Transfer of glucose hydrogens via acetyl-CoA, malonyl-CoA, and NADPH to fatty acids during de novo lipogenesis

Abstract

Deuterated water (²H₂O) is widely used for measuring de novo lipogenesis (DNL). ²H is incorporated into fatty acids via exchange between body water and the hydrogens of acetyl-CoA, malonyl-CoA, and NADPH. Previous studies concluded that these exchanges are incomplete; therefore, fatty acid ²H enrichment requires correcting. In mice, we measured the ²H enrichment of fatty acid positions 2 and 3 and methyl hydrogens from [U-²H₇]glucose to determine ²H transfer from glucose to fatty acid via malonyl-CoA, NADPH, and acetyl-CoA, respectively. Positional fatty acid ²H enrichments were compared with ¹³C enrichment of the same sites from an equivalent amount of [U-¹³C₆]glucose provided alongside the [U-²H₇]glucose tracer. Transfer of glucose ²H to fatty acid position 2 and methyl sites was low (²H enrichment of 0.06 ± 0.01 and 0.14 ± 0.01 relative to ¹³C) indicating extensive exchange at both malonyl- and acetyl-CoA, respectively. Transfer of glucose ²H into fatty acid position 3 was more extensive (0.46 ± 0.04 relative to ¹³C, P < 10⁻⁵ vs. position 2), indicating a more limited exchange of those glucose hydrogens that were transferred via NADPH. However, mice provided with [U-¹³C₆]glucose and ²H₂O had equivalent ²H enrichments of fatty acid positions 2 and 3, suggesting that in this setting, NADPH and body water ²H had exchanged extensively. This is explained by contributions of substrates other than exogenous glucose to DNL coupled with their extensive ²H enrichment from ²H₂O prior to DNL. Under such conditions, ²H enrichment of fatty acids from ²H₂O does not need correction.

There is currently high interest in the measurement of de novo lipogenesis (DNL) to better understand its role in the dyslipidemia of diseases such as type 2 diabetes and fatty liver disease (1–3). Fractional DNL rates can be measured from incorporation of deuterated water (²H₂O) into fatty acids (4, 5), an inexpensive and simple method that can be applied to humans, animal models, and cell cultures. ²H enrichment of fatty acids from ²H₂O is conventionally measured by MS, where all the fatty acid chain hydrogens are considered as a single traceable entity (4). While this provides amplification of the m+1 signal arising from ²H incorporation, it does not resolve the carbon-bound fatty acid hydrogens according to their metabolic sources (see Fig. 1). It was reported that in rats, about 30% of plasma triglyceride palmitate hydrogens had not exchanged with ²H body water (6, 7), while for palmitate derived from cultured cells, this nonexchanged fraction was even higher (8). Therefore, for DNL measurement, the ²H enrichment of fatty acids measured by MS needs to be corrected by a predetermined factor related to the number of deuterium atoms that were incorporated per molecule of fatty acid, referred to as N (6). Stoichiometric ²H enrichment of fatty acids from ²H₂O is conditional on full exchange between the hydrogens of water and those of the acetyl-CoA methyls, the malonyl-CoA methylenes, and the reducing hydrogen of NADPH. A less than theoretical ²H enrichment of the fatty acids implies that hydrogen exchange is incomplete, but to what degree this occurs for each of the metabolic precursors is not known. To address this, we provided mice with [U-²H₇]glucose and performed ²H NMR analysis of liver triglyceride (5, 9, 10) to determine the extent to which the ²H were transferred into positions 2 and 3 and the terminal methyl hydrogens of hepatic fatty acids: each position reflecting ²H transfer from glucose via malonyl-CoA, NADPH, and acetyl-CoA, respectively. Our data indicate that there was a substantial transfer of glucose hydrogen to newly synthesized fatty acids via NADPH, corresponding to a limited exchange with water hydrogen, but relatively low transfer via malonyl-CoA and acetyl-CoA intermediates, indicating extensive hydrogen exchange at these loci of the DNL pathway.

