Recent work demonstrated that the febrile response to peripheral immune stimulation with proinflammatory cytokine IL-1β or bacterial wall lipopolysaccharide (LPS) is mediated by induced synthesis of prostaglandin E2 by the terminal enzyme microsomal prostaglandin E synthase-1 (mPGES-1). The present study examined whether a similar mechanism might also mediate the anorexia induced by these inflammatory agents. Transgenic mice with a deletion of the Ptges gene, which encodes mPGES-1, and wild-type controls were injected intraperitoneally with IL-1β, LPS, or saline. Mice were free fed, and food intake was continuously monitored with an automated system for 12 h. Body weight was recorded every 24 h for 4 days. The IL-1β induced anorexia in wild-type but not knock-out mice, and so it was almost completely dependent on mPGES-1. In contrast, LPS induced anorexia of the same magnitude in both phenotypes, and hence it was independent of mPGES-1. However, when the mice were prestarved for 22 h, LPS induced anorexia and concomitant body weight loss in the knock-out animals that was attenuated compared with the wild-type controls. These data suggest that IL-1β and LPS induce anorexia by distinct immune-to-brain signaling pathways and that the anorexia induced by LPS is mediated by a mechanism different from the fever induced by LPS. However, nutritional state and/or motivational factors also seem to influence the pathways for immune signaling to the brain. Furthermore, both IL-1β and LPS caused reduced meal size but not meal frequency, suggesting that both agents exerted an anhedonic effect during these experimental conditions.
- food intake
- body weight
- prostaglandin E2
anorexia is part of the centrally mediated acute-phase inflammatory response, which also includes the symptoms of fever, hyperalgesia, inactivity, sleepiness, social avoidance, and stress-hormone release (55). While it was obscure for a long time how peripherally occurring inflammatory events were monitored by the brain, the latter being an immune privileged organ protected by a blood-brain barrier that is impermeable for blood-born proinflammatory substances, recent studies of the mechanisms of fever (see Ref. 54) have revealed how peripheral inflammatory signals can gain access to the central nervous system and which neural circuits are activated by these signals. In particular, recent evidence indicates that fever is critically mediated by induced PGE2 synthesis along the blood-brain barrier (15) and that a population of target neurons lies in the thermoregulatory center of the brain and has specific PGE2 receptors (60, 68). Other studies have identified central nervous structures that are activated during conditions of anorexia, in both acute and chronic models (12, 21, 32, 34, 49, 56); however, little is known yet about the signaling pathways from the periphery to the brain that are mediating the anorexigenic response to inflammation. Several hypotheses have been put forward in addition to prostaglandin signaling via the blood-brain barrier, including the idea that peripheral nerves are activated by proinflammatory messengers and the possibility that these messengers have direct actions on the brain circumventricular organs (11, 14).
In the present study, we induced anorexia with intraperitoneal (IP) injection of lipopolysaccharide (LPS), which mimics endotoxemia, or the proinflammatory cytokine IL-1β, and examined food intake and body weight in mice with deletion of the Ptges gene that encodes for microsomal prostaglandin E synthase-1 (mPGES-1), the inducible terminal enzyme in the PGE2 synthesizing pathways (30). Both LPS and IL-1β are potent anorexigens that reduce food intake in rodents (26, 27). Use of this knockout mouse enabled us to determine to what extent anorexia induced by inflammation is similar to inflammation-induced fever in its dependency on mPGES-1 and hence on induced PGE2 synthesis. Our findings demonstrate that, while the anorexia elicited by IL-1β seems to be similarly dependent on mPGES-1 as the febrile response, anorexia induced by LPS, in contrast to LPS-induced fever (15), is largely independent of mPGES-1 and hence must be mediated by other pathways than those involving induced PGE2 synthesis.
MATERIALS AND METHODS
A total of 111 female mice with a deletion of the Ptges gene (mPGES-1 knock-out mice) and wild-type controls of the DBA1/lacJ strain (67) was used in this study. Age ranged between 9 and 25 wk (see Table 1). The animals were housed one per cage in a pathogen-free facility at an ambient temperature of 25 or 30°C (±1°C) and on a 12:12-h light-dark cycle (lights on at 0800). Cages were enriched with nesting material, and food and water were available ad libitum, unless otherwise stated. All experimental procedures were approved by the Animal Care and Use Committee at the Linköping University.
