INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The immunodeficiency produced by human immunodeficiency virus (HIV) infection predisposes the lung to both community-acquired and opportunistic infectious agents. The incidence of cigarette smoking among HIV-infected individuals is notably higher than in the general population (1). It has been shown that the incidence of bronchitis (4), bacterial pneumonia (3, 5, 6), and lung burden of HIV (7) is increased in HIV-seropositive smokers. These findings suggest that immune defects exist in HIV-infected smokers. Yet HIV-seronegative smokers have significantly higher percentages of CD4⁺ blood lymphocytes than do nonsmokers (1, 5, 8). Although Park and colleagues (8), and more recently Conley and coworkers (5), have shown that upon seroconversion, HIV-infected smokers experience a marked decrease in CD4⁺ lymphocyte counts, the decline only approaches the profiles of nonsmokers at 2 yr after seroconversion. Thus, as assessed through blood CD4⁺ cell counts, smoking does not appear to induce an immunocompromised state.

In this regard, systemic immune responses do not always reflect specific organ immunity. The compartmentalized increase in CD4⁺ lymphocytes and increased CD4⁺/CD8⁺ ratio seen in the lung, but not in the blood, of patients with sarcoidosis is a case in point (9). Because blood CD4⁺ lymphocyte counts reflect only systemic immune function, it may be necessary to examine organ immunologic indices separately. In the lungs of smokers, macrophages are immunosuppressive for cell-mediated immune responses. On a per-cell basis, smoker's macrophages have a more inhibitory effect on lymphocyte proliferation than do nonsmoker's macrophages (10), and macrophage numbers are greatly increased with smoking (11).

From another perspective, smoker's macrophages may also be impaired in the spontaneous release of cytokines (12, 13). Twigg and coworkers have shown that smoking decreases lung macrophage function in HIV-infected individuals (12). Their findings indicate that smoking decreases alveolar macrophage (AM) accessory cell function and secretion of the proinflammatory cytokines interleukin-1 (IL-1) and IL-6. This finding may be relevant to the understanding of certain HIV-associated pulmonary complications, since critical proinflammatory cytokines (tumor necrosis factor- [TNF-] and IL-1) are important to lung defense against bacteria. For example, mice depleted of functional TNF or IL-1 have increased susceptibility to bacterial pneumonia (14). Furthermore, IL-1 administration protects granulocytopenic mice from lethal gram-negative pneumonia (17). These findings suggest that smoking may alter immunologic mechanisms and suppress host defense in the alveolar environment, and thereby contribute to bacterial pneumonia.

In the current investigation, we examined the concept that smoking may induce a compartmentalized immunosuppressive environment in the lungs of HIV-infected individuals. We extended the original observations of Twigg and colleagues (12) by delineating the effects of smoking on various lung lymphocyte populations and by studying the secretion of both IL-1 and TNF-.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Bronchoalveolar Lavage

Bronchoscopy with bronchoalveolar lavage (BAL) was performed as we have previously described (18). Briefly, after informed consent, subjects underwent standard bronchoscopy with BAL consisting of sequentially instilling and aspirating sterile saline in five 20-ml aliquots into the right middle or lingular bronchus from the wedged position. Recovered lavage fluid was passed through a single layer of sterile surgical gauze to remove mucus and particulate matter. Fluid aliquots were immediately taken for cell counting, lymphocyte phenotyping, harvesting of messenger RNA (mRNA), and culture (see the following discussion).

Cell Processing

Total cell counts and differential counts in the recovered fluid were made by direct cell hemocytometry and Diff-Quik stainings of cytospin preparations as we have previously described (19). This is the standard technique that has been adopted by the American Thoracic Society (20) and accredited in clinical pathology for the hospital laboratory analysis of BAL cells. In addition to cell counting and cell differential determinations, a sample of BAL fluid (BALF) containing approximately 1 million cells was sent to the Ohio State University Hospital cellular immunology laboratory for T-lymphocyte counts and subtyping (see the following discussion). Another 1 million to 5 million cells were lysed in Trizol reagent (Gibco-BRL, Grand Isle, NY) for phenol/chloroform RNA extraction. The remainder of cells in BALF specimens was processed by washing thrice with Hanks' balanced salt solution (HBSS) and culturing in RPMI 1640 medium, 5% fetal bovine serum (FBS), and gentamicin.

