Real‐Time NMR Spectroscopy for Studying Metabolism

Abstract Current metabolomics approaches utilize cellular metabolite extracts, are destructive, and require high cell numbers. We introduce here an approach that enables the monitoring of cellular metabolism at lower cell numbers by observing the consumption/production of different metabolites over several kinetic data points of up to 48 hours. Our approach does not influence cellular viability, as we optimized the cellular matrix in comparison to other materials used in a variety of in‐cell NMR spectroscopy experiments. We are able to monitor real‐time metabolism of primary patient cells, which are extremely sensitive to external stress. Measurements are set up in an interleaved manner with short acquisition times (approximately 7 minutes per sample), which allows the monitoring of up to 15 patient samples simultaneously. Further, we implemented our approach for performing tracer‐based assays. Our approach will be important not only in the metabolomics fields, but also in individualized diagnostics.

Abstract: Current metabolomics approaches utilizec ellular metabolite extracts,a re destructive,a nd require high cell numbers.W ei ntroduce here an approach that enables the monitoring of cellular metabolism at lower cell numbers by observing the consumption/production of different metabolites over several kinetic data points of up to 48 hours.O ur approach does not influence cellular viability,asweoptimized the cellular matrix in comparison to other materials used in avariety of in-cell NMR spectroscopyexperiments.Weare able to monitor real-time metabolism of primary patient cells,which are extremely sensitive to external stress.Measurements are set up in an interleaved manner with short acquisition times (approximately 7minutes per sample), which allows the monitoring of up to 15 patient samples simultaneously.Further, we implemented our approach for performing tracer-based assays.O ur approach will be important not only in the metabolomics fields,but also in individualizeddiagnostics.
Over the last decade,m etabolomics,t he study of cellular metabolism, has become increasingly important. Metabolomic studies address how cells fulfil their energy needs: metabolic pathways for energy production are elucidated by quantification of metabolite concentration. Modes of metabolic rewiring that cells undergo to overcome nutrient deprivation and cellular stress can be detected.
Recently,i th as been shown that changes in metabolism are av ulnerability that can be targeted in cancer cells (reviewed in ref. [1,2]). In fact, the metabolism of malignant cells is different from healthy cells as these cells reprogram their metabolic pathways to fulfil the high energy demands of highly proliferating cells and to develop resistance to drug treatment. [3,4] Metabolism targeting is becoming ac ore research area in therapeutics development for different cancers,including acute myeloid leukemia (AML), ahematological malignancyt hat results in uncontrolled cellular proliferation. [5] In fact, several inhibitors of metabolism are currently being evaluated in clinical trials (l-asparaginase and CPI-613) [4,[6][7][8] and some others have already been approved for AML treatment (Venetoclax and isocitrate dehydrogenase (IDH) inhibitors). [9,10] Nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry are prime technologies to phenotype the metabolism of different cancer cell types.NMR spectroscopy provides remarkably reproducible results,g reat ease of sample preparation, and the possibility of preserving samples over extended periods of time. [11] Using 1D and 2D isotopefiltered experiments,d ifferent metabolic pathways can be simultaneously tracked when using isotope-labeled precursor metabolites. [12] Currently,N MR metabolomics samples are prepared by harvesting cells,e xtracting their metabolic content, and quantifying the change in their concentration. [13] However,a sm etabolism is ah ighly dynamic process,t he concentrations can change rapidly over time which makes it difficult and labor-intensive to make metabolite extracts at different time points to accurately assign metabolic fluctuations over atime course.
Another layer of complexity is added when investigating metabolic profiles under different conditions (for example, adaption to hypoxic conditions), where one needs to differentiate between acute metabolic response,a daptations,a nd chronic rewiring in the cells.Up-to-now,such studies require high cell numbers (approximately 1 10 7 cells) [14] for NMR spectroscopic analysis,w hich are often difficult to obtain when studying primary patient cells,m aking NMR spectroscopy unattractive for this kind of samples.M oreover, materials used for sample preparation, in particular agarose gels in previously described methods for monitoring live-cell metabolism, [15][16][17][18] can be cell-unfriendly,c an further lead to reduced metabolite diffusion rates and induce environmental stress that obscures the real metabolic fingerprint of the cell. [17] Such agarose preparations,h owever, are commonly used also for in-cell NMR spectroscopy,a lthough it may compromise cell viability. [19,20] To address these challenges,w ei ntroduce an automated real-time NMR spectroscopy approach, which enables live monitoring of metabolism changes in viable AML cells.T he newly developed method allowed us to monitor the metabolism of primary patient cells in an automated fashion, extending this method to individualized diagnostics required for personalized medicine approaches.I np rinciple,o ur method allows for as imultaneous interleaved measurement of several patient samples (10-15 samples), due to the short NMR measurement time of 7minutes.For ethical reasons,we demonstrate this experimental schedule,h owever, not on different primary patient samples but apply the acquisition scheme to primary cells from asingle patient.
Different to previous experimental designs, [13] the newly developed approach is not destructive,s ince cells are preserved and used again for other experiments or diagnostic procedures (low TMSP (trimethylsilylpropanoica cid) and D 2 Oc oncentrations are reported to be non-toxic). [21,22] Furthermore,itneeds asmall number of cells (approximately 5 10 5 cells or even fewer) compared to (approximately 1 10 7 cells) required for current metabolites extraction settings.
As ample changer supplemented with temperature control typically set to 37 8 8Ca nd ar obot that alternates the samples without temperature change into the spectrometer has been used ( Figure 1A). Several spectra are recorded over time to detect changes in the uptake and efflux of the individual metabolites ( Figure 1B). To prevent cell sedimentation in the NMR tube,w eo ptimized our approach by preparing samples in acell culture media with ac ell-friendly matrix. We first investigated the impact of agarose,awidely used material for NMR metabolomics and in-cell experiments.W eo bserved as ignificant impact on cellular ATP levels (a measure of viability,F igure 2A). To overcome this, we replaced agarose by 40 %m ethylcellulose media as am atrix. Methylcellulose medium is usually used for assays of highly sensitive cells,i ncluding stem cells assays.I ndeed, methylcellulose did not influence the cellular viability or metabolism ( Figure 2B). In fact, ac oncurrently emerging  report has shown the advantages of using methylcellulose for studying protein interactions in living cells. [23] Ther eliability of the approach was further validated by investigating the cellular responses to interferences with ac ertain signalling/ survival pathway known to influence cellular metabolism. Internal tandem repeats (ITDs) in the fms-like tyrosine kinase (FLT3-ITD,c ommonly mutated in AML [5] )a re reported to induce high glucose uptake.F LT3i nhibition leads to reduction in glucose consumption and hence,reduced glycolytic activity. [24,25] Midostaurin, an FDA-approved FLT3 inhibitor, [26] was used and its effects on metabolism were evaluated using the presented approach. TheF LT3-ITD positive cell line MOLM-13 showed the expected druginduced metabolic shifts of reduction in glucose uptake (higher retention of glucose in the media) in the midostaurin group ( Figure 2C).
To measure O 2 and pH levels in the NMR tube,w e performed measurements of both 48 hours after plating the cells.O xygen levels were between 1.4 %a nd 3.2 %. Physiologically,1 -6 %o xygen is usually the normal oxygen level encountered by the leukemic cells in the bone marrow, depending on the distance from the endosteal niche. [27,28] Hence,o ur technique reflects metabolic measurements of physiological-like conditions.T he pH value was around 7.4, 0.2 pH units below that of the standard cell culture conditions (pH 7.6), thus,within the normal range of blood pH. [29] Moreover,this approach was successfully applied to study mononuclear cells isolated from bone marrow aspirates of an

