To determine the source(s) of blood and very low density lipoprotein (VLDL)-triglyceride glycerol during fasting, four men ingested 2H2O from 14 to 20 h into a 60-h fast to achieve ∼0.5% body water enrichment. At 60 h of fasting, glycerol flux was measured using [2-14C]glycerol. Blood was taken for measurement of2H enrichment at carbon 6 of glucose and at carbon 3 of free glycerol and VLDL-triglyceride glycerol. 2H enrichment of the 2 hydrogens bound to carbon 3 of VLDL-triglyceride glycerol was 105 ± 2% of the 2H enrichment of the 2 hydrogens bound to carbon 6 of glucose, indicating isotopic equilibrium between hepatic glyceraldehyde 3-P and glycerol 3-P. The2H enrichment of the 2 hydrogens bound to carbon 3 of free glycerol was 17 ± 3% of VLDL-triglyceride glycerol, indicating that a significant percentage of free glycerol in blood originated from the hydrolysis of circulating VLDL-triglyceride or a pool of glycerol with similar 2H enrichment. Glycerol flux was 6.3 ± 1.1 μmol · kg−1 · min−1. Glycerol appearing from nonadipose tissue sources was then ∼1.1 μmol · kg−1 · min−1. Seven other subjects were fasted for 12, 42, and 60 h. A small percentage of glycerol in the circulation after 12 h of fasting was enriched with 2H. The enrichment of the 2 hydrogens bound to carbon 3 of free glycerol in the longer periods of fasting was ∼16% of the enrichment of the 2 hydrogens bound to carbon 6 of glucose. Therefore, as much as 15–20% of systemic glycerol turnover during fasting is not from lipolysis of adipose tissue triglyceride.
- very low density lipoprotein
- deuterated water
measurements of systemic rates of appearance (Ra) of glycerol by isotope tracer techniques have long served to estimate overall lipolysis in humans (5). In those estimates, glycerol is assumed to arise predominantly, if not exclusively, from adipose tissue lipolysis. Lipolysis in tissues, such as muscle, and/or release of glycerol via the hydrolysis of circulating very low density lipoprotein (VLDL) could be alternate sources of the glycerol. The possible contribution of glycolysis to systemic glycerol appearance has been examined. DeFreitas and DePocas (6) ascribed 17% of glycerol's Ra in 24-h-fasted rats to the formation of glycerol from glycolytic intermediates, measured using [U-14C]glucose. However, Nurjhan et al. (27) found that, in dogs fasted for 18 h, only 1.6 ± 0.6% of plasma glycerol was from plasma glucose, again measured using [U-14C]glucose. The potential contribution of nonadipose tissue sources, such as VLDL-triglyceride hydrolysis to systemic glycerol appearance in humans, is not known but could be a significant source of glycerol, because VLDL-triglyceride production rates ranging from ∼0.35 to 1.0 μmol·kg−1·min−1 have been reported (11, 31). The release of circulating triglyceride glycerol into the circulation could then supply ∼20% of reported systemic glycerol appearance rates (26).
The potential contribution of circulating triglyceride hydrolysis to free glycerol production is of special interest during fasting. There is extensive evidence for more fatty acid being released by lipolysis than utilized during fasting and also in the postabsorptive state (16, 17). The excess fatty acid must be reesterifed, resulting in a “triglyceride cycle.” Fatty acids esterified in liver are released as VLDL-triglyceride. Those triglycerides on hydrolysis could release glycerol into the circulation. The released fatty acids could then be either oxidized or redeposited in triglyceride.
Glycerol 3-P used in esterification of free fatty acids (FFA) to triglyceride in liver can arise by phosphorylation of glycerol taken up from the systemic circulation, as well as from dihydroxyacetone 3-P (DHAP). DHAP is in extensive equilibrium with glyceraldehyde 3-P (GAP) (22), and GAP is the precursor of carbons 4, 5, and 6 of glucose formed via gluconeogenesis. When we administered 2H2O to fasted humans, the 2H enrichment of the two hydrogens bound to carbon 6 of blood glucose formed via gluconeogenesis was ∼80% of that in body water (19). If glycerol 3-Pextensively equilibrates with GAP, the 2H enrichment of the 2 hydrogens bound to carbon 3 of the glycerol 3-P, used in the formation of VLDL, should be about the same as that of the hydrogens bound to carbon 6 of the glucose.
