CASK ortholog Lin-2 was discovered in cell-lineage screen in C. elegans nearly 36 years ago. Interest in CASK was revived by three simultaneously published articles in the year 1996 in three different animal models. One of the major interests surrounded CASK’s ability to interact with neurexin, Mint, and veli. These interactions suggested that CASK may be a synaptic scaffold. Lack of specific synaptic defect in different animal models, however, has made this hypothesis untenable. Characterization of the CASK knockout mice, discovery of CASK as a specialized protein kinase, and its association with secondary microcephaly and pontocerebellar hypoplasia have revitalized interest in this enigmatic signaling molecule. Besides synaptic scaffolding, numerous studies have pointed to cellular role of CASK in gene expression, cell proliferation, and intracellular trafficking. Despite availability of large amount of biochemical, structural, and functional data, the precise role of CASK still eludes us. It is likely that CASK may perform different function in different cell types. In this chapter, I describe what is currently known about CASK from different experimental systems.
CASK (calcium/calmodulin-dependent protein serine kinase) is an X-linked gene that codes a MAGUK (membrane-associated guanylate kinase) family protein. CASK gene evolved early in animal kingdom even before the origin of three distinct germ layers, and in mammals, it is expressed nearly ubiquitously in all tissues (Hata et al. 1996; Mukherjee et al. 2010). Most animals do not have sex chromosomes; CASK orthologs are therefore typically found in autosome in most animal classes. In humans, a pseudogene remnant of CASK (CASKP) is present in the X-degenerate sequence of the Y-chromosome, signifying the descent of X- and Y-chromosome from ancient autosomes. Although the size of the CASK cDNA is only ~2.7 kbp, human CASK gene is rather large (~408 kbp) and it is comprised of 27exons (LaConte and Mukherjee 2013). In its evolutionary history, CASK gene has undergone a massive expansion (~55 folds) mostly due to increase in the size of its introns. The vertebrate CASK orthologs are significantly larger than their invertebrate counterparts (LaConte and Mukherjee 2013). The increased intronic length may serve a variety of regulatory functions in gene expression, including but not restricted to tissue specific expression and regulation of splicing. For instance, human CASK gene is known to produce many more transcripts (~17) than those produced by some invertebrate CASK orthologs (~6). Besides tissue expression pattern and splicing, intron sequences of CASK may also serve as inducible exons. Intriguingly, mammalian CASK gene locus harbors two orphan rhodopsin-like GPCRs (G-protein-coupled receptors) within its introns (LaConte and Mukherjee 2013). Like many other genes, CASK may also have undergone a duplication event in vertebrates; however, instead of generating further isoforms of CASK, this event may have led to origin of MPP1 (membrane palmitoylated protein 1) gene. Thus, CASK has no known genetic isoforms in any animal.
CASK Protein Structure
CASK protein is composed of a CaMK (Calcium/calmodulin-dependent kinase) domain at its N-terminus followed by two L27 (lin-2, lin-7) domains, a PDZ (PSD-95, Dlg, ZO1) domain, a SH3 (src homology 3) domain, and GuK (guanylate kinase) domain (LaConte and Mukherjee 2013; Hsueh et al. 1998; LaConte et al. 2014). Of these domains, the CaMK domain is unique to CASK; the other domains being found in other members of MAGUK family. The domains and their arrangement in the CASK protein are completely conserved in all animals (Fig. 1). The structure of some of the domains of CASK has been solved at atomic resolution.
The CaMK domain of CASK has remained most intriguing due to its uniqueness among all known MAGUKs. This domain has a strong homology with CaMKII (calcium/calmodulin-dependent kinase II) and shares 44% identity with this protein (Mukherjee et al. 2008). In fact, in Drosophila upon its discovery, CASK ortholog was thought to be an isoform of Drosophila CaMK and termed as Caki (Martin and Ollo 1996). The crystal structure of CASK CaMK domain revealed that it adopts a canonical eukaryotic protein kinase bilobal structure with the N-terminal lobe dominated by five-stranded beta sheet structure and C-terminal lobe being predominantly helical (Mukherjee et al. 2008). The two lobes in eukaryotic protein kinases may acquire either an active “closed conformation” or an inactive “open conformation” relative to each other. Most protein kinases are kept in an inactive conformation, turning on only in presence of appropriate messenger. CASK CaMK domain maintains a constitutively active conformation and a high affinity for adenine nucleotides (Mukherjee et al. 2008). However, evolutionarily CASK CaMK domain have four critical substitutions in the residues lining the nucleotide-binding pocket compared to other CaMK domains that renders the ATP binding to CASK divalent ion sensitive (Mukherjee et al. 2010). Since protein kinases use magnesium, a divalent ion as a cofactor, the overall kinetics of CASK CaMK domain is very slow (Mukherjee et al. 2008). The kinase function of CASK CaMK domain is therefore not likely to be sustained by enzymic turnover as discussed later.
