INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

THE ROUTE OF GLUCOSE DELIVERY is one of the key determinants of net hepatic glucose uptake (NHGU), with NHGU being enhanced two- to threefold during infusion of glucose into the portal vein vs. a peripheral vein (11). The ability of intraportal glucose delivery to increase NHGU is not dictated by the absolute glucose concentration in the portal vein but is instead a function of the magnitude of the negative arterial-portal (A-P) glucose gradient. It seems likely, therefore, that after portal glucose delivery, the portal glucose level is sensed and compared with the arterial glucose level, which is simultaneously monitored at some as yet undetermined site. Three possible arterial reference sites have been suggested: the hepatic artery (6, 24), the arterial blood supply of the hypothalamus (4), and the arterial blood reaching the carotid bodies (2, 3).

Using isolated perfused rat liver, Gardemann et al. (6) and Stumpel and Jungermann (24) showed that a negative glucose gradient between the hepatic artery and the portal vein could be transformed into a metabolic signal locally within the liver and that the action of acetylcholine on the hepatocyte seemed to be involved in this signal transduction. Recently, Horikawa et al. (7) reported that both portal vein and hepatic arterial glucose infusion stimulated NHGU in the conscious dog. These authors went on to suggest that glucose sensors within the liver, rather than the portal vein, are involved in the augmentation of NHGU.

Data also exist to support the contention that the portal glucose level is compared with an arterial glucose level sensed outside the liver. Results from Matsuhisa et al. (9) suggested that minimizing the glucose gradient between the portal vein and the central nervous system diminished hepatic glucose uptake. Our previous study (8) overcame many of the problems associated with their study design, however, and failed to confirm their results. In our study, there was no diminution of NHGU during portal glucose delivery when glucose was infused simultaneously into the carotid and vertebral arteries to maintain the brain isoglycemic with the portal vein (8). Other studies (2, 3) have shown that changes in the blood glucose concentration within the carotid body can affect glucose homeostasis. This implies that the carotid bodies provide another potential reference site for arterial glucose sensing. However, infusion of glucose into the carotid and vertebral arteries, as in our earlier study (8), would also alter the glucose level in the carotid bodies. Our earlier data, therefore, do not support carotid body involvement in the operation of the portal signal.

The aim of the present study was to clarify this issue by determining whether elimination of the hepatic A-P venous glucose gradient within the liver would alter the changes in net hepatic and peripheral glucose uptake induced by portal glucose delivery in conscious dogs.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals and surgical procedures. Studies were carried out on twenty-one 42-h-fasted conscious mongrel dogs of either sex weighing 21-28 kg. All animals were maintained on a diet of meat (Kal-Kan, Vernon, CA) and chow (Purina Lab Canine Diet no. 5006; Purina Mills, St. Louis, MO) composed of 34% protein, 14.5% fat, 46% carbohydrate, and 5.5% fiber based on dry weight. The protocol was approved by the Vanderbilt University Medical Center Animal Care Committee, and the animals were housed according to the guidelines of the American Association for the Accreditation of Laboratory Animal Care International. Approximately 16 days before the study, each dog underwent surgery under general anesthesia for placement of sampling catheters into a hepatic vein, the portal vein, and a femoral artery and infusion catheters in a splenic and a jejunal vein (11). A Silastic catheter was inserted into the gastroduodenal artery, and its tip was advanced 3-4 cm retrograde into the common hepatic artery. The distal end of the gastroduodenal artery was then ligated to ensure total delivery of glucose to the hepatic bed. Ultrasonic flow probes (Transonic Systems, Ithaca, NY) were positioned around the portal vein and hepatic artery. The proximal ends of the probes and catheters were placed in subcutaneous pockets (11).

