Stage-dependent involvement of ADAM10 and its significance in epileptic seizures

2.1 Clinical clues from the concurrence of AD and epilepsy

Traditionally, AD and epilepsy do not belong to the same classification of brain disorders, which are individually characterized by cognitive decline and recurrent seizures respectively. However, an increasing amount of evidence has supported their intriguing relationship in the past decades. In fact, seizure morbidity ranges from 1%-22% of AD patients, and seizure incidence varies from 4.8 to 11.9/1,000 person-years in AD patients, which is 2- to 6-fold higher than the rates in age-matched normal individuals.1 In a large cohort of patients with autosomal dominant early-onset AD, the seizure incidence reached nearly 50% after an average follow-up of 8.4 years.2 Seizures and reduced seizure thresholds have been further confirmed in transgenic mouse models for familial AD.16 In addition, Down syndrome (DS) is a disease that simultaneously possesses the typical symptoms of AD and epilepsy, with epileptic seizures usually emerging over the course of dementia.15 All of these findings suggest the prevalence of epileptic seizures in AD.

On the other hand, community-dwelling elders with partial epilepsy have been found to exhibit cognitive decline in comparison to healthy controls and an even greater decline in executive function than do patients with mild cognitive impairment, which is known as a precursor to AD.17, 18 As a pathological marker of AD, Aβ aggregation in resected cerebral cortices of patients with temporal lobe epilepsy was significantly greater than that in autopsy controls.19 In a 50-year follow-up of childhood-onset epilepsy patients and matched controls, a high level of Aβ burden was observed in the prefrontal cortex, parietal cortex and posterior cingulate/precuneus of patients but not of controls,20 an observation that was similar to the findings in patients with preclinical AD.21, 22 In addition, antiepileptic treatments have been found to display a positive effect on cognitive function in animal and human studies.23 Hence, epileptic activities might be a driving factor for the incidence of AD in patients with epilepsy.

2.2 Emerging concern for stage-dependent ADAM10 in epilepsy

Considering the concurrence of AD and epilepsy, concern has emerged in recent years regarding whether these conditions share certain pathogeneses, and clarifying this issue might contribute to understanding epileptic seizures from a novel perspective. Interestingly, a series of cohort studies found risk factors associated with the genesis of seizures in AD, particularly the seemingly contradictory and independent risk factors of young age and severe dementia.11-15

Indeed, patients with genetic/familial forms of AD who carry mutations in APP, presenilin-1 (PS1) and −2 (PS2) are particularly prone to epileptic seizures. Aβ, acting as a downstream molecule correlating with the severity of dementia, likely plays a key role in epilepsy. Moreover, the percentage of patients with epileptic seizures has been found to reach one-third in patients with APP duplications, but no patient was found to have APP mutations over a 5-year observation period.2 Intriguingly, little neuronal loss appears in APP transgenic mice with spontaneous seizures, implying a prominent role of Aβ in synchronized abnormal discharges by a direct excitatory effect on brain networks rather than end-stage degeneration.1, 24 Despite these findings, the severity of dementia often increases with age because of chronic Aβ aggregation, which is not inconsistent with the predisposition to epileptic seizures at younger ages, indicating the involvement of other pathogenic effects in early-onset epilepsy. Among various hypotheses, ADAM10 is a promising candidate to bridge the apparent contradiction in risk factors for seizures in AD because of its stage-dependent involvement in both amyloidogenic processes and cortical dysplasia. Further investigation of this hypothesis would assist in addressing whether certain pathogenic mechanisms are shared between AD and epilepsy.