Quantification of BAL Cells through Fluorescence-activated Cell Sorting

Fresh BAL cells were analyzed with a fluorescence activated cell sorter (FACS; Coulter, Inc., Hialeah, FL) in the Ohio State University Hospital cellular immunology laboratory. Cells were analyzed through dual staining procedures (with directly fluorosceinated antibodies) for the relative frequencies of the following phenotypes: CD3⁺/ CD4⁺ (T lymphocytes of the helper phenotype), CD3⁺/CD8⁺ (T lymphocytes of the suppressor/cytotoxic phenotype), and S6F1⁺/CD8⁺ (activated CD8⁺ lymphocytes of the cytotoxic phenotype) (21, 22).

Cell Culture for Cytokine Production

Lung lavage cells were cultured at 1 × 10⁶ macrophages/ml in RPMI 1640 with 5% FBS and gentamicin either alone or with 100 ng/ml of lipopoly saccharide (LPS) (Escherichia coli; Difco, Detroit, MI) for 18 h. In select experiments, the macrophages were purified by 1-h adherence to tissue culture plastic plate wells prior to the addition of LPS. After 18 h, supernatants were harvested, the nonadherent cells removed by centrifugation and the supernatant frozen at 80° C until assay.

Enzyme-linked Immunoassay for Cytokines

The enzyme-linked immunoassay (ELISA) for IL-8 has been described in detail (23, 24). Briefly, a monoclonal antibody to mature IL-1 is used as the capture antibody (B1), and a rabbit polyclonal antibody to mature IL-1 (Rc) is used to detect captured antigen. This assay is highly specific and is sensitive to IL-1 at a concentration of 50 pg/ml. The TNF- ELISA utilizes a mouse monoclonal antibody (Mo199; Boehringer Mannheim) as the capture antibody and a rabbit polyclonal antibody to TNF- (developed in our laboratory) as the antibody for detection. This assay is sensitive to TNF- at 100 pg/ml, and is highly specific for TNF- (25).

Reverse Transcription-Polymerase Chain Reaction for Quantitation of IL-1 and TNF- mRNA

Measurement of constitutive mRNA for the proinflammatory cytokines was done with RNA harvested from fresh BAL cells. RNA isolation was done with Trizol reagent, as outlined earlier. This RNA was quantified by measurement of optical density at 260 nm before first strand DNA synthesis. The reverse transcription (RT) step of the RT-polymerase chain reaction used 3 µg of total RNA, random hexamer primers (Gibco-BRL), ribonuclease inhibitor (RNasin) (Promega, Madison, WI), 2.5 mM MgCl₂, and Super Script II reverse transcriptase enzyme (Gibco-BRL), which generates sufficient DNA for approximately 30 PCRs. The RT-PCR was done with a DNA Thermal Cycler 480 (Perkin Elmer, Foster City, CA).

Quantitative PCR

Quantitative PCR was done with the MIMIC system (Clontech, Palo Alto, CA), which had been previously standardized for IL-1 and  actin. In this procedure, the concentration of MIMIC reagent (typically in attomoles) that produces a signal equal to that of the test cDNA determines the amount of specific mRNA. In brief, increasing concentrations of the complementary DNA (cDNA) analogue (MIMIC) of the mRNA for the cytokine of interest were added to constant amounts of the sample. PCR with first strand cDNA was then done for 35 cycles, with melting at 94° C for 45 s, annealing at 60° C for 45 s, and primer extension at 72° C for 2 min, using buffer conditions outlined by the manufacturer. The resultant amplified cDNA from the MIMIC process and first stand cDNA test sample was quantified by ethidium bromide staining of 1% agarose gel.

Definition of Smoking Status

Subjects self-reported their smoking status category as current smoker, ex-smoker, or never smoker. Ex-smokers were also asked to identify the time that had passed since they had quit. Subjects who reported never smoking in their lifetime or who had quit for at least 2 yr were categorized as nonsmokers. To eliminate the residual effects of smoking, we excluded subjects who had quit smoking less than 2 yr earlier from this evaluation. Those subjects who reported current smoking were classified as smokers.

