Hyperpolarized [1-¹³C] pyruvate MRSI to detect metabolic changes in liver in a methionine and choline-deficient diet rat model of fatty liver disease

Abstract

Purpose:

Nonalcoholic fatty liver disease is an important cause of chronic liver disease. There are limited methods for monitoring metabolic changes during progression to steatohepatitis. Hyperpolarized ¹³C MRSI (HP ¹³C MRSI) was used to measure metabolic changes in a rodent model of fatty liver disease.

Methods:

Fifteen Wistar rats were placed on a methionine- and choline-deficient (MCD) diet for 1–18 weeks. HP ¹³C MRSI, T₂-weighted imaging, and fat-fraction measurements were obtained at 3 T. Serum aspartate aminotransaminase, alanine aminotransaminase, and triglycerides were measured. Animals were sacrificed for histology and measurement of tissue lactate dehydrogenase (LDH) activity.

Results:

Animals lost significant weight (13.6%±2.34%), an expected characteristic of the MCD diet. Steatosis, inflammation, and mild fibrosis were observed. Liver fat fraction was 31.7% ± 4.5% after 4 weeks and 22.2% ± 4.3% after 9 weeks. Lactate-to-pyruvate and alanine-to-pyruvate ratios decreased significantly over the study course; were negatively correlated with aspartate aminotransaminase and alanine aminotransaminase (r = −[0.39–0.61]); and were positively correlated with triglycerides (r = 0.59–0.60). Despite observed decreases in hyperpolarized lactate signal, LDH activity increased by a factor of 3 in MCD diet-fed animals. Observed decreases in lactate and alanine hyperpolarized signals on the MCD diet stand in contrast to other studies of liver injury, where lactate and alanine increased. Observed hyperpolarized metabolite changes were not explained by alterations in LDH activity, suggesting that changes may reflect co-factor depletion known to occur as a result of oxidative stress in the MCD diet.

Conclusion:

HP ¹³C MRSI can noninvasively measure metabolic changes in the MCD model of chronic liver disease.

1 |. INTRODUCTION

Nonalcoholic fatty liver disease (NAFLD) is a common cause of chronic liver dysfunction and a significant public health problem. Nonalcoholic steatohepatitis (NASH) defines the progressive and more severe form of NAFLD and can eventually lead to cirrhosis, liver failure, and hepatocellular carcinoma. There is therefore an unmet clinical need to develop methods to noninvasively identify patients with NASH, to determine intervention and treatment.

The precise mechanisms involved in the progression of steatosis to NASH are not well understood. The methionine- and choline-deficient (MCD) diet is a reproducible and widely used model of NASH that induces macrovesicular fat, ballooning of hepatocytes, and fibrosis in rodent livers.^1,2 The MCD diet is characterized by altered phosphatidylcholine synthesis and impaired very-low-density lipoprotein secretion, resulting in the accumulation of triglycerides (TG) within hepatocytes.² Moreover, the MCD diet alters mitochondrial function, provoking impairment in the removal of fatty acids³ and changes in metabolic pathways that contribute to the liver injury. Studies in rodents fed MCD diet have associated lipotoxicity and oxidative stress to the pathogenesis of NASH.⁴

Methods are limited for noninvasively monitoring the progression of NAFLD into NASH, and biopsy remains the gold standard in diagnosis and prognosis. Hyperpolarized ¹³C MRSI (HP ¹³C MRSI) is a novel technique for noninvasive characterization of in vivo metabolism in a variety of organs and pathologic conditions, based on ¹³C-labeled small molecule probes. [1-¹³C]pyruvate has been widely investigated because of its high polarization, long T₁ (longitudinal relaxation time~70s at 3 T), and critical biochemical role in central carbon metabolism. [1-¹³C]pyruvate is rapidly metabolized to [1-¹³C]lactate and [1-¹³C]alanine via the enzymes lactate dehydrogenase (LDH) and alanine transaminase (ALT), respectively. Previous animal studies have reported elevated levels of [1-¹³C]lactate production in toxic hepatic injury due to CCl₄⁵ as well as liver injury induced by a high-fat diet (HFD).⁶ Moreover, a recent study evaluating hepatic redox status using HP [1-¹³C]dehydroascorbate (DHA) in mice fed the MCD diet reported a decreased conversion of [1-¹³C]DHA to [1-¹³C]Vitamin C.⁷ These findings show clear potential for noninvasive molecular imaging detection of liver injury based on metabolic changes accessible via HP ¹³C MRSI.

