INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Exacerbations of chronic obstructive pulmonary disease (COPD) are common, particularly during the winter months, and are associated with neutrophilic inflammation in the airways. An exacerbation may be precipitated by viral infection or environmental pollution. Bacterial infection, when it occurs, is often a secondary event and is present in about 50% of patients (23, 24, 38, 42). Nontypeable Haemophilus influenzae, Streptococcus pneumoniae, and Moraxella catarrhalis are the commonest bacterial pathogens causing airway infections. Antibiotics are usually prescribed because of the lack of clinical features that can be used to discriminate between those patients who are infected and those who are not (10). In two recent studies between 13 and 25% of patients with an infective exacerbation had to attend a second consultation because of a poor response to the initial treatment (3, 22). This might be explained by bacterial resistance to the prescribed antibiotic since the incidence of beta-lactamase production by H. influenzae and M. catarrhalis and the incidence of pneumococcal penicillin resistance are increasing (4). Another explanation might be that airway inflammation is poorly controlled, despite successful eradication of any bacterial infection that is present.

Macrolide antibiotics are commonly prescribed during exacerbations of COPD and have an advantage in that they are active against atypical bacteria such as mycoplasma and chlamydia. However, with the exception of certain newer macrolides, most have relatively poor activity against H. influenzae (4). Macrolides have several properties other than their antibiotic action that might be important during exacerbations of COPD. They have been shown in vitro to be anti-inflammatory, stimulate ciliary beat frequency (CBF), reduce mucus production, and decrease toxin production by bacteria (2, 11, 13, 16, 18, 21, 26, 31-33, 36). Dirithromycin is a macrolide antibiotic which is usually given once daily and which achieves levels of 1 to 2 µg/ml in the respiratory mucosa. The MIC of dirithromycin for H. influenzae is variable between 1 and 8 µg/ml (6).

H. influenzae infection of the respiratory mucosa results in the slowing of the CBF, damage to the epithelium, and the loss of ciliated cells (27, 34). We have used several in vitro assays to assess the effect of dirithromycin on the interaction of H. influenzae with the respiratory mucosa.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacteria. H. influenzae SH9, a nontypeable clinical isolate that has previously been studied in our laboratory (27, 34), was cultured at 37°C with agitation for 24 h in brain heart infusion broth supplemented with hemin (5 µg/ml) and nicotinamide (10 µg/ml). The broth culture contained approximately 10¹⁰ cfu ml¹ and was centrifuged and filtered (pore size, 0.45 µm) to produce a sterile culture filtrate. The bacterial pellet from 2 ml of an overnight broth culture was washed twice through 10 ml of phosphate-buffered saline (PBS; Oxoid, Basingstoke, United Kingdom) and resuspended in 150 µl of PBS, and viable counts were determined. The MIC of dirithromycin for SH9 was 1 µg/ml.

Effects of dirithromycin and amoxicillin on the damage caused by H. influenzae broth culture filtrate to human nasal epithelium. Strips of normal human nasal ciliated epithelium were obtained with a cytology brush (31, 39) from both inferior turbinates of healthy volunteers who had been free of respiratory infection for at least 4 weeks. This procedure has been approved by the Royal Brompton Hospital Ethics Committee. The strips were dispersed by gentle agitation of the brush in 2 ml of cell culture medium 199 with Earle's salts and HEPES (N-hydroxyethylpiperazine-N'-2-ethanesulfonic acid; Flow Laboratories, Irvine, Scotland, United Kingdom). The cells were preincubated at 37°C with dirithromycin (1 µg/ml) or medium 199 for 4 h prior to the addition of the strain SH9 culture filtrate. For each experiment (n = 6), three sealed microscope coverslip-slide preparations containing approximately equal numbers of strips of nasal epithelium were constructed: medium 199 alone (control), SH9 culture filtrate (1:1) with epithelium in medium 199, and SH9 culture filtrate (1:1) with epithelium preincubated with dirithromycin in medium 199. CBF was measured by a photometric technique (39), and epithelial integrity was assessed by light microscopy (1, 31) at hourly intervals for 4 h. At each time point 10 recordings of CBF were made from the same sites on the epithelium, and each site was recorded as either intact (smooth outline of epithelium) or disrupted (outline uneven due to disruption of the integrity of the epithelium and cell projection). The same experiments were performed with amoxicillin (2 µg/ml). We also investigated whether brain heart infusion broth supplemented with hemin and nicotinamide (1:1 with medium 199) itself affected CBF or epithelial integrity (n = 6). At the end of the dirithromycin experiments the epithelium was fixed in 2.5% sodium cacodylate-buffered gluteraldehyde and was prepared for transmission electron microscopy to assess ultrastructural damage as described previously (1, 9).

