ADAM10 and its role in Alzheimer disease

Abstract

Introduction

The amyloidogenic pathway of amyloid precursor protein (APP) processing is well known in the pathogenesis and therapeutics of Alzheimer disease (AD), whereas the non-amyloidogenic pathway has been less studied. ADAM10 is the main α-secretase responsible for this pathway in the human brain.

Content

ADAM10 belongs to a family of transmembrane proteins with catalytic activity. It acts as an α-secretase on APP and many other substrates, some of which are particularly relevant in the central nervous system. The ADAM10 gene has been identified in genome-wide association studies of patients with AD; mutations have been reported in families with strong functional support but incomplete segregation; and haploinsufficiency has been reported in a family carrying a nonsense mutation. However, genetic studies of AD cohorts have not identified causal variants. ADAM10 levels in biological fluids show conflicting results, except in platelets, where patients with AD consistently exhibit reduced levels. Stimulation of ADAM10 as a therapeutic target offers new opportunities through various components and such measures as physical exercise. However, only one (positive) clinical trial has been published to date, using the retinoid acitretin.

Conclusions

ADAM10 plays a fundamental role in brain function, and sufficient studies support its involvement in AD pathogenesis. There are only isolated examples of ADAM10 mutations as a genetic cause, but these encourage continued screening in familial AD. ADAM10 levels in platelets could be considered as a biomarker. The enhancement of ADAM10 expression in AD remains a therapeutic target that requires further research.

Resumen

Introducción

La vía amiloidogénica de procesamiento de APP es bien conocida a nivel de patogenia y terapéutica en la EA, mientras que la vía no-amiloidogénica ha sido menos estudiada. ADAM10 es la principal α-secretasa responsable de esta vía en el cerebro humano.

Desarrollo

ADAM10 pertenece a una familia de proteínas transmembrana con actividad catalítica. Actúa como α-secretasa sobre APP y otros muchos sustratos, algunos de ellos especialmente relevantes en el SNC. El gen ADAM10 ha sido señalado en estudios GWAS de EA; se han comunicado mutaciones en agregados familiares con respaldo funcional robusto pero segregación incompleta; y se ha demostrado el efecto de haploinsuficiencia de una mutación nonsense en una familia. Sin embargo, estudios genéticos de cohortes con EA no han encontrado variantes causales. Los niveles de ADAM10 en fluidos biológicos muestran resultados contradictorios excepto en plaquetas, donde los pacientes con EA tienen una reducción consistente. La estimulación de ADAM10 como diana terapéutica plantea posibilidades de desarrollo a través de varios componentes y medidas como el ejercicio físico, pero hasta ahora hay un solo ensayo clínico (positivo) con el retinoide Acitretino.

Conclusiones

ADAM10 tiene un papel fundamental en el funcionamiento cerebral y suficientes estudios apoyan su implicación en la patogenia de la EA. Como causa genética hay sólo ejemplos puntuales pero que animan a continuar su screening en EA familiar. Los niveles de ADAM10 en plaquetas podrían plantearse como biomarcador. La potenciación de ADAM10 en la EA es una diana terapéutica pendiente de más investigación.

Introduction

In 1906, Alois Alzheimer was the first author to report a clinico-pathological description of the condition that, a few years later, Emil Kraepelin named after him, Alzheimer disease (AD).

AD has traditionally been characterised by the presence of 2 anatomopathological findings: neuritic senile plaques (accumulation of misfolded amyloid protein) and neurofibrillary tangles (accumulation of hyperphosphorylated tau protein).¹ Both are accurate markers for the definitive diagnosis of the disease. However, CSF protein panels and positron emission tomography with amyloid tracers have also been considered as AD biomarkers in the last decade^2,3; these techniques enable in-vivo biological diagnosis according to the amyloid-tau-neurodegeneration profile (A-T-N).⁴

The pathophysiology and progression of senile plaques and neurofibrillary tangles is not fully understood. Much of the current research is based on the amyloid hypothesis, which considers dyshomeostasis to be an essential phenomenon in the production and clearance of β-amyloid (Aβ) fragments.⁵ This would cause accumulation of Aβ peptides in the form of senile plaques, triggering the AD pathological cascade as much as 2-3 decades before symptom onset. β-amyloid peptides are generated after processing of amyloid precursor protein (APP). The functions of this transmembrane protein are not fully understood, although it is known to be processed through 2 alternative metabolic pathways, classified as amyloidogenic and non-amyloidogenic according to whether Aβ₁-₄₂ peptides, the substrate of amyloid plaques, are generated.