HP [1-¹³C]pyruvate MRSI is readily translatable, with 15 research centers already conducting human studies at the time of writing. As a step toward clinical translation, we applied HP ¹³C MRSI to image changes in the real-time hepatic metabolism of pyruvate along the progression of NAFLD to NASH in animals fed the MCD diet, as compared with existing histologic and serum markers for liver disease.

2 |. METHODS

2.1 |. Animals

All experiments were approved by the University of California, San Francisco Institutional Animal Care and Use Committee. A total of sixteen 12-week-old adult male Wistar rats (Charles River Laboratories, Wilmington, MA, USA) were used for all studies (Table S1). Fifteen rats were placed on a MCD diet (Teklad, TD.90262; Envigo, Indianapolis, IN, USA) for 1–18 weeks. A single untreated rat was reserved for baseline histology. (The overall study schema is shown in Figure S1). A subgroup of 9 of 15 rats that received the MCD diet were used for MRI experiments. Six of 15 rats received the MCD diet but were not imaged due to logistical challenges related to the coronavirus disease 2019 pandemic as well as MR equipment downtime. These animals were sacrificed at various time points to provide histologic reference data. Animals were weighed every 2 weeks.

2.2 |. Serum analysis

Blood samples were collected from rat tail veins in serum separator tubes (Becton, Dickinson and Company, Franklin Lakes, NJ, USA) before each MRI time point, and serum was isolated after 30–60 min using centrifugation at 1000–1500 g. In the group of six animals that did not receive MRI, serology was performed before the animal was killed and the liver harvested. Liver injury was evaluated using standard clinical laboratory assays to quantify the serum levels of ALT, aspartate aminotransferase (AST), and triglycerides (TG).

2.3 |. Histology

At each time point, a select number of rats was randomly chosen for histology, being humanely euthanized following standard Institutional Animal Care and Use Committee procedures and livers quickly harvested. A portion of each liver was fixed in 10% buffered neutral formalin and embedded in paraffin. The paraffin blocks were cut into 4 μm sections for staining. Hematoxylin and eosin (H&E), trichrome, and Sirius red stains were performed. A board certified GI/Liver pathologist with extensive experience in animal models of liver disease reviewed each slide and assigned grades based on: the location and quantity of steatosis (grade 0: <5%, grade 1: 5%–33%, grade 2: 33%–66%, grade 3: 67%–100%), the distribution and quantity of inflammation (based on number of foci per ×200 field), the presence of liver injury (ballooning, acidophil bodies, pigmented macrophages, Kupffer cell hyperplasia), as well as the location and pattern of fibrosis. Inflammation that showed heterogeneous patterns was graded as such, intermediate between the two grades it contained.

2.4 |. Enzyme assays

Portions of each liver were processed for biochemical assays. Lactate dehydrogenase activity was measured as described previously.⁸ Briefly, liver-tissue samples weighing between 4 and 18 g were homogenized using a Tissuelyser LT (Qiagen, Germantown, MD, USA) in cell lysis buffer (Cell Signaling Technology, Danvers, MA, USA). Following centrifugation, the supernatant was diluted with phosphate-buffered saline and mixed with pyruvate at varying concentrations in a reaction buffer containing 200 μM NADH. The LDH activity was then measured spectrophotometrically by quantifying a linear decrease in NADH absorbance at λ = 339 nm using a microplate reader. The maximum velocity (V_max) and the Michaelis–Menten constant (K_m) were estimated using a Lineweaver-Burk plot. Values were normalized to total protein.

2.5 |. MR studies

Rats were anesthetized with 1.5%–2% isoflurane. MR scanning was performed on a 3T Bruker system (BioSpec 70/30; Bruker, Germany) with separate ¹H and ¹³C transmit/receive quadrature volume coils. Rat livers were imaged at baseline, week 2, week 4, week 6, week 9 and week 18, as shown in Figure S1.

Liver fat was quantified using an axial 2D multislice multi-echo gradient echo pulse sequence covering the whole liver with following parameters: TE = 2.2, 3.3, 4.4, 5.5, 6.6, and 7.7 ms, TR = 200 ms, flip angle = 15°, and slice thickness = 2 mm.

