In vivo ²H/¹³C flux analysis in metabolism research

Abstract

Identifying the factors and mechanisms that regulate metabolism under normal and diseased states requires methods to quantify metabolic fluxes of live tissues within their physiological milieu. A number of recent developments have expanded the reach and depth of isotope-based in vivo flux analysis, which have in turn challenged existing dogmas in metabolism research. First, minimally invasive techniques of intravenous isotope infusion and sampling have advanced in vivo metabolic tracer studies in animal models and human subjects. Second, recent breakthroughs in analytical instrumentation have expanded the scope of isotope labeling measurements and reduced sample volume requirements. Third, innovative modeling approaches and publicly available software tools have facilitated rigorous analysis of sophisticated experimental designs involving multiple tracers and expansive metabolomics datasets. These developments have enabled comprehensive in vivo quantification of metabolic fluxes in specific tissues and have set the stage for integrated multi-tissue flux assays.

Introduction

Understanding how metabolism is regulated under both normal and pathological conditions is essential for developing strategies to prevent, diagnose, and treat metabolic diseases. Because metabolic flux control is distributed amongst several distinct tissues and organ systems, it is best studied using an in vivo model system [1]. However, in vivo studies often infer metabolic pathway alterations indirectly from changes in enzyme expression, rather than from direct measurements of metabolic flux. This can be misleading, since metabolic enzymes are tightly regulated by allosteric feedback, post-translational modifications, and substrate availability. Therefore, mRNA or protein abundances often do not correlate strongly with pathway fluxes [2]. Furthermore, metabolic fluxes cannot be determined solely from static measurements of metabolite pool sizes [3]. Although radiotracer methods can detect tissue-specific substrate uptake or disposal with high sensitivity, they lack the data richness needed to quantitatively track the flux of substrates through intermediary metabolic pathways. In contrast, stable isotope tracers can distinguish flux contributions from different metabolic pathways based on specific labeling patterns that are detectable by mass spectrometry (MS) or nuclear magnetic resonance (NMR) spectroscopy. In vivo metabolic flux analysis (MFA) with stable isotopes requires an integration of physiological, surgical, analytical chemistry, and mathematical modeling expertise that has historically restricted its use to specialist labs (Figure 1). Here, we review advancements in each of these areas that are lowering the implementation barriers to in vivo MFA and highlight several recent applications that are expanding the limits of metabolic tracer technologies.

Figure 1.

Recent technical advancements have lowered the barriers to in vivo ²H/¹³C MFA.

(1) Simultaneous administration of multiple isotope tracers and development of intravenous infusion and sampling techniques have enabled concurrent quantification of overlapping metabolic pathways from a single experiment. (2) Improvements in the sensitivity and specificity of analytical platforms have increased the breadth of metabolite measurements. (3) Innovative modeling approaches and publicly available MFA software packages have facilitated the analysis of complex isotope labeling datasets. Created with BioRender.com.

Advancements in tracer experiment design and surgical techniques enable simultaneous quantification of multiple metabolic pathways in vivo

Proper selection of metabolic tracer(s) is critical for quantifying specific pathway activities. Single tracer experiments, while simpler to analyze and interpret, may fail to provide sufficient information to simultaneously assess flux through multiple overlapping pathways. By leveraging advancements in the field of optimal experiment design [4], modern in vivo MFA studies typically infuse a cocktail of different isotope tracers that have been tailored to the pathways of interest. However, deconvolution of the complicated enrichment patterns that result from multiple simultaneous tracers requires sophisticated mathematical models and computational tools (discussed further below). Administration of multiple stable isotopes to rodents has enabled the concurrent assessment of glycolytic/gluconeogenic, TCA cycle, and anaplerotic fluxes in liver and cardiac tissue [5,6], thus minimizing the number of animals required for comprehensive studies of in vivo metabolism. Similar approaches have been adapted to human subjects, where combined administration of ²H and ¹³C tracers has been used to quantify glucose turnover, hepatic TCA cycle activity and ketone turnover during starvation [7] or obesity [8•]. However, the high cost, administrative burden, and minimally invasive sampling required for human subjects research have meanwhile prompted innovation and miniaturization of surgical procedures required for in vivo metabolic tracer studies in rodents. For example, implantation of dual arterial-venous catheters have enabled simultaneous tracer infusion and plasma sampling in conscious, unrestrained mice [9]. Such techniques are critically important for in vivo studies of metabolism, since physiological alterations caused by anesthesia or stress at the time of sample collection can lead to unacceptable variability that obscures the experimental effects under investigation (Table 1).