Documentation of Smoking Status

To confirm the validity of self-reported smoking status, saliva samples for cotinine analyses were taken from those subjects who were enrolled in data collection at the Ohio State University General Clinical Research Center during June and July 1995 (n = 38). Each subject provided a saliva sample and filled out a questionnaire that described his or her current smoking rate (number of cigarettes per day). On the basis of self-report data, 17 of the 38 subjects were classified as smokers (44.7%). No subject reported using other forms of tobacco (e.g., cigar, pipe, or smokeless tobacco). Cotinine analyses were completed in a blinded manner on the saliva sample obtained from each of the 38 subjects as previously described (26). Eighteen of the 38 subjects' samples contained cotinine at levels greater than 14 ng/ml, which represents the cutoff level that discriminates between smoking and nonsmoking status (27). Cotinine levels ranged from 45 to 586 ng/ml, with a mean level of 210.7 ng/ml (SD = 129.1). With cotonine validation, the smoking prevalence rate for these subjects increased to 47.4%. Only one subject's self-reported smoking status and cotinine-validated status were not in agreement; 17 of the 18 subjects (94.4%) accurately self-reported themselves as smokers, as evidenced by the presence of cotinine in their saliva. These results demonstrate that self-reporting of smoking is a valid measure of smoking, and that the majority of participants enrolled in our clinical study reported their smoking status accurately.

Statistical Methods

Data in the text are expressed as mean ± SEM and data in tables are expressed as mean ± SD. Statistical comparisons were made with the t test for groups > 30 subjects and when variances were equal (as determined by p > 0.05 on Levene's test) and with Welch's analysis of variance (ANOVA) when the variances were not equal, using SAS JMP Sofware version 3.2 (SAS Institute, Cary, NC). For the cytokine measurements, because there were a large number of undetectable values, comparisons were made with the chi-square and Wilcoxon's rank sum tests for the detectable values. The values for HIV-negative individuals were analyzed with Wilcoxon's rank sum test because of the small number of samples.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Population Characteristics

Lung alveolar cell numbers, lung cell cytokine production, and lung diffusion capacity (DL_CO) measurements in 92 HIV-positive individuals who were asymptomatic and had no history of prior lung infection were evaluated as part of a larger study to characterize the natural history of HIV infection. Subjects with a history of injection drug use were excluded from this evaluation because they were more frequent in the smoker population in a ratio of three to one. The data shown in Table 1 characterize the demographics of the studied population. None of the demographic characteristics differed significantly. For comparison, 21 HIV-negative individuals were also studied concurrently (see the following discussion).

Lung Cell Differential Counts in BALF

All study subjects underwent standard bronchoalveolar lavage (BAL) with 100 ml of fluid. Table 2 shows the observed BAL cell differential counts. Smokers had higher percentages of macrophages and lower percentages of lymphocytes (p < 0.0001).

CD4⁺ Cell Counts in Blood and Lung

Since CD4⁺ lymphocytes are important regulators of the immune response, CD4⁺ T cells were counted in the blood and lung of HIV-positive individuals in this evaluation. It should be noted that CD4⁺ cell counts in smokers' blood were somewhat higher than those in nonsmokers' blood (Figure 1A), but the differences were not statistically significant (359 ± 33 cells/µl versus 278 ± 42 cells/µl; p = 0.13, t test). The lack of a deficit in CD4⁺ cell counts in smokers' blood as compared with nonsmokers' blood suggests that smoking does not directly impair systemic lymphocyte responses. In contrast to the blood CD4⁺ cell counts, the lung BALF cell CD4⁺ cell counts of smokers were significantly depressed (4.9 ± 0.8 × 10³ cell/ml versus 22.2 ± 5.8 cells × 10³/ml BALF; p = 0.006, Welch's ANOVA, smokers versus nonsmokers, respectively; Figure 1B). In addition to the findings mentioned, there was also a trend toward a lower CD4⁺/CD8⁺ cell ratio (T-helper/ T-suppressor ratio) among BALF lymphocytes (0.27 ± 0.04 versus 0.45 ± 0.12, smokers versus nonsmokers; p = 0.155, Welch's ANOVA; Figure 1C). These findings imply that smokers may have a compartmentalized deficit in cell-mediated immunity in the lung alveolus, a site critical in first line defense against infection.

View larger version (12K): [in this window] [in a new window]   Figure 1.   CD4+ cell counts in blood and lung. (A) CD4+ T-cell numbers in each millliter of blood of nonsmokers (n = 34) and smokers (n = 58). Note that in contrast to the result for the lung, CD4+ cell counts in smokers' blood trended nonsignificantly toward higher values (p = 0.13). (B) CD4+ T-cell numbers in the same population of smokers and nonsmokers as in (A), counted per milliliter of BALF recovered; **p = 0.0003. (C ) Helper/suppressor ratio for lung T lymphocytes in smokers and nonsmokers. The average ratio of CD4+ to CD8+ lymphocytes is presented for nonsmokers (n = 30; open bars) and smokers (n = 53; solid bars); (p = 0.155).

