INTRODUCTION

TOP ABSTRACT INTRODUCTION GENERAL METHODS RESULTS DISCUSSION REFERENCES

IT IS GENERALLY ASSUMED that some direct stimulus consequences of the accumulation of ingested fluid in the gastrointestinal tract play a major role in the termination of a meal. If this were so, then when postingestional stimulation is eliminated or minimized (Sclafani and Nissenbaum, Ref. 9) by the sham feeding procedure, intake should increase immediately to maximum in the first sham feeding test. Contrary to this expectation, all but one of the studies that have measured changes in intake as rats gain sham feeding experience are consistent in showing that it takes three or more consecutive sham feeding tests before intake reaches maximum and stabilizes (2, 3, 5, 7, 10, 11).

Davis and Campbell (2) reported that when rats were given a series of sham feeding tests after stabilization of intake in real feeding tests with either sweetened condensed milk or Vivasorb, a chemically defined diet, it took three to four sham feeding tests before intake reached maximum. After real and sham feeding experience with one of these test solutions, the rats were given real feeding tests with the other. When real intake had stabilized, the rats were then given five sham feeding tests with the second test solution. The rats showed the same pattern of increasing intake with sham feeding experience with the second solution. Davis and Campbell (2) suggested that the progressive increase in sham intake reflected the extinction of a conditioned control of intake acquired in preceding real feeding tests rather than behavioral adaptation to the novelty of sham feeding.

Young et al. (11) reported the same phenomenon in rats sham feeding the chemically defined diet EC116, and Mook et al. (7) observed it when rats were tested with 2 M glucose. Weingarten and Kulikovsky (10) and Davis and Smith (3) reported that rats tested with 0.8 M sucrose in the sham feeding procedure also required three to four sham feeding tests before intake reached maximum. The one exception to this rule of progressive increase in intake with sham feeding experience is the report by Mook and colleagues (7). They reported that when 0.25 M glucose was used as the test stimulus, sham intake was maximum on the first sham feeding test. Because 0.25 M glucose is much less concentrated than any of the other solutions used in the sham feeding studies, this result raised the possibility that the progressive increase in sham intake depends on the concentration of the test solution.

To investigate this possibility, we measured the intake of rats in five consecutive sham feeding tests using a series of dilutions of sweetened condensed milk ranging from very dilute (16:1) to very concentrated (1:2). We used milk as the test solution for two reasons. One was that the original study reporting the phenomenon (2) used one part water to one part milk. The other was that this diet is a commonly used one for studying intake control mechanisms in the rat, so that the results obtained would have more general interest than if we had used a solution containing only a single nutrient.

	GENERAL METHODS

TOP ABSTRACT INTRODUCTION GENERAL METHODS RESULTS DISCUSSION REFERENCES

Subjects. Male albino rats of the Sprague-Dawley strain (200-300 g) obtained from Taconic Farms, NY, were initially assigned to the experiment. They were housed individually in plastic cages and maintained on a 12:12-h light-dark cycle (lights off at 1900) at a temperature of 22 ± 1°C. The rats had ad libitum access to pelleted food (Purina Rat Chow 5001) and tap water, except for a single water deprivation period before testing. All animals were surgically implanted with a gastric fistula according to the method of Schneider et al. (8) and were allowed 2 wk to recover from the surgery before being tested.

Testing procedures. Testing was carried out in 24-cm wide by 20-cm high by 29-cm deep cages. Before surgery the rats were trained to drink from a metal tube centered on the front of the cage, 15 cm above the floor. In the first training session the rats were offered water in the drinking tubes after 17 h of water deprivation. If a rat did not drink from the drinking tube during the 30-min test period, it was tested again on the following day under the same conditions. Most rats learned to approach and drink from the drinking tube in the first test, all by the second. From this point on the rats were tested at 1300 after a 4-h period of food, but not water, deprivation in the testing cages with only the test solution available. After a test, the rats were returned to their home cage where food and water were available ad libitum until 0900 the following day when food was removed from the home cages.

