INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

SEVERAL STUDIES HAVE SHOWN that nutrient infusion into the hepatic portal vein reduces the amount consumed in short-term intake tests, whereas comparable jugular infusions are without effect (e.g., Refs. 3, 21, 24, 25, 31-33). Such findings support the suggestion that the liver contributes to the control of meal size. Little is known, however, about the behavioral mechanisms that underlie the intake effects observed. This issue is addressed in the present study through a detailed behavioral analysis of rats ingesting glucose solution. Analyses of the temporal pattern of spout licking ("lick microstructure") over the course of short-term intake tests have yielded useful inferences about the effects of various treatments on the oral (intake excitatory) and postingestive (intake inhibitory) factors that codetermine the overall meal size (e.g., Refs. 1, 10-12, 18). Thus the initial (e.g., 1st min) licking rate and the average duration of lick bursts during the meal have been taken as indicators of the rat's hedonic response to the taste stimulus (see Refs. 13-15, 28, 34). In addition, the change (decline) in licking rate over the course of the meal has been taken as a correlate of the intake-inhibitory impact of accumulating postingestive load (e.g., Refs. 9, 10, 14, 28).

Results from a study by Campbell and Davis (7; see also Ref. 21) are consistent with the suggestion that the portal infusion effect on meal size is mediated, at least in part, by an action on taste-evaluative processes. They showed that hepatic-portal glucose infusions suppressed the rat's rate of spout licking in a 2-min glucose access test. It seems reasonable to expect then that licking in the initial segment of a complete meal would be similarly affected by hepatic-portal glucose infusion. Here we evaluate this expectation in rats ingesting 12.5% (0.7 M) glucose and determine whether the treatment reduces the mean burst duration, an outcome that would also be consistent with the notion that portal nutrient infusion affects taste evaluation (28). Additionally, the effects of hepatic-portal glucose infusion on the impact of the accumulating postingestive load of an ongoing meal have not been evaluated in detail (21). The statistical analysis of licking over the course of a single meal presented here provides an opportunity to address the respective influences of portal glucose infusion on oral and postingestive signal processing and their relative contributions to the treatment's overall effect on meal size.

The use of lick pattern analysis to interpret the effect of portal nutrient infusion on meal size presumes that an intake effect can be reliably obtained. The literature, however, is inconsistent in this regard. A number of investigators have failed to obtain an intake-suppressive effect of portal nutrient infusion (e.g., Refs. 2, 4, 5, 29, 30, 35, 37, 38). Such failures have been attributed variously to the animal's nutritional status (3, 21, 22), the testing paradigms used (31), and the portal infusion protocol (3, 5). In the present study, we attempted in several ways to maximize the likelihood of obtaining the effect. First, to control for possible infusion carryover effects to the next testing day, we used the design strategy recommended by Tordoff and Friedman (31) whereby mock (no infusion) conditions are run between infusion (saline and glucose) conditions. Second, because of potential interactions between portal nutrient infusion and physiological state we examined both nondeprived (experiment 1) and food-deprived (experiment 2) rats. Finally, we attended to individual rat differences, and found a strong correlate of the sensitivity to portal glucose infusion in the amount consumed under baseline conditions. This correlation between baseline intake and intake suppression after portal glucose infusion may offer insight into the general problem of treatment effectiveness. For this study, it provided a rationale for focusing the lick-microstructure analysis on the subset of rats for which a prominent portal-glucose infusion effect on meal size was obtained.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Surgery

Apparatus

Procedure

Behavioral measures. Session volume intake was measured as the difference between pre- and posttest weight of the fluid reservoir's content (adjusted for g/ml ratio of the fluid tested). The actual average lick volume (drop size) was then calculated (session intake volume divided by total session lick count) for evaluation of system precision and for derivation of meal size and ingestion rate parameters.

