Intellectual disability: dendritic anomalies and emerging genetic perspectives

Abstract

Intellectual disability (ID) corresponds to several neurodevelopmental disorders of heterogeneous origin in which cognitive deficits are commonly associated with abnormalities of dendrites and dendritic spines. These histological changes in the brain serve as a proxy for underlying deficits in neuronal network connectivity, mostly a result of genetic factors. Historically, chromosomal abnormalities have been reported by conventional karyotyping, targeted fluorescence in situ hybridization (FISH), and chromosomal microarray analysis. More recently, cytogenomic mapping, whole-exome sequencing, and bioinformatic mining has led to the identification of novel candidate genes, including genes involved in neuritogenesis, dendrite maintenance, and synaptic plasticity. Greater understanding of the roles of these putative ID genes and their functional interactions might boost investigations into determining the plausible link between cellular and behavioral alterations as well as the mechanisms contributing to the cognitive impairment observed in ID. Genetic data combined with histological abnormalities, clinical presentation, and transgenic animal models provide support for the primacy of dysregulation in dendrite structure and function as the basis for the cognitive deficits observed in ID. In this review, we highlight the importance of dendrite pathophysiology in the etiologies of four prototypical ID syndromes, namely Down Syndrome (DS), Rett Syndrome (RTT), Digeorge Syndrome (DGS) and Fragile X Syndrome (FXS). Clinical characteristics of ID have also been reported in individuals with deletions in the long arm of chromosome 10 (the q26.2/q26.3), a region containing the gene for the collapsin response mediator protein 3 (CRMP3), also known as dihydropyrimidinase-related protein-4 (DRP-4, DPYSL4), which is involved in dendritogenesis. Following a discussion of clinical and genetic findings in these syndromes and their preclinical animal models, we lionize CRMP3/DPYSL4 as a novel candidate gene for ID that may be ripe for therapeutic intervention.

Etiology of intellectual disability (ID)

Intellectual Disability is defined in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), by three primary criteria: 1) onset before age 18; 2) an intelligence quotient (IQ) below 70; 3) impairment of general mental ability that impacts adaptive functioning in three domains including conceptual, social, and practical skills. Neurodevelopmental abnormalities with cognitive and communication deficits are widely evident in children diagnosed with ID. Roughly two-thirds of ID patients present with a mild-to-moderate level of impairment and one-third develop a severe-to-profound level of impairment. In addition, these patients are at risk for developing various comorbidities, such as sensory impairments, convulsive disorders, behavioral disturbances, and psychiatric illnesses. Development of the human brain results from highly complex and coordinated activity involving genetic and environmental processes. This varied network of interactions suggests that the underlying causative factors leading to ID are heterogenous in nature. Numerous studies have provided evidence that ID is not a single disease, but rather a common hallmark of a collection of heterogeneous disorders [96]. The pathogenesis of these developmental disorders can best be viewed through the lens of altered intracellular signaling cascades, which arise from environmental stimuli or genetic factors, and culminate in the development of a pathological phenotype. The most common environmental factors leading to ID include exposure to neurotoxic compounds (alcohol, lead, mercury, drugs), malnutrition during pregnancy, and peri/post-natal asphyxia. These factors play a major role in contributing to the prevalence of mild forms of ID [44, 101]. In contrast to environmental factors, genetic causes can include a single genetic defect, such as deletion or mutation of a single gene (SYN1, POLR3B, SLC5A2; SYNGAP1, DYRK1A, SCN2). In addition, chromosomal deletion, aneusomies, inversion, and structural abnormalities may provoke more severe forms of ID [23, 72]. Commonly occurring forms include, but are not limited to, DS, RTT, DGS, and FXS. Laboratory findings and clinical data suggest that aberrant activity of metabolic, transcriptional, and signaling pathways are the most common molecular factors involved in the development of the various forms of ID [16, 224]. Neuropathologically, alterations in dendritic tree morphology have been reported in human neurological and psychiatric diseases and are particularly prevalent in those disorders associated with ID [51, 137].

Dendrite physiology: from specification to dysgenesis in ID

Neurons are characterized by a highly polarized cellular morphology, composed of an axon and dendrites, which are critical for the formation of neuronal circuits. These processes are the essential structural elements that allow information to be transmitted and processed in the nervous system. Dendrites are highly branched structures with characteristic arborization and morphology that can be distinguished after 5 days in a morphogenetic model of cultured hippocampal neurons (stage 4)[55]. The specific pattern of a dendritic tree can often be associated with a neuronal cell type, such as Purkinje or cortical neurons, and their physical structure reflects the complexity of their computational task. During development, initial wiring of the nervous system depends on precise refinement and remodeling of dendritic arbors through extension, addition, elongation, retraction, and pruning of individual dendrites [122]. These processes are highly dynamic, and dendritic structure is intimately influenced by local synaptic activity. Additionally, dendrites are crucial for the spatial and temporal integration of upstream information within neuronal circuits. The appearance of dendritic retraction, together with synaptic disassembly prior to soma/axon loss, correlates with the development of substantial functional deficits and is associated with early signs of pathology in a number of neurodegenerative disorders [58, 114].Various intrinsic signaling proteins mediate temporal and spatial control of dendritic arborization, which ensures dendritic growth is correctly generated and maintained over a lifetime [127, 166]. The roles of these molecules are functionally intertwined and highly regulated, which allows for tight control over dendritic extension and arborization. The plethora of molecular regulators of dendritic structure include neuromodulators and neurotransmitters, such as 5-hydoxytryptamine (5-HT) and gamma aminobutyric acid (GABA), as well as cell surface receptors, including neuropilin/plexin, Notch, Trks, and EpH receptors [75, 178, 205]. Ligands for these receptors, like the neurotrophic factors ciliary neurotrophic factor (CNTF), insulin-like growth factor (IGF), nerve growth factor (NGF), and brain-derived neurotrophic factor (BDNF) [29, 169], or the semaphorins, are fundamental to regulation of dendritic structure. Regulators also include cell adhesion molecules L1CAM, Cadherin-catenin adhesion complex, SAX7, and MNR-1[205]. Signaling molecules, such as CRMPs [172], Calmodulin Kinase 2b (CaMKIIb), and CRP1, as well as calcium signaling proteins and transcription factors NF-kB, CuX1-2, Neurogenin-2, CREB, and CREST [1, 37, 52] have critical roles. Cytoskeletal proteins are the final effectors supporting the morphology of dendrites, and synaptic scaffolding proteins, PSD95 and Cypin [2, 135, 157], as well as microtubule and tubulin polymerization regulators, APC, EB3, XMAP215, TACC3, and MAP1, are involved in regulation of dendritic structure.