To adapt the rats to the test solution, they were given access to the assigned solution under real feeding conditions in daily 30-min tests until intake stabilized. Intake was judged to be stable when there was a no more than 10% variation in intake over three consecutive 30-min tests. This criterion was typically achieved by the fifth test. In no case were more than seven tests required to achieve it. After adaptation to real feeding, the rats were given five consecutive sham feeding tests.

In all tests, real and sham, a collecting tube, occluded in real feeding and open in sham feeding tests, was attached to the gastric fistula. This collecting tube directed fluid drained from the stomach during the sham feeding tests to a pan located below the testing cage. At the end of each test, the volume of fluid in the pan was measured. If the volume in the pan was <100% of the volume ingested, all the data for that animal were excluded from data analysis.

Rats were assigned to six groups of twelve each for testing with a different dilution of Eagle brand sweetened condensed milk. The dilutions were 16:1, 8:1, 4:1, 2:1, 1:1, 1:2 (distilled water-to-milk ratios). The caloric content of the 1:1 dilution was ~1.7 kcal/ml. Some of the rats were lost during the experiment because of problems related to the implantation and use of the gastric fistula either during or after surgery. Others were rejected because of failure to recover all the fluid ingested from the stomach during sham feeding tests. Of the original 72 rats assigned to the experiment, 13 were eliminated from data analysis for one of these reasons, leaving 59 that completed the real feeding and five sham feeding tests successfully. The final group numbers for the six groups were: 1:2, 12; 1:1, 9; 2:1, 11; 4:1, 7; 8:1, 10; and 16:1, 10.

The group of rats tested with the most concentrated solution (1:2) was given an additional five consecutive sham feeding tests. We did this because sham intake in the first five tests was relatively low, and we wanted to be sure that intake had reached maximum.

At the completion of this experiment, an additional group of 10 rats were tested with the 1:2 dilution in five real followed by five sham feeding tests. This was done because the intake in the real and sham feeding tests with this solution in the original experiment was lower than expected. We wanted to be certain that the results obtained from that group in the original study were not unique to those particular rats.

Data recording and analysis. The test cages were equipped with a computer-controlled lickometer (Henderson, DiLog Instruments, Tallahassee, FL), which passed a current of <60 nA through the rat each time its tongue made contact with the drinking tube. During the test, the time of each tongue contact, to the nearest millisecond, was stored in a data array in the computer memory. At the end of the test, a file of these contact times was created for each rat for later analysis.

Independent-measures ANOVAs were used to analyze the differences in intake across the six groups, and dependent-measures ANOVAs were used to analyze differences across tests within each group. Statistical analysis was carried out with the Systat 7.0 statistical analysis software program, and graphics were created with Sigma Plot for Windows 2.0. Calculation of the microstructure of licking behavior variables was done with the Quick Lick software program (1).

	RESULTS

TOP ABSTRACT INTRODUCTION GENERAL METHODS RESULTS DISCUSSION REFERENCES

Significantly more milk was ingested in the first sham feeding test than in the preceding real feeding test across the concentration range [F(1,53) = 198.5, P < 0.001; Fig. 1]. There was also a significant interaction between intake and the concentration of the test solution [F(5,53) = 5.3, P = 0.001; Fig. 1]. This interaction occurred because intake decreased monotonically over the concentration range in the real feeding tests [F(1,53) = 66.9, P < 0.001; Fig. 1], but not in the first sham feeding tests, where a significant quadratic trend [F(1,53) = 24.3, P < 0.001; Fig. 1] was present. Post hoc analysis showed that the intake of the three lowest concentrations in the first sham feeding tests was a significant increasing linear function of concentration [F(1,24) = 4.9, P = 0.037], but not in the preceding real feeding tests [F(1,24) = 1.7, P = 0.210]. In contrast, the intakes of the three most concentrated solutions in both the real and sham feeding tests changed in parallel; there was a significant linear decrease in intake in both types of tests over this range [real, F(1,29) = 40.3, P < 0.001; sham, F(1,29) = 30.3, P < 0.001].

