INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

ADMINISTRATION OF THE fructose analog 2,5-anhydro-D-mannitol (2,5-AM) stimulates eating behavior in rats by its action in the liver (23, 24). 2,5-AM is phosphorylated to the mono- and bisphosphate forms in liver (17) but is not metabolized further (8). As a result, 2,5-AM decreases liver ATP because phosphate that would otherwise be available for ATP synthesis is trapped in the phosphorylated forms of 2,5-AM (12, see also Ref. 22). The eating response to 2,5-AM appears to be triggered by this decrease in liver ATP. Injection of exogenous phosphate along with 2,5-AM prevents both the reduction in liver ATP and the increase in food intake, suggesting that the decrease in ATP due to phosphate-trapping underlies the eating response (13). A decline in liver ATP, as opposed to reduced phosphate, appears critical for stimulation of eating behavior because administration of L-ethionine, which decreases liver ATP by trapping adenine, also increases food intake in rats (14).

Additional evidence that 2,5-AM triggers eating by decreasing liver ATP stems from observations suggesting that the changes in liver metabolism and behavior occur within a similar time frame (3, 12). Liver ATP, measured enzymatically in freeze-clamped liver, declines within 30 min and food intake increases within 1 h after intraperitoneal injection of 2,5-AM (13, 23). With intravenous infusion of 2,5-AM, liver ATP, measured in anesthetized rats using ³¹P-nuclear magnetic resonance (NMR) spectroscopy, first decreases and then partially recovers (3, 12). When conscious animals are given the same dose of 2,5-AM by the same route, their eating response parallels this time course of changes in liver ATP (3).

Although suggestive, these observations do not address important questions about the temporal relationship between the decrease in liver ATP and stimulation of food intake after 2,5-AM treatment. First, they do not establish whether the decline in ATP precedes the initiation of eating as would be expected if the metabolic change triggered the behavioral response. Second, because the time course of the change in liver ATP induced by 2,5-AM injection was assessed only in anesthetized animals, these earlier observations do not clarify how well the time course of the eating response tracks that of the metabolic response under conditions relevant to the behavior (i.e., in conscious animals). Third, the evidence that the metabolic and behavioral effects of 2,5-AM show similar temporal patterns was based on correlative comparisons of data from two different experiments; there was no direct test of whether initiation of the eating response to 2,5-AM at a given time after treatment depends on the level of liver ATP. Fourth, the specificity of the relationship between liver ATP content and eating behavior was not evaluated; for example, changes in the tissue concentration of other adenine nucleotides, which can vary along with that of ATP, may also track the eating response to 2,5-AM.

The present experiments were designed to address these issues and thereby more directly assess the temporal relationship between changes in liver ATP and food intake in rats given 2,5-AM. To facilitate the analysis of this relationship, we determined the effect of 2,5-AM on eating behavior and then at a later time measured hepatic ATP in the same animals under similar conditions. In one experiment, the time courses of these responses to 2,5-AM were examined to establish whether the change in liver ATP precedes the eating response. In two other experiments, the eating response was probed at different times after 2,5-AM injection by withholding food for various periods before the intake test to determine whether the behavioral response to the analog waxes and wanes along with the changes in liver ATP levels over time. In addition, adenosine mono- and diphosphate were measured to determine whether stimulation of food intake by 2,5-AM can be attributed to changes in adenine nucleotides other than ATP.

	METHODS AND PROCEDURES

Top Abstract Introduction Methods Results Discussion References

Subjects

Tissue Collection and Extraction

Assays

Experiment 1: Time Course

Fifty-eight rats were assigned to three groups for eventual tissue collection at one of three time points. Each group was then randomly divided into two groups that were injected intraperitoneally with either isotonic saline or 2,5-AM (300 mg/kg in saline, 2 ml/kg) ~2 h after the start of the daylight period. Food intakes (to the nearest 0.1 g corrected for spillage) were measured 15, 45, and 90 min after injections. One week later, rats were reinjected at the same time with vehicle or 2,5-AM and deprived of food, and their livers were removed 15, 45, or 90 min later.