Experimental design. At 120 min, a primed (36 µCi) continuous (0.3 µCi/min) peripheral infusion of D-[3-³H]glucose and a continuous peripheral infusion of indocyanine green dye [(ICG) Sigma Chemicals, St. Louis, MO, 4 µg·kg¹·min¹] were begun in all three protocols. The ICG infusion allowed confirmation of hepatic vein catheter placement and provided a second measurement of hepatic blood flow. After 80 min (from 120 to 40 min) of dye equilibration, there was a 40-min (from 40 min to time 0) basal period followed by two 90-min experimental periods. At time 0, constant infusions of several solutions were begun. Somatostatin (0.8 µg·kg¹·min¹; Bachem, Torrance, CA) was infused to suppress endogenous insulin and glucagon secretion. Insulin (7.2 pmol·kg¹·min¹) and glucagon (0.65 ng·kg¹·min¹; both from Eli Lilly, Indianapolis, IN) were infused intraportally at rates designed to elevate insulin three- to fourfold and to keep glucagon basal. In addition, a primed continuous peripheral infusion of 50% dextrose was begun so that the arterial blood glucose level could be quickly clamped at a value approximately equal to twofold basal. In the first experimental period, glucose was infused only through a peripheral vein to double the glucose load to the liver in all groups. In the control study (CONT, n = 7) the conditions established in the first test period were continued throughout the second experimental period. During the second experimental period of the other two groups, glucose was infused intraportally (22.2 µmol·kg¹·min¹). In one group, saline was concurrently infused into the hepatic artery (PD, n = 7), but in the other group, glucose was concurrently infused into the hepatic artery (8.0 ± 0.5 µmol·kg¹·min¹) to eliminate the glucose gradient between the portal vein and the hepatic artery (PD+HAD, n = 7). The peripheral glucose infusion rate was modified as required to maintain the hepatic glucose load equal to twofold basal in each protocol. Dextrose, 20 and 5%, was used for the portal and hepatic arterial glucose infusions, respectively, and p-aminohippuric acid (PAH; Sigma Chemicals, St. Louis, MO) was added to the infusates to assess mixing of the infused glucose with blood in the portal and hepatic veins, as described previously (11, 18). Blood sampling was performed as previously described (8).

Processing and analysis of samples. Plasma glucose was assayed by the glucose oxidase method with a Beckman glucose analyzer (Fullerton, CA). Plasma insulin and glucagon concentrations were determined by RIA, as previously described (8). Blood glucose and blood lactate levels were determined by fluorometric enzymatic assays on perchloric acid-treated samples, as previously described (8). PAH was also measured in perchloric acid-deproteinized blood (11, 18).

Calculations. When substrates are infused into the portal vein, the possibility of poor mixing with the blood in the laminar flow of the portal circulation is of concern. In addition, multiple branching patterns of the hepatic artery in the dog (21) raise concern about the mixing of infusates with the blood of the hepatic artery. In both PD and PD+HAD, the mixing of the infusate with portal vein blood was assessed by comparing the portal PAH infusion rate with the calculated appearance rate of PAH in the portal vein [the difference between the rate of PAH exiting in portal blood and the rate of PAH entering the splanchnic bed through the arterial system (11, 18)], as shown in the equation below
where A and P represent the femoral arterial and portal venous blood concentrations, and PBF represents portal blood flow.

1

1

1

1

1

1

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Plasma insulin and glucagon concentrations. Arterial (data not shown) and liver sinusoidal insulin concentrations rose nearly three- to fourfold in all groups. Glucagon levels, on the other hand, remained basal. Neither hormone differed between groups (Fig. 1).

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Calculated plasma insulin (top) and glucagon (bottom) levels in the hepatic sinusoid in 42-h-fasted conscious dogs during the basal and two experimental periods in PD (portal vein glucose delivery), PD+HAD (portal vein and hepatic artery glucose delivery), and CONT (control) groups (n = 7/group). SRIF, somatostatin; PO+HA Glucose, intraportal and hepatic arterial glucose infusion. There are no significant differences between groups.

