Abstract
Purpose: Adenocarcinoma with bronchioloalveolar carcinoma (BAC) features is a subtype of non–small cell lung cancers characterized by an intense inflammatory reaction composed of macrophages and neutrophils and by a distinct natural history with intrapulmonary spread leading to death due to respiratory failure. We hypothesized that neutrophils could promote aerogenous spread of lung adenocarcinoma with BAC features.
Experimental Design: We examined the effect of neutrophils on A549 cell line detachment in vitro and we quantified desquamation of tumor cells on tumor tissue (n = 25) and on matched bronchioloalveolar lavage (n = 17) in vivo in a series of patients with adenocarcinoma with BAC features.
Results: Neutrophils induced A549 detachment mediated by signals through cell-to-cell contact. Detached A549 cells were still viable and able to proliferate in vitro. Neutralization studies identified several membrane-bound molecules involved in detachment (i.e., intercellular adhesion molecule-1/lymphocyte function-associated antigen-1, tumor necrosis factor α/tumor necrosis factor α receptor inhibitor, interleukin-1α /interleukin-1α receptor, and neutrophil elastase). In tumor tissue, shedding was detected in all samples, with a median shedding score of 42% (range, 4-95%). Micropapillary clusters were detected in 23 of the 25 tumor tissue samples, with a median micropapillary score of 1.40 (range, 0-2.1), and tumor cells were detected in 7 of 17 lavages. The micropapillary score was associated with a high neutrophil count in bronchioloalveolar lavage (P = 0.051). The shedding cell percentage was a significant factor in shorter survival (P = 0.034, univariate Cox analysis).
Conclusions: Tumor shedding is induced by neutrophils. It is a significant factor of shorter survival and may be an important event in adenocarcinoma progression.
- adenocarcinoma
- bronchioloalveolar carcinoma
- neutrophils
- aerogenous spread
- tumor cell shedding
Lung cancer is the most common fatal malignancy worldwide. Approximately 80% to 90% of cases involve non–small cell lung cancer, and among non–small cell lung cancers, adenocarcinoma is the most common histologic type.
Bronchioloalveolar carcinoma (BAC) of the lung is a subtype of lung adenocarcinoma. Although traditionally grouped with other non–small cell lung cancers, BAC has unique morphologic features and clinical behavior related to its aerogenous spreading. At an early stage of diagnosis, focal BAC is usually cured by surgical resection, whereas at a later stage (1, 2), the multifocal or diffuse form is characterized by pulmonary progression and death due to respiratory insufficiency rather than to dissemination (3). For these patients, optimal therapy still needs to be defined, although several recent studies report that some patients are showing significant responses to tyrosine kinase inhibitors of epidermal growth factor receptor (4).
Since the latest revisions in the WHO/International Association for the Study of Lung Cancer classification in 1999 and 2004, most adenocarcinomas with aerogenous spread have to be referred to as adenocarcinomas with BAC features instead of pure BAC because of the presence of a mixed histologic subtype or stromal, nodal, pleural, and vascular invasion (5). Adenocarcinomas with BAC features seem to be increasing in incidence in the past several years, especially in nonsmoking females (6). In addition, with the introduction of spiral computed tomography for lung cancer screening and early detection programs, small peripheral lesions are being detected, many of which are adenocarcinomas with BAC features (1).
Adenocarcinomas with BAC features are characterized by an intense inflammatory reaction, especially composed of alveolar neutrophils and macrophages (7, 8). Increased numbers of tumor-infiltrating neutrophils are linked to an unfavorable outcome (9). We showed that the tumor environment drove local neutrophil recruitment and activation via C-X-C release of chemokines, such as interleukin (IL)-8 (CXCL-8/IL-8), but also prolonged alveolar neutrophil survival through production of soluble antiapoptotic factors (granulocyte macrophage colony-stimulating factor and granulocyte colony-stimulating factor; refs. 10, 11). We postulated that neutrophils might participate in the aerogenous spread of lung adenocarcinoma and reported previously that tumor-infiltrating neutrophils released soluble factors favoring tumor cell migration along the alveolar basal membrane (12).
In the study reported herein, we examined in vitro the effect of neutrophils on tumor cells and showed that neutrophils were able to induce detachment of tumor cells. We characterized the viability of detached tumor cells and analyzed molecules involved in this detachment. We examined in vivo shedding of tumor cells in a series of patients with adenocarcinoma with BAC features and showed that it was associated with shorter outcome.
Materials and Methods
Cells and reagents
Peripheral blood polymorphonuclear (PMN) neutrophils were isolated from peripheral blood of healthy volunteers by means of density gradient centrifugation (PMN cell separation medium, Eurobio). PMNs were separated from erythrocytes by hypotonic shock and washed thrice in sterile saline. This method yielded >97% pure PMNs, as assessed by May-Grunwald-Giemsa staining. PMNs were resuspended in DMEM (Life Technologies, Invitrogen) with 5 mmol/L HEPES, 2 mmol/L l-glutamine, 105 units/L penicillin, 100 mg/L streptomycin, and 2% fetal bovine serum, hereafter referred to as complete medium. To obtain PMN conditioned medium, PMNs were plated at a density of 6 × 106/mL in complete medium with 2% fetal bovine serum in 24-well culture plates (0.5 mL/plate) for 24 h at 37°C in humidified 5% CO2/95% air. Medium was then harvested, centrifuged, and stored at −80°C until use.