Fig. 1.

Sources of hydrogens for the synthesis of palmitoyl-CoA from glucose and body water via fatty acyl-CoA synthase, with the butyryl acyl carrier protein (Butyryl-ACP) intermediate also shown. When glucose is metabolized via glycolysis, hydrogens 1 and 6 (shown in blue) are transferred to pyruvate and then to acetyl-CoA and malonyl-CoA (represented by blue italic). During the first cycle of fatty acid synthesis, acetyl-CoA and malonyl-CoA hydrogens are incorporated into the terminal methyl group and one of the position 2 hydrogens of butyryl-ACP, respectively. In the palmitoyl-CoA product, this corresponds to the terminal methyl hydrogens and one of the methylene hydrogens attached to each even-numbered carbon. When glucose is metabolized via the pentose phosphate pathway (PPP), hydrogens 1 and 3 (shown in red) are transferred to NADPH. During the first cycle of fatty acid synthesis, these are incorporated into both position 3 hydrogens of butyryl-ACP and subsequently into both methylene hydrogens attached to each odd-numbered carbon of palmitoyl-CoA. The remaining hydrogen in position 2 of butyryl-ACP is derived from body water (shown in gray) and is incorporated into one of the methylene hydrogens attached to each even-numbered carbon of palmitoyl-CoA. The color scheme indicates possible labeling sites only and is not meant to represent the stoichiometry of hydrogen transfer.

METHODS

Materials

[U-²H₇]- and [U-¹³C₆]glucose at 98–99% enrichment were obtained from Sigma-Aldrich. [3-²H]glucose at 98% enrichment was obtained from Omicron Biochemicals, South Bend, IN.

Animal studies

Animal studies were approved by the University of Coimbra Ethics Committee on Animal Studies (ORBEA) and the Portuguese National Authority for Animal Health (DGAV), approval code 0421/000/000/2013. Adult male C57BL/6J mice were obtained from Charles River Laboratories, Barcelona, Spain, and housed at the University of Coimbra Faculty of Medicine Bioterium, where they were maintained with a 12 h light/12 h dark cycle. Upon delivery to the Bioterium, mice were provided a 2 week interval for acclimation, with free access to water and standard chow. Following acclimation, the drinking water was supplemented with glucose and fructose (15.0 g of each sugar to 100 g of water). In the first set of studies, the glucose was enriched to 20% each with [U-²H₇]glucose and [U-¹³C₆]glucose. In a second set of studies, glucose was enriched to 16% each with [3-²H]glucose and [U-¹³C₆]glucose. In the third set of experiments, mice were administered an intraperitoneal injection of 99.9% ²H₂O containing 9 mg NaCl per milliliter at a dose of 4 g/100 g body mass at the start of the dark period. For these animals, the drinking water was enriched to 5% with ²H₂O and the glucose was enriched to 50% with [U-¹³C₆]glucose but not with ²H. Animals were allowed to feed and drink ad libitum during the entire 12 h dark period and were euthanized by cervical dislocation the following morning. Livers and adipose tissue depots were freeze-clamped and stored at −80°C until further processing for triglyceride extraction and purification.