2.3 Epileptic phenotypes of ADAM10 in transgenic models

ADAM10 is a key α-secretase responsible for physiological shedding of the APP ectodomain, for which it competes and thereby inhibits Aβ deposition owing to β-site cleavage. Thus, ADAM10 plays a key role in AD pathology. Notably, transgenic mice with Adam10dn/APP experience more severe seizures and longer recovery times. These animals also show more significant degeneration in the hippocampal region compared to transgenic mice that moderately overexpress Adam10 (Adam10 mo)/APP and APP mice with endogenous ADAM10 levels.4 Furthermore, the loss of function of ADAM10 was observed to be related to increased seizure-associated mortality after weaning in an ADAM10 conditional knockout (cKO) mouse model.25 In addition, increased β-site cleavage products and epileptic seizures were observed in APPSWE mice that were developed to simulate extracellular Aβ aggregation.26 All of these findings support the notion that ADAM10 could protect against epileptic seizures in AD.

3.1 Amyloidogenic processes

Because of the competitive role of ADAM10 in inhibiting Aβ aggregation, the decreased function of this molecule seems to be a key factor for amyloidogenic processes.27-31 Thus, interest has increased into whether the Aβ burden mediated by ADAM10 is involved in epileptic seizures as well as in cognitive decline in AD. In fact, a correlation between the frequency of epileptiform-like discharges and the number of Aβ plaques has been observed in an APP/PS1 mouse model of AD. Moreover, epileptic activity appeared when N-(2-chloroethyl)-N-ethyl-bromobenzylamine was employed to form Aβ plaques in the cortex.5 Hypersynchronous activities underlying epileptic seizures even preceded Aβ plaques and memory impairment and could be improved by passive immunization with an anti-human APP/Aβ antibody,6, 7 indicating that early stages of aggregation before the development of Aβ plaques, such as the formation of Aβ tetramers, might initiate the process of epilepsy pathology.

Excessive excitatory transmission is necessary in the process of epileptic seizures, and Aβ deposition is significantly increased in the temporal cortex and the hippocampus of patients with refractory epilepsy.32 Notably, γ-aminobutyric acid (GABA) terminals are in contact with the somas and proximal axons of the cerebral cortex in layers II-VI but are replaced by Aβ deposits around the membranes of neurons, primarily pyramidal cells, in AD.33 As these GABA synapses exert inhibitory effects on the action potentials of excitatory pyramidal cells, the loss of function of these synapses would lead to the hyperactivity of the neurons surrounded by Aβ deposits. These previous studies have supplied a clue regarding the pathological role of Aβ-mediated excitotoxicity in epileptic seizures.

Recently, a humanized monoclonal antibody, solanezumab, which was designed to increase Aβ clearance, failed to improve the cognitive decline in a double-blind, placebo-controlled, phase 3 trial with a total of 2129 AD patients.34 Considering that epileptic seizures are comorbid with dementia in AD, we must doubt the role of Aβ burden in epilepsy. Interestingly, transgenic mice overexpressing the APP intracellular domain (AICD) and the binding partner of this domain, Fe65, displayed abnormal spiking episodes. Moreover, these mice were susceptible to a seizure-causing stimulus. Importantly, this phenomenon was not observed in transgenic mice overexpressing Aβ and harbouring a mutation (D664A) in the AICD. These results indicated that in addition to Aβ in amyloidogenic processes, AICD might be another modulator associated with epileptic seizures. Hence, targeting the integral modulation of Aβ and the AICD could be a strategy for treating epilepsy, and the upstream molecule ADAM10 could be a promising candidate for modulating amyloidogenic processes.

3.2 Cortical dysplasia

In addition to APP, many other ADAM10 substrates have been reported,35 and many more could remain undiscovered given the extensive effects ADAM10 exerts via the shedding of the ectodomain.36 These known and potential substrates speak to the variety of roles of ADAM10 in biological activities. In light of the complex interaction between ADAM10 substrates in vivo, it is necessary to comprehensively describe the major biological effects of ADAM10, as well as its classical role in amyloidogenic processes.