Monitoring of Food Intake
Food intake was measured by an automated feeding monitoring system (AccuScan Instruments, Columbus, OH). The system, consisting of 12 Plexiglas cages with separate food chambers filled with powdered food (R70, Lactamin, Norrköping, Sweden), allowed the mice ad libitum access to food while minimizing the animals' ability to remove food for storage elsewhere in the cage. Food chambers were available to the mice via a wooden ramp. The food tray was standing on a balance (resolution 0.01 g) that was coupled to a computer. Feeding data were collected and processed using software DietDat and DietMax (AccuScan Instruments). Data were retrieved every 1.5 s and included information on whether the scale was in a stable or unstable mode, the weight of the food that had been removed, and the time points for these events. All animals were allowed to acclimatize to the cages 1 wk before the experiment. Food trays were refilled once per day, and food intake and spill of food were studied during 1 wk before the experiment. Mice that had visible spill of food in the cage or beside the balance were excluded. Food intake data were imported to a database (Microsoft Access 2003; Microsoft, Redmond, WA) and programmed scripts (Visual Basic, Microsoft) to obtain accumulated food intake every one-half hour, meal duration, meal size, and meal frequency.
The following criteria were used to define a meal. A meal was considered to have taken place whenever at least 0.05 g of food was removed from the tray. A meal was considered terminated if a new removal was not initiated within 10 min. Otherwise, it was considered as a single meal. By using these criteria, in average, we included 94.5% (SD 4.5) of the total amount of ingested food during the test period.
Experiment 1: Food Intake in Naive Mice
Food intake was continuously measured in knock-out and wild-type mice during 3 consecutive days after the acclimatization week. Food trays were refilled once per day, but mice were not handled. To match the experimental conditions described below, room temperature was set to 25 ± 1°C or to 30 ± 1°C in two different sessions involving different sets of mice.
Experiment 2: IP Injection of IL-1β
Mice kept at a room temperature of 25 ± 1°C were injected IP with 600 ng of recombinant mouse IL-1β expressed in E. coli (R&D Systems, Minneapolis, MN; endotoxin content <0.1 ng/μg of IL-1β), diluted in 100 μl saline. Control animals received 100 μl saline. The dose of IL-1β was based on the results of a previous study from this laboratory, in which it was shown to induce a robust febrile response (58). Injections were performed between 1845 and 1900. Food trays were withdrawn during 1 h after injection. Then light was turned off, food reintroduced, and food intake measured for 12 h. By this setup, we ensured that the injected IL-1β had elicited a systemic response by the time we started monitoring the food intake, and that other effects on food intake, such as those due to the stress induced by the handling of the animals during the injection, were minimized. The body weight of the mice was recorded at the time of injection, and 12, 24, 48, 72, and 96 h after food was reintroduced (i.e., 1 h after injection).
Experiment 3: IP Injection of LPS
Two micrograms of LPS (Sigma, St. Louis, MO; 0111:B4) were diluted in 100 μl saline and injected IP. Control animals were injected with the vehicle only. The LPS dose was based on a previous study from this laboratory in which it was shown to elicit a robust febrile response (15). All other experimental conditions were the same as in experiment 2.
Experiment 4: Food Deprivation and Subsequent LPS Injection
Because both genotypes (Ptges−/− and Ptges+/+) in experiment 3 displayed a similar, and pronounced, anorexia following LPS injection (see results), we also studied their response to LPS following food deprivation to examine if an enhanced feeding drive would reveal any differences in food intake between the two genotypes. Animals were food deprived for 22 h before they were given an IP injection of either 2-μg LPS diluted in 100-μl saline or vehicle only, and food intake was recorded as described above. The body weight of the mice was recorded at the start of fasting period (at 2200 the day before injection), at the time of injection, and 12, 24, 48, 72, and 96 h after food was reintroduced. Animals were kept at 30°C to avoid their entering a status of torpor (for review, see Ref. 48).