For each HP ¹³C MRSI experiment, [1-¹³C]pyruvate was polarized for 1 h using a HyperSense DNP polarizer (3.35 T, 1.4° K; Oxford Instruments) and then dissolved in mL of buffer (80 mM NaOH in Tris–HCl). All animals were injected intravenously with about 2.3 mL of 80 mM [1-¹³C]pyruvate over 12 s, and the sequence was initiated 20 s after the start of the injection. Two-dimensional ¹³C CSI was acquired with a TR = 66.4 ms, TE = 1.2 ms, center-out encoding ordering with slice thickness = 8 mm, FOV = 80 × 80 mm, matrix = 8 × 8, flip angle = 5°, number of spectral points = 128, spectral bandwidth = 2000 Hz, and spectral resolution = 7.81 Hz/point. Fifteen dynamic time points were acquired over 45 s.

2.6 |. Data analysis

Image processing was performed using MATLAB 2015b (MathWorks, Natick, MA). Fat signal fraction (FSF) maps were generated from multi-echo data using the fat-water toolbox from the ISMRM⁹ and a regularized formulation and discretized graphcut algorithm.¹⁰ Regions of interest (ROIs) were manually drawn for all axial liver slices.

¹³C metabolite maps were generated using the open-source SIVIC package¹¹ and calculated by integrating the spectral peaks of lactate, alanine, and pyruvate over time for each voxel. The area under the curve (AUC), calculated over 45 s of the acquisitions for each metabolite within a manually prescribed liver ROI, was used to calculate the lactate to pyruvate (L/P) ratio and the alanine to pyruvate (A/P) ratio.

2.7 |. Statistical analysis

Statistical analysis was performed with Prism software (version 9.0; GraphPad Software). Results were expressed as mean ± SEM. Data normality was confirmed using the Shapiro-Wilks test. Subsequently, mixed-effects analysis with Tukey’s multiple comparisons tests were used to determine statistical significance for weight, fat signal fraction, L/P AUC ratio, A/P AUC ratio, ALT, AST, and TG among time points (p < 0.05 was considered a significant difference). Pearson correlation coefficients were based on the subset of animals with complete HP-MRSI data and serum biomarkers sampling during the different imaging time points. Assessment for potential data outliers was performed using the ROUT test.

3 |. RESULTS

3.1 |. MCD diet–induced body weight loss and liver injury after 4 weeks

Compared with baseline, significant weight loss was observed at all time points after 4 weeks on MCD diet: Week 4: −3.08% ± 0.43% (p = 0.0008); Week 6: −4.78% ± 0.27% (p < 0.0001); Week 9: −12.72% ± 3.90% (p < 0.0001); and Weeks 18–20: 92% ± 2.77% (p < 0.0001) (Figure 1A). ALT and AST both increased until Week 9 on diet (Figure 1B,C) and then decreased between Weeks 9 and 18 (p = 0.029). ALT and AST at 18 weeks (ALT: 114.70 ± 7.96 U/L; AST: 126.40 ± 9.86 U/L) were no longer significantly different from baseline (ALT: 41.50 ± 3.56 U/L; AST: 80.40 ± 6.89 U/L). Serum TG levels were significantly decreased at all time points beginning at Week 4 on the diet (Figure 1D).

FIGURE 1.

(A) Body weight, (B) serum alanine aminotransferase (ALT), (C) serum aspartate aminotransaminase (AST), and (D) serum triglycerides as a function of time on the methionine- and choline-deficient diet. **p < 0.005; ***p < 0.0001 compared with baseline.

3.2 |. Substantial inflammation and mild fibrosis were observed between 4 and 6 weeks on diet

Histologic steatosis peaked early in response to MCD feeding, but liver injury and fibrosis emerged and progressed at later time points (Figure 2, Table 1). Histologic evaluation of liver sections showed severe steatosis (in 85%–98% of animals) with panacinar macrovesicular steatosis after 2 weeks on MCD diet. Hepatocyte ballooning and mild lobular inflammation were noted after 2 weeks and increased to moderate inflammation on average by Weeks 6–18. Centrizonal pericellular fibrosis was first observed between Weeks 4 and 6, and by Week 6 progressed to early septal fibrosis. Mild fibrosis was also seen around the periportal zones in MCD diet–fed animals. Although inflammation was present earlier, the combination of histologic features required to make a diagnosis of NASH was seen only after Week 6. No major progression of liver injury was observed between Weeks 6 and 18 on the diet.