Blood glucose levels, the A-P glucose gradient and hepatic blood flow. Peripheral glucose infusion in the first test period doubled the blood glucose levels (Fig. 2) such that the arterial, portal venous, and hepatic venous glucose levels were not significantly different (NS) among the three groups. In the second test period, the arterial blood glucose level was slightly higher in CONT than in the other two groups (9.3 ± 0.2* vs. 8.6 ± 0.2 and 8.3 ± 0.2 mM, * P < 0.05 vs. PD and PD+HAD, respectively). The portal and hepatic glucose levels, on the other hand, were indistinguishable in the three groups. In PD, intraportal glucose infusion switched the A-P blood glucose gradient from 0.10 ± 0.04 (period 1) to 0.75 ± 0.08 mM and thereby presented the liver with a portal signal (Fig. 3B). In PD+HAD, the hepatic arterial glucose infusion offset the portal glucose infusion so that a positive glucose gradient was maintained between the hepatic artery and the portal vein, even in the latter experimental period (0.22 ± 0.03 vs. 0.55 ± 0.08 mM; Fig. 3A). However, there was still a negative glucose gradient between the femoral artery and the portal vein (0.57 ± 0.10 mM). A positive A-P glucose gradient was maintained throughout the experiment in CONT (0.11 ± 0.03 and 0.14 ± 0.07 mM in the two test periods, respectively).

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Blood glucose levels in the femoral artery, portal vein, and hepatic vein in 42-h-fasted conscious dogs during the basal and two experimental periods in PD, PD+HAD, and CONT groups (n = 7/group). * P < 0.05 vs. CONT at this time point.

View larger version (29K): [in this window] [in a new window]   Fig. 3.   Blood glucose levels in the hepatic arteries (calculated), portal vein, and femoral artery in 42-h-fasted conscious dogs during the 4 sampling points in the second experimental period in PD+HAD (A) and PD (B); n = 7/group.

1

1

Hepatic glucose load, net hepatic glucose balance, and net hepatic fractional extraction of glucose. The hepatic glucose loads (HGL) did not differ (268 ± 29 and 264 ± 24 µmol·kg¹·min¹) between the two test periods in CONT. The HGLs in the other two groups were similar, although the HGL rose slightly in period 2 in these groups (252 ± 19 and 281 ± 18 vs. 233 ± 19 and 278 ± 21*µmol·kg¹·min¹ in PD and PD+HAD; * P < 0.05 vs. period 1; Fig. 4, top). Net hepatic glucose balance (NHGB, i.e., net hepatic glucose output and NHGU) is shown in Fig. 4 (bottom). In the basal period, net hepatic glucose output did not differ between PD, PD+HAD, and CONT (9.7 ± 1.1, 9.7 ± 1.6, and 9.3 ± 1.7 µmol·kg¹·min¹, respectively). Peripheral glucose infusion in the presence of hyperinsulinemia resulted in similar rates of NHGU in all three groups (11.5 ± 1.6, 9.0 ± 2.3, and 10.1 ± 2.2 µmol·kg¹·min¹ in PD, PD+HAD, and CONT, respectively). During the latter experimental period, NHGU increased to 23.8 ± 3.3 µmol·kg¹·min¹ in PD (12.4 ± 3.2 µmol·kg¹·min¹, P < 0.05), to 13.8 ± 4.2 µmol·kg¹·min¹ in PD+HAD (4.9 ± 2.4 µmol·kg¹·min¹, NS), and to 12.8 ± 2.1 µmol·kg¹·min¹ in CONT ( 2.7 ± 1.5 µmol·kg¹·min¹, NS). NHGU did not differ between PD+HAD and CONT at any time, indicating that elimination of the glucose difference between the hepatic artery and the portal vein markedly reduced the effect of portal glucose delivery on NHGU (Fig. 4, bottom). When the data were analyzed with D (rather than I), NHGU increased to 20.5 ± 2.9 µmol·kg¹·min¹ in PD (9.1 ± 2.4 µmol·kg¹·min¹, P < 0.05), to 8.4 ± 2.7 µmol·kg¹·min¹ in PD+HAD (0.3 ± 0.7 µmol·kg¹·min¹, NS), and to 12.8 ± 2.1 µmol·kg¹·min¹ in CONT ( 2.7 ± 1.5 µmol·kg¹·min¹, NS).

View larger version (29K): [in this window] [in a new window]   Fig. 4.   Hepatic glucose load (top) and net hepatic glucose balance (bottom) in 42-h-fasted conscious dogs during the basal and two experimental periods in PD, PD+HAD, and CONT (n = 7/group).  P < 0.05, CONT vs. PD at this time point; * P < 0.05, PD+HAD vs. PD at this time point.