Most dendrites have spines that can be classified into four morphological categories including, stubby, thin, filopodial, and mushroom. These structures serve as the main postsynaptic contact sites of glutamatergic excitatory synapses in the CNS [32]. Cytoskeletal elements within dendritic spines are primarily composed of cortactin, ARP2/3 complex, PSD 95/93, cofilin, actin filaments, and the regulatory Rho GTPases Rac, Rho, and Cdc42 [123]. These postsynaptic dendritic protrusions are important regulators of physiological Ca²⁺ gradients and serve to amplify local levels of calcium, which mediate activation of downstream calcium-dependent signaling pathways that are essential for synaptic plasticity [158]. Despite the small size of spines, a multitude of molecules have been implicated in their development and remodeling.

It is therefore tempting to assume that various etiological changes, such as environmental and genetic factors, will impact dendrite and spine structure and influence cognitive function [26, 153]. This is particularly true when considering neuronal mechanisms underlying ID, where defects in neuronal structure, network connectivity, and plasticity are critical features associated with common clinical presentations. In line with this concept of structural abnormality, alteration of the dendritic arbor is a consistent feature exhibited by patients with ID across multiple clinical presentations, as assessed on post-mortem neural tissues, with significant alterations of several important CNS areas, including the hippocampus, dentate gyrus, and cerebral cortex [180, 208]. Golgi staining of pyramidal neurons in the hippocampus and cerebral cortex in patients with DS, RTT, DGS and FXS, revealed a reduction in dendritic branching, arborization, and abnormalities in the morphology and density of dendritic spines compared to normal controls [22, 43]. This finding is highly suggestive of altered neuronal connectivity in these syndromes. Pathological spine loss can be caused by altered expression of various signaling proteins, such as Rho-GEF (KALRN), Rho-GAP (OPHN1), Cdc42-GEF (ARHGEF9), and Rac-GAP (OCRL1) [39, 161]. Moreover, SAP102 (DLG3; XLID) and PSD95 (DLG4) are associated with anchoring NMDA receptors as well as other major PSD proteins, such as HOMER, CaMKII, SynGAP, DLG-associated protein (DLGAP/GKAP) and SHANKs. Remarkably, a high number of de novo mutations and genomic aberrations have been identified in SHANK3, SHANK2, SYNGAP1, and GKAP in patients with ID, suggesting that alterations in glutamatergic neurons contribute, at least in part, to the cognitive impairment. Indeed in ID, dendritic abnormalities appear to correlate, to some extent, with the cognitive impairment profile of the patient [141].

Phenotypic features of patients with prototypic ID syndromes

ID of genetic origin, including DS, RTT, DGS and FXS, present with specific clinical and histopathological features and, although abnormalities can extend to many organs, we limited this review to those involving the nervous system.

DS.

A hallmark observation in neonatal DS patients is the presence of hypotonia, which is responsible for delayed motor development. Functionally, the IQ in these patients decreases over the course of the first decade of life and reaches a plateau that continues into adulthood with deficits in working memory/spatial skills, and reading/language [9]. Studies have reported microcephaly, reduced hippocampal and cerebral cortex size, and reduced cortical thickness. Neuropathological findings include abnormal neuronal differentiation, abnormal cortical lamination, decreased dendritic branching, reduced spine density, and altered dendritic arborization and spine morphology [192]. Aberrant electrophysiological properties of CNS neurons have also been reported. Patients also develop neuropathology closely resembling that of Alzheimer’s disease (AD) by the age of 50-60 years [214].

RTT.

RTT is almost exclusively found in female patients, with signs of the disorder often starting to emerge only after 6–18 months of life [40, 71], although they can appear ealier in the congenital variant. The onset of mental decline coincides with the emergence of motor abnormalities characterized by apraxia and often an ataxic gait that may develop into a Parkinsonian phenotype [94]. Patients exhibit partial or complete loss of spoken language, as well as a lack of hand movements. Many children with RTT display reduced brain volume, which is consistent with a smaller head circumference. This phenotype arises from the relatively small somatic volume of neurons associated with dense packing of these smaller cells in the cerebral cortex, cerebellum, and hippocampus, and reduced dendritic arborization. Neuropathologically, reduced dendritic morphological complexity is present throughout life. Apical dendrites from neurons of the hippocampus, prefrontal, motor, and middle temporal cortical areas are often asymmetric in structure and significantly reduced in number with some segments completely devoid of spines [11, 152]. In addition, phosphorylation of MAPs, which acts to stabilize microtubules, is decreased in postmortem tissues from these patients.

DGS.

