The LPS-induced neutrophil recruitment into rat air pouches is mediated by TNFα: likely macrophage origin

The role of resident cells during the lipopolysaccharide (LPS)-induced neutrophil recruitment into rat air pouches was investigated. In this model, LPS (Escherichia coli, O55: B5 strain; 2–2000 ng) induced a dose– and time-dependent neutrophil recruitment accompanied by the generation of a tumour necrosis factor-α (TNFα)-like activity. Dexamethasone (0.05–5 mug) and cycloheximide (6 ng), injected 2 h before LPS into the pouches, inhibited the neutrophil recruitment and the generation of the TNFα-like activity, while the H1-receptor antagonist mepyramine (1 and 4 mg/kg, i.p., 0.5 h before LPS) and the PAF-receptor antagonist WEB 2170 (0.05 and 1 mg/kg, i.p., 0.5 h before LPS) had no effect. Purified alveolar macrophages (AM) were used to replenish the pouches of cycloheximide-treated recipient rats. AM provided by PBS-treated animals led to the recovery of the LPS-induced neutrophil recruitment and of the TNFα-like formation contrasting with those from cycloheximide-treated animals (1 mg/kg, i.p.). When delivered in situ, liposome-encapsulated clodronate, a macrophage depletor, significantly impaired both the LPSinduced neutrophil recruitment and the TNFα-like activity. An anti-murine TNFα polyclonal antibody (0.5 h before LPS) was also effective. These results emphasize the pivotal role of macrophages for LPS-induced neutrophil recruitment via the formation of TNFα.


Introduction
Leukocyte recruitment 1,2 and the formation of pro-in ammatory mediators, including different cytokines, are the hallmark of an in ammatory response. The latter is characterized by vasodilatation, rapidly followed by neutrophil adhesion to endothelium and migration into the perivascular connective tissue.
In vitro studies have shown that the bacterial endotoxin (lipopolysaccharide, LPS) is not by itself a chemotactic factor, 3 even though LPS may interact with neutrophils via CD14 and the LPS-binding protein (LBP) to express CR3 activity which mediates neutrophil adhesion. 4,5 In fact, local cell targets seem to be more relevant in vivo and in particular, resident macrophages are believed to play a pivotal role in the recognition and the transduction of the effects of LPS (for review, see Manthey and Vogel 6 ) since they produce the chemotactic mediators IL-1a and b, TNFa, IL-8 and MIP-1 and 2 as well as LTB 4 and PAF. Furthermore, LPS elicits a selective and transient cycloheximide-dependent collagenase synthesis by macrophages, providing thus another potential tissue damaging factor. 7 Initially described as a potent in ammatory cytokine derived from LPS-activated macrophages, TNFa may also account for neutropenia, neutrophilia 8,9 and neutrophil recruitment 2,10 -12 and activation. 13,14 TNFa also induces the expression of cell surface molecules, leading to adherence on endothelial cells. 15 As an early cytokine, TNFa is released extracellularly within 15 min after its gene transcription upon exposure to in ammatory stimuli. 16 Its broad spectrum of activities and the amounts produced by macrophages stimulated by various products (up to 2% of their total biosynthesis) suggest that TNFa is an important mediator of LPS-induced in ammatory response. 17 Glucocorticoids such as dexamethasone inhibit the production of circulating TNFa in mice, rats 18 and guinea-pigs 19 treated with LPS. Dexamethasone also alters the phagocytic functions of macrophages in vitro 20 and contributes to protect rat macrophages against LPS-induced TNFa production in vitro, even though less so in vivo. 21 The cutaneous air pouch provides a virtual cavity which can be tailored as a migration chamber covered innerly with a lining membrane, the facsimile synovium. 22 Since the walls of the pouch are formed by macrophages and broblasts and a few mastocytes, we took advantage of this model to study the mode of action of LPS. Because of the potential damage engendered by the neutrophil towards connective tissue, 23 we investigated the dependency of LPS-induced neutrophil emigration upon resident cells and bring evidence that macrophages and TNFa are respectively the target and the mediator of neutrophil recruitment following LPS injection into rat air pouches.

