Gluconeogenesis and glucuronidation in liver in vivo and the heterogeneity of hepatocyte function.

In order to examine metabolic zonation in human liver, glycerol, which labels carbons 2 and 5 of glucose-6-P, and [1-C]lactate, which labels carbons 3 and 4 of glucose-6-P, in the process of gluconeogenesis, were infused intravenously into healthy subjects who ingested acetaminophen and had fasted 36 h. Distributions of C were determined in glucose in blood and in the glucuronic acid moiety of acetaminophen glucuronide excreted in urine. Ratios of C in carbons 2 and 5 to C in carbons 3 and 4 were significantly higher in blood glucose than in glucuronide. Since glucose and glucuronic acid are formed from glucose-6-P in liver without randomization of carbon, the differences in the ratios indicate that the pool of glucose-6-P in liver is not homogeneous. The glucuronide sampled glucose-6-P with more label from lactate than glycerol compared to the glucose-6-P sampled by the glucose. The apparent explanation is the greater decrease in glycerol compared with lactate concentration as blood streams from the periportal to the perivenous zones of the liver lobule. Glucuronidation is then expressed in humans relatively more in the perivenous than periportal zones and gluconeogenesis from glycerol more in the periportal than perivenous zones.

In order to examine metabolic zonation in human liver, [2-14 C]glycerol, which labels carbons 2 and 5 of glucose-6-P, and [1-14 C]lactate, which labels carbons 3 and 4 of glucose-6-P, in the process of gluconeogenesis, were infused intravenously into healthy subjects who ingested acetaminophen and had fasted 36 h. Distributions of 14 C were determined in glucose in blood and in the glucuronic acid moiety of acetaminophen glucuronide excreted in urine. Ratios of 14 C in carbons 2 and 5 to 14 C in carbons 3 and 4 were significantly higher in blood glucose than in glucuronide. Since glucose and glucuronic acid are formed from glucose-6-P in liver without randomization of carbon, the differences in the ratios indicate that the pool of glucose-6-P in liver is not homogeneous. The glucuronide sampled glucose-6-P with more label from lactate than glycerol compared to the glucose-6-P sampled by the glucose. The apparent explanation is the greater decrease in glycerol compared with lactate concentration as blood streams from the periportal to the perivenous zones of the liver lobule. Glucuronidation is then expressed in humans relatively more in the perivenous than periportal zones and gluconeogenesis from glycerol more in the periportal than perivenous zones.
Hepatocytes from the periportal and perivenous zones of livers of animals have been found by in vitro techniques to have differing metabolic capacities (1,2). In gluconeogenesis, glucose-6-P formed from gluconeogenic substrates is hydrolyzed to glucose. In glucuronidation, glucuronic acid from UDP-glucuronic acid is conjugated. The UDP-glucuronic acid is formed from glucose-6-P with glucose-1-P and UDP-glucose as intermediates. Thus, both glucose and glucuronic acid are formed from glucose-6-P with carbon skeletons unchanged.
In humans fasted for 12 h and 60 h, the concentration of glycerol, but not lactate, is much lower in hepatic vein than arterial blood (3)(4)(5). This means that during fasting periportal, as compared to perivenous, cells of human liver are exposed to relatively higher glycerol than lactate concentrations. Since a higher capacity for gluconeogenesis is reported to exist in periportal than perivenous cells of animals and for glucuronidation in perivenous than periportal cells (1, 2), we postulated that if carbon-labeled lactate and glycerol were administered to humans and metabolic zonation similarly exists in human liver in vivo, more label from the glycerol would be found in glucose than in a glucuronide relative to label from the lactate.
We infused into fasted normal subjects, given acetaminophen, [2-14 C]glycerol, which labels carbons 2 and 5 of glucose-6-P, and [1-14 C]lactate, which labels carbons 3 and 4 of glucose-6-P (Fig. 1). The ratios of the 14 C in carbons 2 and 5 to the 14 C in carbons 3 and 4 of blood glucose were compared with the ratios in the acetaminophen glucuronide excreted in urine.

