Issue:

When Haruko Obokata announced her STAP breakthrough, she had no idea that Twitter would be her downfall

Twitter, blogs and Facebook have revolutionized the way people gather and exchange information, contributing to the fading influence of traditional news outlets. Now these same forces are reshaping the way scientists communicate, and the unfortunate saga of Haruko Obokata and her STAP cells is a prime example. Within days of her breakthrough announcement, self-appointed scientific watchdogs worldwide were raising questions about the images in her papers in tweets and blog posts, and working scientists were exchanging notes on their attempts to reproduce her results.

Obokata, of Kobe’s RIKEN Center for Developmental Biology along with colleagues at other institutions in Japan and at Harvard Medical School in Boston announced on Jan. 29 that briefly putting blood cells from newborn mice in an acid bath and then carefully nurturing them could generate pluripotent stem cells. Such cells are theoretically capable of developing into any of the cells within a body, and researchers envision the possibility of growing replacement tissue for damaged or diseased body parts as being key to new treatments for spinal cord injuries, neurodegenerative diseases such as Alzheimer’s, and diabetes. In two papers published online at the journal Nature, Obokata and her colleagues called their new method “stimulus-triggered acquisition of pluripotency,” or STAP.

Previously known methods of generating stem cells are time-consuming, expensive efforts requiring sophisticated laboratory skills and specialized equipment. The STAP technique appeared to be an astoundingly simple, quick and inexpensive alternative. This was truly a breakthrough scientific achievement one that was celebrated with banner headlines worldwide.

A LAB OF PASTEL COLORS
In Japan, there was also understandable pride in a local doing well. And Obokata young, attractive and female was a refreshing spot of color in Japan’s staid scientific establishment, a point bolstered by a number of endearing personality quirks.

But while the local press was reporting how she painted her lab in pastel colors and wore a traditional kapogi apron instead of a lab coat, scientists around the world were growing skeptical. Though the reports had been through Nature’s peer review pro-cess involving vetting by a small number of senior experts in the field, researchers had begun raising questions. One of the first of these came up on the same day Obokata’s papers appeared online. “I am struggling to understand the results presented in figure 1i,” wrote an anonymous contributor to a website called PubPeer.

PubPeer was set up in October 2012 to “create an online community that uses the publication of scientific results as an opening for fruitful discussion among scientists,” according to a statement on the website. In effect, PubPeer allows ordinary researchers to critique scientific papers after publication. Those running the site and most posters remain anonymous to avoid “negative effects on their scientific careers.” For “negative effects,” read lawsuits and retribution by senior scientists who might be able to influence decisions on funding and employment.

That first comment on Figure 1i in the Obokata paper put others on the case, and on Feb. 4 someone spotted the problem: The figure shows the results of a certain type of genetic test. But instead of being the results of just one such test, as is typical, it appears that one part of a picture from a separate test was spliced into the image to look as if it belongs there. Such image manipulation is strictly taboo.

SCIENCE SLEUTHS ANONYMOUS
Japanese watchdogs were also at work. The most prolific of these scientific sleuths runs Twitter and Facebook accounts and blogs with the handle “11Jigen,” sometimes written as Juuichi Jigen, which means “11 dimensions” in Japanese. Juuichi Jigen, who has over 14,000 Twitter followers, adopted this nom de guerre after exposing alleged misconduct by a Uni-versity of Tokyo researcher who claimed to have developed an 11-dimensional theory of the universe.

Juuichi Jigen is a life science researcher who started uncovering scientific misconduct in 2010 after he “wasted time and money” attempting to reproduce results published by a certain Japanese researcher. Juuichi Jigen documented alleged problems with the researcher's papers, notified the researcher’s university and posted all the evidence on a website. Many of the researcher’s papers were subsequently retracted and he resigned his position. He was just the first on a long list of disgraced scientists listed on Juuichi Jigen’s websites.

On Feb. 14, Juuichi Jigen tweeted that a section of Obokata’s research article describing a certain laboratory procedure appeared to have been copied and pasted from a previously published paper by a different research group. (This was first mentioned on yet another internet bulletin board.) He also identified problems with other images in the Nature papers as well as in a previous Obokata paper. It is difficult to be sure exactly who was the first to spot any given problem, as the watchdogs retweet and cross-post each other’s findings. But most, if not all of the questions about images and plagiarism originated with people like Juuichi Jigen and the contributors to PubPeer.