Experiment 2: Delayed Access to Food After 2,5-AM Injection

Two experiments were performed in which the design of the behavioral testing differed slightly. In the first experiment (2a), a between-subjects design was used in which separate groups of rats were tested for their eating response to saline or 2,5-AM after either a 1- or 4-h delay. In the second experiment (2b), a crossover design was used in which each animal was tested for its eating response to saline and 2,5-AM (within-subjects component) at either 1 or 4 h after injection (between-subjects component).

Experiment 2a. Rats (n = 39) were divided into four groups (n = 9-10 each) matched for body weights and food intake. Two groups were deprived of food, injected with either saline or 2,5-AM (300 mg/kg ip) ~2 h into the daylight period, and given food again 1 h later, whereas two other groups were deprived of food, injected with either saline or 2,5-AM, and given food 4 h later. Food intakes were measured 1, 2, and 3 h after food was returned. One week after this behavioral test, rats were deprived of food, reinjected with either saline or 2,5-AM, and killed 1 or 4 h later for collection of liver tissue (i.e., at the time when the earlier food intake tests began).

Experiment 2b. Rats (n = 38) were assigned to two matched groups. In two tests conducted 1 wk apart, rats in the two groups were deprived of food and injected with either saline or 2,5-AM (300 mg/kg ip) in a counterbalanced order across the two test days. One group of rats was refed 1 h after injection, whereas the other group was refed 4 h after injection. Food intakes were measured 1, 2, and 3 h after refeeding. One week after behavioral testing was completed, one-half of the rats in each group were injected with 2,5-AM or saline and then, after either a 1- or 4-h delay without access to food, killed for collection of liver samples (i.e., at the time when the earlier food intake tests began).

Data Presentation and Analysis

Differences in the proportion of rats per group initiating eating behavior was determined with the use of ² analysis. All other data were analyzed by analysis of variance, and differences between individual groups were assessed by post hoc t-tests.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Experiment 1: Time Course

View larger version (18K): [in this window] [in a new window]   Fig. 1.   Effect of 2,5-anhydro-D-mannitol (2,5-AM) on eating behavior. Food intake (A) and percent of animals eating 0.3 g of food or more (B) during different intervals after injection of 2,5-AM (300 mg/kg ip). Food intake values are means ± SE of 29 rats per group. * Significantly different from saline-injected rats, P < 0.01.

Injection of 2,5-AM significantly decreased liver ATP compared with saline injection at all time points [Fig. 2; F(1,47) = 39.8, P < 0.00001]. ATP levels in 2,5-AM-treated rats, which did not differ statistically across time, decreased from those in saline-treated rats by 45, 63, and 64% at 15, 45, and 90 min after injection, respectively. Liver ADP level was significantly lower in rats given 2,5-AM as opposed to saline [Fig. 2; F(1,47) = 4.6, P < 0.05]; however, analysis of data for each time point showed that this decrease in liver ADP was statistically significant only at 45 min after injection. Injection of 2,5-AM had no effect on liver AMP levels at any of the time points examined (Fig. 2).

View larger version (20K): [in this window] [in a new window]   Fig. 2.   Liver adenine nucleotides at different times after injection of 2,5-AM (300 mg/kg ip). A: ATP. B: ADP. C: AMP. Values are means ± SE of 8-10 rats per group. * Significantly different from saline-injected rats, P < 0.05.

The energy charge of liver was lower at all time points in rats given 2,5-AM as compared with those given saline [Table 1; F(1,47) = 40.2, P < 0.00001]; decreases in energy charge of 15, 26, and 26% were observed at 15, 45, and 90 min postinjection, respectively. The ATP-to-ADP ratio in liver was lower in rats treated with 2,5-AM compared with those given saline [Table 1; F(1,47) = 16.6, P < 0.001]. The ATP-to-ADP ratio decreased by 46, 50, and 66% at 15, 45, and 90 min after 2,5-AM injection, respectively, although only the changes at the later two time points were statistically significant.

