INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE ADIPOSTAT HYPOTHESIS was formulated over 40 years ago to explain control of body weight and food intake (30), and, later, it was invoked to explain the link between body fat content and reproductive function (24, 31). With regard to reproduction, the adipostat hypothesis stems from the observation that body fat content is often positively correlated with measures of pubertal development in women, domestic animals, laboratory rats, and most other species (23). Based solely on these correlations, it was hypothesized that the attainment of a particular level of adiposity generates some sort of signal that permits normal reproductive function. When levels of adiposity fall below a critical threshold, a hypothetical signal is detected and purportedly sent to the hypothalamic-pituitary-gonadal system, thereby inhibiting ovulatory cycles. Similarly, it has been hypothesized that signals originating in white adipose tissue "regulate" food intake in service of an optimal body weight and body fat content (30).

An important component of the adipostatic model is a hormonal or neural signal that accurately reflects the level of adiposity and acts on targets in the hypothalamic-pituitary-gonadal system. For many years, it was commonly held that the critical signal for the initiation of puberty was estradiol generated by changes in the aromatization of testosterone in adipocytes (23). Substantiation of this idea would require a plausible neuroendocrine mechanism linking adipocyte estradiol synthesis to the hypothalamic-pituitary-gonadal system. More recently, the pancreatic hormone insulin was a candidate for the adipostatic signal (3, 29, 40, 41, 57, 68). In the last three years, a vast number of studies have focused on the role of leptin, the ob protein secreted predominantly by white adipose tissue cells (70). Plasma leptin concentrations are positively correlated with adiposity in human beings and rodents (2, 18, 37), and leptin treatment can reverse the obesity and infertility characteristic of ob/ob mice, mice homozygous for a mutant ob allele (13, 25, 46, 66). Systemic treatment with leptin results in decreased food intake and body weight and reverses the effects of underfeeding on several aspects of reproduction in both lean and obese laboratory rodents (1, 4, 15, 16, 69). A small number of adult human subjects with mutations of the ob gene were identified and found to be markedly obese and to have reproductive abnormalities (42, 61).

The obesity and infertility in animals homozygous for the mutant ob allele provide evidence that at least some leptin is required for normal energy balance and reproduction. However, data from animals lacking leptin cannot provide strong evidence that changes in leptin levels within physiological ranges control food intake and reproduction in populations of wild-type organisms. Despite this fact, leptin is most often depicted as an adipostatic signal (13, 15, 16, 18, 19, 46). The adipostatic hypothesis is either stated explicitly in words or implied by illustrations that show arrows coming from white adipose tissue directly to the brain and then leading to "food intake" or "reproduction" (e.g., see Ref. 19). Two specific implications of the adipostatic hypothesis are that 1) plasma leptin levels accurately reflect body fat content and 2) decreased plasma leptin concentrations characteristic of lean fasted animals necessarily result in increased food intake and inhibited reproductive processes. The present experiments are designed to test these assumptions using Syrian hamsters as a model system.

Syrian hamsters are unusual in that they fail to show compensatory hyperphagia after fasting (58). According to the adipostatic hypothesis, lack of compensatory hyperphagia might result from an attenuated leptin response to fasting (i.e., a failure of leptin levels to drop after fasting). Another characteristic of this species is the occurrence of fasting-induced anestrus. A 48-h period of fasting during days 1 and 2 of the estrous cycle blocks the next expected period of estrus, ovulation, and the underlying neuroendocrine processes (43, 44, 54). Anestrus occurs when the 48-h period of fasting occurs on days 1 and 2 of the estrous cycle but not when fasting occurs on days 2 and 3, 3 and 4, or 4 and 1 (44). Fasting-induced anestrus occurs in hamsters with a low but not a high body fat content (54). According to the adipostatic hypothesis, fat hamsters would be protected from fasting-induced anestrus due to high leptin or insulin levels as a result of their high body fat content. To test predictions about the adipostatic hypothesis of reproduction and food intake, we examined plasma leptin and insulin concentrations at time points during the fasting period in same-aged hamsters that were either fat or lean before fasting. In addition, plasma hormone levels were measured in groups of fasted hamsters at various time points after refeeding.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Animals and housing. Animal care and use was in accordance with guidelines of the United States Department of Agriculture, National Institutes of Health, and the Lehigh University Institutional Animal Care and Use Committee. Adult female Syrian hamsters, Lak:LVG, were purchased from Charles River Breeding Laboratories (Wilmington, MA). They were individually housed in polypropylene cages with wire lids in a room maintained at 22°C on a 16:8-h light-dark cycle with "lights on" at 0700 and were fed Purina Rodent Chow (no. 5001) in either pelleted (lean hamsters) or powdered (fat hamsters) form for 2-4 wk before blood sampling.