DGS is a frequent cause of ID, with an estimated prevalence of approximately 1/3000 – 1/4000 live births [139]. Patients often exhibit complex and variable phenotypes, including developmental delay, hearing loss, as well as anomalies of the eye, craniofacial structures, and musculoskeletal systems. Approximately one-third of patients with DGS display mild to moderate ID, often associated with psychiatric disorders, such as schizophrenia and attention-deficit/hyperactivity disorder (ADHD) [206]. A global reduction in brain volume is observed in these patients, with widespread reductions in frontal, temporal, parietal, and occipital lobes, as well as atrophy of the cerebellum and hippocampus [92]. Post-mortem studies have revealed abnormalities in neuronal migration, organization, and connectivity [184]. Functional MRI reported patterns of hypoconnectivity in large-scale neuronal networks, which are necessary for higher order cognitive functions [110].

FXS.

FXS patients displaying an ID phenotype are found at a rate of approximately 1:7000 in males and 1:11000 in females [143]. They experience learning deficits, delayed language acquisition, impaired motor skills, and repetitive behavior. In addition, they exhibit a range of psychiatric symptoms including anxiety, poor social interaction, hyper-arousal/sensitivity, disrupted sleep, ADHD, and autism spectrum symptoms [130]. Male FXS patients demonstrate severe disability, while female patients are often less affected due to mosaicism resulting from X-chromosome inactivation, which randomly occurs early in embryogenesis [116]. During adulthood, FXS patients may be at increased risk of parkinsonism, and in some cases, dementia [187, 203]. These patients display impaired neurite extension with global failure of normal spine maturation and pruning during development that persists throughout adulthood.

Genetic basis of specific ID syndromes

Next-generation sequencing (NGS), using either whole exome or genome panels, as well as chromosomal microarray analysis (CMA), have emerged as powerful new tools for identification of genomic abnormalities associated with a wide range of ID syndromes. Genomic assays include single nucleotide polymorphism (SNP) arrays and comparative genomic hybridization (CGH) arrays, both of which are useful for detection of genomic copy number variants (CNV), such as microdeletions and microduplications. Together with karyotyping, these techniques have revealed that the frequency of disease-causing CNVs is highest (20%-25%) in children with moderate to severe ID and is accompanied by malformations/dysmorphic features and specific genetic mutations.

DS.

The phenotypic variability present in patients with this syndrome is due to the presence of three copies of functional genes on HSA21 loci. HSA21 is the smallest human chromosome with a density of 15 genes per megabase (Mb) and remains among the poorest chromosomes in terms of functional DNA elements per Mb. However, HSA21 shows significant enrichment for encoded proteins involved in the regulation and development of cytoskeletal structures [8]. These cytoskeletal proteins are known to be involved in neuronal development and neurological disorders, particularly in Alzheimer’s neuropathology. At least five genes located on chromosome 21 may affect neuronal development, including SIM2, DYRK1A, GART, PCP4, DSCAM [198]. The transmembrane protein DSCAM contains an intracellular domain that interacts with a number of known kinases, including PAK1, FAK, FYN, JNK, and p38 [125], as well as PSD95, and the MAGI scaffolding proteins. Importantly, overexpression of DSCAM can contribute to the intellectual disability phenotype of DS patients [182]. In this context, the reduced dendritic complexity observed in cortical and hippocampal neurons of DS patients is consistent with the reduced dendritic arborization observed following DSCAM overexpression in mouse hippocampal neurons [5, 64]. Alternatively, data from studies comparing post-mortem transcriptomes obtained from 58 ID patients with euploid controls showed that some of the dysregulated protein coding genes do not map to HSA21. Interestingly, amyloid precursor protein (APP) is encoded on chromosome 21 at location 21q21.3, and trisomy of this protein may explain the increased frequency of dementia in DS patients.

RTT.

Three genes, MECP2, CDKL5 and FOXG1, have been reported to cause RTT. Interestingly, expression of the well-known transcriptional regulator MeCP2 (Methyl-CpG-binding protein 2) correlates with neuronal differentiation and postnatal development of the CNS, which suggests a role in neuronal maintenance and activity [61, 71, 89]. MeCP2 has two known isoforms, MeCP2-e1 and MeCP2-e2, which result from differential splicing of exons 1 and 2, respectively. The existence of these isoforms contributes to the diverse functions associated with MECP2 expression, and mutations in exon 1 have been observed in RTT [3, 42, 197]. However, in a majority of cases, patients diagnosed with typical RTT have mutations in MECP2-e1/e2, which could disrupt synaptic proliferation and pruning during the restricted developmental window that coincides with peak synapse formation in the cerebral cortex between 7–18 months of age. RTT with early epilepsy, and the congenital variant of RTT, are mainly due to variations in the CDKL5 and FOXG1 genes [24, 103, 156], respectively. Additionally, because the coding region for MeCP2 is located on the X-chromosome, mutations in MECP2 contribute to the characteristic features of RTT. Since males only have one copy of the X-chromosome, mutations in males are generally incompatible with life and as such, RTT affects mainly females, although it has been reported that, in rare cases, male patients can survive until adulthood. Recently, integrative analysis of omics data has identified new genes and new signaling pathways that could be affected in RTT. Specifically, Sp1 and REST, known cis-regulatory elements for MeCP2, in addition to SIN3A and SMC3, which are parts of the MeCP2-HDAC protein complex [126], have been identified using this strategy [6, 20, 115], Other functions attributed to MeCP2 include histone deacetylase-independent repression, maintenance of DNA methylation, histone methylation, assembly of secondary chromatin structure independently of methyl CpG DNA binding capability, and interactions with RNA splicing factors. Furthermore, the expression of MeF2C, which is also known to cause ID and psychomotor impairment, is inhibited by MeCP2 together with the HDAC complex [126].

DGS.

