Skin molecule maps using mass spectrometry

Bouslimani et al. (1) have produced a wealth of new chemical information about human skin using impressive analytical chemistry, data handling, and big data representation. The authors examine skin cells, skin tissue and, overwhelmingly, swabs of selected regions of the skin. These samples, microbial strains, and beauty product samples were fractionated by chromatography and tandem MS (MS/MS) data were recorded. A unique set of MS/MS transitions was obtained by pairwise comparison and a metabolomics database helped to identify individual compounds and create molecular networks. Despite the massive amounts of data taken, there is neither a claim of complete coverage nor are biological variations explored; most of the identified compounds originate in personal hygiene and beauty products. There are also quirky observations, such as finding the MALDI matrix sinapic acid on a subject’s hands (could this person be a chemist?). Implicit in the study are questions of whether in beauty care we are repeating past mistakes with chemical overuse. The study certainly informs us on what is left on our skin when we wash away the products that we dab, spray, smear, or deluge on ourselves.

Understanding the chemistry of our skin is incomplete without the study of the microfauna that we host. Public interest in microbiology is rising; see, for example, the opening of the “first museum of microbes,” Micropia, in Amsterdam. Scientific interest is also high, exemplified by the launch of the NIH microbiome project. A multiomics approach is needed to acquire the chemical and microbial information on the multispecies arena of our skin and Bouslimani et al. (1) use 16s rRNA sequencing for microbial identification.

Advances in analytical chemistry have played a key role in development of the omics sciences. Modern chemical instrumentation allows the acquisition of large and multidimensional datasets. Incorporation of spatial information provides an additional level of understanding. Mass spectrometry requires minute samples and provides a vast amount of chemical information. However, more data do not necessarily translate to better understanding, so parallel development of data analysis strategies capable to assessing the complexity of multiomics data are vital. The relevant information may be unraveled through patterns of similarity or dissimilarity present among samples and variables, rather than following the variation in a single “channel” of information (e.g., molecule); organisms function systemically and have evolved complex dynamic interrelationships between molecules and their biological function. Data mining is one approach to explore large amounts of data in which biological relationships exist, and some methods are inspired by biology itself: genetic algorithms simulate the evolution of an organism to adapt to a given environment; artificial neural networks that emulate the brain’s biology influence machine learning. Bouslimani et al. (1) use MS/MS-based networking to provide a global view of the molecular composition of a skin area. Mass spectrometric patterns are aligned and clustered to unravel biosynthetic and metabolic pathways. All this precedes the actual characterization of the molecules themselves, a situation that parallels that of early genome-sequencing projects. Ultimately, the question “why?” remains to be answered, even when given more information and better methods of data analysis. Regardless, Bouslimani et al. reveal the chemical and biological richness of our skin, graphically representing the molecular networks, although much of the MS/MS data are uncharacterized.

The aim of this commentary is to discuss emerging ambient tandem mass spectrometry methods and their applicability to large-scale projects like skin atlas development (Fig. 1). The key method, MS/MS, goes back more than 60 y. It first flourished in the 1970s when used to fragment mass-selected ions in gaseous collisions to characterize their structures from their product ion mass spectra (2). With the development of chemical ionization and other soft ionization methods, sets of molecules could be examined as their ionic surrogates, allowing MS/MS to be used to characterize individual mixture constituents (3). An added reason for the importance of MS/MS in the examination of complex analysis is that signal-to-noise ratios typically increase as one goes from MS to MS/MS (and to MS³) experiments, even though total signals fall significantly. This counterintuitive feature of MS/MS proved vital in the development of proteomics, starting with the use of triple quadrupoles (4) and four-sector instruments (5). Dorrestein and coworkers, both in the current report (1) and earlier work (6), use MS/MS fragmentation patterns to identify compounds with common structural elements. Metabolic processes impose chemical relationships that are readily recognized by common sequential fragmentations associated with particular substructures.

Fig. 1.