Definitions. A licking burst was defined as two or more consecutive licks with no interlick interval exceeding 1 s. Meal onset was defined as the first lick of a burst containing at least 15 licks. This strategy was employed to remove artifactual registrations attributable to manual insertion of the spout or to nonlingual spout contacts. Meal's end was defined as the time of occurrence (after meal onset) of the last lick that preceded the first pause 600 s. (There were no instances of a test session ending before a 600-s pause was encountered.) Meal size (ml) was then calculated by multiplying the number of licks in the meal (all licks from meal onset to a pause 600 s) and the average lick volume for that test session. Likewise, ingestion rate was calculated by multiplying the number of licks in the analysis period (whole meal; single minutes) by the session-average lick volume and dividing by the duration of the analysis period.

Dependent measures. Measures selected for analysis were meal size (ml), meal duration (s), latency to meal onset (s), meal average burst duration (s), number of bursts per meal, meal average ingestion rate (ml/min), and ingestion rate for each minute of the meal. In addition, within-burst lick frequency (licks/s) was taken as the reciprocal of the mode of the interlick interval distribution for the test meal.

Experimental Design

Experiment 2. Experimentally naive rats received a series of eight consecutive daily 90-min test sessions. Portal infusions of glucose or saline (see Experiment 1) were delivered on days 2 and 6, with condition order counterbalanced across rats. Rats were deprived of food for 20.5 h before the portal glucose or saline infusion. Mock infusion tests were conducted on the remaining test days, with ad libitum food access between these tests. Only data from rats that completed all experimental conditions with patent catheter placements (n = 14) were included in the analysis.

Statistical Analysis

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experiment 1: Nondeprived Rats

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Mean (±SE) 12.5% glucose intake after portal saline infusion (solid bar), portal glucose infusion (open bar), and mock test infusions (hatched bars) in nondeprived rats (n = 16).

The extent to which glucose infusion reduced intake relative to saline baseline was positively and significantly correlated with the amount consumed under the saline infusion condition [saline condition intake (ml) × saline condition minus glucose condition intake (ml): Pearson r = 0.85; P < 0.0001; see Fig. 2]. Baseline intake was also significantly correlated with the treatment effect when the effect was expressed as percent difference between infusion conditions [(saline condition minus glucose condition intake) · 100/saline condition intake; Pearson r = 0.66, P < 0.005]. Individual differences in pretest body weight were not a significant correlate of either baseline intake (r = 0.40; NS) or of the difference (ml) in intake between glucose and saline infusion conditions (r = 0.32; NS).

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Mean (±SE) 12.5% glucose intake after portal infusion of saline (solid bar) and glucose (open bar) for nondeprived rats whose baseline intakes fell in upper half of intake range (A; n = 8) and for those falling in lower half of range (B; n = 8). C: scatter plot of intake after portal saline infusion (x-axis) versus magnitude of treatment effect on intake [i.e., difference (ml) between intakes under saline and glucose infusion conditions (glucose  saline); y-axis]. Line of best fit is shown (slope = 0.85 per ml baseline intake; r = 0.85, P < 0.0001).

The relationship between individual differences in baseline intake and degree of intake suppression after glucose infusion prompted us to perform a median split of the overall group based on saline baseline intake (median = 9.46 ml) and contrast results for the "higher-baseline" and "lower-baseline" subgroups. For the higher-baseline subgroup, there was a significant overall effect of condition on amount consumed [F(5,35) = 5.55, P < 0.001; Fig. 2]. The 48% reduction in intake from saline to glucose infusion conditions was significant [F(1,7) = 18.07, P < 0.01]. (Intake under glucose infusion was also significantly different from intakes under any and all mock infusion conditions. There were no significant differences between saline and mock infusion conditions.) For the lower-baseline subgroup, intakes under saline and glucose infusion conditions did not significantly differ [F(1,7) = 1.14, NS]. There was an overall condition effect on intake in this subgroup [F(5,35) = 6.63, P < 0.0001], however, which was largely attributable to somewhat lower intakes under saline and glucose infusion conditions (mean = 7.4 ml) than under mock infusion conditions [mean = 9.2 ml; F(1,7) = 10.12; P <0.02].

As described, the higher- and lower-baseline groups were composed on the basis of individual rat intakes under the saline infusion condition. It is important to add that the individual-subject differences were stable across saline and mock infusion conditions, as shown by a significant coefficient of concordance [Kendall's W = 0.413; ²(15) = 30.98, P < 0.01]. It follows, then, that results similar to those described above should be obtained if the split was made on intakes under any baseline (saline or mock) condition. We explicitly evaluated this suggestion with the grouping based on intake under the first mock infusion test day (i.e., before any portal infusion was delivered) and found a nearly identical pattern of results.

