Guest post: SmartFlares fail to reflect their target transcripts levels

This is a guest post by Maria Czarnek and Joanna Bereta, who have just published the following article in Scientific Reports entitled SmartFlares fail to reflect their target transcripts levels. 

We got the idea of using SmartFlare probes when working on generating knockout cells. In the era of CRISPR-Cas9 genome editing, the possibility of sorting out knockout cells based on their low target transcript content (mRNAs that contain premature stop codons are removed in a process called nonsense-mediated decay) instead of time-consuming testing of dozens or thousands of clones would be a great step forward. SmartFlare probes seemed to be just the ticket: no transfection, lysis or fixation needed; moreover, the probes were supposed to eventually leave the cells. We were full of hope as the first probes arrived.

We added the probes to the cells, examined the fluorescence using flow cytometer and… nothing. There was absolutely no difference between SmartFlare fluorescence in CRISPR-Cas9-treated and wild type cells. We expected that a small fraction of the cells would exhibit strongly reduced fluorescence, since previously generated knockout cell clones usually contained only 10-20% of mRNA of the targeted gene compared to that in WT cells. We chose not to give up and decided to generate knockout cell clones and test SmartFlares again, but it didn’t change a thing. It still wasn’t working the way it was supposed to. Stimulation of mRNA expression brought no change to cell-associated SmartFlare fluorescence either.

We also failed in getting any technical support from Merck, although eventually after 3 months of our pestering them, they accepted our complaint. A thorough combing through the Internet finally led us to Raphael’s blog. We realized that we were not the only ones who had been struggling with SmartFlares. We decided not to put our results into mothballs but to purchase additional probes and test whether this faulty flare constituted a single case. Four more probes gave exactly the same results: no correlation between SmartFlares fluorescence and transcripts levels.

We were aware that publishing our results would not be easy. We started with Scientific Reports and were positively surprised by the encouraging response from the Editor and Reviewers. Some of the comments helped us to refine the manuscript. However, some unfortunate circumstances pushed our revised manuscript into the hands of another Editor, who immediately rejected it suggesting that all our results could easily be explained by the use of degraded flares. It took an additional five months and two appeals to finally publish our manuscript. Meantime, the number of SmartFlare probes with different specificities offered in Merck’s on-line catalog dropped down from a few hundreds [Note from Raphael: two years ago, Northwestern boasted about “1,700 commercial forms of NanoFlares sold under the SmartFlare“] to about  20.

Although the second Editor caused us a great deal of trouble, we must admit that thanks to their critique we discovered that various SmartFlare probes were labelled unequally, which could be observed both when fluorescence was measured at basal conditions and when oligonucleotides were liberated from the gold nanoparticles with dithiothreitol (DTT). It is crucial in the case of the scramble probe, which serves as a control of background fluorescence arising from unspecific unquenching of the fluorophore. In our hands scramble probes (both Cy3- and Cy5-labelled) displayed about half of the RFU of the gene-specific probes. These results indicate that any differences between fluorescence of the target-specific probe and the scramble probe may simply result from the different degrees of fluorescent labelling of the probes rather than from the presence of target RNA and differences between specific versus non-specific fluorescence unquenching.

To the best of our knowledge, there are only a few papers showing positive results obtained with SmartFlare probes and we cannot help the feeling that the results might have been misinterpreted. We hope that our contribution to the discussion about SmartFlares (and spherical nucleic acids in general) will help other scientists decide whether to spend their time and money working with this technology, which in our opinion is both questionable and unconvincing. And if they do, we suggest they check the level of probes’ labelling before drawing any conclusions or, even better, before starting the experiments.