Table 1.

Comparison of publicly available metabolic flux analysis software packages

Analytical breakthroughs increase breadth of isotope measurements and reduce sample size requirements

Early attempts at in vivo MFA relied on measurements of a small number of highly abundant metabolites (e.g. glucose, lactate, glutamine, alanine). Recent advancements in analytical platforms have revolutionized metabolomics studies by reducing sample size requirements and expanding measurements to include previously undetectable metabolites. Despite its inherently low sensitivity, NMR has been widely used for in vivo MFA because of its ability to assess position-specific isotope enrichments and directly differentiate between ²H and ¹³C nuclei [10]. Recently, use of hyperpolarized (HP) ¹³C magnetic resonance imaging (MRI) has improved the sensitivity of NMR by 10 000 fold, enabling in vivo probing of metabolic processes in real time [11,12]. Such advancements have set the stage for emerging applications in characterizing metabolic alterations involved in cancer [13], cardiac dysfunction [14] and neurological diseases [15]. However, a principal limitation of HP tracers is their short hyperpolarization lifetime, which typically restricts analysis to the initial steps of a pathway. For more conventional ¹³C tracer studies, improvements in decoupling approaches for 2D ¹H–¹³C heteronuclear single quantum coherence (HSQC) NMR have enabled more accurate quantification of metabolites and their in vivo enrichment [16]. An underappreciated aspect of NMR is its ability to accurately quantify low isotope enrichments (e.g. 0.1%) that are below the noise threshold of typical MS measurements [17•]. This is a major advantage for some in vivo studies, especially in human subjects, where cost and safety constraints limit the total amount of tracer that can be administered.

MS-based platforms can detect low-abundance metabolites with much higher sensitivity compared to NMR instruments and are thus becoming increasingly popular for in vivo flux characterization, especially for mouse studies and other situations where sample volumes are limited [5]. Gas chromatography-mass spectrometry (GC–MS), widely used for measuring semi-volatile compounds such as fatty acids and organic acids, has experienced a renaissance due to improvements in sample derivatization techniques [18] and addition of tandem MS (MS/MS) [19] and time of flight (ToF) capabilities [20]. The advent of electrospray ionization (ESI) has concurrently revolutionized the use of liquid chromatography (LC)–MS and LC–MS/MS in biomedical applications, chiefly due to its versatility and limited sample preparation requirements [21,22]. Recent advancements in hydrophilic interaction liquid chromatography (HILIC) have further improved isotopic spectral accuracy of various LC–MS platforms [23]. The ability of high-resolution MS to distinguish between ²H-labeled and ¹³C-labelled metabolites based on their mass defects allows contributions from multiple tracers to be directly quantified in the same sample [24]. Fragmentation of parent metabolites by MS/MS instruments can provide additional information about the position of labeled atoms in isotopically enriched metabolites [25]. Therefore, high-resolution MS/MS analysis combines some of the most attractive features of NMR—extensive positional labeling information and ability to distinguish different isotopic nuclei—with the high sensitivity that is achievable by MS instruments. Such enhancements enable accurate measurement of metabolite abundance and isotope enrichment from plasma and tissues samples collected from in vivo tracer experiments [26•,27]. The reader is referred to more in-depth reviews for an examination of the unique strengths and weaknesses of MS and NMR metabolomics platforms [28,29].

Advances in modeling approaches and software tools facilitate rigorous analysis of metabolic tracer experiments

Metabolic tracer experiments hinge on the fact that enzymes rearrange substrate atoms in unique and predictable ways. Consequently, the isotope labeling patterns that emerge in downstream products following tracer administration encode detailed information about the activity of upstream metabolic pathways and their relative fluxes. In some cases, it is possible to infer information about pathway activity or metabolite turnover based on a qualitative inspection of the isotope enrichment data. However, because metabolic pathways interact to rearrange substrate atoms in complex ways, it is often necessary to use mathematical models to determine metabolic fluxes from isotope labeling data. This is especially true when analyzing complex datasets involving multiple tracers [17•], integrating measurements of numerous metabolites and their adducts [30] or derivatives [5], and accounting for the added complexity of reversible isotope exchange [31] and secondary tracer recycling [26•] that inevitably occur during in vivo tracer studies. In these situations, intuitive inspection of the isotope labeling data can frequently lead to erroneous or incomplete conclusions.

