Interleukin-1 -induced Anorexia in the Rat
Influence of Prostaglandins
Marc K. Hellerstein, Simin Nikbin Meydani, Mohsen Meydani, Ken Wu, and Charles A. Dinarello
U. S. Department ofAgriculture Human Nutrition Research Center on Aging at Tufts University, Department of Medicine, Division of
Geographic Medicine and Infectious Diseases, Tufts University School of Medicine and New England Medical Center, Boston,
Massachusetts 02111; Department of Nutritional Sciences, University of California at Berkeley, Berkeley, California 94720;
and Department ofMedicine, Division of Endocrinology and Metabolism, San Francisco General Hospital, University ofCalifornia at
San Francisco, San Francisco, California 94110
Abstract
The anorexia associated with acute and chronic inflammatory
or infectious conditions is poorly understood. Our objectives
were to explore the anorexigenic effects of interleukin-1 (IL-1)
in the rat. Recombinant human (rh) IL-1,6, murine (rm) IL-la
and to a lesser extent rhIL-la significantly reduced food intake
at 24.0 gg/kg i.p. but not at lower doses, in young (200-250 g)
meal-fed rats on chow diets. The anorexic effect appears to be
mediated by prostaglandins since pretreatment with ibuprofen
completely blocked it, and a fish oil based diet abolished it, in
comparison to corn oil or chow diets. Fish oil feeding also
decreased basal and IL-1 stimulated prostaglandin E2 production by tissues in vitro (liver, brain, peritoneal macrophages)
and in the whole body. Constant intravenous infusions of lower
doses of IL-1 also diminished food intake, though intravenous
boluses did not (reflecting rapid renal clearance). Chronic daily
administration of IL-1 caused persistent inhibition of food intake for 7-17 d in chow and corn oil fed rats, but had no effect
in fish oil fed rats. There was an attenuation of the effect
(tachyphylaxis) after 7 d in corn oil and chow fed rats, but
slowed weight gain and lower final weights were observed after
17-32 d of daily IL-1. Old (18-20 mo Fisher 344) rats showed
less sensitivity to IL-1 induced anorexia. In conclusion, IL-1 is
anorexigenic in the rat, but this is influenced by the structural
form of IL-1, the route and chronicity of administration, the
source of dietary fat, and the age of the animal. The ability of
prior fat intake to influence the anorexic response to IL-1
represents a novel nutrient-nutrient interaction with potential
therapeutic implications.
Introduction
Both acute and chronic inflammatory conditions as well as
infectious diseases are associated with alterations in nutrition
and metabolism (1, 2). One of the most important of these is
decreased food intake (anorexia), which can progress to caPortions of this work have appeared previously in abstract form. 1988.
FASEB (Fed. Am. Soc. Exp. Biol.) J. 2:Al 198. (Abstr.)
Address reprint requests to Dr. Hellerstein, Room 4101, Koret
Center for Human Nutrition, San Francisco General Hospital, 1001
Potrero Avenue, University of California, San Francisco, CA 941 10.
Receivedfor publication 2 June 1988 and in revisedform 7 March
1989.
J. Clin. Invest.
© The American Society for Clinical Investigation, Inc.
0021-9738/89/07/0228/08 $2.00
Volume 84, July 1989, 228-235
228
chexia and death by starvation (3, 4). With its high prevalence
in human immunodeficiency virus-infected patients (5-7), cachexia is becoming an increasingly important problem, yet it
remains poorly understood. The availability of recombinant
cytokines (8, 9) allows experimental testing of new hypotheses
about the mediators and mechanisms of the anorexia of inflammation.
Tumor necrosis factor/cachectin (TNF)' has received attention in this regard (9-12). Administration of a solution
containing macrophage secretory products including TNF
suppresses food intake in mice (10) but these preparations
contain other cytokines such as IL- I13 (13) in addition to TNF.
Tumor cells secreting TNF can induce cachexia in mice (1 1)
but again the presence of other mediators cannot be excluded.
When given in sublethal doses to mice (12, 13), recombinant
TNF results in decreased appetite and weight loss. However,
the effect appears to be short lived ( 12) in addition to requiring
escalating doses (13).
The effect of IL-l on food intake in rats is largely unknown (14). McCarthy et al. (15) reported decreased intake of
a liquid diet in rats given recombinant murine IL-la, but the
effect was relatively minor, lasting only for the first hour of
refeeding. More recently (16), rhIL-lB administered to ad lib.
fed mice at 6 ,g/kg per d in two divided doses was reported to
decrease food intake by 10- 15%, a similar quantitative effect
as for rhIL- 1 a but significantly less than recombinant murine
IL-la given at similar doses (40-50% reduction in intake). The
effect of IL- 1(3 on food intake is relevant to interpretation of
TNF effects, since TNF stimulates the release of IL- 1f (16, 17).
Our objectives in these studies were, first, to develop a
quantitative, reproducible and physiologically relevant animal
model for the anorexia of inflammatory illnesses, using IL-l.
Dose-response relationships, potential mediation by eicosanoid pathways, involvement of central or peripheral appetite
regulatory mechanisms, effects of normal aging, differences
between the two structural forms of IL-I and persistence of
any effects during chronic IL- 1 administration were then examined.
Methods
Male Sprague-Dawley rats (Charles River Breeding Laboratories, Wilmington, MA) were trained to a single daily meal (meal-fed) after
reaching 150 g body wt. Meal-feeding has advantages over ad lib.
feeding for studies of drug or hormone effects on appetite (18, 19), in
that the effects of anorexigens with a short biological duration of action
1. Abbreviations used in this paper: C.O., corn oil; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; F.O., fish oil; LA, linoleic
acid; TNF, tumor necrosis factor.
M. K Hellerstein, S. N. Meydani, M. Meydani, K. Wu, and C. A. Dinarello
are not lost over a 24-h period, since all intake occurs over 3 h. Mealfed animals grow well and eat amounts comparable to ad lib.-fed rats
(20), and the metabolic and enzymatic consequences of meal-feeding
have been well characterized (21, 22). Studies were performed at
200-250 g body wt. Aged Fisher 344 rats (18-20 mo old, National
Institute of Aging) were used in some studies. All animals were housed
in individual cages in a temperature controlled room (220C) with a
12-h dark/light cycle daily. Animals were fed nonpurified diets consisting of ground Purina Rat Chow (Ralston Purina Co., St. Louis,
MO), or purified nutritionally adequate diets containing 10% by
weight of corn oil (C.O.) (Mazola; Best Foods, Englewood Cliffs, NJ) or
1.2% C.O. plus 8.8% fish oil (F.O.) (MaxEPA, a gift from R. P. Scherer
Co., Troy, MI) for 6 wk. The details of diet preparation and precautions to minimize oxidation of oils have been previously published
(22). As previously reported (23), C.O. contained 65.6±0.6% linoleic
acid (LA) while MaxEPA contained 2.0±1.4% LA, 16.7±1.8% eicosapentaenoic acid (EPA), and 12.2±1.1% docosahexanoic acid (DHA).
