INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Respiratory tract infections are the leading cause of death from infectious disease, and an important source of morbidity and mortality. Nearly 1 million annual hospitalizations and 4 million outpatient health care encounters in the United States are the result of lower respiratory tract infections (1). Despite the staggering clinical and economic impact of pneumonia, there is limited basic information about its resolution. The elderly take longer to clear pneumonia and have greater disability and mortality from it than do younger individuals (1, 2). Although host factors and organism virulence are the major determinants of both radiographic clearing and clinical recovery from pneumonia, the cellular and molecular processes involved in normal resolution of the disease are poorly understood. For example, Mycoplasma pneumoniae pneumonia resolves quickly, whereas pneumococcal pneumonia may take months to resolve, leaving significant residual chest-radiographic abnormalities (2). Substantial cost to the health care system is incurred in the evaluation and treatment of pneumonia that is prolonged either clinically or radiologically. Therefore, enhanced understanding of the mechanisms leading to resolution of pneumonia is critical.

Bacterial infection generally leads to cell injury and cell death within the affected organ, and the nature and severity of this response may be an important determinant of outcome. The two modes of cell death that have been described, necrosis and apoptosis, are distinctly different. Necrosis occurs when the cell dies as a result of physical disruption of its plasma membrane. In contrast, apoptosis is a "cell suicide" that includes several morphologic and biochemical features, among which are cell shrinkage, nuclear condensation, protease activation, and internucleosomal fragmentation of DNA. During resolution of lung injury from the acute respiratory distress syndrome in humans, apoptosis may play a role in the selective removal of granulation tissue (3). Although cell death during acute lung injury (ALI) or pneumonia has classically been defined as necrotic, we have recently reported that apoptosis of lung cells is a prominent component of ALI, and also occurs within hours of initial infection in an experimental model of pneumonia (4). However, no one has studied apoptosis during the resolution of pneumonia.

In an attempt to understand the role of apoptosis in the outcome of pneumonia, we studied two models of bacterial pneumonia in experimental animals (5, 6). In both models, infection causes acute pneumonitis in rats that receive an intratracheal bolus of bacteria. However, infection with Streptococcus sanguis resolves over a period of 3 wk, whereas infection with Streptococcus pneumoniae type 25 persists and causes the lungs to become fibrotic. In the present study, we examined these models of pneumonia, using the terminal deoxynucleotidyl transferase-uridine nucleotide-end labeling (TUNEL) technique in association with computer-aided image analysis. In addition to in situ detection of DNA fragmentation, we used expression of the cyclin-dependent kinase-5 (Cdk5) gene as a marker for apoptotic cells. Members of the family of cyclin dependent kinases play an essential role in progression of the cell cycle, and some have also been implicated in cell death. Cdk5, originally identified by its homology to Cdk2 (7), is highly upregulated during cell death (8, 9). Cdk5 does not appear to have a role in the cell cycle, and was originally found mainly in association with neuronal differentiation (10). However, its highest level of expression is seen in cell death. This increase in the expression of Cdk5 is posttranscriptional, resulting in increases at the protein level (but not the messenger RNA [mRNA] level) and an upregulation of the kinase activity of Cdk5 (8, 9). Zakeri and coworkers have shown that upregulation of Cdk5 occurs in terms of both protein level and activity during apoptosis in a variety of models, including the developing mouse embryo (8, 9, 11), the postcastration hormonoprivic state in prostatic epithelia (9), and cell death induced either by retinoic acid (8) or cyclophosphamide (L. Lin and Z. Zakeri, unpublished observations). Others have reported an increase of Cdk5 after induction of apoptosis in rat substantia nigra (12). In addition, regulation of the Cdk5 protein level and kinase activity associated with cell death has been found to play roles in many neurodegenerative diseases, such as Alzheimer's (13) and Parkinson's diseases (12, 14). In this article we report that Cdk5 expression is also upregulated during apoptosis induced by bacterial infection.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pneumonia Models

S. sanguis or S. pneumoniae type 25 (~ 5 × 10⁶ cfu) were introduced in an equal volume of bacteriologic mucin (Difco Laboratories, Detroit, MI) into the left main bronchus of female, pathogen-free Wistar rats via a tracheostomy, as described previously (5). Control rats were injected in an identical manner with 6% mucin in sterile culture medium. Animals were killed by exsanguination under anesthesia in groups of three to five at various times over a period of 8 d after inoculation of either S. sanguis or S. pneumoniae type 25. The lungs were then fixed by perfusion with modified Karnovsky's fixative (15). The perfused lung was further fixed by immersion for 4 h in modified Karnovsky's fixative, followed by immersion in 10% buffered formalin for 48 h, and was then embedded in paraffin.