Blood sampling and hormone assays. In experiment 2, hamsters received an overdose of pentobarbital sodium (100 mg/kg body wt) injected intraperitoneally. Approximately 1.5 ml blood was drawn by cardiac puncture with heparinized syringes, and the carcasses were processed for analysis of body composition. In experiment 3, for repeated blood sampling, hamsters were lightly anesthetized using metofane inhalant or intraperitoneal injections of pentobarbital sodium (50-70 mg/kg body wt). Approximately ~0.5- to 1.0-ml blood samples were drawn by cardiac puncture with heparinized syringes and stored in EDTA-coated tubes. After centrifugation, leptin and insulin concentrations were determined by RIA from duplicate 100-µl plasma samples using the I¹²⁵ Multispecies Leptin Kit and Rat Insulin Kit from Linco Research (St. Charles, MO). The multispecies leptin assay is highly specific, cross-reacting at <1% with hormones other than leptin. Coefficients of variation were 10% or less, and intra-assay variation was <1%.

Carcass analysis. After blood sampling, hamsters were overdosed with pentobarbital sodium (100 mg/kg body wt), and retroperitoneal and parametrial fat pads were excised, weighed, and returned to the carcasses. The carcasses were analyzed for water, fat, and lean body mass as described previously (52, 53). Briefly, viscera were removed, and preweighed carcasses were baked at 70°C until they reached a consistent weight. Water, lipid, and fat-free dry mass were determined by a modification of the procedure of Leshner et al. (33).

Statistical analysis. In experiment 1, the number of hamsters showing 4-day estrous cycles was analyzed with the R X C test for independence using the G statistic (similar to a chi square test; see Ref. 60). In experiment 2, percent body fat was analyzed by a three-way ANOVA, where the three factors were time (12, 24, 36, and 48 h after the start of fasting), body composition (fat or lean), and food (fasted or fed). When main effects were significant, ANOVA was followed by paired comparisons of fat/fasted vs. fat/fed, fat/fasted vs. lean/fasted, and lean/fasted vs. lean/fed. These comparisons were made within each time point. For example, fat/fasted was compared with lean/fasted within the 12-h time point. Data for plasma leptin and plasma insulin displayed unequal variances (P < 0.01) and significant mean-variance correlations (P < 0.01); thus, these data were transformed to the log scale before analysis (60). Log-transformed scores met all of the assumptions of the ANOVA. For the same reasons, ANOVA for body fat content and fat pad weights was performed on square root-transformed scores. All data are graphed in the linear scale for ease of presentation. In a separate analysis, data from all of the fed (fat and lean) groups were pooled to examine the statistical correlation between body fat levels and plasma hormone levels. In experiment 3, one-way ANOVA was used to compare hormone levels of the fed, fasted, and refed groups within each time point, followed by Student-Newman-Keuls post hoc comparisons when the main effects were significant.