FIGURE 2.

Representative hematoxylin and eosin (H&E) and sirius red images are shown for untreated control and methionine- and choline-deficient (MCD) diet–treated rats for 4, 9, and 18 weeks. Insets showing ballooning and inflammation are shown for Weeks 9 and 18. MCD diet–induced severe steatosis, inflammation, ballooning, and fibrosis over time. The white bar represents 100 μm.

TABLE 1.

Histological findings in each animal at time of sacrifice and liver harvest.

Animal number	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
Weeks on diet	0	2	2	4	4	6	6	6	6	6	9	9	9	18	18	18
Total steatosis % of cells
Macrovesicular	100%	98%	98%	90%	95%	96%	85%	100%	100%	96%	96%	100%	100%	97%	94%	94%
Microvesicular	100%	98%	98%	90%	95%	96%	85%	100%	100%	96%	96%	100%	100%	92%	91%	92%
Inflammation
Lobular inflammation	2	1	1	2	1	1	2	2	2.5	2.5	1	2	2	2	2	2
Portal inflammation	1	0	0	1	0	0	1	1	1	1	0	1	1	1	1	1
Liver cell injury
Ballooning	0	0	0	1	0	0	1	0	0	0	0	0	0	1	1	1
Acidophil bodies	0	0	1	1	1	1	1	0	0	0	1	0	0	1	1	1
Pigmented macrophage	0	0	0	0	0	0	0	0	0	0	0	0	0	1	1	1
Ductular reaction	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
Fibrosis (PicroSirusRed stain) Grade	0.5	0.5	0.5	1	0	0.5	1.5	2	2	2	0.5	2	2	1.5	1.5	1.5

Note: Pathological grading as follows: lobular inflammation: Grade 0 = no foci; Grade 1: < 2 foci per ×200 field; Grade 2: 2–4 foci per ×200 field; Grade 3: >4 foci per ×200 field. Portal inflammation: 0 = none to minimal; 1 = greater than minimal. Ballooning: 0 = none; 1 = few; 2 = many. Acidophil bodies: 0 = none; 1 = present. Pigmented macrophage: 0 = none; 1 = present. Fibrosis: 0 = none; 1 = mild (centrizonal); 2 = early septal; 3 = early bridging; 4 = cirrhosis.

3.3 |. Fat signal fraction in rat livers increased starting after 2 weeks on diet

The average FSF in the liver increased by a factor of more than 14 between baseline and Week 4 on the MCD diet (baseline: 2.22% ± 0.44%; Week 4 = 31.74% ± 4.48%, p < 0.0001; Figure 3). FSF peaked at Week 4 and then declined significantly at Week 9 (24.35% ± 3.86%, p = 0.02) and Week 18 (22.18% ± 4.27%, p = 0.008). Although the FSF in these later weeks were approximately 28% lower than the peak values, they remained more than 10 times higher than baseline.

FIGURE 3.

(A) Coronal images showing representative fat signal fraction (FSF) maps of a rat liver at baseline and (B) Week 4 on a methionine- and choline-deficient (MCD) diet. (C) Average FSF in the liver as a function of time on MCD diet. ***p < 0.0001 compared with baseline. See Table S2 for p-values between other time points.

3.4 |. Lactate and alanine production from pyruvate in rat liver decreased starting after 2–4 weeks on diet

Maps of the HP ¹³C L/P and A/P ratios showed higher metabolite signals in the anterior and peripheral aspect of the liver (Figure 4C–F). Similar spatial variation was seen across all animals and time points. After 4 weeks on diet, lactate and alanine metabolite maps showed clear decreases in the levels of these metabolites throughout the liver. These results were also seen spectroscopically by examining single voxels in the liver (Figure 5). After normalizing the curves based on total pyruvate, the alanine and lactate signals clearly decreased after 4 weeks.

FIGURE 4.

(A,D) T₂-weighted axial MRI image of the center of a rat liver at baseline (A) and after 4 weeks (D) on methionine- and choline-deficient (MCD) diet. (B,E) Spatial map of total area under the curve for lactate-to-pyruvate ratio at baseline (B) and after 4 weeks (E) on MCD diet. (C,F) Spatial map of total area under the curve for alanine-to-pyruvate ratio at baseline (C) and after 4 weeks (F) on MCD diet. White squares in panels (A)–(F) indicate the voxel locations for the spectra showed in Figure 6. Color maps have been interpolated to match the spatial resolution of the proton images. The color-mapped regions from the metabolite maps (B–E) illustrate the extent of the regions of interest used to compute the average metabolite signals in the liver. Of note, the spatial variations within the liver metabolite maps (D,F) at 4 weeks are not well seen, because the color map scale was chosen to match the metabolite maps at baseline.

