Susceptibility of lung epithelium to neutrophil elastase: protection by native inhibitors

The development of emphysema is thought to be due to an imbalance of proteases (especially neutrophil elastase [NE]) and antiproteases with loosening of the respiratory epithelium as an early event. We investigated the effect of NE on respiratory epithelial cell adherence in vitro , in the presence of varying concentrations and combinations of native inhibitors, α-1-proteinase inhibitor (PI) and secretory leukoprotease inhibitor (SLPI). SLPI was two to 12 times more effective than PI at preventing the effects of NE, especially when enzyme:inhibitor ratios were almost equivalent. Even when the concentration of SLPI was only 10% of the total (as in normal peripheral lung secretions), it gave greater protection than PI alone. This suggests that SLPI plays an important role in controlling neutrophil elastaseinduced inflammation and tissue damage.


Introduction
Historically neutrophil elastase (NE) has been regarded as the primary mediator of tissue damage which occurs during pulmonary emphysema, notably by degradation of extracellular matrix. 1 More recent evidence suggests that NE activates certain metalloproteinases whilst inactivating their inhibitors 2,3 and that metalloproteinase levels are raised in emphysematous patients 4 suggesting that NE and metalloproteinases act in concert in the aetiology/pathology of this disease. In addition, NE has been shown to play a pro-in ammatory role in pulmonary in ammation involving neutrophil in ux. For example, activation of complement to C5a 5 and stimulation of IL-8 synthesis and secretion by airway epithelial cells 6 would amplify neutrophil in ux, whilst its ability to reduce airway epithelial cell ciliary beat frequency and secretagogue activity 7,8 would contribute to reduced clearance during diseases such as chronic bronchitis and cystic brosis. Thus NE activity is normally strictly regulated by systemic and locally produced inhibitors. SLPI (12 kD) is a reversible inhibitor of NE, synthesized and secreted by non-ciliated airway and bronchiolar secretory cells; 9 SLPI will inhibit NE even when bound to a substrate. 10 In contrast, PI (52 kD) is the major serum-derived inhibitor of NE and transudes into the intersti-tium and airspaces of the lung. It complexes with, and irreversibly inhibits, free NE 10 as well as dissociating and inactivating NE bound to other, reversible, inhibitors e.g. SLPI, 11 prior to clearance of NE:PI complexes from the circulation. Ela n (12 ±14 kD) is synthesized by both Clara-like and alveolar type II pneumocyte-like cell lines; 12 the mechanism of action of ela n has not been completely characterized. Analysis of the antineutrophil elastase screen in human bronchoalveolar lavage shows that although PI forms a relatively high proportion of this screen 13 (except in PI de ciency, which predisposes to the development of emphysema), the other locally produced inhibitors are present at levels that suggest they also play a signi cant role in controlling the action of NE in epithelial lining uid. 14 Like PI, SLPI and ela n may also vary in concentration between healthy individuals and according to the disease state of the lung. 13,15 Thus it seems likely that control of extracellular NE could depend not only on the relative levels and synchronized action of these different inhibitors but also on the levels and site of deposition of NE i.e. within the epithelial lining uid or interstitium.
Induction of SLPI mRNA levels in primary human airway epithelial cells, 16 and retention of SLPI activity by proteolysed SLPI fragments following exposure to NE, 17 suggests that increased SLPI levels provide a sustained local anti-NE defence mechanism against excessive pulmonary NE activity. In addition, aerosolized rSLPI stimulates glutathione levels in sheep epithelial lining uid 18 and downregulates IL-8 levels in respiratory secretions from patients with cystic brosis, 19 suggesting that SLPI has an important anti-in ammatory role in the lung.
Previous investigations comparing the ability of PI and SLPI to prevent neutrophil-mediated proteolysis of brinogen 20,21 and neutrophil-induced detachment of human umbilical endothelial cells 22 have shown that SLPI is a better inhibitor than PI when used at the same concentration but separate from each other. SLPI and PI were never used in combination, as is known to occur in vivo. Although it is likely that in the absence of suf cient antiprotease protection, NE also disrupts peripheral structural lung cell integrity during emphysema, there are no quantitative studies to illustrate this or to show how PI and SLPI, alone or combined, inhibit the effects of NE on peripheral lung cells.
Two important epithelial cells found in the area affected by emphysema, the respiratory bronchioles and alveolae, are the non-ciliated bronchiolar secretory cell (NCBSC) and type II pneumocyte respectively. Apart from producing SLPI, the NCBSC produces other proteins, such as surfactant apoproteins 23 -25 and Clara cell secretory protein, 26 while a major function of the type II cell is to produce pulmonary surfactant, 27 the extracellular lining layer responsible for maintaining reduced surface tension and preventing alveolar collapse. Both cell types are involved in the metabolism of xenobiotics 28,29 as well as acting as progenitors to type I alveolar epithelial cells (type II 30 ) and ciliated cells (NCBSC 31 ). Damage to these epithelial cells could therefore have serious consequences in terms of lung defence and regeneration of damaged tissue. Adequate inhibition of NE during in ammatory cell in ux to the respiratory units would therefore be critical in preventing the occurrence of emphysema and in ammatory lung disease processes. 32 It is not possible to investigate the exact relationship between changes in NE load, inhibitor pro le and lung tissue damage that may be relevant to the development of emphysema in an in vivo model. Consequently, we have adapted a model that was previously used to assess bronchiolar 33 and alveolar 34 toxicity in vitro, to study the effect of NE on respiratory epithelial cells under de ned experimental conditions. The speci c aim was to determine the effect of NE on pulmonary epithelial type II and NCBSC cell detachment from (i.e. damage), and their ability to adhere to (i.e. repair), extracellular matrix in vitro. The use of matrix was regarded as crucial as, in vivo, these cells would either be in contact with, or re-populate, extracellular matrix of either epithelial or endothelial origin. Therefore, matrices generated by both epithelial and endothelial cells were examined. As our previous studies on human bronchoalveolar lavage showed that the ratio of PI to SLPI varied, 13 the protective effect of PI and SLPI on NE-induced damage was also investigated when used alone and when combined in ratios representing that in normal bronchoalveolar lavage (PI:SLPI 9:1 13 ) and bronchoalveolar lavage from smokers with obstructive lung disease (PI:SLPI 1:1 13 ).