Statistical analysis was performed in SPSS, version 12.0.1 (SPSS, Chicago, IL) or Microsoft Excel 2003 (Microsoft). For statistical analysis of accumulated food intake and relative weight change at a given time point, the ratio R = (μa/μb)/(μc/μd) was studied, where μ is the expected population mean of LPS/IL-1β-treated (a) or saline-treated (b) knock-out mice, and LPS/IL-1β-treated (c) and saline-treated (d) wild-type mice. A Taylor expansion gives an approximately unbiased estimator of R and its variance, with degrees of freedom calculated with Satterthwait's method. At every measured time point, a one-tailed t-test was used to test the null hypothesis R = 1 vs. the alternative R > 1 (i.e., the gene deletion resulted in an attenuated anorexic response and weight loss in response to LPS/IL-1β injection). In addition to this, in experiment 2, one-tailed t-test was used to compare saline vs. IL-1β-treated knock-out mice with respect to accumulated food intake. Furthermore, in experiment 3, a two-way ANOVA with treatment and genotype as fixed factors was used to compare whether LPS treatment had an effect on accumulated food intake after 12 h and on relative weight change at every measured time point.
Other statistical analyses were done by a one-way ANOVA, followed by post hoc t-test using, where appropriate, the Bonferroni method to correct for multiple comparisons.
Data are expressed as means ± SE, unless otherwise explained. A P value < 0.05 was considered significant.
Experiment 1: Food Intake in Naive Ptges+/+ and Ptges−/− Mice
To ensure that food intake during normal conditions did not differ between wild-type and knock-out mice, we measured food intake for 72 h at two different room temperatures (25 and 30°C). Accumulated total food intake, as well as accumulated total food intake in relation to body weight, was similar in the two genotypes at both ambient temperatures (Fig. 1). As expected, accumulated food intake was higher at 25°C than at 30°C, reflecting increased metabolism at a nonthermoneutral temperature (reviewed in Ref. 48). In addition, as expected, animals consumed food mostly during the dark phase, although a smaller amount was consumed during the late phase of the light period, as described previously (37). The differences in the amount of food ingested at the different temperature settings were primarily due to an increased meal frequency at the lower temperature, whereas meal size was largely unaffected (data not shown).
Experiment 2: Food Intake in IL-1β-injected Ptges+/+ and Ptges−/− Mice
Accumulated food intake.
Wild-type mice injected with IL-1β displayed a robust anorectic response lasting throughout the observation period (Fig. 2A). Thus, at 4 h after food had been reintroduced (i.e., 5 h postinjection), IL-1β-injected wild-type mice showed a 60% reduction of food intake compared with saline-injected wild-type controls. At 8 h, the difference was ∼50%, and at 12 h (i.e., the end of observation period), it was 40%. In sharp contrast, IL-1β-injected knock-out mice displayed a complete lack of anorectic behavior, compared with their saline-injected controls, during the earliest phase of the experimental period (0–2.5 h after food presentation). At 3–8 h after food presentation, knock-out mice displayed an attenuated anorectic response but caught up and had consumed almost the same amount of food at the end of the experimental period (9–12 h) as the saline-injected controls. The difference between the two genotypes in their response to IL-1β was statistically significant or close to statistically significant during the first 4.5 h after food had been reintroduced (P = 0.03–0.1) and was statistically significant during the remaining observation period (P = 0.005–0.03).
Wild-type mice injected with IL-1β displayed a lower average body weight than saline-injected wild-type mice during the observation period of 4 days postinjection and did not regain their initial body weight during this period (Fig. 2B). In contrast, IL-1β-injected knock-out mice showed in average only a small weight reduction compared with their saline-injected controls. However, the differences between genotypes were not statistically significant.
The total size of detected individual meals by the defined criteria was 96.1% (SD 3.6) of the difference between start and end-point values recorded by the balances. Total meal durations were different between groups [F(3,25) = 3.1; P = 0.04], and post hoc test revealed a statistically significant difference between saline and IL-1β-injected wild-type mice (P = 0.004) but not between the saline and IL-1β-injected knock-out mice (P = 0.78). There were no significant differences in meal frequency between the different groups of animals (IL-1β- or saline-injected wild-type and knock-out mice, respectively), but there was a tendency that mean meal size and mean meal duration differed [F(3,25) = 2.4; P = 0.1, and F(3,25) = 2.4; P = 0.09, respectively], being smaller and shorter in IL-1β-injected wild-type mice than in saline-injected wild-type mice, respectively (see Table 2).