View larger version (16K): [in this window] [in a new window]   Fig. 5.   Net hepatic fractional extraction in 42-h-fasted conscious dogs during the basal and two experimental periods in PD, PD+HAD, and CONT (n = 7/group).  P < 0.05, CONT vs. PD at this time point; * P < 0.05, PD+HAD vs. PD at this time point.

Nonhepatic glucose uptake. Peripheral glucose infusion during period 1 resulted in nonhepatic glucose uptake (non-HGU; Fig. 6) of 26.3 ± 3.7, 36.1 ± 3.8, and 28.9 ± 1.2 µmol·kg¹·min¹ in PD, PD+HAD, and CONT, respectively. In period 2, non-HGU was 24.5 ± 3.7, 53.3 ± 8.5, and 39.5 ± 4.3 µmol·kg¹·min¹ in PD, PD+HAD, and CONT, respectively. Non-HGU rose by 10.5 ± 4.2 µmol·kg¹·min¹ in period 2 of CONT. In PD, non-HGU fell by 1.7 ± 4.7 µmol·kg¹·min¹ in period 2, and thus the net effect of the portal signal was to decrease non-HGU by 12.4 ± 4.7 µmol·kg¹·min¹. In PD+HAD, non-HGU rose by 17.5 ± 5.9 µmol·kg¹·min¹ in period 2. The change in non-HGU did not differ between PD+HAD and CONT. Thus hepatic arterial glucose delivery completely blocked the suppressive effect of portal glucose delivery on non-HGU.

View larger version (27K): [in this window] [in a new window]   Fig. 6.   Nonhepatic glucose uptake in 42-h-fasted conscious dogs during the basal and two experimental periods in PD, PD+HAD, and CONT (n = 7/group).  P < 0.05, CONT vs. PD at this time point; * P < 0.05, PD+HAD vs. PD at this time point.

1

1

Net hepatic lactate balance. In response to peripheral glucose infusion, net hepatic lactate balance switched from uptake (10.8 ± 1.7, 8.7 ± 1.3 and 5.9 ± 2.2 µmol·kg¹·min¹) to output (4.4 ± 1.3, 3.6 ± 2.4 and 5.8 ± 1.8 µmol·kg¹·min¹) in the first test period of PD, PD+HAD, and CONT, respectively. Net hepatic lactate output fell slightly in the latter experimental period but was not significantly different among the three groups (Table 1). Blood lactate levels in the femoral artery, portal vein, and hepatic vein were also not different among the three groups (data not shown).

Net hepatic balance of glucose equivalents. The net balance of glucose equivalents across the liver represents the combination of glucose and lactate balance (after the latter is converted to glucose equivalents) and serves as an index of glycogen deposition. The net balance of glucose equivalents across the liver switched from output (4.0 ± 1.3, 5.4 ± 1.7, and 9.0 ± 1.7 µmol·kg¹·min¹) to uptake (9.2 ± 1.9, 6.9 ± 1.8, and 7.2 ± 1.8 µmol·kg¹·min¹) in PD, PD+HAD, and CONT, respectively, in response to peripheral glucose infusion (Fig. 7). In the latter experimental period, the uptakes of glucose equivalents in the three groups were 22.1 ± 3.2,* 13.8 ± 4.0, and 10.9 ± 1.6 µmol·kg¹·min¹, respectively. Hepatic arterial glucose infusion markedly reduced the stimulatory effect of the portal signal on glycogen deposition (* P < 0.05 vs. the other two groups).

View larger version (33K): [in this window] [in a new window]   Fig. 7.   Net hepatic balance of glucose equivalents in 42-h-fasted conscious dogs during the basal and two experimental periods in PD, PD+HAD, and CONT (n = 7/group). The net balance of glucose equivalents is calculated as the sum of the net balance of glucose and lactate, the latter converted to glucose equivalents.  P < 0.05, CONT vs. PD at this time point; * P < 0.05, PD+HAD vs. PD at this time point.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Previous studies have demonstrated that the portal signal is an important component of the metabolic response to feeding (4, 11, 12, 18). We have established that a negative A-P gradient is necessary for the initiation of the portal signal (19). However, the reference glucose concentration against which the portal glucose concentration is compared is still not clear. This study sought to determine whether elimination of a negative glucose gradient between the hepatic artery and the portal vein would alter the effects of the portal signal on glucose uptake by the liver and/or peripheral tissues. The present data demonstrate that the effects of the portal signal on hepatic and peripheral glucose uptake and on the stimulation of hepatic glycogen storage were markedly reduced by eliminating the hepatic A-P glucose gradient within the liver in conscious dogs.