Triglyceride extraction and purification

Liver triglyceride was extracted and purified as previously described (10). Briefly, livers were powdered under liquid nitrogen and then rapidly mixed with HPLC-grade methanol (4.6 ml/g) followed by methyl-tert-butyl ether (MTBE) (15.4 ml/g). The mixture was placed in a shaker for 1 h at room temperature and then centrifuged at 13,000 g for 10 min. The liquid fraction was collected and phase separation was induced by adding 4 ml of distilled water to the liquid fraction and letting it rest at room temperature for 10 min. The liquid was then centrifuged for 10 min at 1,000 g. The upper organic phase containing the lipids was carefully separated and dried under nitrogen gas in a dark glass vial. Triglycerides from the dried lipid extracts were purified with a solid phase extraction (SPE) process. Discovery DSC-Si SPE cartridges (2 g/12 ml) were washed with 8 ml of hexane/MTBE (96/4; v/v) followed by 24 ml of hexane. The dried lipids were resuspended in 800 μl of hexane/MTBE (200/3; v/v) and added to the column after washing. The lipid vials were washed with a further 500 μl solvent to quantitatively transfer the lipids to the column. Triglycerides were eluted with 32 ml of hexane/MTBE (96/4; v/v), collected in 4 ml fractions. Fractions containing triglycerides were identified by TLC. A few microliters of the eluted fractions were spotted on the TLC plate alongside triglyceride standards and the plate was developed with petroleum ether/diethyl ether/acetic acid (7.0/1.0/0.1; v/v/v). After drying, lipid spots were visualized by iodine vapor. The triglyceride-containing fractions were pooled and dried under nitrogen gas and stored at −20°C until ready for NMR analysis.

NMR analysis

Purified triglycerides were dissolved in ∼0.5 ml CHCl₃. To these, 25 μl of a pyrazine standard enriched to 1% with pyrazine-d₄ and dissolved in CHCl₃ (0.07 g pyrazine/1 g CHCl₃), and 50 μl of C₆F₆ were added. ¹H and ²H NMR spectra were acquired with an 11.7 T Bruker Avance III HD system using a dedicated 5 mm ²H-probe with ¹⁹F lock and ¹H-decoupling coil as previously described. ¹H spectra at 500.1 MHz were acquired with a 90° pulse, 10 kHz spectral width, 3 s acquisition time, and 5 s pulse delay. Sixteen free-induction decays (fid) were collected for each spectrum. ²H NMR spectra at 76.7 MHz were obtained with a 90° pulse, a 1,230 Hz sweep width, an acquisition time of 0.37 s, and a pulse delay of 0.1 s. Between 10,000 and 20,000 fid were acquired for each spectrum. Correction factors were applied to all ²H triglyceride signals to adjust their intensities relative to the partially saturated ²H pyrazine standard signal. These were obtained as mean values from a set of seven liver triglyceride samples obtained from mice administered with ²H₂O. For each sample, a spectrum was acquired with the described parameters and immediately followed by a spectrum acquired under the same parameters with the exception of the acquisition time and pulse delay, which were set to 1 s and 8 s, respectively. The correction factors for the ²H signals in fatty acid position 2, position 3, and methyl hydrogens were 0.51, 0.52, and 0.88, respectively. For ¹³C isotopomer analysis by ¹³C NMR, dried triglyceride samples were dissolved in 0.2 ml 99.96% enriched CDCl₃ (Sigma-Aldrich) and placed in 3 mm NMR tubes. ¹³C NMR spectra were acquired at 150.8 MHz with an Agilent V600 spectrometer equipped with a 3 mm broadband probe. Spectra were acquired with a 70° pulse, an acquisition time of 2.5 s, and a 0.5 s pulse delay. For each spectrum, 2,000–4,000 fid were collected.

Quantification of triglyceride positional ²H and ¹³C enrichments

Positional ²H enrichments of triglyceride fatty acids were quantified by analysis of ¹H and ²H NMR triglyceride spectra as previously described (5, 10). From the methyl and carboxyl ¹³C NMR resonances, positional ¹³C enrichments of fatty acids were estimated from the ratio of ¹³C-¹³C-spin-coupled doublet signals (representing positional isotopomers derived from [1,2-¹³C₂]acetyl-CoA) to the singlet signal, representing the 1.11% natural abundance ¹³C (see supplemental Fig. S1). From the methyl singlet and doublet ¹³C NMR signals, the ¹³C enrichment of fatty acids from [1,2-¹³C₂]acetyl-CoA in the methyl position was calculated as follows:

$$\begin{array}{l}{\text{Excess}}^{\text{13}}\text{C enrichment of fatty acid methyls}(\%)\\ =\mathrm{Methyl}\mathrm{D/Methyl\; S}\times 1.11\end{array}$$

where Methyl D and Methyl S are the doublet and singlet components, respectively, of the ¹³C-signal of the fatty acid terminal methyls and 1.11 represents the background ¹³C enrichment (%).

