Introduction

Abstract Introduction Materials and Methods Results Discussion References

Tumor necrosis factor- (TNF-) (1) is a cytokine with broad biologic activity, which participates in the lung immune and inflammatory response to microbial challenge (1). It acts in part by upregulating leukocyte adhesins, including E-selectin and intercellular adhesion molecule-1 (ICAM-1) on the vascular endothelium, stimulating the release of chemotactic factors and enhancing phagocyte function. The role of TNF- in lung inflammation is not fully defined, and appears to vary with the nature of the microbial challenge. On the basis of studies in which the actions of TNF- were inhibited with blocking antibodies, Denis and colleagues found that this cytokine plays a critical role in lung inflammation in response to Micropolyspora faeni, an agent that causes hypersensitivity pneumonitis in mice (4, 5). TNF- may play a complex and variable role in lung defense against gram-negative bacterial pathogens. Blocking antibodies to TNF- and soluble TNF-receptor antagonists impair the clearance of Klebsiella pneumoniae, Legionella pneumophila, and Pseudomonas aeruginosa from the lung, but this was not always associated with impaired polymorphonuclear leukocyte (PMN) recruitment (6). The role of TNF- in lipopolysaccharide (LPS)-induced lung inflammation is less clear (8, 11). Whether the differences in the role of TNF- in these studies reflect true differences, methodologic differences, or both, the results suggest considerable complexity in the role of TNF- in lung defense, inflammation, and injury in response to microbial challenge.

There is also considerable evidence that the lung response to microbial challenge is compartmentalized (14- 16), although this may be incomplete in the case of progressive or overwhelming infection or injury. For example, intratracheal challenge of rats with LPS results in increased TNF- in the bronchoalveolar lavage fluid (BALF), alveolar macrophages (AM), and lung tissue, but not in the blood, whereas systemic administration of LPS results in increased TNF- levels in the serum only (11, 16, 17). Aerosol administration of interferon- (IFN-) to human volunteers increases lung expression of 10-kD inflammatory protein (IP-10) (an IFN--dependent protein) without a concomitant increase in the blood, whereas systemic administration of IFN- has no effect on the lung (15). Previous attempts to address the role of TNF- in lung inflammation have used local or systemic administration of inhibitors or of an adenovirus directing expression of a TNF- antagonist (6, 18). These studies addressed the pathophysiology of pulmonary inflammation without directly addressing the question of whether the effects of TNF- were mediated locally or systemically. Similarly, although other investigators have expressed soluble inhibitors of TNF- globally in transgenic mice, using the cytomegalovirus (CMV) or alpha-1-antitrypsin (₁-AT) promoters (8, 22, 25), these mice are not useful for studies seeking to test tissue-specific actions of this cytokine.

To address this question, we generated mice that expressed a soluble TNF- inhibitor (soluble TNF receptor type II-Fc fusion protein [sTNFRIIFc]) transgene under the control of the human surfactant apoprotein C promoter (SPCTNFRIIFc mice). STNFRIIFc is a fusion protein containing the human type II TNF receptor (TNFRII), which is known to bind murine TNF- (33), fused to the Fc portion of immunoglobulin G1 (IgG₁) (27); fusion of the receptor to the Fc domain of IgG creates a dimeric ligand that binds TNF with much greater avidity and has a longer half-life in vivo than does the TNF receptor itself (26, 27, 30). In accord with the known specificity of the human surfactant apoprotein C promoter, the sTNFRIIFc transgene was expressed selectively in the distal airway and alveolar epithelium of the lung, resulting in substantial concentrations of the inhibitor in the airways and minimal concentrations in the blood. The mice bearing this transgene were challenged by local administration of M. faeni antigen and LPS. These two agents were chosen as model microbial agonists, since the importance of TNF- in lung inflammation in response to these agents appears to differ, as noted previously (4, 5, 8, 11). Results in SPCTNFRIIFc transgenic mice were compared with those in wild-type controls, with those in type 1 TNFR knockout mice (TNFRI^/ mice, in which most if not all proinflammatory actions of TNF- are impaired) (34, 35), and with those in mice given recombinant sTNFRIIFc intravenously, which preferentially inhibited systemic more than pulmonary actions of TNF-. The results suggested that PMN recruitment in response to M. faeni antigen was largely mediated through local actions of TNF- in the distal lung, whereas the response to LPS appeared to be more dependent on extrapulmonary actions of this cytokine. Thus, depending on the nature of the challenge, the site of TNF- action varied.

