On February 25, the Journal of Neuroscience quietly posted on its website its retraction of a 2011 paper that questioned the existence of Aβ aggregates inside neurons. In a terse statement, the journal, a publication of the Society for Neuroscience (SfN), cited “the journal’s findings of data misrepresentation” as the reason, and noted that it no longer considered the results of the study to be reliable. Unlike past retractions in the Alzheimer’s field, in this case the journal and the authors are publicly at odds. Senior authors Virginia Lee and John Trojanowski at the University of Pennsylvania Perelman School of Medicine in Philadelphia acknowledge mistakes in two figures. However, they assert that retraction was not necessary, as they believe the conclusions of the paper remain sound. An inquiry by UPenn supported their view. In addition, the authors said they have been banned from publishing in the journal for several years.
This issue has received coverage online, including in Retraction Watch, the blog DrugMonkey, and the Philadelphia Inquirer. Alzforum held off covering the retraction to give the Journal time to comment. For the first few days, representatives of J.Neurosci and SfN flatly declined, citing confidentiality, but later requested time to issue a general statement on standards of accuracy in the scientific literature, which was posted March 4.
At the time of its publication, the paper set off intense debate with its claim that commonly used anti-Aβ antibodies detected amyloid precursor protein (APP), rather than Aβ, inside neurons of 3xTg AD mice. In essence, the authors asserted that the finding of intraneuronal Aβ had been an artifact. The result challenged a previous paper by Frank LaFerla and colleagues at the University of California, Irvine, which had reported extensive intraneuronal Aβ accumulation in these mice (see Oddo et al., 2003, and Nov 2002 Webinar). Alzforum covered the issue in a June 2011 Webinar, which featured vigorous debate and was followed by spirited written commentary for some weeks.
Has time and more research told whether the paper was right? Not really. While some studies published since 2011 agree with Lee and Trojanowski that APP, or longer fragments of it, makes up the bulk of intraneuronal amyloid staining, most papers also report at least low levels of Aβ inside cells. Other recent papers flag intraneuronal Aβ as a key culprit in damaging neurons and synapses. “I think most researchers acknowledge the existence of intraneuronal Aβ aggregates as a pathologic feature, but still question its pathogenic relevance,” Lars Nilsson at Oslo University, Norway, wrote to Alzforum.
Now the retraction of the 2011 paper adds another wrinkle to an already complicated story. It is unclear what first drew the journal’s scrutiny. Citing confidentiality, SfN and journal staff declined to say whether they routinely check for image manipulation, or whether someone inside or outside the journal raised a flag. Some journals use automated software to detect problems with figures. Such an analysis likely would have turned up the errors. In this paper, they amounted to three instances of a duplicated image being used to represent different data. Specifically, panels E and F of Figure 1, and panels E and H, as well as F and G, in Figure 4, depict the same microscopic field, in some cases rotated or shown at different magnifications, even though the captions identified the images as coming from mice of different ages and showing progression. Lee said she first became aware of the issue in February 2014, when she received a letter from the SfN Ethics Committee pointing out the image duplications.
The committee also informed the Perelman School of Medicine. The school conducted a preliminary inquiry, which is standard procedure in such cases, executive vice dean Glen Gaulton told Alzforum. The inquiry committee examined lab notebooks and raw data and interviewed the authors, including first author Matthew Winton, who left research in 2008 and now works as a financial consultant for a pharmaceutical company in the Boston area. According to Gaulton, Winton acknowledged making mistakes in image selection during revisions on the paper. Gaulton noted that an earlier, draft version of the paper contained the correct images. Winton did not respond to Alzforum’s emailed requests for comment. The UPenn committee concluded that the mistakes were unintentional and did not alter the findings of the paper. Gaulton said they reported this to the SfN Ethics Committee in a letter published by Retraction Watch. Lee said she offered to submit a correction to the journal, but was informed last month that it would retract the paper instead.