View larger version (21K): [in this window] [in a new window]   Fig. 1.   Volume ingested as a function of milk dilution for last real and first and fifth sham feeding tests. Data points above label "Rep" on x-axis are from experiment with the 1:2 dilution in 10 additional rats that replicated results from first group of 12 rats (see GENERAL METHODS).

The rats tested with the most concentrated solution (1:2) were given five additional sham feeding tests, because sham intake in the fifth sham feeding test was considerably lower than all but the sham intake of the most dilute solution (Fig. 1). We wanted to be sure that intake had reached asymptote. In fact, sham intake had stabilized by the fifth sham feeding test because, although intake increased slightly from the fifth to tenth sham feeding test (27.3 ± 2.3 to 30.3 ± 1.3 ml), the increase was not statistically significant [t(11) = 1.1, P = 0.300].

The results obtained in the second experiment, which replicated the conditions used in the group tested with the 1:2 solution, were essentially the same as those obtained in the first (Fig. 1). There was no overall statistically significant difference between the two replications in the volume ingested over the six tests (last real and 5 sham feeding tests) [F(1,200) = 1.1, P = 0.315]. The interaction between test and replication was not significant either [F(5,100) = 1.4, P = 0.245]. The magnitude of the increase in intake from the last real to the first sham feeding test in the replication (7.8 ± 0.7 ml) was very close to that obtained in the first experiment (6.3 ± 1.4 ml). The magnitude of the increase in sham intake over the five sham feeding tests in the replication (8.7 ± 2.9 ml) was also a very similar to that obtained in the original study (8.0 ± 3.6 ml).

Overall there was a significant increase in intake with sham feeding experience, because significantly more was ingested in the fifth sham feeding test than in the first [F(1,53) = 8.5, P < 0.001; Fig. 1]. A significant interaction between intake and concentration [F(5,53) = 6.1, P < 0.001; Fig. 1], however, showed that this was not true for all concentrations. The interaction occurred because with the three lowest concentrations there was no significant difference in intake between the first and last sham-tests [F(1,24) = 0.9; Fig. 1], but with the three highest concentrations there was [F(1,29) = 62.4, P < 0.001; Fig. 1]. With the two most concentrated solutions, significantly more milk was ingested in the last than in the first sham feeding test [1:2, t(11) = 6.0, P < 0.001; 1:1, t(8) = 9.2, P < 0.001]. More of the 2:1 dilution was ingested in the fifth than first sham feeding test (Fig. 1), a difference that was statistically significant by a one-tailed test [t(10) = 2.1, P = 0.031] but not quite by a two-tailed test (P = 0.062).

To measure the effect of direct negative feedback derived from the accumulation of the ingested solution in the gastrointestinal tract, we calculated the difference in intake between the first sham and last real feeding tests. The absolute magnitude of this difference depended on the concentration of milk [F(5,53) = 4.1, P = 0.003] and showed a significant inverted "V"-shaped quadratic function of concentration [F(1,53) = 18.7, P < 0.001; Fig. 2, top]. Because the intakes in both types of test varied with milk concentration (Fig. 1) and because the difference between them did as well (Fig. 2, top), we expressed the real intake at each concentration as a percentage of the sham intake of the corresponding concentration (Fig. 2, bottom). Adjusting the real intakes of each test solution in this way showed that, with the exception of the weakest concentration, the suppressing effect of direct postingestional stimulation on real feeding was independent of the concentration of the fluid in the gastrointestinal tract. With the five most concentrated solutions, gastrointestinal filling suppressed intake by 55% with no significant effect of concentration [F(4,44) = 0.2]. With the weakest solution (16:1), the suppression of intake by negative feedback was only 30%, significantly less than the suppression caused by any of the other concentrations (largest Bonferroni adjusted P = 0.025).

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Difference in intake between first and fifth sham and last real feeding test as a function of concentration of milk expressed as absolute difference (top) or as a percentage of intake in last real feeding test (bottom).