The F.O. diet, therefore, provided about 1% by weight of LA. Food was
provided daily in spill-proof glass containers placed inside cages. Water
was provided ad lib.
RhIL- I( was produced, purified and characterized as described (8)
and consisted of the carboxy-terminal 157 amino acids. This and subsequent lots were active in the rabbit pyrogen test at 100 mg/kg and the
mouse lymphocyte activating factor (LAF) assay. The endotoxin concentration in the present preparation was < 60 pg/mg. The specific
activity as measured by half-maximal units (LAF) on C3H/He5 mouse
thymocytes was 5.5 X 107 U/mg. The rhIL-lI3 was diluted in 0.1%
serum which had been dialyzed overnight at 4VC then heated to 560C
for 20 min. The diluted rhIL-1If was administered either intravenously
or intraperitoneally. RhIL-la and recombinant murine (rm) IL-la
were generously provided by Dr. Peter Lomedico (Hoffmann-La
Roche, Nutley, NJ). The activity of rmIL-la is 1.3 X 108 U/mg
(D 1O.G4. 1 cells, where I LAF unit = 20-50 DI 0 units) and the endotoxin concentration is 0.55 ng/mg protein (LAL). The activity of
rhIL-la is 2.5 X I09 U/mg (D0 assay) and the endotoxin concentration is 65 pg/mg protein (LAL). Meal-fed rats (Chow, C.O., or F.O.)
given 0.1% serum i.p. or heat inactivated rhIL- IB i.p. 1 h premeal had
no change in food intake from baseline.
The technique for placement and maintenance of indwelling intrajugular silastic catheters has been described elsewhere (25). Catheterization may depress food intake for 24-48 h, therefore no studies
were performed for at least 72-96 h after catheter placement, until
food intake had stabilized at precatheterization values. A parenteral
preparation of ibuprofen (50 mg/ml) was the gift of Upjohn Co. (Kalamazoo, MI) and was administered intravenously or intraperitoneally
in sterile 0.45% saline.
48-h urine was collected on dry ice from rats before and after
intraperitoneal injection of rhIL-l. The urinary metabolite of PGF2a
[ 13,14-dihydro- l 5ketoPGF2a (MetF2a)], an indicator of total body
PG production, was extracted using a Sep-pak C18 cartridge (Waters
Associates, Milford, MA) as described by Powell (26). MetF2a in the
extracts was measured by RIA as described previously (27), and urinary creatinine was measured by Roche Cobas Fara Centrifugal Analyzer (Nutley, NJ) using Roche Diagnostic Systems Reagent and procedure number 44905, a modification of the method of Larson (28).
Rats were anesthetized with metaphane and peritoneal leukocytes
were obtained by lavaging the peritoneal cavity with 50 ml of Ca2l
and Mg2e free PBS solution. The cell pellet was obtained by centrifugation at 1,200 rpm for 10 min. Cells were washed two times in RPMI
1640 (Gibco Laboratories, Grand Island, NY) and resuspended in
RPMI 1640 at a concentration of 5 X 106 cells/ml. 100 gl of cell
suspension (5 X 105 cells) was cultured in 96-well culture plates (Becton-Dickinson Co., Oxnard, CA) in the presence of 0.5% fetal bovine
serum (Gibco Laboratories) and 50 gl of 0.1% BSA (Sigma Chemical
Co., St. Louis, MO) or 50 ,l of 60 ng/ml rhIL-lIfl solution in 0.1% BSA
(final rhIL-113 concentration 15 ng/ml) for 24 h in a 37°C, 5% CO2
humidified incubator. Cell-free supernatant was saved at -70'C for
PGE2 analysis. PGE2 was analyzed by RIA (27). PGE2 antibody had a
-
cross-reactivity of 5.6% with PGE3 standard (Cayman Chemical Co.,
Ann Arbor, MI). Blood was collected from the vena cava and brain
dissected on an ice-cold platform and immersed with ice-cold buffer.
The mid-brain (including hypothalamus, striatum, and hippocampus)
and brain stem (including medulla oblongata and pons) were dissected
with precooled tools. 10% homogenates from brain regions and liver
were incubated for 30 and 10 min, respectively, in a shaking water bath
at 370C. PGE2 was measured in the supernatant of the incubation
media. Student's t test or one-way ANOVA (when significant, followed
up by Tukey-HSD procedure with a procedure-wise error rate of 0.05)
was used to determine significance of differences between the means of
the dietary groups. When unequal variances were present, logarithmic
transformation of the data was performed. If this resolved the unequal
variances, statistics were performed on the transformed data.
Results
First, the anorexigenic dose of intraperitoneal rhIL- 13 in
meal-fed rats was established. Administration of 0.1% serum,
alone or with 80 ng (400 ng/kg) or 240-400 ng (1.2-2.0 ug/kg)
of rhIL- 13 i.p. 1 h before the daily meal had no effect on food
intake in young (200 g) chow fed rats (Fig. 1). Administration
of 800 ng (4 gg/kg) rhIL- 13 had a marked anorexic effect (Fig.
1), decreasing food intake by 37.5±6.6% (P < 0.005) in this set
of rats. Food intake increased to 97.0±3.3% of baseline values
by the next day (NS vs. baseline). Boiled rhIL- 13 given at 4-6
ug/kg i.p. had no anorexic effect. Because many, but not all of
the diverse effects of IL-lI3 are mediated by PGE2 (29), we
administered the cyclooxygenase inhibitor ibuprofen 10
mg/kg i.v. 10 min before injection of 800 ng rhIL-lB i.p. This
completely blocked the acute anorexic effect of rhIL-l1( (Fig.
1). Acetaminophen, which preferentially inhibits brain cyclooxygenase relative to that in peripheral tissues (30, 31), did
not block rhIL- 13 anorexia (not shown).
