Say no to DMSO: and remember make your materials and methods clear

DMSO is used in many protocols as a ‘universal solvent’, it’s great for solublising small molecules and you've almost certainly used DMSO when freezing cells or in PCR. Unfortunately it’s relative ubiquity of use in research labs means it’s negative impacts on some compounds can be forgotten. A recent paper in Cancer Research by Michael Gottesman’s lab at NCI highlights one of these: that DMSO inactivates platinum drugs like cisplatin. They demonstrated that DMSO inhibits the cytotoxicity of platinum therapy, and reported that a literature review found up to one third of researchers are using DMSO and make the claim that this may render “a substantial portion of the literature on cisplatin uninterpretable”. Up to 50% of papers did not report the solvent that was used to dissolve cisplatin, making reproducibility of their results potentially impossible.

DMSO inhibits platinum therapy cytotoxicity

Interestingly a 2010 article in PlOS One showed that methyl sulfone may have clinical potential as a non-toxic agent effective against metastatic melanoma. Methyl sufone and DMSO are very very similar.

• DMSO (C₂H₆SO)
• Methyl sulfone, C₂H₆SO₂

The authors were drawn to the issue by the high rate of DMSO:cisplatin use reported in the literature, but also by issues they found in developing a high-throughput cisplatin screen where the robots used are optimised with DMSO as the solvent. They also point out that other groups developing alternative platinum therapeutics might be missing good leads due to the ubiquitous use of DMSO. The final paragraph of the discussion is very strongly worded, mch more so than the average paper. The authors state “the practice of dissolving platinum drugs in DMSO must cease” [my highlight]. But given that they provide guidelines on what to use instead, it seems appropriate that people take notice.

I don’t use Cisplatin, why should I be reading this: This paper points to two main issues. The first is a lack of quality in reporting experimental methods, it’s a message we should all listen to. The second is that a lack of understanding of this issue even though it had been previously reported up to twenty years ago.

We could all do with being clearer in our materials and methods sections. The use of supplementary files allows almost unrestricted descriptions of what we did. I find it exasperating that it can be impossible to find out what method someone actually used to create their NGS library or exactly how they sequenced it and how many reads they generated per sample. And it can be impossible to get a clear picture of the experimental design and replication used. George Church does methods right, we don’t need to know which tube you used (it’s only a 1.5ml Eppendorf after all), but the polymerase or the use of single- or paired-end reads is important. Lets all try a bit harder on our next papers.