Role of interleukin-1 and tumour necrosis factor in leukocyte recruitment to acute dermal inflammation

The cytokines IL-1 and TNF-α are involved in inflammation and their production is stimulated by various agents, especially endotoxin (LPS). Here, using the human IL-1 receptor antagonist (IL-1RA) and a new monoclonal antibody (mAb 7F11) to rabbit TNF, the role of endogenous IL-l and TNF production in acute (3h) leukocyte (PMNL) recruitment to dermal inflammation in rabbits has been studied. IL-1RA inhibited by 27% the PMNL accumulation in reactions induced by killed Escherichia coli (p < 0.05) but not by LPS. The monoclonal antibody to TNF inhibited by 27% and 38% (p < 0.002) the PMNL accumulation in LPS and E. coli reactions respectively, but a combination of the mAb with IL-1RA was not more effective. Treatment of human umbilical vein endothelium with LPS for 3 h activated endothelium to induce PMNL transendothelial migration in vitro, which was not inhibited by IL-1RA, antibody to TNF-α, IL-1 or to IL-8. In conclusion, TNF and IL-1 may partially mediate acute PMNL infiltration in vivo to LPS and Gram negative bacteria, but there is a major IL-1/TNF independent mechanism, at least in dermal inflammation, which may be due to direct LPS activation of the microvasculature or perhaps the generation of cytokines other than IL-1 and TNF.


Introduction
The acute inflammatory reaction is usually characterized by rapid, early (1-2 h) infiltration of the tissue by polymorphonuclear leukocytes (PMNL) and an increase in vascular permeability. Subsequently, vascular and tissue injury may result. A host of mediators and mediator systems have been implicated in this response, including vasoactive amines, lipids and proteins, some of which are generated by activation of cascade systems such as the complement and coagulation pathways. 2 More recently, some members of the cytokine family of proteins have been found to have inflammatory properties. [3][4][5][6] Among these, interleukin-1 (IL-1) and turnout necrosis factor cz (TNF) are synthesized rapidly, especially by monocytes, resident tissue macrophages and to a lesser extent by PMNL, lymphocytes, and connective tissue cells. 7,s Both IL-1 and TNF can subsequently active other cells, such as vascular endothelium to become adhesive for PMNL and mononuclear leukocytes, and to induce the synthesis of cytokines directly chemotactic for leukocytes such as IL-8 and monocyte chemotactic protein-1 (MCP-1).6'9 Among the cytokines, IL-1 and TNF-oq are likely the earliest cytokines in the cytokine cascade to be synthesized during acute inflammatory reactions. It is now well recognized that vascular and tissue injury during acute inflammation is in part dependent on the invasion of the involved tissues by PMNL. As part of on-going investigations into the mechanisms regulating PMNL migration into inflammation, an acute dermal inflammation model in rabbits has been used in this study. The role of endogenously produced IL-1 and TNF in the acute PMNL accumulation in dermal inflammatory reactions induced by a variety of agents has been investigated. For this purpose, recombinant human IL-1 receptor antagonist (IL-1RA) 1'11 and a new monoclonal antibody (mAb) which neutralizes rabbit TNF, were employed to block the activity of endogenously produced rabbit IL-1 and TNF. The results indicate that of the reactions studied, TNF in particular, contributes significantly but only to a minor extent, to the PMNL infiltration during killed Escherichia co/i and endotoxin (LPS) induced inflammation.

Materials and Methods
Dermal inflammatory reactions: New Zealand White rabbits of either sex (3-4 kg) had their backs shaved the day before experiments. On the day of study, 51Cr labelled blood leukocytes (see below), were injected i.v. and up to 56 skin sites were injected intradermally (i.d.) (0.2ml) with inflammatory stimuli in a random pattern using 30-gauge needles as described previously, az'13 Three hours later, the A. C. Issekutz, N. Lopes and T. B. Issekutz animal was sacrificed with an intravenous (i.v.) overdose of pentobarbital, the skin was removed, and the lesions punched out using a 16 mm leather punch. Radioactivity in the skin samples was determined in an LKB gamma spectrometer.