Infection with either S. sanguis or S. pneumoniae type 25 results in acute pneumonitis within 24 h of infection. However, although the S. sanguis infection resolves over the course of 3 wk (5), the S. pneumoniae infection does not, and the lungs become fibrotic at 1 wk (6). We will refer to these models as resolving and nonresolving pneumonia, respectively. In the resolving model, the animals developed a typical confluent bronchopneumonia (6), which involves 90% or more of the infected lung, within 24 h. The bacteria are virtually eliminated by 2 d. The animals thereafter regain appetite and body weight, and the pneumonia resolves by 8 d. Except for an abscess that typically forms at the extreme base, the lungs appear morphologically normal. This abscess heals, by 21 d, leaving a small scar.

In the nonresolving model, infection is initially identical. However, these animals continue to remain sick, and typically lose 10 to 15% of their body weight by 4 d. The bacteria persist and proliferate, and by 8 d the lungs show evidence of fibrosis, with the left lung remaining firm and shrunken (6). The salient features of these models are illustrated in Figure 1.

View larger version (40K): [in this window] [in a new window]   Figure 1.   Features of the resolving and nonresolving models of pneumonia. Flow chart illustrating the pathogenesis of disease in rats infected with either S. sanguis or S. pneumoniae type 25 over the course of 8 d.

Identification of Apoptotic Cells

Tissue sections (4 to 5 µm) were mounted on slides pretreated with 3-aminopropylethoxysilane (Digene Diagnostics, Inc., Beltsville, MD). The slides were baked for 30 min at 60° C and then washed twice in fresh xylene for 5 min each to remove paraffin. The slides were rehydrated through a series of graded alcohols, and each slide was washed in distilled water for 3 min. The slides were then either stained with hematoxylin and eosin (H&E), assayed with a modified TUNEL method, or double-labeled (immunohistochemistry for Cdk5 in combination with the TUNEL assay). Cell types undergoing apoptosis were determined on the basis of location and morphology.

The TUNEL assay was performed on the tissue sections as described (16), with the following modifications: digoxigenin-deoxyridine triphosphate (Dig-dUTP) (25 mM; Boehringer Mannheim, Indianapolis, IN) was used in the end-labeling mixture. Labeled DNA was detected with an alkaline phosphatase-conjugated anti-DIG antibody (Boeringher Mannheim) in 1% bovine serum albumin-Tris-buffered saline (TBS) (100 mM Tris, pH 8.0; 150 mM NaCl) for 1 h at 37° C, followed by washing three times in TBS. This was followed by a 15-min wash in alkaline phosphatase buffer (100 mM Tris-HCl, pH 9.5; 100 mM NaCl; 50 mM MgCl₂). The color reaction was run for approximately 30 min in the dark, after which the slides were washed in water and the sections were mounted with Aqua-Polymount (Polysciences, Warrington, PA). Alternatively, sections were incubated with 20 µg/ ml rhodamine-conjugated anti-DIG antibody, washed as previously described, counterstained with 2 µg/ml 4',6-diamidine-2-phenylindole-dihydrochloride (DAPI) to stain all nuclei, and mounted with Immu-mount (Shandon-Lipshaw, Pittsburgh, PA). Sections were viewed and micrographed with an Optiphot microscope (Nikon, Melville, NY) equipped with epifluorescence optics and the UFX-II camera system (Nikon Inc.). Rhodamine fluorescence was viewed with the Nikon G-2A filter set (excitation at 510 to 560 nm, barrier at 590 nm); DAPI fluorescence was viewed with the Nikon UV-2A filter set (excitation at 330 to 380 nm, barrier at 420 nm).

