13C magnetic resonance spectroscopic imaging of hyperpolarized [1‐13C, U‐2H5] ethanol oxidation can be used to assess aldehyde dehydrogenase activity in vivo

Aldehyde dehydrogenase (ALDH2) is an emerging drug target for the treatment of heart disease, cocaine and alcohol dependence, and conditions caused by genetic polymorphisms in ALDH2. Noninvasive measurement of ALDH2 activity in vivo could inform the development of these drugs and accelerate their translation to the clinic.


INTRODUCTION
Ethanol is oxidized by the liver to produce acetate, the majority of which exits via the hepatic vein and is metabolized further in other tissues (1). At low to medium blood concentrations of ethanol (up to 24 mM in humans) the bulk of its oxidation to acetaldehyde and further to acetate is via the coupled reactions catalyzed by the NAD þ -dependent dehydrogenases: cytosolic alcohol dehydrogenase (ADH) and mitochondrial aldehyde dehydrogenase (ALDH2) (2,3). Microsomal NADPH-dependent cytochrome CYP2E1 and peroxisomal H 2 O 2 -dependent catalase can also oxidize ethanol to acetaldehyde, their relative contributions increasing, for example, during exposure to high-doses of ethanol and in starvation, respectively (4,5). Ethanol oxidation by hepatic dehydrogenases causes an acute increase in the NADH/NAD þ ratio, leading to an increase in cytosolic lactate concentrations and disturbances of lipid metabolism, with consequent onset of alcoholic fatty liver disease (2,6). Adaptive induction of hepatic CYP2E1 activity by chronic ethanol exposure, which increases the liver's capacity to produce acetaldehyde from ethanol, is accompanied by a decreasing capacity of the liver mitochondria to oxidize this acetaldehyde. This decreased capacity results from functional and structural impairment caused by exposure to the elevated acetaldehyde concentrations (7)(8)(9). Inefficient disposal of acetaldehyde and its accumulation, both intrahepatically and extrahepatically, is the direct cause of liver cirrhosis, and is implicated in alcohol-induced cardiomyopathy and pancreatitis, and in some types of cancer, particularly in the upper aerodigestive tract (10,11).
The capacity to oxidize endogenous and exogenous aldehydes, conferred on extrahepatic tissues by expression of ALDH2, is fundamental for their normal function. Pharmacological activation of ALDH2 protects the myocardium from ischemic injury by stimulating the oxidative activity of ALDH2 toward 4-hydroxy-2-nonenal, a toxic aldehyde produced in ischemic heart (12)(13)(14). ALDH2 is expressed in brain tissue, where it is essential for the maintenance of dopamine homeostasis and the oxidation of the endogenous aldehydes 3,4-dihydroxyphenylacetaldehyde and 3,4-dihydroxyphenylglycolaldehyde. Accumulation of 3,4dihydroxyphenylacetaldehyde, 3,4-dihydroxyphenylglycolaldehyde, and dopamine is neurotoxic and disturbances of their metabolism are implicated in neurodegenerative disorders such as Parkinson's disease and in Alzheimer's disease (15,16). In addition, dopamine signaling in brain plays a major role in establishing addictive behaviors and in triggering relapse (17,18).
The role of ALDH2 in human health is illustrated by the inability of individuals carrying the ALDH2*2 allele, which encodes an inactive E487K mutant, to dispose efficiently of toxic aldehydes. ALDH2*2 allele carriers are at higher risk of increased morbidity after myocardial infarction and are more likely to develop late-onset Alzheimer's disease and cancers in the upper aerodigestive tract (10,12,19,20). The number of ALDH2*2 allele carriers worldwide is estimated to be at least 540 million, comprising approximately 40% of native East Asian populations (20).