On the basis of that reasoning, this study was undertaken to determine the extent to which intrahepatic 3 carbon compounds serve as a common precursor for gluconeogenesis and glycerogenesis.2H2O was administered to subjects fasted for 60 h. The enrichment in the hydrogens bound to carbon 6 of glucose was compared with the enrichment in the hydrogens bound to carbon 3 of glycerol from VLDL. Also, the enrichment at carbon 3 of free glycerol in the circulation was determined in those subjects and other subjects fasted for 12, 42, or 60 h. The extent of those enrichments provides the measure of the nonadipose origin of free glycerol in the circulation. That is so because, in view of the size of triglyceride stores, it would be virtually impossible to enrich adipose tissue glycerol to a measurable extent during those periods of fasting.
MATERIALS AND METHODS
The study was performed at the Mayo Clinic in Rochester, MN, and the Karolinska Hospital in Stockholm, Sweden. The eight men and three women studied were normal healthy volunteers. The study was approved by the human investigation committees of the institutions, and informed consent was obtained from the volunteers.
Four men, of ages 21–32 yr and body mass indexes (BMI) 22.0–27.3, were admitted in the evening to the Mayo General Clinical Research Center (GCRC). Before the start of the study, their body fat was measured using dual-energy X-ray absorptiometry (13). They ingested dinner at 6 PM and then fasted. Fourteen hours after the meal, they ingested 5 ml/kg body water of 99% enriched 2H2O, divided into four equal doses (1.25 ml/kg body water) given at 2-h intervals over 6 h. Body water was estimated at 72% of fat-free mass. The dose of2H2O was expected to increase the enrichment of the hydrogens of body water to ∼0.5%. To maintain that enrichment, the subjects drank only water (or other noncaloric beverages) 0.5% enriched with 2H2O. After subjects fasted for 60 h, a catheter was placed in a forearm vein, and a 3-h infusion of [2-14C]glycerol (0.3 μCi/min) (Amersham, Arlington Heights, IL) was begun. Oxygen consumption (V˙o 2) and CO2 production (V˙co 2) were measured by indirect calorimetry with a DeltaTrac Metabolic Cart (Yorba Linda, CA), with the volunteers resting quietly in bed.
Blood, 250 ml, was collected from a radial artery over the last 20 min of the 3-h infusion. Enrichments were measured of the hydrogens bound to carbons 1 and 3 of free glycerol in the blood, and of glycerol from VLDL, as well as of the hydrogens bound to carbon 6 of the blood glucose. Enrichment in plasma water was determined as the measure of body water enrichment. Arterial blood was collected to provide a better measure than venous blood of the glycerol Ra. Concentrations of β-hydroxybutyrate and insulin in plasma were also determined.
Seven subjects were admitted to the Clinical Research Unit at the Karolinska Hospital in the evening. They ingested a dinner of 12–14 kcal/kg body wt, composed of 48% carbohydrate, 19% protein, and 33% fat. Then they fasted. Two of the subjects, men aged 21 and 29 yr with BMIs of 23.1 and 25.4, drank2H2O, 2.5 ml/kg body water at 3 and 7 h into the fast. Body water was estimated at 60% of body weight (20). Five hours later, i.e., after 12 h of fasting, 250 ml of blood were drawn from an antecubital arm vein for measurement of 2H enrichments at carbons 1 and 3 of blood glycerol and at carbon 6 of blood glucose. One milliliter of blood was collected for measurement in plasma of the enrichment in body water. Two other men, ages 22 and 26 yr with BMIs of 22.8 and 23.1, were treated in the same manner but fasted for 60 h, and 150 ml of blood were drawn. Three women, aged 31–39 yr with BMIs of 22.2–25.3, were treated in the same manner except that the 2H2O was ingested 5 and 9 h after the meal, body water was estimated at 50% of body weight, and 140 ml of blood were drawn after 42 h of fasting and only for measurements of enrichments in glycerol. For those three subjects, measurements of enrichments in blood glucose and in body water, determined in urine collected at 42 h, have previously been reported (20). During fasting, the subjects were allowed to drink water 0.5% enriched in 2H2O ad libitum.
