Immunobiol., vol. 186, pp. 199-213 (1992)

Laboratoire d'Immunopharmacologie Experimentale, CNRS UPR 405, Paris, France

Indirect and Selective Down-Regulation of Serum Tumor Necrosis Factor-a Release by Interleukin-l~ CAROLE LE CONTEL, FRANCINE PARANT,

and MONIQUE P ARANT

Received December 30, 1991 . Accepted in Revised Form March 20, 1992

Abstract A selective inhibition of LPS-induced tumor necrosis factor-a (TNF) response in mice was caused by an injection of recombinant human interleukin-l (IL-l). The decrease in serum TNF level reached 70 to 80 percent of the controls receiving LPS alone when IL-l was given simultaneously or prior to the challenge. At the same time serum IL-6 release was more elevated. Ex vivo assays have s hown that macrophages from IL-l treated animals did not respond to LPS when stimulated immediately after harvesting but recovered their normal responsiveness after being cultured for 2 hou rs and then washed. In vitro with or without addition of IL-I , mouse elicited macrophages responded equally to LPS in releasing TNF. In the absence of adirect and lasting effect onTNF-producing cells, the host reaction responsible for the inhibitory effect of IL-I could be related to the overproduction of corticosterone that occurred after IL-l injection, since it was not observed in adrenalectomized animals. Indeed the blockade of corticoid secretion by indomethacin prevented the inhibition of TNF production induced by IL-l administrati on before LPS challenge. TNF administration did not result in elevation of corticosterone level and in contrast to IL-l enhanced the TNF response to LPS injection. In vitro and ex vivo assays have shown this enhanced response to LPS was linked to a direct and prolonged effect of TNF on TNF-producing ceUs. Muramyl dipeptide (MOP) which was used as a known priming agent for enhanced cytokine release had a similar effect on TNF-producing cells. Under conditions which resulted in an increase in TNF titer in the blood, the level of IL-6 also was elevated but experiments with mice given IL-I indicate that the release of IL-6 was not dependent o n TNF production.

Introduction It has been established that cytokines are essential in maintammg homeostasis especially in regulating the immune system. Overproduction of cytokines has been associated with some pathologic states (1, 2) suggesting a strict control of the magnitude of the production under normal conditions. A cytokine cannot only induce the production of a cascade of other cytokines but also reinforce or inhibit one another. In this intricate Abbreviations: TNF = tumor necrosis factor; IL = interleukin; TGF = transforming growth factor; LPS = lipopolysaccharide; Adx = adrenalectomized; MDP = muramyl dipeptide; HPA = hypothalamic-pituitary-adrenal; CRH = corticotropin-releasing hormone; ACTH = adrenocorticotropic hormone

200 . C.

LE

CONTEL, F. PARANT, and M. PARANT

network, tumor necrosis factor-a (TNF-a), a pleiotropic peptide secreted mainly by activated mononuclear phagocytes, was considered to play a central role in inducing the release of interleukin-1 (IL-1) and interleukin-6 (IL-6) during bacteremia (3). Kinetic studies have demonstrated clear sequences in cytokine production following an injection of bacterial agents such as lipopolysaccharide (LPS), showing that IL-1 and IL-6 release was maximum when the level of TNF has declined (4). A number of studies have shown that TNF production could be inhibited in vitro or in vivo by some cytokines, namely interleukin 4 (IL-4), transforming growth factor-~ (TGF-~) and recently in vitro by IL-10 (5-7). Furthermore, IL-6, a cytokine induced by TNF, displayed an inhibitory effect on TNF production, suggesting a negative feedback loop generated by TNF itself (8,9). In contrast, inoculation of IFN-y into mice resulted in increased yields of serum TNF following an LPS challenge (10). Xenobiotics capable of activating macrophages such as muramyl dipeptide (MDP) have also been shown to induce the release of greater levels of circulating TNF in response to LPS (11). In a previous work, we have reported that recombinant TNF itself displayed a priming effect for an enhanced release of TNF and IL-6 in the blood of mice receiving LPS a few hours later. Moreover, although TNF and IL-1 share many biologic properties, their effect on LPS-induced cytokine production was opposite. Administration of IL-1 before LPS resulted in depression of serum TNF whereas the release of IL-6 was not affected (12). The present study was designed to evaluate the influence of in vivo pretreatment with IL-1 or TNF on TNF-producing cells. The results indicate that the priming effect of TNF or MDP was related to the modulation of cell function that could be demonstrated in ex vivo experiments. The depression in LPS-induced TNF response caused by pretreatment with IL-1 was indirect and seemed to be associated with an overproduction of serum corticosterone, indicating that, like IL-6, IL-1 may provide a negative feedback signal in the regulation of TNF production.