The 22q11.2 microdeletion associated with DGS is one of the most common genetic microdeletion syndromes in humans. However, the relative importance of individual genes on chromosome 22 that contribute to cognitive and behavioral problems in patients with DGS has yet to be elucidated. Most of the coding region of chromosome 22 (~90%) is 3 Mb and contains approximately 60 genes, while the 1.5 Mb deletion (~8%) contains 28 genes critical for DGS [56, 149]. The decrease in gene dosage resulting from the deletion disrupts mitochondrial activity and limits neurite growth. Recent evidence suggests that ZDDHC8, DGCR8, and DGCT8 can compromise neuronal circuits in the hippocampus [138]. RANPB1, TRMT2A, HIRA, CDC45L, and UFD1L are additional genetic candidates that potentially contribute to DGS by altering cortical development. In addition, TBX1 is implicated in several clinical features including the formation of the craniofacial region, thymus, parathyroid gland, and the aortic arch, which are organs and structures affected in the 22q11.2 deletion. Furthermore, in the hippocampus, RTNR4 influences dendritic spine morphology and limits activity-dependent synaptic strengthening [30]. Finally, COMT, PRODH, and PIK4CA, have been investigated in the context of schizophrenia as the 22q11.2 deletion was the CNV the most significantly associated with the disease. DGS is one of the rare examples in humans of a cytogenetic abnormality occurring in conjunction with a psychiatric disease. Phenotypic pathology in a subset of DGS patients has also been associated with mutations on the non-deleted allele, leading to unmasking of additional conditions [140]. This suggests that variability in the deletion size and breakpoint locations may play an important role in various DGS phenotypes.

FXS.

Approximately 1100 genes have been annotated on the X chromosome, of which at least 800 are protein-coding, and approximately 696 are expressed in the brain. There are two clinical presentations associated with FXS. The first is a neurodevelopmental disorder that manifests following full mutation of the FMR1 gene, which is located on the q27.3 locus of the X chromosome in FXS patients. The second, known as Fragile X-associated tremor/ataxia syndrome (FXTAS), is a neurodegenerative disease experienced during aging of premutation carriers with the FMR1 gene. The molecular mechanism underlying silencing of the FMR1 gene arises from expansions in the length of the FMR1-5’untranslated region (UTR). Traditional FXS patients carry a full-length mutation, which is composed of >200 CGG repeats [83, 143]. FXS and FXTAS patients express differing levels of CGG repeats, which alter the translation of the FMR protein (FMRP), with deficiency observed in FXS and increases found in FXTAS. FMRP is an RNA-binding protein that controls the translation of a large array of mRNAs, which are critical for dendritic arborization and spine structure [54, 165]. In addition, FMRP modulates local protein synthesis in dendrites and optimizes the activity of several signaling cascades through interactions with Rac1, Arc, PSD95, MAP1B, CaMKII, calbindin, and cadherins, which are essential for spine morphogenesis [14, 77]. Furthermore, FMRP induces age-dependent impairments in synaptic plasticity, specifically in LTP function.

Evidence suggests that loss of function (LOF) associated with a subset of genes may give rise to a large variety of brain disorder phenotypes. For example, ID and schizophrenic patients, which also display dendritic abnormalities, have overlapping LOF mutations in POGZ, SCN2A and SYNGAP1 genes. Additionally, ITSN, which encodes intersectin -a specific guanine nucleotide exchange factor for Cdc42- has been implicated in both AD and DS. Mutations of these genes produce the dystrophic dendritic spine phenotype present in various forms of ID. This suggests that different categories of proteins that are altered in ID may operate in the same molecular and cellular processes, which points toward the next challenge of identifying a common molecular basis among various ID syndromes.

Animal models of ID

The interpretation of genetic causality associated with ID is not straightforward, as patients may have comorbid and unknown genomic abnormalities. Extensive efforts have been undertaken to generate mouse models with the aim of unravelling the underlying physiopathology of ID. Genetically null mouse models cannot completely recapitulate the complexity of human disease [18] and there are examples of non-concordant, and even paradoxical, relationships between human and murine findings. However, animal models frequently allow the identification of a single genetic target, such as through rescue of phenotypic abnormalities by experimental gene transfer, which can contribute to deciphering the detailed genetic processes underlying disease pathogenesis [65, 66].

DS mouse models.

Because DS is a contiguous genetic syndrome that spans the entire 35 Mb of the long arm of HSA21 [215], there is no simple or complete trisomy mouse model of DS. In the past, several DS mouse models (Ts65Dn, Ts1Cje, Tc(Hsa21)1TybEmcf, Dp(10/16/17)1Yey) [46, 87] have been generated, with each displaying trisomy of a different subset of HSA21 genes or their orthologs. Trisomy of multiple genes, including Dyrk1A, Synj1, and Sim2 have been found to cause learning and memory defects in Ts65Dn mice, which suggests the possibility that overexpression of these genes may be associated with learning disability in patients with DS [10, 175]. In addition, several Cre/loxP and transchromosomic transgenic mice bearing the entire, or a sub-segment of the HSA21 region is sufficient to cause a DS phenotype, including altered dendritic structure, abnormal LTP, and cognitive deficits [199, 217]. Importantly, overexpression of specific genes like Dyrk1a revealed that DYRK1A is implicated in changes to GABAergic- and glutamatergic-related protein expression. Changes in the expression profiles of these proteins are important because they are critically involved in cognitive function [68, 76, 175]. A meta-analysis of data obtained from proteomic and transcriptomic approaches identified several signaling pathways which are perturbed in DS mouse models [79] and DS patients [59, 154]: specifically, a subset of genes on HSA21 was shown to be altered (SOD1, APP, DONSON, TIAM1, COL6A2, ITSN1 and BACE2), as well as the BDNF-dependent pathway, which is involved in neuronal growth, differentiation, and survival. Finally, expression of CRMP3, which regulates dendrites, is decreased in the hippocampus of 141G6 and Ts65Dn transgenic DS mouse models [204].

RTT mouse models.

