Enrichment of hepatic glycogen and plasma glucose from H₂18O informs gluconeogenic and indirect pathway fluxes in naturally feeding mice

Abstract Deuterated water (2H2O) is a widely used tracer of carbohydrate biosynthesis in both preclinical and clinical settings, but the significant kinetic isotope effects (KIE) of 2H can distort metabolic information and mediate toxicity. 18O‐water (H2 18O) has no significant KIE and is incorporated into specific carbohydrate oxygens via well‐defined mechanisms, but to date it has not been evaluated in any animal model. Mice were given H2 18O during overnight feeding and 18O‐enrichments of liver glycogen, triglyceride glycerol (TG), and blood glucose were quantified by 13C NMR and mass spectrometry (MS). Enrichment of oxygens 5 and 6 relative to body water informed indirect pathway contributions from the Krebs cycle and triose phosphate sources. Compared with mice fed normal chow (NC), mice whose NC was supplemented with a fructose/glucose mix (i.e., a high sugar [HS] diet) had significantly higher indirect pathway contributions from triose phosphate sources, consistent with fructose glycogenesis. Blood glucose and liver TG 18O‐enrichments were quantified by MS. Blood glucose 18O‐enrichment was significantly higher for HS versus NC mice and was consistent with gluconeogenic fructose metabolism. TG 18O‐enrichment was extensive for both NC and HS mice, indicating a high turnover of liver triglyceride, independent of diet. Thus H2 18O informs hepatic carbohydrate biosynthesis in similar detail to 2H2O but without KIE‐associated risks.

H or 2 H has been widely used for the study of biosynthetic activities in whole organisms, including the endogenous synthesis of glucose and glycogen. [1][2][3][4][5] An important advantage of this approach over methods involving 13 C-labeled substrates is that labeled water rapidly distributes throughout the entire organism, thereby providing constant precursor enrichment for all tissues that can be easily measured from blood, saliva, or urine. Moreover, for both glucose and glycogen, the incorporation of water hydrogen into certain product positions is independent of precursor sources, while the enrichment of other positions is specific to a particular precursor source. 4,6,7 Thus, this approach provides precise estimates of fractional synthetic or turnover rates of glucose or glycogen irrespective of the available precursors, while at the same time resolving the contributions of different precursors to the overall synthesis rate. The principal uncertainties associated with the use of hydrogen isotopes in this manner include significant discrimination against enrichment of certain product positions because of strong kinetic isotope effects (KIE). 8 Discrimination of 2 H incorporation from water into position 2 of glycogen has been observed in both mice and fish. 9,10 Nevertheless, methods for quantifying positional 2 H-enrichments of glucose and glycogen have been developed for both mass spectrometry (MS) 11,12 and nuclear magnetic resonance (NMR) spectroscopy. 7,[13][14][15] The oxygens of water also undergo exchanges and/or additions with precursors of glucose and glycogen synthesis, hence in the presence of 18 O-enriched water (H 2 18 O), the hexose oxygens become enriched with 18 O (Figure 1). In contrast to hydrogen isotopes, the much smaller relative mass difference between 18 O and 16 O means the absence of significant isotope discrimination because of KIE. We recently demonstrated 18 O-incorporation from H 2 18 O into the oxygens of glucose-6-phosphate in hemolysate preparations that was in accordance with known exchange mechanisms acting at the triose and hexose phosphate levels. 16 These measurements were performed by 13 C NMR analysis of the monoacetone glucose derivative and quantification of the isotope-shifted 18 O-signals. The objective of the present study was to translate this approach into animal models and test the applicability of H 2 18 O as a tracer for hepatic carbohydrate biosynthesis. Because the 18 O-enrichment levels achievable in vivo are typically less than in vitro, we sought to increase the sensitivity of our 13 C NMR measurements through a novel glucose derivative: 3,5,6-tri-O-[1-13 C]acetyl monoacetone glucose (TAMAG). For this derivative, the number of 13 C reporter nuclei for oxygens 3, 5, and 6 is amplified by a factor of nearly 100 over the background 13  O. Mice were euthanized the next morning and biological fluids and tissues were collected. Arterial blood was collected and immediately centrifuged at 2000 Â g for 5 min at 4 C and plasma was isolated and stored at À80 C for further experimental analysis, including for the analysis of body water 18 O-enrichment. Liver and other tissues, including adipose tissue depots, were freeze-clamped and stored at À80 C until further processing.