§ Day of IL-
o:
80 no 240-400mg
(6)
(6)
m00ng
(10)
Administration
DayFoliowinglL-1
Administration
800 g
* IBU
(8)
Figure 1. Effect of intraperitoneal rhIL-l1( alone and with intravenous ibuprofen on food intake in meal-fed rats. Rats weighing
175-225 g were studied. Housing was in individual cages in a temperature controlled room (220C) with a 12-h dark/light cycle daily.
Catheterization may depress food intake for 24-48 h, therefore no
studies were performed for several days after catheter placement,
until food intake had stabilized at precatheterization values. RhIL-lfI
was administered intraperitoneally 1 h before the daily meal. Baseline food intake represents the mean±SEM daily intake of the previous 7-10 d. P values are in comparison to baseline food intake,
using a two-tailed I test. Ibuprofen (IBU) was administered at a dose
of 10 mg/kg i.v., 10 min before administration of rhIL-Ifl. *P < 0.005.
Interleukin-I Anorexia in the Rat
229
Different structural forms and species sources of recombinant IL- I were also tested. RhIL- 1 a was less anorexigenic than
rhIL-1I3 in chow-fed rats (although this did not reach statistical
significance), while the effect of rmIL- la was similar to that of
rhIL-1 I3 (Table I). Anorexia from intraperitoneal rmIL-la, as
from intraperitoneal rhIL-1I, was completely blocked by ibuprofen 10 mg/kg i.v. (not shown). To exclude any possible
effects due to the intravenous route of ibuprofen administration or due to catheterization itself, ibuprofen 10 mg/kg or
0.9% saline were administered to noncatheterized rats by the
intraperitoneal route 1 h before rmIL-la 6 jm/kg i.p., (Fig. 2).
IP ibuprofen completely abolished rmIL-la anorexia (food
intake 99.4% with ibuprofen/IL-l, 56.3% with saline/IL-l).
Next, the effect of dietary fat source on IL- 1 anorexia was
examined (Table I). Exogenous N-3 fatty acids, which are present at high levels in cold water F.O. (24) are known to reduce
PGE2 synthesis in rats (32) as well humans (33). Groups of rats
were chronically fed chow or purified diets based on AIN diet
recommendations for rats (34) containing 10% by weight of
either F.O. or C.O. Animals were begun on purified diets at
75-100 g body wt and continued for up to 6-10 wk. Composition of tissue fat in growing rats has been shown to be altered
by dietary fat source (35). Serum tocopherol levels in these
F.O., C.O. and chow fed rats were comparable (not shown).
The effect of rhIL-1I3 4 jg/kg or rmIL-la 6 ,jg/kg injections on
Table I. Effect of Dietary Fat Source and Recombinant IL-I
Structural Form and Species Source on rIL-J Induced Anorexia
Dietary group
Chow
IL- I form
Effect on food intake
n
% Decrease Mean±SE
rhIL-11,
(12)
rmIL-la
(41)
rhIL- I a
(16)
Corn oil
rhIL-113
(12)
rmIL-1a
(12)
rhIL- I a
18.9±3.2%$
23.6±3.2%$
12.7±3.6%
32.1±5.3%$
44.8±5.5%*
8.5±4.6%
211
* 8sseline I
Figure 2. Effect of pre-
Postl-treatment with intraper-
7
_S
S.
itoneal ibuprofen (10
mg/kg) or 0.9% saline 1
h before administration
1
of rmILla 6jg/gkg on
Intake
food intake in chow fed
rats. Ibuprofen or 0.9%
5
saline were given in 1
ml volume. Baseline
values were calculated
the mean intake for
0.3SI
/Ltas
0.12 Saillne/1L.I
lbuprafanflt.- I
(N=6)
(N1-I)
the previous 9 d for
each animal. Values
shown as mean±SE for each group. P values are calculated by twotailed t test. **P < 0.01; NS, not significant.
15
food intake was significantly different between the F.O. and
C.O. groups with chow-fed rats in between (Table I). Administration of 0.9% saline i.p. had no effect on food intake in any
group. The effect of dietary fat source was similar for rmIL- 1 a
as for rhIL- 1, (Table I). To test the hypothesis that this nutrient-nutrient interaction (prior dietary fat source influencing
food intake response to IL- 1 administration) is mediated by
eicosanoid pathways, PGE2 synthesis was determined in vitro
and in vivo in basal and IL- I stimulated states. Consistent
with the model, in vivo production of PGs of the 2 series,
represented by the urinary excretion of the PGE2a metabolite
13,14-dihydro- 1 Sketo-PGF2a (MetF2a) was higher in C.O. fed
rats after rhIL- 1, injection than in F.O. fed rats (Table II).
Moreover, peritoneal leukocytes from C.O. fed rats synthesized more PGE2 than F.O. fed rats in vitro, both in the presence and absence of rhIL- 113 (Table III). Brain stem and liver
homogenates from F.O. fed rats also produced significantly
less PGE2 than C.O. and chow fed rats (Table III), although no
such difference was observed in mid-brain.
When rhIL- 113 was given intravenously as a 800-ng (4
jig/kg) bolus i h before the daily meal, there was no effect on
food intake (Table IV). The same dose administered by constant intravenous infusion at 200 ng/h over 4 h beginning I h
before the meal reduced food intake by 34.8±15.9% (P
< 0.005), which was not significantly different from equal
doses given intraperitoneally. When rhIL- 11 was infused
overnight at 80 ng/h for the 20 h before and during the daily
(5)
Fish oil
rhIL-1#
6.4±6.5%
(1 1)
rmIL-la
(12)
4.9±4.6%
rhIL-la
-1.1±5.3%
(6)
Rats were fed fish oil, corn oil, or chow diets as described in the text.
RhIL-1,B 4 ug/kg, rmIL- la 6 ug/kg, or rhIL- la 6 jg/kg i.p. were administered 1 h before the daily meal. Changes in food intake were
calculated by comparison to the animal's mean daily intake over the
preceding 7 d. P values are calculated using one-way ANOVA followed up by Tukey-HSD procedure when significant. P value for
overall effect of dietary fat or IL- I source on food intake was 0.0001
by one-way ANOVA.
* P < 0.05 vs. all other groups.
t P < 0.05 vs. fish oil fed (all IL- I sources).