In mice, knock-down of the X-linked Mecp2 gene recapitulates many hallmark symptoms of RTT, such as decreased numbers and regional loss of dendritic spines in the frontal cortex. Also, loss of Mecp2 in Mecp2^tm1.1Bird [80, 133] and Mecp2^tm1.1Jae [28] transgenic mice results in abnormal dendritic spine morphology, decreased pyramidal dendritic arbor complexity, spine density [15, 25] as well as altered glial cell structure and function. MeCP2 present in astrocytes and microglia may indirectly regulate dendrites and synapses [108]. These mice also display higher spine turnover, with a net bias favoring spine elimination, reflecting the persistence of an immature mental state [106]. At four weeks of age, null mice begin to develop uncoordinated gait, hypoactivity, tremor, irregular breathing, and hindlimb clasping reminiscent of the hand stereotypies seen in RTT patients. As in humans, these mice exhibit abnormalities in neuronal metabolism [70], increased oxidative stress [42], increased serum cholesterol and triglycerides [21, 115], and abnormal mitochondrial structure with these phenotypes progressing in severity until death [189]. In an attempt to reverse the RTT phenotype, delayed expression of Mecp2 after the onset of neurological symptoms drastically decreased ataxia severity, hindlimb clasping, tremor, and irregular breathing, while also extending the lifespan of the animals. Additionally, overexpression of Mecp2 in Wild Type (WT) mice induced dendritic overgrowth, increasing the number of higher order branches of apical dendritic arbors in layer V pyramidal neurons. These findings suggest that lack of MeCP2 alone does not produce irreversible neuronal damage, which highlights the potential benefits of a therapeutic intervention targeting MECP2 in patients with RTT [207].

DGS mouse models.

The close homology between human chromosome 22 and portions of mouse chromosome 16 permits causal modeling of CNVs using targeted murine mutagenesis [128, 142]. The Df(16)A^+/− mouse model, which carries a hemizygous 1.3 Mb chromosomal deficiency (Df(16)A^+/−) syntenic to the 1.5 Mb human 22q11.2 deletion and encompassing 27 genes, has been generated using chromosomal engineering techniques [132]. Behavioral characterization of Df(16)A^+/− mice has revealed specific deficits in sensorimotor gating, working memory, and fear conditioning. Furthermore, several studies of Df(16)A^+/− mice have shown modest alterations in corticogenesis, relatively widespread deficits in dendritic complexity and spine density, as well as region-specific alterations in short- and long-term synaptic plasticity. Df(16)A^+/− mice also display transcriptomic alterations in the prefrontal cortex and hippocampus [63, 162]. Furthermore, diminished synchrony between the hippocampus and medial prefrontal cortex during a memory retrieval task is indicative of long-range functional connectivity impairments resulting from 22q11.2 deletions. Additionally, Df(16)A^+/− mutant mice lacking Comt show impaired emotional reactivity and increased aggressive behaviors [107]. The Df(16)A^+/− mutant is among several transgenic lines that model the 22q11.2 microdeletion (LgDel, Df1, Dgcr8), and these mice have demonstrated alterations in brain microRNA biogenesis. Dgcr8 haploinsufficiency may contribute to these alterations through down-regulation of a specific subset of microRNA. Specifically, miR-185 was the top-scoring down-regulated microRNA in the prefrontal cortex and hippocampus, regions which are also implicated in the pathology of schizophrenia. Furthermore, this reduction in miR-185 expression contributed to deficits in dendrite and spine development in hippocampal neurons [212].

FXS mouse models.

To examine the function of FMRP in FXS, mouse models have been generated with either complete or conditional ablation of Fmr1 [78, 113]. These Fmr1^−/− mice exhibit impairment of neurogenesis in the dentate gyrus, defects in neurite extension, deficits in various learning and memory tasks, as well as dysfunctional synaptic plasticity (LTP and LTD) in several brain structures. Collectively, this evidence suggests that the cognitive impairment observed in FXS is likely due to synaptic dysfunction [19]. Selectively increasing Htt expression, a long-non-coding RNA, using a modified CRISPR/Cas9 (dCas9–VP64–SAM) approach, rescued dendritic maturation deficits observed in Fmr1^−/− neurons [129]. Fmr1^−/− mice display only mild phenotypic characteristics, which are also seen in patients with FXS. This suggests a compensatory mechanism, likely through activity of FXR2P, which has a high degree of homology with FMRP. Additionally, Fmr1^−/− Fxr2^−/− double KO mice have demonstrated hyperactivity, increased anxiety, reduced PPI, impaired LTP, as well as altered spatial associative learning and memory. Together, these findings support the conclusion of a functional relationship between FXR2P and FMRP. Additionally, these double KO mice exhibited significant reductions in dendritic complexity, total dendrite length, and number of branching points that underscores how the interaction between FMRP and FXR can potentially contribute to the development of FXS phenotypes [194]. Furthermore, CGG knock-in mice, which serve as models for FXTAS, have been generated to study dendritic dystrophy [82].

Emerging mechanism: chromosome 10 (q26) deletion, characteristic phenotypes, and CRMP3/DPYSL4 as a novel candidate gene for ID