Activity of NE and antiproteases
The relative activity of commercially obtained NE (Elastin Products Comp. Inc., MO, USA) was assayed using the method of Nakajima et al. 35 The activity of PI (Sigma, A9024) and SLPI (puri ed in this laboratory as described previously 14 ) was assessed against this NE and the molar concentration of active inhibitor could then be determined on the basis that PI and SLPI inhibit NE on a 1:1 molar basis.
These assays were carried out immediately before each experiment and inhibitory capacity was also assessed after incubation under experimental conditions at 378 C (without cells) for 18 h.

Animals and removal of lungs for epithelial cell preparation
Male mice (BALB/C 25 ±30 g) and rats (WAG 250 ±300 g), bred in-house, were killed by a lethal intra-peritoneal injection of phenobarbitone/0.15 M saline (1:1 v/ v) containing 300 units/ ml heparin. The lungs were perfused free of blood by gravity feed of 0.15 M NaCl via the pulmonary artery and carefully removed intact from the cavity with a tracheal cannula still tied in place. The excised lungs were lavaged to remove free alveolar cells and pulmonary airway secretions (4 3 0·6 ml [mice] and 4 3 10 ml [rats] of sterile 0.15 M NaCl).

Isolation of mouse NCBSC
NCBSC were isolated using the method of Oreffo et al. 36 The nal cell sample was suspended in DCCM-1 medium (Biological Industries, Glasgow) containing 0.4% Ultroser G (Gibco, Paisley) and plated according to the protocol described below.

Isolation of rat type II cells
Type II cells were isolated using the method of Richards et al. 37 The nal cell pellet was processed in the same way as the NCBSC.

Identi® cation of NCBSC
Functional NCBSC may be distinguished from other cells present in the preparation by their high levels of nitroblue tetrazolium (NBT) reductase. The method of Oreffo et al. 36 was used to estimate the proportion of NCBSC present in the original isolate as well as those remaining after treatment with enzyme and inhibitors.