Effect of dirithromycin on normal CBF. For each experiment (n = 6), four sealed microscope coverslip-slide preparations containing approximately equal numbers of strips of nasal epithelium were constructed: medium 199 alone (control) or medium 199 containing dirithromycin (0.125, 0.5, or 8.0 µg/ml). CBF was measured hourly for 4 h.

Assessment of tissue by transmission electron microscopy. All specimens were coded prior to analysis. Each cell was scored for the following features: projection from the epithelial surface was recorded as 0 (cell in line with epithelial surface), +, or ++ (cell still in contact with the epithelium but almost completely extruded); cytoplasmic blebbing originating from the luminal surface was recorded as 0 (absent), minor, or major (large blebs); and mitochondrial damage evidenced by swelling and disruption of the cristae was recorded as present or absent. Cytoplasmic blebbing and mitochondrial damage were scored separately for ciliated and unciliated cells.

Effect of dirithromycin on sputum transportability. A previously described bovine trachea model was used to examine the effect of dirithromycin on sputum transportability (37). Mucus depletion was achieved by overnight incubation of bovine trachea at 37°C in PBS with penicillin (60 µg/ml) and streptomycin (100 µg/ml), with or without dirithromycin (1 µg/ml), followed by repeated passage of bovine mucus for 4 h in the same solution. The rates of movement (transportability) of each of 17 sputum samples (12 from patients with cystic fibrosis and 5 from patients with bronchiectasis) along the mucosa were measured on two 8-cm lengths of trachea taken from the same animal, one of which had been treated with dirithromycin. The transportability rate was calculated from a mean of at least four measurements taken at 15-s intervals. The transportability index was calculated by expressing the transportability rate of each sputum sample as a percentage of the transportability rate of bovine mucus.

Organ cultures. The organ culture method has been described previously (9, 14, 35). Briefly, human adenoid tissue was resected and transported to the laboratory in minimal essential medium (MEM; Gibco, Paisley, United Kingdom) containing antibiotics (50 µg of streptomycin per ml, 30 µg of penicillin per ml, and 50 µg of gentamicin per ml). Dissection was performed in antibiotic medium to yield small squares approximately 4 mm² in area and 2 to 3 mm in thickness. The tissue was screened for ciliary activity in order to select tissue squares with at least one fully ciliated edge. The tissue was immersed in antibiotic medium for at least 4 h in order to remove commensal bacteria and was then immersed for at least 1 h in medium alone in order to remove the antibiotics.

Effect of culturing of H. influenzae with sub-MICs of dirithromycin prior to infection of the respiratory mucosa. For each experiment (n = 6) four organ cultures were prepared: control tissue, tissue infected with H. influenzae, and tissue infected with H. influenzae which had been cultured overnight with dirithromycin (0.125 or 0.5 µg/ml). Two microliters of washed bacteria in PBS or PBS alone was gently pipetted onto the surface of the appropriate organ cultures, which were then incubated in a humidified atmosphere of 5% CO₂ at 37°C for 24 h. At the end of each experiment each of the four edges of the organ cultures was touched with a sterile loop and plated onto Levinthal agar, which had been produced by supplementing brain heart infusion agar (Oxoid) with lysed horse blood (Sigma, Poole, Dorset, United Kingdom) and nicotinamide, in order to assess the sterility of control organ cultures and the purity of H. influenzae growth in the infected organ cultures. The filter paper strip was then cut near the tissue with a sterile blade, removed with the tissue attached, and fixed for scanning electron microscopy as described previously (9, 14, 35).

Effect of incubating tissue with dirithromycin before infection with H. influenzae. For each experiment (n = 6) five organ cultures were constructed: control tissue, tissue infected with H. influenzae but not incubated with antibiotic, and tissue preincubated with dirithromycin (0.125, 0.5, or 1.0 µg/ml) prior to infection with H. influenzae. Appropriate tissue squares were preincubated with 4 ml of MEM containing dirithromycin (0.125, 0.5, or 1 µg/ml) for 4 h prior to assembly of the organ cultures. During this time other tissue squares were preincubated in MEM alone. Appropriate organ cultures were infected with 2 µl of washed bacteria in PBS or PBS alone. All organ cultures were incubated in a humidified atmosphere at 37°C in 5% CO₂ for 24 h. Tissue squares were tested for the sterility or purity of H. influenzae growth and fixed for scanning electron microscopy as described previously (9, 14, 35). The effect of the antibiotic that had penetrated into the adenoid tissue on bacterial growth was determined by mixing 1 ml of an overnight culture of H. influenzae with 10 ml of melted Levinthal agar before pouring the contents into the plates. Organ cultures incubated with dirithromycin (0.125, 0.5, or 1.0 µg/ml) for 4 h were placed in the center of a plate, which was then incubated overnight in 5% CO₂ at 37°C and examined for a zone of inhibition of H. influenzae growth.