In our study, we performed a literature review of the role in AD pathophysiology of the ADAM10 protein, the main brain α-secretase and the main component of APP processing through the non-amyloidogenic pathway. We also reviewed articles indexed in PubMed and including the keywords “ADAM10” and “Alzheimer” (both original and review articles), and selected those considered most relevant for clarifying the role of ADAM10 gene variants in the development of AD, as well as the current state of research into biomarkers of the non-amyloidogenic pathway in different biological samples.

The amyloid hypothesis and APP processing

In the amyloidogenic pathway, APP is initially cleaved by β-secretase (BACE) and subsequently by γ-secretase, leading to the formation of Aβ peptides (Fig. 1). Depending on the site of APP processing through this pathway, Aβ fragments of different lengths are created: Aβ₁-₄₀ (the most abundant) and Aβ₁-₄₂ (less abundant, accounting to approximately 10% of Aβ₁-₄₀), which is more insoluble and is closely related to the formation of senile plaques. Cleavage by BACE produces a soluble residual fragment of APP, known as sAPPβ.

Figure 1.

Amyloidogenic and non-amyloidogenic pathways of APP processing.

Created using the Biorender software.

This amyloidogenic pathway is well understood, as the known genetic causes of early-onset autosomal dominant AD affect components of this pathway. Mutations in the APP gene involve exons 16 and 17, which encode the protein region between the BACE and γ-secretase processing sites,⁶ whereas mutations in the PSEN1 and PSEN2 genes affect proteins that are part of the γ-secretase complex.⁷ At the biomarker level, the amyloidogenic pathway is well-documented, with a decrease in the CSF Aβ₄₂/Aβ₄₀ ratio (normal value > 0.06) being considered to indicate the presence of cerebral amyloidosis, with > 85% sensitivity and specificity.⁸ This pathway has also been analysed as a therapeutic target, with drugs aiming to treat abnormal accumulation of Aβ peptides (passive immunotherapy with anti-amyloid antibodies)⁹ and decreased BACE activity.¹⁰

The non-amyloidogenic pathway occurs through the action of α-secretase; the main protein performing this function in the brain is ADAM10. This pathway is probably less well understood due to the lack of clinical or genetic examples of AD affecting its components. However, it is reasonable to consider its role in its pathogenesis as well as its usefulness as a therapeutic target. A defect in α-secretase functioning may incline APP processing towards the amyloidogenic pathway and may represent a significant risk factor for the development of AD. Therefore, we analysed ADAM10 and its role in AD from a genetic viewpoint, as well as its possible usefulness as a biomarker in different biological fluids.

Structure and function of the ADAM10 protein

ADAM10 is a protein of 748 amino acids that is considered the main α-secretase in the human brain. It belongs to the ADAM (A Disintegrin And Metalloproteinase) family, a group of transmembrane proteins with proteolytic activity on multiple substrates participating in numerous physiological and pathological processes.¹¹ Its activity as protease consists of cleaving the ectodomain part of proteins adhered to the cell membrane (ectodomain shedding), so that they can initiate their biological signalling. ADAM proteins are characterised by preserved amino acid domains that include an N-terminal sequence (necessary to direct proteins towards the secretory pathway), a prodomain that is responsible for maintaining inactivity of the catalytic region, a metalloprotein domain, a disintegrin domain, a cysteine-rich domain, a transmembrane domain, and a cytoplasmic domain (Fig. 2). Western blot characterisation of ADAM10 reveals 3 species with different molecular weights: an immature full-length form retaining the prodomain (proADAM10; ∼80 kDa), a mature unprocessed full-length form (ADAM10f; ∼55 kDa), and a truncated large soluble form released from the membrane (sADAM10; ∼50 kDa), which is described as an inactive form in some studies (Fig. 2).¹²

Figure 2.

