¹H-NMR and Hyperpolarized ¹³C-NMR Assays of Pyruvate-Lactate Exhange: a comparative study

Abstract

Pyruvate-lactate exchange is mediated by the enzyme lactate dehydrogenase (LDH) and is central to the altered energy metabolism in cancer cells. Measurement of exchange kinetics using hyperpolarized ¹³C NMR has provided a biomarker of response to novel therapeutics. In this study we investigated an alternative in vitro ¹H assay, using [3-¹³C]pyruvate, and compared the measured kinetics with a hyperpolarized ¹³C-NMR assay, using [1-¹³C]pyruvate, under the same conditions in human colorectal carcinoma SW1222 cells. The apparent forward reaction rate constants (k_PL) derived from the two assays showed no significant difference, and both assays had similar reproducibility (k_PL = 0.506 ± 0.054 and k_PL = 0.441 ± 0.090 nmol/s/10⁶ cells, (mean ± standard deviation, n = 3); ¹H, ¹³C assays respectively). The apparent backward reaction rate constant (k_LP) could only be measured with good reproducibility using the ¹H-NMR assay (k_LP = 0.376 ± 0.091 nmol/s/10⁶ cells, (mean ± standard deviation, n = 3)). The ¹H-NMR assay has adequate sensitivity to measure real-time pyruvate-lactate exchange kinetics in vitro, offering a complementary and accessible assay of apparent LDH activity.

Introduction

There is a resurgence of interest in the area of cancer metabolism, including understanding deregulated metabolic pathways, which confer growth advantages to tumors (1). One metabolic adaptation to energy metabolism in cancer is known as the Warburg Effect (2), where glycolysis readily occurs even in the presence of oxygen; this classical hallmark of cancer has been associated with the activity and expression of the key enzyme lactate dehydrogenase (LDH), where increased enzyme activity has been associated with more aggressive disease (3-5). There is a need, therefore, for robust assays of enzyme kinetics, historically carried out by spectrophotometric analysis (6); however, the spectrophotometric assays cannot be performed on whole, viable cells and are consequently an indirect method of evaluating response to targeted therapies (7,8). In contrary to the spectrophotometric technique, NMR has facilitated measurements of real-time enzyme kinetics from viable, whole cells both in vitro and in vivo, and has been used to identify putative therapeutic metabolic targets in cancers (2,9-12).

In this study we have expanded upon an observation by Day et al. (13) and evaluated a ¹H-NMR assay of apparent LDH activity, and qualified it against a hyperpolarized ¹³C NMR measurement using Dynamic Nuclear Polarization (DNP) in the same tumor cell line.

Methods

Cell Preparation for NMR

SW1222 cells were seeded at a density of 3×10⁴ cm⁻² and cultured in DMEM medium (Invitrogen, UK) with the addition of 10% foetal bovine serum (FBS, Invitrogen, UK), 1% penicillin and streptomycin (Invitrogen, UK) and 1% L-glutamine (Invitrogen, UK) at 37°C in an atmosphere containing 5% CO₂. The floating cell fraction was discarded, and cells were rinsed twice with phosphate buffered saline (PBS). Cells were trypsinised and centrifuged for five minutes at 190 g. Cell pellets were immediately resuspended in 500 μl serum-free medium and NMR studies were carried out within 10 minutes of cell harvesting.

All NMR experiments were performed on an 11.7 T Bruker DMX spectrometer (Bruker Biospin, Germany).

¹H-NMR assay

100 μl D₂O and 100 μl of a solution containing 50 mM [3-¹³C]pyruvate (99% isotopically enriched, Sigma-Aldrich, UK) and 50 mM unlabelled, [¹²C]lactate (in PBS, pH 7) was mixed with a 500 μl cell suspension of 37 ± 5 × 10⁶ SW1222 cells (n=3) in a 5 mm NMR tube. Proton spectra were recorded at 500 MHz, 37°C, with 128 spectra acquired sequentially every 5 s using a 60° pulse-and-acquire sequence (1 transient, 16k time domain points, 10 kHz spectral width, 5 s recovery time). A presaturation pulse (noesygppr1d) was used to achieve good water suppression and a flat spectral baseline.