VLDL particles were isolated using standard ultracentrifugation techniques (30) from 100 ml of blood plasma from each of the four subjects studied at the Mayo GCRC. Glucose and glycerol were isolated from the infranatant after protein precipitation by addition of Ba(OH)2 and ZnSO4 (33). The supernatant obtained on centrifugation, combined with an aqueous wash of the precipitate, was deionized by passage through a column of AG 1-X8 in the formate form over AG 50W-X8 in the hydrogen form (Bio-Rad, Hercules, CA) (18). The column was washed with water and the effluent evaporated to dryness. Separation and purification of the glycerol and glucose in the residue after the evaporation were by HPLC (27). The residue was applied to a Bio-Rad HPX-87P column with water at 80°C as solvent and a flow rate of 0.5 ml/min. Glucose eluted at 16–20 min and glycerol at 24–28 min. The glucose fraction was identified using a glucose oxidase assay (YSI 2300 Analyzer, Yellow Springs Instrument, Yellow Springs, OH) and the glycerol fraction by an ultraviolet enzymatic assay (8). The glucose fraction was rechromatographed using the HPX-87P system. To remove contaminating glucose (27), the glycerol fraction was subjected to two more applications by use of the HPX-87P system. Then the glycerol peak was chromatographed using an HPX-87C column with water at 85°C as solvent and a flow rate of 0.6 ml/min. On subjecting labeled glucose added to unlabeled glycerol to the procedure, <0.01% of the label added was in glycerol. The yield of glycerol was determined using the enzymatic assay. From the ∼1 mg of glycerol in 250 ml of blood collected after 12 of fasting and in 140–150 ml of blood collected after long-term fasting, the yield of the purified glycerol was ∼0.5 mg. Glycerol and glucose were isolated from the blood collected from the subjects in Sweden in the same manner as from the subjects at the Mayo Clinic, i.e., deproteinization with Ba(OH)2 and ZnSO4, deionization, and purification by multiple applications via the HPLC systems.
The glyceride of the VLDL particles was hydrolyzed by heating at 70°C for 1 h in 0.5 N alcoholic KOH (8). The cooled hydrolysate was neutralized with perchloric acid, and the potassium perchlorate precipitate was removed by centrifugation. The supernatant, after addition of water, was deionized on the ion exchange column and evaporated to dryness. Glycerol was isolated from the residue by use of the HPX-87P system.
Glycerol, 0.15 mg, from the VLDL hydrolysate, from the infranatant in the VLDL isolation, and from the blood, was oxidized with periodate to form formaldehyde from carbon 1 and from carbon 3 of the glycerol with their the two hydrogens (Fig. 1). The formaldehyde was collected by distillation. Ammonia was added to form hexamethylenetetramine (HMT) (19). Also, 0.3 mg of glycerol was incubated with ATP and glycerokinase (Sigma Chemical, St. Louis, MO) to yield glycerol 3-P (Fig. 1). The reaction mixture was applied to the AG 1X-8 column in the formate form, and the column was washed with water and 1 N and then 4 N formic acid. The glycerol 3-P eluted in the 4 N fraction. The fraction was evaporated to dryness and applied to the HPX-87C system. Glycerol 3-P eluted at 9–12 min (27). It was treated with periodate to form formaldehyde containing carbon 1 of the glycerol with its two hydrogens. An HMT was also made from that formaldehyde. About 0.5 mg of glucose from infranatant and blood was oxidized with periodate to yield carbon 6 with its two hydrogens as formaldehyde (19). That formaldehyde was also converted to HMT.