Materials and Methods Mice

Pathogen-free OFI female mice (2 months old) were purchased from Iffa Credo (SaintGermain sur l'Abresle, France). Adrenalectomy was performed under ether anesthesia 2 days before the experiment. Adrenalectomized mice (Adx) and sham-operated controls were given 1 % NaCI solution to drink. Reagems Salmonella enteritidis LPS extracted by the phenol water procedure was obtained from Difco Laboratory (Detroit, MI, USA). MDP (NAcMur-L-Ala-D-iGln) was kindly provided by Vacsyn (Paris, France). They were injected by the intravenous route. Indomethacin (Idm) (Sigma Chemical Co, St. Louis, MO, USA) was dissolved in NazCO) (\ mg/ml), diluted in saline and injected by the intraperitoneal route at 300 fLg/0.5 ml. Human recombinant TNF-a (rHuTNF) was purchased from Biotrans (Los Angeles, CA, USA). Under our experimental

Endogenous Corticoids and Cytokine Regulation . 201 conditions its specific activity was 1.6 x 107 U / mg. The specific activity o fmurine recombinant TNF-a (rMuTNF, Biogent) was 1.06 X 108 U /mg. Human recombinant IL-la and IL-1/3 were kindly supplied by Roussel Uclaf Research Center (Romainville, France) and had the same specific activity of about 107 U/mg. Murine recombinant IL-6 (Innogenetics, Belgium), used as a standard in bioassay, had a specific activity of about 4 X 108 U/mg. Monokines did not show any significant contamination with endotoxin « 0.1 nglmg protein) as determined by the Limulus amebocyte assay. They were injected intravenously. Monoclonal anti-rHuTNF IgGl antibody was kindly supplied by Roussel Uclaf. Rabbit polyclonal IgG raised against murine rTNF-a was a gift of Dr. W . FIERS and did not inhibit murine lymphotoxin or human rHuTNF-a. Corticosterone determination Mice were killed by decapitatio n to collect the blood, and aliquots of serum were stored frozen at -20°C until the assay . Corticosterone level was evaluated by using a radioimmunoassay (RIA) developed by Roussel Uclaf Research Center and performed in the presence of an excess of blocking agent which prevents corticosterone binding to endogenous protein. Serial dilutions of serum samples and of standard corticosterone were incubated with 125 1_ corticosterone and with an antiserum specific for the steroid. The separation of antibodybound and free radiolabeled corticosterone was achieved by using a precipitating second antibody. After overnight incubation and centrifugation, supernatants were decanted by inverting the tubes. The amount of precipitated radioactivity was determined using a gamma counter. Corticosterone concentration in serum samples was determined from the standard curve ranging from 0.04 to 40 ng/m!. Peritoneal ceJJs Mice were injected i.p . with 1.5 ml thioglycollate broth (Diagnostics Pasteur, France) and elicited peritoneal cells were harvested by lavage of the peritoneal cavity 4 days later. Percent of macrophages was evaluated by a-naphthyl non-specific esterase staining and found about 90 %. Cells were suspended in serum-free RPMI 1640 medium, supplemented with Hepes (10mM), glutamine (1 %), gentamycin (50 [l.g/ml), penicillin (100 UUml), streptomycin (50 [l.g/ml). They were cultured in 24-well culture plates (NUNC, Denmark) at 106 cells/well/m!. Total peritoneal cells were incubated for 2 h at 37"C in a 5 % CO 2 atmosphere to obtain adherent cells. Total cells or adherent cells were pretreated for 3 h with rHuIL-l/3 (0.3 [l.g/ml corresponding to 3000 U), rHuTNF-a (0.3 [l.g/ml corresponding to 4800 U), MDP (100 [l.g/ml) or medium, then challenged with LPS during 18 h. In some experiments thioglycollate-treated mice received rHuIL-I/3 (1 [l.g), rHuTNF-a (1 [l.g), MDP (100 [l.g) or saline intravenously three hours before harvesting the cells. After incubation for 18 h with o r without LPS (1 [l.g/ ml), culture supernatant fluid s were collected, and stored fro zen at -20 °C. Cell viabilities were determined by trypan blue exclusion and were > 95 % at the end of the experiments. Cells were then washed extensively wi th RPM I m edium, frozen at -20°C in medium, then plates were thrice frozen and thawed. Cell lysates were harvested, and tested for cell-associated TNF activity. Cytokine bioassays Measurements were performed in supernatant fluids from macrophages or in mouse sera that were de complemented before testin g (12, 13). The TNF activity was measured essentially as described b y RUFF and GIFFORD (14) in the presence of 1 [l.g/m l actinomycin D (Sigma). Briefly, cytokine level was evaluated by a cytotoxic assay performed on L929-a cells line as already described (12). N eutralization with antibody was performed by adding antibody solution to pretested samples that have been adjusted to about 10 U/ml of cytotoxic activity. Then the mixture was added to L929-a cells in 96-well plate. Monoclonal and polyclonal unrelated antibodies were used as controls. IL-6 level was measured by a tritiated thymidine uptake assay using the specific IL-6-dependent hybridoma cell line 7TDI. This mouse-mouse hybrid developed at the Ludwig Institute for Cancer Research was kindly provided by Dr. VAN SNICK (Brussels, Belgium) and did not show any thymidine incorporation under IL-l