For the higher-baseline subgroup, there was no overall effect of condition on meal duration [F(5,35) = 1.58, NS; Table 1] or on latency [F(5,35) = 1.60, NS; Table 1]. Although meal duration was reduced by 26% under the glucose compared with the saline infusion, the difference was not statistically significant according to the planned comparison [F(1,7) = 0.49, NS]. For the lower-baseline subgroup, the overall ANOVAs and planned comparisons yielded no significant effect on meal duration [F(5,35) = 1.18, NS] or on latency [F(5,35) = 1.66, NS], respectively.

The average rate of ingestion (meal size divided by meal duration) was reduced 25% in the higher-baseline subgroup and was increased 22% in the lower-baseline subgroup under glucose relative to saline infusion conditions (Table 1). These differences, however, were not statistically significant [higher baseline: F(5,35) = 2.53, NS; lower baseline: F(5,35) = 1.62, NS].

Lick microstructure. Analyses of behavioral results for the lower-baseline subgroup yielded no significant effect of treatment condition for any parameter (mean values for the two venous infusion conditions are presented in Table 1). The remainder of this section therefore concerns results obtained in the higher-baseline subgroup.

The ANOVA yielded no significant effect on within-burst lick frequency [F(5,35) = 0.66, NS] or on number of bursts in the meal [F(5,35) = 0.46, NS]. The mean differences between saline and glucose infusion conditions for these two parameters were quite small (Table 1) and not significant according to the respective planned comparisons.

There was a large difference in mean burst duration between the two portal infusion conditions (Table 1). However, significant results were not obtained with the overall ANOVA [F(5,35) = 0.78, NS] or for the planned comparison between infusion conditions [F(1,7) = 3.74, P < 0.12].

First minute lick rate did not vary significantly as a function of test condition [F(5,35) = 0.78, NS] and did not differ between glucose and saline infusion conditions according to the planned comparison [F(1,7) = 0.32, NS]. The ingestion rate (minute-by-minute) curves for glucose and saline conditions, however, did diverge in the subsequent minutes (Fig. 3). Inspection of the group ingestion rate curves (Fig. 3) suggests that the separation, in general, was sustained, but only until about the 12th min of the meal. Such group curves, however, should not be taken at face value insofar as the later descent of the curves reflects, to a large extent, the growing number of rats whose meals have already ended. This subject attrition issue prompted us to limit the ANOVA to the 5-min range, throughout which all rats in both conditions were actively ingesting. The two-way ANOVA (infusion condition × minute) on ingestion rate for the first 5 min of the respective meals showed no overall effect of minute [F(4,28) = 1.95, P = 0.14] but a significant main effect of infusion condition [F(1,7) = 18.02, P < 0.004] and a significant two-factor interaction [F(4,28) = 7.56, P < 0.0003]. Post hoc tests showed that ingestion rate became significantly slower for glucose compared with saline infusion condition by the 3rd min of the meal and remained so for the 4th and 5th min.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Average minute-by-minute ingestion rate for higher-baseline subgroup of experiment 1 after portal infusion of saline () and glucose () infusion. Error bars (±SE) are shown for first 5 min, during which all rats were ingesting. For each subsequent minute, symbol represents mean intake of all 8 rats, and adjacent value indicates number of rats that were ingesting during that minute (i.e., whose meals had not yet ended).

Experiment 2: Deprived Rats

Despite no effect on group-average meal size, glucose infusion did systematically affect meal size in a manner that depended on the amount that individual rats ingested under the saline infusion condition (r = 0.57, P < 0.035; see Fig. 4). The correlation between saline baseline intake and degree of intake suppression under glucose infusion expressed as a percent reduction from baseline was also significant (r = 0.64, P < 0.01). Pretest body weight was not a significant correlate of baseline intake (r = 0.23, NS) or of the difference in intake between glucose and saline infusion conditions (r = 0.12; NS).

