All animal experiments were conducted in accordance with federal regulations (Swiss Federal Act on Animal Protection) and were authorized by the Service de la consommation et des affaires vétérinaires (SCAV – EXPANIM) of the Canton of Vaud, Switzerland. Male Wistar (n = 14, 250.9 ± 15.9 g) were anesthetized with isoflurane in a 1:1 mixture of air and oxygen and fitted with catheters in the femoral arteries and vein. Blood was sampled from an arterial catheter following each hyperpolarized experiment and analyzed immediately using an epoc system with blood gas, electrolyte and metabolite measurement cards (Siemens Healthcare AG, Zurich, Switzerland). The respiration rate was maintained around 60 breaths per minute by adjusting the isoflurane level, and body temperature was kept between 37.5 and 38.5°C using warm water circulating through silicone tubing placed next to the rat. Breathing was monitored with a pneumatic pillow coupled to a pressure transducer and body temperature with a rectal probe. Both were connected to a monitoring system (SA Instruments, Inc., Stony Brook, NY, United States). The fasted group was deprived of food overnight for 12 h. When used, the PEPCK inhibitor 3-mercaptopicolinic acid (3-MPA) (TRC, Toronto, ON, Canada) was administered intraperitoneally (100 mg/kg, neutralized, in PBS) 1 h prior to the hyperpolarized pyruvate infusion. In five experiments, the rats were scanned twice with hyperpolarized pyruvate, with 3-MPA treatment prior to the second infusion. Animals were sacrificed after scanning under 4% isoflurane anesthesia by intravenous sodium pentobarbital.

Ten microliters of 1-¹³C pyruvic acid (Sigma-Aldrich Chemie GmbH, Buchs SG, Switzerland) formulated with 21 mM OX63 trityl radical (Albeda Research ApS, Copenhagen, Denmark) was frozen in a sample cup, along with 13 μl of 10 M NaOH for neutralization, and polarized in a custom-built instrument (7 T and 1.05 K) (Comment et al., 2007; Cheng et al., 2013) with microwave irradiation (196.59 GHz, 55 mW). The sample was dissolved with 5.5 ml of hot 47 mM sodium phosphate, 100 mM NaCl, 2.7 mM KCl, 0.3 mM EDTA, pH 7.4 in D₂O, rapidly transferred to the scanner magnet, and 1.3 ml was infused intravenously (∼0.13 mmol/kg).

Data were acquired in a 9.4 T horizontal-bore magnet (Magnex Scientific, Abingdon, United Kingdom) with a VNMRS console (Varian, Palo Alto, CA, United States) and VnmrJ 3.2 (Agilent Technologies, Santa Clara, CA, United States). A surface coil equipped with a single loop tuned to ¹H (400.2 MHz) and two 16 mm loops in quadrature tuned to ¹³C (100.67 MHz) was placed over the left kidney (to avoid signals from the liver), its position verified by ¹H GEMS MRI, and shimmed using FASTESTMAP (Gruetter and Tkác, 2000). A series of ¹³C magnetic resonance spectra were acquired starting with the pyruvate infusion (∼2.9 s repetition time, with respiratory gating and cardiac triggering, 30° BIR4 adiabatic excitation, 20161.3 Hz spectral width, 8258 complex points, WALTZ-16 ¹H decoupling).

Spectral signals were fit using Bayes (Washington University, St. Louis, MO, United States¹). Both the individual spectral time series and sums of spectra containing hyperpolarized signal from each infusion were analyzed. The metabolite areas under the curve for each series were calculated using the signal amplitudes and spectrum acquisition times. Metabolite signals were analyzed normalized to both the sum of all hyperpolarized signals and to the sum of all metabolite signals. Statistical significance was assessed by two-way ANOVA in R² with Bonferroni correction to account for multiple comparisons. Graphical error bars and values in the text and tables report the mean ± SD.