	Materials and Methods

Abstract Introduction Materials and Methods Results Discussion References

Determination of sTNFRIIFc Bioactivity In Vitro

The ability of recombinant sTNFRIIFc to inhibit the activity of murine TNF- was determined by cytolytic assay, using the mouse fibrosarcoma line WEHI 164 clone 13 (36). The WEHI cells (2 × 10⁴ cells/well) were incubated for 24 h in 96-well plates with serial dilutions of recombinant murine (and for comparison human) TNF- (Genzyme, Cambridge, MA) and different concentrations of sTNFRIIFc. Following this, Alamar blue (AccuMed International Inc., Westlake, OH) was added at a 1:10 final concentration to each well, and the plates were read 3 h later at dual wavelengths of 570/600 nm on an EL 312e microplate reader (Biotech Instruments, Inc., Winooski, VT). The cytotoxic effects of TNF- and the ability of sTNFRIIFc to inhibit these effects were expressed as percent of surviving cells. The experiment was repeated three times for each condition.

Wild-Type and TNFRI Knockout Mice

Normal C57BL/6 and B6D2F2 mice were purchased from Charles River (Charles River Breeding Laboratories, Wilmington, MA) and Jackson Laboratories (Bar Harbor, ME). Mice in which the type I TNF- receptor has been knocked out (TNFRI^/ mice) were generated at Immunex Corporation (Seattle, WA), on a pure C57BL/6 background (37).

Generation and Initial Characterization of SPCTNFRIIFc Transgenic Mice

Generation of SPCTNFRIIFc mice. The t-intron and poly-A fragments of SV-40 were inserted into the SacI restriction site of the pBSK II plasmid (Stratagene, La Jolla, CA). A 3.7-kb fragment from the 5' flanking region of the human surfactant protein C (SPC) gene (38) (a generous gift from J. Whitsett, Cincinnati Children's Hospital, Cincinnati, OH) was excised with HindIII, filled in with Klenow fragment, and cloned into the EcoRV site of the pBSKII plasmid (Stratagene) containing the SV-40 t-intron and poly-A fragments. The complementary DNA (cDNA) for a fusion protein of the extracellular domain of human TNFRII and the Fc portion of human IgG₁ (Immunex) (27) was cloned into the NotI site of this plasmid between the SPC promoter and t-intron. The transgene construct was excised with SalI/AflI and injected into the male pronucleus of B6D2F2 mouse embryos, using standard techniques (39).

Identification of transgene-positive mice. Mice harboring the transgene construct were identified by analysis of tail biopsy DNA by polymerase chain reaction (PCR) and Southern blot analysis. Tail biopsies were digested overnight at 55°C with proteinase K, and the resultant DNA was purified by ethanol precipitation. For PCR analysis, DNA (100 ng) was then amplified, using primers derived from the region flanking the junction of the cDNA for sTNFRII (5'-AACAGAACCGCATCTGCACC-3') and that of the Fc portion of human IgG₁ (5'-CGTGCTGTTGTACTGCTCCTCC-3'). The resultant 700-bp PCR product was gel-electrophoresed and stained with ethidium bromide. For Southern blot analysis, 7.5 µg of tail biopsy DNA was digested with BamHI and subjected to Southern blot analysis using a random ³²P-labeled, 500-bp BamHI/BssHII fragment of huTNFRIIFc cDNA as a probe.

Northern blot analysis. Ten micrograms of RNA were extracted and purified from a variety of tissues, using the guanidine isothiocyanate CsCl method (40), and were then transferred to Nytran filters (Schleicher & Schuell, Keene, NH). Blots were probed with the cDNA fragment described previously. Equality of sample loading was judged by reprobing these filters with a random ³²P-labeled cDNA fragment of elongation factor-1 (EF), as previously described (41).

Immunohistochemistry. The left lung from transgenic and wild-type littermate control mice was harvested and fixed in methyl Carnoy's reagent. Tissues were embedded, sectioned, and then blocked with 5% nonfat dry milk in phosphate-buffered saline (PBS) for 30 min at room temperature. To detect the transgene product, sheep antihuman TNFRIIFc p6 polyclonal antibody (Immunex), diluted 1:400 in PBS and 0.05% Tween 20, was then applied for 60 min at room temperature. PBS served as the control. For detection, biotinylated rabbit antisheep Ig (Vector Laboratories, Burlingame, CA) was diluted and applied according to the manufacturer's instructions, and the tissue preparations were incubated at room temperature for 60 min. The tissue sections were then incubated with an avidin-biotin conjugate/alkaline phosphatase (ABC- AP) kit (Vector) for 60 min at room temperature, followed by AP-substrate treatment (Vector Red Kit) for 15 min at room temperature. The tissue sections were then counterstained with methylene blue (Loeffler) for 15 s.