For double-labeling, deparaffinized sections were prepared as described previously (8). These sections were visualized on a Meridian Ultima confocal microscope (Meridian Instruments, Inc., Okemos, MI) equipped with the appropriate filters. The slides used for detection of only one parameter were used as controls both for evaluating background and for calculating the percent overlap from one channel to another.

Image Capture and Analysis

One tissue section from each of two different rat lungs was analyzed for each time point in the following manner: Three "optical cross-sections," starting from the base and moving to the apex of the lung, were captured with a ×20 objective on the Nikon Optiphot microscope equipped with a DAGE72 CCD camera (DAGE-MTI, Inc., Michigan City, IN) in conjunction with the MetaMorph software package (Universal Imaging, West Chester, PA). Under these conditions, each field corresponds to an area of 0.2 mm², and at least 120 fields from each animal, were captured for analysis.

Apoptotic nuclei were scored objectively with the MetaMorph software. Fields from each animal were captured with identical camera and software settings. Threshold values were determined by first measuring the average gray-scale value of four fields in a control section. The threshold values were then set to be at least 10 units greater than the average background. In addition, a size filter range of  10 to  300 (pixels) was used during the analysis, in order to avoid spurious signal from staining anomalies. To quantify the extent of apoptosis, the number of TUNEL-positive nuclei per field was determined. At least 3,900 cells from each animal were examined.

Statistical Analysis

Analysis of variance was done with the SAS software package (SAS Institute, Cary, NC) on an IBM-compatible computer. Descriptive statistics and Student's t tests were performed on the data from all fields, to analyze differences between the different models at single time points, using Microsoft Excel (version 4.0) software (Microsoft, Seattle, WA) on a PowerPC Macintosh computer (Apple Computer, Cupertino, CA).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Apoptosis in Control Lungs and During Acute Pneumonitis

We have recently reported that apoptosis is a component of the earlier phases of infection by S. sanguis (4), which led us to hypothesize that apoptosis might be a distinguishing feature of resolving as compared with nonresolving pneumonia in these models. To test this idea, we utilized the TUNEL technique. The TUNEL technique is an in situ technique that labels free, 3'-OH ends of DNA generated by endonucleases that are activated during apoptosis (16). For this study, Dig-dUTP was used in the labeling reaction, and labeled DNA was detected either colorimetrically or fluorometrically (see METHODS).

To determine the extent of apoptosis resulting from the injection procedure, we injected control animals with vehicle alone. Very few apoptotic nuclei were detected in animals injected with vehicle alone (data not shown).

Figures 2A and 2B show that at 8 h after infection, during the acute stage of infection, the lungs were filled with fluid in both the resolving and nonresolving models. Regions of apoptotic activity involving predominantly bronchiolar and alveolar epithelia were evident at this time point (Figures 2C and 2D).

View larger version (86K): [in this window] [in a new window]   Figure 2.   Apoptosis of epithelial cells at 8 h after infection, during the pneumonitic phase of infection. (A) H&E-stained lungs infected with S. sanguis. (B) TUNEL assay of a section adjacent to the one shown in (A). Arrows point to TUNEL-positive bronchiolar epithelial cells. (C ). H&E-stained lungs infected with S. pneumoniae type 25. (D) TUNEL assay of a section adjacent to the one shown in (C ). Arrows point to TUNEL-positive bronchiolar epithelial cells; arrowhead points to TUNEL-positive alveolar epithelial cell. Scale bar = 10 µm.

Progression of Apoptosis in Pneumonia

A distinguishing feature of the resolving model of pneumonia used in our study is the occurrence of an abscess at the base of the lung. Organizing but not yet formed abscesses were apparent at 2 d. Figure 3A shows a portion of an organizing abscess as well as an injured region of parenchyma. In areas outside the abscess, apoptotic nuclei were found in alveolar epithelium (Figure 3B). Higher-magnification views showed that the apoptotic nuclei within the abscess consisted predominately of neutrophils (Figure 3D). There were no abscesses in lungs with nonresolving injury (6), yet apoptosis was evident in these lungs, albeit in a patchy pattern of apoptotic nuclei throughout the lung. The apoptotic cells detected were desquamated alveolar epithelial cells, alveolar macrophages, and neutrophils (data not shown).