Experiment 1. Fasting-induced anestrus occurs in hamsters fasted on days 1 and 2 of the cycle but not in hamsters fasted on days 2 and 3, 3 and 4, or 4 and 1 (44). This suggests that fasting-induced anestrus requires a stimulus that occurs some time between the beginning of day 2 and persists until the end of day 2 of the estrous cycle. It has been presumed that this stimulus inhibits pulsatile luteinizing hormone (LH) secretion and that prolonged inhibition of LH pulses inhibits follicle development. A pilot experiment examined the duration of fasting necessary to induce anestrus. Twenty estrous-cycling hamsters that were between 85 and 92 g in body weight were divided into three groups (n = 6 or 7/group) that were fasted for either 24, 36, or 48 h beginning 0700 on day 1 of the estrous cycle. The 48-h-fasted group received food on 0700 on day 3. This experimental paradigm is similar to that used in our previous work on metabolic control of estrous cyclicity (54). To examine the importance of having food during the evening of day 4 of the estrous cycle, 40 hamsters with body weights of 85-99 g were divided into four groups (n = 5-12/group) that did not differ significantly in body weight and that were fasted for either 24, 36, 48, or 60 h beginning at 2200 on day 4 of the estrous cycle. The 60-h fast ended on the morning of day 3 of the estrous cycle and corresponds to the fasting period used in previous experiments on the effects of leptin treatment (51). All hamsters were checked for estrous behavior on the evening of day 4 of the next estrous cycle and for vaginal discharge on the morning of day 1 of the next expected cycle.

Previous work showed that leptin treatment during the fasting period on days 1 and 2 of the estrous cycle prevented fasting-induced anestrus (51). In the present experiment, we examined whether increased levels of plasma leptin on days 3 and 4 of the cycle are important for normal estrous cyclicity. Hamsters were fasted and treated intraperitoneally with either leptin (Peprotech, 5 mg/kg body wt) or the phosphate buffer vehicle (pH = 7.5) two times per day at 0800 and 2000 on days 1 and 2 of the cycle. A third group was fasted on days 1 and 2 and was treated with the same dose of leptin two times per day on days 3 and 4 of the estrous cycle.

Experiment 2. Experiment 2 was designed to examine the natural endogenous changes in plasma insulin and leptin that occur during fasting in fat and lean hamsters. The experiment began just before the onset of the dark period on day 4 of the estrous cycle (2300). Two groups of hamsters were fed diets known to result in either a high or low body fat content (54). One-half of the lean (82-95 g body wt) and one-half of the fat hamsters (130-160 g in body wt) were either fasted or fed. The fasted groups were placed in a clean cage without food, whereas the fed groups were placed in a clean cage with ad libitum access to chow pellets inside the cage. Hamsters from each group received an overdose of pentobarbital sodium at either 12, 24, 36, or 48 h after the start of fasting. Approximately 1.5 ml blood were drawn, and the carcasses were processed for composition analysis. Thus, at each of the four time points, there were four groups of hamsters: lean/fasted, fat/fasted, lean/fed, and fat/fed with 7-12 hamsters in each of these 16 groups. The experiment was conducted in two replicates. The first replicate included hamsters from the 12- and 24-h group. The second replicate included hamsters from the 36- and 48-h group.

Experiment 3. In experiment 2, differences in plasma leptin did not accurately reflect body fat content in fasted hamsters. Thus it seems that the plasma leptin level does not serve as a signal of long-term energy storage. An alternative possibility is that plasma leptin levels serve as signals for short-term changes in energy availability. For example, plasma leptin concentration might increase rapidly in response to meals. If so, meal-induced increases in plasma leptin levels would result in normal estrous cyclicity in hamsters fasted for 24 h and then refed. Short-term increases in plasma hormones in response to meals might not be detectable in the lean/fed hamsters from experiment 2 because blood sampling was not fixed with regard to meal times. The present experiment examined the time course of leptin responsiveness to fasting and refeeding.

Sixteen hamsters (90-110 g body wt) were placed in a clean cage without food and were fasted for 24 h. Next, one-half of the hamsters received chow pellets inside their cage while the others continued to fast. Blood samples were drawn from lightly anesthetized hamsters at 3, 6, and 12 h after the food was placed in the cages of the fed group. Plasma was analyzed for insulin and leptin concentrations. Hormone assays and statistical analyses were carried out as in experiment 2.