Using brain samples from transgenic Adam10dn mice, microarray analyses were performed to evaluate the influence of insufficient ADAM10 on mRNA expression profiles in the brain. A total of 143 genes were observed to be differentially expressed in transgenic mice compared with the wild-type mice, with most genes being implicated in nervous system development, neuronal projections and synaptic transmission.37 The crucial role of ADAM10 in brain development has been repeatedly described in recent years, and cortical dysplasia is one of the most common aetiologies of intractable epilepsy, especially in early-onset patients.38-41 These previous results imply the involvement of ADAM10 in epilepsy via the modulation of cortical development.

Compared with amyloidogenic processes in elderly individuals, ADAM10-mediated cortical dysplasia is distinct and is most likely caused by a non-amyloidogenic pathogenic mechanism in early-onset epilepsy. Thus, we reviewed the major progress that has been made in understanding the role of ADAM10 in this condition.

3.2.2 Neurite and synapse formation

Neurite outgrowth is a fundamental process for formation of cortical neuron circuits in the brain. Current evidence suggests that IgLON family proteins can be released into the extracellular space via metalloproteinase activity, thereby affecting neuronal maturation.49 In particular, neurons lacking Negr1, a member of the IgLON family, show reductions in both dendrite length and number, indicating that Negr1 promotes neuronal tree growth.50 Intriguingly, ADAM10 can release Negr1 from neuronal membranes, and the pharmacological inhibition of ADAM10 was demonstrated to lead to defects in neurite maturation. Importantly, this effect could be rescued by soluble Negr1 administration.10 Hence, ADAM10 can regulate neurite outgrowth via the ectodomain shedding of Negr1.

Neuronal cell adhesion molecule (NCAM) is an important regulator of neuronal outgrowth and synaptic plasticity. EphrinA/EphA-dependent axon repulsion is crucial for synapse formation, and ephrinA5/EphA3 can trigger NCAM proteolysis via ADAM10 to promote growth cone collapse.51 In cortical neuron cultures, the soluble extracellular domain of NCAM (NCAMed) was demonstrated to inhibit NCAM-dependent neurite branching and outgrowth.52 However, transgenic mice overexpressing soluble NCAMed displayed reduced GABAergic synapses and altered behavioural phenotypes.52 These results indicate that excess levels of soluble NCAMed because of ADAM10 activity can reduce the perisomatic innervation of cortical neurons by inhibiting synapse formation during cortical development.

Synapses are the basic units of information transmission, processing and integration in the nervous system. N-cadherin plays important roles during dendritic arborization, axonal guidance and synaptogenesis. In particular, N-cadherin at synaptic sites is involved in the regulation of cell-cell adhesion and plasticity control. Recent studies have indicated that N-cadherin can be shed by ADAM10, and impairing the localization and activity of ADAM10 at synaptic sites could result in a decreased ability to process N-cadherin, leading to the accumulation of full-length N-cadherin, increases in spine head width, and abnormal modifications of the number and function of glutamatergic AMPA receptors.53 In addition, the roles of PLD1 and PACAP38 in synaptogenesis were also affected by ADAM10-mediated N-cadherin cleavage.54

Neurexins are a type of synaptic cell adhesion molecule involved in the development and maturation of neuronal synapses. Notably, the neurexin Nrxn3β can be processed by ADAM10, and the shedding of the ectodomain of Nrxn3β releases a soluble fragment that increases the spine density of newborn hippocampal neurons,55 indicating that ADAM10 modulates synaptogenesis via ectodomain shedding of Nrxn3β, as well as Negr1, NCAM and N-cadherin.

ADAM10 plays a key role in ectodomain shedding. Thus, the functions of ADAM10 depend on the biological effects of its substrates. Several substrates have been associated with epileptic seizures via their effects on amyloidogenesis in the ageing stage and cortical dysplasia in the developmental stage (Figure 1). In addition, more substrate candidates have been found through mRNA and protein expression profiling of ADAM10-deficient neurons and brain tissues.36, 37 These data together indicate that the pathogenic mechanisms of ADAM10 in epilepsy are likely complex and must be further explored.

