PLOS One Retracts Lung Fibroblast Antioxidant Study After Repeated Image Overlap Concerns
PLOS One has retracted a 2014 paper that claimed a bacterial fucose-rich polysaccharide could stabilize MAPK-mediated Nrf2/Keap1 signaling and blunt hydrogen peroxide-driven apoptosis in human lung fibroblast cells. The retraction rests on a cluster of image integrity concerns spanning multiple figures, including panels that appeared similar, lanes that appeared more similar than independent experimental results should allow, and panels that partially overlapped across different datasets.
The paper sat at the intersection of oxidative stress biology and cell survival signaling, a familiar and clinically relevant framework in lung injury research. Hydrogen peroxide exposure is often used to model oxidative damage in vitro, while the Nrf2/Keap1 axis and MAPK pathways such as p38 and JNK are central to the cellular response against reactive oxygen species. When image-based evidence for those pathways becomes unreliable, confidence in the mechanistic story falls quickly.
PLOS identified concerns in figures 4, 6, 8, 9, 11 through 13, and S3. In figure 4, the H2O2 300 µM HFC 50 µg/ml 24 h panel appeared similar to the H2O2 300 µM HFC 100 µg/ml 6 h panel. In figure 6, the FACS results for H2O2 300 µM 3 h and H2O2 300 µM + HFC 250 µg/ml 6 h appeared more similar than expected from independent samples, as did the H2O2 300 µM + HFC 250 µg/ml 3 h and 12 h panels. Those kinds of similarities matter because flow cytometry readouts are supposed to capture distinct biological states, not recycled visual evidence.
The scrutiny did not stop there.
Further concerns involved lanes that appeared too similar across western blot panels. Figure 8B, figure 9A, figure 12A, figure 13A, and figure S3 all drew attention for patterns suggesting reuse, flipping, or overlap between lanes representing different proteins, cellular compartments, or treatment conditions. Figure 8B bad lane 3, when flipped, appeared similar to bad lane 5 and to the right Nrf2 cytosol lane 4 in figure 13A. Figure 9A cytosol CytC lanes 1 through 3 appeared similar to figure 12A nucleus Nrf2 lanes 1 through 4. Figure 13A p-JNK lanes overlapped visually with p-38 lanes, and figure S3 showed similarity between p-p38 and p-ERK lanes. Western blots depend on separation of signal, not visual recycling; once overlap appears, the molecular readout loses credibility.
Several panels also appeared to partially overlap. Figure 11A active Caspase-3 overlapped with image material from a 2014 PLoS One paper later retracted, and with phospho-p38 and phospho-JNK panels from another retracted PLoS One study. The four right-hand control panels in figure 12 and the four right-hand H2O2 300 µM + HFC 250 µg/ml 24 h panels also appeared to overlap. In practical terms, those are not isolated cosmetic issues. They strike at the core of whether the reported apoptotic and signaling responses were generated from distinct experiments at all.
The editorial decision was made without author resolution. The authors did not respond to the concerns. One author, SRC, did not agree with the retraction, while SS, SB, TKS, RS, RKB, BA, and AB either did not respond directly or could not be reached. PLOS also acknowledged additional overlap concerns raised on PubPeer involving the figure 11A active Caspase-7 panel in the retracted paper and panels in another article, but those concerns were not validated further because the referenced paper could not be accessed.
For readers of the original study, the retraction changes the evidentiary status of a once-neat mechanistic claim. A bacterial polysaccharide was said to scavenge reactive oxygen species directly and to preserve cell survival signaling under oxidative stress, but the figure set supporting that narrative can no longer be treated as dependable. In cell biology, the mechanistic chain is only as persuasive as the integrity of the images that document it: ROS reduction, MAPK modulation, Nrf2 redistribution, and caspase activation all depend on accurate, non-duplicated experimental records.
The retraction also illustrates how image concerns travel across a paper in layers. A single duplicated panel can raise suspicion; repeated similarity across FACS plots, blots, and apoptosis readouts turns suspicion into a broader reliability problem. Once one figure appears to echo another, readers are forced to ask whether the same experimental material was being used to support multiple claims about oxidative injury, mitochondrial stress, and death signaling.
The subject matter of the study remains biologically plausible. Human lung fibroblasts respond to oxidant injury through pathways that converge on apoptosis, and interventions that modulate Nrf2 or MAPKs can, in principle, alter cell fate. But plausibility is not validation. The retraction leaves the paper’s central claims unsupported and removes it from the dependable literature on antioxidant signaling in lung fibroblast models.
Researchers working in oxidative stress and pulmonary cell injury now have one fewer paper to cite when tracing the relationship between polysaccharides, ROS scavenging, and apoptosis. The remaining question is not whether the pathway exists in biology; it does. The question is whether this particular study documented it faithfully. PLOS has answered that question by retracting the article.