2H enrichments (mole percent excess) in the HMTs were determined by mass spectrometry. Electron impact ionization was used, and ions of mass-to-charge ratios, or m/z, of 140 and 141, corresponding to unenriched and enriched (m + 1) molecular ions, were monitored (19). Only an insignificant number of doubly labeled (m + 2) molecules of formaldehyde are formed at a body water enrichment of 0.5%, i.e., 1/200 × 1/200 = 1 in 40,000 molecules. At each assay, HMTs of 0.0625, 0.125, 0.25, 0.50, and 1.00% enrichments, prepared from formaldehyde from [1-2H]sorbitol of known enrichment (3, 20), provided a standard curve. Enrichments reported at carbon 1 and carbon 3 of glycerol and at carbon 6 of glucose in Tables 1-3 in results are then the sum of the enrichments of the two hydrogens bound to those carbons.
Blood glucose concentrations were measured with a Beckman Glucose Analyzer (Beckman Instruments, Fullerton, CA). Hydroxybutyrate concentrations in plasma were measured as described by Cahill et al. (2). Plasma insulin concentrations were measured using a chemiluminescence method with the Access Ultrasensitive Immunoenzymatic assay system (Beckman, Chaska, MN). Plasma glycerol specific activity was determined as described by Judd et al. (15). Glycerol was isolated using HPLC as its tribenzoyl derivative.
Systemic glycerol flux was calculated from the mean steady-state glycerol specific activity (dpm/μmol) and the tracer infusion rate (dpm/min) (15). Fatty acid oxidation rates were calculated using each individual's V˙o 2 andV˙co 2. Nitrogen losses were estimated using values from previous studies (14). Because a glycerol molecule on periodate oxidation yields two formaldehyde molecules, the enrichment reported at C1 + C3 of glycerol in Tables 1-3 is the average of the enrichments of the hydrogens at carbons 1 and 3. Therefore, the enrichment at carbon 3 was calculated by multiplying the average of the enrichments at carbon 1 and carbon 3 by two and subtracting the enrichment at carbon 1. Enrichment in body water was determined using an isotope ratio mass spectrometer.
All results are expressed as means ± SE.
The characteristics of the volunteers fasted for 60 h and studied in the Mayo Clinic are recorded in Table1. As expected, the respiratory quotient and plasma concentrations of glucose and insulin were below those usually seen after an overnight fast, whereas ketone body concentrations and fatty acid oxidation rates were increased.
Systemic glycerol turnover (glycerol Ra), measured using the [2-14C]glycerol tracer, was 522 ± 79 μmol/min in the Mayo subjects. Table 2 records the enrichments of the hydrogens in VLDL-glycerol and in glycerol and glucose isolated from the infranatant obtained in the isolation of the VLDL from the plasma of the four subjects. The enrichment in the two hydrogens bound to carbon 1 of the VLDL-glycerol was very similar to that bound to its carbon 3 (see column a vs. column c in Table 2). The enrichment of the two hydrogens bound to carbon 3 of the VLDL-glycerol was about the same as the enrichment of the two hydrogens bound to carbon 6 of blood glucose (see column cvs. column e). The ratio of c to e was 1.05 ± 0.02 (SE; n = 4). The enrichments of the hydrogens bound to carbons 1 and 3 of glycerol from the VLDL was much more than from the infranatant (columns b vs. d). Thus the ratio of d to e was 0.168 ± 0.032. Presumably, enrichment at carbon 3 of the glycerol was about the same as at carbon 1 (see also Table 3), but only sufficient glycerol was isolated from subject 2 to prepare glycerol 3-P(see legend of Table 2). The 2H enrichment of body water was ∼0.45%. The mean enrichment in free blood glycerol, 16.8% of that in VLDL-glycerol, calculates to a nonadipose tissue glycerol contribution to systemic glycerol Ra of 90 ± 28 μmol/min.