Previous work using animal models, as well as Golgi staining of post-mortem tissues obtained from patients with ID-associated syndromes, has indicated the presence of defects in dendritic growth, which points toward a need to elucidate the molecular mechanisms regulating this process [196]. Converging evidence arising from these experimental and clinical pathophysiological investigations has shown that dendritic malformation and synaptic dysfunction are frequent and prominent findings in ID. Our efforts directed toward uncovering new molecules responsible for dendritogenesis identified CRMP3 as a positive intracellular signaling regulator of dendrites and their spines in hippocampal neurons [68, 169, 174]. We found that CRMP3 has a profound influence on neurite initiation and dendritic complexity in vertebrate hippocampal neurons (Fig.1). Previous observations identified CRMP3 expression in axonal growth cones and the distal tip of dendrites in mammalian neurons [95]. Additionally, our work has revealed the involvement of CRMP3 in the development of dendritic spines (Fig.2) [172]. In a murine model with constitutive loss of CRMP3 expression, Crmp3^−/− mice develop altered LTP and pre-pulse inhibition (PPI) [169, 173, 174], which suggests its involvement in neuronal plasticity and behaviors typically associated with psychiatric illness. Importantly, over-expression of CRMP3 induces dendrite outgrowth through increased Ca²⁺ influx (Fig.2), which can be inhibited using Crmp3 siRNA. In vitro, Crmp3^−/− neurons, or neurons transfected with a dominant-negative truncated Crmp3 (Crmp3ΔC), displayed prototypical “fried-egg” morphology with complete absence of neurite extensions [172]. Additionally, over-expression of CRMP3 prevented prion-induced dendritic dystrophy in cultured neurons. Our observations suggest a distinct, but critical, contribution of CRMP3 activity in dendrite morphogenesis and maintenance [171, 172]. Work performed by other groups has found that CRMP3 interacts with the cytoskeletal proteins tubulin and actin. These studies have also identified that CRMP3 mediates Sema3A/Plexin-A/Nrp1 signaling and interacts with NrCAM, which is linked to Sema3F activity [147, 219]. Furthermore, CRMP3 expression is decreased in brain samples of AD patients, as well as the 3xTg-AD mouse model [60, 98], which both display dendritic dystrophy. It has been suggested that CRMP3 may be critically involved in learning and memory, since it is part of a group of long-lived synaptic proteins [86]. Indeed, treatment with Rosiglitazone, a peroxisome proliferator-activated receptor γ (PPARγ) agonist, increased CRMP3 expression in the hippocampus and improved cognition in the Tg2576 AD mouse model [51]. In contrast, in a mouse model of phenylketonuria, treatment with titanium dioxide nanoparticles (TiO₂NPs) decreased the length of hippocampal dendrites, reduced CRMP3 expression, and impaired behavioral measures of memory [90, 93, 97]. From these findings, it is tempting to hypothesize that CRMP3 may participate in dendritic remodeling and arborization, which are critical for learning and memory processing in vivo.

Figure 1. Expression of CRMP3/DPYSL4 during nervous system development and role in cultured hippocampal neurons.

In situ hybridization of CRMP3 in sagittal sections of embryonic and adult mouse brain. (a) At E12.5, transcripts were detected in dorsal root ganglia (drg) and the central nervous system (CNS). Different regions of the rhombencephalon (Rhomb), mesencephalon (Mesen), and diencephalon (Dien) were stained and are denoted with white arrows. The black arrows denoted the CRMP3 expression in the ventricular neuroepithelia and the cortex (Cx). Staining was also observed in cerebellum, which is marked with unfilled arrows. No staining was observed in the choroid plexus (cp). (b). In sections obtained from adult mice, CRMP3 mRNA was detectable mainly in the cerebellar granular layer, the inferior olive, and the dentate gyrus. (c). Summary of In Situ Hybridization with an antisense riboprobe: highly positive signal (+++); weak signal (+). Representative Crmp3^−/−cells with no visible processes (d; e) and Crmp3^+/− cell with neurites (f; g) using β-galactosidase activity (d; f) or immunostaining with MAP2 antibodies (e; g). Transfected with Fl-CRMP3, neurons are characterized by active transport of the protein, increased formation of lamellipodia, and increased dendritic arborization. Representative CRMP3-over-expressing neuron (h; ov-crmp3) immunostained for Flag (red) and the dendritic marker MAP2 (green). Overlay of image from transfected neurons are in yellow (h, right). Yellow arrows show control (h) or LacZ⁻ (d) cells. White arrows show neurites (f, g).

Figure 2. A proposed model for the regulation of dendrite morphology by CRMP3/DPYSL4 activity.

We propose a model involving CRMP3/DPYSL4 activity underlying neurite initiation and dendrite outgrowth following Ca²⁺ influx. Combinatorial stimulation of CRMP3/DPYSL4 gene expression, or activity by intrinsic and extrinsic factors (1). The localization and distribution of FL-CRMP3-positive puncta in dendrites suggest that CRMP3 could be associated with vesicles or large carrier protein complexes and is then actively transported to dendrites (2). Structural and biochemical studies support the notion that CRMP3 activity might be regulated by phosphorylation or other post-translational modifications (3–4) induced by neurotrophic factors or guidance cues (7). Activated CRMP3/DPYSL4 binds to protein partners and serves as an adaptor in a variety of signaling pathways including Ca²⁺ influx (CaV; 5). Our genetic studies have shown that CRMP3/DPYSL4 contributes to dendritic arborization, dendritic spine genesis (6) and neurite initiation (8). These effects might involve lamellipodia formation, cell migration (9), dendritic arborization, spine density and structure. Collectively, these functional clusters may affect cognition, learning and memory (10).