In the above-mentioned experiments, we were surprised to find no significant increase in plasma leptin at 12 h after refeeding. Thus we repeated the experiment, this time taking blood samples 12 and 24 h after the start of refeeding. Fasted and fed controls were also included. Twenty-one hamsters (82-104 g body wt) were either fasted for 48 h, fasted for 24 h and refed for 24 h, or had access to food ad libitum for 48 h. The fasting began on the morning of estrous cycle day 1. Blood samples were drawn from lightly anesthetized hamsters at 12 and 24 h after the food was placed in the cages of the refed group. A separate experiment was run contemporaneously to ensure that there was no significant effect of repeated blood sampling on plasma leptin concentrations at the latest time point. Fifteen hamsters (80-99 g body wt) were either fasted for 48 h, fasted for 24 h and refed for 24 h, or had access to food ad libitum for 48 h. Blood samples (~1.5 ml) were drawn only one time at the end of the 48-h feeding or fasting period (in the fed and fasted group) or 24 h after refeeding (in the refed group). Plasma was assayed for leptin concentrations as described in experiment 1.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Experiment 1. When fasting began on the morning of day 1 of the estrous cycle, 88% of hamsters that were fasted for 48 h from the morning of day 1 until the morning of day 3 became anestrus (Fig. 1). In contrast, none of the hamsters fasted from the morning of day 1 for 24 or 36 h became anestrous, and this difference was significant (P < 0.01). When fasting began earlier, on the evening of day 4 of the cycle, again the hamsters that received their food back on the morning of day 3 were the only group susceptible to fasting-induced anestrus (those fasted for 60 h; Fig. 1).

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Percent of lean hamsters showing 4-day estrous cycles in groups fasted for different durations beginning on either the morning (lights on) of day 1 of the estrous cycle (A) or beginning on the evening (lights out) of day 4 of the previous estrous cycle (B).

Leptin treatment (5 mg/kg body wt by ip injection) two times per day during fasting on days 1 and 2 of the estrous cycle prevented fasting-induced anestrus, as has been shown previously (51). In contrast, the same leptin treatment on days 3 and 4 of the estrous cycle did not prevent the effects of fasting on days 1 and 2 of the cycle (Fig. 2).

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Percent of lean hamsters showing 4-day estrous cycles in groups fasted on days 1 and 2 of the estrous cycle and treated with either vehicle on days 1 and 2, leptin on days 1 and 2, or leptin on days 3 and 4 of the estrous cycle. Leptin treatment prevented fasting-induced anestrus when given during fasting on days 1 and 2 but not when given on days 3 and 4 of the estrous cycle.

Experiment 2. Percent carcass lipid was positively correlated with plasma leptin concentrations (r = 0.828, P < 0.001, Fig. 3) and plasma insulin concentrations (r = 0.667, P < 0.001, Fig. 3) in Syrian hamsters fed ad libitum. The three-way ANOVA for plasma leptin and insulin levels showed significant main effects of time, body composition, and feeding (all P < 0.01), as well as significant interactions: time × body composition (P < 0.01) and body composition × feeding (P < 0.05). Planned comparisons showed that, in fasted hamsters, detectable decreases in plasma leptin were apparent long before decreases in body fat content (Fig. 4A and Table 1). Beginning 12 h after the start of fasting, plasma leptin and insulin concentrations in the fat/fasted hamsters were significantly lower than those of the fat/fed hamsters (P < 0.001), although indexes of body fat content did not differ significantly between these two groups (Fig. 4A and Table 1). By 36 h after the start of fasting, the plasma leptin and insulin concentrations in fat/fasted hamsters fell almost to the level of that shown by lean/fasted hamsters, with no significant change in adiposity (Fig. 5A and Table 2). In fat/fasted hamsters, plasma leptin concentrations dropped by an average of 64% at 12 h and by an average of 77% at 24 h (Fig. 4). Plasma insulin concentrations dropped by an average of 54 and 58% at 12 and 24 h, respectively (Fig. 4). Despite the dramatic drop in insulin and leptin levels, fat/fasted hamsters had slightly but significantly higher levels of these hormones than lean/fasted hamsters at all time points (Figs. 4 and 5, P < 0.05).