Assessment of tissue by scanning electron microscopy. At the end of each experiment tissue squares were given a coded number by an independent observer so that the original identity of the samples was unknown during analysis. Each tissue square was examined with an Hitachi S-4000 scanning electron microscope (Katsuta-shi, Ibaraki-Ken, Japan) by the same observer. The tissue was initially viewed at a magnification of ×50. A transparent acetate sheet with 100 equal squares was placed over the screen of the visual display unit. Forty grid squares were chosen for further viewing by using a predetermined pattern. This pattern involved the horizontal, vertical, and both diagonal axes and gave a representative survey of the mucosal surface measuring 1.42 × 10⁴ µM². Care was taken to ensure that there was no overlap of squares in the center of the organ culture. Each square was assessed at a magnification of ×3,000, again using the acetate sheet, for the percentage of the surface area occupied by four mucosal features: mucus, ciliated cells, unciliated cells, and damaged epithelium. Extruding cells, cell debris, and loss of epithelium were scored together in the category of damaged epithelium. Unciliated areas were defined as areas not covered by cilia with or without microvilli. When a square contained more than one mucosal feature, the majority feature was scored. Summation of the scores allowed assessment of the percentage of each field that was occupied by each mucosal feature.

Statistics. All values are given as means ± standard errors. Comparison of the CBF and epithelial disruption by light microscopy, the mean percentage of nasal epithelial cells showing each of the parameters assessed by transmission electron microscopy, and the mean percentage of the surface area occupied by each of the four mucosal features in the organ culture model were analyzed by the Mann-Whitney test. The number and distribution of bacteria associated with each of the four mucosal features and the transportability of sputum on the bovine trachea were analyzed by the Wilcoxon signed rank paired test. P values of 0.05 were judged to be significant.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacteria. The mean inoculum sizes of H. influenzae in 2 µl of PBS used to infect the organ cultures were 4.8 × 10⁷ ± 0.6 × 10⁷, 4.3 × 10⁷ ± 0.4 × 10⁷, and 4.7 × 10⁷ ± 0.4 × 10⁷ when bacteria had been cultured in broth alone or in 0.125 or 0.5 µg of dirithromycin per ml, respectively, and 5.4 × 10⁷ ± 2.4 × 10⁷ for experiments in which tissue had been cultured with dirithromycin. There were no significant differences between the sizes of these inocula, and therefore, the results of the experiments could be compared. At 24 h all control organ cultures were sterile and H. influenzae-infected organ cultures gave a pure growth of this organism.

Effects of dirithromycin and amoxicillin on the damage caused by H. influenzae broth culture filtrate to human nasal epithelium. The addition of H. influenzae culture filtrate to human nasal epithelial cells in cell culture medium (1:1) caused a significant (P  0.008) reduction in the CBF (Fig. 1) and a significant (P  0.01) increase in epithelial disruption (Fig. 2) compared to the CBF and epithelial disruption for the control. The fall in CBF was immediate and continued for 4 h, whereas epithelial disruption gradually increased over 3 h. Incubation of the tissue with dirithromycin (1 µg/ml) significantly (P  0.03) reduced the slowing of the CBF and the epithelial disruption caused by the culture filtrate. The CBF slowing was still significantly different from that for the control at 4 h in the presence of dirithromycin. The reduction in epithelial disruption was more complete than the CBF slowing, and after incubation with dirithromycin, epithelial disruption of tissue exposed to the culture filtrate was not significantly different from that for the control. Broth alone (1:1 with medium 199) did not slow CBF over 4 h compared to the CBF with medium alone (the mean CBFs [n = 6] after 4 h were 13.5 ± 0.1 and 13.7 ± 0.2 Hz, respectively), nor did broth affect epithelial integrity because all strips were intact at 4 h.