The type I transmembrane protein ADAM10. Figure captions: A) Diagram depicting the domains of the type I transmembrane protein ADAM10. The arrow indicates the prodomain region where the 3 variants described as pathogenic are located. B) Western blotting for ADAM10 in CSF samples from controls, using a specific antibody kit for the ADAM10 central region, which is common to all variants.

Cys: cysteine; Cyt C: cytoplasmic C terminal; Dis: disintegrin; Pro: prodomain; TM: transmembrane.

As the main α-secretase, ADAM10 plays a crucial role in the non-amyloidogenic processing of APP (Fig. 1). In this process, ADAM10 cleaves APP at a specific codon, thus generating a residual non-amyloid peptide known as sAPPα (soluble fragment produced by α-secretase). Besides the fact that this pathway does not generate Aβ fragments, sAPPα has a neuroprotective and neurotrophic effect, with such benefits as promotion of synaptogenesis and neuroplasticity, as well as antiapoptotic effects.

In addition to its direct activity on APP as an α-secretase, ADAM10 acts as an ectodomain protease for another 90 membrane proteins. One particularly relevant substrate is triggering receptor expressed on myeloid cells 2 (TREM2), which participates in the microglial immune response to the Aβ of senile plaques.¹³ ADAM10 is responsible for releasing the soluble form of TREM2 into the extracellular matrix. By cleaving other proteins, ADAM10 has been shown to participate in maintaining synaptic activity, neurogenesis and gliogenesis during brain development, and neurogenesis in the adult hippocampus.¹⁴

We should also mention the various roles of ADAM10 outside the nervous system. ADAM10 is involved in correct embryogenesis and tissue differentiation, and in the homeostasis and regeneration of several tissues.¹⁵ Its relevance is demonstrated by mortality in ADAM10 knock-out mice, which die in early embryonic stages.¹⁶ It is also involved in modulating the immune response through cleavage of surface proteins in immune cells (receptors and adhesion molecules); in oncological processes, it cleaves such proteins as cadherins, which favour migration of cancer cells, thus contributing to tumour progression and metastasis. All these functions may represent obstacles to the development of ADAM10-mediated therapeutic interventions with relatively selective effects and no adverse effects.

The ADAM10 gene

The role of ADAM10 as the main α-secretase has sparked research interest in the ADAM10 gene (spanning 19 exons, located on chromosome 15q21.3) and its variants as a possible cause of AD pathogenesis. Several genome-wide association studies have consistently documented the association between some ADAM10 variants and AD in numerous well-characterised samples; therefore, ADAM10 is part of the polygenic set that confers late-onset AD its highly heritable nature.^17,18

Two rare ADAM10 variants (Q170H and R181G in the prodomain region, with an allele frequency < 0.01) are associated with late-onset AD, and have been reported in 7 families.¹⁹ Cosegregation of these variants with AD in these families was incomplete, but functional studies supported their pathogenicity and association with the disease. Suh et al.²⁰ showed that mutations promoted AD-type changes in transgenic mice; using cell cultures, the authors demonstrated that both variants impair the molecular chaperone function of the ADAM10 prodomain, attenuate the protein’s α-secretase activity, and shift APP processing towards β-secretase-mediated cleavage, thus increasing the amyloid plaque load (Fig. 3). Presence of mutations also decreased the effect of ADAM10 as a stimulator of adult hippocampal neurogenesis.²⁰

Figure 3.

Experimental studies on the impact of ADAM10 variants. Summary of the experimental studies performed to establish the functional impact of ADAM10 variants described in families with AD. All studies report an imbalance in APP processing, with an increase in processing through the amyloidogenic pathway.