Quantitation of sTNFRIIFc fusion protein in lung epithelial lining fluid. The trachea was cannulated with a rigid, 22-gauge blunt needle, and bronchoalveolar lavage (BAL) was performed with 2 ml of PBS and 0.6 mM ethylenediamine tetraacetic acid (EDTA) at 37°C. Approximately 1.7 to 2.0 ml of fluid was recovered from each mouse.

Characterization of the Lung Inflammatory Response to Intrapulmonary Challenge with M. faeni Antigen and LPS

For challenge with M. faeni antigen, mice were lightly anesthetized with a mixture of xylazine and ketamine administered intraperitoneally. Three doses of 75 µg each of M. faeni antigen (Immunex) were instilled into the nares at 24-h intervals, and mice were analyzed 4 h after the last intranasal dose. In some experiments, 2 µg or 100 µg of sTNFRIIFc was administered to a subset of the mice via the lateral tail vein at 24 h before instillation of the first dose of M. faeni antigen.

For LPS challenge, Escherichia coli LPS 0111:B4 (Sigma) was reconstituted to the desired concentration in 16 ml of PBS and administered to mice with an aerosol device over a 30-min period, as previously described (43). Mice were analyzed at 4 or 24 h after aerosol challenge. In each experiment, four or five mice were analyzed in each group. In one set of experiments, two groups of mice received 100 µg of sTNFRIIFc IV at 24 h before the administration of LPS. These mice were then analyzed at 4 h after aerosolized LPS challenge.

Lung inflammation was assessed by analysis of cellular recruitment into BALF. BAL was performed as described previously, and the total numbers of cells in the BALF were enumerated with a hemacytometer. The BALF was then centrifuged, and aliquots were frozen and stored at 80°C until used. The remaining cell pellet was resuspended in PBS and 0.6 mM EDTA, stained with Diff-Quik (Baxter Healthcare Corp., Miami, FL), and the cell type was determined by microscopy. The quantity of total protein in the BALF was also determined, using the bicinchoninic acid (BCA) protein assay in microtiter trays (Pierce, Rockford, IL). Briefly, 10 µl of BALF was mixed with 200 µl of the BCA protein reagent and incubated at 37°C for 30 min. A standard curve from 20 to 1,000 µg/ml of BSA was generated concurrently. The plates were read at 562 nm on an EL 312e microplate reader, and sample values were determined with Delta Soft3 software. In some experiments, lung expression of ICAM-1 following inflammatory challenge was determined by Northern blot analysis, using 10 µg of total lung RNA per lane. Filters were probed using a random ³²P-labeled, 270-bp ScaI/EcoRI fragment of murine ICAM-1 cDNA. The cDNA was generated by reverse transcription (RT)-PCR, using lung RNA as template and Pfu polymerase (Stratagene, La Jolla, CA); was cloned into PCR3.1 vector (InVitrogen, La Jolla, CA); and was sequenced with the ABI/Prism system (Applied Biosystems Inc., Foster City, CA). The sequence obtained was identical to that of authentic murine ICAM-1 (GenBank no. 51507). The equality of sample loading was determined with an EF cDNA probe as described previously.

Survival of Mice after Systemic LPS Administration

E. coli LPS 0111:B4 was injected intraperitoneally at the desired dose following intraperitoneal injection of 18 mg of d-galactosamine (Sigma) (44, 45). Survival following challenge was monitored for 72 h; preliminary data indicated that no additional deaths occurred beyond a 24-h time point. In one experiment, a subgroup of mice received 100 µg of sTNFRIIFc intravenously 24 h before the LPS/d-galactosamine challenge.

Statistical Evaluation

All values are expressed as means ± SD. A two-tailed Student's t test was used to determine P values, and values of P  0.05 were considered statistically significant.

	Results

Abstract Introduction Materials and Methods Results Discussion References

Bioactivity of sTNFRIIFc

Human sTNFRII binds murine TNF- and inhibits its actions (33). Because dimeric sTNFRIIFc inhibits human TNF- activity more effectively than does sTNFRII monomer (27), sTNFRIIFc should effectively inhibit murine TNF-. The murine WEHI 163 cell cytotoxicity assay was used to test the ability of sTNFRIIFc to inhibit the activity of murine TNF-. sTNFRIIFc inhibited the activity of murine TNF- in a dose-dependent manner at concentrations  50 ng/ml (Figure 1), although at high concentrations of TNF-, inhibition was not complete. Greater concentrations of sTNFRIIFc inhibited the growth of the indicator cell line and could not be tested reliably. These experiments indicated that the sTNFRIIFc fusion protein can inhibit the actions of murine TNF- in vitro, which is consistent with the findings in studies indicating that systemic administration of sTNFRIIFc to mice impedes the lethal effects of endogenous TNF induced by systemic LPS administration (27), and impedes the protective effects of TNF in lung infection with K. pneumoniae (26). These results suggest that transgenic mice expressing sTNFRIIFc in the lung should provide a useful model for evaluating the local versus systemic role of TNF in the lung inflammatory response.