Assuming that fatty acid enrichment from [1,2-¹³C₂]acetyl-CoA via elongation was limited to position 1 and 2 carbons, enrichment of carbon 3 was assumed to be equivalent to that of the terminal methyl carbon.

From analysis of the fatty acid singlet and doublet carboxyl ¹³C-signals, excess enrichment of the position 1 of fatty acids from [1,2-¹³C₂]acetyl-CoA was estimated as follows:

$${\text{Position 1 Excess}}^{\text{13}}\text{C enrichment}(\%)=\mathrm{\Sigma }\mathrm{D}/\mathrm{\Sigma }\mathrm{S}\mathrm{\times }1.11$$

where ΣD and ΣS are the summed doublet and singlet components, respectively, of the ¹³C-carboxyl resonances and 1.11 represents the background ¹³C enrichment (%). Excess enrichment of the fatty acid position 2 carbon was assumed to be equal to that of position 1.

For determining the fractional rate of [U-²H₇]glucose transfer into acetyl- and malonyl-CoA relative to that of [U-¹³C₆]glucose, the ²H/¹³C enrichment ratios of the terminal methyl and carbon 2 positions were divided by 1.5 to account for the fact that there are three ²H for every two ¹³C in the initial glucose mixture. For estimating the fractional rate of [U-²H₇]glucose transfer into NADPH relative to the lipogenic utilization of [U-¹³C₆]glucose, we assumed that two ²H were transferred per glucose molecule corresponding to one ²H per acetyl-CoA derived from glucose.

Statistics

All results are presented as mean ± standard error and comparisons were made by an unpaired two-tailed Student’s t-test performed with Microsoft Excel.

RESULTS

Mice that were provided with the mix of [U-²H₇]- and [U-¹³C₆]glucose yielded liver triglycerides that were enriched in both ¹³C and ²H (Table 1). Both ²H and ¹³C NMR spectra featured composite signals of the inner fatty acid methylenes and well-resolved resonances representing both ends of the fatty acid chain (Fig. 2A, supplemental Fig. S1). Enrichment from [U-¹³C₆]glucose was relatively uniform between terminal and proximal carbons indicating that DNL was the main route for incorporation of glucose carbons into fatty acids with elongation playing a relatively insignificant role.² In comparison, enrichment of fatty acid sites from [U-²H₇]glucose was highly heterogeneous, as seen by the very different intensities of positions 2 and 3 and the terminal methyl resonances (Fig. 2A) and the excess ²H enrichments estimated from these signals (Table 1). Normalizing the fatty acid enrichment from [U-²H₇]glucose to that of [U-¹³C₆]glucose provides a measure of the fractional retention of the [U-²H₇]glucose ²H atoms in a given position (Table 2). These data revealed that for those fatty acids that were synthesized from exogenous glucose, far more ²H was transferred into the position 3 hydrogens compared with either position 2 or the terminal methyl hydrogens. This indicates a greater exchange of ²H and water hydrogens during the conversion of [U-²H₇]glucose to acetyl-CoA and malonyl-CoA compared with hydrogen transfer via pentose phosphate pathway (PPP) oxidation and NADPH. Only one of the fatty acid position 2 hydrogens is derived from malonyl-CoA, the other originates from body water. Therefore, based on the observed fatty acid position 2 ²H/¹³C enrichment ratio of 6%, we can infer that the ²H/¹³C enrichment ratio of the malonyl-CoA precursor was 12%. This is similar to the 14% estimated for the initial acetyl-CoA pool recruited by fatty acyl-CoA synthase (FAS).