4.1 Transcriptional level

In the human genome, ADAM10 is located at 15q21.3 and has a length of 154 744 nucleotides (nts). Within 2 200 nts upstream of the translational start site, a classical CAAT-box at −480 nt in the 5′-flanking region of the human ADAM10 gene was found via computational sequence analysis, representing a potential regulatory element required to initiate gene transcription.56 A series of deletion mutations further identified −508 nt to −300 nt as the core promoter region. Interestingly, a potential Maz binding site (−491 nt), a CAAT-box (−479 nt), a Sp1 binding site (−366 nt) and an USF element (−317 nt) were observed in this region. Removing these binding sites strongly decreased the promoter activity of ADAM10,56 indicating that this region is required for ADAM10 transcription.

In the human genome, numerous genetic variations exist around the core promoter region of ADAM10, but most of these are rare mutations. Only rs514049 A > C and rs653765 G > A (−630 nt and −279 nt upstream of the translation initiation site of ADAM10 respectively) are common variations that may potentially confer genetic susceptibility to ADAM10-related diseases (Figure 1). Intriguingly, linkage disequilibrium between these two sites was observed via haplotype analysis, and compared to the A-G haplotype (rs514049-rs653765), the C-A haplotype showed decreased levels of soluble APPα in AD but higher levels than those in healthy controls.57 This finding implies the differential involvement of TFs in ADAM10 transcriptional regulation. Remarkably, the C-A haplotype (rs514049-rs653765) also showed an association with epilepsy.58 Thus, future efforts should be made to clarify the involvement of TFs in the haplotype.

Considering that BACE1 competes with ADAM10 to shift APP processing from protective α- to neurotoxic β-cleavage, a library composed of 704 expression plasmids encoding human TFs was employed and ultimately found that XBP1 is an exclusive candidate for ADAM10 regulation compared with BACE1. In addition, numerous TFs or inducers, such as XBP1,59 USF1,60 Sox2,61 PPARα,62 melatonin63 and SIRT1, 64 have been reported to promote ADAM10 transcription and to protect against amyloidogenic processes (Figure 1). However, whether these TFs also function in non-amyloidogenic processes, particularly cortical dysplasia, remains unclear. Among humans, mice and rats, the sequence identity is as high as 78% within 500 nts upstream of ADAM10 translational start sites. In particular, the regulatory elements are highly evolutionarily conserved between Homo sapiens and Rattus norvegicus, which would facilitate cross-species experiments involving transcriptional modulation of ADAM10.

4.2 Posttranscriptional level

5.1 Amyloidogenic processes

Amyloidogenic processes gradually develop with age; thus, dementia and complicated seizures often emerge in the elderly, resulting in a long-term window for early intervention. As a key α-secretase, ADAM10 can inhibit the amyloidogenic pathogeneses. Through microarray analyses, ADAM10 overexpression was found to affect a series of pathways, primarily those involved in cell communication, nervous system development, neuronal projections and synaptic transmission. However, proinflammatory and proapoptotic proteins were not up-regulated in ADAM10 transgenic mice,37 indicating that reasonable increases in ADAM10 expression may facilitate seizure control without the risk of significant inflammation- and apoptosis-related damage.

Several TFs, such as XBP1, USF1, Sox2 and PPARα, have been demonstrated to be required for ADAM10 transcription. Thus, these TFs are potential molecules for seizure control. In addition, vitamin A improves the AD-related attenuation of cognitive function in mouse models by activating the downstream retinoic acid receptor, a TF involved in ADAM10 transcription.92 Considering that vitamin A is safe and convenient for administration, the effects of this vitamin on complicated seizures merit identification in future experiments. In addition, the genetic variations that potentially confer susceptibility to epilepsy, such as the C-A haplotype (rs514049-rs653765) around the core promoter region 57, 58 and the six polymorphisms within the ADAM10 5′-UTR G-rich region must also be studied.