Inflammatory stimuli: Recombinant human interleukin-1 (IL-I), which had a specific activity of 4 x 107U/mg, was a gift from Dr D. Urdal (Immunex Corp., Seattle, WA, USA). Recombinant human tumour necrosis factor (TNF-) (specific activity of 5 x 107U/mg) was a gift from Genentech Inc. (South San Francisco, CA, USA).
All of these cytokines contained <1 ng of endotoxin/mg. Each of the cytokines were diluted immediately before use in 0.1% endotoxin-free human serum albumin (Connaught Labs., Don Mills, Ontario, Canada) in phosphate buffered saline (HSA-PBS). E. coli 0111 lipopolysaccharide (LPS) was from List Biologicals (Campbell, CA). E. coli 018.K1HT, an isolate from an infant with meningitis, was grown in broth culture, killed with formaldehyde, washed, and the number of bacteria were estimated spectrophotometrically as described previously, a2 Zymosan activated plasma (ZAP) was generated as previously a4 by incubation of heparinized (5 U/m1) rabbit plasma for 60 rain (37C) with 5 mg/ml Zymosan A (Sigma Chemical Co., St. Louis, MO, USA) followed by removal of zymosan by centrifugation.
The reversed dermal Arthus reaction was induced by i.v. injection of 10 mg human pyrogen free IgG (Sandoglobulin, Sandoz Pharmaceuticals, Dorval, Quebec, Canada), 1 h prior to i.d. injection of heat inactivated hyperimmune rabbit antiserum to human IgG. a3'a5 Rabbit TNF-0 was purified from rabbit macrophage culture supematants. Briefly, rabbit peritoneal macrophages were harvested, following pentobarbital euthanasia, by peritoneal lavage with 300ml Tyrodes solution (2U/ml heparin), 5 days following i.p. injection of 35 ml sterile light mineral oil. a6 The cells were adhered in 175cm / tissue culture flasks (Falcon, Fisher Scientific Co., Dartmouth, Nova Scotia, Canada) for 2 h in RPMI-1640 (Gibco, Grand Island, NY, USA) 5% heat inactivated rabbit serum, then rinsed to remove non-adherent cells. The adherent cells, which were > 95% macrophages, were stimulated with LPS (30 ng/ml) in RPMI-1640 for 6 h at 37C in 5% CO2. The supematant was then harvested, dialysed against water, concentrated by lyophilization 100-200-fold and applied to a Superose-12 FPLC column (Pharmacia Fine Chemicals, Dorval, Quebec) equilibrated with 0.3 M NaC1-0.01 M Na2HPO4, pH 7. Fractions eluting between 40 and 50 kilodaltons (kDa), were rich in TNF activity as assayed by cytotoxicity for the TNF sensitive L929 mouse fibroblast line. a7 The fractions which eluted between 12 and 20 kDa were devoid of IL-1 activity. 8 Anticytokine reagents: Highly purified human recombinant interleukin-1 receptor antagonist (IL-1RA) was a kind gift from Dr R. C. Thompson (Synergen Inc., Boulder, CO). The rat mAb 7Fll, which neutralized rabbit macrophage TNF-0 and lymphocyte derived TNF-fl bioactivity, was rat isotype IgGa according to a commercially available isotyping kit (Serotec Canada, Toronto, Ontario). This antibody was generated by immunization of AO strain male rats with partially purified TNF-z (40-50 kDa Superose-12 fractions). The immunization protocol consisted of covalently coupling 3-5 #g of protein to 1 cm 2 pieces of the APT form of Transabind paper (Schleicher and Schuell Ltd, Keene, NH, USA) according to the instructions of the manufacturer. The immobilized antigen was introduced surgically under anaesthesia into the peritoneal cavity of the rats every 10-14 days together with i.p. injection of polyadenosinepolyuridine (800/tg) (Sigma Chemical Co.) as an adjuvant, a9'2 After a total of four such immunizations, the animals had neutralizing activity in their serum for rabbit TNF-. After a rest period, one rat was booster immunized by intrasplenic injection of 5 g of partially purified antigen. Four days later the spleen was removed, and splenocytes were fused with the P3U1 mouse myeloma using polyethylene glycol 1450 (Merck Frosst, Montreal, Quebec) according to standard techniques. 