Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Tumor necrosis factor (TNF) is a pleiotropic proinflammatory cytokine (1) that has been associated with lung injury since it was originally found to be elevated in the bronchopulmonary secretions of subjects with acute respiratory distress syndrome (ARDS) (2). TNF is a product of inflammatory cells found in the lung, including macrophages (3) and neutrophils (4). It is produced initially in a membrane-associated 26-kD form (mTNF), which is then cleaved enzymatically by TNF--converting enzyme (TACE) into the soluble 17.5-kD cytokine (5). The majority of previous studies investigating TNF activity in lung disease, and all studies to date investigating the role of TNF specifically in ARDS, have measured either the soluble protein (6, 7) or TNF messenger RNA (mRNA) (3). However, it is now established that mTNF is functionally active before processing by TACE, and recent studies using transgenic animals have shown that in the absence of soluble TNF, mTNF can mediate inflammation in experimental arthritis (8) and hepatitis (9).

ARDS is a potentially catastrophic form of lung injury that can arise from a number of direct and indirect initiating events but is commonly associated with sepsis (10). There remains significant mortality despite improvements in intensive therapy. ARDS is an inflammatory disease characterized by an acute neutrophil alveolitis in association with increased pulmonary vascular permeability (11) and fibroproliferation (12), pathologic consequences that have been shown to be associated with high levels of TNF. Several studies comparing plasma and bronchoalveolar lavage fluid (BALF) levels of TNF have reported that the observed increase in TNF is lung-derived (6), and the alveolar macrophage (AM) is the likely cellular source (13). AMs are known to have increased expression of mTNF in lung inflammation (14), and we hypothesized that functionally active mTNF may have profound influence on the alveolar-epithelial interface and may therefore contribute to the diffuse alveolar damage that is a feature of ARDS.

In this study we demonstrate for the first time that mTNF on AMs is biologically active in a bioassay system. We show that mTNF expression is significantly increased in subjects with ARDS compared with At Risk subjects, but there was no significant difference in the levels of biologically active soluble TNF in the BALF or AM supernatants between the groups. We also demonstrate diminished shedding but increased mRNA expression of CD120b-soluble TNF receptor in the AM supernatants derived from subjects with ARDS compared with At Risk subjects, suggesting that increased surface expression of this receptor may be important in mediating the signal of mTNF in ARDS. The present study highlights a potentially important and previously unexplored source of TNF that may be implicated in the pathogenesis of ARDS.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Subjects

We studied 67 subjects in seven groups associated with the development of ARDS/acute lung injury (ALI): pancreatitis (n = 5), emergency aortic aneurysm repair (n = 5), multiple trauma (n = 3), multiple blood transfusion (n = 7), abdominal sepsis (n = 23), pneumonia with sepsis (n = 11), and pneumonia without sepsis (n = 13). Patients were studied on admission to the Intensive Therapy Unit (ITU), Southmead Hospital, Bristol, UK. The extent of injury to trauma patients was assessed by the Injury Severity Score (ISS) (15). An ISS of > 15 was used to define multiple trauma. Sepsis syndrome was defined according to the criteria of Bone and colleagues (16). Patients were defined as ARDS, ALI, or At Risk according to the American-European consensus (17) (see Table 1). Patients with bilateral infiltrates on chest X-ray and a Pa_O2:FI_O2 ratio < 300 mm Hg were defined as having ALI (n = 7), whereas patients with a Pa_O2:FI_O2 ratio < 200 mm Hg were defined as having ARDS (n = 40). Patients without bilateral infiltrates on chest X-ray were defined as At Risk (n = 20). Six patients ventilated postoperatively for elective aortic aneurysm repair, without cardiopulmonary bypass, were included as ventilated control subjects (VC). All the patients were studied within 48 h of arrival on ITU and all the patients were mechanically ventilated at the time of bronchoscopy. The study was approved by the Southmead NHS Trust Ethics Committee.