230
M. K
Table I. Urinary MetF2a, Concentration in Rats Fed Different
Diets before and after Intraperitoneal rhIL-I1 Injection
(Mean±SEM, n = 6)
MetF2, (ng/ng creatinine)
Diet
Before injection
After injection
Fish oil
Corn oil
Chow
0.33±0.02
0.35±0.05
0.35±0.06
0.22±0.04*
0.52±0.16
0.40±0.15
* P < 0.09 vs. corn oil fed rats.
48 h urine was collected on dry ice from rats before and after intraperitoneal injection of rhIL- I#. Logarithmic transformation of data
was used when there were unequal variances, and one-way ANOVA
was then applied with Tukey's followup.
Hellerstein, S. N. Meydani, M. Meydani, K Wu, and C. A. Dinarello
Table III. PGE2 Synthesis by Peritoneal Leukocytes and Liver
and Brain Stem Homogenates of Rats Fed Different Diets
in the Presence or Absence of rhIL-If (Mean±SEM)
Peritoneal macrophages
Control
+rhILlj3
Liver
(n)
(n)
(n)
Diet
pg/S X 106 cells
Fish oil
Corn oil
Brainstem
(n)
ng/g
220±52
328±73*
42±7*
(4)
(5)
(6)
(5)
941±440
1327±493
329±72
18±2
(5)
(5)
Chow
9±1§
(6)
(5)
304±63
19±3
(5)
(5)
Tissues were isolated as described in the text. PGE2 was measured by
RIA (26) in the supernatant of the incubation media. P values are
calculated using two-sample Student's t test for peritoneal macrophage data (two groups) and one-way ANOVA for liver and brainstem data (three groups). Statistics were performed on logarithmically transformed data to resolve unequal variances.
* P < 0.09 vs. corn oil fed.
P < 0.0001 for overall diet effect by ANOVA. Fish oil significantly
different from corn oil and chow (P < 0.05) by Tukey's followup.
§ P < 0.0012 for overall diet effect by ANOVA. Fish oil significantly
different from corn oil and chow (P < 0.05) by Tukey's followup.
meal (1,600 ng total, Table IV), food intake was markedly
decreased, and remained decreased the next day in comparison to preceding and succeeding intakes while catheterized.
Inclusion of ibuprofen at 2.5 mg/kg per h to the overnight
Table IV. Effect of Intravenous rhIL-I1f Alone and with
Intravenous Ibuprofen on Food Intake (g/d) in Meal-fed Rats
Group (n)
Baseline
food
intake
Day I intake
(% Baseline)
Day 2 intake
(% Baseline)
11.0±1.0
12.4±1.8
(113.0±16%)
II Constant infusion at 13.2± 1.8
8.6±2.1t
13.6±0.7
200 ng/h
(65.2±15.9%) (103.0±5.0%)
X 4 h (7)
III Constant infusion at 11.1±0.6
1.3±1.2*
6.1±1.9*
80 ng/h
(11.7±10.8%) (55.0±17.1%)
X 20 h (6)
IV Constant infusion at 12.7±1.0
6.2±1.6t§
8.6±1.9*
80 ng/h X 20 h
(48.8±12.6%) (67.7±14.9%)
+ i.v. ibuprofen
2.5 ng/kg per h
X 20 h (5)
I Bolus 800 ng i.v. (4)
intravenous rhIL-13 infusion partially restored food intake toward baseline values (Table IV), but not completely. RmIL- la
administered at 80 ng/h (400 ng/kg per h) for 20 h also decreased food intake (to 66% baseline) and this was blocked by
co-administration of ibuprofen 2.5 mg/kg per h (intake 108%
baseline).
The question whether normal aging alters the anorexic effect of rhIL- 1 was addressed, because many normal homeostatic responses are diminished with aging (36) and because
there exists some evidence (Hellerstein, M. K., unpublished
observations) that elderly humans are less able to mount other
features of the acute-phase response to inflammation (e.g.,
fever, leukocytosis, etc.). NIA aged, 18-20-mo old chow-fed
Fisher 344 rats had diminished sensitivity to IL-113 anorexia
(8.3±3.9% inhibition of food intake, P < 0.05 vs. young chowfed rats).
Finally, we examined the effect ofchronic IL-l administration on food intake and body weight in rats fed different fat
sources. These experiments consume large amounts of recombinant protein. Accordingly, only a relatively small number of
animals could be studied (5 chow rats, 3 F.O., 3 C.O.). To
minimize the possible confounding effects of antibodies to a
foreign protein developing with chronic administration in this
rodent model, rmIL-la was given. Rats were given daily
rmIL- I a 6 gg/kg or 0. 1% serum i.p. Average food intake over
the first 7 d of IL-1 administration was significantly diminished in chow-fed rats (16.6± 1.0 to 13.5±0.9 g/d, a decrease of
18.7%) and C.O.-fed rats (19.3±2.6 to 13.4±2.3 g/d, a decrease
of 30.6%) but not in F.O.-fed rats (14.2±1.9 to 13.7±2.6 g/d, a
decrease of 3.5%) (Fig. 3). IL-la administration was continued
in the C.O. and chow-fed groups beyond the initial 7 d. The
anorexic effect in both groups was attenuated during days
8-17 of IL-l administration (Fig. 3), and there was no apparent anorexic effect days 18-25 of IL- 1 administration (done in
chow group only). After discontinuation of IL-1 treatment
(days 18-25 for C.O., 26-32 for chow), food intake returned to
baseline levels in both groups (chow 99.4%, C.O. 97.9%). The
chow group was rechallenged with IL- 1 on days 33-40 and
120
110
U Baseline Intake
100
S Baseline0PesFood Intake
0 OtffIL-lIalpha
80
70
60
rhIL-lI was administered intravenously to chronically catheterized
rats. Bolus refers to a single injection 1 h before the daily meal. Constant infusions were performed as detailed in the text, using a Harvard infusion pump. Baseline, day one and day two food intakes and
P-values for comparison were calculated as described in Fig. 1.
* P < 0.025
vs. baseline.
* P < 0.005 vs. baseline.
I P < 0.025 vs. group III, day 1 intake.
hRechallange with IL-I alpha
90
Fish Oil
(N-3)
Corn Ol
(N-3)
Chow
(N=S)
Figure 3. Effect of chronic rmIL- Ia administration on food intake in
fish oil, corn oil, and chow-fed rats. A dose of 6 ug/kg i.p. was given
daily. P values calculated by one-way ANOVA, which revealed an
interaction between diet and time on effect of IL- 1 (e.g., diet alters
pattern over time, P < 0.03). *P < 0.05 vs. baseline intake. Each bar
represents the mean±SE from 21-50 data points (three to five animals X 7-10 d of rmIL-lIa).