View larger version (111K): [in this window] [in a new window]   Figure 3.   Patterns of apoptosis at 2 d after infection. (A ) Low-power magnification montage of two adjacent fields from H&E-stained lungs infected with S. sanguis at 2 d after infection. Boxed region in (A) indicates the area in an organizing abscess (i.e., microabscess) magnified in (C ). The inset in (C ) is a high-power magnification view of the boxed region, and shows the histology of the area corresponding to that in (D). Arrow points to an apoptotic polymorphonuclear neutrophil (PMN) (see [D]). (B and D) TUNEL assay of a section adjacent to the ones shown in (A and C ). Boxed region indicates the area magnified in (D). The inset in (D) is a high-power magnification view of the boxed region, with the arrow pointing to a TUNEL-positive PMN. abs = abscess (microabscess); aw = airway. Scale bar = 150 µm.

By 4 d after infection the resolving model had microabscesses that were more organized than seen earlier, at 2 d (compare Figure 3A with Figure 4A). Very little consolidation or edema was evident in most of the lung. Few if any TUNEL-positive cells were seen in the portions of the lung in which disease had clearly resolved (Figure 4B). However, apoptotic cells were found in the areas adjacent to abscesses, particularly in the alveolar epithelia (Figure 4F). In addition, numerous TUNEL-positive neutrophils and macrophages were found in the abscesses themselves (Figures 4B and 4F). At this time point in the nonresolving model, alveolar septa were thickened, and alveolar spaces were consolidated with polymorphonuclear leukocytes and macrophages (Figures 4C and 4G).

View larger version (142K): [in this window] [in a new window]   Figure 4.   Apoptosis at 4 d after infection. (A and E ) Low-power magnification montages of two adjacent fields from H&E-stained lungs infected with S. sanguis at 4 d after infection (A). Boxed region indicates the area magnified in an organizing abscess and damaged alveoli magnified in (E ). The inset in (E ) highlights a region of damaged alveoli corresponding to the region seen in (F ). (B and F ) TUNEL assay of a section adjacent to the ones shown in (A and E ). Boxed region indicates the area magnified in (F ). The inset in (F ) highlights a region of apoptotic alveolar epithelial cells. (C and G) Low-power magnification montage of two fields of H&E-stained lungs infected with S. pneumoniae type 25. Boxed region indicates the area magnified in (G). The inset in (G) highlights a region corresponding to the region in (H ) (D and H ) TUNEL assays of sections adjacent to the ones shown in (C ) and (G). Arrow in (D) points to an airway where many bronchiolar epithelial cells are TUNEL-positive. Boxed region indicates the area magnified in (H ). The inset in this panel shows apoptotic bronchial epithelial cells (arrows) as well as apoptotic nuclei in the alveoli. abs = abscess; aw = airway; rd = resolved region; da = damaged alveoli. Scale bar = 150 µm.

At 8 d after infection, the resolving lung was largely repaired, with the exception of an abscess that persisted at the base. Histologically, most of the lung appeared normal (5) (Figure 5A). A TUNEL assay of tissue adjacent to the abscess showed very few apoptotic cells (Figure 5B) outside the abscess. However, as in the case of the 4-d time point, apoptosis was still clearly evident in the abscess (Figures 6C and 6D), as well as in the surrounding granulomatous tissue (Figures 5B and 5F). At this time point in the nonresolving model, widespread apoptosis persisted (Figure 5D). Interestingly, apoptosis was less evident in the bronchial epithelia, suggesting that turnover of these cells is complete at this time point (compare Figure 4D and Figure 5D).