Bronchoalveolar Lavage

Each bronchoscopy was performed through an indwelling endotracheal tube. Bronchoalveolar lavage was performed in the right middle lobe. Eight 20-ml aliquots of bicarbonate-buffered isotonic saline were instilled and gently aspirated into a siliconized bottle kept on ice. The chilled BALF was strained through a single layer of coarse gauze to remove mucus clumps and then spun at 400 × g for 10 min to recover cells. The resultant cell-free fluid was stored at 70°C until analysis. The AM population was purified by negative selection with magnetic beads using granulocyte-specific anti-CD66b (Serotec, Kidlington, UK) and T cell- specific anti-CD3 antibodies (Dynal, Bromborough, UK). Minimal binding of AMs was observed. The resultant population was more than 98% pure and 95% viable (estimated by Diff-Quik staining and trypan blue exclusion). AMs were cultured for 24 h at 37°C, 5% CO₂ in the presence or absence of lipopolysaccharide (LPS) in serum-free RPMI 1640 medium supplemented with 100 U/ml penicillin and 100 µg/ml streptomycin (GIBCO-Nunc, Paisley, UK).

Measurement of TNF- Bioactivity

The Wehi 164 clone 13 mouse fibrosarcoma cell line displays dose-dependent cytotoxicity only in response to TNF- (18). Confluent cultures were rinsed with sterile phosphate- buffered saline (PBS), Ca²⁺- and Mg²⁺-free, before addition of trypsin/ethylenediaminetetraacetic acid solution to detach the cells. After centrifugation at 80 × g for 5 min, the cells were resuspended in RPMI 1640 medium containing 20% heat-inactivated fetal calf serum. A total of 50 µl of the cell suspension was added to the wells of 96-well tissue culture microtiter plates (Gibco-Nunc) and the cells were left to adhere for 2 h in an incubator. After cell attachment, 10 µl of 10 µg/ml actinomycin D (Sigma, Poole, UK) was added to each well to arrest cell division. A total of 40 µl of TNF- standards (a gift from Bayer, Newbury, UK; range, 1 to 500 pg/ml) and samples were added in triplicate to bring the final volume in the well to 100 µl. Wehi cultures were then incubated for 20 h at 37°C, 5% CO₂ in a humidified incubator. After this, 25 µl of 5 mg/ml 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) was added to each well and incubated for a further 2 h. MTT dye is metabolized by viable cells to give a purple formazan product. The cells were then lysed with 100 µl of 20% sodium dodecyl sulfate/50% dimethyl formamide, acidified to pH 4.7 with acetic acid. The plates were incubated overnight to dissolve the formazan before being read at 570 nm on a plate reader (Dynatech MR7000). Sample values were extrapolated from the standard curve using the Biolinx package (Dynatech, Billingshurst, UK). This bioassay has a detection limit of 3 pg/ml.

TNF Receptor Enzyme-Linked Immunosorbent Assay

Enzyme-linked immunosorbent assay plates (Nunc Maxisorp) were coated with 5 µg/ml 5R13 monoclonal antibody for CD120a and 5 µg/ml 7R10 monoclonal antibody for CD120b in coating buffer, 100 µl per well, and incubated for 12 h at 4°C. These antibodies (kindly provided by Dr. Sue Stephens, Celltech Therapeutics, Slough, UK) showed no cross-reactivity with other known cytokines, and their activity was not affected in the presence of human plasma or serum. The plates were then blocked with PBS containing 0.5% bovine serum albumin (BSA) for 1 h at 37°C. The wells were then washed three times on a platewasher with PBS containing 0.1% Tween 20. Samples were then added, together with the recombinant CD120a or CD120b, 100 µl per well, and the plates were incubated for 2 h at 37°C. The plates were washed four times in wash buffer, and 100 µl of biotinylated TNF-, 50 ng/ml, was added to each well. The plates were incubated for 1 h at 37°C and then washed four times with wash buffer. The bound TNF was detected with avidin peroxidase (1:400 dilution) and incubated for 30 min at room temperature, and the plates were washed four times. Tetramethyl benzidine substrate (Biostat, Stockport, UK) was then added, 200 µl per well, and the plates were left to develop in the dark for 15 min. The reaction was stopped with 50 µl of 1 M H₂SO₄ and the plates were read at 450 nm (reference filter 630 nm) on a plate reader. Sample values were calculated from the resultant standard curve. The detection limit was 0.03 ng/ml for CD120a and 0.07 ng/ml for CD120b.