Identi® cation of type II cells
The type II cell is the only cell in the lung with appreciable levels of the enzyme alkaline phosphatase, this was therefore used to identify type II cells in the original cell isolate following the method of Miller et al. 38 To estimate the number of cells remaining after enzyme/inhibitor treatment the cells were stained with Giemsa which has the advantage over the alkaline phosphatase stain which does not involve xation and cannot be stored for long periods of time to enable counting.

Experimental protocol
The experimental model was a modi cation of those described previously for toxicological studies on pulmonary alveolar and bronchiolar epithelial cells in vitro. 33,34 To mimic the in vivo conditions of adherence to extracellular matrix of endothelial origin, 96 well plates, in which endothelial cells were previously grown under controlled conditions, were used (Biological Industries). The cells are removed prior to purchase, leaving a coat of endothelial cellderived extracellular matrix. This matrix is referred to as endothelial matrix in this study. It is similar in its composition to naturally occurring endothelial cell basement membrane which contains type IV collagen, laminin, heparan sulphate proteoglycan, bronectin, entactin and nidogen. To mimic the in vivo situation of adherence to epithelial matrix, untreated 96 well plates were coated with Matrigel (Gibco, Paisley), an extracellular matrix of basement membrane synthesized by Engelbreth-Holm-Swarm mouse sarcoma consisting of laminin, type IV collagen, heparan sulphate proteoglycan, entactin and nidogen. This matrix is referred to as epithelial matrix in this study.
Cells were plated onto 96 well plates, 5 3 10 4 cells/ 100 ml DCCM-1 + 0.4% Ultroser G (media)/well, which were coated with either endothelial or epithelial matrix. Cells were studied either prior to adherence or following adherence, as follows: (a) adherent cells, allowed to adhere for 16 h, the media removed, 100 ml of fresh media added to each well and exposed to NE and inhibitors for 2 h; (b) freshly pl ated cells, treated with enzyme and inhibitors at the time of plating and incubated for 18 h.

Effect of NE and inhibitors on cell adherence/detachment
Fifty ml of media containing 0, 2, 10, 20 and 40 pmol active inhibitor in each of the following combinations-PI alone, SLPI alone, 9 PI:1 SLPI and 1 PI:1 SLPI-were then added. Immediately after, 50 ml of media containing 0, 3, 6, 12 and 24 pmol active NE were added to each dose and each combination of inhibitor. Thus cells were incubated (i) alone, (ii) with all doses of NE, and (iii) with each dose of NE mixed with every dose and combination of inhibitor. After 2 (adherent cells) or 18 h (freshly plated cells; see above), the incubation media and any non-adherent or dislodged cells were removed. Non-adherent cells were assayed for viability with trypan blue, while adherent cells were stained and counted as described above. This experiment was performed in triplicate for each set of conditions on three separate occasions.

Enumeration of cells
After staining, a representative tract of the culture was selected across the diameter of the well to account for any plating anomalies. Three different areas were counted along this tract (for control cultures this represents 800 ±1000 cells) and from this the mean number of cells/ unit area could be ascertained. Within any one well the standard error of the mean, in control cultures, or those treated with the same dose of HNE (where applicable) did not exceed 15%. The number of cells/ unit area determined for the control culture was taken as 100% and the number of attached cells/ unit area in the presence of elastase and/or inhibitors was expressed as a percentage of the control value.
All data obtained from these experiments were analysed using analysis of variance (BMDP Statistical Software, University of California).
Release by NE of endothelial and epithelial matrix from tissue culture plastic Matrix that is not populated by adherent cells may be susceptible to NE (e.g. in studies of non-adherent cells) and is likely to be degraded/ modi ed and contribute to changes in epithelial cell adherence. Inhibitors may modify the effect of NE on the matrix. Thus, both endothelial and epithelial matrix-coated plates were exposed to NE and inhibitors in an identical manner to that described above for non-adherent cells but in the absence of cells, to examine the effect of NE on the matrix alone. After 18 h, the medium was removed, the wells washed in PBS buffer and the remaining matrix protein level assayed in situ using the Bicinchoninic acid (BCA) protein assay as described in the manufacturer's instruction manual (Pierce, Rockford, Illinois, USA). Absorbance was measured at 562 nm on a Labsystems MCC340 plate reader and protein levels expressed as a percentage of the nonproteolysed control value.