Approaches combining next-generation sequencing, cytogenomic mapping, and DECIPHER/IPA/GRAIL bioinformatics analyses have identified new chromosomic rearrangement, such as deletion or ring chromosome phenotypes. These studies have also identified chromosomal imbalance, along with new gene networks [84, 102, 223], and highlighted CRMP3/DPYSL4 as a new candidate gene for ID in data obtained from 1,354 patients. Additionally, these genetic/proteomic approaches have demonstrated interactions with CRMP2, DYNC11-1, HDGFRP3, and NrCAM, which is an integral component of the Sema3F receptor complex [49, 219]. Furthermore, recent studies using AGH have also detected CRMP3/DPYSL4 deletion in children with idiopathic ID [68]. Previously, we mapped the Crmp3/Dpysl4 gene onto chromosome 7 in mice [170] where it is linked to Mgmt, IGF2 and Drd2. However, in humans, we found that CRMP3/DPYSL4 is located on the distal region of chromosome 10 (10q.26) (Fig.3) [91]. Human chromosome 10 has 1.31 x 10⁸ nucleotides, containing approximately 1,300 genes, of which 816 are protein coding and 430 are pseudogenes [48]. Most of these genes have both strands overlapping coding genes, with 67 that are likely antisense transcripts. Using the published draft of the complete human genome (International Human Genome Sequencing Consortium, 2001) [53], it is possible to correlate regions of chromosomal deletion or imbalance with corresponding gene sequences, and thus identify genes potentially responsible for clinical phenotypes (Table 1). Previous cytogenetic studies have shown that the 10q25-10qter deletion syndrome is clinically heterogeneous with common clinical features including ADHD [36], ID, craniofacial anomalies, growth failure, and congenital heart defects [117, 211]. Because the size and location of the deletions are variable, it remains unclear if the variability and severity of the phenotype correlates with specific genetic segments or critical genomic coding regions. However, the presence of a clearly recognizable phenotype, despite deletion size variability, has led to the hypothesis that deletion of a critical region could produce the characteristic clinical picture. Studies using AGH have shown that patients with 10q25.2-10qter deletion, including the CRMP3/DPYSL4 gene, have a recognizable ID phenotype [99] (Table 2). No other known genes in this region are clear candidates for the dendritic anomalies and characteristic phenotype observed in the patients harboring this deletion. Table 1 lists genes of interest, either up or downstream of CRMP3/DPYSL4, and Table 2 reports selected genes on chromosome 10q terminal involved in brain disorders. These characteristic phenotypes are found in patients with Ring Chromosome 10, which is a disorder that occurs following breakage and rejoining of both chromosome arms, as well as patients with a linear subterminal deletion of chromosome 10. Patients with deletion of the CRMP3/DPYSL4 gene (10q26.3) always present with an associated ID phenotype. Interestingly, a subset of patients without CRMP3/DPYSL4 gene deletion do not present with ID syndromes (Table 2). Collectively, these observations demonstrate an intriguing correlation, suggesting that a critical region containing the CRMP3/DPYSL4 gene could contribute to the characteristic ID phenotypes seen in the 10q26 deletion. A role for CRMP3 in dendrite dystrophy in ID would be in agreement with its altered expression in AD patients, as well as in AD/PKU mouse models, which are characterized by dendrite degeneration.

Figure 3. Schematic representation of gene distribution on chromosome 10.

Mapping the q26 terminal area (a) and the genes involved in brain disorders (b).

Table 1.

Schematic presentation of the 10q25.2-q26.3 region and gene positioning from top to bottom, with clinical features.

Genes of interest on Chromosome 10 (q26.1-26.3)
Genes	Functions	References
PAX2	Chronic kidney diseases	Deng et al; 2019. Ref.50
EMX2	Transcription Factor	Cecchi et al; 2002. Ref.31
GLRX3	Vesicoureteric reflux	Darlow et al; 2017. Ref.41
EBF3	Facial dysmorphisms; ID candidate	Lopes et al; 2017. Ref.131
MGMT	Glioblastomas biomarkers	Jovcevska; 2019. Ref.105
MDR1	Amplified in glioblastomas	Suzuki et al; 2004. Ref. 193
ADAM12	Cancer; arthritis; cardiac hypertrophy	Ray et al; 2011. Ref.177
DOCK1	Cranio facial dysmorphism; ID candidate	Yatsenko et al; 2009. Ref.221
TXNL2	Breast, colorectal cancers	Qu et al; 2011. Ref.167
FGFR2	Microcephaly; ID candidate	Sangu et al; 2016. Ref.185
PTPRE	Obesity-induced disease development	Kogelman et al; 2016 Ref.112
MK167	Cancer risk factor	Joensuu et al; 2004 Ref.104
TCERG1	Apoptotic agent; spliceosome	Montes et al; 2015 Ref.148
BNIP3	Mitophagy; cancer	Panigrahi et al; 2019 Ref.159
JAKMIP3	Candidate for COPD	Mostafaei et al; 2018 Ref.150
DPYSL4/CRMP3	Dendrite maintenance; AD, ID candidate	Quach et al; 2020 Ref.174
LRRC27	Intra-ocular pressure; glaucoma	Woo Kim et al; 2020 Ref.216
PWWP2B	Histone modifications regulation	Zhang et al; 2018 Ref.225
NKX6-2	Hypomyelination; spastic ataxia	Chelban et al; 2020 Ref.27
GPR123	Candidate gene for asthma	Tremblay et al; 2008 Ref. 200
VENTX	Putative tumor suppressor	Gao et al; 2016 Ref.67
MIR202	Reduced pancreatic tumor	Mody et al; 2017 Ref.146
ADAM8	Neuro-inflammation; asthma	Knolle et al; 2009 Ref.111
TUBGCP2	Neuronal migration defect	Mitani et al; 2019 Ref.145
CALY	Attention deficit hyperactivity disorders	Plaisancie et al; 2014 Ref.163
PRAP	Developmental regulation during pregnancy	Duan et al; 1997 Ref. 57
ECHS1	Deafness; epilepsy; optic atrophy	Haack et al; 2015 Ref.81
MTG1	Cardiac hypertrophy; mitochondrial function	Xu et al; 2019 Ref.219
SPRN	Influence in TSE susceptibility	Lampo et al; 2007 Ref.118
CYP2E1	Alcohol abuse/immune interactions	Ratna et al; 2017 Ref.176
FRG2B	Neuro-endocrine tumors	Snezhkina et al; 2019 Ref.190
UTF	Germ cell lineage development	Kasowitz et al; 2017 Ref.109
PAOX	Over-expressed in myelofibrosis	Salati et al; 2016 Ref.183
SYCE	Non-obstructive azoospermia	Maor-Sagie et al; 2015 Ref.134
PPP2R2	Mitochondrial fragmentation; autophagy	Fang et al; 2013 Ref. 62
DRD1IP	Attention deficit/hyperactivity disorders	Laurin et al; 2007 Ref.121

Table 2.