View larger version (12K): [in this window] [in a new window]   Fig. 3.   Plasma leptin (A) and insulin (B) plotted as a function of percent carcass lipid in Syrian hamsters with ad libitum access to Purina Laboratory Rodent Chow pellets (no. 5001).

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Mean ± SE percent carcass lipid, plasma insulin (ng/ml), and plasma leptin concentration (ng/ml human equivalent) in four groups of hamsters: 1) fat/fasted hamsters were 130-160 g body weight before the start of the experiment and were fasted on days 1 and 2 of the estrous cycle, 2) fat/fed hamsters were fat hamsters with ad libitum access to food during the same time period, 3) lean/fasted hamsters were 82-95 g and fasted, and 4) lean/fed were lean hamsters that had ad libitum access to food throughout the same time period. In A, the fasting period was for 12 h, and, in B, the fasting period was for 24 h. A Significantly different from fat/fed; B significantly different from fat/fasted; C significantly different from lean/fed at P < 0.05.

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Mean ± SE percent carcass lipid, plasma insulin (ng/ml), and plasma leptin concentration (ng/ml human equivalent) in four groups of hamsters: 1) fat/fasted hamsters were 130-160 g body weight before the start of the experiment and were fasted on days 1 and 2 of the estrous cycle, 2) fat/fed hamsters were fat hamsters with ad libitum access to food during the same time period, 3) lean/fasted hamsters were 82-95 g and fasted, and 4) lean/fed were lean hamsters that had ad libitum access to food throughout the same time period. In A, the fasting period was for 36 h, and, in B, the fasting period was for 48 h. A Significantly different from fat/fed; B significantly different from fat/fasted; C significantly different from lean/fed at P < 0.05.

In contrast to fat/fasted hamsters, lean/fasted hamsters actually lost body fat, and leptin levels decreased more slowly. At 12 h, differences in body fat content and plasma leptin were not significant (Fig. 4A). However, at 24 h, lean/fasted hamsters showed a 41.6% decrease in body fat content and a 30% drop in plasma leptin concentrations relative to the lean/fed hamsters, and these decreases were significant (P < 0.03; Fig. 4A). Plasma insulin concentrations mirrored food intake more closely than they reflected adiposity but continued to decrease to the lowest levels in lean/fasted hamsters (Fig. 4). Plasma insulin concentrations were significantly lower in lean/fasted hamsters compared with all other groups at all time points (P < 0.001).

Experiment 3. Fasted, refed hamsters ate an average of 3.4 g of chow before the first blood sampling and ate between 0.1 and 1 g thereafter (data not shown). Plasma insulin secretion increased significantly by 12 h after the start of refeeding (P < 0.001, Fig. 6B). There were no significant differences in plasma leptin concentration between fasted and refed hamsters at 3, 6, or 12 h after the start of refeeding (Fig. 6A).

View larger version (19K): [in this window] [in a new window]   Fig. 6.   Mean ± SE plasma leptin concentration (A) and plasma insulin concentrations (B) in lean adult female Syrian hamsters fed ad libitum (ad lib; hatched bars) or fasted for 24 h and then either refed for 24 h (filled bars) or fasted for 48 h (open bars). Blood samples were taken by cardiac puncture in lightly anesthetized hamsters at 3, 6, and 12 h after the start of refeeding. There were no significant differences between fasted and refed Syrian hamsters in plasma leptin levels. Insulin levels in refed hamsters were significantly higher than those of fasted hamsters by 12 h after the start of refeeding (* P < 0.01).

