19 or Pcdh-y10; J.A.W., unpublished data) probes have confirmed that neurons can express more than one gene from each cluster. An elegant and technically impressive recent study41 used single-cell RT-PCR of neurons and polymorphisms that can distinguish between two mouse strains to show that expression of Pcdh-a V exons is monoallelic but combinatorial. Single cerebellar Purkinje cells were found to express 2-4 Pcdh-a V exons; while the gene clusters of both alleles were active in a given neuron, transcripts of each individual V exon derived solely from one allele41. This differs from previously reported forms of monoallelic gene expression and, given the presence of nonsynonymous single-nucleotide polymorphisms in human and mouse Pcdh-aV exons33'42'43, suggests yet another mechanism that may increase the diversity of protocadherin interactions.

The hypothesis that such interactions might play a role in the establishment of synaptic connections is supported by studies using antibodies raised against the constant domains of a- or y-Pcdhs. Subcellular fractionation of brain lysates followed by Western blotting has consistently found that a-Pcdh19,44 and y-Pcdh10,44,45 proteins are concentrated in synaptic membrane and post-synaptic density (PSD) fractions. In addition, proteomic methods have identified y-Pcdhs as components of the "presynaptic particle web" that provides a scaffold for organizing neurotransmitter release46. Consistent with this biochemical analysis, immunostaining of brain sections with antibodies against the constant domains of a- or y-Pcdhs yields a punctate, neuropil-like pattern19,44. Our studies using a mouse line in which the gene for green fluorescent protein (GFP) is fused to the third Pcdh-Y constant exon10, found a similar pattern of fusion protein localization and further indicated that y-Pcdhs are highly concentrated in synapse-rich zones such as the retinal inner and outer plexiform layers. Within these zones, some y-Pcdh puncta overlap with puncta labeled by antibodies against presynaptic (SV2, synaptophysin) and postsynaptic (PSD-95, gephyrin) proteins, but there is clearly substantial extrasynaptic y-Pcdh as well10. This suggests that the y-Pcdhs may localize to particular types of synapses, which is consistent with immunostaining studies of cultured hippocampal neurons. In hippocampal cultures, the y-Pcdhs are highly expressed by inhibitory interneurons, but in an apparently nonsynaptic pattern, while the half of the spiny excitatory neurons that express y-Pcdhs localize them to only a portion of their synapses45.

These data have been confirmed and extended by Phillips et al.45, who performed immunogold electron microscopy with anti-y-Pcdh constant domain antibodies. This work clearly showed that while some hippocampal synapses contain y-Pcdh protein at both the pre- and postsynaptic membranes, much of the labeling was in "tubulovesicular" structures within axonal terminals and dendritic profiles45. Although the precise function of these structures is not yet clear, they may represent endosomal compartments that could mediate dynamic insertion of y-Pcdhs at synapses, perhaps in response to activity or during synaptic maturation. An a-Pcdh isoform (a4/CNR1) was also detected at the synapse by immunogold electron microscopy19; although a-Pcdhs are clearly located extrasynaptically as well, it is not yet known whether they are found in "tubulovesicular" structures as are the y-Pcdhs. In the chick ciliary ganglion, a-Pcdhs have been reported to be present in axons and perisynaptic sites, rather than directly at the synapse47. There does appear to be substantial co-localization of a- and y-Pcdhs in hippocampal neurons both in vitro and in vivo, and results of a recent study44 suggest that this is more than coincidental. In many cell lines transfected with a construct expressing a full-length a-Pcdh isoform, the exogenous protein failed to localize efficiently to the cell surface35. Murata et al.44 present evidence that efficient surface expression of a-Pcdhs requires a cis-interaction with y-Pcdhs. Antibodies against the a-Pcdhs can co-immunoprecipitate y-Pcdhs in brain lysates, and affinity-tag pull-down assays with truncated Pcdh proteins suggest that both the extracellular and cytoplasmic domains of a- and y-Pcdhs can interact44. Such interactions likely underlie the increased cell-surface delivery of an exogenous a-Pcdh isoform when HEK293T cells are co-transfected with constructs encoding y-Pcdh isoforms44. Whether interactions between a- and y-Pcdhs are important for their surface expression in neurons remains to be seen, but such a mechanism could contribute to the dynamic modulation of protocadherin-mediated adhesion at the synapse.

