Hyperpolarized Micro-NMR Platform for Sensitive Analysis of In Vitro Metabolic Flux in Living Cells

Abstract

Metabolism represents an ensemble of cellular biochemical reactions, and thus metabolic analyses can shed light on the state of cells. Metabolic changes in response to external cues, such as drug treatment, for example, can be rapid and potentially an early indicator of therapeutic response. Unfortunately, conventional techniques to study metabolism, such as optical microscopy or mass spectrometry, have functional limitations in specificity and sensitivity. To address this technical need, we developed a sensitive analytical tool based on nuclear magnetic resonance (NMR) technology, termed hyperpolarized micro-NMR, that enables rapid quantification of multiple metabolic fluxes in a small number of cells, down to 10,000 cells, nondestructively. This analytical capability was achieved by miniaturization of an NMR detection coil along with hyperpolarization of endogenous metabolites. Using this tool, we were able to quantify pyruvate-to-lactate flux in cancer stem cells nondestructively within 2 min, which has not been possible with other techniques. With further optimization, we envision that this novel device could be a powerful analytical platform for sensitive analysis of metabolism in mass-limited samples.

1. Introduction

Metabolism fundamentally regulates our daily lives, as it not only generates energy and building blocks but also plays a major role in redox balance [1] and epigenetic regulation [2]. Due to its relevance in biochemical homeostasis, altered metabolism has been implicated in a wide range of diseases, including cancer [3]. For example, a common metabolic feature of cancer cells is significantly upregulated glucose consumption and lactate generation, regardless of oxygen availability, commonly referred to as the “Warburg effect.” This feature allows cancer cells to meet the metabolic needs for deregulated proliferation [4] and evade immune surveillance in the tumor microenvironment [5]. Targeting metabolic vulnerabilities of cancer cells has emerged as a promising therapeutic approach, with numerous clinical trials inhibiting metabolic enzymes or metabolite transporters underway (e.g., NCT02632708, NCT03875313, NCT01791595).

The recent surge of interest in metabolic dysfunction has led to the advancement of technologies aimed at interrogating metabolism. Particularly, nuclear magnetic resonance (NMR) spectroscopy and magnetic resonance imaging (MRI) techniques in conjunction with hyperpolarization of nuclear spins have emerged as a novel approach, as they allow nondestructive analysis of metabolic “flux” [6]. Metabolic flux, the rate of a metabolic reaction, indicates the activity of a metabolic pathway of interest at a given moment and sensitively changes in response to various factors, including drug treatment. Flux analysis can be beneficial for evaluating treatment response before clinicopathological changes occur and for investigating how cells rewire their metabolism under varying environmental conditions. Hyperpolarization, which increases the polarization level of target nuclei beyond the level at thermodynamic equilibrium, has been reported to enhance the sensitivity of NMR/MRI detection methods by a factor of more than 10,000 [7]. Multiple approaches have been developed to achieve hyperpolarization, including parahydrogen-induced polarization and dissolution dynamic nuclear polarization (dDNP) [8]. dDNP, in particular, achieves the highest polarization level among hyperpolarization methods—up to 70% [9]—and utilizes endogenous metabolites, including pyruvate [10-13] and glutamine [14]. Due to these advantages, several ongoing clinical trials with dDNP technology seek to evaluate its capacity to diagnose disease and monitor treatment (e.g., NCT03830151, NCT03502967, NCT02526368, NCT02647983). However, even with dDNP technology, numerous cells (order of 10⁷ cells) are required to analyze metabolic fluxes, and it has thus been challenging to apply dDNP to studying metabolism in mass-limited samples, including stem cells or tumor organoids.

To address this limitation, we have developed the hyperpolarized micro-NMR platform, which allows for metabolic flux analysis in a small number of cells (down to 10⁴ cells) [15]. This sensitive tool is based on the development of a miniaturized NMR detection coil (micro-coil) optimized for hyperpolarized experiments. As shown in Fig. 1a, the micro-coil is fabricated inside a block made of elastomer, which is mounted on top of our custom-designed printed circuit board. Because the micro-coil, with a detection volume of ~2 μL, is integrated with a microfluidic channel, it is easy to load and unload liquid samples, such as a mixture of cell suspension and hyperpolarized substrates (Fig. 1b). The metabolic flux of pyruvate-to-lactate conversion can be calculated based on the changes in NMR signals. The NMR signal of hyperpolarized molecules decays exponentially with a spin-lattice relaxation constant T₁; signal decays faster than T₁ indicate that a hyperpolarized substrate is being consumed. Figure 2 shows an example of ¹³C-NMR spectra acquired from the hyperpolarized micro-NMR system, as described in Fig. 1b: [1-¹³C] pyruvate signal decreases faster than its exponential decay (Fig. 2b), which indicates that pyruvate was being consumed by these cells, and [1-¹³C] lactate signal decreases slower than its exponential decay (Fig. 2c), which indicates that lactate was being generated by cells (from pyruvate).

