Inflammation and metabolism are closely intertwined processes that influence each other in multiple ways (1, 2). During an inflammatory response, the body requires additional energy to support the activation and proliferation of immune cells involved in the immune response and successful healing. Metabolic pathways are redirected to generate the necessary energy and substrates for these processes (3), while immune cells, especially macrophages and T cells, undergo metabolic reprogramming during inflammation (4). On the other hand, inflammatory cytokines, such as interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-α), can affect insulin sensitivity, glucose, and lipid metabolism (5), potentially leading to insulin resistance. In line with this, chronic low-grade inflammation is linked to various metabolic disorders, including obesity, type 2 diabetes, and cardiovascular diseases (6, 7).

In order to harness this interplay for clinical decision-making, as well as to further delineate the mechanisms on how inflammation and metabolism interact, noninvasive approaches that monitor both processes in vivo would be highly desirable. In this context, the metabolism of glucose is especially of major interest, since, in particular, in the initial inflammatory phase, infiltrating immune cells mainly rely on glycolysis to cover their energy demand (8, 9). Positron emission tomography (PET) using [2-¹⁸F]fluorodeoxyglucose (FDG) is an established and sensitive tool for glucose imaging (10), which has also been applied for visualization of inflammation via the enhanced glucose uptake of the affected tissue (11). However, it has several limitations since it requires radioactive tracers and is restricted to monitor glucose uptake, whereas further metabolic turnover of the ingested glucose remains unclear (12). Furthermore, an increased glucose uptake is not specific for inflamed tissue and occurs also, e.g., in cancer [known as the Warburg effect (13)] and thus is a rather indirect readout for inflammation.

In the present study, the power of multinuclear magnetic resonance (MR) techniques was used to specifically resolve inflammatory and metabolic processes in parallel but at different levels: The ¹⁹F nucleus was employed to track immune cell trafficking, ²H (deuterium) for monitoring metabolic turnover, and conventional ¹H MR imaging (MRI) for anatomical assessment of the inflamed tissue. ¹⁹F MRI emerged over the last decade as a background-free approach for inflammation imaging (14, 15), while ²H MR spectroscopy (MRS) is a novel, noninvasive method for tracking the pathway of deuterated substrates, which has yet be mainly used for analysis of cerebral or cancer metabolism (16–18). Here, these approaches were combined for the first time to monitor metabolic changes via ²H MRS in inflamed tissue identified by ¹H/¹⁹F MRI. To this end, an easy and reproducible murine inflammation model was applied, making use of an implanted Matrigel plug doped with LPS (19) for an efficient recruitment of circulating immune cells into this experimental inflammatory focus.

For induction of defined inflammatory conditions, we made use of a previously described murine model with neck implantation of an LPS-doped Matrigel plug, while using PBS-doped plugs as respective negative control (19). The successful formation of an inflammatory hotspot within the neck of mice was verified by ¹H/¹⁹F MR inflammation imaging (14) 1 day after implantation of the doped Matrigel plugs. To this end, PFCs that are preferentially phagocytosed by circulating neutrophils/monocytes were injected intravenously for visualization of the infiltrating immune cells. Figure 1A shows the anatomical location of the Matrigel plug that clearly emerged as a bright oval structure in T2-weighted ¹H MRI of the neck region. Merging of subsequently acquired ¹H and ¹⁹F images ( Figure 1B ) demonstrated the presence of ¹⁹F-loaded immune cells predominantly in the border region of the LPS-doped Matrigel plug, while no ¹⁹F signal was found in PBS-doped control plugs. These in vivo findings were underpinned by both histology and flow cytometry corroborating the marginal location of immune cells in the plug ( Figure 1C ) and revealing neutrophils as the predominant cell population infiltrating the plug at this early time point after implantation ( Figure 1D ).