View larger version (13K): [in this window] [in a new window]   FIG. 1.   Effect of dirithromycin on the reduction in the CBF of human nasal epithelial cells caused by H. influenzae broth culture filtrate. , nasal epithelial cells with medium 199 alone; , nasal epithelial cells incubated with dirithromycin (1.0 µg/ml) for 4 h prior to mixing (1:1) with H. influenzae broth culture filtrate; , nasal epithelial cells in medium 199 mixed (1:1) with H. influenzae broth culture filtrate; *, P < 0.008 versus medium 199 alone; $, P < 0.04 versus medium 199 alone; +, P < 0.03 versus nasal epithelial cells incubated with dirithromycin prior to incubation with culture filtrate.

View larger version (12K): [in this window] [in a new window]   FIG. 2.   Effect of dirithromycin on the disruption of the integrity of human nasal epithelium caused by H. influenzae broth culture filtrate. , nasal epithelial cells with medium 199 alone; , nasal epithelial cells incubated with dirithromycin (1.0 µg/ml) for 4 h prior to mixing (1:1) with H. influenzae broth culture filtrate; , nasal epithelial cells in medium 199 mixed (1:1) with H. influenzae broth culture filtrate; *, P < 0.01 versus medium 199 alone; +, P < 0.005 versus nasal epithelial cells incubated with dirithromycin prior to incubation with culture filtrate.

View larger version (18K): [in this window] [in a new window]   FIG. 3.   (A) Effect of dirithromycin on cell projection from human nasal epithelium caused by H. influenzae broth culture filtrate. Open bars, nasal epithelial cells with medium 199 alone; bars with diagonal lines, nasal epithelial cells in medium 199 mixed (1:1) with H. influenzae broth culture filtrate; bars with cross-hatching, nasal epithelial cells incubated with dirithromycin (1.0 µg/ml) for 4 h prior to mixing (1:1) with H. influenzae broth culture filtrate; 0, cell in line with epithelial surface; +, cell extruded from epithelial surface; ++, cell still in contact with epithelium but almost completely extruded from epithelial surface; *, P < 0.01 versus medium 199 alone; #, P < 0.03 versus medium 199 alone; +, P < 0.03 versus cells incubated with dirithromycin. (B) Effect of dirithromycin on cytoplasmic blebbing on human nasal epithelial cells caused by H. influenzae broth culture filtrate. The bars are as described above for panel A. *, P < 0.01 versus medium 199 alone; #, P < 0.03 versus medium 199 alone; +, P < 0.03 versus cells incubated with dirithromycin. (C) Effect of dirithromycin on mitochondrial damage in human nasal epithelial cells caused by H. influenzae broth culture filtrate. The bars are as described above for panel A. *, P < 0.01 versus cells with medium 199 alone; $, P < 0.01 versus cells incubated with dirithromycin.

Effect of dirithromycin on CBF and sputum transportability. Dirithromycin (0.125, 1.0, and 8.0 µg/ml) had no effect on the CBF over a 4-h period compared to the CBF for the control. The mean CBFs at 4 h were 16.7 ± 0.5 Hz (medium 199), 16.4 ± 0.5 Hz (0.125 µg/ml), 16.0 ± 0.5 Hz (1.0 µg/ml), and 16.3 ± 0.4 Hz (8.0 µg/ml). There were no significant differences in the CBFs at any time point. Incubation of bovine trachea with dirithromycin (1 µg/ml) had no effect on the transportability of sputum samples. In the absence of dirithromycin the mean transportability index was 35 (range, 11 to 59; standard error, 3.3), whereas it was 31 (range, 9 to 52; standard error, 3.4) with dirithromycin.

Effect of culturing of H. influenzae with sub-MICs of dirithromycin prior to infection of the respiratory mucosa. H. influenzae infection of organ cultures for 24 h caused a significant (P  0.01) increase in mucosal damage (70.6% ± 8.7% of the organ culture surface) and a significant (P  0.05) decrease in ciliated (1.7% ± 0.8%) and unciliated (23.9% ± 7.5%) cells compared to the results for the control (13.6% ± 5.9%, 26.3% ± 7.3%, and 56.3% ± 5.8%, respectively). Bacteria adhered to the damaged epithelium and to a lesser extent to unciliated cells and mucus. Epithelial tight junctions were frequently separated from each other, and H. influenzae was seen adhering in the gaps between cells. Culturing of H. influenzae with dirithromycin (0.125 and 0.5 µg/ml) prior to infection of human respiratory mucosa did not affect the mucosal damage caused by the infection. The percentages of the organ culture surface occupied by mucosal damage and ciliated and unciliated cells were 49.2% ± 8.3%, 9.6% ± 5.2%, and 37.1% ± 4.1%, respectively, for 0.125 µg of dirithromycin per ml and 63.1% ± 7.9%, 3.9% ± 1.8%, and 27.4% ± 7.4%, respectively, for 0.5 µg of dirithromycin per ml. Dirithromycin (0.125 and 0.5 µg/ml) had no effect on the total number or density of H. influenzae adhering to each mucosal feature (data not shown).

