I hate to be like this. But yesterday a number of chemists began noticing this paper, and after having a look, I think it really needs to be mentioned. (Update: if that link is down, and it’s been spotty under all the attention, try this one). It’s a group from Kuala Lumpur, the pharmacology department of the University of Malaya, and it details the effect of a particular compound on various cell lines (particularly hepatocarcinoma-derived ones).
Fine. I’ll say up front that I don’t find papers like this particularly interesting, but that’s just my own opinion. A lot of compounds will affect various cell lines in vitro, and by now I’ve seen enough of them that I don’t feel the need to read about more examples unless there’s something really notable about what’s going on. At the same time, though, there’s nothing erroneous about saying “Here’s this compound and here’s what it does to cells”. Uninteresting does not mean wrong. I will say, though, that I find that rationale behind the paper to be rather thin – as far as I can see, it comes down to “Piperazine compounds are known to affect liver cells – here’s a piperazine compound, and here’s what it does to liver cells”. But again, that’s not wrong.
So let’s get down to the parts that are wrong. This is the structure of the compound of interest, as rendered in the paper’s supporting information. And this is a warning sign. That structure has been drawn by someone who has little or no experience with organic chemistry or with a structure-drawing program. The bond lengths and angles are all skrinked around to no purpose. The connectivity of the structure is still there, but the other message this drawing conveys is “I don’t know what I’m doing”. You’d say the same thing if someone produced a picture that is supposed to be an outline of the state of Florida, but shows the panhandle curving back around towards the Keys. It doesn’t do that; Pensacola isn’t over that way. In the same way, that carbonyl carbon isn’t trying to flee the rest of the molecule by lengthening its bonds and angling them off into the distance (and that’s not the only funny-looking bond in the structure by any means).
Well, fine, there are ugly structures all over the place, drawn (more often than not) by biologists and appearing in biology-oriented journals where most of the readers don’t even notice. But more serious problems are coming. You’ll note that no stereochemistry is shown for those hydroxyl groups on the cyclohexane rings, so in theory this could be a mixture of eight compounds. When you go to the experimental section of the paper, though, you get this for the synthesis:
Using tert-Butyl piperazine-1-carboxylate, a Buchwald coupling was performed. Deprotection of the carbamate was achieved using 20% HCl in methanol. Subsequent amidation using triethylamine and various acyl chlorides yielded the final compounds that were tested after purification and structure validation using NMR.
That’s it. And that “procedure”, unacceptably sparse as it is, clearly will not make that structure. Buchwald couplings make bonds between nitrogens and aryl rings, so one’s first guess is that someone really didn’t know organic chemistry and meant for both of those cyclohexyls to be phenyls. But hold on – the structure, as drawn, is C17H30N2O5, for a molecular weight of 342.4. The aromatic version, if that’s what was meant, comes to C17H18N2O5, 330.3. But the experimental says that they calculated the elemental analysis for C20H20N2O2 (320.4), and it came out correct. What compound is that? Not what’s drawn, that’s for sure. It gets even worse when you look at the mass spec in the supplementary information. There’s a big, strong peak at 395, which the authors apparently think is just fine to put next to a structure that doesn’t match it in any way, neither of which match the experimental section of the paper itself.
They also have proton and carbon NMR spectra, and I don’t suppose that it’ll come as a surprise at this point that neither of those appear to match the structure given, either. The carbon NMR certainly tells you immediately that those rings aren’t cyclohexyls, as drawn, but it also tells you that there are apparently two carbonyls in the molecule (or at least some other carbon out in that region), which isn’t right, either. The proton NMR doesn’t remotely match anything discussed so far. There are aromatic peaks out at 8.7 (pyridine?), a singlet at 9.4 (?), and a doublet out at 14.3 (phenol OH? carboxylic acid? who knows?). Nothing adds up. Any organic chemist who looked over this stuff for thirty seconds would know this, which tells us that no organic chemist ever did so – not the authors, not the reviewers.
We’ll get back to that topic of reviewers in a moment. For now, though, let’s walk away from the wreckage that is the paper’s chemistry content and turn to its biology, which is clearly its main emphasis. Take a look at figure 5. As a commenter on the paper (see the bottom of the page at its link above) caught, this figure supposedly describes the effect of the compound at two different concentrations on three different cell lines. But all the images are identical. The same cells are in the same places, which is impossible. All that’s been added in each frame are the “features” pointed out by the red arrows. One does not need to be a cell biologist to notice this. But neither the authors nor the reviewers thought anything of it.
There’s more like this. Figure 6 is full of copy-and-pasted cells, in a manner so blatant that a high school student would not expect to get away with it. Figure 10 has whole panels duplicated between what are supposed to be different cell lines. I don’t even want to start looking at the Western blots – why would you trust them when the rest of the paper is such a mess?
Enough. The authors should be ashamed of sending a manuscript out like this, because its mistakes are so numerous and so obvious as to make outright fraud – deliriously incompetent fraud – the first explanation that comes to mind. The reviewers, if there were any, should also be ashamed for letting something like this pass. Who looked at this stuff? Now, Scientific Reports is a catch-all journal, for sure, the Nature Publishing Group’s answer to PLOS ONE. In theory, it’s not supposed to review for impact in any way, just for scientific accuracy. Once that’s established, the authors pay the fee and the paper gets published, and the whole thing is open-access.
I don’t have a problem with that model. But it doesn’t work if you don’t review the damned papers. I don’t think that Nature wants to compete head-to-head with all the junk publishers of this world, but if they really want to, this is how you’d do it. Wave everything through, publish crap, cash the checks. They need to get their act together before they do any more damage to their good name.
Update: alert reader Adrian Roitberg has sent along a link to this paper, published in January from the same authors in Malaysia in the journal Frontiers in Pharmacology. Note that Figure 2 in this second paper contains some of the same images that are found in Figure 5 of the Scientific Reports paper. Unfortunately, they’re supposed to be images of completely different cell lines. Similarly, Figure 4 of the Frontiers in Pharmacology has images that are reproduced in the SR paper’s Figure 3, and again, they’re supposed to be completely different cells from a completely different experiment.
Want to see something even more stupid? Not only are these images being blatantly recycled between papers, but in the Frontiers in Pharmacology paper, Figure 4, the images for the HT-29 cells are the same images for the WiDr cells, just rotated 90 degrees. Such a time-saver! I get the impression that this lab has only taken about half a dozen cell images, which keep getting rotated and recolored as needed.
Update 2: Retraction Watch is on this, too. Yet another paper from the same group also shows duplicated and shared images as well. Amazingly, the lead author is denying any problems at all, saying that well of course these cells look the same – they’re in the same stage of growth, right? That explanation shows all the believability and finesse of the original image manipulations. I look forward to seeing what comes next.