| Liver glycogen and triglyceride purification
Frozen liver powder from each animal (approximately 500 mg) was treated with methanol (4.6 ml/g) and methyl tert-butyl ether (15.4 ml/g) for lipid extraction. Glycogen was extracted from the pellet by treatment with 30% potassium hydroxide (10 ml/g) at 70 C for 1 h. The mixture was treated with 8 ml/g 6% Na 2 SO 4 and glycogen was precipitated with 50 ml/g ethanol. After centrifugation, the solid residue was resuspended in water and the pH was adjusted to 8. After drying, the pellet was resuspended in 5 ml of 0.05 M acetate buffer, pH 4.5. For glucose quantification, 100 μl of each sample was collected prior to glycogen digestion. To the remaining glycogen, 200 U of amyloglucosidase from Aspergillus niger (59.9 U/mg; Fluka, NC, USA) dissolved in the acetate buffer was added and incubated for 5 h at 55 C. After collection of 100 μl for glycogen quantification, the supernatant was deproteinized and evaporated. Glycogen was quantified by analyzing the glucose in both 100-μl aliquots using a Miura 200 (ISE, S.r.l, Rome, Italy). Triglycerides were purified from the evaporated organic fraction with solid-phase extraction as previously described. 17 The triglycerides were then transesterified with sodium methoxide in methanol 18 and the aqueous fraction containing glycerol was evaporated.

| Blood sampling and processing
Blood was harvested without anticoagulants. A few microliters of blood was placed on 903 Protein Saver Snap Apart Card filter paper (Whatman, GE Healthcare Ltd, Cardiff, UK), which was then dried at room temperature and stored in a desiccator 19

| Derivatization of glucose to TAMAG
Glucose from both liver glycogen and blood extracts was derivatized to TAMAG. Each lyophilized extract was vigorously mixed with 5 ml of acetone containing 4% sulfuric acid (v/v) for 8 h. The mixture was stirred overnight at room temperature to yield diacetone glucose. The acetonation reaction was quenched by adding 5 ml of water and the pH was adjusted to 2 with 1 M Na 2 CO 3 . The newly formed diacetone glucose was hydro-

| 13 C NMR spectroscopy and data analysis
Proton-decoupled 13 C NMR spectra of TAMAG prepared from liver glycogen and blood glucose were acquired at 25 C with a Bruker Avance III 800 MHz NMR spectrometer (Billerica, MA, USA) operating at 201. 16 MHz for 13 C. A 5-mm TXI cryoprobe was utilized to record 13 C NMR data. 13 C NMR spectra were obtained with a 45-degree pulse, an acquisition time of 2.5 s, and an interpulse delay of 0.5 s. The number of acquisitions ranged from 2000 to 3000. All spectra were analyzed with ACD/NMR Processor 12 spectral analysis software (Advanced Chemistry Development, Inc., Toronto, ON, Canada). The relative areas of 18 O-isotopic-shifted and parent 13 C signals were calculated by peak fitting. Positional 18 Oenrichment (%) was calculated as follows:  were water and acetonitrile, respectively.
The LC was programmed as follows: 80%-30% of B (0-9 min), 30%-20% of B (9-10 min), 20% of B (10-11 min), and 20%-5% of B (11-12 min). The ionization source ESI Turbo V was operated in the negative mode set to an ion spray voltage of 4500 V, 35 psi for nebulizer gas 1, 0 psi for the nebulizer gas 2, 30 psi for the curtain gas, and a temperature of 450 C. All molecules were analyzed by multiple reaction monitoring (MRM) settings Q1 and Q3 at unit resolution, the entrance potential at À4 eV, and the collision gas was at À3 psi. The MRM detection window was set to 1235 ms and the target scan time to 60 ms. The collision energy was set at À14 eV and declustering potential at À50 eV for all transitions. This method has been previously validated for analysis of [U-13 C]-and [6,6-2 H]glucose enrichments in DBSs of rats administered with these tracers. 19 Peak area ratios were calculated from integrated peak areas of the analyte and internal standard on MultiQuant 2.1.1 (AB Sciex). Quantification was performed for glucose M + 0, M + 2, M + 4, and M + 6, by monitoring the sum of the transitions of each glucose species (Table S1).
Enrichment of glucose M + 2 was corrected for background isotope contributions using data from the glucose M calibration curve. 2.9 | GC-MS of DBSs and data analysis A 6.5-mm diameter circle from the DBS was placed in an Eppendorf tube, to which 40 μl of water was added, followed by 400 μl of ethanol.
The solution was agitated for 45 min at room temperature then centrifuged at 9300 Â g for 5 min with a benchtop Eppendorf centrifuge. The The following parameters were used for GC-MS analysis of glycerol 3-TBDMS derivative: the initial oven temperature of GC was 60 C for 1 min followed by a ramp of 10 C/min up to 325 C and finally a hold time of 5 min. 22 The isotopic 18 O and natural abundance correction was performed as for the glucose aldonitrile pentapropionate fragments.
All comparisons of 18 O enrichments between different glucose-6-phosphate positions and between diets were evaluated by the Mann-Whitney test. The significance level was set as two-sided with α = 0.05.