View larger version (17K): [in this window] [in a new window]   Figure 1.   Assessment of the ability of sTNFRIIFc to block TNF--induced cytotoxicity in the WEHI 163 cell line. WEHI cells (2 × 104 cells/well) were incubated for 24 h with serial dilutions of recombinant murine TNF- and different concentrations of sTNFRIIFc. Results of TNF--induced cytotoxicity, and the ability of sTNFRIIFc to inhibit this effect, are expressed as percent surviving cells ± SD. The experiment was repeated three times for each condition.

Characterization of Transgenic Mice

Four lines of SPCTNFRIIFc transgenic mice were generated (data not shown), and one line was selected for further studies on the basis of its favorable pattern of transgene expression. Northern blot analysis of 10 µg of RNA from various tissues derived from this line of SPCTNFRIIFc transgenic mice and from wild-type littermate controls revealed expression of the transgene only in the lungs of the transgenic mice (Figure 2a). Although some inequality of RNA loading between tissues was demonstrated with the EF probe (Figure 2b), prolonged exposure on Phosphoimager plates failed to demonstrate expression of the transgene in any other tissues. Immunohistochemistry (Figures 3a through 3d) revealed the sTNFRIIFc protein in the distal airways, alveoli, epithelial cells, and adjacent interstitium of SPCTNFRIIFc transgenic mice but not in those of littermate controls, as predicted for genes expressed under the SPC promoter (46, 47). The sTNFRIIFc protein concentration in lung epithelial lining fluid (ELF) of the SPCTNFRIIFc mice (range: 0.63 to 9.97 µg/ml) was 10 to 40 times that in the serum (range: 5 to 252 ng/ml) (Figure 4). Levels of the sTNFRIIFc protein were below the level of detection in the serum and BALF of wild-type littermate controls (data not shown). By comparison with Figure 1, these results suggest that concentrations of sTNFRIIFc in the lung should be sufficient to inhibit, at least partly, concentrations of TNF commonly found in the airways or lungs of rodents with model infectious or inflammatory challenges (4-6, 13; Ulich, 1993 #39).

View larger version (136K): [in this window] [in a new window]   Figure 2.   Northern blot analysis of SPCTNFRIIFc and wild-type littermate mice. (a) RNA from various tissues probed with a BamHI/BssHII fragment from sTNFRIIFc cDNA. (b) The same blot as in a reprobed with EF-1 to determine equality of sample loading.

View larger version (145K): [in this window] [in a new window]   View larger version (133K): [in this window] [in a new window]   View larger version (131K): [in this window] [in a new window]   View larger version (142K): [in this window] [in a new window]   View larger version (145K): [in this window] [in a new window]   View larger version (143K): [in this window] [in a new window]   Figure 3.   Immunohistochemistry of SPCTNFRIIFc and wild-type littermate lungs. Representative lung sections from SPCTNFRIIFc mice (A through C, E) and from littermate control mice (D, F ) stained with sheep antihuman TNFRIIFc (A through D) and PBS (E, F ). Arrow ends show alveolar epithelial cells staining strongly for sTNFRIIFc.

View larger version (12K): [in this window] [in a new window]   Figure 4.   Concentration of sTNFRIIFc in serum and ELF of the lung. Values were determined by ELISA on serum and BALF of 15 animals, after which BALF were corrected, with urea as a standard, to determine ELF concentration. Values are means ± SD.

Intranasal Challenge with M. faeni Antigen

M. faeni is an etiologic agent of hypersensitivity pneumonitis (4, 5), and when instilled intranasally in mice induces pulmonary inflammation and interstitial fibrosis. TNF- was shown by antibody blocking studies to be critical for the induction of lung inflammation in response to M. faeni antigen (4, 5). To determine whether local blockade of TNF action in the distal lung (in SPCTNFRIIFc mice), or global inhibition of TNF action (in TNFRI^/ mice), was required to reduce lung inflammation in response to M. faeni antigen, SPCTNFRIIFc, TNFRI^/, and wild-type control mice were challenged intranasally with this antigen. Figure 5 shows the combined results of four experiments with four SPCTNFRIIFc and four wild-type mice and three TNFRI^/ mice per group in each experiment. There were few PMN in the lungs of wild-type, SPCTNFRIIFc, or TNFRI^/ mice given saline intranasally (< 8% PMN and < 0.4 × 10⁴ PMN/ml BALF). Following M. faeni antigen challenge, the total number of PMN (Figure 5a) and percent PMN (Figure 5b) in the BALF of control mice increased markedly. PMN recruitment was essentially abolished in TNFRI^/ mice as compared with wild-type control mice (P < 0.001). PMN recruitment was substantially reduced in the SPCTNFRIIFc mice (P = 0.01 for total PMN and P < 0.001 for % PMN in the BALF versus wild-type control mice), but not to the same degree as in the TNFRI^/ mice (P < 0.01 versus SPCTNFRIIFc mice). The total cellular influx into BALF was significantly reduced only for the TNFRI^/ mice (P < 0.001) (data not shown). As another measure of airway inflammation/injury, the total protein concentration in the BALF was determined. The protein concentration in the BALF increased by only approximately twofold in response to challenge with M. faeni antigen, and was similar in each of the groups (data not shown). Thus, the influx of PMN, but not the modest increase in BALF protein, was TNF dependent.