Recently, widespread concern arose over the failure of solanezumab to improve cognitive decline in an AD clinical trial.34 Compared to the concentrations in the peripheral blood, the solanezumab concentration in the brain was extremely low because of the blood-brain barrier (BBB). Thus, there may have been insufficient levels to function, which are considered an important reason for the failure of the trial. Interestingly, miRNAs are key modulators at the posttranscriptional level, having the ability to bypass the BBB via intranasal administration.93 Thus, miRNAs might be promising candidates for seizure control. Several miRNAs, such as miR-144, miR-221, miR-1306, miR-103 and miR-107, have been observed to function in AD. In addition, as primary drugs for dementia that induce acetylcholinesterase (AChE) overproduction, AChE inhibitors were observed to increase ADAM10 activity by promoting ADAM10 trafficking.94 However, it is uncertain whether these molecules are also involved in complicated seizures.

Notably, Adam10dn mice without transgenic APP overexpression experienced seizures for shorter durations and showed less neuronal cell death and neuroinflammation after kainate injection than did wild-type mice. This result indicates that the antiepileptic effects of ADAM10 that act via amyloidogenic processes most likely depend on APP overexpression.4 Hence, the incremental modulation of ADAM10 in amyloidogenic processes seems to be limited to treating epileptic seizures in AD.

5.2 Cortical dysplasia

Unlike amyloidogenic processes, cortical development begins during the embryonic period and continues after birth; thus, cortical dysplasia usually occurs because of aetiological injuries during the cortical development. In clinical practice, epileptic seizures triggered by cortical dysplasia are resistant to almost all current drugs. From this perspective, epilepsy should be considered a neurodevelopmental disorder, especially early-onset epilepsy.95 As pathological changes in the cortex are difficult to reverse, surgical resection is considered a strategy to treat cortical dysplasia-related epilepsy. A series of clinical studies have demonstrated favourable outcome after surgical resection with the help of image positioning.96 However, the manifestation of cortical dysplasia is not always typical and is often uncertain in clinical practice. Thus, there is an urgent need to determine how these patients can be identified, particularly in the early processes of cortical dysplasia, as this recognition is necessary for early intervention.

Genetic screening has been widely used for the prenatal diagnosis of developmental disorders in clinical practice, and this approach should supply clues for the early recognition of cortical dysplasia. As ADAM10 is a key modulator of cortical dysplasia, screening for ADAM10 genetic variations might contribute to early recognition of high-risk individuals with cortical dysplasia-related epilepsy. In fact, there are numerous polymorphic sites with potentially relevant functionality, such as rs514049, which is near the core promoter region, as well as rs653765, rs886775779, rs1022312149, rs367745138, rs900600176, rs373778366 and rs199972921, all within the ADAM10 5′-UTR. The association between these polymorphisms and cortical dysplasia warrants future investigation.

In addition to regulatory region polymorphisms, exon mutations are also important; however, missense mutations of ADAM10 in epilepsy patients have not been reported in whole-exome sequencing clinical studies. These results seem to cast doubt on the genetic involvement of ADAM10 in cortical dysplasia. In our opinion, the negative results of whole-exome sequencing can be readily explained. Given that ADAM10 is a key modulator of embryonic development, missense mutations in ADAM10 exons most likely result in the severe impairment of its function, followed by embryonic or neonatal death before being enrolled in whole-exome sequencing.

As mentioned above, surgical resection is an effective therapy for cortical dysplasia, but the development of drugs to block pathologic discharges that result from cortical dysplasia should not be ignored. In recent years, a number of differentially expressed proteins underlying childhood cortical dysplasia with epilepsy were identified via iTRAQ-facilitated proteomic profiling, which provides a proteomic profile after disease onset.97 In addition, miR-323a-5p and miR-4521 expression levels are aberrant in epilepsy patients with focal cortical dysplasia.98, 99 Hence, in combination with iTRAQ proteomic profiling and genetic screening for the early detection of cortical dysplasia, a comprehensive investigation of these miRNAs might represent a method for treating seizures in these patients.