When this experiment was repeated and plasma leptin levels were sampled at 12 or 24 h after the start of refeeding, again, plasma leptin concentrations were not higher in refed compared with fasted hamsters at either time point (Fig. 7A). The unusually high variance in plasma leptin concentrations in the fed group at the 24-h time point in Fig. 7A can be accounted for by one hamster with a value of 18 ng/ml. In the final experiment, when blood was sampled only one time, 24 h after refeeding (48 h after fasting), the one-way ANOVA showed a significant effect of treatment (P < 0.05). Fed hamsters showed a trend toward a significantly higher level of plasma leptin than fasted hamsters (P < 0.056), and, again, there was no significant difference in plasma leptin concentration between fasted and refed hamsters which ate 8.6 ± 0.7 g (Fig. 7B).

View larger version (19K): [in this window] [in a new window]   Fig. 7.   Mean ± SE plasma leptin concentration in lean adult female Syrian hamsters fed ad libitum (filled bars) or fasted for 24 h and then either refed for 24 h (hatched bars) or fasted for 48 h (open bars). A: repeated blood samples were taken by cardiac puncture. B: one blood sample was taken by puncture in lightly anesthetized hamsters at 24 h after the start of refeeding. There were no significant differences between fasted and refed Syrian hamsters in plasma leptin levels.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The first important result of the present study was the demonstration that plasma leptin and insulin concentrations were not a simple function of adiposity. In fat hamsters, plasma leptin and insulin concentrations plummeted 12 h after the start of fasting, with no significant changes in body fat content. Similar results were seen when ob mRNA was measured in fasted male Syrian hamsters. Adipose tissue ob mRNA levels in hamsters fasted for 48 h were significantly lower than in hamsters fed ad libitum, although differences in body fat content were small (39). Thus transcription of the leptin gene (39) and plasma leptin and insulin concentrations are more closely associated with short-term energy balance than with long-term changes in adiposity, as has been suggested by others (48, 62). The first assumption of the adipostatic hypothesis, that plasma leptin and insulin concentrations accurately mirror body fat content, is not supported by data from fat/fasted Syrian hamsters (Figs. 4 and 5), lean rats, lean mice, and human subjects (2, 5, 7, 26, 32, 50, 62).

Second, the lack of postfast hyperphagia in Syrian hamsters cannot be explained by a failure of this species to show fasting-induced decreases in either ob mRNA (39) or plasma leptin concentrations (Fig. 4). Instead, some investigators have attributed the lack of postfast hyperphagia to the failure of neurons in the arcuate nucleus of the hypothalamus to secrete neuropeptide Y in response to fasting (39). Furthermore, because fat hamsters fail to show fasting-induced anestrus, it appears that a dramatic drop in plasma leptin and insulin is not a sufficient condition for anestrus. Thus decreases in plasma insulin and leptin concentration and ob mRNA do not provide a "starvation" signal that increases food intake and inhibits reproduction in Syrian hamsters.

The lean/fasted group showed a small significant decrease in body fat content and plasma leptin concentrations relative to the groups that would be expected to show normal estrous cycles (Figs. 4 and 5). Plasma leptin levels dropped only 30%, and a significant decrease relative to fed controls was not apparent until 24 h after fasting (Fig. 4). Plasma insulin levels dropped more rapidly than plasma leptin levels, and levels of these hormones were significantly lower than those of lean/fed hamsters throughout days 1 and 2 of the estrous cycle. These results are in keeping with the adipostatic hypothesis in that insulin and leptin levels were lowest in the only group that was expected to become anestrus. The adipostatic hypothesis did not stand up to further scrutiny in experiment 3, however.

The third important finding of these experiments was that elevated plasma leptin concentrations (higher than those of fasted hamsters) were not necessary for normal estrous cycles. Forty-eight-hour-fasted and 24-hour-fasted hamsters did not differ significantly in plasma leptin levels on days 1 and 2 of the estrous cycle (Figs. 6 and 7), and yet 48-h-fasted hamsters became anestrus, whereas 24-h-fasted hamsters showed normal estrous cycles (Fig. 1). The lack of increase in plasma leptin concentrations after refeeding cannot be attributed to the effects of repeated sampling on leptin synthesis and secretion because, in the last experiment, blood samples were taken only one time (Fig. 7B). Again, plasma leptin concentrations in refed hamsters were not significantly higher than those of fasted hamsters at 24 h after refeeding (48 h after fasting, Fig. 7B).