View larger version (175K): [in this window] [in a new window]   Figure 5.   Apoptosis at 8 d after infection. (A) Low-power magnification montages of two adjacent fields from H&E-stained lungs infected with S. sanguis at 8 d after infection (A). Boxed region indicates the area magnified in the abscess and the abscess wall in (E  ). The inset of (E  ) highlights an area of the abscess corresponding to the region seen in (F  ). The white arrow in (F  ) indicates an apoptotic fibroblast. (B and F ) TUNEL assays of sections adjacent to the ones shown in (A and E  ). Boxed region indicates the area magnified in (F  ). The inset in (F  ) highlights a region in the abscess with apoptotic polymorphonuclear neutrophils (PMNs) (black arrow) and apoptotic fibroblasts (white arrow). (C ). Low-power magnification montage of two fields of H&E-stained lungs infected with S. pneumoniae type 25. Boxed region indicates the area of fibrotic tissue magnified in (G). The inset in (G) highlights a region corresponding to the region seen in (H ). The white arrow points to an apoptotic fibroblast shown in (H ). (D and H ) TUNEL assay of a section adjacent to the one shown in (C and G). The boxed region indicates the area magnified in (H ). The inset in this panel shows apoptosis in fibrotic tissue, with the arrow indicating an apoptotic fibroblast. abs = abscess; aw = airway; rd = resolved region; gr = granulomatous tissue. Scale bar = 150 µm.

View larger version (185K): [in this window] [in a new window]   Figure 6.   Apoptosis with abscess at 8 d after infection. Low-power magnification photomicrographs of abscess caused by S. sanguis stained with H&E (A), and an adjacent TUNEL-assayed section (B). (C and D) Higher-power magnifications of (A) and (B), respectively. Black arrow shows an apoptotic PMN. Abbreviations: abs = abscess; rd = resolved region. Scale bar = 150 µm.

Computer-aided Image Analysis

We used computer-aided image analysis to determine whether apoptosis differed quantitatively in the two models used in our study. Table 1 is a summary of the results of this investigation. This table shows that the amount of apoptosis (apoptotic nuclei/field) was very similar during the acute phase of infection (8 h) in both models. The amount of apoptosis increased and peaked over the first 4 d in both models, and there was more apoptotic activity in the nonresolving model at all time points except at 8 h.

Although Table 1 is an overview of the apoptotic activity in the two models we examined, it does not show differences in the pattern of apoptotic activity indicated by the photomicrographs. To illustrate these differences, we plotted the number of apoptotic nuclei per field from a representative set of consecutive fields at each time point, starting at the base and moving to the apex of the lung (Figure 7). A different pattern emerged as early as 2 d after infection, when the response appeared to be biphasic in the resolving model but more homogeneous in the nonresolving model (Figure 7B). The high apoptotic activity in the base of the lung showing resolution was due to the forming abscess, whereas the apical areas contained regions of damaged alveoli. The apoptotic activity was limited to the abscess to an even greater extent at 4 d and 8 d (Figures 7C and 7D, respectively). In the nonresolving model, apoptosis persisted and became more widespread by 8 d.

View larger version (26K): [in this window] [in a new window]   Figure 7.   Graphic representation of a quantitative analysis of apoptosis in experimental pneumonia. Scattergrams are derived from nonoverlapping adjacent fields, as described in METHODS, from one animal infected with either S. sanguis (closed squares) or S. pneumoniae type 25 (open circles) at: (A) 8 h after infection, (B) 2 d after infection, (C ) 4 d after infection, (D) 8 d after infection. Open squares in (C ) and (D) represent the abscess region in S. sanguis-infected animals. Each data point represents an area of 0.2 mm2. Similar results were obtained when the extent of apoptosis was calculated as the percent of TUNEL-positive nuclei per total number of DAPI-stained nuclei (data not shown).

Cdk5 and TUNEL Colocalization

Reports in the literature have suggested that the TUNEL assay detects both necrotic and apoptotic nuclei (17). Since Cdk5 expression has been shown to be an early marker of apoptosis in other systems (8, 9, 11), we performed a Cdk5-TUNEL double-labeling assay followed by confocal microscopy. We used sections taken 4 d after infection, since our image-analysis data indicated that this time point had the highest amount of apoptotic activity. Comparison of the TUNEL assay (Figure 8A) and Cdk5 staining results (Figure 8B) suggested that the TUNEL-positive cells were a subset of the cells that were Cdk5 positive. To ascertain which cells were double-positive (TUNEL and Cdk5) as opposed to single positive (for either Cdk5 or TUNEL), we superimposed the respective images (Figures 8C and 8D). The view at a lower magnification shows that most of the cells that were TUNEL-positive were also Cdk5 positive, as indicated by the yellow color. Of the singly-positive cells, there appeared to be qualitatively more cells that were Cdk5-positive only (Figure 8C and arrowhead in Figure 8D). This is not surprising, since Cdk5 immunohistochemistry identifies cells earlier in the apoptotic program than does the TUNEL technique. However, it does suggest that we may have been underestimating the amount of apoptosis when using the TUNEL assay. The occasional cell that was TUNEL-positive and Cdk5-negative appeared morphologically to be apoptotic (see arrows in Figure 8D), suggesting that these were cells in the later stages of apoptosis.