The intake of those rats in the groups tested with the three highest concentrations increased with sham feeding testing (Figs. 1, and 2, top). Therefore, to assess the effect of both the direct and labile negative feedback on real intake, it was expressed as a percentage of intake in the fifth sham feeding test (Fig. 2, bottom). Statistical analysis showed that there was a statistically significant effect of concentration [F(5,53) = 7.4, P < 0.001] on this measure. Post hoc analysis showed that real intake as a percentage of sham intake was significantly smaller with the two most concentrated solutions than with the other concentrations [F(1,57) = 32.1, P < 0.001], but this measure of real intake of these two concentrations was not significant [t(19) = 0.13; Fig. 2, bottom].

Davis and Campbell (2) reported that the increase in sham intake that occurred with sham feeding experience was, at least partially, due to an increase in the rate of licking during the early part of the test. To determine if this was true for the three most concentrated solutions, we divided the number of licks made by each rat in the first and fifth sham feeding tests in half. We then analyzed statistically the change in the number of licks from the first to fifth sham feeding tests for both the first and second halves of the tests. With the two most concentrated solutions, the rat made significantly more licks in both halves of the fifth sham feeding test than in the first sham feeding test [1:2, F(1,11) = 20.5, P = 0.001; 1:1, F(1,8) = 14.2, P = 0.005; Fig. 3].

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Number of licks recorded in first half () and second half () in first and fifth sham feeding tests for 3 most concentrated solutions. A, 1:2; B, 1:1; C: 2:1, water:milk. * Statistically significant difference between 2 halves of corresponding sham feeding tests;  statistically significant increase in number of licks from first to fifth test in corresponding half of test.

With the milk diluted 2:1 there was an interaction between tests and the halves of the tests [F(1,10) = 4.8, P = 0.053]. The interaction occurred because, although the number of licks in the first half of the tests did not increase across the sham feeding tests [t(10) = 0.04l Fig. 3], they did in the second [t(10) = 3.8, P = 0.004; Fig. 3]. There were no significant changes from the first to the fifth sham feeding tests in the number of licks in either half of the tests with any of the three lowest concentration test solutions (largest F ratio = 0.829).

With those solutions where there was a significant increase in the number of licks with increasing sham feeding experience, the increase could have been due to either an increase in the size or number of clusters [the size of a cluster is the number of consecutive licks before a pause of 0.5 s (Davis and Smith, Ref. 4)]. To determine which of these alternatives was correct, we calculated the size and number of clusters in the two halves of the first and fifth sham feeding tests with those solutions. With the two most concentrated solutions, the increase in number of licks from the first to fifth sham feeding test in the two halves occurred exclusively because of an increase in the number of clusters (no F ratio was <14.7, which, with 1 and 8 degrees of freedom, had a P value of 0.005). With the most concentrated solution, there was a significant decrease in the size of the clusters in both halves of the test from the first to fifth sham feeding test [F(1,11) = 9.2, P = 0.011; Fig. 4]. The size of the clusters decreased from the first to fifth sham feeding test in both halves of the test with the 1:1 dilution, but was not quite statistically significant [F(1,8) = 4.9, P = 0.057; Fig. 4]. There was no statistically significant change from the first to fifth sham feeding test in either the number of clusters [F(1,10) = 2.3, P = 0.160] or their size [F(1,10) = 2.0, P = 0.190] with the 2:1 dilution (Fig. 4).

View larger version (14K): [in this window] [in a new window]   Fig. 4.   Size of clusters recorded in first half () and second half () in first and fifth sham feeding tests for 3 most concentrated solutions. A, 1:2; B, 1:1; C, 2:1, water:milk.  Statistically significant decrease in size of clusters from first to fifth test in corresponding half of test.