View larger version (13K): [in this window] [in a new window]   Fig. 4.   Mean (±SE) 12.5% glucose intake after portal infusion of saline (solid bar) and glucose (open bar) for deprived rats whose baseline intakes fell in upper half of intake range (A; n = 7) and for those falling in lower half of range (B; n = 7). C: scatter plot of intake after portal saline infusion (x-axis) versus magnitude of treatment effect on intake [i.e., difference (ml) between intakes under saline and glucose infusion conditions (glucose  saline); y-axis]. Line of best fit is shown (slope = 0.79 per ml baseline intake; r = 0.57, P <0.035).

The individual differences in intake were stable across saline and the six mock infusion conditions, as shown by a significant coefficient of concordance [Kendall's W = 0.506; ²(13) = 46.05, P < 0.0001].

As in experiment 1, the overall group was divided in half on the basis of intake under the saline infusion condition (median = 15.3 ml). Glucose infusion tended to reduce intake relative to saline values in the higher-baseline subgroup (from 18.57 to 14.54 ml), but tended to increase intake in the lower-baseline subgroup (from 11.84 to 14.33 ml; see Fig. 4). These trends, however, did not achieve statistical significance according to the respective planned comparisons [higher baseline: F(1,6) = 5.60, P < 0.10; lower baseline: F(1,6) = 2.22, P < 0.25].

None of the analyses (whole group ANOVA, ANOVAs for higher- and lower-baseline subgroups, planned comparisons) yielded significant effects on meal duration, latency, average ingestion rate, or lick microstructural parameters.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Hepatic-portal glucose infusion reduced glucose intake in nondeprived rats (experiment 1). The effect was significant for the overall group (24% suppression relative to saline infusion), but the brunt of the effect was carried by a subset of the rats tested. A strong correlate of the degree of intake suppression was the intake magnitude under baseline conditions (Fig. 2). This correlation did not result from a "floor effect" on intake reduction (i.e., rats with small baseline intake could only show a small absolute suppression), because it held when intake suppression was taken as a percentage of baseline intake. We therefore divided the overall group in half on the basis of intake under the saline infusion condition and found that portal glucose infusion reduced intake of rats in the higher-baseline subgroup by 48% after portal glucose infusion. Rats in the lower-baseline subgroup, by contrast, showed no effect of portal glucose infusion on meal size. The individual rat differences in sensitivity to portal nutrient infusion may be relevant to the inconsistent literature on this treatment's intake effects (see discussion below). The outcome of the median split also prompted us to emphasize, for purposes of discussion of the lick-microstructure results, those rats for which a prominent portal glucose infusion effect on intake was obtained.

Behavioral Correlates of Taste Evaluation and Satiation

Another parameter sometimes taken as a sensitive measure of taste evaluation is the average burst duration for the meal (9). Spector et al. (28), for example, showed that increases in initial lick (ingestion) rate as sucrose concentration was raised, reached asymptote at ~0.1 M, whereas average burst duration increased linearly over the full concentration range (0.03-1.0 M) tested. Here we found that burst duration was not significantly reduced by portal glucose infusion. There was, however, a trend in this direction of appreciable magnitude (50% reduction) that, moreover, was observed only in the higher-baseline subgroup of experiment 1. We view this trend as suggestive and therefore hesitate to rule out the possibility that portal nutrient infusion affects taste processing under the present testing conditions.

More clear was an action of portal nutrient infusion on the satiation process. Such judgments are typically derived from differences in the slopes of ingestion rate curves for treatment and control conditions. This was apparent here in the form of a significant interaction between venous infusion condition and ingestion rate over the first 5 min of the meal, due to the greater decline in ingestion rate over minutes when glucose was intraportally delivered. Post hoc comparisons showed that it was not until the 3rd min of the meal that ingestion rates under the two infusion conditions significantly diverged (see Fig. 3), suggesting a synergy between portal nutrient treatment and the intake-inhibitory influence of the accumulating postingestive load.