FIGURE 5.

(A,B) Dynamic metabolite signal obtained from the liver voxel shown by the white square in figure at baseline (A) and after 4 weeks (B) on diet. (C,D) Dynamic spectra obtained from the liver voxel shown by the white square in figure at baseline (C) and after 4 weeks (D) on diet. A, alanine peak; L, lactate peak; P, pyruvate peak.

When metabolite levels were averaged over the entire liver, the average L/P ratio decreased after 2 weeks on diet (p = 0.03), with the lowest L/P ratio observed at Week 4 (p < 0.0001; Figure 6A). After reaching a minimum value at Week 4, the L/P ratio increased slightly (8.6%) and remained steady between Weeks 6 and 18. Although L/P ratio for later time points (Weeks 6–18) were higher than at Week 4, L/P ratios for all times points were lower than at baseline.

FIGURE 6.

(A) Ratios of the areas under the curve for the total lactate and pyruvate measured over the entire time course (L/P AUC ratio), as a function of the number of weeks on the methionine- and choline-deficient (MCD) diet. (B) Ratios of the areas under the curve for the total alanine and pyruvate measured over the entire time course (A/P AUC ratio), as a function of the number of weeks on the MCD diet. Mean ± SEM are shown. *p < 0.05; **p < 0.005; ***p < 0.0001, compared with baseline. See Table S2 for p-values between other time points.

The A/P ratios in rat liver similarly dropped between baseline and Week 4 (p = 0.02) on diet and remained low at Week 6 (p = 0.04) (Figure 6B). For the other time points, A/P values tended to be lower compared with those calculated at the baseline, although not statistically significant.

3.5 |. Metabolite levels correlated with serum markers of liver injury

For those time points when animals underwent HP ¹³C MRSI, L/P and A/P AUC ratios were compared with serum measures of liver injury (Figure 7). Both L/P and A/P AUC ratios were negatively correlated with AST (L/P vs. AST r = −0.61, A/P vs. AST r = −0.57) and ALT (L/P vs. ALT r = −0.39, A/P vs. AST r =−0.62) (all p < 0.05). L/P and A/P were both positively correlated with serum TG (L/P vs. TG r = 0.60, A/P vs. TG r = 0.59) (p < 0.002). Of note, in two animals there were single baseline laboratory values (AST = 762 U/L in one, TG = 145 mg/DL in the other) that were several SDs beyond the mean and, as all the other laboratory measures in these other animals were normal, these two single values were treated as outliers and not included in the analysis.

FIGURE 7.

(A,C,E) Ratio of the areas under the curves for lactate and pyruvate as a function of serum aspartate aminotransferase (AST) (A), alanine aminotransferase (ALT) (C), and triglycerides (E). (B,D,F) Ratio of the areas under the curves for lactate and pyruvate as a function of serum AST (B), ALT (D), and triglycerides (F). Pearson’s correlations with p < 0.05 were considered significant. Single anomalous values of AST and triglycerides at baseline in (E) and (F) were attributed to measurement error and not included in analysis. A/P, alanine and pyruvate; L/P, lactate and pyruvate.

3.6 |. LDH activity in the liver increased in rats on MCD diet

To determine whether the observed changes in HP ¹³C lactate labeling might be explained by changes in LDH, we measured LDH activity in animals for which livers were extracted for histology. For LDH activity assays, the average protein concentration of sample extracts was 220 ± 67 μg/mL, as determined from Bradford protein assays. LDH activity (V_max) increased by a factor of 3, from 0.6 ± 0.25 μM NADH/min/[protein] at baseline to a peak value of 1.88 ± 0.13 at 9 weeks (p < 0.05) μM NADH/min/[protein]. After 18 weeks, activity had decreased from the peak but remained slightly more than twice the baseline value (Figure 8).

FIGURE 8.

Lactate dehydrogenase (LDH) activity as measured by V_max in liver tissue as a function of time on methionine- and choline-deficient diet. Statistical significance is measured compared with baseline (*p < 0.05, **p < 0.005).

