Antisense oligonucleotides could squelch the production of toxic Aβ, suggests a new study led by Michelle Hastings at the Rosalind Franklin University of Medicine and Science in Chicago. The researchers created an ASO that fools the splicing machinery into skipping exon 17 when creating amyloid precursor protein mRNA. Exon 17 encodes all but the first 17 amino acids of the Aβ peptide. After treatment with splice-switching oligonucleotides (SSOs), Aβ levels dropped in fibroblasts from patients with Down's syndrome, and in wild-type mouse brains. The work is described in the June 6 Molecular Therapy.

  • Splice-switching oligonucleotides cut out more than half of the Aβ peptide.
  • The SSOs reduced Aβ42 production in fibroblasts from Down's syndrome patients.
  • Aβ levels in mouse brain dropped after intraventricular injection of an SSO.

“It seems like a logical approach and the study was well done,” said Adrian Krainer, Cold Spring Harbor Laboratory in New York. Krainer co-developed nusinersen, an FDA-approved oligonucleotide that alters mRNA splicing to treat spinal muscular atrophy.

Researchers have long sought to shut off production of toxic Aβ peptides by creating inhibitors of the β- and γ-secretases that process APP. Now first author Jennifer Chang, in collaboration with scientists at Ionis Pharmaceuticals, Carlsbad, California, have targeted APP itself. “The idea of ‘tweaking’ the APP molecule has the potential to overcome some of the side-effects seen with (β-secretase) inhibition, primarily due to the promiscuous nature of the latter,” wrote Jichao Sun and Subhojit Roy, University of Wisconsin, Madison (full comment below).

Aβ Cleavage Spliced Out.

Intraventricular injection of SSO 15-31 causes skipping of APP exon 15, the ortholog of human exon 17, in the cortices of newborn (top, left) and adult (bottom, left) mice. Correspondingly, Aβ42 levels drop in the hippocampi. [Courtesy of Chang et al., Molecular Therapy, 2018.]

Chang’s SSOs result in an APP missing 49 amino acids, including the last 25 of Aβ and its γ-secretase cleavage sites. Like nusinersen, these SSOs are 18-mers, chemically modified with 2’-O-methoxyethyl groups and phosphorothioate bonds to protect them from nuclease degradation and reduce non-specific binding to proteins. Chang tested a series of 36 SSOs that spanned exon 17 and neighboring 5’ and 3’ intron sequences. The most potent, SSO 17-3, had a half-maximal effective dose of around 43 nM in human embryonic kidney cells.

“We focused on SSOs that modulate splicing because they have demonstrated efficacy for treating neurodegeneration in humans, as shown with nusinersen,” said Hastings. “Also, we thought we might retain some of APP’s normal function by altering it without downregulating its expression.”

The researchers tested SSO 17-3 in fibroblasts from patients with Down's syndrome, which carry an extra copy of the APP gene. The oligonucleotide treatment caused Aβ42 in the fibroblast media to drop by 45 percent to a level seen in media from control cells.

To test the approach in mice, Chang designed ASOs to target mouse APP exon 15, the equivalent of the human exon 17. Mouse exon 15 encodes most of the Aβ peptide and its γ-cleavage sites. Using the same oligonucleotide design strategy, the scientists created SSO 15-31, which induces exon-skipping in a dose-dependent manner. The researchers injected 25 or 50 mg or a control SSO into the ventricles of newborn mice. Lo and behold, three weeks later, about 15 percent of APP transcripts in both the cortex and hippocampus were missing exon 15. Immunohistochemistry with an antibody against ASOs modified with 2’-O-methoxyethyl groups revealed that the SSO was distributed widely across the mice’s hippocampus and cortex.

The researchers then treated a second set of mice at birth and waited to allow endogenous Aβ42 to accumulate. The SSOs were still present four months later and APPΔex15 comprised approximately 11 percent of the total APP transcripts in the cortex at that time, indicating a sustained effect. Aβ42 levels dropped by approximately half in the hippocampus. The researchers then repeated the experiment in two-month-old mice. Because these animals were older, they waited only three weeks after treatment to assess both APPΔex15 and Aβ42 levels. The former comprised about 15 percent of total cortical APP mRNA and Aβ42 levels dropped nearly 90 percent in the hippocampus (image above).

Takaomi Saido, RIKEN Brain Science Institute, Wako, Japan, was perplexed by the modest effects of the ASO on splicing versus the robust drop in Aβ42 levels. Hastings speculated that the spliced variant might interfere with normal processing of full-length APP. Krainer said that, given the unknown half-lives of the mRNAs, proteins, and kinetics of their processing, it is hard to know what is really going on. After all, the outcome measurements reported in this initial paper represent but a single snapshots in time. “But clearly things are going in the right direction, and that’s a good sign,” Krainer said.