Metabolic models used for MFA are specific to the system under investigation. Every reaction in the model is associated with an annotated enzyme or transport process, and atom rearrangements are assigned to each reaction based on its biochemical mechanism. This information is used to enumerate mass balances and isotopomer balances that describe the conservation of atoms within the metabolic network. In some cases, these balances can be distilled to closed-form equations that relate isotope enrichment measurements to pathway fluxes. While convenient to use, these simplified equations involve implicit assumptions and approximations that may not be appropriate or fully validated under the conditions of interest. An alternative approach uses least-squares regression to obtain a best-fit flux solution that provides optimal agreement between model-predicted and experimentally determined isotopomer measurements. The model parameters are iteratively adjusted, and the balance equations are repeatedly solved until the measurement residuals are minimized.

Model-based regression approaches account for the complexities of in vivo stable isotope experiments and can rigorously test assumptions used in the calculation of metabolic fluxes. The availability of flexible software tools (Table 1) for simulating metabolic tracer experiments [32], extracting isotopic enrichments from metabolomics datasets [33], and estimating fluxes from isotopomer measurements [34–36] now make sophisticated MFA workflows increasingly practicable. Comprehensive isotopomer modeling has the potential to reconcile apparently divergent results and identify flux estimates that are sensitive to methodological differences or, conversely, are robust to a variety of study designs and assumptions [26•]. The large amount of isotopomer data obtainable from each sample results in a highly overdetermined flux solution that can be statistically assessed to detect errors in measurements or model formulation [37]. Furthermore, regression approaches can accommodate a broad range of modeling assumptions, isotope tracers, and measurement inputs without the need to introduce ad hoc mathematical approximations. As a result, models for in vivo MFA can be readily adapted to a broad range of study designs and physiological conditions.

MFA technologies are essential for assessing tissue-specific metabolic regulation in vivo

Because the liver is a key metabolic hub of the body, the value of in vivo MFA has been most evident in the field of hepatology research. Decades of work by various groups has led to refined methods for assessing in vivo gluconeogenesis, glycogenolysis, anaplerosis, TCA cycle, lipid biosynthesis, fat oxidation, and ketogenesis fluxes in the liver. Recent studies have applied combinations of ²H and ¹³C tracers to assess changes in hepatic oxidative and glucose metabolism in response to dietary interventions [38,39] or pharmacological treatments [40–43] aimed at inhibiting progression of nonalcoholic fatty liver disease (NAFLD). Surprisingly, some interventions predicted to reduce NAFLD severity, such as vitamin E treatment [40] or ketogenic diet feeding [38], actually exacerbated dysregulation of oxidative metabolism in the liver. Similar techniques have also been applied to identify early indicators of hepatotoxicity due to chemical exposures in rats [44]. Additionally, MFA studies have revealed that the first-line diabetes drug metformin reduces hepatic glucose production by inhibition of fructose-1,6-bisphosphatase 1 activity [45], whereas cotadutide, a dual agonist of glucagon-like protein-1 and glucagon receptors, improved NAFLD phenotypes by enhancing hepatic glycogen flux [46].

A major advantage of in vivo mouse studies is the wide array of genetic models available, which enables specific metabolic and signaling pathways to be dissected with exquisite precision. For example, disruption of liver ketogenesis doubled TCA cycle flux and significantly increased glycogenolysis in fed animals, indicating the important role of this pathway in controlling hepatic energetics and glucose production [47•]. Other ²H/¹³C MFA studies have uncovered that reduction of liver methyltransferase activity inhibits hepatic glycogenolysis and gluconeogenesis [48], while liver-specific AMPK knockout blunts the glycogenolytic response to exercise [49]. Cappel et al. recently showed that liver-specific knockout of a key anaplerotic enzyme, pyruvate carboxylase (PC), impaired gluconeogenesis and TCA cycle flux and led to a compensatory increase in ketogenesis and renal gluconeogenesis [50••]. While mouse studies are useful for fundamental research, however, translation of findings to the clinic requires research on human subjects. The Roden group administered ²H₂O combined with [6,6-²H₂]glucose during hyperinsulinemic-euglycemic clamps to examine the acute effects of a Mediterranean-like diet [51] or a diet high in saturated fats on hepatic fluxes [52]. Another study found that increased lipogenesis in NAFLD patients was not stimulated by insulin action but rather by an increase in lipogenic substrate availability [53]. Stable isotope tracer studies performed under ketotic (24-hour fasted) conditions revealed that NAFLD patients had elevated oxidative metabolism and diminished capacity to safely dispose of excess fatty acids as ketones [8•]. In contrast, NMR-based studies in healthy individuals after a 60-hour fast found that starvation led to decreased endogenous glucose production and mitochondrial oxidation that was attributed to a reduction in alanine turnover [7].