The recently developed ALDH2 activator, Alda-1, was shown to confer protection against myocardial ischemic damage in a rat model of infarction (12,13) and a ALDH2specific inhibitor, CVT-10216, inhibited addictive substance seeking and relapse in rat models of cocaine addiction and alcoholism (17,18). In addition, the catalytic activity of the E487K ALDH2 mutant can be partially restored by Alda-1, which could have important implications for clinical management of diseases affecting carriers of the ALDH2*2 allele (20). A noninvasive method for measuring ALDH activity in vivo would inform preclinical development of these drugs and could accelerate their translation to the clinic.
Much of what is known about ethanol metabolism derives from ex vivo studies on perfused organs, isolated hepatocytes, and tissue samples (6,21,22). Investigation of ethanol metabolism in vivo has been limited to magnetic resonance spectroscopy (MRS) studies, which have been used, for example, to investigate the distribution of 13 C label in rat brain metabolite pools following systemic administration of [1-13 C] ethanol (23). The main limitation of using 13 C MRS in this way is its low sensitivity, which results in relatively low temporal and spatial resolution.
The recent introduction of dissolution dynamic nuclear polarization (24), which can increase the sensitivity of 13 C MRS by at least four orders of magnitude, has made possible measurements of the metabolism of many 13 C-labeled substrates in vivo (25,26). In this process, the 13 C-labeled substrate is mixed with a stable radical and rapidly frozen to form a glass. The sample is then cooled to $1.2 K, at which temperature the electron spins on the radical become fully polarized. Microwave irradiation is then used to transfer this electron spin polarization to the 13 C nuclear spins (24), following which the sample is warmed rapidly to room temperature using superheated water, with substantial retention of the nuclear spin polarization (24). The increase in polarization is sufficient to allow imaging, using 13 C magnetic resonance spectroscopic imaging, of the hyperpolarized tracer and its metabolic products following injection into a biological system. Despite the relatively rapid decay of the nuclear spin polarization in the liquid state, which is the principal limitation of the technique, dissolution dynamic nuclear polarization has already translated to the clinic, where hyperpolarized [1-13 C] pyruvate was used to detect tumors in the prostate (27).
The activity of ALDH2 has been estimated indirectly from measurements of hyperpolarized 13 C label exchange between injected [1-13 C] pyruvate and endogenous lactate, in the reaction catalyzed by lactate dehydrogenase. This exchange was increased in rat liver by ethanol injection, which increased the NADH/NAD þ ratio, and was decreased by the ALDH2 inhibitor, disulfiram. An increase in the NADH/NAD þ ratio strongly stimulates the lactate dehydrogenase-catalyzed exchange (28). However, this indirect measurement assumes that during ethanol metabolism the change in the mitochondrial NADH/NAD þ ratio mirrors its cytosolic equivalent (29). This is not the case, for example, in physiological states where the relative contribution of NAD þ -independent pathways of ethanol oxidation to acetaldehyde, such as that catalyzed by CYP2E1, becomes important (3).
We show here that oxidation of hyperpolarized perdeuterated [1-13 C] ethanol ([1-13 C, U-2 H 5 ] ethanol) to [1-13 C] acetate, in the coupled reactions catalyzed by cytosolic ADH and mitochondrial ALDH2, can be measured and imaged directly in mouse liver in vivo using 13 C MRS/I. The influence of ALDH2 activity on this label flux was demonstrated by inhibiting the enzyme with the irreversible inhibitor, disulfiram (22).