BALF Cell Recovery (Smokers versus Nonsmokers)

Although there was a deficit in the numbers of CD4⁺ lymphocytes recovered from the lungs of HIV-positive smokers, there was at least a twofold greater number of recovered AM in smokers as compared with nonsmokers. Smokers had significantly higher numbers of recovered BALF cells (34 ± 3.9 × 10⁶ cells/total lavage versus 16 ± 2.0 × 10⁶ cells/total lavage [mean ± SEM]; p < 0.0001, Welch's ANOVA). Figure 2 compares the results for the study population for the major cell types recovered by BAL, expressed as absolute cell counts/ml of BALF. Smokers had higher numbers of total cells recovered/ml BALF (742 ± 74 × 10³ cells/ml versus 339 ± 37 × 10³ cells/ml BALF; p < 0.0001, Welch's ANOVA) and higher numbers of macrophages/ml (656 ± 65 × 10³ cells/ml versus 236 ± 26 × 10³ cells/ml BALF; p < 0.0001, Welch's ANOVA, smokers versus nonsmokers, respectively). Since macrophages have also been characterized as immunosuppressive cells in cell-mediated immune responses (10), these increases in macrophage number may also promote an immunosuppressive environment in smokers.

View larger version (12K): [in this window] [in a new window]   Figure 2.   Total cells counts in nonsmokers' and smokers' BALF. The graph shows the mean ± SEM of absolute cell counts for total cells recovered and macrophages and lymphocytes from BALF of nonsmokers (n = 34; open bars) and smokers (n = 58; solid bars) expressed as × 103 cells/ml BALF; **significantly lower values for smokers versus nonsmokers (p = 0.0001).

Lymphocyte Subtypes Recovered

The average number of lymphocytes recovered was depressed in smokers as compared with nonsmokers (40 ± 6 × 10³ cells/ ml versus 96 ± 18 × 10³ cells/ml BALF recovered, p = 0.006, Welch's ANOVA; Figure 3A). Lymphocytes were further subdivided into CD8⁺ cells and the CD8⁺ subtypes S6F1⁺ (which phenotypically marks the cytotoxic cells) and S6F1 (which denote the suppressor cells). The mean numbers of lymphocytes of the subtypes CD4⁺/CD3⁺ and CD8⁺/CD3⁺, and of both the S6F1⁺ and S6F1 CD8⁺ subtypes, were also depressed in smokers versus nonsmokers, respectively, as follows: CD4⁺/CD3⁺ = 4.9 ± 0.8 × 10³ cells/ml versus 22.2 ± 5.8 × 10³ cells/ml BALF recovered, p = 0.006, Welch's ANOVA; CD8/CD3 = 26.1 ± 4.4 × 10³ cells/ml versus 63.0 ± 13.9 × 10³ cells/ml of BALF recovered, p = 0.0158 Welch's ANOVA; S6F1⁺/CD8⁺ = 16.1 ± 3.2 × 10³ cells/ml versus 35.6 ± 10.0 × 10³ cells/ml BALF recovered, p = 0.0725, Welch's ANOVA; and S6F1/CD8⁺ = 4.3 ± 0.9 × 10³ cells/ml versus 15.6 ± 3.8 × 10³ cells/ml BALF recovered, p = 0.0070, Welch's ANOVA (Figure 3A).

View larger version (13K): [in this window] [in a new window]   Figure 3.   Lymphocyte subtypes in BALF. Mean ± SEM for lymphocytes, CD4+/CD3+ cells, CD8+/CD3+ cells, and cytotoxic phenotype of CD8+ cells as measured by S6F1 surface expression. (A) HIV-positive subjects' BALF. Results for HIV-infected nonsmokers (n = 34; open bars) and HIV-infected smokers (n = 58; solid bars) (see text for statistics). (B) Normal BALF. Results for uninfected control nonsmokers (n = 12) and smokers (n = 9).