TABLE 1.

Fatty acid ¹³C and ²H enrichments

Experiment	Isotope	Fatty Acid Positional Enrichment
		Methyl	Position 2	Position 3
20% [U-13C6]glucose + 20% [U-2H7]glucose	13C	0.99 ± 0.06	1.04 ± 0.07a	0.99 ± 0.06b
	2H	0.23 ± 0.02	0.09 ± 0.01	0.49 ± 0.05
16% [U-13C6]glucose + 16% [3-2H]glucose	13C	0.77 ± 0.02	0.83 ± 0.03	0.77 ± 0.02
	2H	0.01 ± 0.01	n.d.	0.17 ± 0.02

Fatty acid ¹³C and ²H enrichments from a group of eleven mice provided with glucose enriched to 20% with [U-¹³C₆]- and [U-²H₇]glucose and a second group of five mice provided with glucose enriched to 16% with [U-¹³C₆]- and [3-²H]glucose. n.d., not detected (signal-to-noise ratio <3:1).

^a13C-excess enrichment of fatty acid position 1 carboxyl carbons was assumed to be equivalent to that of position 2.

^b13C-excess enrichment of fatty acid position 3 was assumed to be equivalent to that of the methyl carbon.

Fig. 2.

A: A representative ²H NMR spectrum of purified liver triglyceride acquired from a mouse provided with glucose enriched with equimolar amounts of [U-²H₇]glucose and [U-¹³C₆]glucose. B: A ²H NMR spectrum of triglyceride acquired from a mouse provided with glucose enriched with equimolar amounts of [3-²H]glucose and [U-¹³C₆]glucose. C: A ²H NMR spectrum of triglyceride acquired from a mouse provided with ²H₂O and [U-¹³C₆]glucose. For the spectra, the signals corresponding to ²H in the fatty acid terminal methyl (TM) as well as in fatty acid position 2 (H2) and position 3 (H3) are indicated in bold. Also indicated are signals from ²H in fatty acid olefinic sites and glyceryl position 2 (OL-G2), ²H from glycerol sn1,3 positions (G1,3), ²H bound to the allylic carbon of monounsaturated fatty acids (AL), and ²H bound to the inner methylene carbons (ME).

TABLE 2.

²H/¹³C enrichment ratios for fatty acid position 2, position 3, and terminal methyl

	Position 2	Position 3	Methyl
[U-2H7]/[U-13C6]glucose	0.06 ± 0.01a	0.46 ± 0.04b	0.14 ± 0.01
[3-2H]/[U-13C6]glucose	Not determined	0.22 ± 0.02	Not determined

¹³C/²H enrichment ratios for fatty acid positions 2 and 3 and methyls estimated from the group of eleven mice provided with [U-²H₇]- and [U-¹³C₆]glucose and that of fatty position 3 from the group of five mice provided with [3-²H]- and [U-¹³C₆]glucose.

^aP < 0.005 compared with methyl.

^bP < 10⁻⁵ compared with methyl.

We performed an additional a set of studies where [U-¹³C₆]glucose was accompanied by [3-²H]glucose, where incorporation of ²H into fatty acids from this precursor occurs exclusively via NADPH. Taking into account the lower ²H/¹³C precursor enrichment (16% versus 20% for the [U-²H₇]glucose/[U-¹³C₆]glucose experiments), the fatty acid ¹³C enrichment distributions were consistent with those of the [U-²H₇]glucose/[U-¹³C₆]glucose study. This indicates that for both studies, exogenous glucose had been utilized to the same extent for DNL. Figure 2B shows a representative ²H NMR spectrum of liver triglyceride from a mouse provided with the [3-²H]glucose/[U-¹³C₆]glucose mixture. In accordance with the predicted metabolic fate of the ²H label, fatty acids were enriched in position 3, while enrichment of position 2 was not detectable and the terminal methyl had a vestigial ²H signal. Given the substantial transfer of glucose hydrogens into the position 3 of fatty acid relative to position 2, it might be expected that for mice administered with ²H₂O, enrichment of fatty acid position 3 would also be less than that of position 2. However, as shown by Fig. 2C, the intensities of position 2 and 3 signals were similar, and there was no significant difference between the ²H enrichments quantified for each site (1.79 ± 0.19% and 1.81 ± 0.19% for positions 2 and 3, respectively).