202 . C.

LE CONTEL,

F. PARANT, and M. PARANT

stimulation (IS). The specificity of bioassays was confirmed with the cytokines used in the study. All results were expressed in units/m!.

Results

Extent of IL-l-induced depression in serum TNF release after an LPS challenge We have already reported that administration of IL-1 to mice three hours before LPS injection markedly decreased the yield of serum TNF bioactivity but did no~ modify the time course of release. Thus, the peak of serum TNF production was always found 1.5 h post LPS and TNF activity was nearly undetectable by 3 hours (12). The following experiment was designed to d~termine TNF and IL-6 activity in mice receiving 25 I-lg LPS at various time intervals after a single intravenous injection of rHuIL-1~. In

A

1000

E

"::J u..

Z

500

f-

E OJ

L

'"

(/)

a

Saline

0

3

5

8

20

hours

Time of rHulL -1 pretreatment before the LPS challenge

B

15

...,

~

..

I 0

CE

10

"::J to

I =o!

5

E

.,~

Vl

a Saline

0

3

5

8

20

hours

Time ,of rHulL-l pretreatm ent before the LPS challenge

Figure 1. Influence of rHuIL-I pretreatment on LPS-induced release in the blood ot TNl:' (A) or IL-6 (B). IL-ll3 (l 'llg) was injected i.v. at various times prior to an LPS (25 Ilg) i.v. challenge to groups of 5 mice. Serum was collected 1.5 h later and each bar represents the mean ± SD. ". = P < 0.05, "." = P < 0.01 using Student's t test compared to saline-treated groups.

Endogenous Corticoids and Cytokine Regulation . 203

the assay reported in Figure 1, blood was collected 1.5 h after LPS injection, time which was found to correspond to the peak of TNF response in the various groups. TNF values obtained with the bioassay show that a marked reduction (about 70 %) was observed in the titer of circulating TNF activity upon LPS stimulation when mice were treated with IL-1 either simultaneously or 3-5 h before the challenge. A significant degree of reduction (40 %) was still shown when the time interval of IL-1 pretreatment was 20 hours (Fig. 1a). Additional assays were performed in mice pretreated with various doses of IL-1 and no clear dose-dependent response was found. With a smaller dose (0.3 !!g/mouse) the TNF level produced by LPS injection was not altered (551 Ulml versus 578 Ulml in LPS controls). With doses larger than 1 !!g, the inhibitory effect was not increased (not shown). IL-6 concentration was determined in the same serum samples although the time of bleeding in the assay did not correspond to the peak of IL-6 production (4). Under our experimental conditions, maximal increase in circulating IL-6 level was found 2.5-3 h post LPS and concentration measured 1.5 h after injection was about 50 % of the peak response (12). As reported in Figure 1b, the yield of IL-6 was higher in IL-1-pretreated groups as compared with controls receiving LPS alone, and particularly when the cytokine was injected simultaneously or 3 h before LPS.

In vitro effect of IL-J on TNF production by mouse macrophages The influence of IL-1 on LPS-induced responses of murine elicited peritoneal macrophages was examined with total or adherent cell population. It was compared with the effect of TNF or MDP. Cells were cultured for 3 h with the various agents before the addition of LPS, and then incubated for an 18-h period. Table 1. Influence of pretreatment on LPS-induced cytokine production by mouse elicited peritoneal cells Pretreatment' in viero

Medium rHuIL-1 rHuTNF MDP Medium rHuIL-1 rHuTNF MDP

Challenge

None None None None LPS LPS LPS LPS

Cytokine level at 18 h (U/ml)b Adherent cells Total cells TNF IL-6 TNF

IL-6