21 Following fusion, the lymphocytes were plated into flat bottom 96-well microtitre plates (NUNC, Gibco) at a density of 2 x 10 cells/well in Dulbecco Modified Eagle's Medium (DMEM; Gibco) with nonessential amino acids, Na pyruvate and 10% foetal calf serum (FCS; Hyclone, Logan, UT, USA), into wells containing a mixture of 5 x 105 irradiated mouse thymocytes and splenocytes prepared 24 h previously as feeder cells. 2a Supernatants from hybridoma-containing wells were screened for neutralization of the cytotoxic activity of rabbit TNF-0q and one of the positive hybridomas, 7Fll was cloned by limiting dilution. This mAb neutralized both rabbit macrophage derived TNF-z and mitogen (Con A and PHA). stimulated rabbit lymphocyte derived TNF activity, presumably TNF-/. 7 However, mAb 7Fll was specific for rabbit TNF in that it did not neutralize mouse, rat or human TNF0 activity. The mAb was purified by affinity chromatography on Protein G Sepharose from cells grown in serum free medium (DMEM supplemented with 2% Nutridoma, Boehringer Mannheim, Montreal, Quebec). Binding of the IgG was at pH7 in PBS and elution was with 0.1 M glycine-HC1, pH 3.5 followed by pH 2.7 buffers, as recommended by the manufacturer (Pharmacia Fine Chemicals, Dorval, Quebec). The purified mAb preparation had a titre of 30000 neutralizing units/ml. Polyclonal rabbit antibody to human IL-10 and IL-1/ and to human TNF-0 were kind gifts from Dr C. Dinarello

Results
Effect of IL-1 receptor antagonist and antibody to TNF on derma/PMNL infiltration: In initial experiments the eflqcacy of recombinant human IL-1 receptor antagonist (IL-1RA) and the mAb to rabbit TNF for blocking the in vivo activity of IL-1, and rabbit TNFfor inducing PMNL accumulation in skin sites were tested. As can be seen in Fig. 1 We next investigated the effects of IL-1RA and mAb to TNF on the accumulation of PMNL in skin sites in the rabbits injected with E. coli LPS or killed E. coil bacteria. As can be seen on the lower panel in Fig. 1, the leukocyte accumulation induced by LPS was not significantly inhibited by the co-injection of IL-1RA, but was significantly inhibited by mAb to rabbit TNF. The combination of IL-1RA with mAb to TNF did not result in greater inhibition than did mAb to TNF alone. In response to i.d. injection of killed E. coli, both IL-1RA and TNF mAb significantly inhibited PMNL accumulation. The combination of IL-1RA with anti-TNF did not inhibit to any greater extent than either alone. In these experiments PMNL accumulation was measured during the first 3 h after i.d. injection of the skin sites, since it has been shown previously that maximum PMNL accumula-tion occurs during this time in these types of lesions. 12 All combinations of agents were studied in the same rabbit with triplicate or quadruplicate skin sites being injected and statistical analysis was by paired t-test. The dose of IL-1RA and mAb to TNF were optimized for inhibitory activity and the level of inhibition with IL-1RA did not increase upon increasing the dose of IL-1RA to 1000 rig/site or by increasing the mAb dose to 2500 NU per site.