DISCUSSION

Since the pioneering studies of Beylot and coworkers (4, 6), ²H₂O has been extensively used as a tracer for quantifying fractional DNL rates. From the beginning, it was understood that ²H incorporation into fatty acids might be limited by a) incomplete exchange of body water and metabolite hydrogens and/or b) discrimination against the incorporation of ²H into these precursors or intermediates of FAS due to kinetic isotope effects. In the present study, by quantifying fatty acid positional enrichments corresponding to the transfer of ²H from [U-²H₇]glucose via acetyl-CoA, malonyl-CoA, and NADPH, the extent of glucose hydrogen exchange during its conversion to fatty acids via each intermediate was evaluated.

Exchange of glucose ²H during the formation of acetyl-CoA and malonyl-CoA

Glycolysis of [U-²H₇]glucose yields one pyruvate with a single ²H and a second pyruvate with two ²H in the methyl position. Therefore, there are 1.5 equivalents of ²H for each pyruvate. If this pyruvate is recruited for lipogenesis via acetyl-CoA, citrate, and cytosolic acetyl-CoA, one of the methyl hydrogens undergoes obligatory exchange with water, resulting in one equivalent of ²H per cytosolic acetyl-CoA. While pyruvate methyl hydrogens can undergo extensive exchange with water via alanine aminotransferase (11), there is evidence that not all intracellular pyruvate participates in this process. For example, in perfused rat hearts supplied with [3-²H₃,1-¹³C]pyruvate, Funk et al. (12) observed a significantly higher retention of ²H in lactate compared with alanine at 3 and 6 min of perfusion. Moreover, pyruvate molecules with ²H in the methyl position may be metabolized at different rates to those without ²H. In the same setting, [U-²H₇,U-¹³C₆]glucose was less efficiently incorporated into the Krebs cycle compared with [U-¹³C₆]glucose (13), suggesting that the presence of ²H in the methyl hydrogens of acetyl-CoA was attenuating its conversion to citrate via citrate synthase. Thus, while there is an efficient exchange mechanism for incorporating ²H from water into the methyl sites of pyruvate, there may be intracellular pyruvate pools that do not experience this exchange. Moreover, pyruvate molecules that do become enriched with ²H may be transformed into citrate and cytosolic acetyl-CoA at slower rates in comparison to their nondeuterated counterparts. In our study, we found that for the initial FAS-bound acetyl-CoA that was derived from exogenous [U-²H₇]glucose, 86% of ²H had been exchanged for ¹H. From analysis of cytosolic acetyl-CoA ²H enrichment from ²H₂O via chemical biopsy, Duarte et al. (5) concluded that exchange between the methyl hydrogens of this precursor and body water was ∼100%. In our experimental setting, exogenous glucose contributes a minor fraction (∼10%) of hepatic DNL (14). In the presence of ²H₂O, it is probable that endogenous DNL sources, including acetyl-CoA derived from fructose, lactate, and microbial acetate, will be already enriched to a significant extent with ²H prior to being recruited for DNL. For these reasons, we believe that the overall hydrogen exchange fraction of the lipogenic acetyl-CoA pool is closer to 100% than to 86%. Thus, in our experimental setting, animals provided with ²H₂O would be expected to have near-equivalent ²H enrichments of acetyl-CoA methyl hydrogens and body water. For [U-²H₇]glucose that was metabolized to fatty acid via malonyl-CoA, exchange of ²H with ¹H was similar to that observed for acetyl-CoA. Therefore, enrichment of malonyl-CoA from ²H₂O would also be expected to approach that of body water.