View larger version (23K): [in this window] [in a new window]   Figure 5.   PMN recruitment to the lungs following intranasal M. faeni antigen challenge. (a) Total PMN in BALF. (b) % PMN in BALF. Each mouse received three doses of 75 µg each of M. faeni antigen antigen and was analyzed 4 h after the last dose. The figure represents the compiled results of four experiments, each of which included four wild-type, four SPCTNFRIIFc, and three TNFR/ mice per group. A subgroup of wild-type and SPCTNFRIIFc mice (Rx) in two separate experiments were pretreated intravenously with 100 µg of sTNFRIIFc at 24 h before the initial M. faeni antigen challenge. Results for TNFRI/ mice that are significantly different from those for wild-type mice are indicated by * (P < 0.001), and significant differences between SPCTNFRIIFc and wild-type mice are indicated by  (P = 0.01) and  (P < 0.001). Significant differences between TNFRI/ and SPCTNFRIIFc mice are indicated by § (P < 0.01). Significant differences between mice treated intravenously with sTNFRIIFc and wild-type mice are indicated by ** (P < 0.05) and (P < 0.01).

The more complete reduction in PMN recruitment in the TNFRI^/ mice than in the SPCTNFRIIFc mice may reflect: (1) an extrapulmonary role for TNF-; or (2) incomplete neutralization of TNF- by sTNFRIIFc, owing to inadequate local concentrations of the antagonist in the airways and lung tissue. To explore these possibilities, in two of the four M. faeni antigen experiments, some mice were given 100 µg or 2 µg of sTNFRIIFc intravenously. High-dose intravenous sTNFRIIFc was given to achieve a greater extrapulmonary than intrapulmonary blockade of TNF-, and low-dose sTNFRIIFc was given to determine whether the level of sTNFRIIFc present in the serum of the SPCTNFRIIFc mice was sufficient to modulate PMN recruitment. At the time of killing, 96 h after intravenous injection of sTNFRIIFc, the ELF concentrations of sTNFRIIFc in mice given 100 µg or 2 µg of the inhibitor were 0.4 to 1.8 µg/ml and < 62.5 pg/ml, respectively, and the serum levels were 1.5 to 11.1 µg/ml and 4 to 33 ng/ml, respectively. Thus, mice given 100 µg of sTNFRIIFc intravenously had ELF concentrations of sTNFRIIFc similar to those of SPCTNFRIIFc mice, and much greater serum levels (Figure 4), whereas mice receiving 2 µg of sTNFRIIFc intravenously had serum levels similar to those of SPCTNFRIIFc mice, and lacked detectable sTNFRIIFc in their BALF.

Administration of 100 µg of sTNFRIIFc intravenously to wild-type mice prior to M. faeni antigen challenge diminished PMN recruitment to the lungs as compared with that in untreated wild-type mice (Figure 5; for % PMN in the BALF: P = 0.06; for total PMN in BALF: P = 0.008); inhibition of PMN was less complete than in the TNFRI^/ mice (Figure 5; for % PMN in BALF and total PMN in BALF: P < 0.001) but similar recruitment to that observed in the SPCTNFRIIFc mice. Administration of 100 µg of sTNFRIIFc to SPCTNFRIIFc mice did not further diminish PMN recruitment (Figure 5). Administration of 2 µg of sTNFRIIFc intravenously to wild-type mice had no effect on PMN recruitment (data not shown; one experiment with four mice per group). In this series of experiments, PMN recruitment was inhibited to a similar degree in SPCTNFRIIFc and mice treated with high-dose (100 µg) sTNFRIIFc intravenously. These groups had similar ELF concentrations of sTNFRIIFc, but the mice given high-dose sTNFRIIFc intravenously had much higher serum concentrations. Thus, inhibition of PMN recruitment to the lungs following M. faeni antigen challenge correlated with ELF and not with serum concentrations of sTNFRIIFc.