Hyperpolarized ¹³C-NMR assay

18 mg [1-¹³C]pyruvic acid (99% isotopically enriched, Sigma-Aldrich, UK) containing 15 mM trityl free radical OX63 (Oxford Instruments, UK) was polarised in a HyperSense® DNP polariser (Oxford Instruments Molecular Biotools Ltd, UK) for 1 hour; average polarization is typically of the order P = 10%, corresponding to an enhancement factor of ~8000 compared to the thermal signal acquired on the same sample at 11.7 T. The hyperpolarized sample was dissolved in 4 ml aqueous buffer (50 mM sodium lactate, 50 mM sodium hydroxide, 1 mM EDTA) resulting in a 46 ± 3 mM pyruvate solution at pH 7. A volume of this solution (100 μl) was mixed with a 500 μl cell suspension of 35 ± 2 × 10⁶ SW1222 cells (n = 3) in a 5 mm NMR tube. The reaction was monitored at 125 MHz, 37°C, with 128 spectra acquired sequentially every 2.0 s using a 10° pulse-and-acquire sequence (1 transient, 16k time domain points, a 19 kHz spectral width).

Kinetic Modeling

Lactate dehydrogenase activity can be described by a two-way chemical exchange model (scheme 1), where forward and backward apparent reaction rate constants are given by k_PL and k_LP respectively. The exchange reaction can be separated into two pools containing either the ¹³C or ¹²C spins, the sum of which represents the total pool sizes e.g. P(t) = ¹³P(t) + ¹²P(t) and L(t) = ¹³L(t) + ¹²L(t). For the ¹H assay, the exchange reaction can be written as a set of differential equations:

$$\frac{d}{dt}\left[\begin{array}{c}\hfill {}^{13}P\left(t\right)\hfill \\ \hfill {}^{13}L\left(t\right)\hfill \\ \hfill {}^{12}P\left(t\right)\hfill \\ \hfill {}^{12}L\left(t\right)\hfill \end{array}\right]=\left[\begin{array}{cccc}\hfill -k_{PL}\hfill & \hfill k_{LP}\hfill & \hfill 0\hfill & \hfill 0\hfill \\ \hfill k_{Pl}\hfill & \hfill -k_{LP}\hfill & \hfill 0\hfill & \hfill 0\hfill \\ \hfill 0\hfill & \hfill 0\hfill & \hfill -k_{PL}\hfill & \hfill k_{LP}\hfill \\ \hfill 0\hfill & \hfill 0\hfill & \hfill k_{PL}\hfill & \hfill -k_{LP}\hfill \end{array}\right]\left[\begin{array}{c}\hfill {}^{13}P\left(t\right)\hfill \\ \hfill {}^{13}L\left(t\right)\hfill \\ \hfill {}^{12}P\left(t\right)\hfill \\ \hfill {}^{12}L\left(t\right)\hfill \end{array}\right]$$ (1)

Where P(t) and L(t) represent the ¹H z-magnetizations of pyruvate and lactate signals respectively within the ¹³C and ¹²C pools. Whilst the above matrix appears block-diagonal owing to the absence of interconversion between the ¹³C and ¹²C they are connected via the presence of the common cofactor NAD(H). At the time-point of the injection of [3-¹³C]pyruvate (t = 0) the only non-zero pools are ¹³P(0) and ¹²L(0) and can be solved to yield time dependencies for [¹³C] and [¹²C]pyruvate and lactate signals as follows:

$${}^{13}P\left(t\right)=\frac{{}^{13}P\left(0\right)(k_{LP}+k_{PL}{e}^{-t(k_{PL}+k_{LP})})}{(k_{PL}+k_{LP})}$$ (2.1)

$${}^{13}L\left(t\right)=\frac{{}^{13}P\left(0\right)k_{PL}(1-e^{-t(k_{PL}+k_{LP})})}{(k_{PL}+k_{LP})}$$ (2.2)

$${}^{12}P\left(t\right)=\frac{{}^{12}L\left(0\right)k_{LP}(1-e^{-t(k_{PL}+k_{LP})})}{(k_{PL}+k_{LP})}$$ (2.3)

$${}^{12}L\left(t\right)=\frac{{}^{12}L\left(0\right)(k_{PL}+k_{LP}{e}^{-t(k_{PL}+k_{LP})})}{(k_{PL}+k_{LP})}$$ (2.4)

Scheme 1.