4 |. DISCUSSION

In this work we studied the progression of NAFLD to NASH using HP [1-¹³C]pyruvate MRSI imaging in a rat MCD diet model. Although this animal model does not match the obese phenotype that commonly accompanies human NASH (in fact, the animals lose weight), the MCD model has been proposed as a useful tool for understanding the progression of NAFLD because of its ability to mimic the histological changes of human NASH as well as its ability to induce oxidative stress.^4,12 Indeed, in our study, we observed substantial hepatic inflammation after Week 4 and mild fibrosis after Week 6 on diet, features that have been linked to fat accumulation, reactive oxygen species production, and oxidative stress.^3,13,14

The mechanism of weight loss in rodents fed MCD diet is still unclear and a matter of debate. Impairment of the gene-encoding stearoyl-coenzyme A desaturase-1 during MCD feeding has been suggested as a key factor leading to hypermetabolism, which includes increased total body energy expenditure and up-regulated fatty acid β-oxidation.² However, several studies have found a down-regulation of fatty acid β-oxidation because of an impairment of mitochondrial enzymes carnitine palmitoyltransferase 1 and 2, involved in the transport of long-chain fatty acids across the mitochondrial membranes.^13,15,16

In the present work we used male Wistar rats, which have been shown to develop more steatosis and more severe liver injury than female rats or rats of other strains following MCD feeding.¹ Our MRI results showed that MCD feeding lead to a significantly increased fat signal fraction in rat livers, which matched the severe steatosis observed histologically. This hepatic steatosis could be the outcome of lipid accumulation resulting from phosphatidylcholine deficiency halting lipid export, with low serum TG levels observed after 2 weeks on diet. In line with this, prior work in rodents fed MCD diet have reported that decreased TG serum levels produce an imbalance between fatty acid influx and export as a component of very-low-density lipoprotein particles, contributing to the development of NAFLD.^15,17

There is increasing interest in using HP ¹³C MRSI to diagnose and monitor liver injury noninvasively.¹⁷ In this work, we observed a decrease in lactate and alanine production from pyruvate within the livers of rats fed the MCD diet. This result was somewhat surprising given that a prior study of rats fed a high-fat diet (HFD) found increased hyperpolarized pyruvate-to-lactate conversion in livers of rats who developed fatty liver disease (although not necessarily NASH).⁶ In addition, toxic liver injury as seen in CCl₄ injection has also been shown to increase hyperpolarized lactate signal.⁵ Several reasons could potentially explain the difference between the HFD study and our findings. The first is related to the differences in the biochemical and pathological characteristics of these two NAFLD models, with more severe injury occurring in the MCD model. Only minor changes in serum ALT levels were observed in rats fed HFD, suggesting minimal liver injury, whereas using the MCD diet we observed significantly increased ALT levels, indicating more substantial liver injury.^1,2,12

Another surprising result was that despite the observed decrease in hyperpolarized ¹³C lactate signal observed for animals on the MCD diet, the activity of LDH (required to catalyze the conversion of pyruvate to lactate) actually increased. This somewhat contradictory result may be explained by decrease in available cofactors such as NADH. The conversion of pyruvate to lactate via LDH requires conversion of the cofactor NADH to NAD+. The oxidative stress and inflammatory injury accompanying the MCD diet has been shown to induce hyperactivation of poly ADP-ribose polymerase¹² and is associated with depletion of ATP and NAD⁺. The depletion of this cofactor would explain the decreased conversion of pyruvate to lactate observed in our study. Supporting this, we found that the ex vivo activity of LDH in liver tissues (where excess NADH was available) was increased. The presence of oxidative stress and depleted glutathione also has been shown in vivo using HP ¹³C DHA in mice fed MCD diet.⁷ In this earlier study, decreased conversion of HP ¹³C DHA to ¹³C-labeled vitamin C was observed after 2 weeks on the MCD diet, with restoration of normal levels after stopping the diet for 1 week.

Although depletions in cofactors may explain decreases in HP ¹³C lactate signal, they do not fully explain the observed decreases in HP ¹³C alanine signal (because NADH is not required). The explanation for the observed decrease in alanine is uncertain and requires further study. Because the observed HP ¹³C metabolic signal is highly influenced by the MCT transporters,¹⁸ changes in transporter activity could also potentially contribute to the observed results.