One may entertain the simple notion that the liver itself is a source of signals that contribute to the termination of an ongoing meal and that the portal nutrient infusion affects intake directly by exaggerating the liver's normal contribution to the satiation process. Our previous work (3), however, suggests that an appropriate interpretation of the hepatic contribution to meal size control is not likely to be so straightforward. We showed, using the intraoral intake model, that portal glucose infusions that begin with the onset of ingestion do not affect the amount consumed. To be effective at limiting meal size, portal infusions had to begin ~1 h before the onset of meal taking. The degree of intake suppression, moreover, did not vary with the amount of glucose infused into the portal vein. Thus the same degree of intake suppression was obtained with a 2-h premeal glucose infusion that was sustained during the intake test as was obtained with a 15-min infusion initiated 1 h before the test. These characteristics contrast with the results from studies where nutrient delivered to or withdrawn from the stomach and/or duodenum affects intake in a graded manner, as a function of the amount infused during the test (e.g., Refs. 8, 16, 17, 36). We argued on the basis of these results that the action(s) of effective portal infusions probably do not reflect the engagement of a primary hepatic satiation mechanism but rather affect the amount that will be ingested in a subsequent meal. In this light, the present data suggest that, via as yet unidentified central and/or peripheral mechanisms, effective portal infusions proactively influence satiating signals or the interpretation of such signals derived from prehepatic sources.

Interaction Between Deprivation State and the Intake Effect of Portal Nutrient Infusion

Others have investigated the ingestive effects of portal nutrient infusion in deprived animals, with conflicting results. Portal nutrient infusion effects have been obtained in the deprived rabbit (22). Nondeprived rabbits, in fact, failed to show the effect (22). A survey of the literature, however, shows that the majority of studies employing deprivation failed to obtain an effect of portal nutrient infusion on intake (e.g., Refs. 2, 4, 5, 29, 37; cf. Refs. 24, 25). These failures may reflect methodological choices (e.g., infusion parameters, test paradigm, etc.; see Ref. 33) or, alternatively, may be related specifically to the metabolic/physiological sequella of food deprivation. Here comparable designs were used in the respective experiments with nondeprived and deprived rats. Our finding that an effect of portal nutrient infusion on average meal size was obtained only in nondeprived rats is consistent with the suggestion that deprivation per se reduced the potential of hepatic-portal nutrient infusion to suppress intake under the current testing conditions.

Although deprivation muted or eliminated the portal nutrient effect for the group taken as a whole, a significant correlation was obtained linking the amount consumed under the saline baseline condition with intake change after treatment. This regression result can be contrasted with that obtained in nondeprived subjects. One can see from Figs. 2 and 4 that the slopes of the respective best-fit lines for deprived and nondeprived rats were quite comparable (respectively, 0.79 and 0.85 ml intake suppression per additional ml baseline intake). The best-fit lines, however, occupied different positions. Their separation can be characterized in terms of the respective points of intersection with the dashed horizontal ("zero suppression") line. For the deprived group, we observe a 7.7 ml rightward shift in the baseline intake point, above which intake is suppressed and below which increased intake is the expected outcome of portal nutrient infusion. An overall treatment effect arose for the nondeprived group, where the distribution of the baseline intakes was clearly biased to the right of the point of intersection. For the deprived rats, the distribution of baseline intakes was more balanced across the intersection point, resulting in no net effect of portal nutrient infusion.

The individual difference correlations suggest that a common factor or factors underlie both the sensitivity to portal nutrient infusion and the tendency of individuals to ingest either more or less glucose during baseline tests. The identity of the relevant factor, however, is a matter of speculation. We know that body weight is not a relevant variable, as it was not a significant correlate of either baseline intake or sensitivity to the treatment. Variables that may merit attention include baseline insulin level, and insulin response to portal glucose infusion and to ingested glucose (see Refs. 6, 23). Portal nutrient infusion triggers a cascade of events [including changes in portal glucose and insulin levels, net hepatic glucose uptake, glycogen synthesis, hepatic oxidative metabolism, and hepatic-vagal afferent activity (e.g., Refs. 20, 23, 26, 32)], some unknown subset of which must be relevant to the intake response. When the relevant variables are identified we will be in a better position to address the individual differences in response to portal nutrient infusion, the influence of deprivation, and the interaction of these factors and the central processes by which taste and postingestive signals are integrated to affect meal size.