Experiment 2: Delayed Access to Food After 2,5-AM

View larger version (15K): [in this window] [in a new window]   Fig. 3.   Effect of delayed access to food on the eating response to injection of 2,5-AM. Different groups of rats were injected with 2,5-AM (300 mg/kg) or saline and then given food either 1 or 4 h later. A: food intake over the 3 h after access to food. B: percent of animals eating 0.3 g of food or more in the first hour access to food. No rats given saline ate after a 1-h delay. Food intake values are means ± SE of 9 or 10 rats per group. * Significantly different from saline-injected rats, P < 0.05.

Liver ATP was lower in rats given 2,5-AM compared with those given saline [Fig. 4; F(1,36) = 33.8, P < 0.0001]. Although liver ATP was lower in rats both 1 and 4 h after 2,5-AM compared with those given saline [F(1,36) = 3.0, P = 0.09 for treatment × time postinjection interaction], the decrease in ATP at 4 h after 2,5-AM treatment was approximately one-half that at 1 h postinjection (16 vs. 29%). Liver ATP levels were significantly higher 4 h after 2,5-AM injection than they were at 1 h [F(1,36) = 6.1, P < 0.02]. Because ATP levels were similar in rats given saline and killed at 1 and 4 h, this difference in ATP at 1 and 4 h appeared to be due primarily to a higher level of liver ATP in rats killed 4 h after 2,5-AM injection compared with rats killed 1 h postinjection [t(18) = 2.28, P < 0.05].

View larger version (18K): [in this window] [in a new window]   Fig. 4.   Liver adenine nucleotides at different times after injection of 2,5-AM (300 mg/kg ip). A: ATP. B: ADP. C: AMP. Values are means ± SE of 9 or 10 rats per group. * Significantly different from saline-injected rats, P < 0.05.

Liver ADP levels were significantly higher in rats given 2,5-AM compared with those injected with saline [Fig. 4; F(1,36) = 5.7, P < 0.03]. Although analysis of variance indicated that liver ADP was increased in both groups given 2,5-AM relative to those given saline regardless of the time delay after treatment [F(1,36) = 3.1, P = 0.09 for treatment × time postinjection interaction], the effect of 2,5-AM appeared to be due to an increase in liver ADP levels of 2,5-AM-treated rats given 2,5-AM and killed 4 h later relative to their saline controls [t(18) = 2.71, P < 0.02] as liver ADP was similar in rats given 2,5-AM and saline and killed 1 h later. Liver ADP levels were higher in rats killed 4 h after injection than they were in rats killed 1 h postinjection [t(18) = P < 0.02]. Rats injected with 2,5-AM had higher liver AMP levels compared with those given saline at 4 h, but not at 1 h, after injection [Fig. 4; F(1,36) = 6.4, P < 0.02 for treatment × time postinjection interaction].

Compared with saline treatment, administration of 2,5-AM lowered the hepatic energy charge to a similar degree 1 and 4 h after injection [Table 1; F(1,36) = 18.0, P < 0.0002]. The liver ATP-to-ADP ratio was lower in rats given 2,5-AM than it was in those injected with saline regardless of the time after injection [Table 1; F(1,36) = 26.3, P < 0.0001].

Experiment 2b. Injection of 2,5-AM increased food intake relative to injection of saline when food was withheld for either 1 or 4 h after treatment [Fig. 5; F(1,37) = 19.3, 16.4, and 14.1, P < 0.001 for 1-, 2-, and 3-h cumulative food intake, respectively]. However, during the 3-h observation period, rats given access to food 1 h after injection increased food intake after 2,5-AM treatment relative to saline injection more than those given food after 4 h [42 vs. 113% increase for 1- and 4-h delay, respectively; t(37) = 2.20, P < 0.05]. A greater proportion of rats initiated eating behavior in the first hour of refeeding after 2,5-AM injection than they did after saline treatment regardless of whether access to food was delayed for 1 or 4 h [Fig. 5; ² = 5.2, P < 0.05 and 6.7, P < 0.01 for 1- and 4-h delay].

