To the Editor:
Once considered mere metabolic waste, the products of heme catabolism are now increasingly viewed as molecules with a mission.1 Bilirubin, potent interceptor of free radicals, hit the spotlight first, but the focus has now widened to embrace the two primary products of the process, CO and biliverdin. A recent example is the report of the effects of CO and biliverdin on immune-related liver injury.2 However, the findings of that paper and an earlier related paper3 are seriously undermined by a fundamental methodological flaw that may not be evident to most readers and seems to have been overlooked or ignored by the authors. The flaw concerns the nature and identity of the “biliverdin” used for the study and on which major conclusions of the paper ultimately depend. Described as originating from Sigma Chemical Co., the “biliverdin” was administered to mice intraperitoneally at a dose of 25 mg/kg dissolved in saline. The volumes injected and the concentration of pigment in the injectate were not given.
To my knowledge, Sigma has never sold biliverdin. At one time it did sell “biliverdin dihydrochloride,” which is not the same as the free acid and would have different solubility properties. More importantly, the material that was sold was clearly labeled as “Approx. 80%.” Material of even that degree and uncertainty of purity is unsuitable and inadequate for reliable biomedical investigations, especially those dealing with immune responses, cytokines, and enzyme induction. However, it has been known for years that “Approx. 80%” is an optimistic estimate, some batches of the product containing far less than 80%.4–6 Figure 1 shows thin-layer chromatograms of authentic biliverdin and of two batches of Sigma “biliverdin dihydrochloride” (Catalog no. B-3753). The commercial product is clearly not homogenous and contains a great deal of (greenish) material that is not biliverdin and mainly remains at the origin. Analysis of the same two batches by high-pressure liquid chromatography (HPLC) indicated that one contained only ≈38% biliverdin and the other ≈68%. In each case the biliverdin was a mixture of IIIα, IXα, and XIIIα isomers (which do not separate on thin-layer chromatography), making the actual content of biliverdin IXα, the natural biosynthetic isomer, even less than 38% and 68% respectively.
Sigma stopped making and selling “biliverdin dihydrochloride” about 6 years ago and no longer offer it in the Sigma-Aldrich online catalog. Thus, the experiments described in the HEPATOLOGY paper2 are based not only on material of uncertain identity, but also on a product that is no longer available, making them independently unverifiable. Without reliable experimental evidence that the material used was authentic biliverdin of a high degree of purity, it is impossible to judge the scientific significance of the recent studies.2, 3 A similar caveat applies to other papers in the literature that have unquestioningly used the Sigma product or similar “∼80% pure” material from other vendors.7, 8 That is not to say that biliverdin does not have the biological effects suggested—it might have even stronger effects—only that its true biological activity cannot be divined from the experiments as described.
The biological activities of the impurities in Sigma “biliverdin dihydrochloride” are unknown. Biliverdin is poorly soluble in water, and therefore preparation of injectates by dissolving the pigment directly in saline, as described,2, 3 would be impracticable or impossible with pure material unless a co-solvent or relatively vast volumes of saline were used.
When assessing the purity of bile pigments it is important to be aware that criteria such as HPLC, absorbance spectrophotometry, and mass spectrometry are inadequate and potentially misleading when considered alone. For example, a single peak on HPLC does not necessarily signify a pure compound because impurities may be retained on the column, the system may be inadequate, or a detector optimized for biliverdin may not “see” impurities. Absorption spectrophotometry is also unreliable because bile pigments have broad absorbance bands and the spectra of impurities may resemble those of biliverdin itself. To establish purity, several methods need to be used and the material should, at least, give satisfactory C,H,N analysis. Fortunately, simple, inexpensive thin-layer chromatography procedures can be useful informative indicators of purity (as shown in Fig. 1), if appropriate systems are used.
Investigators can hardly be expected to analyze all solvents and reagents, and for common materials this is generally unnecessary. But bottle labels and suppliers' descriptions and analyses are not infallible and it is foolish (and irresponsible) not to confirm the identity and suitability of key materials before their use in extensive investigations. Otherwise, as in the present instance2 and earlier work on “biliverdin,”3, 7, 8 much effort may be wasted to no useful scientific end, cluttering the literature with questionable, possibly unreproducible, findings. In the case of the recent HEPATOLOGY papers,2, 3 studies done with grossly impure material could be particularly misleading. It is not only in the computer world that the Gigo effect9 applies.