METHODS
All chemicals were purchased from Sigma Aldrich (Gillingham, Dorset, UK), unless stated otherwise.

C T 1 Measurements In Vitro
Measurements were made at 11 T (Bruker Avance II þ 500 MHz spectrometer (Bruker Spectrospin Ltd., Coventry, UK)) using inversion recovery sequences in samples containing approximately 10 mM

Hyperpolarization of [1-13 C, U-2 H 5 ] Ethanol
A 2:3 glycerol: [1-13 C, U-2 H 5 ] ethanol mixture, containing 15 mM trityl radical (OX063: tris(8-carboxy-2,2,6,6-tetra-(hydroxyethyl)-benzo-[1,2-4,50]-bis-(1,3)-dithiole-4-yl)methyl, sodium salt; GE Healthcare, Little Chalfont, UK) and 1.5 mM gadoteric acid (DOTAREM; Guebert, Roissy, France), was polarized for approximately 120 min in a 3.35 T Hypersense polarizer (Oxford Instruments, Abingdon, Oxfordshire, UK) using irradiation at 94.124 GHz. The dissolution buffer (PBS, pH 7.5, 100 mg/L EDTA) was heated to 180 C and pressurized to 10 bar before dissolving the frozen polarized samples. The dissolved sample contained 110 mM [1-13 C, U-2 H 5 ] ethanol. For measurements of the liquid-state polarization, a sample of [1-13 C, U-2 H 5 ] ethanol was polarized and dissolved as described above. Two milliliters of the sample, containing approximately 60 mM [1-13 C, U-2 H 5 ] ethanol were injected into 2 mL of PBS buffer, pH 7.4 in a 10-mm NMR tube placed inside a vertical bore 9.4 T NMR spectrometer with a broadband probe tuned to 13 C (Varian NMR Instruments, Palo Alto, CA). Immediately following injection, and approximately 12 s after dissolution, 180 single transient spectra, with a pulse flip angle of 6 and 8-kHz spectral width, were acquired with a TR of 1 s. Three 90 pulses were then applied to eliminate the remaining hyperpolarization. A 13 C spectrum (the sum of 128 transients) was then acquired using a 6 pulse, a spectral width of 8 kHz, and a TR of 300 s. The temperature of the sample was kept at 37 C. The liquid-state polarization was calculated from the ratio of the maximal integrated intensity of the hyperpolarized [1-13 C, U-2 H 5 ] ethanol signal to the integrated intensity of the thermal equilibrium signal corrected for the number of transients.

C MRS Measurements of Ethanol Oxidation In Vitro
Approximately 300 units (2 mg) of ADH from Saccharomyces cerevisiae were dissolved in 100 mL of 0.05 M sodium phosphate buffer, pH 7.5 and mixed with 1.9 mL of 0.05 M pyrophosphate buffer, pH 9.0, containing 1 mg/mL of bovine serum albumin and 15 mM NAD þ (Roche Applied Science, Basel, Switzerland). The sample was then placed immediately in a 9.4 T NMR spectrometer with a 10-mm broadband probe tuned to 13 C (Varian NMR Instruments, Palo Alto, CA). Within approximately 3 min from placing the tube inside the magnet, and approximately 12 s after dissolution, 2 mL of the dissolved polarized sample containing 110 mM [1-13 C, U-2 H 5 ] ethanol was injected into the tube. A series of four transient spectra (sweep width 16 kHz), which were centered either at 240 ppm (the acetaldehyde region of the spectrum) or at 0 ppm (the ethanol region of the spectrum), were then acquired with a flip angle of 12 . Four acetaldehyde spectra were acquired followed by a single ethanol spectrum (TR 1.0 s) and this sequence was repeated over a period of 320 s. The sample temperature was maintained at 37 C.

C MRS Measurements of Ethanol Oxidation In Vivo
Animal experiments were approved by local ethical review committees and performed in accordance with the Animals (Scientific Procedures) Act of 1986. Female CD1 mice (6-8 weeks old) were used in all experiments. Animals were anaesthetized using 3% isoflurane vapor (Isoflo, Abbotts Laboratories Ltd., Maidenhead, UK) in an oxygen/air mixture. Anesthesia was maintained during the MR experiments with 1-2% isoflurane vapor in an oxygen/air mixture delivered via a face mask. A catheter was inserted in a tail vein and the mouse placed in a heated cradle. Throughout the experiment animal breathing rate and core body temperature were monitored using a Biotrig physiological monitor (Small Animal Instruments, Stony Brook, NY). These were within the range of 60-100 bpm and 36-37 C, respectively. A 20-mm-diameter surface coil (Rapid Biomedical GmbH, Rimpar, Germany) was placed over the liver and the entire assembly placed in a 13 C/ 1 H volume coil (Rapid Biomedical, Germany), in a 7 T horizontal bore magnet (Varian, Palo Alto, CA). Livers were localized in transverse 1 H images, acquired using a spin-echo pulse sequence (TR, 1.5 s; echo time (TE), 10 ms; field of view, 40 Â 40 mm 2 ; data matrix, 128 Â 128; slice thickness, 2 mm; 15 slices). Immediately after dissolution, 13 C spectroscopic acquisitions were started and 10 mL of the dissolved sample per gram body weight, typically 240 mL in total, were injected via the tail vein over a period of 2 s. Single transient spectra (sweep width 6 kHz; the delay from the middle of the excitation pulse to beginning of data acquisition was 280 ms) from the entire sensitive volume of the surface coil were acquired using a frequency-selective excitation pulse (450 ms three-lobe sinc pulse) centered either at 180 ppm (the acetate region of the spectrum) or at 60 ppm (the ethanol region of the spectrum) with a nominal flip angle of 10 . A series of ethanol and acetate spectra were acquired from one animal, where four acetate spectra were acquired followed by a single ethanol spectrum (TR 1.0 s) and this pattern was repeated over a period of 120 s (Fig. 1). In animals in which ALDH2 was inhibited with disulfiram (n ¼ 7), spectra were collected using a repeated acquisition pattern, where nine acetate spectra were acquired followed by a single ethanol spectrum (TR 1.1 s). Within each repeat, every spectral acquisition was preceded by a saturation pulse (1 s hard pulse with a B 1 field of 100 Hz) at 100 kHz (control saturations 1 and 2), À1840 Hz (control saturation 3), 1840 Hz (saturation at the acetaldehyde resonance frequency), or at 100 kHz (control saturation 4) from the acetate resonance frequency. In three animals, two chemical shift images (FOV 40 Â 40 mm, 32 Â 32 data matrix, TR 30 ms, TE 0.58 ms) were acquired, starting 15 s after ethanol injection, where the images were acquired first from acetate and then ethanol. All data were phase and baseline corrected and the signal integrals for ethanol and acetate measured.