The effect of IL-1RA and mAb to TNF on three other inflammatory reactions was also studied, namely that induced by i.d. injection of zymosan particles, by the reversed passive Arthus reaction and by ZAP, the active principle of which is CSade Arg. 14 As shown in Fig. 2 have shown that this activated endothelium not only promotes PMNL adhesion but also PMNL transendothelial migration in vitro. 9'23 Gram negative bacterial LPS has also been show to activate endothelium for adhesion of PMNL, but the role of endothelium generated cytokines such as IL-1 or TNF, which could directly activate the endothelium, is not clear. 9 Therefore, the effect of LPS on PMNL interaction directly with vascular endothelial cells was studied using an in vitro system which allows quantitation of PMNL transendothelial migration. Endothelium stimulated with LPS for 3 h resulted in PMNL transendothelial migration as shown in Fig. 3, even though the LPS was removed from the system. An LPS concentration of 0.3 ng/ml was sufficient to induce a maximal response. These results are comparable to previous findings using this system with IL-1 or TNF0 to activate the endothelium, i.e. 2-4 h of stimulation was optimal for observing PMNL transendothelial migration and periods less than 2 h or longer than 8 h resulted in a decrease in the migration response 23 (and not shown). In contrast to the 3 h required to activate endothelium with LPS for supporting this transendothelial migration, PMNL migration across the activated endothelium was rapid, plateauing between 60 to 90 min (not shown) as it was with IL-1 and TNFinduced PMNL migration reported previously. 3 There was no migration of PMNL in response to any of the concentrations of LPS, in the absence of an endothelial monolayer 24 (and not shown).
It has been shown previously that endothelium can be activated to secrete cytokines such as IL-1, IL-8 and possibly TNF. 6'9 Therefore, the possibility was examined that the observed effects of LPS may have been due to the endogenous production of one of these cytokines by the endothelium and secondary activation of the endothelium in an autocrine fashion for supporting PMNL migration. Table I shows the effects of IL-1RA, and antibodies to IL-1, IL-113, TNFor IL-8 on LPS stimulated PMNL transendothelial migration. The IL-1RA and antibodies were present during the 3 h LPS stimulation and also during the 75 min PMNL migration, to ensure maximal neutralization of any endogenously produced IL-1, TNF or IL-8.  Although the antibodies to TNF-, IL-1, IL-8 and IL-1RA each effectively blocked (>95%) PMNL migration induced by the corresponding exogenously applied cytokine (not shown), the LPS induced activation of endothelium for PMNL transendothelial migration was not significantly inhibited by any of the agents even in combination.

Discussion
The major findings in this study are that IL-1RA and mAb to TNF inhibited PMNL accumulation in inflammatory reactions induced in the dermis by killed E. coli, mAb to TNF inhibited the LPS reaction but, even in combination, these agents only partially inhibited PMNL accumulation in these reactions (Fig. 1). This finding was surprising because IL-1 and TNFinduce PMNL infiltration into the skin upon intradermal injection and they have synergistic effects. 3'4 The results suggest that endogenous production of TNF, at least in dermal inflammation may be more important than IL-1 in LPS and perhaps also in E. coli inflammatory reactions, the latter being partly dependent on LPS shed from the cell wall of the bacteria. 24 Another explanation for our results, namely that IL-1RA may not be active as an antagonist of the rabbit IL-1 receptor, is unlikely because the human 1L-1RA effectively inhibited the PMNL accumulation in rabbit skin induced by i.d. injection of IL-1 and IL-lfl and systemic or local treatment in various species with human IL-1RA has inhibited PMNL infiltration e.g. in LPS induced alveolitis in rats and immune complex colitis in rabbits. 