Map of distribution of sinapic acid on hand by swabbing, then LC-MS/MS, and Firmicutes sp. on feet by 16s rRNA [from Bouslimani et al. (1)]. Spray-based methods of ambient ionization of biological surfaces for diagnosis of strep throat (Upper) (12) and detection of tumor margins in tissue (Lower) (14, 15).

Although omics analysis relies on MS/MS, virtually all omics MS workflows include one or more steps of separation, typically chromatographic, because the direct ionization of complex mixtures introduces complex phenomena like competitive ionization. However, with increasing interest in complex macroscopic systems that include multiple types of biological materials, typified by an earlier study of Watrous and Dorrestein (7), the number of samples that must be examined, the speed required per sample, and the need to work nondestructively or minimally destructively, often precludes the use of separation methods. This means that even greater reliance must be placed on tandem and multistage MS. It also means that the traditional and almost universal approach in MS of introducing a sample into a vacuum must be discarded. Fortunately, methods of examining samples in the ambient environment without prior preparation or accompanying separation are available. The question we raise is whether ambient MS performed without prior sample preparation might be suitable for future studies of the type of Bouslimani et al. (1).

Ambient ionization (8), with its feature of examination of unmodified samples from their native environment, represents a major change in the range of systems on which chemical information can be sought. There are now dozens of ambient ionization methods (9), all unified by maximizing the speed and simplicity of MS analysis. Most samples of chemical interest are complex mixtures and the skin’s chemical and microbiome registry should be ideal for the molecular specificity and sensitivity of MS. Furthermore, correlation between chemistry and spatial location is possible via ambient MS imaging or 3D reconstruction of spatially registered MS data, as shown in Bouslimani et al. (1).

Ambient techniques have already been used for imaging and analysis of complex biological materials (9), including plant leaves, tissues of various organs, microscopic embryos, and microorganism cultures.

Bouslimani et al. have produced a wealth of new chemical information about human skin using impressive analytical chemistry, data handling, and big data representation.

Integration of chemical and complementary information, such as genetics, morphological features, biochemical mechanism, or spatial location, provides a powerful means by which to network and better understand dynamics and interrelated biological systems. Telling early examples are the use of desorption electrospray ionization to follow lipid changes in conjunction with gene-expression analysis in microscopic samples, such as individual oocytes and preimplantation embryos (10), and to reveal interspecies interactions and the chemical language of symbiosis or hostility (7).

Ambient MS is extremely effective for small molecules and lipids. The structural complexity of lipids results from a series of shared molecular building blocks that can be accessed by multistage MS and molecular networking. This analytical approach facilitates understanding of connections and cross-talk in the metabolism of species under physiological and pathological conditions. Indeed, as in other omics sciences, lipid research is driven by advances in MS, by increased understanding of lipid synthesis, metabolism, interactions with proteins, and regulation of gene expression, and by systematization of structural information into global datasets and freeware for statistical and pathway analysis of lipids and other small molecules, whose development was parallel to and inspired by work in proteomics (5, 6). Fundamental contributions to lipidomics MS by Brown and Murphy (11) and others have created a strong infrastructure in lipid biochemistry covering thousands of molecules, methods, pathways, publications, and mass spectra.

The swab approach of Bouslimani et al. (1) can be compared with recent methods where MS ionization occurs directly from the swab, leaving out the chromatography step. Such a touch-and-spray approach (Fig. 1) allows noninvasive sampling in vivo and is currently being tested in strep throat diagnosis (12). Spatial and chemical information coregistration allows MS data to be used to answer questions like, “is this area cancerous?” The recognition of the diagnostic role of lipids is a recent phenomenon, but lipid MS now provides tools for improved tumor surgery via molecular detection of cancer margins (13, 14) and discovery of new endogenous and onco-metabolite markers (15). Ambient MS techniques fit well in the surging effort to gain chemical information on complex biological systems and, in future, one can visualize use of portable MS/MS instruments. Ambient ionization tandem MS contrasts strikingly to the standard paradigm for omics chemical analysis in that chromatography is dispensed with, speed is of the essence, and complex MS/MS methods are relied upon.