The results from the present study suggest that the portal signal is generated within the liver itself. They are consistent with the work of Gardemann et al. (6) and Stumpel and Jungermann (24) in the perfused liver. However, the intrahepatic mechanism by which the portal signal is generated and the way in which it is transduced into an effect on hepatic glucose uptake are still unknown. Stumpel et al. suggested that the parasympathetic nervous system may be involved. In their studies with isolated perfused rat livers, the increment in insulin-stimulated glucose uptake induced by a negative A-P glucose gradient was markedly reduced by either portal or arterial delivery of atropine. In addition, they showed that the effect of a negative A-P glucose gradient on NHGU could be mimicked by addition of acetylcholine to either the portal or arterial perfusate. The combination of acetylcholine and adrenergic blockers infused intraportally in the conscious dog rapidly stimulated NHGU in the presence of hyperinsulinemia and hyperglycemia, but adrenergic blockade alone did not alter NHGU (23). The above change in NHGU was consistent with those induced by the portal signal in terms of both time course and magnitude. These observations suggest that the muscarinic nervous system within the liver is somehow involved in the generation of the hepatic effect of the portal signal and/or in the transduction of its effect into a biological action.

Neurophysiological data indicate that glucose-responsive neurons in specific hypothalamic regions and in the dorsomedial medulla oblongata modulate glucose metabolism in some organs (liver and pancreas) through autonomic efferent nerves (16, 17). However, only one study (9) has suggested that the head glucose level is involved in the generation of the portal signal. Furthermore, the quantitative accuracy of that study has been questioned on several grounds, including the mixing of their infusate in portal blood (8). In a previous study designed to examine the same question, we infused glucose (22.2 µmol·kg¹·min¹) intraportally at a rate known to increase NHGU maximally in the presence of a fourfold rise in insulin. Unlike Matsuhisa et al. (9), we used a rate that minimized potential problems of portal glucose mixing (a favorable noise-to-signal ratio). In addition, we used an independent method (PAH assay) to assess the mixing of the infused glucose in the portal vein and the hepatic vein, and this in turn allowed evaluation of the potential effects of imperfect mixing in the two individual protocols. Also, we infused glucose into the carotid and vertebral arteries bilaterally instead of unilaterally, as had Matsuhisa et al., to ensure a uniform elimination of the difference between brain and portal vein glucose. Our results clearly demonstrated that, under hyperglycemic hyperinsulinemic conditions, raising the head arterial glucose level did not modify the increase in NHGU seen in response to portal glucose delivery. Thus the brain arterial glucose level does not appear to provide the reference information used in generation of the portal signal in conscious dogs. To date, our data suggest that the liver, not the brain, is the organ critical for orchestration of the distribution of dietary glucose after a carbohydrate meal.

A recent report by Horikawa et al. (7) suggested that augmentation of hepatic glucose uptake was not dependent on the sign of the A-P venous glucose gradient and that a difference in either direction (A > P or P > A) was effective. They used conscious dogs in an experiment consisting of a 30-min control and three 90-min test periods. After the control period, glucose (55.6 µmol·kg¹·min¹) was first infused via the superior mesenteric vein; it was then infused into both the superior mesenteric vein and the gastroduodenal artery (27.8 µmol·kg¹·min¹ in each vessel); and finally, it was infused solely into the gastroduodenal artery (55.6 µmol·kg¹·min¹). Unfortunately, the data from that study must be questioned on several grounds. First, the authors did not assess the mixing of their infusate with blood in any vessel, and poor portal glucose mixing was likely, given the extremely high glucose infusion rates used. This is further evident from the inconsistency between NHGU calculated by the direct and indirect methods in the second test period. When the indirect method of glucose balance calculation was used, it appeared that NHGU decreased 50% during concurrent infusion of glucose into the portal and arterial systems, suggesting an inhibition of NHGU by hepatic arterial glucose infusion. On the other hand, no decrease in NHGU was evident when the direct method of calculation was used, making it difficult to draw a conclusion. Second, the pancreatic hormones were not clamped, and as a result they varied somewhat over the course of the study. Third, the authors failed to include a control group that would account for changes over time during the study. Finally, the plasma glucose values were not in steady state, thus limiting the accuracy of the arteriovenous difference calculation.