Other genetic studies have examined the presence of certain single-nucleotide polymorphisms (SNP), mainly in the ADAM10 promoter region. Three studies performed in Chinese populations did not find differences in the distribution of genotypes between patients with AD and controls,²¹ although some differences were observed when stratifying by age of onset or APOE-ε4 allele.²² Two studies have found an association between the rs2305421 SNP and development of AD.^19,23

The most significant correlation between pathogenicity and an ADAM10 mutation is probably the one described by our study group in a small family with early- and late-onset AD, presenting a dominant inheritance pattern and with affected individuals showing the p.Tyr167*(Y167*) mutation.²⁴ This is a nonsense mutation that generates a premature stop codon with a deleterious change in protein length. In silico prediction programs assign the highest pathogenic score to this type of mutation. Furthermore, the p.Tyr167* mutation is not included in public databases and was not reported after a large-scale analysis of sporadic cases of AD.²⁵ We did not identify more cases with this variant after examining 200 cases from our cohort of familial AD and 274 cases from an independent cohort. We should underscore that the mutation is located in the same prodomain region and very close to the other 2 rare, probably pathogenic mutations (Q170H and R181 G) described in groups of familial late-onset AD.¹⁹

In this family of p.Tyr167* mutation carriers, in addition to the essential biomarkers of AD (Aβ₁-₄₀, Aβ₁-₄₂, and their ratio; total tau, and p-tau, which showed values corresponding to A + T+N+ in accordance with the diagnostic schema⁴), we analysed ADAM10 levels and APP processing products (sAPPα and sAPPβ) in the CSF of 2 brothers with AD versus controls and cases with sporadic AD. Findings were compatible with ADAM10 haploinsufficiency. We observed a 50% decrease in all ADAM10 species in carriers as compared to controls, sAPPα levels in the CSF also showed a significant, selective reduction of approximately 70% with respect to controls and patients with sporadic AD. sAPPβ levels in carriers of the mutations were in the highest ranges, but showed no statistically significant differences with regard to patients with sporadic AD and controls, probably due to the small sample size.

Fig. 3 summarises the experimental studies performed to establish the functional impact of ADAM10 variants described in families with AD. Overall, these findings suggest selective impairment of the non-amyloidogenic pathway with an imbalance favouring APP processing through the amyloidogenic pathway, and support the involvement of these variants in the development of the disease.

Subsequently, in a larger study, we sequenced the entire ADAM10 gene (exonic and adjacent intronic regions) in a cohort of 100 patients with AD and positive family history, mostly with late onset. We suggest the possibility that some cases of familial late-onset AD are associated with the presence of rare variants (allele frequency < 0.01) of the ADAM10 gene that cause more significant age-related attrition or impairment of α-secretase activity in comparison with cases of AD in carriers of more frequent variants. We decided to compare the sequencing of cases of AD against results in 96 cognitively preserved individuals aged 90-99 years, thus comparing 2 opposing populations, one with sensitivity and the other with resistance to AD, in order to optimise the differences.²⁶ In our limited sample, we observed only one exonic variant, Q170H, in 3 patients with AD, but also in one cognitively healthy individual. Between 10% and 15% of the members of both groups were carriers of different intronic variants whose significance is still unknown. A variant in the 3′UTR promoter region, associated with ADAM10 transcription, was observed in 6 cognitively healthy individuals versus 1 patient with AD (P = .06).²⁶

ADAM10 as a biomarker

Several studies have assessed the behaviour of ADAM10 in different biological samples to analyse its role in the pathophysiology of AD and examine its potential as biomarker. Table 1 includes the most relevant findings in the samples analysed.

Table 1.

Studies of ADAM10 in AD.

Sample analysed	Cases/Analysis	Results	Ref.
CSF	Sporadic AD vs controls	ADAM10 reduction in AD	^12
	ADAM10 mutation vs sporadic AD vs controls	Reduced ADAM10 in mutation carriers and sporadic AD vs controls	^24
	Sporadic AD vs controls	Increased soluble ADAM10 in AD	^30
	Sporadic AD vs familial AD vs controls	Similar levels in all 3 groups	^26
Plasma	Sporadic AD vs controls	Increased ADAM10 in AD, but the protein is inactive	^30
	MCI vs patients showing physical frailty	Increased ADAM10 in MCI	^31
	Longitudinal cohort study of patients aged > 60 years	ADAM10 increase is associated with a decrease in the MMSE score.	^32
Platelets	Sporadic AD vs controls	ADAM10 reduction in AD	^31,33–35
	Cognitively preserved subjects of different ages	ADAM10 increase in older patients	^36
Brain tissue	mRNA in the hippocampus in sporadic AD vs controls	Twofold increase in AD vs controls, regardless of severity	^28
	ADAM10 levels in the hippocampus and cerebellum of carriers of ADAM10 promoter haplotype	Haplotype carriers present lower amyloid levels, increased CSF sAPPα levels, and higher ADAM10 levels in the hippocampus.	^27
DNA	Assessment of the SNP rs2305421 and AD in a cohort	Positive statistical association between the SNP rs2305421 and late-onset AD	^33
	Assessment of the pathogenicity of the Q170H and R181 G mutations	Mutations are associated with development of AD in cell cultures and knock-out mice	^20
	Genome-wide association study in patients with AD	AD is associated with ADAM10 variants	^17,18
	Comparison between carriers of the novel Tyr167* mutation vs AD vs controls	Reduced levels of ADAM10 and its metabolic products in carriers, segregation	^24
	Comparison of exomes in patients with AD vs controls	Non-significant trend towards a correlation between ADAM10 mutations and AD	^18