5. GENETIC ANALYSIS OF y-PROTOCADHERIN FUNCTION

Despite the great interest in the clustered protocadherins, we still know relatively little about their function. The multiplicity of Pcdh-a, -ft, and -Ygenes makes genetic redundancy likely, such that ablation of any single isoform by traditional gene knockout methods might not produce any detectable phenotype. In addition, because there are few antibodies against individual protocadherin isoforms available, we know almost nothing about the neuronal subsets that express each particular isoform and thus about the types of connections they make. This would make it difficult to know where to look for defects in any single Pcdh gene knockout animals, should such defects be subtle. Conversely, since multiple Pcdh genes are expressed by (probably) every neuron in the nervous system, forced over- or mis-expression of any single isoform might not have any observable effect on neuronal function. Our own work has avoided these problems by taking advantage of mouse mutants, produced in the course of gene expression studies31, in which the entire Pcdh-y gene cluster has been deleted or otherwise disrupted9,10. Analysis of these Pcdh- ^mutant mice has provided the first genetic evidence that at least some of the clustered protocadherins play critical roles in nervous system development and function.

The Pcdh-y deletion mutant was produced in mouse embryonic stem (ES) cells by sequential targeting to insert loxP DNA sequences both upstream of V exon yA1 and within C exon 310,31. Subsequent expression of Cre recombinase in the double-targeted ES cells led to excision of the entire ~300 kb Pcdh-ylocus. The resulting ES cells were used to generate chimeric mice, which were then bred to obtain germline transmission. Homozygous Pcdh-y deletion mice were born alive in the expected Mendelian ratio, but died within hours after birth. Mutant pups exhibited multiple signs of severe neurological defects: a hunched posture, repetitive limb tremor, lack of voluntary movements, and nearly nonexistent reflexes and withdrawal behaviors. No obvious defects were found in non-neural organs, and analysis of intramuscular nerves and neuromuscular junctions suggested that they had formed during embryonic development and were grossly normal. Within the CNS, however, severe neuronal defects were readily apparent10.

*The work described in this section and published in refs. (9) and (10) was performed while the author was a postdoctoral fellow in the laboratory of Dr. Joshua R. Sanes (Washington University School of Medicine; now at Harvard University), in collaboration with Dr. Xiaozhong Wang (Baylor College of Medicine; now at Northwestern University).

Figure 9.2. Genetically Dissociable Neurodegeneration and Synaptic Defects in Mice Lacking the Pcdh-Y Locus. (A-D) Sections through the spinal cords of P0 control, Pcdh-Y deletion homozygotes (del/del), and Pcdh- Ydeletion/Bax knockout double mutants (del/del; Bax -/-), stained for Nissl substance (A,B) to reveal cell bodies or with antibodies to neurofilaments (C,D) to label axonal fiber tracts. The reduction in spinal cord size due to neuron and fiber tract loss observed in Pcdh- Ydeletion mice (A,C) is rescued by additional deletion of Bax (B,D). Double mutants still, however, display severe reductions in both excitatory (PSD-95+) and inhibitory (gephyrin +) synaptic puncta in spinal cord sections (E). These reductions are not quite as severe as those in the Pcdh- Ydeletion mutants, in which the death of neurons exacerbates synapse loss, but are still very significant, approaching 50% (see ref. 9). Bar = 100 ^m in (A-D), 12 ^m in (E).

Figure 9.2. Genetically Dissociable Neurodegeneration and Synaptic Defects in Mice Lacking the Pcdh-Y Locus. (A-D) Sections through the spinal cords of P0 control, Pcdh-Y deletion homozygotes (del/del), and Pcdh- Ydeletion/Bax knockout double mutants (del/del; Bax -/-), stained for Nissl substance (A,B) to reveal cell bodies or with antibodies to neurofilaments (C,D) to label axonal fiber tracts. The reduction in spinal cord size due to neuron and fiber tract loss observed in Pcdh- Ydeletion mice (A,C) is rescued by additional deletion of Bax (B,D). Double mutants still, however, display severe reductions in both excitatory (PSD-95+) and inhibitory (gephyrin +) synaptic puncta in spinal cord sections (E). These reductions are not quite as severe as those in the Pcdh- Ydeletion mutants, in which the death of neurons exacerbates synapse loss, but are still very significant, approaching 50% (see ref. 9). Bar = 100 ^m in (A-D), 12 ^m in (E).