Interleukin-J Anorexia in the Rat
231
there was no anorexic effect observed (intake 1011.8% with or
without IL- 1). Body weights for chow fed rats giveen IL-la are
shown in Fig. 4. Initially, weight gain is entirely pirevented but
after 7-10 d weight gain is again observed, though the slope of
gain is less than it was before rmIL- la administration and is
less than in paired rats (average weight gain duriing IL- 1 administration was 1.5±0.2 g/d, P < 0.05 vs. pre--IL- 1 weight
gain [3.6±0.2 g/d] and P < 0.05 vs. control group weight gain
[2.7±0.2 g/d]). Similar results were observed in 1the C.O.-fed
group (not shown).
Discussion
These results demonstrate that the meal-fed ralt is a useful
animal model for quantifying the effects of purificad mediators
on food intake and that rhIL-l(3 and rmIL-1a reduce food
intake in the rat at doses 2 4 ug/kg i.p. PGE2 pr()ductiOn appears to be required for the IL- 1 effect, based on 1the ability of
intravenous or intraperitoneal ibuprofen to blo(ck the effect
completely, the ability of chronic F.O. feeding 1to markedly
decrease it, and the observation that F.O. fed rats have signifi
cantly lower in vivo and in vitro PGE2 producti on The anorexic effect of rhIL-lI3 and rmIL- la were similarl modulated
by F.O. feeding and ibuprofen treatment.
The interaction between IL- 1 secretion and P4GiE2 production is complex. PGE2 may inhibit further IL- 1 secretion by
endothelial cells and macrophages (37), represeniting a short
negative feedback loop. Since not all the actionss of IL- are
mediated by PGE2 (e.g., certain immunostimulastory effects,
reference 29), inhibition of macrophage and other tissue PGE2
production (Table III) may increase IL- 1 secreti(on and nonPGE2 mediated IL- 1 effects, even as it decrease-s those mediated by PGE2. Proper classification of IL- 1 ano4rexia a presumably undesirable effect in the clinical setting,;as PGE2 mediated is therefore important. It is worth inoting that
cyclooxygenase inhibition also prevents a numbeir of effects of
TNF (38, 39) and endotoxin (39) without prevernting the endotoxin-induced rise in circulating TNF (39).
iy
300 r
OW* ,"le
Begin IL-
,e-a*
or 0. 1%Serum
l4
Injections
Cr -0-C
-- - Control Rats
- -- IL-I Rats
Average
Body
Weight
(grams)
200
[
*ff4184Alao
aos3°°
a0
.
,
10
5
20
25
30
35
Days
Figure 4. Effect of chronic rmIL- I a administration on
in chow-fed rats. A dose of 6 tog/kg i.p. was given daily.bAverage
weight gain of IL-I group was 1.5±0.2 g/d during IL-I administration, P < 0.05 vs. weight gain pre-IL- 1 (3.6±0.2 g/d) anid P < 0.05
vs. control group (2.7±0.2 g/d), by two-tailed t test.
232
M. K Hellerstein,
S.
Comparisons with previous studies of the effect of IL- I on
food intake in the rat are complicated by methodologic considerations. The report of McCarthy et al. (40) that intracerebroventricularly administered endotoxin and "IL-I" induced
fever but did not suppress intake of a liquid diet in previously
fasted rats must be interpreted with caution, since human
monocyte supernatant containing IL- 1 was administered, not
recombinant IL- 1. The brief and relatively minor effect of
rmIL- la on intake of a liquid diet reported in the rat (14) in
comparison to the more robust effects on total daily intake
shown here with rhIL-l( and rmIL-la has several possible
explanations. The liquid diet is not a physiologic form of food
and may modify any effects of IL- 1 on appetite that are mediated by suppression of gastric motility (41). Due to rapid
clearance of administered IL- 1f (37), the meal-feeding model
may be superior for revealing IL-1,8 anorexia. The increased
hepatic lipogenesis of meal-fed rats (19-21) may also increase
sensitivity to IL- 1 anorexia, if the mechanism is a peripheral
one and involves alterations in hepatic carbohydrate utilization or triose metabolism, as appears to underlie several other
metabolic models of anorexia (17, 18, 42-46). The ability of
cytokines to stimulate hepatic lipogenesis (47) in the rat is
suggestive in this regard. The difference is unlikely to be due to
the form of IL- 1 used, since we (Table I) and Moldawer et al.
(16) have found rmIL- 1 a to be an effective anorexigen in rodents.
Our observation of differences between rhIL-l1B and
rhIL-1a (Table I) is consistent with other reports of different
activities of IL-la and in some systems (e.g., 48) and may
reflect different affinities of IL-la and IL-113 for the IL-1 receptor, and different abilities to generate PGE2. The difference
between rhIL- 1 a and rmIL- 1 a emphasizes the importance of
species source. The molecular basis of these variations in actions between IL- 1 from different species is unknown, although the profound effect of dietary fat source on both
rhIL-lI3 and rmIL-la anorexia suggests a shared mechanism
for their anorexigenic actions.
The question whether IL- 1 anorexia operates by a central
or peripheral mechanism can not be definitively answered yet.
The failure of intravenous boluses of rhIL-l(I (800 ng) to cause
any decrease in food intake (Table IV) is probably due to rapid
renal clearance of IL- 1( (37), in view of the ability of intravenous rhIL- 1(3 or rmIL- 1 a continuously infused at lower doses
to suppress food intake. The inability of acetaminophen,
which preferentially inhibits brain cyclooxygenase relative to
that in peripheral tissues (30, 31), to block rhIL-lB anorexia
supports a peripheral mechanism. The inability of intravenous
ibuprofen to completely block the anorexic effect of prolonged, low dose rhIL- 1( infusion also suggests that other
mechanisms besides PGE2 production may be involved in this
setting, although other explanations are possible since PG production was not measured. These questions require further
study.
The finding that normal aging decreases sensitivity to the
anorexic effect of rhIL-1( is of interest. One might have expected a greater effect on food intake in these old, presumably
less resilient animals. The fact that the opposite was observed
indirectly supports the IL- 1(3 effect as being part of a "designed" physiologic response, which is attenuated with aging,
in contrast to a model wherein anorexia is a "nonspecific"
effect of illness.
