Cracking the Mori case: A reviewer describes how manipulated images came to light

We’ve had two questions since learning of the fraud case involving Naoki Mori: Who discovered the manipulations? And how?

We now have answers. We recently received an e-mail from a researcher who specializes in Mori’s field — cancer viruses — and who claims to have been a reviewer of a paper he submitted early last year to a journal in his field. (We’re obscuring some details to maintain our source’s anonymity.)

During what turned out to be a “painful” review of the manuscript:

I noticed problems with duplicated control actin bands within the same manuscript—sometimes within the same figure!  It was astonishing.

Our source describes Mori as someone “who characteristically publishes very weak papers. And a lot of them.” Those two facts raised this scientist’s suspicions that the dodgy actin bands weren’t simply a one-off problem.  Going online, the reviewer found and printed out a slew of Mori’s previous publications.

For two days, I worked in my office with several of my graduate students, sitting on the floor, and cutting out figures from the numerous Mori papers. We created piles of figures that had many of the same ß-actin RT-PCR “control” bands used throughout.  We found that several versions of the ß-actin RT-PCR and Western “control” bands in many published papers, as well as GAPDH bands, reused EMSA data [a technique for studying how proteins interact with DNA or RNA molecules], and on and on. Importantly, there were examples where the same bands were deliberately altered by increasing the width of a band series to give the impression that the bands were more intense.  Yet a pixel-by-pixel examination of the band series very clearly shows the bands are identical in origin, and simply altered by stretching vertically. Mori was very, very intentional — and cavalier — in his efforts to fabricate and falsify.

We asked Jeff Perkel, our resident molecular biologist, to help us translate. Here’s his take on why what Mori (and at least one co-author, Mariko Tomita), was doing:

These controls are important, because they show that the amount of input nucleic acids in each reaction is identical; GAPDH is what’s called a housekeeping gene — its expression is expected to be constant across experimental conditions (eg, if you have a sample that doesn’t show a particular band, is that a true negative, or is it because the reaction didn’t work, or you didn’t add sample to that reaction?)

It sounds as if not only did the authors of the paper do that with RT-PCR data, but also with Western blots (that is, blots measuring protein abundance) and EMSA experiments (which indicate the presence of DNA-binding proteins). In all cases, controls are needed to show that each lane’s reaction worked, and that they contain equal input amounts of material.

The second allegation, “where the same bands were deliberately altered by increasing the width of a band series …” is also a case of creative Photoshopping. In this case, they evidently wanted to demonstrate that there was more of a given band (i.e., more protein, RNA, or DNA) in a given lane than there really was. That’s different from what some researchers legitimately do, which is to modulate the brightness and contrast of a gel to highlight certain features — that is done uniformly across a gel, generally, whereas these researchers specifically gamed just one feature — what my wife and I used to call the Sharpie effect (i.e., drawing on a piece of X-ray film with a Sharpie to bring out certain features).

Armed with evidence, the reviewer notified the journal. The response wasn’t exactly reassuring.

Yet oddly, they said that they agreed there was a problem with the submitted manuscript under review and that ASM [American Society of Microbiology] had sufficient evidence to reject it [which it did]. But they did not express interest in pursuing it beyond ASM. I felt like so much more needed to be done.

(See comment below from the ASM.)

More included contacting the editor of Blood, which had recently posted two of Mori’s articles online ahead of print (those papers have since been retracted).

Amazingly, Mori’s university was contacted immediately and an investigation ensued. I say amazing, as this is the fifth serious case of scientific misconduct I’ve encountered in my reviewing of grants for NIH and reviewing manuscripts since 2004.

That was back in mid-March 2010. According to our source, Mori’s institution, the University of the Ryukyus, looked at 50 of his papers, finding 37 that had “anomalies.” By August, as we now know, Mori was out of a job.

Such rapid resolution is almost unheard of in the U.S., according to our source.

Even when a case is blatant, universities in the United States are required to perform an internal investigation, and they decide whether or not fraud has occurred.  And I know that many universities decide that the individual is innocent, since a finding of guilt by one of “their own” faculty is a mark on the University, bad media coverage, and may result in the loss of indirect costs from grants should the PI be fired.  ORI [Office of Research Integrity] is very well aware of this problem, yet they are powerless to do anything.  Fortunately, in four of the five serious cases of misconduct I’ve uncovered, four of the identified investigators eventually left their universities — all quietly except for Mori [who, as we reported last week (hat tip, ScienceInsider) has lawyered up to try to get his position back.]

We recognize that some of the retractions we cover are more humorous or ironic than scientifically significant. But the fallout from Mori’s deception and other sweeping cases of fraud has consequences beyond the damaged reputations of researchers and the frustrations of journals and publishers.

We’ll give our source the last word here, because we can’t express it any better:

I have had graduate students base their hypotheses and experiments on data from the Mori papers … What a waste. I am angry.  Mori, like several of the others I’ve stumbled upon, has been doing this data fabrication for so long that he’d become cavalier about it.  He just didn’t think he would ever be caught.