EXPERIMENTAL PROCEDURES
Subjects-The subjects were four healthy women volunteers, ages 26 -44 years, on weight-maintaining diets, and with a mean body mass index (kg/m 2 ) of 19.8 (range 19.4 -20.1). The experimental protocol was approved by the Human Investigation Committees at the Karolinska Hospital and University Hospitals of Cleveland. Informed consent was obtained from each subject.
Protocol-The subjects fasted for 36 h. A catheter was then inserted into a peripheral vein of one arm for blood sampling and into a peripheral vein of the other arm for infusion. In the first subject we infused 10 Ci of [2-14 C]glycerol, and in each of the other three subjects 25 Ci, in trace quantity (specific activity 40 mCi/mmol; purchased from ICN Biomedical Inc., Irvine, CA). In the first subject we also infused 60 Ci of [1-14 C]lactate, and in each of the other three subjects 55 Ci (10 Ci as a bolus), again in trace quantity (specific activity 61 mCi/mmol; also purchased from ICN Biomedical Inc.). The labeled glycerol and lactate were in solution in sterile isotonic saline, shown to be pyrogen-free, and the infusion rate was constant at between 10 and 15 ml/h. Duration of infusion in the first two subjects was 5 h and in the other two subjects 8 h. The first two subjects ingested 0.5 g of acetaminophen at the beginning and at 1 and 2 h into the infusion, the other two subjects at 3, 4, and 5 h into the infusion.
Blood samples, each 75 ml, were drawn between 3 and 5 h in the first two subjects and between 5 and 8 h in the other two subjects. Subjects were encouraged during the infusions to drink about 240 ml of water/h. Urine was collected from the first two subjects between 2-3.5 h and 3.5-5 h and in the other two between 5-6.5 h and 6.5-8 h. One ml blood samples were collected at the beginning and end of each infusion for determination of plasma glucose concentration.
Analyses-Glucose concentration in plasma was determined using glucose oxidase. Glucose was isolated from blood as described previously (6). Briefly, the blood at the time of collection was deproteinized by the addition of ZnSO 4 and Ba(OH) 2 . The precipitate was removed by centrifugation and the supernatant frozen until it could be processed. It was deionized by passage through a column of ion exchange resins. Glucose in the effluent was isolated using preparative paper chromatography followed by high pressure liquid chromatography.
The acetaminophen glucuronide was isolated, reduced to its glucoside, and glucose isolated from the glucoside as described previously (7,8). Briefly, urine at the time of collection was taken to pH 4.5 and frozen until processed further. The urine was concentrated, then brought to its original volume with methanol, and the precipitate that formed was discarded. The methanol was evaporated and the resulting concentrate made basic and applied to a column of AG1-X8 in the acetate form (Bio-Rad). The column was washed with water and then increasing concentrations of acetic acid. The fraction eluted containing the glucuronide, identified using carbazole (7), was evaporated to dryness. The glucuronide was reduced with diborane to acetaminophen glucoside and the glucoside hydrolyzed with ␤-glucosidase. Glucose from the deionized hydrolysate was isolated as just described for glucose from blood.
A portion of each glucose from blood and urine was combusted to CO 2 . The remainder of each glucose was degraded to yield each of its six carbons as CO 2 . Each CO 2 was assayed for 14 C specific activity (6,7).
Calculations-The specific activity of each carbon of glucose divided by the sum of the specific activities of the six carbons of glucose time 100 is recorded as the percentage of 14 C in that carbon. The sum of the six specific activities divided by the specific activity of the glucose carbons, determined by the combustion of the glucose to CO 2 , ϫ 100, is recorded as percentage of recovery. That provides a measure of the adequacy of the degradation.
Statistics-The significance of the differences in the ratios of incorporations of 14 C in blood glucose to those in urinary glucuronide and other differences were assessed by a two-tailed paired t test (9).

