Tuesday, September 29, 2015

Stacking the deck with images?

I wish that I could solicit input on this question as an unbiased observer. Anyone who's read my previous posts knows that I am anything but. I considered presenting this data absent the source, but anyone with 1% of Sherlock Holmes' sleuthing ability would track down the paper I am referring to in a nanosecond.  Going back and forth with the EIC at ACS Applied Materials and Interfaces recently (maybe more on this later) forced me to look at aspects of this paper in greater detail than I would have otherwise. As I documented before, one of the reviewers fixated on the cell imaging aspects as a defense of the results, and a rationale to reject my Comment. In preparing an appeal, I looked more carefully at the cell images

Here are cells treated with Fe3+ and probe:

The difference between a) and b) is 1 hr of incubation time. Ref: ACS Appl. Mater. Interfaces20146, 18408–18412

Here is the control of cells treated only with probe:
The difference between d) and e) is 1 hr of incubation time. Ref: ACS Appl. Mater. Interfaces20146, 18408–18412

Notice any differences? I am not very experience with growing cells, but to me it looks like the confluency is much higher and the cell size is larger in the images of the iron-treated cells (Figure a-c), than the sensor-only cells (Figure d-f). This could create the illusion of a greater fluorescence response in the iron-treated cells. There is simply more sensor present per unit area, and there is clearly basal fluorescence. In fact, I might argue that the sensor-only cells (f) are not even healthy compared to the Fe3+-treated cells (c). It also appears that the fluorescence images of the sensor-only cells are out of focus. Even the brighter cells in the middle of the field look fuzzy. Again, this could make the response look more dramatic than the reality. There are other problems with this experiment related to metal transport, but I don't want to conflate imaging with biology. I have a biological interpretation backed by literature for what they observe in the Fe3+-treated cells.

I am interested if anyone concurs, or if I am the victim of confirmation bias. There have been several notorious examples of image manipulation including cut & paste to make nanochopsticks and reusing old data. This isn't that kind of malfeasance, but could it be stacking the deck to fit a preconceived model? How say you?

Friday, September 11, 2015

Fe(III) sensor saga continues

A few months back, I blogged about a paper on iron sensing by Belfield in ACS Applied Materials & Interfaces, and my efforts to bring my concerns about the validity of the conclusions to the attention of the editor. While those efforts were failing to get traction, I brought up similar concerns about a paper in EurJIC. The editor at EurJIC requested that I write a peer-reviewed "Correspondance" on the issues after the author was unable to satisfactorily respond to the criticisms in private emails. Having started the correspondence process with EurJIC, I realized this was also the only recourse I had left at ACS-AMI. A similar mechanism was mentioned in passing by the EIC of ACS-AMI as the only option once a paper was published in the journal, so I decided to submit a "Comment" to ACS-AMI

Obviously, since both papers deal with Fe(III) probes, there are some similarities in the problems as in many other published reports. There are some notable differences as well. In short, the EurJIC paper utilizes PBS buffer and reports a fluorescence quenching mechanism. I indicated evidence of particulate iron and speculated about alternate explanations for the fluorescence changes. The ACS-AMI paper reports fluorescence enhancement in unbuffered water, which I concluded is a clear indication of protonation instead of Fe(III) binding. I very confident in my critique of the ACS-AMI paper because this is exactly the same problem that my group investigated previously with another (very similar) probe. I received a rejection of the comment Friday. Two positive reviews, both suggesting minor revisions, and one rejection. 

Although anonymous, I have suspicions that the rejection was written by an author of the original paper. I conclude this because 1) the authors should have an opportunity to respond in such a situation, and 2) the phrasing of the arguments. I wish that I was surprised that the EIC decided to side with the single "Do not publish" referee and didn't leave an opening for appeal. From the beginning I surmised that the EIC had little interest in criticism of the paper. As a reminder, the lead author of the paper in question is a member of the ACS-AMI editorial advisory board. The dissenting referee made some flimsy arguments against Fe(III) hydrolysis, and focused on the "validity" of the cell imaging studies as a reason to reject the Comment. I specifically did not address cell imaging because sources of false positives in imaging studies are elusive, and more often unrecognized. In the Cu(I)-sensing field, as well as sensing for other metal ions, colloidal aggregates are just recently being recognized as a significant (if not primary) reason why many probes respond  in cells. This is one possible explanation for the imaging results using the Belfield probe, but it is equally likely there is an unknown mechanism that can produce false positive in Fe(III) systems.