Cytokine Production by BALF Cells from Smokers and Nonsmokers

As in the case of the lymphocyte subpopulations discussed previously, spontaneous IL-1 and TNF- production by the BALF cells of smokers was markedly decreased. In these experiments, freshly harvested BALF cells were cultured overnight in RPMI 1640 medium and 5% FBS, either alone or with 1 µg/ml of LPS. Cell-free supernatants were then analyzed for IL-1 and TNF- with a sandwich ELISA. Table 3 shows that the average spontaneous cytokine release was lower in smokers than in nonsmokers.

Because a large fraction of BALF samples showed no spontaneous IL-1 or TNF- release, smokers and nonsmokers were also compared for this tendency. Analysis of 56 smokers and 31 nonsmokers samples revealed 37 smokers' (66% of smokers) and 14 nonsmokers' (45% of nonsmokers) samples without detectable spontaneous IL-1 release, whereas 19 smokers' (34% of smokers) and 17 nonsmokers' (55% of nonsmokers) samples had detectable IL-1 release (chi-square = 3.597; df = 1, p = 0.058). When only those samples with detectable spontaneous IL-1 release were evaluated, the mean release showed a trend toward lower values in smokers (p = 0.124, Wilcoxon's rank sum test), but the difference was not significant. For spontaneous detectable TNF- release, smokers released lower levels than did nonsmokers (p = 0.045, Wilcoxon's rank sum test). In contrast, stimulated cytokine release did not differ in smokers and nonsmokers (p = 0.28 and p = 0.68 for IL-1 and TNF-, respectively; Table 3) and was detectable in all samples tested.

In an effort to more accurately reflect the status of proinflammatory cytokine production in vivo in the lungs of smokers and nonsmokers, we also used quantitative PCR for IL-1 in a subgroup of the study population. Although the difference was not statistically significant, the ratio of molecules of IL-1/actin mRNA for smokers (n = 11) was 0.74 ± 0.36, and for nonsmokers (n = 4) was 1.51 ± 1.7, which is consistent with the results for the spontaneous production of IL-1 in culture.

Effect of Smoking on Lung Inflammatory Milieu in Control Subjects

In order to put the effect of smoking in HIV-infected individuals into perspective, we concurrently studied 21 HIV-negative subjects. The characteristics of this control population are outlined in Table 4. As can be seen, smoking was associated with significant increases in the percentages of macrophages and decreases in the percentages of lymphocytes and in the helper suppressor T-cell ratio in lung BALF cell populations. As in HIV-positive subjects, there was a nonsignificant increase in blood CD4⁺ cell counts in HIV-negative smokers. Furthermore, spontaneous IL-1 and TNF- production was low in the smokers, but this was not statistically significant.

Regarding lung lymphocyte subtypes, Figure 3B shows the lymphocyte data from the normal control population and demonstrates that the major difference between HIV-infected subjects and uninfected controls was in the number of CD8⁺ cells recovered. In contrast to the HIV-infected populations, normal subjects did not show an effect of smoking on the absolute numbers of CD8⁺ cells recovered (9.7 ± 2.4 × 10³ cells/ ml versus 10.2 ± 2.0 × 10³ cells/ml, smokers versus nonsmokers, respectively; p = 0.8639, t test). Furthermore, there was no significant difference in the numbers of S6F1⁺ CD8⁺ lymphocytes (p = 0.325, t test) or S6F1 CD8⁺ lymphocytes (p = 0.0783, Welch's ANOVA) in uninfected smokers and nonsmokers. On the other hand, as seen in our HIV-infected subjects, smoking was associated with a decrease in the numbers of BALF CD4⁺ lymphocytes (5.4 ± 3.4 × 10³/cells/ml versus 16.0 ± 3.0 × 10³ cells/ml BALF; p = 0.0247, Welch's ANOVA). There was also a nonsignificant trend toward fewer total lymphocytes/ml in normal smokers (17 ± 3.6 × 10³ cells/ml versus 30 ± 4.8 × 10³ cells/ml BALF; p = 0.0644, Welch's ANOVA).