²H transfer from glucose to fatty acids via NADPH

²H that were metabolized from glucose to fatty acids via NADPH were more highly retained in comparison to those transferred via acetyl-CoA and malonyl-CoA. During PPP oxidation, the first and third hydrogens of glucose are transferred to NADPH, hence [U-²H₇]glucose contributes two ²H, while [3-²H]glucose only contributes one. In accordance, transfer of ²H to fatty acid position 3 from [U-²H₇]glucose was found to be approximately twice that from [3-²H]glucose (46% vs. 22%, see Table 2). This indicates that under our study conditions, there was no additional loss of ²H from glucose position 1 compared with position 3, for example, by exchange of glucose-6-phosphate with fructose-6-phosphate and mannose-6-phosphate (15). Direct studies of intracellular NADPH enrichment from ²H₂O report a greater degree of exchange between the reducing hydrogen and body water than might be expected based on our observations (16). One explanation for this is that the PPP is not the sole source of intracellular NADPH, as illustrated by Fig. 3. In mitochondria, NADPH can be generated from NADH by nicotinamide nucleotide transhydrogenase. In mitochondria as well as cytosol, NADPH can also be generated via NADP⁺-malic enzyme and NADP⁺-isocitrate dehydrogenase. In all of these cases, the likelihood that the hydrogen that was transferred to NADP⁺ had previously exchanged with body water is high. For NADPH formed via the NADP⁺-malic enzyme, the reducing hydrogen originates from hydrogen 2 of malate, which in turn originated from water during the hydration of fumarate. If the malate is metabolized via the Krebs cycle to citrate and isocitrate, this hydrogen is transferred to NADPH via NADP⁺-isocitrate dehydrogenase. The bulk of mitochondrial NADH hydrogens are also derived from Krebs cycle intermediates, such as malate, whose hydrogens are highly exchanged with those of water. In addition, NADPH and water hydrogens may also be exchanged via NADP⁺-linked redox enzymes (16, 17). Another consideration, and an important caveat of our assessment of hydrogen exchange at the level of NADPH, is that fatty acid ²H enrichment from [U-²H₇]glucose via this pathway was quantified relative to [U-¹³C₆]glucose conversion to fatty acid via malonyl-CoA. If the rate of glucose oxidation by the PPP was different to that of its conversion to fatty acid via malonyl-CoA, then the fractional rate of ²H/¹H exchange at the level of NADPH would also differ from the ²H/¹³C ratio of fatty acid position 3.³ However, this does not alter the conclusion that the relative amounts of ²H transferred from glucose to fatty acid via NADPH were significantly higher than those transferred via acetyl-CoA and malonyl-CoA intermediates.

Fig. 3.

Schematic of intracellular NADPH hydrogen sources. On the left-hand side are the pathways that are involved in the transfer of water hydrogen to NADPH (depicted in gray). These include cycling between NADPH and other redox cofactors such as FAD whose hydrogens are exchanged with those of water; generation of NADPH from malate and isocitrate via NADP⁺-malic enzyme and NADP⁺-isocitrate dehydrogenase, respectively; and exchange of NADH and NADPH via transhydrogenase. On the right-hand side are the two principal pathways that transfer hydrogens to NADPH from nutrient substrates, namely, glucose and serine.

Our observation of extensive ²H transfer from glucose to fatty acid synthesis via NADPH suggests that this process is tightly coupled to the extent that NADPH generated by the PPP does not fully mix with other intracellular NADPH pools. This apparent compartmentation of PPP and lipogenic NADPH metabolism has also been reported in other laboratories using different isotopic approaches and models. In mice with fatty liver, hepatic PPP fluxes measured by a ¹³C-isotopomer approach were found to be highly correlated with lipogenesis but were not associated with antioxidant activity (18). This suggests that in the postprandial liver, NADPH derived from the PPP is prioritized for DNL rather than being utilized for antioxidant defense, such as the reduction of oxidized glutathione. In the H1299 tumor cell line supplied with various ²H-enriched substrates, Lewis et al. (19) also demonstrated a substantial transfer of ²H label from [3-²H]glucose to fatty acids, while fatty acid ²H enrichment via NADPH from [2,3,3-²H₃]serine was much more limited.