CASK has two L27 domains termed as L27-n (N-terminus) and L27-c (C-terminus). NMR and CD spectroscopy suggest by themselves L27 domains are largely unfolded and do not form homo-oligomers (Feng et al. 2004). L27 domains have been assumed to be critical for organizing scaffolding proteins via oligomerization. Two different NMR structure have been solved for the L27 domains of CASK: (1) L27-n complexed with L27 domain of SAP97 and (2) L27-c domain complexed with L27 domain of lin-7 (veli) (Feng et al. 2004, 2005). Both structures show a tetramer composed of two heterodimers. Each L27 domain of CASK in these complexes is composed of three α-helices. The first two helices of each L27 domains in a heterodimer are packed in a four-helical bundle. The third helix of each L27 domain forms an additional four-helical bundle to form a dimer of the heterodimers (Feng et al. 2004).
The central region of CASK has a PDZ domain. The PDZ domain has also been referred to as DHR (dlg homologous region) or GLGF (glycine leucine glycine phenylalanine) motif and is made of 80–90 residues and thought to be important for signal transduction. PDZ domains have been classified into two basic varieties; the CASK PDZ belongs to class II of PDZ domains. Despite the low sequence similarities (~20%) between the two classes, the overall molecular structures of PDZ domains are pretty similar. CASK PDZ domain consists of a core of six β strands and two α-helices (Daniels et al. 1998). The significant differences from the Class I PDZ domains can be found in the residues lining surface for ligand binding in CASK PDZ domain. While Class I PDZ domain harbors a hydrophobic pocket for position 0 of the ligand and a hydrophilic pocket for position −2 of the ligand, CASK PDZ domain harbors hydrophobic pockets for bulky residues for both 0 and −2 positions of ligands (Daniels et al. 1998).
The C-terminus of CASK is composed of a SH3 domain and GuK domain with a Hook motif nestled in between. Although crystal structure of CASK SH3 domain is not available, SH3-GuK domain of another MAGUK protein PSD-95 (postsynaptic density 95) has been solved by two independent laboratories (Tavares et al. 2001; McGee et al. 2001). Both the structures indicate strong intramolecular interaction between the two domains. This interaction is maintained primarily via formation of an antiparallel double-stranded β-sheet, where one beta sheet is contributed by C-terminus of the Hook motif and the second beta sheet is contributed by 12 residues C-terminus to the GuK domain. Since SH3 domain themselves are beta-barrel folds comprised of 5–6 beta strands, it has been argued that in MAGUK proteins SH3 domain is formed by noncontiguous primary sequence interspersed by Hook motif and GuK domain (McGee et al. 2001). SH3-GuK domain may thus form an integrated unit. The isolated GuK domain of CASK itself has been crystallized and its structure solved at an atomic resolution (Li et al. 2002). The structure of CASK GuK domain bears similarities with yeast guanylate kinase and can be divided essentially into three parts. The core comprising of five parallel beta strands and four alpha helices, the GMP-binding region comprising of four strands of beta sheet and a single alpha helix, and the LID region composed of a long loop and alpha helix. The LID region of CASK GuK domain has a structure distinct from that of yeast guanylate kinase (Li et al. 2002).
Despite having such structural information, it is difficult to speculate what conformation the full-length CASK adopts and what would be the relation between the various domains. This problem is further amplified by the fact that under physiological conditions CASK is unlikely to be present in an uncomplexed, monodispersed form (LaConte et al. 2016), interaction with a variety of binding partners and small molecules as well as posttranslational modifications are likely to affect the overall fold of CASK. Therefore, although the structural information provides a basic framework to probe molecular and cellular function of CASK, they might not inform us adequately on the physiological function of CASK. Modeling experiments using primary sequences of various CASK ortholog suggest that the structures described above may have been optimized early in animal evolution, indicating an overall conservation of molecular function (LaConte and Mukherjee 2013).