HP ¹³C bicarbonate, which can be sometimes observed as a downstream metabolite of HP ¹³C pyruvate, was not observed in our studies, likely due to SNR effects.

In this study, we administered a fixed dose of HP ¹³C pyruvate, which in our laboratory is a standard dose independent of weight. Weight loss experienced over the course of the study could potentially affect results, especially because it has been shown that observed HP ¹³C pyruvate metabolism is transport limited, and we give an excess of HP ¹³C pyruvate.¹⁸ Although this may have affected our results, we expect the effect to have been small, given that the changes in metabolite levels occurred early (weeks 2–4), whereas most weight loss did not occur until late (weeks 4–12).

L/P and A/P ratio maps all showed a spatial gradient with higher levels at the anterior parts of the liver and lower levels in the posterior parts. This gradient was observed even after the animals were placed on diet. This gradient was most likely caused by partial volume effects related to the low spatial resolution of our acquisition (8 × 8 matrix size), with high intravascular pyruvate levels in large vessels leaking into adjacent voxels.

This study had several limitations. We did not have measurements of hyperpolarized ¹³C pyruvate metabolism in rats that were fed normal chow diet. However, in prior studies of diet-induced fatty liver disease, there was no observed change in lactate or alanine relative to baseline in control groups that did not receive the intervention.⁶ In another study of high-fat diet, the control group showed a small increase in lactate production at 2 weeks, although statistical comparison was not made.¹⁹ Lactate production did not change at subsequent time points, and there was no change at all in alanine production. Therefore, we are confident that the results observed principally reflect the effects of our intervention. In addition, multiple experiments using rapid sequential injections of hyperpolarized ¹³C pyruvate have shown minimal perturbation of metabolite signal, suggesting that intermittent injections of hyperpolarized ¹³C pyruvate in this study are unlikely to affect our observations.²⁰ We did not have direct measurements of the redox ratio NAD⁺/NADH in the rat livers we imaged, but NAD⁺ levels have been measured previously and noted to decrease in the context of the MCD diet in mice.¹² Moreover, our longitudinal study design precluded us from performing histological analysis for each animal in which we performed HP MRSI. Instead, we performed histology in a select subset of animals, and we measured serum markers of liver injury at every time point, a standard noninvasive technique that is used widely for evaluating liver function. Future work should focus on validating our findings and obtaining complementary data on the changes in hepatic metabolism in NAFLD using other ¹³C-labeled probes such as [2-¹³C]dihydroxyacetone. HP [2-¹³C]dihydroxyacetone has been recently proposed to probe hepatic TG-precursor synthesis and glucose metabolism.²¹

Finally, this work supports the growing literature that HP ¹³C MRSI can detect metabolic changes due to liver injury. At the microscopic scale, liver is involved in energy regulation, glucose metabolism, and amino acid metabolism. Understanding these metabolic pathways could identify new therapeutic targets for NAFLD or NASH, involving anti-hyperglycemics, insulin sensitizers, gluconeogenesis inhibitors, and modulators of lipid metabolism. Thus, HP ¹³C MRSI and its different probes could potentially help understand and diagnose metabolic diseases, as well as monitor treatment response in patients.

5 |. CONCLUSION

In this study, we detected changes in liver fat as well as altered metabolic conversion of pyruvate-to-lactate in a dietary model of NASH. This suggests that HP ¹³C pyruvate metabolism is sensitive to the effects of oxidative stress in the liver. The fact that changes in pyruvate metabolism with the MCD diet are contrary to that seen in simple fatty liver disease suggests that careful analysis will be required to understand the changes of HP ¹³C pyruvate metabolism in human patients with NASH.

Supplementary Material

supplementary info

Figure S1. Timeline and details of the experimental acquisition. Blood was obtained from all surviving animals at all time points. At each time point, select animals were sacrificed to determine histology.

Figure S2. Representative MR spectrum in a 2D-CSI voxel (summed over 15 dynamic time points) at baseline (A) and after 4 weeks on methionine- and choline-deficient (MCD) diet (B).

Table S1. Description of procedures performed on each animal.

Table S2. Tukey’s multiple comparisons test for weight, fat signal fraction, lactate, and pyruvate measured over the entire time course (L/P AUC ratio), alanine and pyruvate measured over the entire time course (A/P AUC ratio), alanine aminotransaminase (ALT), aspartate aminotransaminase (AST), and triglycerides.