Effect of incubation of tissue with dirithromycin before infection with H. influenzae. H. influenzae infection of the respiratory mucosa caused a significant (P  0.03) increase in mucosal damage and a significant (P  0.03) decrease in the numbers of both ciliated and unciliated cells compared to the results for the control, as in the previous series of experiments (Table 1). Preincubation of the tissue with dirithromycin (all concentrations) prior to bacterial infection significantly (P  0.05) reduced the amount of epithelial damage caused by H. influenzae infection and the loss of both ciliated and unciliated cells. However, the protection afforded by dirithromycin was not complete since tissue incubated with 0.125 and 1.0 µg of dirithromycin per ml had significantly (P  0.03) elevated levels of damaged epithelium compared to that for the control, and tissue incubated with all three concentrations of dirithromycin had significantly (P  0.03) fewer ciliated cells compared to the numbers for the control.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Host-antibiotic interactions are important since they influence the outcome of treatment of an infection. A number of studies have shown that macrolide antibiotics have an anti-inflammatory effect which is independent of their antibiotic action or any influence on corticosteroid metabolism (5). Bronchial infections usually remain confined to the mucosa, and in patients with COPD many of them will resolve spontaneously. The bacteria that cause bronchial infections are usually upper respiratory commensal organisms or opportunistic pathogens such as Pseudomonas aeruginosa. When bacterial infection persists it usually reflects the severity of the impairment of the lung defenses rather than the virulence of the microorganism. Bacterial infection stimulates a host inflammatory response which, if it fails to clear the infection, becomes chronic and can cause tissue damage. Large numbers of activated neutrophils are attracted into the airway by host and bacterial chemotactic factors. Activated neutrophils do not differentiate between bacteria and bystander lung tissue, damaging the latter by spilling proteinase enzymes and reactive oxygen species during phagocytosis. This process has been termed a "vicious cycle" since tissue damage further impairs the lung defenses promoting continued infection (7, 24, 38). This hypothesis has led to the use of corticosteroids and nonsteroidal anti-inflammatory agents to augment antibiotic treatment of chronic bronchial infections (19, 28, 30).

The anti-inflammatory effects of macrolides have led to clinical trials of these drugs for the treatment of asthma (5). In chronic bronchial infections macrolides might combine their antibiotic action with an anti-inflammatory effect which could reduce host-mediated tissue damage. Diffuse panbronchiolitis is a disease with distinct features described in Japan. It is characterized by chronic inflammation of the respiratory bronchioles and the infiltration of chronic inflammatory cells. The disease progresses insidiously and results in respiratory failure due to chronic lower respiratory tract infection (12). Long-term, low-dose erythromycin therapy has been shown to be an effective treatment for this condition (12, 13, 20, 25, 41). The anti-inflammatory effects of macrolides include suppression of neutrophil activity (2, 16, 21, 36), inhibition of the production of proinflammatory mediators (18, 26, 32), and decreased lymphocyte proliferation (42). Two other properties of macrolides may be beneficial during bronchial infections. Erythromycin, but not other antibiotics, inhibited the secretion of mucus from human airways in vitro, and erythromycin and roxithromycin stimulated the ciliary motility of rabbit epithelium in vitro (11, 33). Macrolides have also been shown to reduce the level of toxin production by P. aeruginosa without inhibiting bacterial replication (31).

In the present study, we have shown that dirithromycin, a macrolide antibiotic, but not amoxicillin, a beta-lactam antibiotic, reduced the slowing of the CBF and the damage caused by a H. influenzae broth culture filtrate to nasal epithelium. The H. influenzae products which slow the CBF and which damage the epithelium have not been fully characterized, but the factors have a low molecular weight and are heat stable (17). H. influenzae lipopolysaccharide has also been shown to cause epithelial damage (8, 15). The mechanism by which dirithromycin protects the epithelial cells against the toxic effects of the culture filtrate is unknown. The nasal brushing technique yields strips of epithelial cells, and since the donors are healthy volunteers with no recent history of respiratory infection, very few inflammatory cells are present in the preparation. This makes it very unlikely that the beneficial effect of dirithromycin on epithelial cells occurred indirectly via an effect on inflammatory cells.