N. Meydani, M. Meydani, K Wu, and C. A. Dinarello
By analogy to TNF, the possibility needs to be considered
that IL- 1 induced suppression of food intake is a toxic effect
rather than a primary anorexigenic effect. Daily administration of recombinant human TNF-a to rodents (12) initially
caused acute gastrointestinal inflammation, with edema and
hemorrhage, followed by rapid recovery within 24-48 h (tolerance). Food intake and weight loss followed a similar time
course, with an acute effect followed by rapid tolerance (12).
However, IL- 1 anorexia is unlikely to be due to acute gastrointestinal toxicity, for several reasons. First, extensive toxicity
studies in mice and rats using 100-fold higher IL-l doses than
were used here have failed to reveal any organ toxicity (Dinarello, C. A., unpublished observations). In fact it is quite difficult to kill a nonadrenalectomized rodent with massive doses
of IL- 1, in contrast to TNF. Second, one would expect gastrointestinal toxicity to be worsened by prostaglandin inhibition,
rather than ameliorated, since PGE2 is cytoprotective in the
gut (49, 50). Finally, the anorexia and slowing of body weight
gain that we observed were not restricted to the first 24-48 h of
IL-l administration, but persisted for at least 7 d (Figs. 3 and
4), and withholding of IL- 1 for a day during chronic administration immediately restored food intake to normal (not
shown).
The effects of chronic rmIL- 1 a administration are notable
for several reasons. The issue of tachyphylaxis to cytokine anorexia is central if a physiologic role for cytokines in the anorexia of chronic disease is to be considered tenable. Our observations indicate that tachyphylaxis to IL- 1 anorexia (attenuation of anorexigenic effects and return of some weight gain)
does occur, but equally importantly show at least a 7-d effect
for the anorexia at a constant rmIL- 1 a dose (Fig. 3) and a net
lower weight in the IL- 1 treated rats compared to controls after
several weeks (Fig. 4). Thus, IL- 1 by itself has long-term nutritional consequences when administered chronically, in this
model. Also, the dietary fat effects observed in acute food
intake experiments were reproduced for at least the first 7 d of
treatment, supporting the physiologic relevance of the acute
data. The mechanism for tachyphylaxis was not established by
these studies (e.g., development of antibodies to the recombinant IL- 1, a well-documented event even in the same species,
receptor down-regulation, metabolic adaptations, etc.). The
question whether synergy with other cytokines (51, 52) would
overcome this attenuation is currently under investigation.
Placed in the larger picture, these findings add support to
the concept that nutrients and inflammation are intimately
intertwined, in a complex bidirectional fashion mediated by
cytokines (Fig. 5). Thus, dietary protein and source of fatty
acids may alter IL-1 release (53, 54), dietary fats may alter
SUBSTRATE
AVAILABILITY
( NUTRITION )
PROTEIN
CYTOKINE RELEASE
FAT
CYTOKINE ACTIONS
CHO
L
-
ANOREXIA
ALTERED METABOLISM:
-
NITROGEN WASTING
-
PYRUVATE OXIDATION
-
FAILURE OF KETOSIS
-
HYPERLIPIDEMIA
Figure 5. Cytokine-mediated interactions between nutrients and inflammation. See text for details.
ml
cytokine actions (see above), carbohydrates may in theory be
involved as signals in their anorexic effects (18, 42-46), and
inflammatory mediators not only alter metabolism of all
classes of substrates (47, 55-57) but reduce their overall intake
(cause anorexia). Inclusion of IL-1 anorexia in this scheme is
now necessary. From a clinical perspective, perhaps most interesting is the ability of a dietary supplement to diminish IL- 1
anorexia. This would be a simple and safe intervention in
patients experiencing (or expected to experience) anorexia and
weight loss in association with an inflammatory illness or injury. If IL-l is involved in the human syndrome and acts
through PGE2 generation, as is the case in the rat, therapeutic
trials of F.O. supplementation may be rewarding. By analogy
with F.O. supplementation used for other clinical purposes
(58) it is unlikely that humans would need to ingest a diet with
80% of the fat as F.O. to test the hypothesis. Human volunteers
adding 3 g/d of EPA in F.O. capsules to their otherwise normal
western diets demonstrated a 50% reduction in IL- 1 production when their blood leukocytes were stimulated ex vivo (59).
From an experimental point of view, the importance of species
source and structural form of IL- 1 used, dose, chronicity, and
route of administration, feeding regimen, dietary fat source
and age of animals in determining the effect of IL- 1 on food
intake needs to be taken into account and specified in future
studies.
Acknowledgments
The authors are grateful to Dr. J. Dupont of Iowa State University and
M. M. Mathias of Colorado State University for providing PGE2 antibody, to Dr. J. D. Davies of R. P. Scherer Co., Troy, MI for supplying
MaxEPA, to Denise Cesar and Paul Bizinkauskas for technical assistance, and to Mark Hudes for statistical consultation.
Supported by grant R87SF091, provided by the State of California
and allocated on the recommendation of the University of California
Universitywide Task Force on AIDS (M. Hellerstein), USDA Contract
53-3K06-5-10 (S. Meydani and M. Meydani), and National Institutes
of Health grant Al 15614 (C. Dinarello).
References
1. Beisel, W. R. 1975. Metabolic response to infection. Annu. Rev.
Med. 26:9.
2. Hornstein, I. 1977. Impact of infection on nutritional status
(Symposium). Am. J. Clin. Nutr. 30:1206-1210.
3. Theologides, A. 1979. Cancer cachexia. Cancer 43:2004-2012.
4. Theologides, A. 1986. Anorexins, asthenins and cachectins in
cancer. Am. J. Med. 81:696-698.
5. Kotler, D. P., J. Wang, and R. N. Pierson. 1985. Body composition studies in patients with acquired immunodeficiency syndrome.
Am. J. Clin. Nutr. 42:1255-1264.
6. Chlebowski, R. T. 1985. Significance of altered nutritional status
in AIDS. Nutr. Cancer. 7:85-91.
7. Serwadda, D., N. K. Sewankambo, J. W. Carswell, A. C. Bayley,
R. S. Tedder, R. A. Weiss, R. D. Mugerwa, A. L. Wegaba, G. B. Kirya,
R. G. Downing, S. A. Clayden, and A. G. Dalgleish. 1985. Slim disease:
a new disease in Uganda and its association with HTLV-III infection.
Lancet. ii:849-852.