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Effect of delayed access to food on the eating response to injection of 2,5-AM. Rats were injected with 2,5-AM (300 mg/kg) or saline on two separate days and then given food either 1 or 4 h later. A: food intake over the 3 h after access to food. B: percent of animals eating 0.3 g of food or more in the first hour access to food. Food intake values are means ± SE of 19 or 20 rats per group. * Significantly different from saline-injected rats, P < 0.05.

Injection of 2,5-AM lowered liver ATP levels compared with saline injection only in rats killed at 1 h, but not 4 h, after treatment [Fig. 6; F(1,34) = 7.33, P < 0.05 for treatment × time postinjection interaction]. Liver ADP levels were significantly reduced (by 29%) 1 h after 2,5-AM injection relative to saline injection, but were increased (by 16%) 4 h after 2,5-AM treatment [Fig. 6; F(1,34) = 22.4, P < 0.0001 for treatment × time postinjection interaction]. Liver AMP levels did not differ among the groups (Fig. 6).

View larger version (17K): [in this window] [in a new window]   Fig. 6.   Liver adenine nucleotides at different times after injection of 2,5-AM (300 mg/kg ip). A: ATP. B: ADP. C: AMP. Values are means ± SE of 8-10 rats per group. * Significantly different from saline-injected rats, P < 0.05.

Hepatic energy charge was lower in rats given 2,5-AM as opposed to saline at both 1 and 4 h after injection [Table 1; F(1,34) = 6.8, P < 0.05] and was higher in rats regardless of treatment at 4 h than it was at 1 h after injection [F(1,34) = 4.3, P < 0.05]. The ratio of liver ATP-to-ADP was reduced in rats given 2,5-AM as compared with those given saline regardless of the time after injection [Table 1; F(1,34) = 23.0, P < 0.0001].

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

The results suggest that the eating response to injection of 2,5-AM is associated with a decrease in liver ATP. Measurement of food intake and hepatic adenine nucleotides under similar testing conditions indicated that rats initiate eating behavior in response to 2,5-AM administration after liver ATP levels decline. Furthermore, if rats are denied access to food after 2,5-AM injection, the eating response is triggered as long as ATP remains low, but wanes as liver ATP levels return to normal. Although the behavioral response to 2,5-AM treatment did not track changes in liver AMP, ATP-to-ADP ratio, or energy charge, the results raise the possibility that changes in liver ADP could also provide a signal for food intake after injection of the analog.

A temporal relationship between the increase in food intake and decrease in liver ATP after administration of 2,5-AM was suggested by results of previous studies showing that these two effects occur within the same general time frame (3, 12). A closer examination of the time course of the behavioral and metabolic effects of the analog in the present study confirms that suggestion by showing that the decline in liver ATP precedes the initiation of the eating response. Intraperitoneal injection of 2,5-AM decreased liver ATP within 15 min, whereas the eating response was initiated between 15 and 45 min after treatment. The fact that liver ATP falls before rats initiate eating behavior is consistent with the hypothesis that a decrease in hepatic ATP or some related event plays a causal role in triggering food intake after administration of 2,5-AM (3, 7, 12, 13).

Although rats ate between 15 and 45 min after 2,5-AM treatment, liver ATP levels remained low for at least 90 min when rats were given the analog but not allowed to eat. This suggests that the decrease in ATP triggers eating behavior but does not sustain it. Presumably, other factors may operate to inhibit food intake despite a continuing reduction in liver ATP. However, it is possible that ATP simply remained low because rats were not permitted to eat; if they had eaten, ATP content in liver may have increased. This would suggest that consumption of food after 2,5-AM treatment removes the stimulus to eat by reversing the fall in liver ATP. Additional studies are needed to determine the effect of food intake on the eating response to 2,5-AM injection.