What about potential side effects? In cell culture experiments, Hastings found APPΔex17 in the media, rather than anchored to the membrane, probably because the spliced isoform lacks the protein’s transmembrane domain. “This is likely to severely impact its normal function, as almost all known physiologic functions of APP are related to its membrane targeting,” wrote Sun and Roy. So far, Hastings has observed no ill effects in mice. SSO-treated animals maintained a normal weight, the astrocytosis marker glial fibrillary acidic protein, and the microgliosis marker allograft inflammatory factor 1, were normal. Hastings agreed that more extensive safety tests are needed.

The next step will be to test the SSOs in mouse models of AD and Down’s syndrome, said Hastings. They plan to use models carrying the full-length human APP gene, monitoring for Aβ42 levels, amyloid plaques, and behavior. The new findings join a growing number of efforts to tackle neurodegenerative disorders, including AD, with ASOs (May 2018 news). “It is a really exciting time, there are a lot of opportunities,” said Hastings.—Marina Chicurel


  1. This is a very interesting study by Jennifer Chang, Michelle Hastings and colleagues, targeting APP using splice-switching antisense oligonucleotides. Traditionally, the field has focused on developing small molecule inhibitors of BACE; and although the biology of APP is relatively well understood, few studies have looked at APP as an alternative target. The idea of “tweaking” the APP molecule has the potential to overcome some of the side effects seen with BACE-1 inhibition, primarily due to the promiscuous nature of the latter.

    The work by Chang et al. is rigorous, and there is no question that the strategy is working. The fact that splice-switching antisense oligonucleotides have recently been approved for the treatment of SMA highlights the overall power of this strategy as a therapeutic tool. One limitation is that this strategy entirely abolishes the membrane-binding of APP and its proteolytic processing. This is likely to severely impact its normal function, as almost all known physiologic functions of APP are related to its membrane-targeting. As the authors’ point out, these approaches will be useful in cases where there is APP over-expression – such as Down’s syndrome or familial APP-duplication – since in these cases, reduction in copy number would restore the protein to normal levels. However, there is no over-expression of APP in the vast majority of Alzheimer’s cases, which are sporadic. Many studies have shown a role of the APP N-terminus in axon growth and signaling, and it is hard to imagine that an absence of a membrane-targeted APP would not negatively impact neuronal physiology.

    The main difference between this study and our recent preprint publication (Sun et al., 2018) is that we used CRISPR/Cas9 to edit the extreme C-terminus of APP, preserving the N-terminus and the membrane-targeting domains. This C-terminus editing specifically manipulates the balance between APP α/β-cleavage. As a result, our strategy inhibits the amyloidogenic (β-cleavage) pathway and enhanced the non-amyloidogenic (protective, α-cleavage) pathway. We also did a battery of physiologic studies and off-target analyses to ensure that the physiology of neurons was not severely disrupted by our approach. Similar experiments are also needed with the antisense oligonucleotide approach. Nevertheless, these recent studies using contemporary genetic tools are an exciting new development in the field, with significant therapeutic potential.

  2. This approach is highly interesting - as it offers a radical way of suppressing the generation of Aβ and thereby reducing its associated neurotoxicity. The decrease of murine Aβ levels upon treatment of wild type mice was surprisingly robust even when the ASOs were delivered into the ventricles of adult animals, which is promising from a therapeutic perspective. Moreover, since it has already been proven feasible to use intrathecally delivered ASOs for the treatment of CNS disorders - as demonstrated by the development of nusinersen against SMN2 for spinal muscular dystrophy - it should be relatively straightforward to take the APP exon skipping ASOs to the clinical trial stage.

    However, potentially negative consequences of excluding exon 17 from APP mRNA will have to be carefully considered. Not only may the generation of a novel form of APP cause unforeseen harm, but also the complete suppression of amyloid-beta could have adverse effects - e.g. by removing a molecule that may have important antimicrobial effects in the CNS.

Make a Comment

To make a comment you must login or register.


News Citations

  1. At AAN, Sights Set on Antisense Therapies for Diseases of the Brain

Further Reading


  1. . Pharmacology of Antisense Drugs. Annu Rev Pharmacol Toxicol. 2017 Jan 6;57:81-105. Epub 2016 Oct 10 PubMed.

Primary Papers

  1. . Targeting Amyloid-β Precursor Protein, APP, Splicing with Antisense Oligonucleotides Reduces Toxic Amyloid-β Production. Mol Ther. 2018 Jun 6;26(6):1539-1551. Epub 2018 Mar 6 PubMed.