A negative correlation between ADAM10 activity and the amount of senile plaques has been described in the brain tissue of patients with sporadic AD. The study by Bekris et al.²⁷ reported elevated levels of ADAM10 in the hippocampus and sAPPα in the CSF of patients with AD with low amyloid burden, in comparison to those with a high burden, thus confirming that increased ADAM10 activity is associated with decreased Aβ deposition. However, studies focused on ADAM10 mRNA have reported a twofold increase in its levels in the hippocampus of patients with AD with respect to controls, suggesting that it may be a compensatory mechanism.²⁸ The influence of the APOE genotype has also been analysed in ADAM10 expression and activity, with significantly lower ADAM10 activity in the brains of patients with AD who were carriers of the APOE-ε4 allele.²⁹ This may influence the pathogenesis of AD in 2 ways, by increasing Aβ synthesis and by reducing the neurotrophic effects of sAPPα.

Divergent results are reported from the quantification of ADAM10 levels in the CSF. Some authors observed a decrease in CSF levels of all 3 ADAM10 species, in Western blotting studies of patients with AD.⁹ Other researchers have reported a relative increase in levels of soluble ADAM10, which was observed to be inactive.³⁰ As ADAM10 is active as a membrane-anchored protease, soluble ADAM10 is likely to represent an inactive protein, and increased levels could be linked to a reduction in overall enzyme activity.

Our group analysed CSF samples from 82 patients with AD and performed sequencing studies of the ADAM10 gene to establish correlations between rare variants of the gene and protein levels.²⁶ We used a commercial ELISA kit with a specific range for CSF samples; however, ADAM10 levels were very similar in patients and controls, with no differences observed between groups. Neither did we observe lower levels in carriers of rare variants (Q170H and intronic variants). The variable findings obtained with the ELISA kit when repeating the analysis of some samples, together with the variability of ranges obtained between batches, suggests that ADAM10 quantification were not reproducible.

Regarding plasma levels, there is some evidence of increased levels of ADAM10 in AD, although only one study has directly assessed a cohort with a definitive diagnosis of AD.³⁰ Two other studies showed a similar trend: increased plasma levels of ADAM10 in older patients with mild cognitive impairment and mild AD with respect to controls,³¹ and a negative correlation between ADAM10 levels and MMSE scores.³² We should underscore that, like in the CSF analysis, Pereira et al.³⁰ found that the increase in plasma levels of ADAM10 corresponded to the soluble and inactive form.

ADAM10 expression in platelets (detected by Western blotting) suggests that this may be the biological sample offering the most reliable data. Several studies describe a consistent decrease in patients with AD,^33–35 which is also associated with a decrease in the sAPPα fragment in platelets and CSF of the same individuals.³³ Furthermore, platelets have been analysed in cognitively preserved individuals of different ages, with ADAM10 levels being lower in younger subjects than in those in their 80 s. This has been interpreted as a protective or resistance effect against AD, provided by α-secretase hyperactivity.³⁶

ADAM as a therapeutic target in AD

Despite the high potential of increased ADAM10 to fight against Aβ peptide deposition, we should underscore the lack of data on ADAM10 as therapeutic target. The work by Peron et al.³⁷ reviews some substances that may act at the transcriptional level to increase ADAM10 activity. These substances include retinoids. Retinoic acid is a vitamin A derivative obtained from the diet, which, in addition to its potent antioxidant action, is an inducer of ADAM10 promoter activity. There is only one phase-II clinical trial, published in 2014, on the use of acitretin (a retinoic acid derivative that increases expression of the gene) in AD, which reported positive results.³⁸ The endpoint of this trial was an analytical marker: sAPPα levels in the CSF.

