Many biomarkers of inherited metabolic diseases can be diagnosed in high-throughput targeted analyses that are often sufficiently routine for large scale screening. However, the development of such protocols requires the knowledge of diagnostic biomarkers which, by definition, must be first identified in (un)targeted analyses of patient body fluids. Currently, (ultra) high performance liquid chromatography/high-resolution (tandem) mass spectrometry ((U)HPLC/MS
(n)) is one of the most successful techniques for untargeted profiling of patient body fluids in order to identify small molecule metabolites that correlate with, and are specific to, a given disorder and can thus be used as biomarkers. Another untargeted approach is nuclear magnetic resonance (NMR) spectroscopy, though this technique is limited to metabolites that are above the low micromolar (μM) range and is thus blind to a significant portion of physiologically-relevant concentrations. LC/MS
(n), on the other hand, is an ultrasensitive technique and can routinely detect thousands of compounds in a body fluid down to the low nanomolar (nM) range, though it lacks the specificity to molecular structure that is often necessary to distinguish isobaric compounds. Thus, the techniques are complementary in the sense that LC/MS
(n) is very good at detecting mass/charge (m/z) features in a sample that correlate with a particular disorder and NMR is very good at identifying the molecular structures of unknowns. This is illustrated in the recent identification of a new inborn error of metabolism (IEM), NANS deficiency (van Karnebeek et al
2016). The metabolite identified as a biomarker of the disease was detected by untargeted metabolite screening using LC/QTOF-MS, where its concentration was well within the sensitivity of the technique, but could not be unambiguously identified from its structural analogues on the basis of retention time, m/z ratio or fragmentation behavior. Thus, NMR was used to identify the molecular structure to correspond to N-acetylmannosamine by comparison to reference material; however, it was close to the limits of sensitivity for NMR. Thus, in order to push the limits of biomarker discovery, advances in analytical methods that combine both extreme sensitivity and highly specific molecular structure information have a critical role. Here, we outline the potential that infrared ion spectroscopy (IR-IS) has to fill such a role in the state-of-the-art toolbox of metabolomics researchers.
Infrared (vibrational) spectroscopy is a well-established technique that is often used in the mid-IR region, or the vibrational fingerprint region, to probe the fundamental vibrations of molecules. This technique is often used, for example, in the organic chemistry laboratory, to determine the presence or absence of particular functional groups in a molecule in either the solid, liquid or gas phase. This is possible because the IR features of different functional groups are found at characteristic (and very often unique) IR frequencies and make it possible to generate a fingerprint or signature of a particular molecule if a range of IR frequencies (for example, 1800–600 cm
−1) is sampled. These types of measurements are based on detecting the extent of attenuation of the IR light after it passes through a sample where photons are absorbed when they are at resonant frequencies with one of the vibrations of the molecule. Infrared ion spectroscopy (IR-IS) is the combination of ion trap mass spectrometry and infrared spectroscopy, selectively generating an IR spectrum for any mass-isolated ion in the MS spectrum (see figures below). In contrast to the standard IR absorption spectroscopy described above, IR-IS detects the absorption of IR photons on the basis of the photofragmentation they induce in a set of precursor ions trapped in a mass spectrometer. This approach exploits the mass spectrometer as an extremely sensitive detector and is necessary because the number densities of ions (of the same charge) required to produce a measurable attenuation of the IR light cannot be reached. Measurements are conducted directly inside the mass spectrometer and thus maintain full sensitivity and require no additional experiments outside of the mass spectrometer. The IR spectrum of a given m/z feature, however, adds an orthogonal level of molecular structure information in the form of an IR fingerprint. IR-IS has evolved in little more than a decade from an experimental technique in academic labs studying fundamental molecular physics (Lemaire et al
2002; Aleese et al
2006; Oomens et al
2006; Fridgen
2009; Polfer and Oomens
2009) to its current state, in which it is a demonstrated (bio)analytical technique for the identification of small molecules in complex mixtures (Martens et al
2017a,
b), though still limited to user facilities or state-of-the-art laboratories housing the most advanced tuneable infrared laser technology. In the past decade, IR-IS has been used to identify and characterize the molecular structures of many classes of chemical compounds including amino acids (Polfer et al
2005,
2006a; Correia et al
2008; Rodgers et al
2008; Oomens and Steill
2009; Scuderi et al
2011), nucleotides and bases (Salpin et al
2007; Chiavarino et al
2013), peptides and proteins (Balaj et al
2008; Polfer et al
2008; Yoon et al
2008; Fukui and Takahashi
2012; Martens et al
2012,
2015,
2016a; Stedwell et al
2012; Scuderi et al
2015; Dunbar et al
2017), saccharides (Martens et al
2017b; Polfer et al
2006b; Cagmat et al
2010; Contreras et al
2012; Schindler et al
2014,
2017), neurotransmitters (Lagutschenkov et al
2010,
2011), and a variety of other mainly small organic compounds and reaction products (Martens et al
2017a; MacAleese and Maître
2007; Rummel et al
2011; De Petris et al
2013,
2016b; Warnke et al
2015; Cismesia et al
2016; Seo et al
2016; Schäfer et al
2017; Gorlova et al
2017). Here we outline various ways IR-IS can currently be utilized by the metabolomics community. As an example, we focus largely on a biomarker of the IEM glutaric aciduria (GA), glutaric acid, and an isobaric metabolite, ethylmalonic acid (itself a biomarker for, among others, the IEM short-chain acyl-CoA dehydrogenase deficiency, (SCAD)). As well, we demonstrate an aspect of IR-IS that sets it apart from other tandem MS techniques; IR spectra can accurately and routinely be calculated in silico for entirely arbitrary molecular structures, unlocking the potential for reference standard free identification. As an additional example, we use a set of saccharides to illustrate how IR-IS can be applied to identifying enantiomeric compounds. Finally, in the same set of compounds we show how functional group information can be obtained for a complete unknown, which is valuable information for, as an example, narrowing down a list of candidate structures resulting from a database search result. This is especially relevant for the application of untargeted metabolomics in body fluid samples in individual patients. In such analyses many features occur that cannot be assigned by current database searches thus hampering further progression of such approaches in current clinical practice.
IR-IS is a potentially valuable new approach to the challenge of identifying the many unknowns that arise from untargeted metabolomics analyses of body fluids. Enabling annotation of the most important “unknown features” in such body fluid analyses may reveal previously unrecognized IEMs and is critical for the field to make steps forward in using untargeted metabolomics in research and diagnostics of IEMs. This article shows that, in the view of the authors, IR-IS is, already in its current state, a valuable and accessible tool for molecular identification to the metabolomics community and it outlines several paths forward to expand this role. We discuss accessibility of the IR-IS technique both in terms of sample submission to laboratories operating as international user facilities, and possibilities arising from recent technological advances enabling table-top IR-IS installations in state-of-the-art (bio)analytical laboratories.