Experiment 3: Food Intake in LPS-injected Ptges+/+ and Ptges−/− Mice
Accumulated food intake.
LPS injection resulted in pronounced anorexia in both wild-type and knock-out mice (Fig. 3A). At 4 h after food had been reintroduced (i.e., 5 h postinjection), LPS-injected mice, irrespective of genotype, had consumed 85% less food than saline-injected mice. At 8 h, the reduction in food intake was ∼70%, and at the end of the observation period (at 12 h) it was 45%. Two-way ANOVA revealed a significant reduction of accumulated food intake in LPS-treated animals during the 12-h period [F(1,29) = 41, P < 0.001].
Both genotypes displayed a pronounced and similar weight reduction subsequent to LPS injection compared with saline-injected controls (Fig. 3B). It was most pronounced at 24 h [F(1,31) = 33; P < 0.001] and 48 h [F(1,31) = 45; P < 0.001] but still statistically significant at 4 days postinjection [F(1,27) = 4.4; P = 0.05] (Fig. 3B).
The total size of detected individual meals by the defined criteria was 96.7% (SD 5) of the difference between start and end-point values recorded by the balances. There was a significant difference in meal frequency among the groups [F(3,25) = 3.0; P = 0.05]. Post hoc test revealed a tendency (P = 0.1) that knock-out mice treated with LPS had a lower meal frequency than saline-injected knockout mice, whereas wild-type mice did not differ (P = 0.39) (see Table 2). However, there was a significant difference among groups with regard to total meal duration [F(3,25) = 9.9; P < 0.001]. Post hoc test revealed that LPS-injected mice of both genotypes ate for a considerable shorted period than their saline-injected controls (P = 0.006 in both cases). While both mean meal size and mean meal duration were reduced in the LPS-injected animals, these differences were not statistically significant [F(3,25) = 2.2; P = 0.1, and F(3,25) = 1.7; P = 0.18, respectively].
Experiment 4: Food Intake in LPS-injected Food-deprived Ptges+/+ and Ptges−/− Mice
Because Ptges−/− mice showed to be just as sensitive to LPS as wild-type mice with respect to the anorectic response, being in sharp contrast to what was found after IL-1β injection, we food-deprived the mice for 22 h before injecting them with LPS. The rationale was to elicit an enhanced feeding drive to examine if any differences in the anorectic response to LPS between genotypes would be revealed during such a condition. To avoid that the mice entered the state of torpor, involving profound reduction in metabolism and body temperature (see Ref. 48), this experiment was performed at a thermoneutral temperature (30°C).
Accumulated food intake.
Food-restricted wild-type mice injected with LPS showed a robust anorexia during the entire experimental period. In comparison, while knock-out mice also displayed an anorectic response, this response was attenuated (Fig. 4A). At 4 h after food had been reintroduced, wild-type mice injected with LPS displayed an 85% reduction in accumulated food intake compared with saline-injected controls, whereas the reduction in food intake among the knock-out mice was ∼70%. At 8 and 12 h, the corresponding figures were 70 and 50%, and 60 and 40%, respectively. The difference between the genotypes was statistically different from 2.5 h after food had been reintroduced (P = 0.001–0.05) and was statistically significant or close to statistically significant during the preceding period (P = 0.01–0.07).
Fasting of the mice reduced body weight to the same degree in both genotypes (Fig. 4B), indicating that metabolic rate during this experimental condition did not differ between genotypes. In addition, as seen in Fig. 4B, the weight difference between LPS-treated and saline-treated mice was more pronounced in wild-type mice than in knock-out mice, being consistent with the attenuated anorectic response in the latter phenotype. Statistical analysis showed a significant difference between genotypes at 12 h and at 1, 2, and 3 days (P = 0.002–0.003), but not at 4 days (P = 0.09).