Cell isolation
The proportion of type II cells present in the original isolate was 95 ±99% whilst the proportion of NCBSC was 70 ±75%. Fig. 1 shows representative cultures, growing on endothelial matrix (type II, Fig. 1A; NCBSC, Fig. 1D).

Effect of NE on ECM and matrigel
Elastase caused the release of endothelial matrix from tissue culture plastic in a dose-dependent manner; even 3 pmol of enzyme caused 35% degradation whilst 12 and 24 pmol caused over 80% degradation (Fig. 2). There was no release of epithelial matrix (data not shown).

Effect of PI and SLPI on NE-induced endothelial matrix degradation
There was no loss of NE activity or inhibitory capacity of PI or SLPI incubated in tissue culture media for 18 h (data not shown). When endothelial matrix was incubated with low doses of NE, 10 pmol or more of PI or SLPI performed similarly and prevented, or signi cantly reduced, endothelial matrix degradation (Fig. 2). However, the effect of higher doses of NE could not be completely prevented by the addition of a molar excess of PI or SLPI, although SLPI was more effective.

NE-induced detachment of adherent cells
Three pmol of NE did not induce epithelial cell detachment and no further studies were carried out with this dose of enzyme (data not shown). Although 6 pmol NE did not affect NCBSC or cause type II cell detachment from endothelial matrix, this dose of enzyme did cause type II cell detachment from epithelial matrix (~40% detachment, Fig. 3A). Furthermore, type II cells were more susceptible than NCBSC to NEinduced detachment from both matrices after exposure to 12 pmol NE (P , 0·001). Twentyfour pmol NE had a similar effect on both cell types causing 70 ±80% detachment from both matrices.
Thus, unlike type II cells, detachment of NCBSC was not differentially affected by the matrix to which they adhered but was related to dose (Fig. 3A). This effect of NE is illustrated in Fig. 1.

Effect of NE on cell adherence of freshly plated cells
The effect of NE on the adhesion of freshly plated cells is shown in Fig. 3B. Since the initial study with adherent cells showed that 6 pmol NE had little effect, the effect of this dose of enzyme was not examined.
The adherence of freshly plated type II cells on to endothelial matrix was reduced to 33% and 7% of untreated cells when exposed to 12 and 24 pmol NE respectively, while the adherence of NCBSC was virtually abolished (Fig.  3B). The effect of 12 pmol NE on type II cell adherence to epithelial matrix was similar to that seen on endothelial matrix (28% of control). However, NCBSC adherence to epithelial matrix was less affected (Fig. 3B). The effect of 24 pmol NE on the adherence of both cell types to epithelial matrix was similar (~12% of control).

Differential effect of NE on freshly plated compared with adherent epithelial cells
The effects of 24 pmol NE on adherence of freshly plated cells was more marked than that on cell detachment; thus in every equivalent combination of cell type, matrix and dose of enzyme, proportionally more freshly plated cells

Mediators of In¯ammation´Vol 6´1997 349
were prevented from adhering than were displaced by NE (Fig. 3).

Prevention of NE-induced changes in cell adherence by PI and SLPI
Two pmol of inhibitor had a negligible effect on NE-induced cell attachment or detachment. The effect of 6 pmol of NE was ef ciently prevented by all levels and combinations of inhibitors (data not shown). The effect of inhibitors on 12 and 24 pmol NE were similar, and related to the ratio of enzyme to inhibitor. Consequently the data for 24 pmol NE is presented as this allows discussion of a wider range of inhibitor to enzyme ratios. The protective effects of inhibitor are illustrated pictorially in Fig. 1.