8. Dinarello, C. A., J. G. Cannon, J. W. Mier, H. A. Bernheim, G.
LoPreste, D. L. Lynn, R. N. Love, A. C. Webb, P. E. Auron, R. C.
Reuben, A. Rich, S. M. Wolff, and S. D. Putney. 1986. Multiple
biological activities of human recombinant interleukin 1. J. Clin. In-
vest. 77:1734-1739.
9. Beutler, B., and A. Cerami. 1987. Cachectin: more than a tumor
necrosis factor. N. Engl. J. Med. 316:379-385.
10. Cerami, A., Y. Ikeda, N. LeTrang, P. J. Hotez, and B. Beutler.
Interleukin-l Anorexia in the Rat
233
1985. Weight loss associated with an endotoxin induced mediator
from peritoneal macrophages. The role of cachectin (tumor necrosis
factor). Immunol. Lett. 11: 173-177.
11. Oliff, A., D. Defeo-Jones, M. Bover, D. Martinez, D. Kiefer, G.
Vuocolo, A. Wolfe, and S. H. Socher. 1987. Tumors secreting human
TNF/cachetin induce cachexia in mice. Cell. 50:555-563.
12. Patton, J. S., P. M. Peters, J. McCabe, D. Crase, S. Hansen,
A. B. Chen, and D. Liggit. 1987. Development of partial tolerance to
the gastrointestinal effects of high doses of recombinant tumor necrosis
factor a in rodents. J. Clin. Invest. 80:1587-1596.
13. Tracey, K., H. Wei, K. Manogue, G. Kuo, S. F. Lowry, and A.
Cerami. 1988. Weight loss, cachexia and inflammation in rats treated
with recombinant cachectin. J. Exp. Med. 167:1211-1227.
14. Dinarello, C. A. 1989. Interleukin-l and its biologically related
cytokines. Adv. Immunol. 44:153-205.
15. McCarthy, D. O., M. J. Kluger, and A. J. Vander. 1985. Suppression of food intake during infections: is interleukin-l involved?
Am. J. Clin. Nutr. 42:1179-1182.
16. Moldawer, L. L., C. Anderson, J. Gelin, and K. G. Lundhold.
1988. Regulation of food intake and hepatic protein synthesis by recombinant-derived cytokines. Am. J. Physiol. 254:G450-G456.
17. Dinarello, C. A., J. G. Cannon, S. M. Wolff, H. A. Bernheim, B.
Beutler, A. Cerami, I. S. Figari, M. A. Palladino, Jr., and J. V. O'Connor. 1986. Tumor necrosis factor (cachectin) is an endogenous pyrogen and induces production of interleukin 1. J. Exp. Med. 163:14331450.
18. Sullivan, A. C., J. Triscari, J. G. Hamilton, and 0. N. Miller.
1973. Effect of (-)-hydroxycitrate upon the accumulation of lipid in
the rat: II. Appetite. Lipids. 9:129-134.
19. Hellerstein, M. K., and Y. Xie. 1987. Effect of 3-mercaptopicolinic acid on food intake in normal, hydroxycitrate treated, vagotomized and hypoglycemic rats. Clin. Res. 35:366a. (Abstr.)
20. Leveille, G. A. 1972. The long-term effects of meal-eating on
lipogenesis, enzyme activity and longevity in the rat. J. Nutr.
102:549-556.
21. Fabry, P. 1967. Metabolic consequences of the pattern of food
intake. In Handbook of Physiology, Section 6, Alimentary Canal. Vol.
1. C. F. Code, editor. Am. Physiol. Soc., Washington, DC. 31.
22. Tepperman, J., and H. Tepperman. 1958. Effects of antecedent
food intake pattern on hepatic lipogenesis. Am. J. Physiol. 193:55-64.
23. Meydani, S. N., A. C. Shapiro, M. Meydani, J. B. MaCauley,
and J. B. Blumberg. 1987. Effect of age and dietary fat (fish, corn, and
coconut oils) on tocopherol status of C57BL6NIA mice. Lipids.
22:345-350.
24. Cathcart, E. S., C. A. Leslie, S. N. Meydani, and K. C. Hayes.
1987. A fish oil diet retards experimental amyloidosis, modulates lymphocyte function and decreases macrophage arachidonate metabolism
in mice. J. Immunol. 139:1850-1855.
25. Hellerstein, M. K., D. J. Greenblatt, and H. N. Munro. 1986.
Glycoconjugates as non-invasive probes of intrahepatic metabolism.
The pathways of glucose entry into compartmentalized hepatic UDPglucose pools during feeding induced glycogen accumulation. Proc.
Nati. Acad. Sci. USA. 83:7044-7048.
26. Powell, S. S. 1980. Rapid extraction of oxygenated metabolites
of arachidonic acid from biological samples using octadecylsilyl silica.
Prostaglandins. 20:947-957.
27. Meydani, S. N., and J. Dupont. 1982. Effect of zinc deficiency
on prostaglandin levels in different organs of the rat. J. Nutr.
112:1098-1104.
28. Larson, K. 1972. Creatinine assay by a reaction-kinetic principle. Clin. Chim. Acta. 41:209-217.
29. Dinarello, C. A., S. 0. Marnoy, and L. J. Rosenwasser. 1983.
Role of arachidonate metabolism in the immunoregulatory function
of human leukocytic pyrogen/lymphocyte-activating factor/interleukin 1. J. Immunol. 130:890-895.
30. Flower, R. J., and J. R. Vane. 1972. Inhibition of prostaglandin
synthetase in brain explains the antipyretic activity of paracetamol
(4-acetaminophenol). Nature (Lond.). 240:410-411.
234
31. Ziel, R., and P. Krupp. 1976. Mechanism of action of antipyretic drugs. In Brain Dysfunction in Infantile Febrile Convulsions.
M. A. B. Brazier and F. Coceani, editors. Raven Press, New York.
141-153.
32. Prickett, J. D., D. E. Trantum, and D. V. Robinson. 1984.
Dietary fish oil augments the induction of arthritis in rats. J. Immunol.
132:725-730.
33. Lee, T. H., R. L. Hoover, J. D. Williams, R. I. Sperling, J.
Ravalese, B. W. Spur, D. R. Robinson, E. J. Corey, R. A. Lewis, and
K. F. Austen. 1985. Effect of dietary enrichment with eicosapentaenoic
acide and docosahexanoic acid on in vitro neutrophil and monocyte
leukotriene generation and neutrophil function. N. Engl. J. Med.