CASK Interacting Partners
Besides these interactions, some novel interactions have also been documented in invertebrate animal models, notably Let23 interaction with lin-2 in C. elegans and CaMKII interaction with dCASK in Drosophila (Lu et al. 2003; Hoskins et al. 1996). Apart from the interactions published in various previous works, using high-throughput proteomic experiments in Drosophila, we also found CASK may interact with many metabolic proteins (Mukherjee et al. 2014). Thus the molecular interactions of CASK suggest that it may have a very complex function. Similarly, CASK has been suggested to be localized in presynapse, postsynapse, nucleus (Hsueh et al. 2000), cytoplasm (LaConte et al. 2016; Srivastava et al. 2016), and mitochondria (Mukherjee et al. 2014). Adding to the complexity, CASK is expressed by neurons as well as astrocytes and oligodendrocytes. CASK has been also shown to have many posttranslational modifications; CASK may be substrate of kinases and known to get phosphorylated (Samuels et al. 2007); CASK is also known to be sumoylated, ubiquitinated, and acetylated (Mukherjee et al. 2014).
Putative Molecular Functions of CASK
Ubiquitous expression of CASK, with different subcellular localization, multiple protein interaction, and posttranslational modifications all suggest a rather complex function of CASK. Indeed, CASK is thought to play a variety of cellular function which involves epithelial patterning, cell proliferation, cell polarization, wound healing and gap junction function, insulin signaling, insulin secretion, sperm motility, gene expression, and most importantly neuronal synapse formation (Butz et al. 1998). Ever since the description of CASK interaction with neurexin and syndecan, most studies have examined the role of CASK in synapse formation and maintenance. Role of CASK has been examined in context of both pre- and postsynapse. CASK may play a role in synapse formation which is regulated by CDK5 phosphorylation (Samuels et al. 2007), in neurotransmitter release, in coupling action potential to release (Slawson et al. 2014), in axonal branching, in protein and synaptic vesicle trafficking, in calcium channel targeting to presynapse. At postsynaptic side, CASK may help to link SAP97 to AMPA and NMDA receptor, NMDA receptor trafficking, syndecan-mediated signaling (Hu et al. 2016), activity-dependent gating of CaMKII (Lu et al. 2003), and organization of neuromuscular junction together with dlg. Despite the long literature on molecular function of CASK, its cellular role till date is largely unknown. This is especially so in context of examination of genetic models. Lin-2, the CASK ortholog in C. elegans was one of the first genes to be described in context of a cell-lineage screen (and hence lin-2) (Horvitz and Sulston 1980). Loss of function of lin-2 produced vulval phenotypes (multiple vulva and vulvaless). However, there were no neuronal or synaptic phenotypes in this organism (Hoskins et al. 1996). In Drosophila, CASK is expressed to the highest level in the eye and nervous system. Initial experiments were done on a transheterozygote model where neighboring genes are also affected; however, analysis suggested that absence of CASK may affect locomotion in Drosophila (Martin and Ollo 1996). This result was later confirmed by specific deletion of CASK β. Despite reduction in locomotion, no major developmental defect was observed in the fly model which also exhibited a normal olfactory habituation (Slawson et al. 2011). Strikingly, the locomotor phenotype could not be rescued by expressing CASK in the motor neurons in these flies clearly indicating that the observed phenotype do not arise due to a specific synaptic defect at the neuromuscular junction (Slawson et al. 2011). Finally, unlike the invertebrate models described above, deletion of CASK is not compatible with life in mouse (Atasoy et al. 2007). This is a fairly unusual phenomenon, typically due to more redundancy in function, often genes that are essential in invertebrate turnout to be nonessential in vertebrates. It is not clear whether this departure from norm represent acquisition of new function by CASK gene or differences in biology between lower organisms and mammals. Although deletion of CASK in mice led to neonatal lethality (died within hours of birth), neuronal function remained fairly normal. No change was observed in neuronal excitability, calcium-dependent presynaptic release, or postsynaptic receptor function (Atasoy et al. 2007). The only impairment was observed in spontaneous release where the glutamatergic miniature frequency was increased whereas gabaergic miniature frequency was found to be reduced. From developmental standpoint, the brains of CASK knockout mice were indistinguishable from wild type at birth and were appropriately laminated. The only developmental defect observed was that of a cleft palate which appeared in nearly 80% of CASK knockout mice (Atasoy et al. 2007). Since the mice without apparent cleft palate also died within hours of birth, cleft palate was not a cause of death per se. Surprisingly, the apoptotic rates within the brain of CASK knockout mice were radically increased, approximately threefold (Atasoy et al. 2007). No change was found in polarity of epithelial cells when CASK was deleted. Thus each of the animal models for CASK deletion experiment provided very different phenotype which makes it very difficult to identify CASK’s cellular role. This is further complicated by the fact that deletion of CASK interacting molecules in mouse does not phenocopy CASK null mice. Till date, many such gene-modified mice have been produced which includes Tbr1 knockout mice, Mint1 knockout mice, veli knockout mice, and CINAP knockout mice.