Animals
Brown-Norway rats (200 ±250 g) (Iffa Credo, France) were allowed to take food and drink ad libitum at room temperature.

Rat air pouch
The air pouch was induced according to Edwards et al. 22 At day 0, rats were anaesthetized with ketamine (50 mg/ kg, i.m.) and their dorsum was thoroughly shaved and gently disinfected with ethanol 708 . Syringes and needles were one-purpose material. Twenty ml of sterile air taken under a laminar ux hood were injected subcutaneously with a gauge 26G 3 0.5"; 0·45 3 12 needle; thus, disruption of the underlying cutaneous connective tissue allowed to make an air-full cavity. At day 5, 10 ml of sterile air were injected in similar conditions in order to maintain pouch patency. At day 7, the pouches were injected with LPS (2-2000 ng/ ml) under a volume of 1 ml. Control animals were injected with the same volume of the LPS vehicle, i.e. phosphate buffered saline (PBS) without Ca 2+ nor Mg 2+ .

Neutrophil migration assessment
At various times (2 h, 24 h, 48 h and 96 h) after LPS injection, animals were killed with an overdose of sodium penthiobarbitone. Four ml of heparinized PBS solution (5 IU/ ml) were injected into the pouch in order to wash the cavity by a gentle massage. The washing solu-tion which was recovered at over 95% was collected into 6 ml-polypropylene round bottom tubes (Falcon8 2063, sterile/gamma irradiated) stored on an ice bath.
Injection of the mastocyte degranulating agent compound 48/ 80, under 1 ml (250 mg/ ml) into the rat air pouch followed the same procedures. This concentration was chosen for its ability to induce granulocyte in ltration in the mouse skin. 24 Leukocyte and differential counts were performed with a cell counter (Coulter © ) and a Cytospin (Hettich Universal © ), respectively, and allowed to calculate the total number of leukocyte neutrophils recovered.
One ml aliquots prepared from the pouch washing were centrifuged (400 g, 10 min at 48 C), pellets and aliquoted supernatants being kept at -408 C until further analysis.
An anti-murine TNFa immunoglobulin preparation was prepared as follows. Female HY/ CR rabbits (2500 g; Charles River, St Aubin les Elboeufs, France) were immunized at 2 weeks intervals by three injections of reduced murine rTNFa (Immungenex, Los Angeles, CA, USA) emulsi ed in adjuvant (Hunter Titermax; CytRx Co., Norcross, Germany): the rst one with 50 mg, the second and the third with 25 mg. The animals were bled 2 weeks after the last injec-tion, and total immunoglobulins were obtained after precipitation with 40% saturation of ammonium sulphate. Puri ed murine polyclonal anti-TNFa (2·5 mg/ ml) was administered into the rat air pouches in a volume of 0.7 ml, 0.5 h before LPS. Control animals were treated under same conditions with the same amounts of preimmune immunoglobulins.

Macrophage replenishment
Rats were anaesthetized (sodium penthiobarbitone, 60 mg/ kg, i.p.), the trachea was cannulated and broncho-alveolar lavages (BAL) with sterile saline under a volume of 6 ml were performed. This procedure was repeated until a nal volume of 36 ml was obtained in a 50 mlpolypropylene graduated conical tube (Falcon © 2098, Blue Max, sterile/gamma irradiated) on ice bath. Immediately after centrifugation (400 g, 5 min at 48 C), the cell suspension underwent an hypotonic lysis with sterile water to remove remaining erythrocytes. Then, cells were counted (Counter Coulter © ) and the suspension was diluted to a nal concentration of 10 6 and 10 4 cells per ml. Purity and viability were assessed by differential count and blue trypan dye exclusion, respectively.
In other experiments, rats were treated with cycloheximide i.p. (1 mg/ kg, 0.5 ml). 25 Control animals received similar volumes of saline. After 2 h, the animals were sacri ced with an overdose of penthiobarbitone and both groups underwent the same procedure to purify the alveolar macrophages.