Consistent with prior reports (Laustsen et al., 2014; von Morze et al., 2017), fasting and the resulting decrease in insulin levels results in lower renal conversion of [1-¹³C]pyruvate to [¹³C]bicarbonate. Normalized to the total metabolite signal of the summed spectra, the untreated fed bicarbonate signal of 0.074 ± 0.022 decreased to 0.023 ± 0.008 in the untreated fasted group (p = 0.00073) (Figure 3A). Conversely, [1-¹³C]aspartate signal increased from 0.031 ± 0.011 in the fed group to 0.068 ± 0.037 in the fasted rats (p = 0.038) (Figure 3B). The fed rats had a slightly higher fraction of alanine (Figure 3D) and lower lactate (Figure 3C), but these differences were not significant.

While 3-MPA treatment did not affect the normalized [¹³C]bicarbonate signal in either the fed or fasted states (Figure 3A), it resulted in a significant decrease in the aspartate C1 signal, which was 2.8-fold lower in the fed case and 15-fold lower in fasted rats (p = 0.0010) (Figure 3B). Lactate C1 as a fraction of the total metabolite signal was 1.3-fold higher (p = 0.014) (Figure 3C) in the 3-MPA-treated rats, and alanine C1 correspondingly trended downward (Figure 3D).

Similar metabolic differences between the groups were also obtained when normalizing the summed spectra metabolite signals to the total hyperpolarized signal including pyruvate (“total carbon”) (Supplementary Figure 1) or using the dynamic spectral signal area-under-the-curve, normalized either to the total metabolite signal or total carbon (Supplementary Figures 2, 3). However, the effect of 3-MPA on the lactate signal normalized to total carbon was not significantly different even though the mean change was almost identical.

Fasting and treatment with 3-MPA had the expected effect on blood glucose and lactate levels (Table 1). Fasting glucose and lactate were lower in fasted rats, but they were not significantly different. 3-MPA treatment did not significantly change glucose and lactate levels in fed rats, but it induced hypoglycemia and a large increase in lactate in fasted rats.

The substantial contribution of PEPCK flux to hyperpolarized pyruvate-to-bicarbonate conversion in the fasted rat liver raised the question whether a similar process was operating in the kidney, given its gluconeogenic function. The marked decrease of pyruvate-to-bicarbonate conversion with fasting and the lack of an effect by 3-MPA treatment shown here clearly demonstrate that any contribution by PEPCK is minor, and indicate that most, if not all, of the bicarbonate is due to PDH flux. The conversion of labeled pyruvate to bicarbonate via PEPCK depends on label exchange to [4-¹³C]oxaloacetate, which involves at least five steps (Figure 1), and it is more challenging to detect in the presence of PDH flux. The [4-¹³C]malate signal provides an indication of the amount of label exchange, and it is much more prominent in the liver than the kidney. Whereas the malate C4 signal normalized to the total metabolite signal was ∼0.025 in the fasted liver (Can et al., 2021) it was considerably lower in the kidney, at 0.007 or less (Supplementary Figure 4). Lower production of [4-¹³C]oxaloacetate in the kidney may therefore account for the apparent lack of [¹³C]bicarbonate resulting from PEPCK flux.

The increased pyruvate-to-aspartate conversion in the fasted rat kidney corresponds to the PC flux required to balance the increase in PEPCK flux in this metabolic state. This, and the greatly diminished production of labeled aspartate following inhibition of PEPCK, particularly in the fasted state, suggests that this signal provides a readout of renal gluconeogenic flux. The decreased aspartate and malate signals following 3-MPA treatment are very likely not due to a decrease in their pool size since their renal concentrations increase in fed (Bennett and Alleyne, 1976) and starved rats following treatment, with aspartate at 3.9-fold and malate at 7.1-fold their untreated levels (Vinay et al., 1980). The increased malate levels may result in the lower labeling of malate and aspartate by inhibiting PC (Scrutton and White, 1974).