In the present study, we made several improvements over the study design of Horikawa et al. (7) to minimize the above errors. First, the hepatic arterial glucose infusion (8.0 ± 0.5 µmol·kg¹·min¹) was kept small to minimize the impact of any mixing problems (i.e., to improve the signal-to-noise ratio). Even under our carefully controlled conditions, mixing of the infusate in the hepatic artery was probably not perfect. This is evidenced by the fact that the positive gradient between the hepatic artery and the portal vein in PD+HAD was slightly greater during the latter experimental period than during the first. Second, we used PAH to assess mixing of both the portal and hepatic arterial infusates and only "mixed" dogs were utilized for data analysis. Third, we used two control groups so that we could isolate the effects of hyperglycemia, hyperinsulinemia, and the portal signal over time. Fourth, we clamped insulin, glucagon, and the HGL so that they did not differ among the protocols. Finally, we made our measurements in a steady-state period. The present results thus clearly demonstrate that a gradient between the hepatic arterial and portal vein glucose levels is critical for the generation of the hepatic and extrahepatic effects of the portal signal.

It is also possible, however, that the portal signal is sensed within the liver and that the transduction of its effect is mediated by the autonomic nervous system outside the liver. This is made more likely by the need to explain the coordinate but discrete responses of the liver and nonhepatic tissues (primarily muscle). It is clear that the glucose portal signal not only enhances hepatic glucose uptake but also suppresses nonhepatic glucose uptake (Fig. 6). If this were not true, the glucose infusion rate would have been the same in PD and PD+HAD. Earlier reports have described the existence of neural pathways that link the liver to the brain and the brain to the liver and various endocrine organs (5, 20). Afferent fibers in the hepatic branch of the vagus nerve (14) and neurons in the lateral hypothalamus (22) can respond to the presence of glucose in the portal vein. Functional studies also have demonstrated that an intact nerve supply to the liver appears to be vital for a normal hepatic response to intraduodenal or intraportal glucose delivery (1, 10). Niijima and colleagues (13, 15) reported that the stimulation of hepatic afferents can alter the efferent activity of the adrenal and vagal pancreatic nerves in the rabbit and rat. There is no doubt that an autonomic link between the liver and peripheral organs exists, but the extent to which it is involved in the initiation of the effects of the portal signal remains to be determined.

Recently, several reports from Xie and Lautt (26, 27) focused on the relationship between peripheral insulin resistance and the activity of hepatic parasympathetic nervous system. They reported that surgical denervation of the hepatic anterior plexus or intraportal atropine infusion reduced the magnitude of insulin's effectiveness in skeletal muscle but had no effect on the liver or gut (25, 26). Complete hepatic denervation plus vagotomy did not cause further impairment, thus indicating that all of the relevant nerves reached the liver via the anterior plexus. Futhermore, intraportal, but not peripheral, acetylcholine infusion reversed insulin resistance produced by liver denervation (27). These studies suggest that the hepatic parasympathetic nerves regulate release of a liver-generated factor that selectively controls insulin effectiveness in skeletal muscle. Because the intrahepatic parasympathetic system has been suggested to be a key determinant in generation and/or transduction of the effect of the portal signal (24), the results from Xie and Lautt (25-27) further imply that the anterior plexus around the hepatic artery might be a potential reference site for generation of the suppressive effect of the portal signal on nonhepatic glucose uptake. This possibility needs to be investigated further.

In summary, the elimination of the negative glucose gradient between hepatic artery and portal vein markedly reduced the effects of the portal signal on hepatic and peripheral glucose uptake. This suggests that the liver plays a primary role in the regulation of postprandial glucose distribution, not only by augmenting its own uptake of glucose but by preventing glucose uptake by skeletal muscle.