Disulfiram Dose-Response Experiments
Animals were given 100 or 600 mg/kg body weight of disulfiram (LKT Laboratories, Inc., St. Paul, MN) as a suspension in 5% w/v of gum arabicum, by oral gavage (30). 13 C MRS experiments were performed approximately 24 h after drug administration. In some experiments, shortly after completion of MRS measurements, animals were killed and their livers rapidly excised, freeze-clamped, and stored at À80 C before homogenization and enzyme assay.

Measurement of Blood Ethanol Level
An appropriate volume of [2-13 C] ethanol, polarized and dissolved to the same concentration as [1-13 C, U-2 H 5 ] ethanol, was injected into the tail vein of animals that had been anesthetized for 30-45 min. Blood was withdrawn by cardiac puncture, which began 75 s after the start of ethanol injection and was completed within 30 s. Individual blood samples were transferred into separate BD Microtainer FE microcollection tubes (BD, Franklin Lakes, NJ), centrifuged at 800g and the plasma removed and frozen in liquid nitrogen. For NMR spectroscopy, plasma samples were thawed and diluted in 200 mM PBS in D 2 O, pH 7.5. The buffer contained 100 mM 2,2,3,3-D4 sodium-3-trimethylsilylpropionate as an intensity and chemical shift standard (Cambridge Isotope Laboratories Inc., Andover, MA). 13 C spectra (the sum of 1024 transients) were acquired under fully relaxed conditions at 125 MHz (Bruker Avance II þ 500 MHz spectrometer) into 25,294 data points using a 90 pulse, a sweep width of 5 kHz, and a TR of 60 s. Integrated intensities of the [2-13 C] ethanol and [2-13 C] acetate resonances were normalized to that of 2,2,3,3-D4 sodium-3trimethylsilylpropionate and multiplied by the methyl 13 C concentration in 100 mM 2,2,3,3-D4 sodium-3trimethylsilylpropionate (3.33 mM) to convert to acetate or ethanol concentrations.

Liver Tissue Extraction
Freeze-clamped livers were homogenized using a manual Potter S homogenizer (Sartorius, Epsom, UK) in seven volumes (mL/gram wet weight of liver) of ice-cold extraction buffer (50 mM Tris-HCl, pH 7.4 containing 0.25 M sucrose, 0.1% v/v 2-mercaptoethanol, 1 mM EDTA, and protease inhibitor cocktail (Roche Applied Science, Basel, Switzerland). Ten strokes of 5 s (each followed by a 5 s pause) at 1000 r.p.m. were used. Homogenates were centrifuged at 800g (Eppendorf 5415D benchtop centrifuge (Eppendorf, Hamburg, Germany)) at 4 C. The supernatant was extracted by the addition of 20% v/v of extraction buffer containing 5% v/v of Triton X-100. The extracts were centrifuged at 16,200g at 4 C to remove debris. Supernatants were kept on ice and used for enzyme assays.