The EIC further cited non-disclosure of the Correspondence at EurJIC as an additional rationalization for rejection. This appears to be a thinly veiled insinuation that I behaved unethically. To be clear, no attempt was made to conceal the existence of the Correspondence. I didn't cite the unpublished Correspondence in the Comment due to the lack of a journal citation at the time of submission, as well as the differences in the two systems. I went to great lengths to perform a non-repetitive analysis, even though the underlying issues of Fe(III) hydrolysis are identical (e.g. I needed to cite the same pKa values for hydrated Fe(III) in both articles and it's difficult to express this uniquely). Owing to the different types of experiments performed in each case, I believe that the Comment highlighted a different set of problems that lead to incorrect conclusions. I am not impartial in this case, but I believe people can still write on the same topic more than once in the scientific literature. Many authors, myself included, have been asked to write reviews on the same subject for different journals and therefore reach different audiences. After receiving referee reports from ACS-AMI, I can certainly imagine having revised the Comment with a citation to the EurJIC Correspondence with some additional discussion to address the referee suggestions, although I feel it would confuse the situation to make a close comparison of 2 systems that exhibit different signal transduction pathways.

Furthermore, the EIC suggested that my group should "[carry] out its own investigation on these sensor systems". If one looks at the synthesis of the Belfield probe, you quickly see that this would be an significant commitment of time. The probe requires a 6-step synthesis with a late stage macrocyclization step that proceeds in a reported 20% yield. The only positive outcome of such an effort would be disproving a published paper. A truly comprehensive investigation would require re-synthesizing and re-analyzing dozens of reported Fe(III) probes. This is not usually the kind of study that the community is excited about publishing. This is definitely not deemed fundable research by granting agencies. My aforementioned Inorg. Chem. paper that included a re-investigation of an existing Fe(III) probe as only a small component of a larger study was deemed unimportant by one referee for exactly this reason. Fortunately, that paper was handled by a diligent, thoughtful editor and was published anyway. Again, I am not impartial, but I feel the alternative resolution was proposed by the EIC at least in part because no rationale PI will undertake such an investigation in the current publishing/funding climate. So where does that leave the ACS-AMI situation? I am seeking the input of my scientific friends and colleagues. I have considered the following options:

1. Do nothing. Let this decision stand and move on to bigger and better things. Certainly the most rational course of action from one perspective. The problem with this choice is that the ACS-AMI paper has been cited by authors on multiple occasions as a defense when I critique their experimental protocols/data interpretation in manuscript reviews. This has happened several times in just the last 3 months. This is how these erroneous ideas have been perpetuated in the literature, and it will only get worse.

2. Conduct a study. In the Cu(I) field, it has been proposed that any imaging study should include a control where cells are treated with a structurally similar probe without metal binding ligands. In the case of the Belfield probe, this probably would be some dialkylanilino-BODIPY derivative. That's a pretty easy easy compound to make, but does not create a complete story without the Belfield probe for side-by-side comparison. This returns to the problem of time and resources. Also, what journal publishes such an investigation? It is highly repetitive to say that I am not impartial, but recent history suggest that ACS-AMI won't be particularly interested in this study.

3. A wikileaks-like information dump. No matter what direction I choose, I feel that it would be justified to "self-publish" the Comment here (although the information is already covered in the blog) as well as the referee reports and EIC feedback from ACS-AMI. I have nothing to hide, but what are the ramifications of such an action? Even with full disclosure, I can't escape the feeling that this is such a niche area that the impact will be minimal unlike more prominent issues that have captured the attention of chemists.

What do you think? I'd love to hear your comments and suggestions.