We have previously shown that an elevation of cyclic AMP levels in epithelial cells protects them against the damage caused by bacterial toxins (9). Takeyama et al. (33) showed that roxithromycin increased intracellular cyclic AMP levels in epithelial cells from rabbit trachea. They hypothesized that the elevation of cyclic AMP levels might occur because roxithromycin altered prostaglandin synthesis by epithelial cells. An alternative mechanism by which macrolides might protect cells from damage was suggested by Anderson et al. (2). They found that pretreatment of sheep erythrocytes with erythromycin, roxithromycin, clarithromycin, or azithromycin increased the resistance of these cells to the haemolytic effect of the bioactive lipids lysophosphatidylcholine, platelet-activating factor, and lyso-platelet-activating factor. The lipophilicity of macrolides suggested that the plasma membrane may be a site of action, and since Anderson et al. (2) could not detect any complex formation between the macrolides and the bioactive lipids, they suggested that the observed protective effects of the macrolides could be mediated via membrane stabilization. Dirithromycin may therefore stabilize the epithelial cell membrane and make it more resistant to the damage caused by bacterial toxins.

We did not find any effect of dirithromycin on the human CBF or on the transport of sputum by the bovine trachea. These results differ from those of Takeyama et al. (33), who showed an increase in the CBF of rabbit epithelial cells exposed to roxithromycin and, to a lesser extent, the CBF of those exposed to erythromycin. This difference could be explained by the species used as the source of cilia, since rabbit and human cilia have given different results in previous studies (29). Alternatively, it could be due to the macrolide antibiotic used in the present study, since Takeyama et al. (33) found that clarithromycin did not have an effect on the CBF of rabbit epithelial cells.

We have previously shown that H. influenzae infection of the respiratory mucosa causes patchy damage to the epithelium after 24 h (27, 34). Bacteria adhered in large numbers to mucus and damaged epithelium but adhered poorly to normal epithelial cells, particularly ciliated cells. In the present study H. influenzae infection produced similar changes. We investigated three concentrations of dirithromycin, all of which reduced the amount of damaged epithelium. A dirithromycin concentration of 1 µg/ml was equal to the MIC for the H. influenzae strain used, and it was therefore not surprising that tissue incubated with this concentration of antibiotic inhibited H. influenzae growth, which could have reduced the amount of damage to the epithelium caused by H. influenzae infection. Tissue incubated with the two lower concentrations of dirithromycin (0.125 and 0.5 µg/ml) did not inhibit H. influenzae growth when the tissue was placed on an agar plate in which H. influenzae was incorporated. However, we cannot exclude the possibility that dirithromycin leaking out from the tissue resulted in a higher concentration immediately adjacent to the tissue surface, and this might have influenced bacterial growth or toxin production in this area. The lowest concentration of dirithromycin (0.125 µg/ml) still protected the epithelium from damage caused by infection. We also found that incubation of H. influenzae with 0.5 µg of dirithromycin per ml prior to infection of the organ culture did not affect the subsequent mucosal damage. These results suggest that the protection afforded by dirithromycin is due to an effect on epithelial cells and not on bacteria.

Ciliated cells were still lost from infected organ cultures preincubated with dirithromycin. These cells are metabolically active and may be more sensitive to bacterial toxins (40). The total number of bacteria adhering to the organ culture was reduced by all three concentrations of dirithromycin. This was partly due to the reduction in the amount of epithelial damage, because the damaged epithelium was a preferred site of bacterial adherence, and was partly due to a reduction in the density of bacteria adhering to mucus (Table 2). Damaged epithelial cells release growth factors for H. influenzae (8), which may explain this latter result. The amount of mucus attached to the adenoid tissue in the cultures was small, perhaps because some was lost during processing for electron microscopy, but we did not find any evidence that dirithromycin reduced mucus production by the organ culture (11).

In summary, we have shown that concentrations of dirithromycin that are achieved in vivo (6) reduce the amount of epithelial damage caused by H. influenzae infection. This effect seems to be due to cytoprotection by dirithromycin. Further work is needed to determine the in vivo relevance of these results and to compare the effects of different macrolide antibiotics. The mechanism of cytoprotection should also be determined.