Comparison of the time course of changes in liver ATP, measured in anesthetized rats using ³¹P-NMR spectroscopy, with changes in eating behavior of conscious rats during intravenous infusion of 2,5-AM suggested that the eating response tracks the decline and recovery of liver ATP (see Ref. 3). The changes in food intake and liver ATP measured in the present study after intraperitoneal injection of 2,5-AM followed different time courses than those seen earlier when 2,5-AM was infused intravenously for an hour (3). Presumably, these differences in time course are based on the use of different routes of 2,5-AM administration, although, as indicated above, evaluation of the earlier observations are clouded by the fact that liver ATP and food intake were measured under different experimental conditions. In addition, the relationship between changes in liver ATP and eating behavior implied by the corresponding time courses were not tested in the earlier experiments.

The effect of 2,5-AM on liver ATP in these experiments was assessed in rats that had been treated with the analog during behavioral testing. It is possible, therefore, that carryover effects of 2,5-AM treatment influenced the time course of changes in liver ATP, thereby distorting the temporal relationship between the metabolic and behavioral responses to the analog. Although the effects of multiple 2,5-AM injections on liver ATP have not been examined directly, this seems unlikely. In experiments similar to those reported here, rats injected intraperitoneally for the first time show a substantial reduction in liver ATP by 30 min posttreatment (13), which is consistent with the present findings. In pilot studies conducted as part of the present work, livers from rats that had not been treated with 2,5-AM previously showed decreases in ATP content 90 min after 2,5-AM injection that were comparable to those reported above and, as mentioned above, eventual recovery of liver ATP levels. In addition, measurement of liver ATP using ³¹P-NMR spectroscopy has shown that liver ATP of previously untreated rats declines significantly within 10-20 min after intraperitoneal injection of 2,5-AM at the dose used in the present studies (unpublished observations). It should also be noted that ~60% of 2,5-AM is excreted in urine within 6 h after intraperitoneal injection (24) and that repeated injections of 2,5-AM in the present experiments were given at 1-wk intervals.

In the present studies, as a direct test of whether the eating response to 2,5-AM follows changes in liver ATP concentration, we varied the timing of feeding tests by denying rats access to food for 1 or 4 h after 2,5-AM injection. Rats increased food intake when access to food was delayed for 1 h after 2,5-AM treatment, at which time liver ATP, measured under similar experimental conditions, was substantially reduced. In this regard, the eating response to intraperitoneal injection of 2,5-AM also persists when food is withheld for 90 min (23) when, as shown here, liver ATP remains low. In contrast, when access to food was delayed for 4 h after 2,5-AM injection, by which time liver ATP levels were largely if not completely restored to normal, the eating response to the analog was attenuated or absent. These results showing a relationship between eating behavior and liver ATP as a function of time after 2,5-AM treatment provide further evidence that food intake is stimulated by a decrease in hepatic ATP in rats given the fructose analog.

In contrast to their response to 2,5-AM, rats reliably increase food intake after injection of insulin or the glucose analog 2-deoxy-D-glucose (2-DG), even when they are denied access to food for up to 6 h posttreatment, by which time the effect of these treatments on blood glucose concentration has lapsed (2, 5, 19). It is possible that insulin and 2-DG have different modes of action than 2,5-AM; for example, 2-DG appears to stimulate food intake via its effect on cerebral metabolism (18), whereas 2,5-AM has a peripheral site of action (24). Alternatively, insulin and 2-DG may produce an as yet unidentified long-lasting change in hepatic metabolism (see Ref. 2) that stimulates eating via a mechanism of action similar to that of 2,5-AM.

