More than two dozen potential Alzheimer's risk factor loci have popped up in GWAS studies so far, but mutations and functional variants in only four genes have ever been pathologically validated for AD—APP, PSEN1, PSEN2, and ApoE. Now, Rudy Tanzi and colleagues from Massachusetts General Hospital in Charlestown claim that a fifth belongs on this list. It is ADAM10, the gene for the main α-secretase that cleaves the amyloid precursor protein (APP) in a non-amyloidogenic fashion. Tanzi’s team previously found two rare mutations in ADAM10 that associate with familial late-onset Alzheimer's disease (LOAD) (see ARF related news story). In a subsequent paper published online September 19 in Neuron, the researchers report murine evidence that these mutations shift APP processing toward β-secretase cleavage, enhance Aβ plaque load and reactive gliosis, and impair neurogenesis.

"These are the first highly penetrant mutations to be associated with late-onset Alzheimer's disease," said Robert Vassar, Northwestern University, Chicago, Illinois, who penned an accompanying preview to appear with the article in the October 16 print version of Neuron. "It is paradigm shifting for the field, because it now makes a direct genetic link between Aβ metabolism and late- onset Alzheimer's disease," he said. Many researchers have argued that while genetic evidence supports the amyloid cascade hypothesis for early onset AD, it does not do so for LOAD. "Those people are going to have a harder time arguing that point with these ADAM10 mutations," Vassar said.

The MGH researchers first discovered the ADAM10 mutations when they went looking for potential AD-associated variants in biologically relevant genes. ADAM 10, which stands for A Disintegrin and Metalloproteinase 10 (ADAM10) cuts APP in the middle of the protein's Aβ region, avoiding creation of the sticky, plaque-forming peptide. The group wondered if mutations hindering the enzyme's activity might lead to AD. After sequencing ADAM10 in the National Institute of Mental Health AD Genetics Initiative families, the scientists found the Q170H and R181G mutations in seven families with LOAD. In cell culture studies, these mutations knocked down ADAM10 activity by two thirds and led to higher Aβ levels relative to the wild-type version. Both mutations appear in the protein's prodomain, a region that gets cleaved off to turn ADAM10 into an active enzyme. The prodomain also helps the protein to fold properly. In this new study, the scientists sought to demonstrate the mutations' effects in vivo.

First author Jaehong Suh and colleagues generated transgenic mice that express either wild-type human ADAM10 or one of the two LOAD mutant forms. The latter produced more products of β−cleavage of APP, fewer products of α-cleavage, and more Aβ in their brains at 12 weeks of age compared to mice expressing wild-type ADAM10. The enzyme's function did not disappear completely in the mutants; rather, that authors report that the results suggest a relative shift from α to β-secretase activity. The ADAM10 mutant mice also showed reduced proliferation and differentiation of newborn neurons in the hippocampus compared to mice that carried wild-type ADAM10.

These mice do not actually model AD, developing neither plaques nor synaptic or behavioral deficits with age. Therefore the researchers crossed them with an AD mouse model to see how the ADAM10 mutations would affect pathogenesis.

The Tg2576 mouse strain overexpresses human APP with the Swedish mutation that leads to early onset AD by increasing β-secretase cleavage. The double-transgenic mutant ADAM 10 mice demonstrated even more β-cleavage and more Aβ in the brain than Tg2576 mice expressing human wild-type ADAM10. By 12 months, mice expressing mutant ADAM10 had a heavier plaque load, as well. As in the single-transgenic ADAM10 mice, enzyme activity did not plummet completely, but dropped somewhat. When the mice were 18 to 20 months old, reactive gliosis was more common in the double-mutant carriers. Together, these results suggest to the authors that these ADAM10 variants lead to reduced function and some of the phenotypes associated with AD.

How do the mutations affect ADAM10 activity? Experiments in the current study suggested they exert their effects before prodomain cleavage, while this region is still attached to the immature protein. At this point in protein maturation, the prodomain aids in ADAM10 folding. To test whether mutations affect that function, the scientists applied just the prodomain of wild-type or mutant ADAM10 protein to APP-overexpressing neuroblastoma cells that produced inactive ADAM10 lacking a prodomain. Only the wild-type version of the ADAM10 prodomain restored enzymatic function. This suggests that the LOAD mutations impair the ability of the prodomain to act as a chaperone for ADAM10 folding, Tanzi said.

"This paper clearly demonstrates these ADAM10 missense mutations cause partial loss of function with respect to APP processing," said Stefan Lichtenthaler, German Center for Neurodegenerative Diseases, Munich. He suggested that researchers next check the cerebrospinal fluid and fibroblasts of people carrying ADAM10 mutations and non-carrying relatives to see if they show similar alterations in APP processing. Would the same profile appear even though these people do not overexpress mutant APP? With CSF, this question could be addressed in a natural model without the complications of double transgenic technology. Lichtenthaler also pointed out that more work needs to be done to work out the relevant mechanism.

The results further suggest that ADAM10 could be a therapeutic target, Tanzi said. Retinoic acid or inhibitors of endogenous ADAM10 suppressors might boost the protein's activity and shift processing away from amyloid production, while improving neurogenesis, the authors wrote. One such compound, acitretin, entered a Phase 2, but no results have been reported .

Not everyone was convinced that these mutations alter ADAM10 function. For instance, Rolf Postina and Elzbieta Kojro of Johannes Gutenberg University, Mainz, Germany, pointed out that the increase in plaque load in the Tg2576 mice expressing a mutant form of ADAM10 was small. Such modest differences could also result from simple differences in the amount of ADAM10 expression, and are not necessarily defects in the function of the enzyme, they wrote to Alzforum in an email. Tanzi countered that his group carefully chose mice that expressed similar levels of ADAM10, and that the differences were larger for total Aβ levels. Even small differences in enzyme function could tip the balance of Aβ processing and lead to late-onset Alzheimer's disease, he said. That might explain why these mutations are associated with disease that manifests clinically in old age.