Angewandte Chemie
Communications 2306 www.angewandte.org AML patient. Ty pically,A ML patient cells are highly sensitive to culture conditions.O nce put in the methylcellulose conditions within the NMR sample tube,c ells behaved completely normal in terms of their metabolic activity.T hey consumed glucose and produced lactate,i ndicating active glucose energy metabolism ( Figure 3). Furthermore,t hey consumed glutamine indicating energy and redox metabolism, branched chain and other amino acids as building blocks for protein biosynthesis,e nergy production and/ or DNA methylation activity.
Finally,w ee xtended the application of our method to perform tracer-based assays.S ince traditional HSQC experiments are time consuming,w hich undermines the real-time characteristics of this approach, apseudo-2D experiment was implemented. Utilizing adouble filtering approach allows the differentiation between protons attached to 12 Co r 13 C, as described. [30] Upon labeling with U-13 C-glucose,anincreased 13 C-glucose ratio in the cells was observed (Figure 4). Quantification of the changes of glucose concentration was subsequently translated in an increased label incorporation in lactate and alanine.
Here,weintroduce acell-friendly approach that facilitates the studying of cellular metabolism in areal-time manner. We demonstrated that this approach does not affect the cellular viability and we could successfully use it to study extremely sensitive cells,such as primary AML cells,thus bringing NMR spectroscopy from the bench closer to the bedside.Moreover, we demonstrated that cells are responsive to any expected changes due to small molecule inhibitors or external stimuli. Finally,w ei mplemented 13 Cf ilter experiments where we could assign in at racer-based setting the label incorporation into the downstream metabolic pathways.