Liposome preparation and experimental design
Multilamellar liposomes were prepared according to Van Rooijen and Van Nieuwmegen. 26 In brief, 75 mg dipalmitoylphosphatidylcholine and 11 mg cholesterol were dissolved in chloroform in a round bottom ask. The thin lm that formed on the walls after rotary evaporation at 458 C was dispersed by gentle shaking for 10 min in 10 ml of PBS (pH 7.4), in order to prepare empty liposomes, and in 10 ml of a solution of 2 g clodronate (dichloromethylene diphosphonate or Cl 2 MDP) in PBS, in order to prepare liposomes with encapsulated Cl 2 MDP. The preparation was kept for 2 h at room temperature and sonicated four times for 5 min at 458 C in a waterbath (50 Hz) and kept at room temperature for a further 2 h. Then, liposomes were ltered through 1.2-mm Minisart NML lters (disposable syringe holders, sterile, pyrogenfree, hydrophilic, Sartorius, Germany) centri-fuged at 100 000 g to 0.5 h, nally resuspended in 5 ml PBS and kept at 48 C.
Rats with 7-day-old air pouches were anaesthetized and injected into this preformed cavity with either 0.3 ml liposome-encapsulated clodronate or 0.3 ml liposome-encapsulated PBS (empty liposome) for 96 h before LPS stimulation as previously described.
In order to overcome the cell counter inability to differentiate the remaining injected liposomes and the LPS-recruited cells, 100 ml aliquots of pouch washing were allowed to stretch on glass slides by cytocentrifugation. After a Diff-Quik ® staining, the slides were observed under light microscope at magni cation 3 1000. A differential leukocyte count was performed (neutrophil, eosinophil, mononuclear cell) taking account of the total elds observed. Thus, the data was expressed as the number of neutrophils per eld.

Data analysis
Experimental value are given as mean SEM. Statistical signi cance of differences between two means of data were evaluated by a Students's t-test for unpaired observations and Pvalues less than 0.05 were considered to be signi cant.
Injected 2 h before LPS, dexamethasone (5000 ng) also inhibited the generation of TNFa like activity (P , 0·05, n = 6) (Fig. 3, inset).    of 10 6 and 10 4 cells per ml and injected into the air pouch previously treated with cycloheximide (6 ng per pouch, 2 h before). This replenishment restored signi cantly the LPS-induced neutrophil recruitment when compared with cycloheximide-treated pouches or with PBS stimulation (Fig. 5). In addition, replenishment with 10 4 AM per ml from cycloheximide-treated animals failed to support the LPS-induced neutrophil recruitment (Fig. 6), contrasting to AM from control animals (P , 0·001 for neutrophils, n = 6).