Aerosolized LPS Challenge

LPS is a potent stimulator of inflammation, acting in part to induce TNF- and to enhance adhesion-molecule expression (48, 49), and may be one of the initial inflammatory stimuli that induces acute respiratory distress syndrome (ARDS) during bacterial sepsis. When mice were challenged with a single aerosol administration of LPS at 50 µg/ml, the total number of leukocytes, and the numbers and percentage of PMN, were significantly lower in the BALF of TNFRI^/ mice than in that of wild-type mice at 4 h after challenge (P = 0.01; Figure 6). Under these conditions the numbers of PMN in the BALF of SPCTNFRIIFc mice were not significantly different from those in wild-type mice (Figure 6), although when assessed at 24 h after the administration of 10 µg/ml or 33 µg/ml of LPS, the percentage but not the total number of PMN was reduced compared with that of wild-type mice (P = 0.003 and P = 0.007, data not shown).

View larger version (25K): [in this window] [in a new window]   Figure 6.   PMN recruitment to the lungs 4 h after aerosol challenge with LPS at 50 µg/ml in wild-type, SPCTNFRIIFc, and TNFRI/ mice. A subgroup of mice (Rx) received 100 µg of sTNFRIIFc intravenously at 24 h prior to challenge. (a) Total PMN in BALF. (b) % PMN in BALF. The results are from one experiment in which there were five mice in each experimental group. Values are means ± SD. Significant differences between wild-type mice and wild-type mice treated with sTNFRIIFc are indicated by * (P  0.05). Significant differences between wild-type mice and TNFRI/ mice are indicated by  (P = 0.01). Significant differences between wild-type mice and SPCTNFRIIFc mice treated with sTNFRIIFc are indicated by § (P < 0.05).

To determine the basis for the greater reduction in PMN recruitment into the airways of TNFRI^/ than of SPCTNFRIIFc mice, a subset of wild-type and SPCTNFRIIFc transgenic mice was given 100 µg of sTNFRIIFc intravenously at 24 h prior to LPS challenge. When assessed at the time of killing, this resulted in much higher serum concentrations of sTNFRIIFc (5 to 40 µg/ml), and in ELF levels similar to those seen in the SPCTNFRIIFc mice (1.8 to 2.2 µg/ml). PMN recruitment was markedly reduced in both groups of mice treated with intravenous sTNFRIIFc as compared with wild-type mice (P < 0.01) (Figure 6), and PMN recruitment in both groups of sTNFRIIFc-treated mice was inhibited as effectively as in the TNFRI^/ mice (Figure 6). LPS administration was associated with an approximately twofold increase in the BALF protein concentration, but this increase did not differ between the groups (data not shown). Thus, TNF contributed substantially to PMN recruitment in response to LPS, but unlike the response to M. faeni antigen, inhibition of PMN recruitment appeared to correlate primarily with serum rather than ELF levels of sTNFRIIFc, and serum concentrations of sTNFRIIFc achieved after intravenous administration of 100 µg of this inhibitor blocked PMN recruitment as completely as did loss of TNFRI expression.

Role of TNF in Regulation of Lung ICAM-1 Expression

TNF- contributes to inflammatory-cell recruitment in part by augmenting the expression of adhesion molecules including ICAM-1, and ICAM-1 appears to contribute to inflammatory-cell recruitment to the lung in response to LPS and gram-negative pneumonia (50). However, in the present study, the abundance of ICAM-1 mRNA in lung tissues of mice challenged with M. faeni antigen or LPS did not correlate with PMN recruitment: ICAM-1 mRNA was not clearly lower in TNFR^/-, SPCTNFRIIFc-, or sTNFRIIFc-treated mice than in controls (data not shown).

Systemic LPS Challenge

To investigate whether the low levels of sTNFRIIFc in the serum of the SPCTNFRIIFc mice were sufficient to inhibit extrapulmonary actions of TNF-, wild-type, SPCTNFRIIFc transgenic, and TNFRI^/ mice were challenged with d-galactosamine, followed by one of three doses of intraperitoneal LPS. Survival of the SPCTNFRIIFc mice and wild-type control mice was similar, indicating no protective effect of sTNFRIIFc at the concentrations found in the blood of SPCTNFRIIFc transgenic mice (Table 1). The TNFRI^/ mice were protected from the effects of LPS as previously described (34, 35). In a subsequent experiment, a subgroup of both wild-type and SPCTNFRIIFc mice were pretreated with 100 µg of sTNFRIIFc given intravenously. No untreated mice survived beyond 7 h, whereas the time to death was slightly prolonged in mice treated with 100 µg of sTNFRIIFc (Table 1). These findings suggest that high systemic concentrations of sTNFRIIFc, but not the low levels of sTNFRIIFc present in the blood of SPCTNFRIIFc mice, modified LPS-induced toxicity, although the effects were incomplete.