In the hyperpolarized ¹³C experiments, the observable magnetization is restricted to only the ¹³C subspace and the rate equations can be written:

$$\frac{d}{dt}\left[\begin{array}{c}\hfill {}^{13}P\left(t\right)\hfill \\ \hfill {}^{13}L\left(t\right)\hfill \end{array}\right]=\left[\begin{array}{cc}\hfill -k_{PL}-r_P\hfill & \hfill k_{LP}\hfill \\ \hfill k_{PL}\hfill & \hfill -k_{LP}-r_L\hfill \end{array}\right]\left[\begin{array}{c}\hfill {}^{13}P\left(t\right)\hfill \\ \hfill {}^{13}L\left(t\right)\hfill \end{array}\right]$$ (3)

where r_P,L = 1/T_1(P,L) denote the relaxation rates of hyperpolarized ¹³C signals of pyruvate and lactate, respectively. The above equation is frequently encountered in NMR relaxation experiments (14) and readily solved by taking the matrix exponential:

$$\frac{d}{dt}\left[\frac{{}^{13}P\left(t\right)}{{}^{13}L\left(t\right)}\right]=e^{\mathrm{L}t}\left[\frac{{}^{13}P\left(0\right)}{{}^{13}L\left(0\right)}\right]={\mathbf{U}}^{-1}{e}^{\mathbf{D}t}{\mathbf{U}}^1\left[\frac{{}^{13}P\left(0\right)}{{}^{13}L\left(0\right)}\right]$$ (4)

Where L is the relaxation matrix in equation (3), D is a diagonal matrix containing the eigenvalues of L, and U is a unitary matrix defined by D = ULU⁻¹. For the case r_P = r_L the solution is analogous to equations (2.1-2.2) convoluted with an additional exponential decay owing to the loss of magnetization. For the general case r_P ≠ r_L the above can be solved to give a full analytical expression (13).

Spectral Analysis

Spectra were apodized to 0.5 Hz (¹H) or 3 Hz (¹³C), phase and baseline corrected, and peaks of interest selected and integrated over the time-course of the experiment. The doublet ¹H resonance from [3-¹³C]pyruvate and the ¹H resonance from [¹²C]pyruvate were integrated and the ¹H(¹³C) integrals were summed. Integrals were fitted simultaneously according to the two-way chemical exchange model, equations 2.1 and 2.3, using Matlab (MathWorks, UK) to determine apparent forward k_PL and backward k_LP rate constants. Rate constants were normalized to cell count for each sample. Integrals of the hyperpolarized [1-¹³C]-resonances of pyruvate and lactate were flip-angle corrected and simultaneously fitted using Matlab to the full analytical solution (13) of the modified Bloch Equations for two-site exchange.

Statistics

Data are presented as mean ± standard deviation. Differences between rates were identified using a Student’s unpaired t-test with a 5% confidence interval.

Results

Following the addition of an equimolar solution of [3-¹³C]pyruvate and [¹²C]lactate to a suspension of SW1222 cells, a time-dependent decrease in the ¹H resonances of both [3-¹³C]pyruvate (2.4ppm, ¹_JC3-H3 = 128.7Hz) and [¹²C]lactate (1.3ppm) was observed with a concomitant increase in the ¹H resonances of [¹²C]pyruvate (2.4ppm) and [3-¹³C]lactate (1.3ppm, ¹J_C3-H3 = 128.7Hz), reflecting LDH-mediated pyruvate-lactate exchange (Fig. 1A). The signal-to-noise of the ¹H [3-¹³C]pyruvate peaks at time t = 0 was 31.5 ± 2.8 (n = 3). The ¹H-NMR assay allowed the detection of both forward and reverse reactions with apparent reaction rates k_PL = 0.506 ± 0.054 nmol/s/10⁶ cells and k_LP = 0.376 ± 0.091 nmol/s/10⁶ cells (n = 3), derived from kinetic modeling of the ¹H data. There was no significant difference between k_PL and k_LP (p = 0.10). The sum of the ¹H(¹³C) and ¹H(¹²C) signals was constant over the time series.

Figure 1.

(A) Real-time ¹H-NMR and (B) hyperpolarized ¹³C-NMR assays of pyruvate-lactate exchange in colorectal SW1222 cancer cells and the corresponding time evolution of the spectral integrals (C, D). (C) Integrals of the ¹H(¹²C) peak, open symbols, and ¹H(¹³C) pyruvate peaks, closed symbols, from the ¹H assay. (D) Integrals of the hyperpolarized [1-¹³C]pyruvate peak, closed symbols, and [1-¹³C]lactate peaks, open symbols, from the ¹³C assay. The solid lines correspond to the fits from the kinetic models described in the text.