The treated fed rats had higher aspartate labeling compared to the treated fasted group, so the aspartate signal may not precisely correspond to gluconeogenesis-driven PC flux, assuming complete or equivalent levels of PEPCK inhibition was achieved in both cases. PC is widely expressed throughout the rat nephron, with activity in the proximal tubule twice that in the distal tubule (Salto et al., 1991). Similarly, aspartate aminotransferase (AST) activity, which interconverts oxaloacetate and aspartate, is also widely expressed, with its activity highest in the distal straight and distal convoluted tubule and slightly lower in the proximal tubule (Chan et al., 1979). Nonetheless, the large decreases in normalized aspartate signal following 3-MPA treatment are consistent with most of the labeled aspartate resulting from gluconeogenic PC flux and therefore occurring in the proximal tubule. The malate–aspartate shuttle (Borst, 2020) plays a critical role in gluconeogenesis from pyruvate by transporting oxaloacetate equivalents from the mitochondria to the cytosol and serving as a bridge between the main sites of PC and PEPCK activity. The decreased aspartate therefore likely results from decreased aspartate turnover or shuttle activity secondary to PEPCK inhibition.

The considerable variation in the normalized aspartate signal in fasted rats (Figure 3B) may reflect its sensitivity to small metabolic differences. But the degree to which hyperpolarized pyruvate-to-aspartate conversion can serve as a general marker of renal gluconeogenesis would need to be validated in a study that measures and relates the pyruvate metabolism to independent measures of renal gluconeogenesis. Other conditions with increased renal gluconeogenesis, such as diabetes, starvation, and acidosis, may result in even greater production of labeled aspartate. Given its lower intensity and proximity to the alanine peak, resolving the aspartate spectral peak requires careful adjustment of magnetic field homogeneity, and it may be difficult to quantitate by spectroscopic imaging. It is nonetheless encouraging that it can be resolved in the rat kidney with a 3T clinical scanner (von Morze et al., 2017).

Although the decreased conversion of pyruvate to alanine with 3-MPA treatment did not reach the threshold for statistical significance by two-way ANOVA, there are clear differences, and the exclusion of the one high outlying point in the fasted 3-MPA group (Figure 3D) would result in a significant difference. A t-test comparing the untreated and treated fed groups alone yields an uncorrected p-value of 0.009, and one can reasonably anticipate seeing lower alanine with 3-MPA treatment in a study with higher statistical power.

The kidney is an anatomically and metabolically complex organ, and different segments of the nephron are functionally distinct and express different ensembles of enzymes to support their metabolic roles (Schmidt and Guder, 1976; Guder and Ross, 1984). The proximal convoluted tubule is the main site of renal gluconeogenesis, while the distal tubule uses oxidizes glucose and lactate as fuel (Klein et al., 1981). These metabolic compartments are schematically illustrated in Figure 4. The observed malate and aspartate are then largely produced in the proximal convoluted tubule, and the distal tubule is likely the main site of bicarbonate production by PDH flux.

The distinct metabolic roles of these nephron segments raises some potential caveats regarding the normalization of the hyperpolarized signals. If most of the filtered pyruvate is reabsorbed in the proximal tubule (Jorgensen and Sheikh, 1988), the pyruvate metabolism attributed to the distal tubule would then result mainly from pyruvate delivered by the capillaries (Ullrich et al., 1984) rather than the filtrate in the lumen. These different routes of substrate delivery may result in different capacities and rates of pyruvate uptake in the various tubule cell populations, and they are another mechanism that may affect the apparent pyruvate metabolism via one pathway over another. With the mean pyruvate signal ranging from 67 to 70% of the total ¹³C signal, a gross change in pyruvate uptake is not apparent in the fasted or 3-MPA treated groups, and the simplifying assumption of analyzing the sum renal metabolites is useful.

In addition to the important limitation imposed by the semi-quantitative nature of hyperpolarized ¹³C metabolism in vivo, the current study relies on the placement of the surface coil to select the renal signals. The left kidney was scanned in order to be better isolated from any signals from the liver, but the coil may also detect signals from the nearby spleen. Additionally, the non-uniform sensitivity of the surface coil results in the closer cortical regions of the kidney having a greater contribution to the measured metabolite signals.

In summary, we demonstrate that the conversion of hyperpolarized [1-¹³C]pyruvate to aspartate in the rat kidney is increased in the fasted state and greatly decreased following PEPCK inhibition, making aspartate a promising metabolic marker of renal gluconeogenesis. By contrast, the decreased renal conversion of pyruvate to bicarbonate in the fasted state was not significantly influenced by inhibition of PEPCK, unlike the case in rat liver, providing further evidence that most of hyperpolarized bicarbonate production in the kidney is due to flux through PDH.