Identification of commonly expressed genes surrounding CRMP3/DPYSL4: delineating the critical region for intellectual disability.

CRMP3/ DPYSL4	Deletion interval	ID	Number of Patients	References
Yes	10q26.2-10q26.3	Yes	4 patients	Plaisancie et al; 2014. Ref 163
Yes	10q26.2-10q26.3	Yes	1 patient	Courtens et al; 2006. Ref 36
Yes	10q25.2-10q26.3	Yes	13 patients	Irving et al; 2003. Ref 99
Yes	10q26-10qter	Yes	1 patient	Turleau et al; 1979. Ref 120
Yes	10q26.13-10qter	Yes	Patient N# 2	Yatsenko et al; 2009. Ref 221
No	10q26.2	No	Patient N#3	Yatsenko et al; 2009. Ref 221
No	10q26.2	No	1 patient	Laudier et al; 2016. . Ref 120
No	10q23.2-10q23.3	No	Patients N#1, 2, 3	Delnatte et al; 2006. Ref 47
No	10q21-10q23	No	12 patients	Bowles et al; 1996. Ref 17
No	10q22.3-10q24.1	No	1 patient	Alimi et al; 2015. Ref 4
No	10q22.1-10q22.3	No	1 patient	Tzschach et al; 2006. Ref 202
No	10q26.2	No	Meta-analysis	Darlow et al; 2017. Ref 41
No	10q24.2-10q24.32	No	1 patient	Hoefele et al; 2012 . Ref 89
No	10q24-q25	No	3 patients	Sanyanusin et al; 1995. Ref 186

An unholy trinity: reflections on the molecular crosstalk between ID, AD, and IQ

Numerous neurodevelopmental disorders that present with comorbid ID and neurodegenerative pathologies (e.g. AD) display similar patterns of altered dendrite and spine structures in the cerebral cortex and hippocampus [73, 191]. However, candidate gene targets, as well as the molecular signaling pathways underlying dendrite and spine dysgenesis are potentially distinct among these disorders. While neurodegenerative diseases are typically considered age-associated disorders where the accumulation and aggregation of pathogenic proteins leads to aberrant regulation of gene expression within dendrites and spines, in neurodevelopmental diseases with associated ID, alterations to dendrites and spines arise from genetic mutations that affect protein expression and gene transcription during the maturation of these structures [88].

Drawing on data from our laboratory, in combination with findings from animal models of various brain disorders, post-mortem tissue studies, and reviews of the current literature describing phenotypes in AD patients and ID patients with chromosome 10 q26.3 terminal deletion, we hypothesize that impaired expression of CRMP3/DPYSL4 may contribute to phenotypes in both ID and AD. Additionally, profiles of dendritic exuberance, positively correlating with performance in learning and memory-related tasks, have been previously posited as a proof of concept regarding this hypothesis [13, 35]. Ultimately, this raises questions of whether this gene and its protein product may also influence some aspects of human intelligence. Although a strict and comprehensive definition of human intelligence has been elusive due to the presence of different types, including sense of reasoning, creativity, and novel problem-solving ability, it is most commonly measured as an “IQ score” using the Wechsler Adult Intelligent Scale (WAIS) [100, 213]. Sequencing the human genome has identified ~30,000 genes, suggesting that the complex functional phenotype of an individual is not only directly related to the number of coding genes, but is also reliant upon genomic and proteomic networks. These networks are regulated through epigenetic factors, in addition to non-linear crosstalk mechanisms, leading to complex interactomes that produce divergent outputs from a small number of input genes [85, 124]. Consequently, it is important to highlight interactions between different levels of molecular regulatory elements and how this could contribute to the functional diversity observed during different periods over the span of a human lifetime. Although there is still much debate regarding the effects of various genes associated with high IQ scores, there are several observations that can be made. First, high IQ scores are linked to larger, more complex dendritic arbors and dendritic spine morphologies in human pyramidal neurons [33, 74, 164] (Fig.4). Second, Sema3F, which is important in the formation and refinement of neuronal networks [26, 45, 147, 180, 181], is associated with intelligence [34] and involved in cognitive decline. Furthermore, in a meta-analysis of 87,740 individuals, the Sema3F signaling pathway was significantly enriched in subjects with high IQ score [222] whereas, a large-scale GWAS meta-analysis involving 269,867 individuals listed CRMP3/DPYSL4 as a candidate gene correlated with high IQ [188]. Molecules like CRMP3, critical for dendrite development, may represent a common final pathway in ID from various etiologies. Additionally, the similar alterations of dendrites and spines found in patients with AD and ID, as well as related animal models, highlights the parallels between these disease states. While these observations are compelling, more comprehensive investigations and further analysis of the vast amounts of genetic information currently available will be required in order to effectively promote a candidate ID-AD-IQ gene to a validated target [210].

Figure 4. Dendritic tree and spine morphology changes associated with intellectual ability.

The model depicts a pyramidal neuron displaying high dendritic spine density, which is correlated with higher levels of intelligence. Lower scores on the IQ assessment, < 70, are associated with lower dendritic arbor complexity as well as decreased spine density in numerous brain regions. These morphological characteristics are highly correlated with scores on the IQ assessment and likely underlie the functional outcomes associated with ID.

Looking forward, it is easy to image CRMP3 as a potential target for gene therapy for a variety of brain disorders associated with these common structural features since CRMP3 induces dendritic outgrowth through increased Ca²⁺ and is essential for maintenance of dendritic spine structural integrity. Future studies regarding the role of CRMP3 in dendritic spine pathology associated with ID are warranted to better characterize this protein in human disease states along with the therapeutic potential offered by understanding its function.