| Analysis of glycogen 18 O-enrichment by 13 C NMR
Both groups of mice had similar liver glycogen content (146 ± 27 μmol/g wet weight NC and 117 ± 34 μmol/g wet weight HS) and the 18 Oenrichment levels of plasma water were also similar (4.75% ± 0.32% NC and 4.95% ± 0.51% HS). The TAMAG derivatives of liver glycogen presented intense and narrow signals for the 13 C-enriched carboxyls bound to oxygens 3, 5, and 6 of the TAMAG hexose moiety with well-resolved 18 O-shifted signals (chemical shift distance between 16

| Blood glucose analysis by GC-MS, LC-MS/MS, and 13 C NMR
TAMAG derivatives prepared from blood glucose yielded 18 O-shifted signals with poorer signal-to-noise ratios compared with those prepared from glycogen because of the lower quantities of the analyte. The 18 O-enrichment distributions in positions 3, 5, and 6 resemble those seen in glycogen, with position 5 being more highly enriched than either 6 or 3 ( Figure S1). The 18 O-enrichment values were also more dispersed compared with those of liver glycogen, therefore the differences in positional 18 O-enrichments within and between the groups were less well defined.
Overall, the enrichment levels of plasma glucose were systematically higher compared with those of glycogen. This may in part be explained by incomplete turnover of liver glycogen during the overnight feeding period, leading to dilution of the 18 O-enrichments by pre-existing unlabeled glycogen.

| Liver triglyceride glycerol 18 O-enrichment analysis by GC-MS
The average 18 O-enrichment of liver triglyceride glycerol oxygens as measured by GC-MS of three different fragments is shown in Table 2. Overall, the average glycerol 18 O-enrichment levels tended to be systematically higher compared with those measured in plasma glucose and indicate a high turnover rate of hepatic triglyceride glycerol during the overnight interval. Unlike plasma glucose, the average glycerol 18 O-enrichment levels did not significantly differ between NC and HS mice for any of the fragments studied. However, the distribution of 18 O-enrichment among the three glycerol oxygens was less even for HS compared with NC mice, as seen by a significantly higher enrichment of the 1-2 (2-3) compared with the 1 (3) fragments.
T A B L E 1 Mean 18 O-enrichment of selected plasma glucose oxygens as measured by LC-MS/MS transitions representing oxygens 1-6, and GC-MS fragments representing oxygens 2-5, 2-4, 4-6, and 5-6 for a group of eight mice fed NC and a second group of eight mice fed a HS diet. Data are normalized to the number of oxygens in each glucose species. Also shown is the plasma water 18 O-enrichment for both groups and estimates of the gluconeogenic contribution to blood glucose appearance calculated from the ratio of mean 18  O is diluted by water (including water generated by respiration) and additionally by respiratory CO 2 and exchange of its oxygens with those of water, while 2 H 2 O is diluted by water alone. This provides the basis for estimating whole-body respiration by measuring the differential dilution of ingested 2 H 2 O and H 2 18 O. 25 In mice, daily CO 2 production represents about 20% of the total oxygen content of body water. 26  Under these conditions, excess 18 O-enrichment levels of plasma glucose and liver glycogen were in the range of 1.5%-4.0%. For the isotope- Blood glucose analysis revealed that gluconeogenesis also accounted for a substantial fraction of postprandial circulating glucose for both the NC and HS groups. This has also been previously reported in rats that were studied under similar postprandial conditions with high sucrose and NC diets, 30 as well as in fasted rats that were given an intraperitoneal glucose load, 31 and following protein meals supplemented with different mixtures of glucose and fructose in overnight-fasted healthy humans. 32 There was a significantly increased gluconeogenic contribution to postprandial glucose levels for HS compared with NC mice, consistent with the gluconeogenic conversion of fructose to glucose. In the study conducted by Barosa et al., the postprandial gluconeogenic fraction was significantly higher in healthy subjects following a protein meal supplemented with high fructose (55% fructose/45% glucose) compared with a meal supplemented with 5% fructose and 95% glucose. 32 However, in rats whose NC was supplemented with sucrose in the drinking water, the total gluconeogenic contribution was not significantly different from rats fed NC alone, but the sources of gluconeogenic carbons were significantly shifted from anaplerotic precursors to those entering at the triose phosphate level, which include fructose. 30 The triglyceride glycerol is derived from dihydroxyacetone phosphate via glycerol-3-phosphate (  O has also been used to measure the turnover of the acyl moieties of membrane phospholipids of peritoneal exudate cells in culture based on carboxyl and water oxygen exchange. 33 However, for the triglycerides in the current study, 18 O-enrichments of the fatty acyl oxygens were assumed to have undergone extensive exchange with those of the reagents during transesterification, hence analysis of the fatty acyl 18  O. This is particularly relevant when applying H 2 18 O to studies of endogenous glucose sources during fasting, where the interval between labeled water administration and sampling of blood glucose may be only a few hours.