Table 3 presents the results for the Swedish subjects fasted for 12, 42, or 60 h, but in whom only enrichments in glycerol and glucose from blood were determined. Again, enrichments of the hydrogens at carbons 1 and 3 of glycerol were about the same (columns a vs. c), except subject 5 for reason(s) unknown. The enrichments of the two hydrogens bound to carbon 6 of glucose were from 0.550 to 0.729% (column d) except after 12 h, when it was about one-half as much. This is in accord with gluconeogenesis contributing almost 100% to glucose production by 42 h of fasting, but only ∼50% after an overnight fast (19). The ratio of the enrichment at carbon 3 of glycerol to that at carbon 6 of glucose at 42 and 60 h of fasting (c/d, means ± SE, n= 5) is 0.162 ± 0.018, similar to the ratio of 0.168 ± 0.032 from the data in Table 2. Data from the two subjects fasted for 12 h are also provided in Table 3. The enrichment at carbon 6 of blood glucose is multiplied by 2 because of dilution of gluconeogenic glucose by unlabeled glucose formed by glycogenolysis. This givesc/d ratios of 0.065 and 0.119. Again, the body water enrichment was ∼0.45%, except less for subject 10.
There are two major findings from the results of Table 2. The first is the very similar enrichments of the two hydrogens bound to carbon 3 of VLDL-glycerol and carbon 6 of glucose after 60 h of fasting. These have a common intermediate (Fig.2), i.e., GAP, from which glycerol 3-P is formed and whose carbon 3 with its hydrogens is the precursor of carbon 6 of glucose 6-P with its hydrogens. The similar enrichments mean that there must be extensive equilibration between GAP and glycerol 3-P, because 2H from2H2O incorporation at carbon 6 of glucose occurs in the equilibration of alanine with pyruvate and of oxaloacetate with fumarate, i.e., before the formation of GAP (4). To the extent that unlabeled glycerol released by lipolysis is converted to glycerol 3-P by liver, that glycerol 3-P must also equilibrate with the GAP. Otherwise the enrichment at carbon 3 of the VLDL-glycerol would be less than that in the glucose. Siler et al. (32) found that, in normal men fasted for 20 h and infused with [2-13C]glycerol, the enrichment of VLDL-glycerol from a mixture of VLDL-triglyceride and phospholipid reached 60% of intrahepatic triose phosphate enrichments, calculated by mass isotopomer distribution analysis.
Near complete equilibration in these pools should be expected. Much of the triose phosphates, DHAP and GAP, probably ∼90% in long-term fasting, is from precursors entering the gluconeogenic process at the level of pyruvate. There is extensive evidence for essentially complete equilibration of DHAP with GAP (20). Glycerol, the precursor of ∼10% of glucose production in long-term fasting, must be converted to glycerol 3-P and then to DHAP (21). The reverse reaction is required for the formation of glycerol 3-P for triglyceride formation other than from glycerol. Hence, the enzyme glycerol-3-P dehydrogenase is an equilibration enzyme with high activity in liver (1). The enrichment at carbon 1 arises (Fig. 3) from reduction of 3-phosphoglyceric acid to GAP by NAD2H, formed in body water enriched with 2H2O, as well as by the isomerization of the GAP to DHAP. If there were complete equilibration, the enrichment of the two hydrogens bound to carbons 1 and 3 of glycerol and carbon 6 of glucose would be about twice the enrichment of the 2H of body water. Enrichment of the two hydrogens bound to carbon 6 of glucose was 82 ± 2% of twice the enrichment of the hydrogen of body water for the six subjects fasted for 60 h. That is in accord with extensive, but not complete, isotopic equilibration of the hydrogens that become bound to carbon 3 of phosphoenolpyruvate, the precursor of the hydrogens at carbon 6 of glucose, and contributions of gluconeogenesis from glycerol, and of glycogenolysis even after 60 h of fasting (4, 19).
The second major finding is that enrichment of the hydrogens bound to glycerol in the circulating blood was ∼16% of that at carbon 6 of the blood glucose and, hence, also that at carbon 3 of the VLDL-glycerol. Given the large stores of triglycerides in adipose tissue (an average of ∼16 kg of body fat in the volunteers studied at the Mayo GCRC), it would be virtually impossible to have enriched adipose tissue glycerol to a measurable extent during the 60 h of fasting. Even under fed conditions, only ∼2% of glucose carbons appear in adipose tissue triglyceride (23), presumably in the glycerol moiety. If labeled glucose were incorporated into the ∼17 mol of triglyceride glycerol stored in the adipose tissue of our volunteers, the fractional turnover rate [5- to 11-mo half-life in men (24)] is such that the release of 2H-labeled glycerol would be negligible. Current evidence is that triglycerides enter a large well-mixed pool inside adipocytes rather than participating in a “last in - first out” phenomenon (25). Therefore, the 2H label in the blood glycerol of our fasted volunteers almost certainly arose from sources other than adipose tissue lipolysis.
