A 50-year-old whose Alzheimer’s arose from a deterministic mutation that runs in the family can look quite similar, clinically speaking, to an 80-year-old who developed sporadic AD. Even so, a new study suggests that early and late-onset forms of the disease may bear little resemblance at the molecular level. Researchers silenced late-onset AD risk genes in a human cell model and found this did not influence amyloid-β production in the same way that familial AD mutations do. “While the early onset AD cases are clearly amyloid dependent, our study shows that in late-onset cases, it is not that simple,” said senior author Lawrence Rajendran, University of Zurich, Switzerland. The findings appeared September 4 in the Proceedings of the National Academy of Sciences USA.

In patients and AD transgenic mice, mutations causing early onset AD drive up the Aβ42/Aβ40 ratio by selectively boosting production of Aβ42, the predominant form of Aβ found in plaques (see Borchelt et al., 1996; Duff et al., 1996; Scheuner et al., 1996). These findings on rare familial AD mutations bolstered the amyloid cascade hypothesis postulating that Aβ deposition drives AD pathogenesis early on, which continues to fuel development of many experimental therapies. Much less is known about the genes linked to some 95 percent of AD cases in the late-onset category. “We asked whether these late-onset AD genes act in the same way (as early onset FAD mutations) to change the Aβ42/40 ratio,” Rajendran said.

To monitor the effects of sporadic AD risk genes on amyloid production, first author Jitin Bali and colleagues used RNA interference to knock down individual LOAD genes in a human cell line (HeLa) overexpressing the Swedish mutant of the amyloid precursor protein (APP). Though non-neuronal, the cell model worked for these studies because it churns out reams of Aβ, making it possible to quantitate several relevant species (Aβ38, 40, and 42) in the same cell culture well. Plus, the HeLa cells are the only model thus far that reliably gives 85-90 percent knockdown efficiency, Rajendran said.

His team chose 24 genes from the AlzGene database that recent genomewide association studies (GWAS) had linked to LOAD. With RNAi, they silenced each gene in the HeLa model. As controls, the scientists introduced RNAs knocking down amyloid precursor protein (APP), β-secretase (BACE1), the γ-secretase component Pen2, and 10 genes that are implicated in other neurodegenerative diseases but have no known polymorphisms linked to AD. By and large, knockdown of the LOAD genes did not increase the Aβ42/40 ratio. Curiously, silencing the LOAD genes did raise Aβ42 levels in 17 of 24 cases. However, Aβ42 also went up when the control genes were silenced, and knocking down many of the LOAD and control genes drove up Aβ40 levels as well. Overall, LOAD genes revealed no specific effect on Aβ42 levels or the Aβ42/40 ratio.

In a separate set of experiments that allowed for potential epistatic interactions by silencing different combinations of genes in the same cells, LOAD genes still had no effect on Aβ42/40 ratios.

“I think this is a very important and creative approach to testing the biological significance of GWAS hits. More studies like this are needed so the AD field can home in on the genes that really matter,” John Trojanowski, University of Pennsylvania School of Medicine, Philadelphia, wrote in an e-mail to Alzforum (see also ARF Webinar). Likewise, Alison Goate of Washington University School of Medicine, St. Louis, Missouri, found the approach “interesting and novel,” noting that few studies have specifically looked at the effects of sporadic AD risk genes.

The present analysis also had shortcomings, some methodological. The RNAi experiments did not control for off-target effects, noted Todd Golde of the University of Florida, Gainesville. Such problems—which occur when an introduced RNA perturbs expression of other genes besides the intended target—have been estimated to affect about 10 percent of silencing RNAs (Qiu et al., 2005).

The experimental setup had other limitations. It only tested for loss-of-function effects, yet “we do not know how late-onset AD susceptibility genes affect AD risk, and it could well be that some of them act by gain of function,” noted Bart De Strooper of KU Leuven, Belgium. “Moreover, the Aβ42/40 ratio is only one aspect of Aβ metabolism,” said De Strooper. Furthermore, Bali and colleagues only measured Aβ production, not clearance, Goate said, even though a recent analysis of Aβ dynamics in sporadic AD patients suggests the problem is more the latter (see ARF related news story).