	Discussion

Abstract Introduction Materials and Methods Results Discussion References

The generation of SPCTNFRIIFc transgenic mice provided a model in which the locus of action of TNF- could be explored through comparison with wild-type and TNFRI^/ mice. The results illustrate the complexity of the pulmonary inflammatory response and suggest that TNF- may act, at least in part, at different sites, depending on the nature of the challenge.

TNF- appeared to play an essential role in PMN recruitment to the lungs in response to M. faeni antigen, because recruitment was nearly abolished in TNFRI^/ mice. These findings are consistent with those in studies by Denis and colleagues, in which an important role for TNF- in M. faeni-antigen-induced hypersensitivity pneumonitis was inferred (4, 5); in their studies, systemic administration of rabbit anti-TNF antibodies in amounts that fully blocked TNF- in BALF markedly reduced cellular recruitment. The present studies extend these findings by addressing the locus of TNF- action. Although the effect was less complete than in TNFRI^/ mice, PMN recruitment into BALF was reduced to a similar degree in SPCTNFRIIFc transgenic mice and in wild-type mice treated with 100 µg of sTNFRIIFc given intravenously. These results paralleled the BALF concentrations of the inhibitor, which were similar in these two groups, but not the serum concentrations, which were ~ 50-fold higher in the mice treated intravenously. Increasing the serum concentrations of sTNFRIIFc in SPCTNFRIIFc mice through intravenous administration of the inhibitor did not further reduce PMN recruitment. These results strongly suggest that PMN recruitment in response to M. faeni antigen was predominantly dependent on effects of TNF- in the distal lung, the site of transgene expression in the SPCTNFRIIFc mice.

In contrast to the experiments with M. faeni antigen, aerosol administration of LPS was followed by a reduction in PMN recruitment to a similar degree in mice given 100 µg of sTNFRIIFc intravenously and in TNFRI^/ mice (70% and 60%, respectively). PMN recruitment was not reduced significantly in SPCTNFRIIFc mice. These results suggest that in response to pulmonary challenge with LPS, TNF- plays an important role in PMN recruitment, but acts at sites other than or in addition to the distal lung. This may at first seem counterintuitive, since the lung inflammatory response to LPS is thought to be largely compartmentalized from the systemic circulation (11, 16, 17). The current results may provide some insight into apparent inconsistencies and paradoxical results in other studies addressing the role of TNF- in lung inflammation in response to LPS.

Kolls and colleagues (8) found that systemic administration of an adenoviral vector transducing soluble dimeric human TNFRIIFc to mice led to very high blood concentrations of this TNF antagonist and nearly abolished PMN recruitment into BALF. In contrast, intratracheal administration of the same vector was less effective. Although this difference may partly reflect the larger amount of virus administered intravenously, these results are consistent with the current finding that high systemic levels of TNF antagonist are necessary to impede the response to intrapulmonary LPS. Intratracheal coadministration of anti-TNF antibody (11, 13) or sTNFRIIFc (13) with LPS markedly reduced TNF bioactivity in BALF but did not reduce PMN recruitment. Conversely, intratracheal coadministration of monomeric human sTNFRI or sTNFRII to rats inhibited PMN recruitment by 30 to 50% at 6 h (but not at 4 h or 12 h), even though this treatment increased TNF activity in BALF. The increase in TNF activity in the BALF may reflect a carrier effect of the soluble monomeric receptors, which when complexed to TNF may delay its clearance and prevent full inhibition of its activity (27, 51, 52). The discordance between the capacity of the various TNF antagonists to inhibit TNF activity in BALF and to inhibit PMN recruitment after intratracheal administration may be related to the distribution of these agents. Soluble monomeric TNF receptors distribute freely across tissues (53), which may have allowed them to move more efficiently than the dimeric TNF-receptor antagonists or anti-TNF antibodies from the airways to other sites at which TNF acted to facilitate neutrophil recruitment. Taken together, the results of the current and past studies are consistent with the notion that TNF- acts systemically rather than (or in addition to) locally to mediate PMN recruitment into the lung in response to LPS.

The conclusions regarding the locus of TNF action in the current study are based on the observations that the SPCTNFRIIFc mice expressed that sTNFRIIFc mRNA solely in the lung, and that sTNFRIIFc was detected only in the distal airways, alveoli, and adjacent lung parenchyma by immunohistochemistry. Small amounts of sTNFRIIFc were present in the blood, probably reflecting bidirectional protein secretion rather than extrapulmonary synthesis. Nonetheless, concentrations of STNFRIIFc in the ELF of the lung exceeded those in the serum by 10- to 40-fold. Two further lines of evidence suggest that the effects of the transgene product were local: (1) Concentrations in the blood had minimal to no inhibitory activity in the in vitro assay; (2) lung inflammatory responses in SPCTNFRIIFc mice were impaired, whereas the systemic response to d-galactosamine/LPS was not affected.