Discussion
The injection of LPS into the rat air pouch resulted in a potent time-and dose-dependent recruitment of neutrophils. Suppression of this recruitment by low amounts of the protein synthesis inhibitor cycloheximide and by dexamethasone, injected into the pouch, suggested the involvement of a local target, capable of producing a secondary mediator. Since the concomitant injection of cycloheximide and LPS failed to block neutrophil recruitment, the 2 h interval required for inhibition is probably accounted for by the time needed for inhibition of protein synthesis. Alternatively, cycloheximide may induce apoptosis, 27,28 an active process requiring protein and RNA synthesis 29 and involving the degradation of nuclear DNA. 30 Although apoptosis is prevented in most cells by cycloheximide, HL-60 cells, 31 hepatocytes, 25,27 thymocytes and macrophages 28 undergo apoptosis when incubated with micromolar doses of this drug. However, apoptosis is unlikely to account for the suppression of neutrophil recruitment in our experiments, since the conditions used here (2 h, 6 ng 200 nmol) were previously shown not to induce apoptosis. 28 Resident mast cells might be affected by cycloheximide and by dexamethasone during LPS-induced neutrophil emigration, as suggested by the results of Matsuda et al. 24 with murine air blebs, a model closely resembling an air pouch but lacking a facsimile synovium. Similarly, Tannenbaum et al., 32 injected compound 48/ 80 into the rat skin at doses 50 times below ours and provoked a marked neutrophilia within 1 ±2 h. Compound 48/ 80 also caused mast cell degranulation, neutrophil adhesion and emigration into the cheek pouch vasculature. 33 Nevertheless, mast cell involvement is unlikely in our model, since the compound 48/ 80 failed to induce a signi cant cell in ltration or generation of TNFa-like activity, throughout all time intervals. A more likely hypothesis is that broblasts and macrophages of the pouch lining tissue account for the LPS-induced production of chemotactic substances and the consequent neutrophil recruitment, in agreement with the anatomopathological studies of Edwards et al. 22 To verify to what extent this might apply to our model, alveolar macrophages were used to replenish the air pouch and indeed inhibition by cycloheximide of neutrophil emigration and generation of TNFa-like activity, was surmounted by transferring fresh alveolar macrophages from control animals, whereas macrophages from cycloheximide-treated animals were not effective, as reported. 12,34 The use of alveolar macrophages rather than other sources of macrophages is supported by practical considerations concerning macrophage purication and their ability to provide large amounts of TNFa 35 or chemoattractants such as MIP-1a. 36 Liposomes are particularly ef cient in delivering water-soluble drugs into phagocytic cells, since phagocytosis is followed by phospholipase-induced disruption of the liposome phospholipid bilayers and the release of entrapped drugs. In particular, liposomeencapsulated clodronate which should deplete air pouch macrophages 37,38 reduced signicantly LPS-induced neutrophil recruitment. Since in separate experiments murine peritoneal macrophages were still absent after 5 days treatment (data not shown), the participation of the air pouch macrophage during neutrophil emigration is thus likely. Clodronate is speci c for macrophages, since it neither affected the neutrophil population nor other cell types. 37,39,40 Since a potential role for histamine and PAF in LPS-induced neutrophil recruitment into rat air pouches was excluded by selective antagonists, the effectiveness of cycloheximide and dexamethasone strongly suggests the involvement of protein synthesis-dependent mediators such as TNFa. As mentioned, the hypothesized participation of mast cells as a major source of preformed TNFa 41 was ruled out by the failure of compound 48/ 80 to induce neutrophil recruitment and detectable TNFa-like activity. Among cells that express the TNF gene, the macrophage is unique, insofar as it is capable of secreting-1000 times more TNFa in response to LPS than any other cell type. 42,43 Furthermore, a TNFa-like activity was detected in the air pouch transplanted with alveolar macrophages from naive, but not from cycloheximide-treated animals. In addition, the rat alveolar macrophage was shown to be an important producer of chemoattractants when stimulated by LPS. 44 Taken together, these results suggest that TNFalike activity accounts for the LPS-induced neutrophil recruitment in our model. This was supported by the ef cacy of an anti-murine polyclonal TNFa antibody to abrogate both LPStriggered neutrophil recruitment and TNFa-like activity. The resident macrophage is the likely source of this cytokine, as its formation was prevented by both in situ treatments with liposome-encapsulated clodronate and polyclonal antibody. Nevertheless, our results do not clarify whether TNFa acts as a direct chemoattractant to promote neutrophil emigration or if an intermediary chemokine is required, such as CINC/ GRO or MIP-1 or 2. Indeed, even though TNFa itself may increase adhesion molecules such as CD11/ 18 to promote neutrophil emigration, 45 several studies seem to plead for a chemokine networking involving either epithelial cell 46 or endothelial cell. 47 In summary, we suggest that neutrophil recruitment induced by LPS is a macrophagedependent event, involving the de novo synthetized TNFa which act directly or via secondary mediators.