	DISCUSSION

TOP ABSTRACT INTRODUCTION GENERAL METHODS RESULTS DISCUSSION REFERENCES

Intake in the first sham feeding tests was significantly greater than in the preceding real feeding test with all six concentrations of milk (Fig. 1). This demonstrates that, regardless of its concentration over this range, the accumulation of milk in the gastrointestinal tract has a direct inhibitory effect on intake. The absolute magnitude of the increase in intake from the last real to the first sham feeding test depended on the concentration of the test solution. It increased with concentration over the three lowest concentrations and then decreased as the concentration increased still further (Fig. 2, top).

With further sham feeding experience, intake continued to increase in the groups tested with the three most concentrated solutions but not in those groups tested with the three most dilute solutions (Fig. 2, top). Thus the phenomenon of increasing intake with sham feeding experience reported to occur with a variety of different types of test solutions depends, in the case of sweetened condensed milk at least, on its concentration. With the three most concentrated solutions the effect was clearly present; with the three least concentrated solutions the effect was completely absent. Thus the type of negative feedback that controls the intake of milk depends on the concentration of the milk. With weaker concentrations this negative feedback is a function only of direct physiological stimulation derived from the accumulation of ingested fluid in the gastrointestinal tract. With the three most concentrated solutions, this direct physiological signal operates in conjunction with a labile negative feedback signal that disappears with sham feeding experience.

Prior research using a concentrated sucrose solution (0.8 M) has shown that this labile type of negative feedback is learned (2, 3, 5, 10). The changes in intake with sham feeding experience with the three most concentrated solutions used in this experiment show the same pattern as that shown with concentrated sucrose. Thus it is likely that the same explanation, extinction of learned negative feedback, can account for the progressive increase in the sham intake of the concentrated solutions of this experiment as well. However, the composition of the test solution in this experiment, consisting of carbohydrates, fats, and proteins, is much more complex than the sucrose used in the other experiments, so further research will be needed to confirm this hypothesis.

The most dilute solution where the labile type of negative feedback occurred was the 2:1 dilution. Therefore, the threshold for the effect occurred between this concentration and the next more dilute one (4:1). The percentage composition of the three major constituents of these two were 1.4 and 2.3% protein, 10.8 and 18% carbohydrate, and 1.6 and 2.7% fat for the weaker and more concentrated forms, respectively. Carbohydrates made the biggest contribution to the whole, suggesting that they may have played a dominant role in creating the labile form of the negative feedback signal. However, because all three were present in the solution, the three in combination may have been responsible for the effect. Repeating this experiment with the individual components alone can help to resolve this question.

Previous studies reported that the increase in intake with sham feeding experience occurred because the rapid decline in the rate of licking early in the meal typical of real feeding was replaced by a more gradual decline in rate of licking as the rats acquired sham feeding experience (2, 3, 5). This was true for the two most concentrated solutions used in this study. From the first to the fifth sham feeding test, there was a significant increase in the number of licks during the first half of tests with those solutions. A new finding was that the number of licks in the second half of these tests also increased significantly from the first to the fifth sham feeding test. This occurred with all three of the most concentrated solutions, the ones where intake increased with sham feeding experience, so an increase in the rate of licking throughout the test made a significant contribution to the increase in intake.

With the 2:1 dilution, the increase in intake with sham feeding experience was due to an increase in the number of licks in the second rather than the first half of the tests. A ceiling effect may be responsible for the lack of an increase in the rate of licking in the first half of the sham feeding test with sham feeding, because the average number of licks in the first half of this test was ~2,800 licks (Fig. 3). This is about the same as that reached by the rats tested with the 1:1 dilution on the fifth sham feeding test.

With the three most dilute solutions, intake in the sham feeding tests was an increasing function of concentration, showing that increasing the concentration of the milk increased it's effectiveness in stimulating ingestion (Fig. 1). On the contrary, with the three most concentrated solutions, intake in the first sham feeding test was a decreasing function of concentration. For the 2:1 and 1:1 dilutions, the relatively low intakes in the first sham feeding tests were due to the presence of labile negative feedback because by the fifth sham feeding test intake was significantly greater than in the first (Fig. 1). The low intake in all tests of the most concentrated solution was unexpected. This might be expected to be the most effective of all in stimulating intake, but it was not. Even after 10 consecutive sham feeding tests, when any residual conditioned negative feedback must have been extinguished completely, intake was not significantly greater than that recorded in the fifth sham feeding test. Note, as well, that in the replication with this dilution the magnitudes of the intakes in all the tests were approximately the same as those recorded in the original test (Fig. 1). The low intakes of this solution appear to be real rather than an anomaly due to the particular animals assigned to this group.