After confirming the presence of inflammatory foci ( Figure 2A ), mice were placed with their neck on a 12 × 8 mm² transmit/receive ²H surface coil inserted into a 30-mm ¹H saw resonator ( Figure 2B ). This setup allowed the unambiguous localization of the Matrigel plug by ¹H MRI ( Figure 2C left) and the sensitive acquisition of ²H MR spectra with the surface coil, whose penetration depth and excitation profile essentially encompassed the Matrigel plug and surrounding scapular muscles ( Figure 2C right). After acquisition of baseline spectra, mice received an i.p. bolus injection of 2 mg/g [6,6-²H₂]glucose per body weight followed by continuous monitoring of ²H MR spectra over 60 min ( Figure 2D ). Figure 3A illustrates the route of the injected ²H label (blue) through glycolysis and TCA cycle with the ²H MRS-detectable metabolites (in the millimolar range) highlighted in red, i.e., glucose, lactate, and glutamate/glutamine (Glx)—the latter cannot be spectrally resolved in vivo because of the broad linewidth of ²H MR spectra. Importantly, during formation of the enol form of acetyl-CoA prior to entry into the TCA cycle, one of the two deuteria can be split off (with a probability of 2/3)—this is eventually found in water and yields only a monolabeled citrate and Glx (not shown in the schematic).

Baseline spectra showed a prominent signal at 4.7 ppm ( Figure 3B bottom lanes) caused by the natural abundant ²H in water. Injection of [6,6-²H₂] glucose gave rise to a fast increase of the corresponding ²H signal at 3.8 ppm and the subsequent appearance of the downstream metabolites Glx (i.e., glutamine + glutamate) and lactate at 2.4 and 1.3 ppm, respectively, accompanied by a continuously increasing water signal. The temporal development of ²H MR spectra over the entire observation period of 60 min is illustrated in Figure 3B with typical examples for mice with Matrigel plugs doped with LPS (right) and PBS as control (left). As can be recognized, the glucose (Glc) signal declined much quicker in the presence of LPS while concomitantly lactate levels were clearly elevated as compared to control conditions. Quantification of spectra for n = 9–12 independent experiments ( Figures 4A–D ) confirmed these findings and calculation of Lac/Glc ratios over the entire observation period after application of the labeled glucose (LPS 1.18 ± 0.08 vs. PBS 0.90 ± 0.12; p < 0.05) further supported the notion of an enhanced lactate formation at the expense of increased glucose consumption under inflammatory conditions. Of note, this was also accompanied by a significantly lower incorporation of ²H label into water ( Figure 4D ).

Taking the naturally occurring ²H signal from water ( Figure 3B bottom lanes) as an internal reference, absolute turnover rates can also be calculated: Assuming a content of 110 M hydrogen in H₂O, an approximately 75% tissue water content (24), and a ²H natural abundance of 0.0115% in water (25, 26), the detected baseline signal reflects an amount of ~9.5 mM deuterons (27). Considering now the total increase of the water signal ( Figure 4D , middle) over the observation period of 60 min, the deuterium turnover from [6,6-²H₂]glucose to water is calculated to be 0.81 ± 0.07 mM/min under control conditions, while it was significantly diminished to 0.64 ± 0.04 mM/min under LPS exposure (p < 0.05, n = 9–12).

In the present work, ²H MRS was applied in combination with ¹⁹F MR inflammation imaging in vivo to monitor the metabolic fate of deuterated glucose within neutrophil-infiltrated inflammatory hotspots noninvasively and in real time. The findings of this study demonstrate that ²H MRS can indeed be successfully used to reveal metabolic signatures in inflammatory environments, specifically identified by hot spot ¹⁹F MRI together with anatomical ¹H MRI.