Measurement of CD120b mRNA

A total of 1 million AMs was lysed in 1 ml of RNAzol B (AMS Biotechnology, Whitney, UK) and the RNA was extracted according to manufacturer's instructions. Purified RNA, 1 µg, was then reverse transcribed using a commercially available kit (Promega Corp., Madison, WI). Polymerase chain reaction (PCR) was performed on the complementary DNA (cDNA) using the following primers: CD120b 5' ATCAGACGTGGTGTGCAAG, 3' GGGTCATGATGACACAGTTCA; -actin 5' CACCTTCTACAATGAGCTGC, 3' CACGTCACACTTCATGATGG. A total of 2 µl of cDNA was added to a 25-µl reaction buffer comprising 1.5 mM MgCl₂, 0.2 mM deoxynucleotide triphosphates, 40 ng primers, and 0.75 U Taq polymerase. Thirty cycles were performed on a thermocycler (Perkin-Elmer, Warrington, UK) for both -actin and CD120b using the following program: 30 s denaturation at 94°C, 45 s annealing at 60°C, and 45 s extension at 72°C. The linearity of the system has been previously validated by serial dilution. PCR products were visualized using electrophoresis of a 1.8% agarose gel stained with ethidium bromide. Gels were photographed under ultraviolet light and densitometric analysis was performed on a calibrated Bio-Rad imaging densitometer using the molecular analyst PC package (Bio-Rad, Hercules, CA).

Detection of mTNF

The expression of mTNF was analyzed on freshly isolated AMs. Before staining, the cells were washed in 40 mM citrate buffer containing 140 mM NaCl to remove any possible receptor-bound TNF. The cells were then incubated with human immunoglobulin (Ig)G to block any nonspecific binding. In the presence of IgG, 1 × 10⁵ AMs were stained using a direct immunofluorescence technique by incubating the cells with 2 µg of fluorescein isothiocyanate-labeled anti-TNF or 2 µg/ml of anti IgG₁ isotype control. After 45 min at 4°C, unbound antibody was washed off with PBS/0.1% BSA, and the cells were resuspended in 500 µl of RPMI medium containing 1% paraformaldehyde. A total of 1 × 10⁴ AMs was acquired by fluorescence-activated cell sorter analysis (Coulter EPICS XL flow cytometer) and the results were assessed using overlaying histograms. To evaluate whether the shift in fluorescence was statistically significant with respect to the isotype control, we compared the mean increase in fluorescence above the isotype control.

Cytotoxic Function of mTNF

The cytotoxicity of the AM-associated mTNF was determined in the Wehi bioassay. Purified AMs were fixed in RPMI/1% paraformaldehyde to prevent any interference from secreted TNF. The AMs were washed four times at 200 × g and the cells resuspended at 2 × 10⁶/ml in RPMI. AMs were then added to a 96-well microtiter plate containing the Wehi cells in doubling dilutions, in the presence or absence of 10 µg/ml anti-TNF antibody (R&D Systems, Abingdon, UK). Plates were incubated and analyzed as described earlier.

Data Analysis

To correct for dilution effects, cytokine values in the BALF were adjusted by comparing BALF and plasma urea values as previously described (19) (see Table 2). Thus, samples are described as concentration per milliliter of epithelial lining fluid (ELF). There was no significant difference in the volumes of ELF returned in the BALF between the four groups. The data are presented as mean values ± standard error of the mean. Cytotoxic function of mTNF was expressed as a cytotoxicity index: [(1  test absorbance) × medium-alone absorbance]. The data, which were normally distributed as determined by the Ryan Joiner normality test, were compared using the Student's t test on Minitab for Windows. A P value less than 0.05 was regarded as significant. Not all of the analyses could be performed in each individual due to the limited amount of material obtained from each subject.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

mTNF Is Present on AMs and Increased in Subjects with ARDS

mTNF expression was observed by flow cytometry on all AMs analyzed (Figure 1). AMs derived from subjects with ARDS (n = 9) had the highest degree of staining, with an increase in mean fluorescence compared with isotype control of 22.3 ± 3.5% compared with 7.09 ± 1.70% for At Risk subjects (n = 9) (P < 0.003) and 10.93 ± 2.01% for ventilated controls (n = 3) (Figure 2).

View larger version (11K): [in this window] [in a new window]   Figure 1.   mTNF expression on AMs derived from subjects with ARDS, At Risk subjects, and VC.