RESULTS
Plasma glucose concentrations at the beginning of the infusions ranged from 3.9 to 5.1 mM. Over the duration of the infusions concentrations declined by 0.5-0.7 mM. Amounts of 14 C in glucoses isolated from some blood samples were small, and therefore degradations of those glucoses were not done. 14 C was in greatest percentages in carbons 2, 3, 4, and 5 (Table I) in accord with the pathways followed by the labeled carbons of the glycerol and lactate (Fig. 1). Recoveries were good. There was only about one-third as much 14 C in carbons 2 and 5 as carbons 3 and 4 of glucose from the first subject, AD. This was the reason for giving 25 Ci rather than 10 Ci of [2-14 C]glycerol to the other three subjects, since the lower the amount of 14 C in a carbon, the more likely an error in the determination of the amount. 14 C in carbon 2 was more than in carbon 5 (p Ͻ 0.001), and 14 C in carbon 4 was more than in carbon 3 (p Ͻ 0.001). This result is in accord with glycerol's entrance into the triose-P pool via dihydroxylacetone-3-P and lactate's via glyceraldehyde-3-P and incomplete isotopic equilibration of the triose phosphates. Distributions in the glucose from blood and from glucuronide in the periods of collection were remarkably similar. The ratios of 14 C in carbons 2 and 5 to 14 C in carbons 3 and 4 were significantly different in blood glucose and in glucose from glucuronide (p Ͻ 0.005), being higher by an average of 30% (range from 26 to 37%). Incorporations into carbons 1 and 6 occur in the formation of glucose-6-P from [2-14 C]lactate formed from the [2-14 C]glycerol. Subtraction of the incorporations of 14 C from the [2-14 C]lactate into carbons 2 through 5 (10, 11) will give a still greater difference in the carbons 2 and 5 to carbons 3 and 4 ratios. DISCUSSION The duration of infusion was increased from 5 to 8 h, with sampling during the last 3 h of infusion, to assure a steady state. However, just the very similar distributions at different times of blood and urine collections for each subject provide that assurance. Since glucose and glucuronic acid are both formed from glucose-6-P without randomization of carbon, the differences in the distributions, reflected in the differing C 2 ϩ C 5 /C 3 ϩ C 4 ratios, mean there was not a homogeneous pool of glucose-6-P in liver. The glucuronide sampled glucose-6-P in hepatocytes where there was overall more label from lactate than glycerol, than was the case for the glucose-6-P the glucose sampled.
The approach we introduce here for testing for metabolic heterogeneity in liver demonstrates differing metabolism in hepatocytes in humans in vivo. That conclusion rests on two assumptions. First, that the distribution of 14 C in blood glucose reflects distribution in glucose-6-P in liver. Liver is the major source of blood glucose, but there is a report of a significant contribution of kidney to glucose production in dogs fasted overnight (12). Second, that the distribution of 14 C in the glucuronide reflects distribution in glucose-6-P in liver. Liver is the major site of glucuronidation, although glucuronidation can occur in other tissues (13). The present results are in accord with our previous finding that when [U-13 C]glycerol and acet-   (14) that rats infused with [2-13 C]glycerol and acetaminophen, and fasted 24 h or infused with fructose, formed glucose and acetaminophen glucuronide from triose-P of the same enrichment. This seeming difference from our findings may be related to the difference in species, a quantity of glycerol rather than a trace amount being given to the rats, the conversion of glycerol to lactate, and the effect of the fructose load when that was also given.
A different gradient in the concentrations of glycerol and lactate along the liver lobule (3)(4)(5) is an apparent major reason for the differing distributions we observe. Other factors, such as differences in the activity of glycerokinase along the liver lobule, could play a role. Hence, hepatocytes in the hepatic vein zone were presumably exposed to more label from the lactate than glycerol compared to hepatocytes in the periportal zone. The glucuronide compared to the glucose then sampled glucose-6-P to a greater extent in the hepatic venous than periportal zone. That conclusion is in accord with a higher capacity for glucuronidation found in rat liver in vitro in hepatic venous than periportal zones.