The obtained results clearly point to an enhanced utilization of glycolysis for energy production in the region of the inflammatory hotspot. Increased lactate levels and enhanced glucose turnover are classical indicators for a predominant anaerobic metabolism, which is further supported by the lower flux of the ²H label to water. In the long term, all deuterons of [6,6-²H₂]glucose will end up in the body water pool, but in the initial phase after bolus injection, the highest probability for the ²H label being transferred to water is given when it enters the aerobic pathway: if the label runs through the TCA cycle, it can be cleaved off either during formation of citrate or when it does not leave the cycle via α-ketoglutarate, but instead passes into fumarate ( Figure 3A ). Thus, the diminished ²H incorporation into water is overtly indicative for a lower utilization of glucose via the TCA cycle. In contrast, under control conditions, glucose turnover within the field of view covered by the surface coil ( Figure 2C ) is mainly determined by aerobic metabolism of the surrounding scapular muscles.

While this finding is not surprising per se, it nonetheless highlights the potential of multinuclear MR approaches to discriminate complex immunometabolic processes at separate levels in parallel (here, ¹H, ²H, and ¹⁹F for anatomy, metabolism, and immune cells, respectively). Of course, compared to FDG-PET (10), the present approach is restricted to metabolites that are present in the millimolar range, but it lacks any harmful radiation and also allows monitoring of downstream metabolites as well as absolute turnover rates. Alternative MR approaches relying on the ¹³C nucleus offer better spectral resolution than ²H, but either require substantial longer acquisition times even without volume selection (normal ¹³C) or are limited to non-ideal substrates (e.g., pyruvate) as well as very short temporal windows of observation (hyperpolarized ¹³C; approximately 1 min). Nevertheless, owing to the dramatically enhanced sensitivity of the latter method, it provides superior spatial resolution within this time frame (28). On the other hand, a variety of deuterated substrates are readily available commercially for ²H MRS and applications for deuterated acetate, choline, fumarate, and β-hydroxybutyrate have already been described (29, 30), but the field is definitely still in its infancy here. Of note, ongoing advances in technology, such as denoising, undersampling, or indirect readout [as used in (28, 30, 31)] will further improve the sensitivity for detection of deuterated metabolites.

In the present study, a ²H surface coil was utilized together with a ¹H volume resonator—while this setup provided excellent coverage and superior sensitivity for the near-surface inflammation model used, it is not suitable for metabolic analysis of deeper organs or tissue. However, this approach can be further developed by designing triple-tuned volume resonators to allow simultaneous acquisition (32) of ¹H/²H/¹⁹F MR images for true temporal co-registration of inflammatory/metabolic processes without penetration limits—not only for basic research to analyze their interwoven relationships in vivo, but also in the long run also for clinical decision-making in immunologic and metabolic disease states, as the feasibility of both ¹⁹F inflammation and deuterium metabolic imaging in the clinical setting has already been demonstrated (16, 33–36).

In summary, the results of this study demonstrate that ²H MRS can be successfully used to reveal real-time metabolic signatures in inflammatory environments—specifically identified by ¹H/¹⁹F MRI. In contrast to the current gold standard FDG-PET that detects glucose uptake only, the present approach discriminates inflammation and metabolism at different levels in parallel and enables the much more meaningful assessment of metabolic turnover through glycolysis and TCA cycle with analysis of downstream metabolites.

In the future, this approach can be employed to noninvasively and longitudinally characterize in depth the immunometabolic interplay in diseases, such as diabetes or obesity, but also its impact on autoimmunity, (innate) immune memory, tumor microenvironment, and hematopoiesis for identification of new therapeutic targets. As the MR techniques used in this study could be readily transferred to human scanners, they could also serve to assess the degree of immunometabolic responses in patients to tailor individual treatment regimens.

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

The animal study was approved by Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen. The study was conducted in accordance with the local legislation and institutional requirements.

VF: Data curation, Formal Analysis, Investigation, Validation, Visualization, Writing – original draft. ST: Data curation, Formal Analysis, Investigation, Validation, Visualization, Writing – review & editing. PB: Methodology, Resources, Writing – review & editing. MG: Conceptualization, Project administration, Supervision, Writing – review & editing. UF: Conceptualization, Funding acquisition, Project administration, Supervision, Validation, Visualization, Writing – review & editing.