Fatty acid enrichment from ²H₂O

We observed extensive exchange of exogenous glucose hydrogens with water during its conversion to fatty acids via both acetyl- and malonyl-CoA. In contrast, glucose hydrogens that were transferred to fatty acids via NADPH experienced more limited exchange. Based on these observations, it would be anticipated that enrichment of the fatty acid position 2 hydrogens from ²H₂O (derived from water and malonyl-CoA) would be equivalent to those of body water, while those of position 3 (derived via NADPH) would be significantly less enriched. Instead, we observed that positions 2 and 3 were enriched to the same level. As mentioned previously, exogenous glucose is a minor contributor of acetyl-CoA to DNL in our experimental setting (although it may have a more substantial contribution to the NADPH equivalents for DNL). Moreover, following ²H₂O administration under both fed and fasted conditions, hepatic glucose-6-phosphate becomes highly enriched with ²H in all positions due to extensive cycling between glucose-6-phosphate and gluconeogenic precursors (20, 21). Under these conditions, NADPH derived from PPP oxidation of glucose-6-phosphate will also become highly enriched with ²H and this will be transferred to the respective fatty acid positions.

Implications for estimating the number of deuterium atoms incorporated per fatty acid

Under our study conditions, the number of deuterium atoms incorporated per fatty acid molecule (N) appears to approach the theoretical value (i.e., 31 for palmitate). Therefore, in this instance, no correction needs to be applied to the observed fatty acid ²H enrichment from ²H₂O. This differs from previous studies in rats, where N was estimated to be 21 and 22 out of 31 (68% and 71% of the theoretical value) by mass isotopomer distribution analysis for plasma and liver triglyceride palmitate, respectively (6, 7). To our knowledge, N has not been determined in the mouse by mass isotopomer distribution analysis, but given its smaller size and higher basal metabolic rate, it might be expected to be higher compared with the rat. The mice in our study also ingested significant amounts of fructose, a sugar that is rapidly metabolized via triose phosphate intermediates to glucose-6-phosphate, lactate, and acetyl-CoA, thereby further promoting ²H enrichment of these metabolites from ²H₂O. It is possible that for mice fed a standard chow diet featuring maltose as the main carbohydrate component, whose digestion yields unlabeled glucose, there could be less complete ²H incorporation from ²H₂O into fatty acids, particularly via NADPH. In cultured cells, N for palmitate was estimated to be 17 (8). This is 55% of the theoretical value and is also substantially lower than that determined for in vivo rat studies. Our data for mice indicate that if glucose was the sole source of DNL in vivo, the value of N would be 24 based on ²H transfer from exogenous [U-²H₇]glucose to fatty acids via acetyl-CoA, malonyl-CoA, and NADPH.⁴ Because there is more opportunity for the loss of glucose ²H by Cori cycling and other inter-organ transfer of glucose metabolites in vivo compared with in vitro, we anticipate that there would be a greater degree of transfer of ²H from glucose to fatty acids in vitro, corresponding to a lower N value.

In conclusion, N may vary considerably according to the type of organism as well as the substrates that were utilized for DNL. Our study indicates that transfer of hydrogen from unlabeled glucose to fatty acids via NADPH is an important factor in the nonstoichiometric incorporation of ²H into fatty acids during DNL. For the measurement of DNL with ²H₂O, the ratio of ²H enrichment in fatty acid positions 2 and 3 as measured by ²H NMR provides a convenient internal check for assessing the extent to which this might have occurred.