ALDH2 Activity Assay
The assay mixture contained 0.5 mM NAD þ , 0.1% v/v of 2-mercaptoethanol, 200 mM KCl in 100 mM sodium pyrophosphate buffer, pH 9.0. The sample of liver extract was preincubated at room temperature for 20 min and the reaction initiated by the addition of propionaldehyde to a final concentration of 20 mM. At this low aldehyde concentration, the assay measures predominantly ALDH2 activity (22,31). The formation of NADH was measured for 5 min after substrate addition as an increase in absorbance at 340 nm using a Sunrise 96well plate reader (Tecan, Mannedorf, Switzerland). One unit of ALDH activity is defined as the amount of enzyme that oxidizes 1 mmol of propionaldehyde (and reduces 1 mmol of NAD þ ) per min. All assays were carried out at 25 C.
Treatment of the animals with disulfiram decreased the maximum acetate signal (normalized to the maximum ethanol signal) by approximately 43% at a dose of 100 mg/kg (n ¼ 1) and by 79 6 8% (SD, n ¼ 3) at 600 mg/kg body weight (Fig. 1c,e). The ALDH2 activity in extracts of livers from control animals was 0.37 6 0.07 units per gram wet weight (U/gww) (SD, n ¼ 3) and 0.12 6 0.10 U/ gww (SD, n ¼ 3) in animals 24 h after treatment with 600 mg/kg body weight disulfiram, a decrease of 68%.
Introduction of a saturation pulse at the resonance frequency of the unobserved [1-13 C] acetaldehyde resulted in a rapid decrease in the acetate signal intensity, when compared with the time course of the acetate signal in the absence of a saturation pulse, and subsequent recovery when saturation was stopped, demonstrating that the acetate was produced via acetaldehyde (Fig. 3). A fit of the acetate signal intensities to the modified Bloch equations for this system is shown in Supporting Information (Fig.  S1). Analysis of the slopes on the log plot between 22 and 33 s (control saturation at À1.8 kHz) and between 33 and 44 s (saturation of the acetaldehyde resonance at þ1.8 kHz) gave values of 0.08 6 0.02 and 0.18 6 0.02 s À1 , respectively (P < 0.01, n ¼ 4), demonstrating that saturation of the acetaldehyde resonance resulted in a significant increase in the rate of decay of the acetate polarization. Chemical shift images showed that the ethanol and acetate signals were mainly from the liver (Fig. 4). 13