Prevention of NE-induced cell detachment by PI and SLPI
Endotheli al m atrix. Forty pmol of any combination of inhibitor prevented the effects of 24 pmol NE on type II cell detachment. Addi-

Mediators of In¯ammation´Vol 6´1997
tion of approximately equimolar concentrations of inhibitor (i.e. 24 pmol NE: 20 pmol inhibitor) partially prevented NE-induced type II cell detachment. This effect was positively related to the level of SLPI, SLPI being approximately 1.5 times as effective as PI alone (Fig. 4A). A small but signi cant effect on NE-induced type II cell detachment was seen with 10 pmol inhibitor.
In contrast, there was no difference between PI and SLPI in their ability to prevent NEinduced NCBSC detachment from endothelial matrix. The reduction in cell detachment in the presence of inhibitor was directly related to inhibitor level, such that with 10 pmol of inhibitor only 50% (compared with 20%) were dislodged, while there was no cell detachment with 40 pmol of inhibitor (Fig. 4B).
Epitheli al m atrix. Forty pmol of any combination of inhibitor prevented NE-induced cell detachment of both type II and NCBSC cells. The action of 24 pmol NE on both cell types could be inhibited by 20 pmol of inhibitor, inhibition being more effective with increasing SLPI to PI ratio (SLPI 1.5 ±2 times more effective than PI, Fig. 4C,D). Interestingly a similar inhibitory pattern was observed when NCBSC cells were exposed to 24 pmol NE and 10 pmol inhibitor.

Prevention of the effects of NE on epithelial cell adherence by PI and SLPI
Endotheli al m atrix. The effects of 24 pmol NE on either cell type could not be completely prevented by 20 or 40 pmol of either inhibitor despite their ability to inhibit NE on a 1:1 molar basis in the absence of cells (Fig. 5A). Forty pmol PI signi cantly prevented the effects on type II cell adherence, allowing adherence to increase from 7%to 50%of the unexposed control level. However, 40 pmol SLPI was more effective, causing a reversal that achieved adherence that was 89%of unexposed control levels (Fig. 5A). A similar but less effective pattern on NE inhibition was observed with NCBSC exposed to 24 pmol NE. Thus, although there was no improvement in adherence of NCBSC after addition of 40 pmol PI, 40 pmol SLPI allowed 51% of the control adherence level. Inhibition was nevertheless  incomplete. For both cell types, the most effective reversal of the effect of NE occurred when SLPI was present. This was most noticeable when there were almost equimolar levels of NE and inhibitor, the ef cacy of the inhibitor increased with increasing levels of SLPI, SLPI being two to 12 times more effective at preventing the effects of NE than PI (Fig. 5A,B).
Epitheli al m atrix. In contrast to cell adherence to endothelial matrix, the effect of 24 pmol of NE on both cell types was almost completely prevented by 40 pmol of any combination of inhibitor whilst the lower doses of inhibitor were also signi cantly more effective at preventing the effects of 24 pmol NE in this situation (Fig. 5). Once again at equimolar concentrations of NE with inhibitor, SLPI was more effective (approximately 1.5 times).