View larger version (25K): [in this window] [in a new window]   Figure 8.   Colocalization of Cdk5 expression and TUNEL-positivity at 4 d after infection. Pseudocolor confocal micrographs of a field from lungs infected with S. pneumoniae type 25. (A) TUNEL assay with fluorescein isothiocyanate-conjugated antibody. (B) Cdk5 expression detected with a Cy3-conjugated antibody. (C) Composite showing colocalization (yellow) of Cdk5 and TUNEL positivity. (D) High-power magnification of region indicated in (C). Arrows indicate TUNEL-positive/Cdk5-negative nuclei; Arrowheads indicate TUNEL-negative/Cdk5-positive cells. Scale bar = 50 µm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study and work in other laboratories have shown that apoptosis is a prominent component of resolution and fibrosis in the lung. In the present study we demonstrated apoptosis during the acute stage of pneumonitis caused either by S. sanguis or S. pneumoniae type 25. Our results in the resolving model suggest that controlled or localized apoptosis is a prominent feature in the resolution of pneumonia. In particular, control of apoptosis in the resolving model is strongly implicated at 8 d after infection, when apoptotic activity is localized to the abscess. In contrast, widespread and persistent apoptosis is associated with nonresolving pneumonia, which is known to progress to fibrosis.

Previous studies have found that the degree of apoptosis in the lung correlates with the severity of the injury incurred by hyperoxia (4, 18). In contrast, quanitative analysis in our study revealed no statistical differences in the degree of apoptosis in the early stages of either of the two models studied. Furthermore, if one included the abscess in the resolving model in the statistical analysis at 8 d, there would be no statistical differences at this stage either. This is clearly a weakness of using the type of quantitative analysis that we used when comparing samples that involve induced apoptosis in different types of cells. This limitation can be solved by double labeling with cell-specific markers and an apoptotic marker. The methodology for this has been reported and, when it is combined with computer-aided image analysis, this technique will provide a powerful tool for delineating apoptosis in the models used in our study and potentially in any system that involves multiple cell types. For example, identifying the cell types involved at the later stages of the nonresolving model would help us determine whether the apoptotic cells are fibroblasts or white cells dying as a result of ongoing inflammation. We are currently refining the methods necessary to address these issues in the lung.

Some of the cyclin-dependent kinases, such as Cdc2 (Cdk1) and Cdk2, which are activated in human immunodeficiency virus type 1 transactivator protein-expressing cells that die (19), may play a role in apoptosis. Cdk5, originally identified by its structural homology to human Cdc2, is highly conserved among vertebrates (20, 21). Although it is a member of the Cdk class of catalytic subunits, Cdk5 is expressed in the embryonic nervous system, in cells that are not proliferating but rather are differentiating (10, 22, 23). Zakeri and coworkers have shown that Cdk5 is uniquely expressed in the dying cells of both normal and abnormal tissues, and that this gene is one of the most abundant genes expressed in dying cells (8, 9). Furthermore, Zakeri and coworkers have demonstrated a one-to-one correlation between cell death and the expression of Cdk5. We demonstrated Cdk5-upregulation during bacterial pneumonia and the colocalization of Cdk5 and TUNEL-positive DNA. In light of the controversy surrounding the specificity of the TUNEL technique, these data indicate that at a minimum, the expression of Cdk5 may be a more sensitive and effective marker of the degree of cell death in situ than is the TUNEL technique. Based on our observations, the therapeutic or preventive potential of regulating Cdk5 expression will depend on the ability to target specific cell types.