Tanzi said his group will soon report more AD-associated mutations in ADAM10 and explore the protease's other physiological substrates to check its feasibility as a drug target. The researchers are also preparing to validate other GWAS hits such as BIN1, CD33, and TREM2 in a similar way: by searching for the variants associated with AD, testing their effects in cell culture, and finding out what they do in animals. "This is a model for going beyond GWAS and entering the era of whole genome sequencing," Tanzi told Alzforum. The strategy applies not just to confirmed GWAS hits, but also to genes with rare but highly penetrant mutations. ADAM10 was one such "little guy" that turns out to contribute to AD genetics, he said. "Now that we see in vivo that ADAM10 mutations are pathogenic, this will hopefully convince the field that this is the fifth AD gene to have mutations that are detrimental in animal models," Tanzi told Alzforum.—Gwyneth Dickey Zakaib


  1. The human ADAM10 mutations Q170H and R181G within the prodomain of the protein have been previously found in several late-onset AD families (Kim et al., 2009). To validate the pathogenicity of the mutations, classical transgenic mice expressing wild type or the LOAD mutations were generated and characterized. It was found that the LOAD mutations attenuate (but do not abolish) ADAM10 alpha-secretase activity, and increase beta-secretase cleavage of APP. Compared to the wild-type situation, the mutants increase Abeta accumulation in AD mouse brains.

    The study is based on dominant-negative effect of ADAM10 expression on endogenous ADAM10 in mice, as has been shown earlier by Postina et al. some years ago (Postina et al, 2004). It again confirms the essential role of ADAM10 (and not of any other suspected ADAM protease) for alpha secretase processing of APP. This highlights its role in human AD patients, urging for more intensive screening for additional ADAM10 mutations.

    Surprisingly, the dominant-negative effects of the transgenic mutation were apparently only affecting the APP substrate, whereas other neuronal substrates such as N-cadherin or Notch appear to be unaffected by these mutations. Perhaps knockin mice mimicking the human pro domain mutations would have been even more convincing.

    The role of ADAM10 in neurogenesis in the hippocampus is interesting but certainly needs additional efforts to understand this at a molecular level. The role of ADAM10 in the adult CNS, including its role in APP processing and endogenous Abeta generation has also been revealed by our group (Prox et al. JN 2013).

    In summary, this is an elegant study providing additional and valuable insight into the function of previously described ADAM10 mutations leading to AD.


    . Potential late-onset Alzheimer's disease-associated mutations in the ADAM10 gene attenuate {alpha}-secretase activity. Hum Mol Genet. 2009 Oct 15;18(20):3987-96. PubMed.

    . A disintegrin-metalloproteinase prevents amyloid plaque formation and hippocampal defects in an Alzheimer disease mouse model. J Clin Invest. 2004 May;113(10):1456-64. PubMed.

    . Postnatal Disruption of the Disintegrin/Metalloproteinase ADAM10 in Brain Causes Epileptic Seizures, Learning Deficits, Altered Spine Morphology, and Defective Synaptic Functions. J Neurosci. 2013 Aug 7;33(32):12915-28, 12928a. PubMed.

  2. We would like to address the concern of Drs. Postina and Kojro, who questioned whether increased Aβ levels in the ADAM10 X Tg2576 mouse brains could be due to differences in ADAM10 expression versus effects of the ADAM10 LOAD mutations. We refer to Figure 1, which clearly shows very similar levels of ADAM10 expression among the three lines that were carefully chosen for wild-type and each ADAM10 mutation. In that same figure, auto-proteolysis of ADAM10 to generate the ADAM10-CTF is clearly impaired by both LOAD mutations. This would not be due to differences in expression levels.

    In these same lines, one can also observe a shift of APP from α- to β-secretase cleavage for both mutations vs. wild-type, with similar levels of ADAM10 expression (Figures 1 and 2). The observation of this shift in the single ADAM10 transgenics (Figure 1), in the absence of mutant APP, also addresses the question of whether the same profile would appear in those who do not over-express mutant APP?

    Finally, we note that the observed moderate (statistically significant) increases in total Aβ (Figure 3) owing to the ADAM10 mutations would be consistent with the average age of onset of 70 years in the seven carrier AD families.

Make a Comment

To make a comment you must login or register.


News Citations

  1. More AD Genetics—A Familial Hit from the ADAMs Family

External Citations

  1. Phase 2

Further Reading


  1. . Potential late-onset Alzheimer's disease-associated mutations in the ADAM10 gene attenuate {alpha}-secretase activity. Hum Mol Genet. 2009 Oct 15;18(20):3987-96. PubMed.
  2. . ADAM10 in Synaptic Physiology and Pathology. Neurodegener Dis. 2013 Sep 4; PubMed.
  3. . Platelet a disintegrin and metallopeptidase 10 expression correlates with clock drawing test scores in Alzheimer's disease. Int J Geriatr Psychiatry. 2013 Aug 22; PubMed.
  4. . Postnatal Disruption of the Disintegrin/Metalloproteinase ADAM10 in Brain Causes Epileptic Seizures, Learning Deficits, Altered Spine Morphology, and Defective Synaptic Functions. J Neurosci. 2013 Aug 7;33(32):12915-28, 12928a. PubMed.

Primary Papers

  1. . ADAM10 Missense Mutations Potentiate β-Amyloid Accumulation by Impairing Prodomain Chaperone Function. Neuron. 2013 Oct 16;80(2):385-401. PubMed.