Effect of Smoking Intensity on Lung Inflammation

Having demonstrated a profound effect of smoking on the lung inflammatory milieu of HIV-positive smokers, we attempted to correlate the intensity of smoking with the inflammatory markers investigated in our study. Smoking intensity was available for 51 of the 58 current smokers. There were 10 individuals smoking 0.15 to 0.5 packs/d, 28 individuals smoking 0.5 to 1.0 packs/d, and 13 individuals smoking between 1.0 and 3.0 packs/d. When the parameters outlined in the study described previously were correlated with the intensity of smoking, there were no significant correlations. This implies that the effects seen with cigarette smoking are not strongly dose-related.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The mechanisms responsible for increased risk of pneumonia in HIV-positive smokers have yet to be completely elucidated. Previous studies have identified the effect of cigarette smoking on blood CD4⁺ cell counts, and have concluded that smoking may not have a marked effect on clinical outcome (1, 2). Conversely, a number of investigations have demonstrated that cigarette smoking is associated with increased risk of pulmonary infectious complications, particularly acute bronchitis and bacterial pneumonia (3, 5, 6, 28).

Our research indicates that cigarette smoking induces a compartmentalized, suppressive lung inflammatory environment in HIV-positive individuals. We demonstrated that cigarette smoking is associated with a marked depression in both the percentage and absolute numbers of lung lymphocytes. Lung CD4⁺ and lung CD8⁺ cell numbers were suppressed by smoking, and lung CD4⁺/CD8⁺ cell ratios trended toward lower values in smokers. HIV-infected smokers had increased numbers of AM recovered by BAL and showed suppressed spontaneous production of the proinflammatory cytokines IL-1 and TNF-. These compartmentalized changes in the lung contrast with the lack of a defect in peripheral blood CD4⁺ cells in cigarette smokers.

Our findings confirm the observation by Twigg and colleagues (12) of a loss of accessory cell function and a decrease in lymphocyte numbers in HIV-seropositive smokers. Our study further characterized lymphocyte subsets into CD4⁺ and CD8⁺ cells, and divided the CD8⁺ population into cells with cytotoxic and those with suppressor phenotypes. Our findings strongly support the notion that cigarette smoking produces a compartmentalized immunosuppression in the lung.

Of note is that the present study used salivary cotinine measurements to confirm the validity of our smoking questionnaire. Previous studies have relied solely on self-report data to determine smoking status. In the present study, salivary cotinine was measured in a subset of subjects to validate the accuracy of self-reporting in the smoking population. Although the validity of self-reporting is questionable in studies that focus on smoking cessation treatment in clinical populations, our findings show that self-reporting may be appropriate in studies that focus on the identification of smoking status as opposed to efficacy of treatment (31). The significance of our observations must remain speculative at this point, but there is circumstantial information implying that our findings with regard to smoking may have pathophysiologic consequences. With respect to the low lung lymphocyte counts found in smokers, it is well recognized that lung lymphocytes are critical to an effective cell mediated immune response (32). The depressed counts of lung CD4⁺ cells with smoking are clearly out of proportion to systemic CD4⁺ counts, suggesting a localized defect in the cell-mediated immune response in the lungs of smokers. Furthermore, it has recently been suggested that CD8⁺ cell counts in blood may reflect the efficiency of viral clearance (33). The lower numbers of CD8⁺ cells in smokers' lungs imply that antiviral activity may be depressed as a result of smoking. In accord with this concept is the finding in previous work that HIV p24 antigen recovery from macrophages of smokers is greater than that from macrophages of nonsmokers (34).

Furthermore, a lack of TNF- and IL-1 may predispose to bacterial infection, as has been shown in animal models (14- 16). Thus, normal early cytokine responses may be able to focus inflammation against early bacterial invasion and thereby prevent major infectious complications. We found that spontaneous cytokine production was deficient in HIV-positive smokers. This measure may be more representative of in vivo baseline responsiveness. Although endotoxin stimulation was done to document the ability of cytocine-producing cells to respond to a large stimulus, such in vitro cytokine induction may not reflect all of the regulatory factors present in the lung milieu.

It is clear that smoking causes additional defects in the lung's defense against bacterial pneumonia beyond those effects reported here. It is important to note that smoking also impairs mucociliary clearance by direct effects on ciliary function (35). In addition, smoking causes chronic bronchitis, providing a milieu for bacterial colonization and secondary infection of distal airways. In this context, the baseline suppressive lung inflammatory environment that we report here for HIV-positive smokers may further increase their risk for lung infection. Moreover, since smoking depresses lung lymphocyte counts in the face of normal peripheral blood lymphocyte counts, the localized immunosuppression that exists in the lungs of cigarette smokers may have been previously underestimated. These findings support the potential importance of smoking cessation in HIV-infected individuals.