Resolution of bacterial pneumonia depends upon multiple factors, including host defenses, organism virulence, and size of the inoculum (5, 6, 24). Apoptosis appears to have a role in limiting the inflammatory response (25). Our observations indicate that in addition to being induced or inhibited by pathogens, apoptosis also has a key function in the resolution of bacterial infection. Numerous viruses and bacteria synthesize products to regulate apoptosis of the infected cell. E1B-19K of adenovirus and p35 of baculovirus are two commonly studied viral genes that inhibit apoptosis in their respective host cells (26, 27). The respective proteins effectively suppress or delay apoptosis long enough for the production of sufficient quantities of progeny. Shigella flexneri, an invasive bacterial species, induces apoptosis of infected macrophages by activating caspase-1/interleukin-1 beta-converting enzyme (28). Invasion plasmid antigen B (IpaB), the bacterial gene that is sufficient for this induction, colocalizes with caspase-1, suggesting a direct activation of caspase-1 (29). Previous studies have shown that only virulent strains of S. pneumoniae anchor the G-protein- coupled platelet-activating factor (PAF) receptor in a human pulmonary epithelial cell line (A549) stimulated with various cytokines (30), and that S. pneumoniae induces apoptosis in A549 cells, with the appearance of internucleosomal cleavage as early as 6 h after infection (31). Whether S. pneumoniae precipitates apoptosis by binding to the PAF receptor remains to be determined. However, since streptococcal pathogenicity is thought to depend on bacterial adherence to extracellular matrix rather than to cells (32), apoptosis may provide a means for the bacteria to expose more binding sites by killing epithelial cells.

Several studies have demonstrated a component of apoptosis during the resolution of type II cell hyperplasia following lung injury (3, 4, 18, 33). Collectively, these studies support the notion that apoptosis is not always beneficial. For example, Bitterman and coworkers have shown that apoptosis is an important component in the resolution of acute respiratory distress syndrome (3). Using bronchoalveolar lavage fluid (BALF) from patients with acute respiratory distress syndrome, they found that the apoptosis-inducing activity was higher in the BALF of patients with resolving ALI. Their results suggest that apoptosis is beneficial for resolution by limiting the fibroproliferative response. Our data in the resolving model are consistent with this hypothesis. However, we also found that widespread (not localized) apoptosis occurs during fibrosis. Results of a study by Uhal and colleagues suggest that fibroblasts, repopulating the lung during fibrotic lung injury, may release apoptotic factor(s) (36). Uhal and colleagues isolated primary fibroblasts from patients with human idiopathic pulmonary fibrosis, and from rats treated with 75% O₂ and paraquat, and from their normal human and rat counterparts (36). Media conditioned by fibroblasts from fibrotic patients or rats induced more apoptosis in alveolar epithelial cells than did media conditioned by fibroblasts from their normal counterparts. With respect to the fibrosing pneumonia model, Rhodes and coworkers proposed that the inability of type 2 pneumocytes to differentiate to type 1 cells and repopulate the epithelium ultimately causes the proliferation of underlying interstitial fibroblasts. On the basis of the location and morphology of the apoptotic cells in the later stages of the nonresolving model, we feel that the majority of these cells are fibroblasts involved in the fibroproliferative response. Whether the apoptosis in nonresolving pneumonia is triggered by the repopulating fibroblasts remains to be determined.

An apparent paradox that arises from our observations is that fibrosis is a proliferative disease yet that cell death is a feature of this model. There have been several reports of increased apoptosis concurrent with an increase in cell division. For example, apoptosis can increase during tumorigenesis (37, 38). Our data indicate that during fibrosis, the relative rate of cell death may be slower than the rate of cell division, resulting in a steady state of fibroproliferation. Perhaps resolution depends on achieving the proper ratio of cell death (to remove damaged tissue) and cell division (to replace the damaged tissue). If this is a maladaptive response to virulent strains of bacteria, then several possible candidate proteins could control apoptosis, including caspases, cytokines, growth factors, and perhaps Cdk5. Identifying the cell type(s) involved in controlling these processes might be the key to determining whether a diseased lung improves or fibroses. Furthermore, delineating the pathway(s) involved may lead to novel therapeutic strategies to promote lung healing.