In addition to NAFLD, in vivo MFA studies are rapidly advancing our understanding of metabolic adaptations that occur in other contexts such as cancer [54], skeletal muscle insulin resistance [55], and cardiovascular disease. As an illustration of the latter, NMR-based stable isotope tracer studies were used to examine the fate of glucose and its role in fueling heart metabolism during pathological cardiac hypertrophy [56] or acute and chronic hypoxia [57]. Stable isotope tracer studies have also proven useful in drug discovery and development, such as assessing the effects of the nicotinic acid receptor agonist Acipimox on cardiac contractility in rats [58]. Hyperpolarized ¹³C-NMR based analysis of metabolic fluxes is also becoming a useful diagnostic tool. In human subjects with type 2 diabetes, it was observed that metabolic flux through cardiac pyruvate dehydrogenase was significantly reduced in both fed and fasted states but significantly increased after an oral glucose challenge compared to age-matched healthy controls [14]. This study was the first attempt to develop a noninvasive method to assess early onset of diabetic cardiomyopathies in type 2 diabetes patients.

Towards multi-tissue MFA of nutrient utilization and disposal in vivo

Building from the success of prior studies to examine flux in specific tissues, recent efforts have applied stable isotopes to assess the carbon sources that fuel core metabolic cycles in vivo. Meta-analysis of data from fifteen nutrient tracers revealed that the majority of circulating carbon flux is carried by two key cycles: glucose-lactate and triglyceride-glycerol-fatty acid [59••]. Futile cycling through these pathways can complicate the interpretation of in vivo tracer data and necessitates the use of rigorous mathematical models to avoid common pitfalls [26•,31,60•]. ¹³C-lactate infusions in mice [61] and human subjects [62] have revealed that circulating lactate acts as a major carbon shuttle between numerous tissues in the body and may serve as the primary fuel for TCA cycle metabolism in some tissues. Lactate is produced rapidly from glucose by certain muscle tissues, whereas many other tissues rely heavily on stored glycogen rather than glucose to sustain glycolytic metabolism in vivo [63•]. Furthermore, because of the dominant contribution of recycled lactate to liver gluconeogenesis, one recent study claims that glycerol is the major contributor of ‘new’ glucose synthesized during fasting [64•]. These innovative strategies to integrate numerous isotope tracer measurements from various plasma and tissue compartments are challenging long-standing dogmas held by the physiology community. We expect they are just the start of an emerging trend toward multi-tissue MFA studies that rely upon metabolic models with gradually expanding details of intermediary metabolism and increasing power to describe a bevy of metabolite measurements and resolve in vivo fluxes that are not observable without the use of stable isotopes (Figure 2).

Figure 2.

In vivo ²H/¹³C MFA has been used to address many important questions about metabolic pathway dynamics and regulation.

(1) Assessment of genetic factors, dietary regimens, and treatment interventions in the control of in vivo metabolism. (2) Quantification of tissue-specific nutrient utilization and intermediary fluxes under normal versus disease conditions. (3) Tracing multi-tissue metabolic networks to elucidate inter-organ exchange fluxes and in vivo interactions. Created with BioRender.com.

Conclusion

Many important questions about pathway dynamics and regulation cannot be answered without the use of stable isotopes. While studies of cultured cells and perfused organs improve our understanding of tissue-specific metabolism, they fail to capture important inter-organ communication through hormonal cues and nutrient exchanges. Recent surgical, analytical, and computational advancements have enabled comprehensive in vivo quantification of metabolic fluxes in specific tissues, as well as whole-body assessment of turnover rates of circulating metabolites and their contributions to multiple tissues. Future advancements in the field may enable simultaneous determination of not only systemic fluxes but also intermediary metabolic fluxes within multiple tissues. The recent emergence of several new NMR and MS technologies also presents an uncapitalized opportunity to assess metabolic fluxes with greater specificity and precision than ever before. However, leveraging these advancements for MFA will require generalized software packages to model the isotopic data obtained from high-resolution MS and MS/MS instruments, as well as advanced NMR systems, and the capability to integrate isotopomer measurements from these various platforms. As these software tools become increasingly available, they will enable a powerful complement of measurements and analyses for comprehensive assessment of in vivo metabolism using stable isotopes. Given the ever-increasing interest in probing intracellular metabolic fluxes and their role in various diseases, we expect in vivo MFA to continue expanding its footprint in pre-clinical and clinical metabolic research.