The amount of phosphorylated 2,5-AM in liver increases during the first 90 min after injection of the analog and then decreases thereafter over the next several hours (24). Because the decline in liver ATP produced by injection of 2,5-AM is due partly to trapping of phosphate in the phosphorylated forms of the analog, the eventual restoration of liver ATP levels after 2,5-AM treatment may be due to dephosphorylation of 2,5-AM or increased uptake of phosphate from the blood. It is also possible that liver ATP levels return to control levels in part because of compensatory changes in fuel metabolism. Administration of 2,5-AM under conditions similar to those used in the present studies produces a shift in metabolism toward fat oxidation (11), which is associated with an increase in the mobilization of endogenous fatty acids (23). This increase in fat fuel utilization apparently serves to meet energy needs because, despite a concomitant decrease in carbohydrate oxidation, whole body oxygen consumption is maintained after injection of 2,5-AM (11). Recent work suggests that the switch toward fat fuel utilization offsets the effect of 2,5-AM on liver ATP (6) and restrains the eating response to the analog (15).

Administration of 2,5-AM produces changes in substrate metabolism that in many respects resemble those seen after a moderate fast (10, 23); for example, the switch to fat oxidation is a hallmark of the metabolic response to food deprivation. The decrease in hepatic energy charge seen after injection of 2,5-AM, which is indicative of a shift toward substrate catabolism in liver (1), is also consistent with the induction of a fasting condition. However, the move toward a catabolic state in liver does not appear to play a causal role in the eating response to 2,5-AM because the decline in energy charge persisted for at least 4 h, after the eating response waned. For the same reason, the shift to fat oxidation after injection of 2,5-AM, which lasts for at least 8 h (11), also does not appear to trigger food intake after administration of the analog. Evidence from this and previous studies (13, 15) indicate that a decrease in liver ATP, rather than to changes in specific metabolic substrates, triggers the eating response to injection of 2,5-AM. Because deprivation of food for only 12 h decreases liver ATP in rats (21), it is possible that both fasting and administration of 2,5-AM stimulate eating behavior by reducing liver ATP.

Injection of 2,5-AM altered the hepatic levels of AMP and ADP in addition to ATP. Liver AMP levels were unaffected by 2,5-AM treatment in the first 90 min postinjection when eating behavior was elicited, suggesting that changes in this nucleotide alone play no direct role in triggering food intake. Changes in liver ADP levels showed a better relationship to the eating response. ADP levels appeared to decline transiently relative to those in saline-injected rats between 15 and 45 min after 2,5-AM administration, during which time food intake was triggered, and were elevated 4 h after 2,5-AM injection when the eating response was attenuated or absent. It cannot be determined from the present experiments whether the apparent transient decline in liver ADP levels occurred before the initiation of the eating response because liver ADP levels decreased during the same interval in which the eating response was observed. It is possible that liver ADP levels may have decreased immediately before rats began to eat during this interval. Liver ADP levels did not differ between saline- and 2,5-AM-injected rats 90 min postinjection even though rats increase food intake in response to 2,5-AM administration when access to food is delayed for this period after injection (23). Although this observation argues against a role for liver ADP levels in the eating response to the analog, further examination of the relationship between ADP and food intake seems warranted.

Adenine nucleotide metabolism is a dynamic process involving complex relationships between cellular concentrations of ATP, ADP, and AMP. Thus, whereas the present findings indicate that the time course of eating behavior after 2,5-AM treatment is tied to changes in liver ATP (and perhaps ADP), it is possible that fluctuations in the relative concentrations of adenine nucleotides may play a part in the behavioral response. The relative concentrations of ATP and ADP are known to have a signaling function (e.g., see Ref. 9) and can exert regulatory control of enzymatic activity (e.g., see Ref. 22). However, in the present experiments, the time course of the eating response after 2,5-AM treatment did not follow changes in the ATP-to-ADP ratio. Similarly, alterations in the hepatic energy charge, which reflects more complex relationships among nucleotides (1), did not show any clear relationship to those in food intake after 2,5-AM treatment. Changes in ATP concentration, production, and metabolism are linked to a variety of cellular processes and to many parameters of cellular energy status. Whether the effect of 2,5-AM on food intake is associated with more subtle interactions between adenine nucleotides or perturbations in some event or process closely related to variations in liver ATP levels remains to be determined.