The total size of detected individual meals by the defined criteria was 91.5% (SD 7.1) of the difference between start and end-point values recorded by the balances. In both genotypes, meal frequency was somewhat lower in LPS-injected animals than in those given saline (see Table 3), but this difference was not statistically significant [F(3,30) = 1.7, P = 0.18]. In contrast, there was a significant difference in mean meal size among groups [F(3,30) = 2.9, P = 0.049], and post hoc analysis revealed a statistically significant difference between saline- and LPS-injected wild-type mice (P = 0.02) and a strong tendency in knock-out mice (P = 0.06), with a reduced meal size in LPS-injected mice. Furthermore, the size of the first meal after injection was significantly different between groups [F(3,30) = 24.1, P < 0.001], with a much smaller meal ingested by LPS-injected wild-type and knock-out mice compared with saline-injected controls (P < 0.001 and P < 0.05, respectively).
In this paper, we demonstrate that mPGES-1, the inducible terminal PGE2 synthesizing enzyme, contributes differentially to anorexia in response to peripheral immune challenge by injection of IL-1β and LPS. Following IL-1β injection, the anorectic response was absent or greatly attenuated in the mPGES-1 knock-out mice. A similar result, although during different experimental conditions, was recently reported by Pecchi et al. (50), who found that food-deprived male mPGES-1 knock-out mice did not display any anorectic response to IL-1β compared with saline during 5 h postinjection. In contrast, as shown by the present study, LPS injection produced a pronounced anorectic response of identical magnitude in both wild-type and mPGES-1 knock-out mice. However, when mice were food-deprived before LPS injection, there was an attenuated anorectic response in the mPGES-1 knock-out mice compared with that seen in wild-type mice. These data suggest that inflammatory anorexia is mediated by different mechanisms, depending on the nature of inflammatory stimulus, and, furthermore, that the mechanism may vary, depending on the nutritional/motivational state of the animal. Hence, while the inflammatory anorexia elicited by peripheral IL-1β seems largely to be dependent on mPGES-1-mediated PGE2 synthesis, the LPS-induced anorexia is independent of this mechanism in free-fed mice but not in prestarved animals.
Recent studies on the role of mPGES-1 in the febrile response have clearly demonstrated that both the IL-1β- and LPS-induced fever are dependent on this enzyme (15, 58), as is the febrile response elicited by subcutaneous injection of turpentine (58), a model for aseptic peripheral inflammation (18). Hence, taken together, the previous and present observations show that the LPS-induced fever and anorexia are elicited through different signaling pathways, at least in free-fed mice. In contrast, the IL-1β-induced fever and anorexia are both largely mediated by the action of mPGES-1, suggesting a common signaling mechanism.
A possible signaling pathway from the periphery to the brain for the immune-induced febrile response has been revealed by recent research. Upon peripheral immune challenge, mPGES-1 has been shown to be induced in brain endothelial cells (8, 12, 70). The same cells also express the inducible isoform of cyclooxygenase, cyclooxygenase-2 (COX-2) (8, 70), as well as receptors for IL-1β (8, 12, 35). The temporal characteristics of the COX-2 and mPGES-1 induction following peripheral immune challenge coincide with induced increase in intracerebral PGE2 levels and the appearance of fever (8, 29, 70), with the possible exception for an early febrile response that has been revealed in certain experimental paradigms (54). These data strongly support the concept that the febrile response to peripheral inflammation is mediated by blood-brain barrier-induced PGE2 synthesis (4, 52, 53, 59) and subsequent ligand binding to intracerebral PGE2 (EP) receptors in the hypothalamic median preoptic region (15, 45, 46, 63, 68).