A difference in plasma leptin levels between 24-h-fasted and 48-h-fasted hamsters might have occurred later, e.g., some time on days 3 and 4 of the estrous cycle. Even if this were true, these elevated plasma leptin concentrations on days 3 and 4 of the estrous cycle are probably not critical for normal estrous cyclicity, because leptin treatment during days 3 and 4 of the cycle did not prevent anestrus in hamsters fasted on days 1 and 2 of the estrous cycle (Fig. 2). Thus levels of plasma leptin typical of the lean fasted hamster do not necessarily serve as a signal mediating fasting-induced anestrus in Syrian hamsters.

Pulsatile LH secretion is inhibited by fasting in most species studied. LH pulses resume within a few hours of a single meal or within minutes of nutrient infusion, far more rapidly than the changes in adiposity, plasma insulin, and plasma leptin found in the present experiments (8-12, 17, 56). The time course of LH secretion in fasted and refed Syrian hamsters is unknown at this time. Hamsters might show rapid restoration of LH pulses as do other rodents and nonhuman primates, or they may be slow to reinstate LH pulses after refeeding, as are human females (35). Because leptin synthesis and secretion is thought to be controlled by the availability of oxidizable metabolic fuels (69), it is reasonable to hypothesize that the slow restoration of plasma leptin concentration in hamsters might be due to the lack of postfast hyperphagia in this species (58). This failure to increase food intake to compensate for a period of fasting, in turn, might be expected to delay the restoration of energy balance and LH secretion. Even so, elevated plasma leptin concentrations (higher than those of fasted hamsters) cannot be necessary for LH pulse frequency and amplitude sufficient for normal follicle development and ovarian steroid secretion. The fasted-refed group showed normal estrous cyclicity even though their leptin levels were no different than those of 48-h-fasted hamsters.

At least one caveat remains. It is possible that significant changes in leptin secretion occur in a pulsatile manner that can only be revealed by more frequent sampling. These pulses would not have to be very large to reach the plasma leptin level of lean/fed hamsters that show normal estrous cycles. Another possibility is that there are changes in leptin secretion or receptor tissues such as brain, liver, or muscle that control reproduction without concomitant changes in plasma leptin concentrations (34, 65).

The time course of changes in insulin in the present study was consistent with a role for this hormone in control of estrous cycles (Figs. 4 and 5). A number of investigators have suggested that insulin is an important signal in control of ovulatory cycles (40, 41). Not all of the data support this idea, however. The rapid meal-induced increases in insulin found in most species are not necessary for the meal-induced increases in pulsatile LH secretion (67). Further work is necessary to show whether meal-induced restoration of reproduction requires meal-induced pulses in insulin in this species.

In summary, rapid decreases in plasma leptin and insulin in fat/fasted hamsters were not associated with anestrus or with fasting-induced hyperphagia. Fasting-induced anestrus in lean Syrian hamsters was not associated with dramatic decreases in plasma leptin levels relative to lean/fed controls. Finally, and most important, elevated plasma leptin concentrations (higher than those of fasted hamsters) were not necessary for normal estrous cycles.