Discussion
Mediators released by neutrophils are thought to make a signi cant contribution to tissue destruction; NE, in particular, has been shown to play an important role in the initiation and perpetuation of an in ammatory response. Studies in experimental animals suggest that loosening of the respiratory epithelium may be an early event in the development of emphysema. 39 We have shown that this could result from release of excessive NE by neutrophils at the apical pulmonary respiratory epithelium. Although there have been numerous studies on the effect of NE on airway epithelial cells 7,40 little is known about its effect on the respiratory epithelium. In this study we examined the effect of NE on adherent respiratory epithelial cells (type II and NCBSC), representing resident epithelial cells in situ, and non-adherent cells, representing dislodged epithelial cells in situ. Dislodged or loosened cells may exist as a result of previous protease action, oxidant action, due to the cytotoxic effects of cigarette smoke, or for other, unknown reasons. As expected, when adherent cells were exposed to NE, dose-dependent detachment of both cell types was observed, although the degree of detachment was greater with type II cells than with NCBSC. In addition, the effect of NE on adherent NCBSC was independent of the source of the extracellular matrix, whereas type II cells were more susceptible to NE-induced detachment from epithelial matrix than from endothelial matrix. This suggests that, in situ, NCBSC are less likely to be displaced by NE than alveolar type II cells, which are most vulnerable when in contact with epithelial cellderived matrix.
In contrast, the effect of NE on cell adherence of freshly plated cells was more marked and suggests that alveolar type II cells would be more likely than NCBSC to 're-populate' denuded, exposed matrix. Since NE degrades endothelial matrix (possibly re ecting degradation of the bronectin component) and NCBSC do not adhere to tissue culture plastic, it seems possible that susceptibility of extracellular matrix to NE and preference of NCBSC for a speci c matrix may exist in vivo and hence be important in repopulation and repair mechanisms. Furthermore, NE may be acting on cell membrane proteins, such as adhesion molecules, since both type II cell and NCBSC adherence to epithelial matrix was inhibited by NE, even though the epithelial matrix used in this study was not signi cantly solubilized by NE under these experimental conditions. Protection, by SLPI, of epithelial cells from NE-induced damage, was equal to, or better than, the serum-derived inhibitor, PI, and was most noticeable when the enzyme:inhibitor ratios were almost equivalent, a situation that might occur in the microenvironment of the phagocyte in vivo. Indeed, PI was often a poor inhibitor of the effects of NE, particularly on adherence of freshly plated cells to endothelial matrix. This probably re ects two factorsendothelial matrix is proteolysed by NE and PI cannot inhibit NE bound to extracellular matrix. 10,32 Thus, prior to cell adherence, NE probably bound rapidly to endothelial matrix and, having done so, could only be inhibited by SLPI (SLPI is believed to bind to extracellular matrix and can inhibit extracellular matrixbound NE). 10,41 This explanation is supported by the observation that, when studied in the absence of cells, more endothelial matrix was found to be solubilized by NE in the presence of PI than SLPI. Such complex interactions may explain why a molar excess of inhibitor would not always prevent NE-induced epithelial cell damage. Reduced levels or activity of SLPI in vivo (which is released both apically and basally, into the interstitial extracellular matrix, by secretory epithelial cells 42 ) might mean that NE that has bound to extracellular matrix cannot be suf ciently inhibited by other inhibi-tors present within the lung tissue and hence that tissue proteolysis will progress. This is supported by a recent study 43 which showed a negative correlation between elevated NE activity and SLPI levels in sputum from patients who had exacerbative chronic obstructive pulmonary disease, suggesting that the condition was worse when SLPI levels were reduced. There was no correlation between PI levels, NE activity and respiratory symptoms, possibly highlighting the signi cance of SLPI in controlling the action of NE and severity of disease.
It is interesting that adherent NCBSC were markedly more resistant to NE than non-adherent NCBSC and also that the difference between the protective effects of PI and SLPI were less noticeable following epithelial cell adherence. As the cells had already adhered, less extracellular matrix or cell surface molecules would be exposed to NE and hence less 'protection' would be required from the inhibitors. PI would be more likely to inhibit unbound NE, having a higher K ass than SLPI for free NE. 11,44 Synthesis and secretion of SLPI by the adherent NCBSC may also provide increased protection from NE, although analysis of NCBSC culture supernatant with antibodies to human SLPI was negative (data not shown). However, this may re ect lack of cross-reactivity between species.
The data suggest that even when the proportion of SLPI in the antiprotease mixture added to the cells is only 10% of the total (cf. normal lung ratios 13,15 ), the degree of protection from NE-induced damage is increased above that of PI alone. Since those with bronchitis and emphysema may have even higher proportions of SLPI (50% 13 ) and in the present study a 1:1 ratio of PI to SLPI conferred greater protection from NE-induced damage, one could hypothesize that increased production of SLPI by the pulmonary epithelium gives extra protection and may be a defence mechanism rather than a non-speci c result of hypersecretion.
In conclusion, this study suggests that incomplete inhibition of NE released at the apical surface of respiratory epithelium may contribute to loosening and possibly detachment of the epithelium. The magnitude of the effect is likely to depend on both the nature of the underlying matrix (i.e. susceptibility to NE) and the differential action of the enzyme on epithelial cells from bronchiolar and alveolar regions of the respiratory unit, including adhesion molecules. Weak epithelial cell adhesion and cell detachment is likely to facilitate access of enzyme to the underlying interstitium and extracellular matrix, including basement membrane and elastic tissue, which are susceptible to proteolysis by NE. Such mechanisms may contribute to the tissue damage that occurs during the development of emphysema. Finally, this study shows that SLPI, particularly in combination with PI, plays a major role in protecting lung cells and interstitial components from the destructive action of NE. We hypothesize that reduced levels or activity of SLPI in vivo may contribute to NE-induced lung damage and neutrophil-mediated lung disease.