Fig. 1.

Hyperpolarized micro-NMR platform, (a) Optical picture of the micro-NMR circuit. Inset shows the micro-coil embedded in a PDMS block, (b) Assay schematic with the hyperpolarized micro-NMR platform. Total assay can be finished within 2 min. B₀, magnetic field. Adapted from reference [15]

Fig. 2.

Analysis of pyruvate-to-lactate metabolic flux in UOK262 cells with the hyperpolarized micro-NMR system, (a) ¹³C-NMR spectra acquired from the hyperpolarized micro-NMR system with [1-¹³C] pyruvate and UOK262 cells (renal cancer cell line). The cell number in the micro-coil was approximately 50,000. Pyr pyruvate, Pyr-H pyruvate hydrate, Lac lactate. (b, c) Comparison of experimental and simulated data of NMR signal from hyperpolarized molecules, [1-¹³C] pyruvate (b) and [1-¹³C] lactate (c). The simulated data models a spin-lattice relaxation decay with a relaxation constant T₁. Adapted from reference [15]

2. Materials

2.1. Micro-coil

1. SYLGARD-184 Silicone Elastomer Clear (Dow).

2. 32-AWG magnetic wire (Belden 8056).

3. Metal rod with a diameter of 1.4 mm.

4. Disposable plastic petri dish with a diameter of 60 mm.

5. Vacuum desiccator.

6. All-purpose glue (e.g., Krazy glue).

7. Razor blade.

2.2. Micro-NMR Circuit

1. Custom-designed printed circuit board.

2. Non-magnetic passive circuit components (e.g., resistor, capacitor, and SMA connector).

3. Soldering equipment.

2.3. Hyperpolarization of Metabolites

1. [1-¹³C] pyruvic acid (Sigma-Aldrich).

2. AH-111501 radical (GE).

3. SPINlab polarizer (GE).

4. 10N sodium hydroxide solution.

5. Trizma hydrochloride (Sigma-Aldrich).

2.4. Hyperpolarization Experiments

1. Centrifuge.

2. MRI scanner (preclinical or clinical version).

3. Methods

3.1. Fabrication of Micro-coil

1. Magnetic wire is tightly wound around a metal rod, and its length determines the sample volume of a micro-coil (Fig. 3a). To maintain the shape of the wire around the rod, a tiny amount of super-glue can be applied to the wire, not between the wire and rod.

2. A SYLGARD-184 Silicone Elastomer Clear kit has two components: elastomer base and curing agent. They are mixed with a ratio of 10:1 (elastomer base:curing agent) in a disposable petri dish (or any disposable container). As the mixing step generates bubbles inside the mixture, the petri dish is put in a vacuum desiccator to remove all the bubbles, which takes approximately 30 min.

3. The elastomer mixture is slowly poured over the wire-wound rod and put inside a vacuum desiccator again, as this pouring step also generates bubbles.

4. Once all the bubbles are removed, the wire-wound rod embedded inside the elastomer mixture is put into an oven at 90 °C overnight, which makes the mixture hardened. See Note 1.

5. The hardened mixture is cut with a razor blade to a certain size and the metal rod needs to be slowly pulled out, which makes a microfluidic channel passing through the micro-solenoid coil (Fig. 3b).

Fig. 3.

Fabrication steps of micro-coil, (a) 32-AWG magnetic wire is wound around a 1.4-mm-diameter metal rod. (b) Once the wire-wound rod is embedded in an elastomer block, the rod is pulled out so that a micro-coil is formed with a microfluidic channel inside

3.2. Fabrication of Micro-NMR Circuit

1. The micro-NMR circuit is a resonance circuit with the micro-coil embedded in the elastomer block and two capacitors (one for resonance tuning and another for resonance matching) (Fig. 4a).

2. To make a resonance circuit with the micro-coil, the C_Mand C_T values need to be calculated (Fig. 4b). As the fabrication step of the micro-coil determines its inductance and resistor values and the magnetic field for NMR analysis determines the resonance frequency ω, the C_M and C_T values can be calculated with the pre-defined values. See Note 2.

3. The fabricated micro-coil and tuning/matching capacitors are soldered into a printed circuit board designed to make a resonance circuit (Fig. 1a).