Perspectives

If hormones such as insulin and leptin are not the critical signals controlling food intake and estrous cyclicity, what are the critical signals? It has been demonstrated in a wide variety of animals that changes in metabolic fuel disposition occur far more rapidly than changes in adiposity, plasma insulin, and plasma leptin levels seen in Fig. 2 (6, 38, 47, 49, 59). In fact, changes from fasting to storage modes of metabolism occur rapidly after refeeding, and the longer the fasting period, the more rapid the metabolic switch occurs. For example, in fasted Syrian hamsters, fasting-induced increases in circulating free fatty acids and ketone bodies are reversed, and lipogenesis in liver is significantly increased within 3 h of refeeding after a 24- or 48-h fast (49), 6 h before the increase in insulin seen in Fig. 6B. In fact, we have seen plasma ketone bodies and free fatty acid levels return to fed levels by 1.5 h after refeeding (Blum and Schneider, unpublished data). This time course dovetails with the time course of LH secretion after refeeding in other species. The signal to reinstate estrous cyclicity may result from increases in metabolic fuel availability and oxidation.

Any similarity in time course of the onset of estrous cyclicity and the changes in disposition of metabolic fuels provides evidence of correlation, not causation. However, there is abundant direct experimental evidence demonstrating that treatments that decrease the availability and oxidation of specific metabolic fuels inhibit estrous cycles in Syrian hamsters (54, and, for review, see Refs. 55, 63, and 64). For example, we have shown that anestrus is induced in fat/fasted hamsters treated during fasting with methylpalmoxirate, a pharmacological agent that blocks fatty acid oxidation, or with 2-deoxy-D-glucose, a glucose analog that blocks glucose oxidation (54). These results demonstrated unequivocally that a high body fat content is not a sufficient condition for normal estrous cyclicity and provided strong evidence for the idea that estrous cyclicity is controlled by the availability and oxidation of metabolic fuels. Exogenous leptin treatment is known to increase metabolic fuel availability and oxidation and thus may prevent fasting-induced anestrus via indirect effects on fuel metabolism (51).

It is important to differentiate between the idea that the metabolic signal is a metabolic fuel and the idea that the signal is an intracellular metabolic event. There is precedence for the idea that metabolic events generate cellular signals. It has been demonstrated that leptin synthesis and secretion is controlled by metabolic fuel available for the hexosamine biosynthetic pathway (65). It is well known that insulin secretion is initiated by intracellular metabolic events associated with glycolysis, rather than by circulating levels of glucose, glucose binding to cell membranes, or glucose transport (27, 36). Thus the increases in plasma insulin concentrations at 12 h after the start of refeeding suggest that an intracellular metabolic event occurred before the increase in insulin synthesis and secretion. We suggest that the primary stimulus for initiation of pulsatile LH secretion and estrous cyclicity also might be an intracellular metabolic event rather than a hormone-receptor interaction or a change in circulating levels of a particular fuel. Metabolic signals that control pulsatile LH secretion might occur in either the brain or periphery. Levels of adiposity, hormones, and circulating fuels may ultimately contribute to control of food intake and reproduction as mediators of the metabolic stimulus, but the proximal signal may be an intracellular metabolic index, such as the level of ATP or the phosphorylation potential (20-22). In cases where leptin affects food intake and reproduction via effects on peripheral or central metabolism, leptin would act as a modulator of the primary metabolic stimulus. It is well known that leptin modulates peripheral fuel oxidation. Low doses of insulin also contribute to fuel oxidation by promoting uptake of glucose into tissues where these fuels are oxidized. Thus above-threshold levels of plasma leptin and/or insulin might elevate metabolic fuel availability and oxidation, thereby acting on reproductive mechanisms indirectly. Increasing plasma leptin concentrations might alter the sensitivity of the hypothalamus to low fuel availability.

In general, it is unlikely that energy balance and reproduction are controlled by a closed-loop system in which levels of body fat produce changes in hormone levels, which in turn affect neuropeptides that control food intake and the hypothalamic-pituitary-gonadal axis. Mechanisms that link energy balance to reproductive processes must employ a sensory detector that translates changes in metabolic fuel availability and oxidation into changes in neural and endocrine processes (20, 21). Signals generated by intracellular changes in metabolic fuel availability and oxidation may be important for control of the secretion of hormones such as leptin and insulin, as well as the release of neuropeptides and neurotransmitters involved in control of food intake, energy balance, and reproduction. These intracellular metabolic processes have received comparatively little attention since the discovery of leptin.