Intravascular hydrolysis of the labeled VLDL-triglyceride is one potential source. We recently observed splanchnic triglyceride release rates of 21 ± 2 μmol/min (∼0.26 μmol · kg−1 · min−1) in 60-h-fasted men (12), whereas Wolfe et al. (34) reported splanchnic triglyceride release of 39 μmol/min in 69-h-fasted men. If all glycerol released from this process entered the circulation, this could account for a significant portion, but likely not all, of the 2H-labeled free glycerol.
Another possible source of 2H-labeled glycerol would be the hydrolysis of intracellular triglycerides in small, rapidly turning over pools. For example, intramyocellular triglycerides provide fatty acids for oxidation but do not release fatty acids into the systemic circulation, even during exercise (10). Glucose used to synthesize the glycerol moiety of intramyocellular triglyceride in our volunteers would be labeled with 2H, and thus any glycerol released from intramyocellular lipolysis would likely be labeled to the same extent as glucose. If labeled to a lesser extent, our estimates of the extent of glycerol release from nonadipose tissue would be underestimates. Of interest, we previously reported renal glycerol release rates of ∼17 μmol/min (21), and more recently of 43 ± 8 μmol/min in 60-h-fasted men (12) in the absence of renal FFA release or triglyceride uptake. This could indicate glycerol release from the kidney, because lipolysis from perirenal fat would be expected to also release FFA. Evidence exists for substrate cycling between glycerol and triose phosphates in perfused liver (28). That could also be another possible source of the 2H-labeled glycerol, except that cycling has not been found to contribute to the appearance of glycerol in systemic circulations in vivo (7, 29).
Glycerol also makes up the backbone of VLDL phospholipids, accounting for ∼25% of glycerol in VLDL (32). The phospholipids were not separated from the VLDL lipid extract before assay. However, liver likely synthesizes VLDL phospholipids by using the same pool of 3 carbon precursors for the glycerol as used for VLDL-triglyceride synthesis. Although there could be exchange of phospholipids between lipoprotein particles via phospholipid transfer protein, low-density lipoprotein phospholipid would also be synthesized in the liver. Some intestinally derived high-density lipoprotein (HDL) particles might have a different phospholipid glycerol enrichment from the apoB-containing lipoprotein phospholipids or VLDL-triglyceride, but the amount of exchange from HDL to VLDL would be expected to be minor. Thus, although we did not measure the VLDL-phospholipid glycerol enrichment separately, we anticipate it would be similar, if not identical, to VLDL-triglyceride glycerol. If VLDL-phospholipid glycerol enrichment were different from VLDL-triglyceride glycerol enrichment, it would likely be less. Then VLDL-triglyceride glycerol enrichment would be underestimated, resulting in an overestimate of the contribution of VLDL-triglyceride glycerol to circulating free glycerol.
The results in Table 3 support the finding in Table 2, that ∼16% of glycerol in long-term fasting is from nonadipose tissue sources. With only two subjects, and perhaps an experimental error in the enrichments in glycerol from subject 5, we can only conclude that at least a small percentage of glycerol in the circulation after 12 h of fasting also arose from nonadipose tissue sources. Lipoprotein lipase has been suggested to be responsible for ∼25% of glycerol released from adipose tissue in normal subjects after an overnight fast (9). This is somewhat greater than our estimate based upon tracer techniques, but it emphasizes that reliance on measurement of glycerol Ra as the sole measure of lipolysis of fat in adipose tissue may be problematic in some circumstances.
This study was supported by National Institutes of Health Grants DK-40484, DK-14507, RR-00583, and AM-07319; the Swedish Medical Research Council; and the Mayo Foundation.
Address for reprint requests and other correspondence: B. R. Landau, Dept. of Medicine, Case Western Reserve Univ. School of Medicine, 10900 Euclid Ave., Cleveland, OH 44106–4951 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2001 the American Physiological Society