Fig. 4.

Design of the micro-NMR circuit, (a) Schematic of the circuit. Z_eff effective impedance of the circuit, C_M capacitance of a matching capacitor, C_T capacitance of a tuning capacitor, L inductance of micro-coil, R resistance of micro-coil, (b) Mathematical analysis of C_M and C_T values. The resonance frequency, ω, and the characteristics of the micro-coil, L and R, are pre-defined based on the system hardware

3.3. Hyperpolarized NMR Experiment with Cells

1. The hyperpolarization technology used in our study is dDNP; we use a SPINlab machine (Fig. 5a) to hyperpolarize [1-¹³C] pyruvic acid.

2. 15 mM of AH-111501 (radical agent) is mixed with [1-¹³C] pyruvic acid thoroughly, and 100 μL of the mixture is loaded into a sample cup. The cup is glued to the fluid path designed for dissolution (Fig. 5b).

3. 40 mM of Trizma hydrochloride solution with 0.4 mM EDTA (a chelating agent) is prepared as a buffer solution for dissolution, 20 mL of which is loaded into a syringe connected to the fluid path (Fig. 5b).

4. The fluid path is air-tightly sealed with Luer fittings and loaded into the SPINlab hyperpolarizer. It takes approximately 2 h to achieve >30% of polarization.

5. After 2 h of hyperpolarization, target cells need to be prepared for metabolic flux analysis; 200–500k cells are collected from the flask, centrifuged, and resuspended in 45 μL of fresh media without fetal bovine serum.

6. When the cells are ready, the dissolution of the hyperpolarized pyruvic acid is initiated; the buffer solution is rapidly pushed through the fluid path, dissolving all the pyruvate-radical mixture from the sample cup to the outside of the hyperpolarizer. The dissolved [1-¹³C] pyruvic acid sample is neutralized with 120 μL of 10 N sodium hydroxide solution. See Note 3.

7. The hyperpolarized and neutralized [1-¹³C] pyruvate is mixed with the cell suspension with a ratio of 1:10, 8 μL of which is loaded into the micro-coil.

8. The micro-NMR circuit system is inserted into the center of the magnetic field in a conventional MRI system, followed by ¹³C-NMR signal acquisition. The NMR signal is acquired every 4 s with a flip angle of 30° radiofrequency pulse.

9. After 2 min of signal acquisition, the micro-NMR circuit system is removed from the MRI system and the cells are collected from the micro-coil for downstream molecular analyses, such as Western blot or PCR. Hyperpolarized NMR analysis is nondestructive and rapid, such that the cell viability after an experiment is unchanged (Fig. 6).

Fig. 5.

Pictures of SPINlab hyperpolarizer and a fluid path for sample loading, (a) A SPINlab machine has four separate channels so that different molecules can be hyperpolarized simultaneously, (b) A fluid path designed for sample loading into a SPINlab has a sample cup, where molecules to be hyperpolarized are loaded, and a syringe, where dissolution buffer solution is loaded. After 1–2 h of hyperpolarization in the SPINlab, the dissolution buffer is warmed and pushed through the sample cup, dissolving all the hyperpolarized molecules out of the SPINlab machine. Adapted from reference [8]

Fig. 6.

Comparison of total cell number and viability before and after hyperpolarized micro-NMR experiments. Two cancer cell lines (K562, chronic myeloid leukemia; UOK262, renal cancer cell) were tested for the experiments, and the cell number and viability were examined before and after. All measurement were conducted in duplicate. Error bars show the standard deviation. Adapted from reference [15]

4. Notes

1. The elastomer block that houses the micro-coil should not have any bubbles, as the air trapped inside bubbles can disrupt magnetic field homogeneity. Also, the bubbles between the wires can distort the overall shape of the micro-coil when the metal rod is removed.

2. The inductance of the micro-coil can be an order of 10⁻⁹ H, which requires relatively higher capacitors (an order of 10⁻⁹ C) for tuning and matching.

3. A hyperpolarized spin state decays with a spin-lattice relaxation time constant T₁, which can be very short—less than 20 s—for certain molecules. The T₁’s of [1-¹³C] pyruvate and [1-¹³C] dehydroascorbate (DHA) are relatively long, >70 s at 1 Tesla magnetic field [16], which motivates the wide use of these molecules in the field of hyperpolarized NMR/MRI. For molecules with a short T₁, such as [6-¹³C] arginine (~7 s), various approaches, including ¹⁵N-enrichment or dissolution with D₂O, have been demonstrated [17, 18].