The nonmonotonic pattern relating sham intake in the fifth test to concentration is similar to that reported by Mindell et al. (6), who studied the relationship of the sham intake of corn oil to its concentration. Sham intake of corn oil in their study was reported to increase monotonically from 0.125 to ~50% and then decrease with further increases in concentration (6). The fat concentration of the most concentrated milk solution used in our study, however, was only ~5%. This is considerably less than the concentration of corn oil, where sham intake began to decline with further increases in concentration. Thus the fat content of the solution is probably not the explanation for the relatively small sham intake of the most concentrated solution in this experiment. An alternative is that the relatively high viscosity of this solution makes it difficult to flow easily from the tube, which in turn, may make the solution somewhat less palatable to the rat. In any event it appears that this very concentrated milk solution is not very effective in stimulating intake, even in the absence of all negative feedback.

An unexpected finding was that, when adjusted for the differences in intake caused by variations in oropharyngeal stimulation, the effect of gastrointestinal stimulation in inhibiting intake was, with the exception of the weakest solution, independent of its concentration (Fig. 2, bottom). Real intake, expressed as a percentage of the intake in the first sham feeding test, was almost exactly 45% of what it would have been in the absence of direct negative feedback. It is remarkable that milk solutions ranging in concentration from 1:8 to 1:2 all had the same relative potency in reducing intake when stimulating the gastrointestinal tract. One might have expected that the more concentrated solutions would have had a greater inhibiting effect on intake than the weaker solutions. On the contrary, for this type of test solution at any rate, direct postingestional stimulation by five of the six solutions produced a 55% inhibition of real intake regardless of the concentration of the milk. The real intake of the weakest solution was reduced by a lesser amount, ~30%.

In summary, the results of this experiment show that the mechanisms that control the ingestion of milk solutions in the rat depend on an interaction between at least three types of stimulus control. These are oropharyngeal stimuli acting to stimulate intake and two different types of negative feedback acting to inhibit it. The control exerted by oropharyngeal stimulation can be seen in the function relating the concentration of the milk to the intake in the fifth sham feeding test. In those tests neither direct physiological nor the labile negative feedback signals could have played any significant role in limiting intake. Therefore, intake had to have been under the sole influence of the oropharynx. The effectiveness of oropharyngeal stimulation to stimulate intake increased linearly over the three weakest solutions. With the next two more concentrated solutions sham intake stabilized, most likely because of a ceiling effect. Ingesting 45 ml in a 30-min test implies an average rate of ingestion of 1.3 ml/min, and, although this is less than the rat's maximum possible rate of 2 ml/min, it is close to it. The low sham intake of the most concentrated solution in the fifth sham feeding test shows that, with this type of test solution at any rate, the effectiveness of oropharyngeal stimulation to stimulate intake does not increase indefinitely with concentration. Whether this is due to mechanics, hedonics, or some other factor is unresolved for now.

It has been known for some time that a labile form of negative feedback exists and plays a role in limiting the intake of some test solutions. It is now clear that this type of negative feedback does not play a role in limiting the intake of the weaker concentrations of milk used in this study. With those solutions the accumulation of ingested fluid alone was sufficient to terminate the meal. Even with the most dilute solution (16:1) where the difference between the last real and fifth sham feeding test was smallest, intake in the real feeding test was still only 70% of that of the sham feeding tests (Fig. 2, bottom).

Thus these results do two things. First, they provide new information about the three controls of intake that can operate during the ingestion of a range of dilutions of milk. Second, they demonstrate the importance of the concentration of nutrients for determining whether a labile, presumably learned, postingestive control occurs.

Perspectives