Stressed out cells sometimes glow under certain wavelengths of light. That phenomenon, a common problem for researchers using fluorescent microscopy, may be at the root of last year’s sensational claims surrounding stimulus-triggered acquisition of pluripotency (STAP) cells, the supposedly powerful stem cells derived using a remarkably simple recipe.
The STAP claims have been shown to be false, and the two Nature papers describing the cells in January 2014 have both been retracted. Now, researchers from seven labs have described what they saw when they tried to replicate the experiments in the weeks and months following the original publications. All report observations that may have misled the researchers who made the original claims, they write in Nature this week—including cells that glow, faintly, under key wavelengths of light.
Their findings don’t exculpate the team that reported the STAP claim, however. More cautious researchers would have realized they were on the wrong track, says Rudolf Jaenisch, a stem cell researcher at the Massachusetts Institute of Technology in Cambridge whose lab spent 2 months trying to make STAP cells and contributed data to the new paper. “If you know what you’re doing, it’s not that difficult,” he says. And other investigations have found evidence suggesting deliberate wrongdoing.
In the STAP papers, researchers in the United States and Japan claimed that treating cells from newborn mice with acid or forcing them through thin pipettes somehow transformed them into pluripotent stem cells, which occur in early embryos and can develop into all of the body’s cell types. The papers generated lots of excitement, not just because the technique was fast and easy, but also because the authors said STAP cells could form placental tissue, which pluripotent cells can’t do. That suggested that the STAP cells were even more versatile than embryonic stem cells.
But the claims quickly fell apart when researchers around the world could not reproduce the experiments. Instead of deriving from newborn mouse tissue, the supposed STAP cells were genetically identical to embryonic stem cell lines present in one author’s lab, according to an investigative committee at RIKEN, the Japanese research organization where lead author Haruko Obokata worked. (A RIKEN team published some of these data in another paper in this week’s issue of Nature.) RIKEN also found evidence of misconduct by Obokata, who resigned in December 2014. Obokata’s supervisor and co-author Yoshiki Sasai committed suicide in August 2014.
Obokata did her first STAP experiments in the lab of anesthesiologist Charles Vacanti of Brigham and Women’s Hospital and Harvard Medical School in Boston. He has continued to stand by the claims and has posted several updated protocols for producing the cells on his lab’s website, the most recent from September 2014. But clues from Vacanti’s lab suggested how he and other authors might have been misled when they tried to create STAP cells.
Around the time the STAP papers were published, stem cell researcher George Daley of Harvard Medical School and his colleagues had developed a method for analyzing the genes active in single stem cells. Daley contacted Vacanti to suggest that the method might help explain what was happening in STAP cells. Vacanti “was completely receptive,” Daley says. “He invited participation in his lab.” A student from Daley’s group, Alejandro De Los Angeles, visited Vacanti’s lab to learn how to make STAP cells. He soon discovered that the team had fallen for an artifact, Daley says.
The STAP team had used mice carrying a so-called reporter gene that would make the animals glow green under certain wavelengths of light if they expressed Oct4, a key gene active in pluripotent stem cells. The acid-exposed cells formed spheres that glowed green, they reported, providing evidence that Oct4 was turned on. When De Los Angeles and others repeated the experiments, they also saw the green spheres but noted that the signal was faint.
As they report this week, the signal was much fainter than the fluorescence from a bona fide embryonic stem cell containing a reporter gene. Instead it was consistent with autofluorescence, the low level of fluorescence produced by many cell types, especially those that are unhealthy or dying. “I think it is difficult to mistake autofluorescence for a genuine [Oct4] signal if you pay attention,” says Hongkui Deng of Peking University in Beijing, a co-author on the replication paper. But inexperienced researchers can be easily misled, Deng says.
The STAP cell team also claimed that their cells could form teratomas, the jumbled tumors of multiple tissue types that pluripotent cells produce when injected under the skin of mice with faulty immune systems. (Teratoma formation is a classical test of pluripotency.) But Vacanti, Obokata, and their colleagues used a slightly altered technique, first seeding the cells into a polymer gel and then inserting that under the skin of mice. When Daley and his colleagues redid that test, the lumps that formed were not teratomas but scar tissue, he says. Vacanti declined to comment on the new papers.
In the months following the STAP papers, several labs made similar observations, Daley says, and scientists began comparing notes. “All of us had detected the artifacts,” he says. “We decided that we should pool our data and go back with a response to Nature.”
The paper also includes a reanalysis of the sequencing data published with the STAP papers. It confirms a finding by the RIKEN panel that the STAP cells reported in the paper were male, although the donor cells were from a female mouse. The authors also found that some supposed STAP cells carried genetic signatures that matched both a line of trophoblast stem cells—the type that produces placenta—as well as an embryonic stem cell line. Those results are unlikely to arise from an accidental mixing of cell lines, Daley says, since these two cell types need different culture conditions to grow in the lab.
To avoid a repeat of the STAP fiasco, Daley, Jaenisch, and others lay out criteria in Nature for future claims of “landmark reprogramming advances.” They suggest a range of tests that should help protect against both misleading assays and cell line contamination; before they publish, researchers should also demonstrate that the technique can be replicated in independent laboratories. “Science is ultimately a self-correcting process,” the authors write, “where the scientific community plays a crucial and collective role.”
With reporting by Dennis Normile.