View larger version (10K): [in this window] [in a new window]   Figure 2.   Percentage increase in mean fluorescence of AMs labeled for mTNF: ARDS (n = 9), ALI (n = 2), At Risk (n = 9), VC (n = 3).

mTNF on AMs Derived from Subjects with ARDS Is Biologically Active In Vitro

Fixed AMs in coculture with Wehi cells displayed cell number-dependent cytotoxicity. The cytotoxic index was increased in subjects with ARDS relative to At Risk subjects (Figure 3). The mean cytotoxicity index was 1.499 ± 0.021 for neat AMs derived from subjects with ARDS (n = 4) compared with 1.127 ± 0.021 for At Risk subjects (n = 4) (P = 0.0004). The observed cytotoxicity was reversed in the presence of neutralizing anti-TNF antibody. mTNF cytotoxicity positively correlated with the mTNF expression determined by flow cytometry (r² = 0.97) (Figure 4).

View larger version (37K): [in this window] [in a new window]   View larger version (30K): [in this window] [in a new window]   Figure 3.   Cytotoxic activity of fixed AMs derived from (A) subjects with ARDS (n = 4) and (B) At Risk subjects (n = 4). *** P < 0.004 ARDS versus At Risk.

View larger version (9K): [in this window] [in a new window]   Figure 4.   Correlation between mTNF expression by flow cytometry and cytotoxicity of AMs in the bioassay system.

TNF- Bioactivity Is Not Significantly Elevated in the ELF or in AM Supernatants Derived from Subjects with ARDS

Although the subjects with ARDS had slightly elevated bioactive TNF in the ELF, there were no significant differences between the groups. There were no significant differences in BALF TNF levels between the groups (see Table 2). The mean ARDS TNF in the ELF was 5.309 ± 1.93 ng/ml (n = 40) compared with 1.009 ± 0.375 in ALI subjects (n = 7), 1.086 ± 0.475 ng/ml for At Risk subjects (n = 20) (P < 0.05), and 0.617 ± 0.341 ng/ml for VC (n = 6) (Figure 5). Spontaneous production of soluble TNF by AMs was increased in the subjects with ARDS and the ALI subjects (mean TNF, 1.414 ± 1.042 and 0.819 ± 0.624 ng/ml, respectively) compared with At Risk subjects and VC (mean TNF, 0.242 ± 0.106 and 0.292 ± 0.16 ng/ml, respectively). This pattern was also observed in AMs stimulated with LPS. These differences, however, did not reach statistical significance (Figure 6).

View larger version (10K): [in this window] [in a new window]   Figure 5.   Bioactive TNF detected in the ELF derived from subjects with ARDS (n = 40), ALI (n = 7), At Risk subjects (n = 20), and VC (n = 6).

View larger version (12K): [in this window] [in a new window]   Figure 6.   Bioactive TNF present in AM supernatants derived from subjects with ARDS (n = 18) and At Risk subjects (n = 10). AMS were either cultured unstimulated or stimulated with 10 µg/ ml LPS for 24 h.

Release of Soluble CD120b Is Diminished in AMs Derived from Subjects with ARDS Compared with At Risk Subjects

Cultured AMs derived from subjects with ARDS released significantly lower levels of CD120b into the supernatant than did cultured AMs from At Risk subjects. The mean CD120b level was 0.264 ± 0.058 ng/ml in ARDS supernatants, compared with 0.593 ± 0.143 ng/ml in At Risk subjects (P < 0.05) (Figure 7). By contrast, there was a significant increase in CD120b mRNA in AMs derived from subjects with ARDS versus At Risk subjects (mean CD120b:-actin ratio, 10.44 ± 2.32 versus 2.71 ± 1.12, P < 0.03) (Figure 8). There were no differences in the levels of either CD120a or CD120b in the BALF (data not shown). There was no difference in the levels of CD120a between the groups.

View larger version (11K): [in this window] [in a new window]   Figure 7.   sTNFR detected in AM supernatants derived from subjects with ARDS (n = 18) and At Risk subjects (n = 10).

