Cellular oxidative phosphorylation involves the production of ATP and MR-detectable metabolic water H₂¹⁷O by consuming glucose and oxygen. In recent development, oxygen-17 NMR has been utilized to obtain the cerebral metabolic rate in vivo by means of inhalation experiments with ¹⁷O₂-enriched gas^[1,2]. The alternative approach of ¹⁷O NMR presented in this work investigates the cellular metabolism on a more fundamental level by directly supplying cells with ¹⁷O-labeled glucose. The potential production of H₂¹⁷O via the metabolization of ¹⁷O-labeled glucose can be attributed to physiological processing.

The aim of this work was to perform ¹⁷O NMR experiments of cellular metabolism by online monitoring of cellular ¹⁷O-labeled glucose consumption in a MR-compatible microbioreactor^[3] platform.

Cell culture and medium: The hepatoma cell line Hep G2 (ATCC, HB-8065, Manassas, VA, U.S.A.) was cultured in either Dulbecco’s Modified Eagle Medium (DMEM) without glucose or Minimal Essential Medium (MEM) with natural glucose (without ¹⁷O-enrichment) and 0.1% phenol red, depending on the experimental purpose. Furthermore, an amount of 1 g/L [6-¹⁷O]glucose with 47% ¹⁷O-enrichment on position 6 (NUKEM Isotopes Imaging GmbH, Alzenau, Germany) was added to the DMEM. Both types of medium have been supplemented with 10% FBS, 1% Penicillin-Streptomycin, 1% Glutamine, 1% non-essential amino acids and 1% sodium pyruvate.

Bioreactor setup: Cells were cultured in a bioreactor containing a 3^D-KITChip^[3]. Medium of either type was pumped through the setup with 400 μl/min. A three-way-valve was installed to facilitate switching between MEM and DMEM.

Hardware: Measurements were conducted on a 9.4 T MR system (BioSpec, Bruker, Ettlingen, Germany). The custom-made RF surface coil was designed to match the dimensions of the bioreactor (c.f. Figure 1).

Measurement protocols: In each experiment, the bioreactor was perfused by two boluses of DMEM separated by MEM. The four measurements shown in Figure 2 have partially different protocols. For each data point, global free induction decay (FID) readout was averaged 2048 (2a-c) or 1024 (2d) times with a repetition time of 73 ms for each average. Signal intensity was obtained from integration of the spectrum. Such FID readout was repeated 52 (2a), 76 (2b-c) or 119 times (2d), resulting in a scan time of 130 min, 189 min or 148 min, respectively.

Each signal course in Figure 2 was normalized to data from the first ten minutes of the measurement and mapped with its perfusion timeline. Time periods during which cells are perfused with either MEM or DMEM overlap due to the gradual nature of medium replacement.

Measurements carried out with cells (2a-c) showed a distinctively different signal course compared with measurements without cells (2d). In the former case, a signal increase of 10–17% (c.f. Table 1) could be discerned to clearly match the in-flow of DMEM containing [6-¹⁷O]glucose. The decrease towards the baseline after the first DMEM bolus as well as the subsequent formation of the second peak showed that [6-¹⁷O]glucose is safely applicable without influencing the cellular vitality. More importantly, the reproducibility was evident.

In measurements without cells, the signal increase during the DMEM phases with [6-¹⁷O]glucose was 2–3%. This small increase could be explained by a combination of two reasons. The natural abundance of H₂¹⁷O varies in the range of 0.037–0.040%^[4]. This relative difference of up to 8% in the most extreme cases might have been responsible for intrinsically different concentrations of the two types of medium. Furthermore, recombination of [6-¹⁷O]glucose with water in the medium under the specific condition (i.e. pH, temperature) of the experiments might have led to an increase in non-metabolic H₂¹⁷O content. Verifications of these arguments will be sought in future experiments. This signal increase of 2-3% was 3-5 times lower than that measured in the presence of cells and only slightly more significant than statistical fluctuation; it was deemed negligible for now. Therefore the observed peaks in the presence of cells could be attributed to the by-product of the metabolization of [6-¹⁷O]glucose.

The in vitro setup introduced in this work showed that there was a direct link between the metabolization of [6-¹⁷O]glucose and the formation of H₂¹⁷O. Precise dosage of [6-¹⁷O]glucose could be employed to provide evidence of cellular vitality. Signal increase was observed to match the perfusion timing accurately with sufficient time resolution. Furthermore, the implementation of [6-¹⁷O]glucose has been demonstrated to not intervene with the inherent cellular physiology, thus proving the method to serve as a neutral observation platform. In conjunction with other x-nuclei methods, the presented ¹⁷O NMR setup has the potential to aid the modeling of fundamental physiological processes and to become a key element in cellular vitality assessment applications.