312:1217-1224.
34. Report of The American Institute of Nutrition. Ad Hoc Committee on Standards for Nutritional Studies. 1977. J. Nutr. 107:13401349.
35. Sprecher, H., A. C. Voss, M. Careaga, and C. Hadjiagapiou.
1987. Interrelationship between polyunsaturated fatty acid and membrane lipid synthesis. In Proceedings of the American Oil Chemists'
Society (AOCS) Short Course on Polyunsaturated Fatty Acids and
Eicosanoids. W. E. M. Lands, editor. AOCS, Champaign, IL. 154-168.
36. Shock, N. W. 1972. In Nutrition in Old Age. L. A. Carlson,
editor. Almquist and Wiksell, Uppsala. 12-23.
37. Dinarello, C. A., T. Ikejima, S. J. C. Warner, S. F. Orencole, G.
Lonneman, J. G. Cannon, and P. Libby. 1987. Interleukin 1 induces
interleukin 1. 1. Induction of circulating interleukin 1 in rabbits in
vivo and in human mononuclear cells in vitro. Am. Assoc. Immunol.
139:1902-1910.
38. Okusawa, S., J. A. Gelfand, T. Ikejima, R. J. Connolly, C. A.
Dinarello. 1988. Interleukin I induces a shock-like state in rabbits.
Synergism with tumor necrosis factor and the effect of cyclooxygenase
inhibition. J. Clin. Invest. 81:1162-1172.
39. Michie, H. R., K. R. Manogue, D. R. Spriggs, A. Revhaug, S.
O'Dwyer, C. A. Dinarello, A. Cerami, S. Wolff, and D. Wilmore. 1988.
Detection of circulating tumor necrosis factor after endotoxin administration. N. Engl. J. Med. 318:1481-1486.
40. McCarthy, D. O., M. J. Kluger, and A. J. Vander. 1985. Effect
of centrally administered interleukin-1 and endotoxin on food intake
of fasted rats. Physiol. Behav. 36:745-749.
41. VanMiert, A. 1980. Fever and gastric function. In Fever. J. M.
Lipton, editor. Raven Press, New York. 57-70.
42. Van Itallie, T. B., N. S. Smith, and D. Quartermain. 1977.
Short-term and long-term components in the regulation of food intake: evidence for a modulatory role of carbohydrate status. Am. J.
Clin. Nutr. 30:742-757.
43. Sullivan, A. C., and R. K. Gruen. 1985. Mechanisms of appetite modulation by drugs. Fed. Proc. 44:139-144.
44. Langhans, W., U. Damaske, and E. Scharrer. 1985. Different
metabolites might reduce food intake by the mitochondrial generation
of reducing equivalents. Appetite. 6:143-152.
45. Russek, M. 1981. Current status of the hepatostatic theory of
food intake. Appetite. 2:137-143, 157-162.
46. Baile, C., W. M. Zinn, andJ. Mayer. 1970. Effects of lactate and
other metabolites on food intake of monkeys. Am. J. Physiol.
219:1606-1613.
47. Feingold, K. R., and C. Grunfeld. 1987. Tumor necrosis factor
-alpha stimulates hepatic lipogenesis in the rat in vivo. J. Clin. Invest.
80:184-190.
48. Lumpkin, M. D. 1987. The regulation of ACTH secretion by
IL-I. Science (Wash. DC). 238:452-454.
49. Robert, A. 1975. An intestinal disease produced experimentally
by a prostaglandin deficiency. Gastroenterology. 69:1045-1047.
50. Robert, A. 1979. Cytoprotection by prostaglandins. Gastroenterology. 77:761-767.
51. Movat, H. Z., C. E. Burrowes, M. I. Cybulsky, and C. A.
Dinarello. 1987. Acute inflammation and a Schwartzman-like reaction induced by interleukin-1 and tumor necrosis factor. Synergistic
M. K Hellerstein, S. N. Meydani, M. Meydani, K Wu, and C. A. Dinarello
action of cytokines in the induction of inflammation and microvascular injury. Am. J. Pathol. 129:463-476.
52. Elias, J. A., K. Gustillo, W. Baeder, and B. Freundlich. 1987.
Synergistic stimulation of fibroblast prostaglandin production by recominant interleukin I and tumor necrosis factor. J. Immunol.
138:3812-3816.
53. Hoffman-Goetz, L., and M. J. Kluger. 1979. Protein deficiency:
its effects on body temperature in health and disease states. Am. J.
Clin. Nutr. 32:1423-1427.
54. Editorial. 1983. Protein malnutrition and leukocyte endogenous mediator. Nutr. Rev. 41:242-244.
55. Kampschmidt, R. F. 1981. Leukocyte endogenous mediator/
endogenous pyrogen. In The physiologic and metabolic responses of
the host to infection and inflammation. M. C. Powanda and P. G.
Canonico, editor. Elsevier/North Holland, Amsterdam. 55-74.
56. Neufeld, H. A., J. A. Pace, and F. E. White. 1976. The effect of
bacterial infections on ketone concentrations in rat liver and blood and
on free fatty acid concentrations in rat blood. Metab. Clin. Exp.
25:877-884.
57. Tredget, E. E., Y. M. Yu, S. Zhong, R. Burini, S. Okusawa, J. A.
Gelfand, C. A. Dinarello, U. R. Young, and J. F. Burke. 1988. Role of
interleukin 1 and tumor necrosis factor on energy metabolism in rabbits. Am. J. Physiol. 255:E760-E768.
58. Kremer, J. M., W. Jubiz, A. Michalek, R. Rynes, L. Bartholomew, J. Bigaouette, M. Timchalk, D. Beeler, and L. Lininger. 1987.
Fish-oil fatty acid supplementation in active rheumatoid arthritis. A
double-blinded, controlled cross-over study. Ann. Intern. Med.
106:497-503.
59. Endres, S., R. Ghorbani, V. E. Kelley, K. Georgilis, G. Lonnemann, J. W. M. Van Der Meer, J. G. Cannon, T. S. Rogers, M. S.
Klempner, P. C. Weber, E. J. Schaefer, S. M. Wolff, and C. A. Dinarello. 1989. The effect of dietary supplementation with n-3 polyunsaturated fatty acids on the synthesis of interleukin- I and tumor necrosis
factor by mononuclear cells. N. Engl. J. Med. 320:265-271.
Interleukin-J Anorexia in the Rat
235