Taken together with the present observation that IL-1β-induced anorexia is largely dependent on mPGES-1, the above data also raise the possibility that this anorexia is similarly mediated by blood-brain barrier-derived PGE2 synthesis. While the brain site mediating such a prostaglandin response remains to be determined, EP receptors have been demonstrated on neurons in brain structures considered to be involved in the control of food intake (7, 13, 44, 71), and some of these neurons have been shown to be activated by anorectic stimuli (12). Furthermore, in both acute and chronic inflammatory models associated with anorexia, COX-2 and mPGES-1 expression has been shown to be induced in these brain structures, implying the presence of PGE2 synthesis (12, 35, 50). However, other mechanisms for IL-1β-induced anorexia must also be considered, in particular, the possibility that peripherally synthesized PGE2 elicits anorexia by activating the vagus nerve. EP receptors of the EP3 subclass are expressed on the vagus nerve, and peripheral administration of IL-1β results in increased discharge of vagus afferents through a cyclooxygenase-dependent mechanism (9). The role of the vagus nerve in cytokine-induced anorexia is, however, disputed. While there is abundant evidence that the vagus nerve is involved in feeding-depressing responses to various gut hormones (e.g., Ref. 62) and also seems to be critical to cytokine-induced conditioned taste aversion (22), in several studies vagotomy has been found to be ineffective in attenuating cytokine-induced anorexia (40, 51, 61), unless animals were prestarved (1). Importantly, and although data are conflicting, the consensus view seems to be that IL-1β-induced fever also is independent of vagotomy (43, 54), at least at moderate to high doses (24). The weight of evidence thus favors the idea that a humoral pathway mediates IL-1β-induced fever and anorexia. Because both phenomena are mPGES-1 dependent, this pathway most likely involves prostaglandin synthesis at the blood-brain barrier interface, as suggested by the IL-1β-induced induction of this enzyme in brain endothelial cells (8, 70).
The present finding that LPS-induced anorexia in free-fed animals was independent of mPGES-1, in contrast to the anorexia elicited by IL-1β, suggests that it is also independent of induced PGE2 synthesis, irrespective of whether this takes place peripherally or centrally. It hence appears that the anorectic response to LPS is not mediated by IL-1β, which is largely consistent with previous observations that IL-1 receptor antagonist completely or partly failed to reverse LPS-induced anorexia, while abolishing the anorexia induced by IL-1β (33, 66). Furthermore, IL-1β knock-out mice exhibited a normal anorexia following systemic LPS administration (19). LPS exerts its effect by binding to the toll-like receptor 4 (TLR-4), which is expressed on peripheral immune cells as well as on nervous structures, both peripherally and centrally, suggesting several alternative and perhaps redundant routes by which LPS may influence central nervous functions. These include the possibility that proinflammatory and anorexigenic cytokines that are released from macrophages, together with IL-1β (6, 20, 36, 47), mediate the LPS-induced anorexia in a prostaglandin-independent manner. LPS may also exert its anorexic effect by an enhanced expression of the cytokine-like adipocyte hormone leptin (23, 25, 57; but cf. Refs. 16, 41), which controls appetite by regulating the expression of neuropeptide Y and α-melanocyte stimulating hormone in the arcuate nucleus of the hypothalamus (10, 69) through signaling pathways that are independent of induced PGE2 synthesis (32). Furthermore, LPS may activate central anorexigenic pathways by binding to TLR-4 on the vagus nerve (28), or directly via central TLR-4s on, e.g., circumventricular organs, such as the area postrema (38), and by these routes influence neuronal signaling in brain stem and forebrain regions critical for feeding behavior (2). In a recent elegant study using chimeric mice, Chakravarty and Herkenham (5) demonstrated a requirement for TLR-4 function in central nervous system-resident cells, independent of systemic cytokine effects, for sustained central nervous system-specific inflammation and corticosterone rise during endotoxemia. A similar approach could help answer the role of peripheral vs. central TLR-4s in the LPS-induced anorexia.