View larger version (7K): [in this window] [in a new window]   Figure 8.   Ratio of CD120b:-actin mRNA in AMs derived from subjects with ARDS (n = 5), ALI (n = 3), At Risk subjects (n = 4), and VC (n = 3).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

This study reports for the first time the presence of increased levels of functional mTNF on the surface of AMs derived from subjects with ARDS. All previous studies looking at TNF in ARDS have measured the soluble form, and although this may be a good marker of systemic TNF activity, it may not necessarily reflect the local activity of TNF within the lung. Many cytokines, including TNF, are able to function in a membrane-associated form by juxtacrine intercellular signaling (20). There has been much recent debate regarding the contribution of soluble TNF to the pathogenesis of acute lung injury and ARDS, inasmuch as it has proven difficult to correlate TNF levels in the plasma or BALF to the degree of mortality or lung injury score. Indeed, it is known that the kinetics of soluble receptor release closely mirror that of TNF in sepsis (23), suggesting that the bioactivity of soluble TNF is restricted almost immediately by the increased availability of its biologic inhibitors. The present paper and another recent study (24) have failed to find increased soluble TNF bioactivity in the lungs of patients with ARDS. It is now apparent, however, that the 26-kD form of TNF may have clinical significance. Although previously unexplored in lung injury, mTNF has been shown to correlate with mortality in septic shock (25). The pathologic potential of mTNF is highlighted by the very close correlation between mTNF expression and cytotoxicity of AMs in the cytotoxicity assay. AMs expressing high levels of mTNF may have a direct interaction with the alveolar capillary barrier, contributing to the diffuse alveolar damage. mTNF may also have an important role in the interaction between AMs and the neutrophil alveolitis characteristic of ARDS.

Although much is now known about cell signaling in response to soluble TNF, there is little information about mTNF signaling through direct cell-cell contact. It has been demonstrated that soluble TNF signals mainly through CD120a, with CD120b ligand passing to enhance the signal (26). Ligand passing is not required for the membrane-anchored TNF, which raises the question of whether there is another role for CD120b. Previous studies have shown that the pathologic effects of mTNF in experimental hepatitis and arthritis require both CD120a and CD120b (8, 27). Another study (9) has demonstrated increased severity of liver damage in transgenic mice overexpressing CD120b and also demonstrated a prominent role for CD120b in the mTNF-dependent upregulation of interferon-. The present study has shown that there is reduced presence of CD120b in the AM supernatants derived from subjects with ARDS. Coupled with the observed increase in CD120b mRNA expression, this suggests that there may be increased retention of CD120b on the surface of AMs in ARDS. The combination of increased mTNF expression and CD120b surface receptor expression would enhance the mTNF signal in these patients.

The availability of CD120b on the cell surface is regulated in part by the catalytic activity of the matrix metalloproteinase (MMP) family of enzymes, of which the adamalysin TACE (also known as ADAM 17) is also a member (28). TACE is thought to be constitutively expressed on all human tissue (5), but there has been very little research into its regulation or expression in the human lung or on macrophages. A recent study has shown that a tissue inhibitor of MMPs (tissue inhibitor of metalloproteinase [TIMP]-3) can inhibit TACE activity in vitro (29). This is a potentially important finding, inasmuch as it has previously been demonstrated that TIMP levels are enhanced in the BALF of subjects with ARDS compared with At Risk subjects (30). Another study has shown that a hydroxymate MMP inhibitor exacerbated mTNF-dependent liver injury in mice, and hypothesized that this could be due to inhibition of TACE and retention of CD120b (31). Our current findings highlight the possibility that the increased TIMP levels previously described in ARDS may contribute to the increased mTNF expression and decreased CD120b shedding that we have observed. It needs to be considered that there is some doubt whether inhibition of TACE does lead to mTNF accumulation, and other pathways have been suggested for its degradation. However, this has yet to be explored in human AMs, and investigation into TACE regulation in these cells is now underway in our laboratory.

In summary, we have found increased mTNF expression on the surface of AMs derived from subjects with ARDS compared with At Risk subjects, which suggests that macrophage-associated TNF activity is upregulated in this syndrome. mTNF on the surface of AMs is functional and therefore has the potential to contribute to lung injury, possibly through increased surface expression of CD120b. We propose that the modulation of TNF- processing in the lung by TACE may be important in pathogenesis of lung injury and warrants further investigation.