DISCUSSION
The disulfiram-sensitive, mitochondrial (low K m ) isoform of ALDH (ALDH2) is primarily responsible for acetaldehyde oxidation in the hepatocyte during ethanol metabolism and accounts for approximately 60-80% of acetaldehyde oxidation in the liver in both healthy humans (34) and in rodents (22). The other enzymes, including the cytosolic (high K m ) ALDH1 isoform, account for the remaining 20-40% (22,31,34,35). Decreasing ALDH activity in this study by 68%, using the irreversible ALDH2 inhibitor disulfiram (22,36), reduced the acetate signal intensity by $79%, and implies, therefore, that the rate of acetate production from hyperpolarized [1-13 C, U-2 H 5 ] ethanol in the liver is determined primarily by ALDH2 activity. This reduction in ALDH2 activity assayed in tissue extracts agrees well with a decrease of approximately 67% measured in rat liver 24 h after administration of 600 mg/kg disulfiram (36). Although the production of acetaldehyde from ethanol was not observed, the level of acetaldehyde being too low to detect, applying a saturation pulse at the acetaldehyde resonance frequency and observing a decrease in the acetate signal intensity confirmed that the production of acetate from [1-13 C, U-2 H 5 ] ethanol proceeded via acetaldehyde. The longer T 1 in [1-13 C, U-2 H 4 ] acetaldehyde, as compared to the protonated form, will contribute to preservation of the hyperpolarization during transfer of the 13 C label from ethanol to acetate. Collapse of the 27 Hz 2 H-13 C coupling in acetaldehyde (Fig. 2), through deuterium decoupling, may enhance direct detection of the acetaldehyde intermediate in vivo. Detection of label flux from ethanol to acetate was also enhanced by the relatively long T 1 of the C1 carbon in acetate, which was 49.2 6 1.1 s (SD, n ¼ 4) in [U-2 H 3 ] acetate in vitro.
Although the activity of mitochondrial ALDH has been shown to govern the rate of acetaldehyde oxidation during ethanol metabolism in rat liver slices (35), studies in perfused liver indicate that ADH activity appears to play some role in determining the rate of ethanol oxidation to acetate (21). However, the dependence of the [1-13 C] acetate signal on ALDH2 activity observed here has shown that the flux between ethanol and acetate is determined primarily by ALDH2 activity. This, further, implies that the delivery of ethanol and its transport into the cell have little influence on the rate of acetate production. The amphipathic nature of ethanol means that it can diffuse freely from the bloodstream into tissues, and therefore, its utilization is less likely to be limited by transport into cells when compared to other hyperpolarized 13 C-labeled tracers that require facilitated membrane transport (28). Assuming that water accounts for approximately 74% of body weight in rodents, injection of 10 mL/g body weight of 110 mM ethanol would be expected to give approximately 1.50 mM of ethanol at equilibrium, with approximately equal concentrations in intracellular and extracellular water (23,37). Consistent with this expectation, we measured a blood concentration of 2.02 mM at 75 s after injection of 10 mL/g body weight of 110 mM ethanol.
Perdeuteration of ethanol, while extending T 1 by a factor of $5, to 56 s, and making possible imaging of the hyperpolarized molecule and its subsequent metabolism in vivo (Fig. 4), could affect the rate of ADH-catalyzed oxidation of ethanol to acetaldehyde through a primary kinetic isotope effect (38). There is no such effect in the subsequent oxidation to acetate, catalyzed by ALDH (39). In stopped-flow pre steady-state kinetic studies on horse liver ADH, a primary kinetic isotope effect was shown to result in an approximately 6.2-fold decrease in the initial rate of oxidation of [U-2 H 5 ] ethanol to acetaldehyde when compared with the protonated form, reflecting the fact that hydride transfer is rate-limiting under these conditions (38). However, under steady-state conditions, dissociation of the binary enzyme-NADH product has been shown to be rate limiting, and therefore, under these circumstances the isotope effect will be much smaller.
In summary, we have measured the oxidation of ethanol to acetate in vivo in real time using hyperpolarized [1-13 C, U-2 H 5 ] ethanol and demonstrated that in the liver in vivo this flux is determined predominantly by ALDH2 activity. As intravenous ethanol administration is a clinically approved medical procedure in, for example, the treatment of ethylene glycol poisoning (40) hyperpolarized [1-13 C, U-2 H 5 ] ethanol could be translated to the clinic, where it could be used as a means of assessing the pharmacodynamic properties of ALDH2-targeted therapies. For example, ALDH2 activators, such as Alda-1, have been proposed for the treatment of Fanconi's anemia, where the accumulation of noxious aldehydes, and specifically acetaldehyde, are particularly toxic (41). A limitation of the method is the requirement that the tissue contain ADH activity, so, for example, it could not be used directly in heart muscle to measure ALDH2 activity as, similarly to most extrahepatic tissues, it lacks ADH (42). Nevertheless, measurements in liver could still be used to assess the efficacy of these drugs. Direct measurement of drug-modulated ALDH2 activity in heart muscle may, however, be possible preclinically as transgenic mouse lines that have high levels of ADH expression in heart muscle have been generated (43). Measurements with hyperpolarized [1-13 C] acetaldehyde would remove the requirement for the presence of ADH activity, however, these are likely to be limited by unacceptable toxicity of the aldehyde (44). designed the study, carried out experiments, analyzed data, wrote the manuscript. M.I.K. advised on experimental design, devised and carried out all MRSI experiments and analyzed the resulting data, and cowrote the manuscript. I.M.R. advised on experimental design, carried out some experiments, analyzed data, and participated in editing of the manuscript. K.N.T. and T.J.L. carried out some experiments and analyzed data. E.M.S. and T.B.R. carried out some experiments. K.M.B. provided scientific advice and directed the study and wrote the final version of the manuscript.