Anyone with sharp eyes or fraud-detection software and sufficient patience can check papers for doctored images and plagiarism. But hands-on researchers were also trying to replicate Obokata's results in the laboratory.

CROWDSOURCING THE EXPERIMENT
To help them compare notes on progress and difficulties, Paul Knoepfler, a stem cell researcher and blogger at the University of California Davis, set up a page on his website dedicated to efforts to reproduce the STAP method. Knoepfler wrote in an email that to the best of his knowledge his was the first attempt at such widespread pooling of a scientific effort. “That’s part of the reason I was excited to give the crowdsource experiment a try on my blog,” he wrote. Over roughly two months, researchers posted 137 comments and questions, and 11 groups posted results, all of them negative.

Kenneth Ka-Ho Lee, a stem cell researcher at the Chinese University of Hong Kong, went further. He decided to go step-by-step through the STAP procedure, and he blogged about it at least daily on a website called ResearchGate, a networking site “built by scientists, for scientists.” Unfortunately, Lee gained unexpected notoriety and uncovered one of the pitfalls of this SNS approach when some of his results were misinterpreted.

One telltale sign of pluripotency is the activity of a certain gene that is often detected though the fluorescence of an attached protein. At one point, Lee reported getting some fluorescence. While it was clear to scientists that the signal was too weak to indicate pluripotency, several media outlets reported that Lee had validated the STAP technique when, in fact, he had concluded the opposite. “I don’t think STAP cells exist and it will be a waste of manpower and research funding to carry on with this experiment any further,” he wrote on ResearchGate on Apr. 3.

Throughout all this, RIKEN was playing catch-up. It launched an internal investigation on Feb. 13 and set up a formal committee on Feb. 17. In a preliminary report released Mar. 13, the committee found that many of the allegations raised by the watchdogs were accurate, including that the Figure 1i image was spliced together and that the description of the laboratory procedure was copied from another paper. On Apr. 1, the committee concluded that some of the problems with the papers constituted research misconduct on the part of Obokata.

INVESTIGATING THE INVESTIGATORS
The story then took another twist. Anonymous whistleblowers raised questions about image manipulation in papers authored by four of the six members of the RIKEN investigating committee. When the evidence surfaced, Shunsuke Ishii, a molecular geneticist at the RIKEN Tsukuba Institute, hastily sent a correction to the journal involved and resigned as head of the committee, which was then considering an appeal by Obokata. He is now under investigation himself. Other committee members have continued to serve, but they are also now facing investigations.

At an Apr. 9 press conference in Osaka, Obokata tearfully admitted that there were errors in the papers. But she said they were innocent mistakes and not the result of deliberate attempts to mislead scientists. She also fiercely defended the bottom line. “STAP cells exist,” she defiantly declared at one point. RIKEN rejected her appeal of the investigative committee finding, and a separate RIKEN committee is now considering disciplinary measures.

As of this writing, no research group has reported reproducing STAP cells. It very difficult to conclusively prove a negative condition in science; that is, it is possible that STAP will be proven to work under certain circumstances. But it now appears that, at the least, STAP is not the simple, quick method the Obokata team thought it to be.

The Obokata papers made such bold claims in such an active and controversial field that scrutiny was inevitable. But it is unlikely that the problems with the images and the method itself would have been identified so quickly without the current social media platforms. “I believe that science is becoming more networked in ways that permit for incredibly rapid communication and interaction in ways that would have seemed unimaginable even just 10 years ago,” Knoepfler wrote in an email. The use of social media by scientists to review scientific papers after they have appeared “is a phenomenon that is here to stay and will only grow in influence,” he added.

The trend might even circle back and force scientific journals to improve the peer review process. It is disturbing “that journals like Nature (probably Cell, and Science, too) have not been carrying out their checking responsibilities seriously,” says Robert Geller, a geophysicist at University of Tokyo who has followed the STAP cell controversy, occasionally tweeting and blogging about it.

No one knows yet to what extent research will change as interconnectivity increases. But it is clear that the traditional peer review process has proven to be no match for social media.

Dennis Normile is the Tokyo correspondent for Science.