While free-fed animals demonstrated an LPS-induced anorectic response that was independent of mPGES-1, prestarved animals differed. Thus, in mPGES-1 knock-out mice that had been food deprived for 22 h at a thermoneutral temperature before LPS injection, the anorectic response was significantly attenuated compared with that seen in wild-type mice, implying that part of the anorexia during this experimental condition was elicited through mPGES-1-mediated PGE2 synthesis. It thus appears that the nutritional status and/or motivational factors affect the anorectic response to LPS differently in the presence or absence of mPGES-1. Food deprivation has been shown to cause a dysregulation in the pattern of cytokine induction by LPS, with highly increased levels of TNF-α and decreased levels of IFN-γ in fasted mice (17). TNF-α induces COX-2 expression in brain endothelial cells (3), indicating that increased TNF-α release by LPS in food-deprived animals could contribute to the LPS-induced anorexia in a prostaglandin-dependent manner. While such or similar mechanisms may help explain the fasting-elicited prostaglandin dependence, it remains obscure how motivational factors influence the immune-to-brain signaling pathway. However, it is clear that motivational factors, and probably also a variety of other experimental conditions, are important in anorexia, and this may help reconcile the present findings with some of the previous literature on COX-2 inhibition of LPS. When mice were given sweetened milk, a highly palatable nutrient that has been shown to overcome the anorexia in certain experimental models (65), the LPS-induced anorexia was attenuated by COX-2 inhibition (64). However, and in good accord with the present findings, the anorexia induced by LPS was reported to be less sensitive to COX-2 inhibition than anorexia induced by IL-1β (65), with a much higher dose being required to attenuate the LPS-induced anorexia than what was needed for the inhibition of the IL-1β-induced anorexia (64). A contribution by prostaglandin-dependent pathways to LPS-induced anorexia in mice was also suggested by Johnson et al. (31), who demonstrated an attenuated anorectic response to LPS following pretreatment with a selective COX-2 inhibitor. In the same study, these authors also showed that COX-2 knock-out mice tended to display less weight loss than wild-type mice after LPS administration. However, a 50-fold higher LPS dose was given than what was administered in the present study, implying that the results are not comparable. Also in rats, COX-2 inhibition has been reported to attenuate the anorectic response to LPS in free-fed animals (42), indicating a contribution to the anorexia by both prostaglandin-dependent and prostaglandin-independent mechanisms in this species.
The observations on meal frequency and meal size obtained in the present study deserve some comment. The prevailing paradigm, based on studies in rats, seems to be that LPS exerts its anorectic effect by reducing meal frequency, while IL-1β is considered to affect both meal size and meal frequency (39). In the present study, we found that IL-1β given to free-fed wild-type mice suppressed the total time spent eating, and further, although not statistically significant, that these mice displayed reductions in individual meal size and mean meal length, whereas there was no reduction in meal frequency. LPS injection to free-fed mice resulted in reduction in mean meal size, mean meal length, and total meal duration, the latter being statistically significant, whereas meal frequency was little affected, except for in knock-out mice. Food-deprived wild-type mice given LPS similarly showed large, and significant, reduction in mean meal size and a strong tendency in the same direction in knock-out mice. In particular, the size of the first meal after injection was much smaller in LPS-injected mice of either phenotype compared with their saline-injected controls. In contrast, there was very little reduction in meal frequency.
These data seem to indicate a different mechanism for both LPS- and IL-1β-induced anorexia in mice than in rats, suggesting that both affect the consummatory phase of the ingestive behavior rather than the food-seeking behavior. It should be noted that, in the present study, food was supplied via a separate food chamber that was available to the mice via a wooden ramp, hence requiring some foraging behavior by the animals. It is of particular interest that food-deprived wild-type and knock-out mice given LPS displayed a reduced meal size, both throughout the experimental period as well as immediately when food was presented, whereas meal frequency was unaffected. This strongly indicates that LPS has an anhedonic effect in this paradigm, and it further emphasizes why factors such as the palatability of the food may be important in anorexia, as discussed above.
In summary, the present study demonstrates that IL-1β and LPS induce anorexia in mice via distinct mechanisms. The anorexia induced by IL-1β is mainly prostaglandin dependent and most likely mediated by a mechanism similar to that critical for the febrile response. In contrast, LPS-induced anorexia largely seems to be prostaglandin independent, but may act through induced prostaglandin synthesis, depending on the nutritional/motivation state of the animal.
This study was supported by grants from the Swedish Research Council (no. 7989), the Swedish Cancer Foundation (no. 4095), government funds for clinically oriented research (ALF) administered by the County Council of Östergötland (LIO-4818), Konung Gustaf V:s 80-årsfond, The County Council of Östergötland (LIO-5355), and the Faculty of Health Sciences.
We thank O. Eriksson for advice and help with statistical calculations.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society