A widely cited 2007 Nature paper about the epigenetics of memory has earned an editorial expression of concern over apparent image manipulation.
The paper’s corresponding author, Li-Huei Tsai, is director of the Picower Institute of Learning and Memory at the Massachusetts Institute of Technology (MIT). She and her colleagues have corrected duplicated images in three other papers this year.
The 2007 study investigated how chromatin alterations affected memory in a mouse model of neurodegeneration that the Tsai Lab created. The authors reported that environmental enrichment or treatment with sodium butyrate, a histone deacetylase inhibitor, correlated with increased chromatin modifications and memory improvement in their mice.
The extent of memory recovery in a model with such extensive neuron and synapse loss was “striking,” says Isabelle Mansuy, who leads a neuroepigenetics lab at the University of Zürich. At the time the study was published, plasticity research was focused on the synapse, so the close association between chromatin dynamics and cognitive performance “was quite innovative conceptually,” says Mansuy, who has cited the paper multiple times. Overall, the study has been cited more than 1,100 times, per Clarivate’s Web of Science.
Five sets of images in this paper appear to be duplicates, though the key findings are still supported by the remaining data and subsequent studies, according to the expression of concern posted on 10 September. All of the paper’s authors agree with the journal’s course of action, which was determined by “the age of the article and the fact that the authors no longer have some of the original raw data,” a spokesperson for Nature told The Transmitter in an email. The spokesperson added that nowadays, the journal uses an artificial-intelligence tool to screen images from all life sciences manuscripts.
There are plenty of opportunities for authors to unknowingly introduce errors along the multistep publication process, Mansuy says. Still, she says she feels Nature’s statement is “of great concern” if the journal’s findings are accurate. “These mistakes are repetitive and leave little room for just chance.”
In response to The Transmitter’s request for more information, a press officer at MIT reiterated the journal’s statement, emphasizing that the concerns did not undermine the core conclusions of the paper, but did not offer further information. Neither Tsai nor the paper’s first author André Fischer, now professor at the German Center for Neurodegenerative Diseases, responded to a request for comment.
The 2007 paper is not the only publication from Tsai’s lab to receive recent scrutiny. The same day Nature posted its expression of concern, Neuron issued a correction to a duplicated image in a paper from 2022. In August, the Tsai Lab amended images in a 2015 paper in the Proceedings of the National Academy of Sciences. And in May, Neuron posted a correction to a 2011 paper from the lab, eight years after Tsai posted on PubPeer that she was working with the journal to address image duplication concerns.
The Transmitter has previously covered other work from the Tsai Lab attempting to demonstrate that induced gamma oscillations could clear brain amyloid; one of those papers was corrected in 2018.
T
he 2007 paper first attracted attention on PubPeer in February 2017, when an anonymous commenter noted possible discrepancies in Western blots illustrating hippocampal histone acetylation; this image was ultimately not listed in the expression of concern.
This February, questions about another figure were raised by data sleuth Kevin Patrick, who posts on PubPeer under the pseudonym Actinopolyspora biskrensis. In his comment, he identified an apparent duplication of Western blot images in two supplementary figures involving different experimental conditions.
One of the duplicate images had been cropped and rotated, suggesting this wasn’t an instance of the authors carelessly uploading the same file twice, Patrick told The Transmitter. “That doesn’t happen by accident,” he says, adding that it is impossible for him to judge whether misconduct had occurred.
Patrick posted his thoughts about the figures on PubPeer and alerted Nature to his findings. Tsai responded to Patrick’s inquiry on PubPeer a few days later. “We are actively searching for the original blots to resolve this issue and ensure accuracy,” she wrote. In April, Nature posted an update indicating that the journal was aware of questions about the paper.
Other than an acknowledgement that the editors had received his report, Patrick says he did not hear back from Nature and only became aware of the journal’s course of action when The Transmitter contacted him 11 September for comment. Looking at the published notice, Patrick says he was surprised to discover that Nature editors had flagged more apparent manipulations than the one he had reported. “I’m not quite sure how I missed them,” he says, adding that he agrees with the journal about the other images.
Patrick says he first noticed the potential duplication in 2023 as he and other sleuths scoured hundreds of Alzheimer’s disease papers for problematic images. This work was included in journalist Charles Piller’s book “Doctored: Fraud, Arrogance, and Tragedy in the Quest to Cure Alzheimer’s,” in which Tsai’s paper was counted among instances of potential data manipulation.
“Sadly, there seems to be very little effort from publishers, journals, or institutions, to go back and clean up past problems, even when they know exactly where they are,” Patrick wrote in an email to The Transmitter.