Sirtuins, particularly SIRT1, may also activate ADAM10 transcription by stimulating the retinoic acid receptor. In animal models of AD, expression of SIRT1 decreases AD pathology burden and behavioural problems, and increases tolerance to neuronal stress. Indirect induction of SIRT1 by natural components such as resveratrol and caloric restriction may also delay neurodegeneration. Furthermore, melatonin has been shown to increase ADAM10 expression, decrease Aβ levels, and improve learning deficits in animal models of AD. SOX-2 protein, which is crucial in neurogenesis and tissue regeneration, is also known to positively regulate ADAM10.

At the post-transcriptional level, several drugs may increase ADAM10 function. In vitro studies have shown that donepezil favours ADAM10 activity and its transportation to the cell membrane, although this does not translate into a detectable disease-modifying effect. Several natural antioxidants that have been studied in AD, such as resveratrol, gingerol, and curcumin, may also present this effect. Statins may favour the enzymatic activity of ADAM10. In vitro studies have shown that reduction of cholesterol by using statins increases the formation of sAPPα and decreases Aβ production. However, there are no definitive clinical data on this matter in humans; a trial with lovastatin showed decreased levels of Aβ peptides in serum but no significant effect on Aβ levels in the brain.

In addition to pharmacological interventions, ADAM10 potentiation may be one of the mechanisms involved in the association between physical exercise and the reduction in the risk of dementia and more benign progression, which has been clearly documented in population-based studies.^39,40 No specific study in humans has assessed the precise effect or mechanism acting on the biological sign of AD, but it has been shown in mice that physical exercise promotes the non-amyloidogenic pathway through the neurotrophic factor BDNF.^41,42 BDNF favours ADAM10 activity and redistribution of the protein from the cell membrane to the intracellular compartment (Golgi apparatus and lysosomes), where it performs part of its catalytic function. This is associated with increased sAPPα levels and decreased Aβ40 and Aβ42 levels in the hippocampus.⁴² Furthermore, the previously mentioned effect attributed to some natural antioxidants may be one of the factors responsible for the protective effects against cognitive impairment in general⁴³ and AD in particular,⁴⁴ in diets rich in such substances, such as the Mediterranean diet.

Table 2 includes the main interventions that may potentially stimulate ADAM10 activity.

Table 2.

Potential stimulators of ADAM10.

Retinoids: acitretin

Sirtuins

Natural antioxidants:

Resveratrol, gingerol, curcumin

Mediterranean diet, caloric restriction

Melatonin

SOX-2 protein

Donepezil

Statins

Physical exercise (BDNF)

Conclusions and future lines of research

The non-amyloidogenic pathway of APP processing and ADAM10 protein, the main α-secretase, offer great opportunities for future research. ADAM10 activity is essential for reducing the generation of Aβ peptides, which has been confirmed both in vitro and in vivo. Genetic studies suggest that some ADAM10 variants predispose to developing AD, and in some families, a direct causal mechanism has been proposed. Without a doubt, this gene should be included in the molecular screening of familial dementias. However, further functional studies are needed to establish correlations between genetic variants, ADAM10 expression, and their impact on APP processing and the formation of amyloid plaques.

Currently, results of ADAM10 quantification in bodily fluids are discordant, possibly because analysis methods are not accurate or reproducible. More precise techniques for differentiating active and inactive ADAM10 species are needed. Platelet ADAM10 levels currently represent the most consistent biomarker.

Lastly, increasing ADAM10 activity is a promising therapeutic target for AD that has received little attention. Some pharmacological and non-pharmacological interventions can favour its transcription or stimulate its activity; these include such a simple measure as physical exercise. However, it would be necessary to selectively regulate its activity towards specific substrates to avoid such adverse effects as tumour dissemination.

Funding

This study has received no specific funding from any public, commercial, or non-profit organisation.