CASK Gene-Associated Pathologies
Role of CASK has also been documented in various cancers; expression level of CASK has been shown to be negatively related to prognosis in colorectal and gastric cancer. However, till date no study directly relates mutations in CASK with cancer. In humans, CASK gene was first identified as an optic nerve atrophy candidate. Many studies since then have identified CASK as an X-linked intellectual disability gene. Since CASK is essential for survival in mammals, deletion or truncation mutations of CASK gene are typically found in females and results in a condition termed as mental retardation and microcephaly with pontine and cerebellar hypoplasia (MICPCH) (Najm et al. 2008; Burglen et al. 2012; Moog et al. 2011). Severe mutations in some cases have been associated with epileptic encephalopathies like Ohtahara syndrome or West syndrome and have a very poor outcome. Missense mutations in CASK gene have been documented among boys with variable intellectual disability, nystagmus, and autistic traits (Hackett et al. 2010). Furthermore, MICPCH has also been documented in males with mosaicism in gene mutation due to somatic mutagenesis during early embryogenesis (Burglen et al. 2012). Besides microcephaly and pontocerebellar hypoplasia, CASK mutated patients exhibit several ocular disorders like optic nerve hypoplasia, optic nerve atrophy, and glaucoma. They also display hypotonia with scoliosis, growth retardation, and seizures (Burglen et al. 2012; Moog et al. 2011). Are the phenotypes seen in human due to loss of CASK function? Our examination of different genetic manipulation of CASK gene in rodents confirms that the CASK-associated pathologies are indeed loss of CASK function phenotype. First of all, CASK heterozygosity in female mice produced many of the phenotypes seen in MICPCH including secondary (postnatal) microcephaly, disproportionate cerebellar hypoplasia, optic nerve hypoplasia, scoliosis, and locomotor abnormalities (Srivastava et al. 2016). Although, seizure were not readily observed in the heterozygous knockout mice, CASK neuronal knockout mice displayed severe seizures to which they succumb by 21 days (Srivastava et al. 2016), indicating that most of the phenotypes described in humans are due to CASK loss of function.
Upon analysis of CASK heterozygous mice two things came as a surprise: (1) defects in metabolism were detected in these mice and (2) the postnatal microcephaly was not a consequence of cell-autonomous function of CASK. Furthermore, all the phenotypes seen in human subjects do not seem to arise from loss of neuronal CASK alone (Srivastava et al. 2016). It is not yet clear whether the metabolic defects are directly related to interaction of CASK with metabolic proteins. However, a non-cell-autonomous function may imply involvement of extracellular factors. At present, the precise mechanism by which loss of CASK disrupts postnatal brain growth remains unknown. Furthermore, the precise mechanism of optic neuropathies and cerebellar hypoplasia also is not clear.
Despite many biochemical, structural, and animal model studies, the function of CASK remains unclear. The clinical importance of this gene emphasizes the necessity to understand the molecular function of CASK. The nonessential function of CASK in invertebrates combined with the non-cell-autonomous function in mammals argues that fundamentally CASK may not be required for cellular survival. The phenotypes observed in humans may therefore be more systemic in nature. It is possible that the differences in biology of various animal models itself leads to different phenotypes observed upon CASK deletion. Alternatively, it is also possible that CASK gained new functions in mammalian brain. More detailed analysis of this gene will be required to dissect its role in development of brain and individual.
Konark Mukherjee is supported by grant from National Eye Institute: R01EY024712-03.
- Burglen L, Chantot-Bastaraud S, Garel C, Milh M, Touraine R, Zanni G, et al. Spectrum of pontocerebellar hypoplasia in 13 girls and boys with CASK mutations: confirmation of a recognizable phenotype and first description of a male mosaic patient. Orphanet J Rare Dis. 2012;7:18.CrossRefPubMedPubMedCentralGoogle Scholar
- Hu HT, Umemori H, Hsueh YP. Postsynaptic SDC2 induces transsynaptic signaling via FGF22 for bidirectional synaptic formation. Sci Rep 2016;6: 33592.Google Scholar
- Slawson JB, Kuklin EA, Mukherjee K, Pirez N, Donelson NC, Griffith LC. Regulation of dopamine release by CASK-beta modulates locomotor initiation in Drosophila melanogaster. Front Behav Neurosci 2014;8: 394.Google Scholar