Another sticking point was the use of a cell model that overexpresses APP with an FAD mutation (APPSwe), while the vast majority of LOAD patients have wild-type APP. Rajendran said that since these assays require robust amounts of Aβ40 and 42 for rigorous statistics, they needed to use this model. He also noted that the Swedish mutation only increases BACE1 cleavage of APP, not γ cleavage, and should not affect the Aβ42/40 ratio. His team is currently repeating the silencing experiments in wild-type mouse neurons. The mouse cells produce normal APP—not whopping levels of mutant APP as in the HeLa model—and have proven much harder to optimize for RNAi knockdown. “We have worked for about a year on this. So far, the best we get is around 85 percent efficiency,” Rajendran told Alzforum.

The current data leave open the possibility that GWAS genes might still play a role in amyloid-induced toxicity or neuroinflammation, despite their lack of effect on Aβ42/40 ratios, Rajendran said. Plus, he noted, “our results caution that current therapeutic strategies aimed solely at removing or inhibiting amyloid production might not be fully effective against late-onset AD.”—Esther Landhuis

Comments

Make a Comment

To make a comment you must login or register.

Comments on this content

  1. This study by Bali et al. provides compelling evidence that downregulation of the late-onset AD (LOAD) risk genes in cell cultures does not alter Aβ generation in a pathologically meaningful way, and suggests that increased Aβ42 levels or Aβ42/40 ratio may not be the driving force of disease pathogenesis in the majority of AD cases (i.e., LOAD cases). Limitations of their experimental approach notwithstanding, it is difficult to argue against the overarching implications of their findings.

    The limitations pointed out by others are valid but irrelevant, or not significant to the authors’ conclusions. RNAi methodology can have off-target effects, but it is almost impossible that reagents used to silent all 24 genes would have similar off-target effects. True, some of the AD risk genes may act by gain of function, but to imagine that all 24 genes do so is highly improbable. To affect Aβ clearance, the AD risk genes must fall in one of three categories: proteases that degrade Aβ, regulators of macrophage-mediated Aβ uptake, or be localized to the blood-brain barrier and mediate Aβ export. Again, it is difficult to see how all 24 AD risk genes examined here fall into one of the three categories.

    Finally, the argument that the use of tissue culture cells overexpressing APPswe mutation somehow invalidates their conclusions (since the vast majority of LOAD cases have wild-type APP) would also invalidate a very large body of published work.

    Thus, if one accepts their conclusion that the 24 AD risk genes studied here do not alter Aβ production in a meaningful way and, as argued above, are unlikely to regulate Aβ clearance, then one must entertain the possibility that some of the AD risk genes act by a mechanism that does not involve Aβ metabolism. Overall, these findings are consistent with the view that AD pathogenesis is a multifactorial process and that not all cases of AD may be explained by the amyloid hypothesis (Pimplikar et al., 2010). The preliminary reports that nasal insulin therapy or intravenous immunoglobulin administration seems to confer beneficial effects in clinical studies supports such a view.
      

    References:

    . Amyloid-independent mechanisms in Alzheimer's disease pathogenesis. J Neurosci. 2010 Nov 10;30(45):14946-54. PubMed.

References

Webinar Citations

  1. Weeding Mendel’s Garden: Can We Hoe Dubious Genetic Associations?

News Citations

  1. Paper Alert: In Vivo Human Data Shows Reduced Aβ Clearance in AD

Paper Citations

  1. . Familial Alzheimer's disease-linked presenilin 1 variants elevate Abeta1-42/1-40 ratio in vitro and in vivo. Neuron. 1996 Nov;17(5):1005-13. PubMed.
  2. . Increased amyloid-beta42(43) in brains of mice expressing mutant presenilin 1. Nature. 1996 Oct 24;383(6602):710-3. PubMed.
  3. . Secreted amyloid beta-protein similar to that in the senile plaques of Alzheimer's disease is increased in vivo by the presenilin 1 and 2 and APP mutations linked to familial Alzheimer's disease. Nat Med. 1996 Aug;2(8):864-70. PubMed.
  4. . A computational study of off-target effects of RNA interference. Nucleic Acids Res. 2005;33(6):1834-47. PubMed.

Other Citations

  1. APPSwe

External Citations

  1. AlzGene database

Further Reading

Papers

  1. . Decreased clearance of CNS beta-amyloid in Alzheimer's disease. Science. 2010 Dec 24;330(6012):1774. PubMed.

Primary Papers

  1. . Role of genes linked to sporadic Alzheimer's disease risk in the production of β-amyloid peptides. Proc Natl Acad Sci U S A. 2012 Sep 18;109(38):15307-11. PubMed.