The spinal cords of Pcdh- Ydeletion homozygotes were greatly reduced in size and neuronal density compared to control littermates, despite the fact that the pups themselves were of normal stature (Figure 9.2A). Analyses of Pcdh-Y deletion mutants at multiple embryonic stages demonstrated that spinal cord neurons proliferated, migrated out of the ventricular zone, extended axons and dendrites and expressed cell-type specific markers normally in the mutant embryos. At late embryonic stages, however, mutant interneurons, particularly those in the intermediate gray and ventral horn, underwent apoptotic cell death, as detected by TUNEL staining, immunoreactivity for cleaved caspase-3, and the prevalence of pyknotic nuclei at both the light and electron microscopic levels10. Consistent with this, mutant spinal cords exhibited multiple signs of neurodegeneration: loss of neurofilament-stained fiber tracts in the lateral and ventral funiculi (Figure 9.2C), reactive gliosis, and macrophage invasion. Similar evidence of normal early development followed by neurodegeneration was found in the brain, with the affected regions being those that are most mature at birth, when the mutants died. Together, these data suggested a previously unhypothesized role for the y-Pcdhs in promoting the survival of particular neuronal populations.

So what about synapses? Electron microscopic analysis uncovered synaptic profiles in thin sections of neonatal mutant spinal cord that were grossly normal in appearance, with a well-defined synaptic cleft, docked vesicles, and a PSD10. This demonstrates that y-Pcdhs are not strictly required for all synapse formation per se. However, such synaptic profiles were much rarer in mutant spinal cord as compared to controls, and immunostaining for presynaptic (synaptophysin) and postsynaptic (PSD-95, gephyrin) markers confirmed a 60-70% reduction in synaptic density (Figure 9.2E). Because so many neurons had been lost by the time of birth, it remained possible that the paucity of synapses was entirely a secondary effect. That this was not the case was, however, suggested by the fact that when synapses were quantified as the number of PSD-95- or gephyrin-positive puncta per neuron, a reduction of similar magnitude was still observed10. Nevertheless, our initial analysis of Pcdh-Y deletion mutants, while demonstrating the critical importance of these genes for neural development, could not definitively identify a synaptic function.

This uncertainty has been resolved recently through the use of two novel genetic approaches9. First, mice harboring the Pcdh-y deletion mutation were crossed with those lacking the gene for Bax, a pro-apoptotic member of the Bcl-2 family in the absence of which much neuronal apoptosis is blocked48. As predicted, this results in a complete rescue of the neurodegenerative phenotype: Doublemutant spinal cords were normal in size, neuron number, and fiber tract density, and lacked any morphological or biochemical evidence of apoptosis (Figure 9.2B,D). However, Pcdh-Ydeletion/Bax knockout mice still died within hours of birth and exhibited neurological defects similar to, if less severe than, those of mice harboring the deletion allele alone. This is apparently due to the fact that, even in the absence of neurodegeneration, loss of Y-Pcdh function leads to severe reductions in synaptic density: PSD-95-positive puncta were reduced by over 35%, and gephyrin or glycine receptor-positive puncta were reduced by 50% in spinal cord sections (Figure 9.2E)9.

A possible caveat in these results is that neurons in the double mutants could be "living dead"; that is, unhealthy in some way, but artificially able to complete the apoptotic program due to the absence of Bax. A second approach using mice harboring a hypomorphic Pcdh-y allele has, however, obviated this concern. The Pcdh-y truncation allele encodes y-Pcdhs lacking the C-terminal 57 amino acids due to insertion of an internal ribosome entry site (IRES)-GFP-LacZ reporter cassette into the third C exon9. This reporter cassette proved to be nonfunctional in the resulting mice, but serendipitously results in a hypomorph that produces only very low levels of y-Pcdh proteins. These Pcdh-Y truncation mice display a dissociation of neurodegenerative and synaptic phenotypes similar to that observed in the Pcdh-Y deletion/Bax knockout double mutants. Their spinal cords appear normal in structure and neuronal density, and levels of apoptosis are low. Nevertheless, Pcdh-Y truncation mice die within hours of birth, exhibiting neurological defects due to a reduction of 40-50% in the density of both excitatory and inhibitory spinal cord synapses9. Unfortunately, the fact that the truncation mutants are likely hypomorphic for all Y-Pcdh proteins makes it impossible to conclude whether the observed synaptic defects are due to the C-terminal truncation that these mice harbor. These results do, however, suggest that the

Control Pcdh-ytr/tr

Defeat Drugs Death

Defeat Drugs Death

This Book Is One Of The Most Valuable Resources In The World When It Comes To Helpful Info On Avoiding And Beating A Fatal Drug Addiction!

Get My Free Ebook


Post a comment