1'25'26 The relative lack of effect of IL-1RA in our experiments, in comparison to these other models is unlikely to be due to the doses employed because concentrations of up to 1000ng per site had no greater inhibitory effect than 200 ng (not shown). Since our mAb neutralizes both and fl species of TNF, it is not possible from these results to conclude that neutralization of TNFwas responsible for the observed inhibition. However, TNF-fl is produced primarily by activated lymphocytes 7 of which there are relatively few in the skin during the acute (< 3 h) phase of inflammation, iv The differences observed in our experiments with IL-1RA and other inflammation models in which 1L-1 appears to play an important role in PMNL recruitment to LPS, for example in peritonitis 28 or alveolitis 2s'29 may be dependent on the tissue and the type of resident cells, especially macrophages available to secrete IL-1, a cytokine which is not readily secreted by all types of macrophages and is released slower or later than TNF-o. 8'3'31 Among skin cells, keratinocytes are known to store IL-1 but apparently do not release this except perhaps upon cellular injury. 32 Langerhans cells have been shown to secrete IL-1 and TNF-oq and mast ceils have recently been shown to secrete preformed stores of TNF-. 33 '34 Secretion of TNFin skin from these cells may occur more rapidly than of IL-1, perhaps accounting for the greater apparent contribution of TNF to PMNL infiltration in the initial 3 h of the reaction. This conclusion is also supported by the observations that TNFis secreted more rapidly (< 1 h) than IL-1 in response to LPS injection systemically 35 or LPS in the cerebrospinal fluid. 36 Additional findings in the dermal inflammation model reported here suggest that neither endogenous production of IL-1 nor TNF contribute to the PMNL infiltration induced by in situ immune complex formation, as in the reversed passive Arthus vasculitis reaction or by zymosan particles (Fig. 2). This is not surprising in that both these reactions are primarily dependent on complement activation 15'3v and likely require the generation of C5aaesArg tO induce PMNL accumulation. The PMNL recruiting activities of ZAP, which is 14 known to be due to C5aae arg, was not inhibited by IL-1RA or by mAb to TNF, as expected and this stimulus served primarily as a negative control and an internal standard.
Finally, the results in Fig. 1 suggest that for LPS and E. coli dermal inflammatory reactions endogenous IL-1 and TNF play minor roles in PMNL recruitment and that additional mechanisms are operative. The killed E. call reaction is complex and involves, in part, complement activation by the cell wall of the bacterium with the generation of C5aa Arg chemotactic activity as well as the release of LPS from the cell wall of the bacterium, which has no direct PMNL chemotactic properties. 14 The response to LPS is also complex. However, two mechanisms can largely be excluded: (a) the LPS used here and at the nanogram concentrations which induce intense inflammatory reactions, has been shown previously not to activate the complement pathway or to generate chemotactic activity for PMNL; 24 and (b) the LPS was not directly chemotactic and did not induce PMNL migration in vitro in the abscence of an endothelium. 24 Endotoxin (LPS) of course is known to be a potent inducer of IL-1 and TNF-o production 4'6-s'35 and via the action of these cytokines could recruit PMNL. In addition, LPS, IL-1 and TNFo all can induce endothelium to produce the PMNL chemotactic cytokine, IL-8, which is also produced by connective tissue cells such as fibroblasts in response to IL-1 or TNF-(but not LPS). 6'8 However, the present findings demonstrate that LPS also can directly activate endothelial cells to promote PMNL transendothelial migration (Fig. 3). This action of LPS, at concentrations <1 ng/ml, does not appear to involve IL-1, TNF-z or IL-8 production by the endothelial cells (Table 1). Such a direct effect of LPS on the microvasculature could, in part explain why only partial inhibition in vivo of PMNL infiltration by the TNF mAb and IL-1RA was observed in these studies and in other LPS induced inflammation models such as in pulmonary inflammation. 25 However, additional cytokines may also be involved in the LPS response such as the 45 kDa protein, previously reported to be produced by LPS stimulated rabbit and human macrophages, which is also active in recruiting PMNL into the tissues. 18 These conclusions of alternate IL-1 and TNFindependent LPS induced mechanisms for PMNL recruitment are supported by most recent observations with LPS induced inflammation in monkey skin, 38 demonstrating that in vivo the endothelium expresses the leukocyte adhesion molecule-E selectin and PMNL infiltration occurs prior to immunohistochemical evidence of IL-1 and TNF<z production.