While SfN’s statement does not directly discuss the incident, it notes that “[SfN policies] address the ethical responsibility of principal investigators, among all authors; highlight that violations may or may not involve any intent to mislead; and encourage the practice of proactive and prompt correction or retraction by authors when an issue becomes evident.” Peggy Mason of the University of Chicago chairs SfN’s ethics committee. She shared her views on data integrity with Alzforum, though she stressed that her opinions do not necessarily represent those of the society. “The ethics committee’s primary goal is to defend the scientific literature. If there are errors in the details of a paper that make it no longer trustable, it needs to be retracted,” Mason said. She noted that this is true regardless of whether or not the errors were intentional. In cases of a single small error, a correction might be appropriate instead, she added. Mason said that ethics committees have a secondary mission to try to prevent the recurrence of problems; to accomplish this, these institutions sometimes impose punitive measures, such as publishing bans.
Researchers in the field expressed surprise at the events and the penalty. They agreed the mistakes were unlikely to be intentional. “If you wanted to misrepresent results, you wouldn’t do it by duplicating figures,” Charlie Glabe at UC Irvine wrote in an email. Nilsson concurred that the errors leave the paper’s conclusions intact, but noted that retraction could still be justified. “As a reader, you need to trust that images shown are correct as described,” he wrote to Alzforum.
What about the central claim that neurons in the triple transgenic mice accumulate APP, not Aβ? Of this, scientists were more skeptical. They note that it is difficult to prove something is not there, because it is always possible the methods used were not sensitive enough to detect it. Several groups since have analyzed the triple transgenic mice and similar models using a variety of methods, and reported the presence of at least some intraneuronal Aβ. For example, Mary Jo LaDu and Leon Tai at the University of Illinois, Chicago, developed the MOAB-2 antibody, which they report recognizes the N-terminal end of Aβ and does not cross-react with APP. This antibody lit up neurons in both triple transgenic and 5xFAD mice brains, they found (see Youmans et al., 2012). However, common anti-Aβ antibodies such as 6E10 and 4G8 cross-react with APP and should not be used to measure Aβ, LaDu told Alzforum.
Likewise, Thomas Bayer and Oliver Wirths at Georg-August-University Göttingen, Germany, used conformation-specific antibodies to detect intraneuronal Aβ in the triple transgenics, although they report that most of it occurs in the cortex, with low levels in hippocampus (see Wirths et al., 2012). Researchers led by Claudio Cuello at McGill University, Montreal, took a different tack, employing high-resolution microscopy, mass spectrometry analysis, and ELISAs to determine that Aβ accumulates inside neurons of a transgenic rat model (see Iulita et al., 2014).
While this recent work contradicts the contention that Aβ does not accumulate inside neurons, several of the papers, including those of Bayer and Wirths and LaDu and Tai, did replicate Lee and Trojanowski’s finding of large amounts of intraneuronal APP in the triple transgenics. Researchers led by Frédéric Checler at the CNRS in Valbonne, France, added a different twist, suggesting that most of the intraneuronal amyloid immunoreactivity in this model comes from the C99 fragment of APP. Intraneuronal Aβ appears only late in life in the triples, the authors reported (see Lauritzen et al., 2012).
Bayer and Wirths summarized the findings from the 3xTg mice in an email to Alzforum, writing, “The main immunostaining detected by pan-Aβ antibodies is due to APP cross-reactivity. The age-dependent APP accumulation in the 3xTg model is unique and has not been observed in other APP transgenic mouse models, at least to our knowledge.” (See full comment below.)
This intraneuronal APP may contribute to pathology, Glabe noted. “It is pathologically misfolded, insoluble and aggregated, in the same fashion as it ends up in plaques,” he wrote. “You see the same type of intraneuronal immunoreactivity in human brain. It just means that not all of the amyloid that accumulates comes from the BACE1/γ-secretase pathway that makes secreted Aβ40 and Aβ42.”