There are several possible explanations for the finding that pulmonary inflammation was less completely blocked in SPCTNFRIIFc mice than in TNFRI^/ mice independent of the type of challenge. Concentrations of sTNFRIIFc in the ELF of SPCTNFRIIFc mice (~ 2 µg/ml) were able to impede substantially the actions of TNF- in vitro (Figure 1), even though complete inhibition was not achieved with TNF at  50 U/ml. Following M. faeni antigen and LPS challenge, higher concentrations of TNF- may have been present locally in the lungs and may not have been fully neutralized in the SPCTNFRIIFc mice or mice given sTNFRIIFc intravenously. TNF bioactivity could not be detected reproducibly in the lungs of challenged SPCTNFRIIFc mice (data not shown). However, immunoreactive TNF- was detected in the BALF of SPCTNFRIIFc mice and persisted at later time points in these animals than in untreated wild-type or TNFRI^/ mice (data not shown), which is consistent with the known prolongation of TNF- half-life when it is complexed with soluble receptors (27, 30, 54). Given the comparable inhibition of LPS-induced lung inflammation in mice treated intravenously with sTNFRIIFc and TNFRI^/ mice, it is unlikely that sTNFRIIFc cannot adequately neutralize murine TNF- in vivo if concentrations of the inhibitor are adequate to block the amount of TNF- present at the relevant sites. Rather, the current data suggest that among mice challenged with LPS, high blood concentrations of sTNFRIIFc in mice treated intravenously with this inhibitor were sufficient to inhibit the activity of TNF- as efficiently as in the TNFRI^/ mice. In the context of the M. faeni antigen challenge, in which it appeared that inhibition of PMN recruitment was primarily if not solely dependent on local effects of sTNFRIIFc, lung concentrations of the latter in the transgenic mice were ~ 10-fold lower than the blood concentrations achieved in mice given 100 µg of sTNFRIIFc intravenously, and may not have been sufficient to inhibit fully the amounts of TNF- present. Alternatively, incomplete inhibition of PMN recruitment in response to M. faeni antigen in the SPCTNFRIIFc- and sTNFRIIFc-treated mice as compared with TNFRI^/ mice could have resulted from the persistence of TNF- complexed to sTNFRIIFc, some of which may have been biologically active in vivo (51).

Our study did not identify the mechanism(s) by which TNF contributed to PMN recruitment into the air spaces. Intratracheal administration of LPS induces an overall increase in expression of ICAM-1 mRNA in the lung (11), which reflects increased expression on the vascular endothelium and type II alveolar epithelial cells (50). It appears that ICAM-1 contributes to PMN influx into the lung in response to LPS and gram-negative bacterial infection, since anti-ICAM-1 antibodies and antisense ICAM-1 oligonucleotides partly inhibit PMN recruitment, although PMN recruitment is not reduced in ICAM-1-deficient mice (50, 57). After challenge with LPS or M. faeni antigen, lung ICAM-1 mRNA abundance was similar in TNFRI^/, SPCTNFRIIFc, and wild-type mice, although it is possible that selective alterations in the expression of ICAM-1 occurred in some cell populations and were not detected by assessment of lung ICAM-1 mRNA abundance. The current results do not argue against a role for ICAM-1 in PMN recruitment in these models (11, 50) but do suggest that TNF facilitates PMN recruitment into the lungs, partly by mechanisms other than the induction of ICAM-1 (2, 3).

In summary, our study demonstrated that TNF- may contribute to lung inflammation by acting at intrapulmonary and/or extrapulmonary sites, depending on the type of challenge. Inhibition of TNF- activity within the lung decreased M. faeni-antigen-induced pulmonary inflammation, whereas global inhibition of TNF- activity was needed to reduce inflammation induced by intrapulmonary administration of LPS. These observations further the understanding of the compartmentalized nature of the pulmonary host response. TNF appears to contribute to lung defenses against a variety of microbes, including Streptococcus pneumoniae, P. aeruginosa, K. pneumoniae, and Mycobacterium tuberculosis (6, 7, 9, 10, 57). SPCTNFRIIFc mice should provide a useful model for determining the locus and mechanisms of action of TNF in pulmonary infections caused by these and other pathogenic bacteria. The development of other models in which tissue-specific roles of cytokines can be addressed may provide a useful approach to developing new targeted immunomodulatory therapies.