Following the addition of an equimolar solution of hyperpolarized [1-¹³C]pyruvate and [¹²C]lactate to a suspension of SW1222 cells, a decrease in the hyperpolarized ¹³C resonance of pyruvate is observed owing to T₁ loss of polarization as well as metabolic conversion, with a concomitant increase in the ¹³C resonance of [1-¹³C]lactate, again reflecting LDH-mediated pyruvate-lactate exchange (Fig. 1A). The signal-to-noise of the [¹³C]pyruvate peak at time t = 0 was measured to be 2186 ± 390 (n=3). Apparent reaction rates k_PL = 0.441 ± 0.090 nmol/s/10⁶ cells and k_LP = 0.289 ± 0.208 nmol/s/10⁶ cells were derived from kinetic modeling of the ¹³C data (p = 0.91) (n = 3). A significant uncertainty was associated with the measurement of the reverse reaction rate constant.

There was no significant difference between k_PL (p = 0.35) and k_LP (p = 0.82), measured from the ¹H and ¹³C assays.

Discussion

In this study we demonstrated that the use of [3-¹³C]pyruvate in combination with ¹H-NMR provides a robust assay to monitor pyruvate-lactate exchange kinetics in vitro. This was compared with the equivalent hyperpolarized ¹³C-NMR assay. While the signal-to-noise ratio of the hyperpolarized ¹³C-NMR assay was, at most, 69 times greater than the ¹H-NMR assay, the ¹H assay provided adequate signal to monitor pyruvate-lactate exchange kinetics in real time. Furthermore, the transient hyperpolarized signal restricts the acquisition window to a few minutes, whereas the ¹H assay allows continuous observation of the enzymatic process.

The ¹H-NMR assay allows direct, and therefore more accurate, measurement of k_LP than the hyperpolarized ¹³C assay. Accurate determination of k_LP from the kinetic modeling of ¹³C-NMR data is challenging because of the initial condition of an empty [¹³C]lactate pool coupled with the decay of all the signals owing to relaxation, and the greater number of model fitting parameters in the hyperpolarized assay. This translates into greater uncertainty in k_LP estimates from the kinetic modeling of ¹³C-NMR compared with the ¹H assay. The accurate determination of k_LP using the hyperpolarized assay has been addressed recently by inverting the magnetization of one of the hyperpolarized metabolite pools, which allows the measurement of forwards and reverse exchange rates (15). A potential solution to allow the direct measurement of k_LP using the hyperpolarized assay is to co-polarize [2-¹³C]pyruvate with [1-¹³C]lactate, allowing simultaneous detection of both reactions.

Additionally, the measured rate of pyruvate-lactate exchange not only reflects LDH activity (16) but is also dependent on the level of LDH cofactors (NAD/NADH), providing information on cellular energetics, the pool sizes of the injected ¹³C species, and the pool sizes of (either endogenous or exogenous) ¹²C metabolites (13,15), as well as the activity and expression of the monocarboxylate transporters, MCT1 and MCT4, that respectively mediate the transport of hyperpolarized pyruvate and lactate into and out of the cells (17). These parameters are often deregulated in cancer, and both assays could be useful to study these important factors, and their modulation, in vitro.

While the ¹H-NMR assay demonstrates several attractive features in vitro, hyperpolarized ¹³C-NMR remains the technique of choice for in vivo measurements, owing to the necessary signal enhancements and large chemical shift range of ¹³C; it also allows detection of the other metabolic fates of pyruvate, such as the formation of alanine and bicarbonate, biomarkers of alanine transaminase activity and pyruvate dehydrogenase (PDH) flux respectively (18-20).

Conclusions

The ¹H-NMR assay of pyruvate-lactate exchange offered a complementary and accessible measurement of apparent LDH activity in live cancer cells in vitro, with the added value of directly measuring the backward reaction.

Figure 2.

LDH-mediated pyruvate-lactate exchange rates measured in SW1222 cells. (A) Forward and reverse apparent rate constants k_PL and k_LP measured from the ¹H-NMR assay (B) Forward and reverse rate constants k_PL and k_LP measured from the hyperpolarized ¹³C-NMR assay. Rates are normalized to cell number in each sample.

Abbreviations

DNP

Dynamic Nuclear Polarization

LDH

lactate dehydrogenase

MCT

monocarboxylate transporter