Human studies have been slow to add evidence for a role for intraneuronal Aβ. In the March 1 Brain, researchers led by Changiz Geula at Northwestern University, Chicago, report seeing Aβ inside basal forebrain cholinergic neurons in people even as young as 20. While its quantity stayed unchanged with age or Alzheimer’s disease, more of it became aggregated into oligomers and larger species, they believe. Because other types of neurons did not accumulate Aβ, the finding might provide a clue to the selective degeneration of cholinergic neurons in AD, the authors suggest (see Baker-Nigh et al., 2015).
Gunnar Gouras at Lund University, Sweden, told Alzforum, “The Winton et al. paper certainly put a damper on the topic of intraneuronal Aβ, but it did not stop research in this area, which has grown at conferences and in papers.”—Madolyn Bowman Rogers
Research Models Citations
- Intraneuronal Aβ Accumulation—More Evidence, Less Controversy?
- Intraneuronal Aβ: Was It APP All Along?
- Winton MJ, Lee EB, Sun E, Wong MM, Leight S, Zhang B, Trojanowski JQ, Lee VM. Intraneuronal APP, not free Aβ peptides in 3xTg-AD mice: implications for tau versus Aβ-mediated Alzheimer neurodegeneration. J Neurosci. 2011 May 25;31(21):7691-9. PubMed.
- Oddo S, Caccamo A, Shepherd JD, Murphy MP, Golde TE, Kayed R, Metherate R, Mattson MP, Akbari Y, LaFerla FM. Triple-transgenic model of Alzheimer's disease with plaques and tangles: intracellular Abeta and synaptic dysfunction. Neuron. 2003 Jul 31;39(3):409-21. PubMed.
- Youmans KL, Tai LM, Kanekiyo T, Stine WB, Michon SC, Nwabuisi-Heath E, Manelli AM, Fu Y, Riordan S, Eimer WA, Binder L, Bu G, Yu C, Hartley DM, LaDu MJ. Intraneuronal Aβ detection in 5xFAD mice by a new Aβ-specific antibody. Mol Neurodegener. 2012;7:8. PubMed.
- Wirths O, Dins A, Bayer TA. AβPP Accumulation and/or Intraneuronal Amyloid-β Accumulation? The 3xTg-AD Mouse Model Revisited. J Alzheimers Dis. 2012 Jan 1;28(4):897-904. PubMed.
- Iulita MF, Allard S, Richter L, Munter LM, Ducatenzeiler A, Weise C, Do Carmo S, Klein WL, Multhaup G, Cuello AC. Intracellular Aβ pathology and early cognitive impairments in a transgenic rat overexpressing human amyloid precursor protein: a multidimensional study. Acta Neuropathol Commun. 2014 Jun 5;2:61. PubMed.
- Lauritzen I, Pardossi-Piquard R, Bauer C, Brigham E, Abraham JD, Ranaldi S, Fraser P, St-George-Hyslop P, Le Thuc O, Espin V, Chami L, Dunys J, Checler F. The β-secretase-derived C-terminal fragment of βAPP, C99, but not Aβ, is a key contributor to early intraneuronal lesions in triple-transgenic mouse hippocampus. J Neurosci. 2012 Nov 14;32(46):16243-55a. PubMed.
- Baker-Nigh A, Vahedi S, Davis EG, Weintraub S, Bigio EH, Klein WL, Geula C. Neuronal amyloid-β accumulation within cholinergic basal forebrain in ageing and Alzheimer's disease. Brain. 2015 Jun;138(Pt 6):1722-37. Epub 2015 Mar 1 PubMed.
- Watch Aβ Aggregate in Real Time Inside Cell’s Acid Vesicles
- SORLA Serves Up Aβ for Destruction
- Like Star Born of Supernova, Plaque Born of Exploded Neuron?
- Intracellular Aβ Causes Neurodegeneration in Mice
- Monomeric Aβ’s Disappearing Act in AD Brain: Two Theories
- SfN: Where, How Does Intraneuronal Aβ Pack Its Punch? Part 1