The A549 cell line (American Type Culture Collection) was originally established in culture from a patient with BAC (13) and was grown in complete medium with 10% fetal bovine serum. The LED 03-04 cell line was established from a primary culture of tumor cells obtained from a bronchioloalveolar lavage specimen recovered from one patient with a mixed adenocarcinoma with BAC and papillary features. Briefly, bronchioloalveolar lavage was centrifuged and alveolar cells (tumor and inflammatory cells) were resuspended in a 75-cm2 sterile culture dish in complete medium supplemented with 10% fetal bovine serum. Then, the culture supernatant was changed twice weekly for 3 months. At that time, we obtained a confluent monolayer of tumor cells. The LED 03-04 cell line expressed cytokeratin 7 (clone OV-TL 12/30, DAKO) by flow cytometry and was used at passage 5.
The BEAS2B cell line (American Type Culture Collection CRL-9609) was originally established from normal human bronchial epithelium obtained from autopsy of noncancerous individuals, transformed with an adenovirus 12-SV40 virus hybrid, and cloned. BEAS2B cells were cultured in BEBM medium supplemented with hormonal cocktail BEGM SingleQuots (Clonetics, Cambrex).
We purchased neutralizing goat polyclonal antibodies directed against human lymphocyte function-associated antigen-1 (LFA-1; CD11a/CD18, DAKO), intercellular adhesion molecule-1 (ICAM-1; R&D Systems Europe), tumor necrosis factor α (TNFα; R&D Systems), IL-1α (R&D Systems), and irrelevant control goat polyclonal antibodies (R&D Systems) for neutralization studies. We used mouse monoclonal antibodies against human neutrophil elastase antibody (clone NP57, DAKO), TNF receptor (TNFR) type 1 (clone 4.12, Zymed, Clinisciences), ICAM-1 (clone My13, Zymed), and an isotype-matched antibody (mouse IgG1 and IgG2aκ, DAKO) for immunostaining. The LabTek Chamber Slide System was from Nalge Nunc International, ATGC Biotechnologies. Culture inserts (BD Falcon cell 0.4-μm pore size) and fibronectin were from BD Biosciences Europe. Paraformaldehyde, neutral red dye reagent, propidium iodide, and α1-anti-trypsin were from Sigma-Aldrich. Annexin V-Fluos was from Roche Diagnostics, Eukitt was from VWR International, and freezing medium (Shandon Cryomatrix) was from Shandon France SA.
Tumor cell detachment assay in vitro
A549 cells, LED 03-04 cells, or BEAS2B cells were resuspended at a density of 2 × 105/mL in respective medium and cultured in 24-well cell culture plates (500 μL/well) for 4 h at 37°C. Plates were precoated with fibronectin at 5 μg/mL for 1 h at room temperature. Confluent monolayers of epithelial cells were exposed to PMN (2 × 105-3 × 106 per well; i.e., PMN/A549 ratio, 2:1-30:1) for variable periods. In some experiments, PMNs were fixed in 1% paraformaldehyde for 1 h at 4°C, then washed with PBS, resuspended in complete medium, and added to the epithelial cell monolayer. PMNs could also be placed in culture inserts. Last, PMN conditioned medium could also be used instead of PMNs.
Quantitation assay of viable adherent cells
To estimate the number of viable adherent cells, we used the ubiquitous lysosomal endogenous enzyme N-acetyl-β-d-hexosaminidase as described previously (14). Briefly, after a variable incubation period with PMNs, supernatants containing desquamated cells were removed from adherent epithelial cells, and substrate solution [7.5 mmol/L p-nitrophenol-N-acetyl-β-d-glucose aminide in 0.1 mol/L citrate buffer (pH 5) and v/v 0.5% Triton X-100] was added (125 μL/well) for 30 min at 37°C. Enzyme activity was blocked by the addition of 200 μL/well of 50 mmol/L glycine buffer (pH 10.4) and 5 mmol/L EDTA. Two hundred microliters were transferred to 96-well plates and absorbance was measured at 450 nm (Titer plate Multiskan MS, Labsystems). Results were expressed as percentage of detached cells calculated as follows:
Quantitation assay of viable detached cells
Neutral red test. Supernatant containing detached cells was harvested and incubated with neutral red dye reagent as described previously (15). Briefly, supernatant was incubated with 0.0025% neutral red (2 vol supernatant/1 vol dye) for 5 min, and then the detached cells were counted using a Malassez hemocytometer. Epithelial cells were easily distinguished from PMN according to cytologic features. Viable cells were red. Results were expressed as percentage of viable epithelial cells calculated as follows:
Annexin V-Fluos expression by flow cytometry. Supernatant containing detached cells was centrifuged at 200 × g for 5 min and washed with cold PBS. The cell pellet was resuspended at a density of 1 × 106 cells/100 μL in labeling buffer [10 mmol/L HEPES/NaOH, 140 mmol/L NaCl, 5 mmol/L CaCl2 (pH 7.4)] containing 20 μL/mL Annexin V-Fluos and 20 μL/mL propidium iodide. Cells were incubated for 10 min at room temperature in the dark and analyzed by flow cytometry (Elite flow cytometer, Beckman Coulter) with gating set on forward scatter and side scatter to separate PMN from A549 cells (16). Apoptotic cells were identified as those with a strong Annexin V-FITC signal and a low propidium iodide signal (early apoptosis) or a strong Annexin V-FITC signal and a strong propidium iodide signal (late apoptosis). A549 cells treated with doxorubicin (2 μg/mL) were used as a positive control for apoptosis.
Bromodeoxyuridine cell proliferation assay. To compare the proliferation rate of detached A549 cells after readherence with A549 still adherent cells, the bromodeoxyuridine (BrdUrd) cell proliferation ELISA kit (Roche Diagnostics) was used as described previously (17) according to the manufacturer's specifications. Briefly, 2.5 × 104 cells/500 μL complete medium were plated into a 24-well plate in triplicate and incubated with BrdUrd for 12 and 36 h. Development was carried out with anti-BrdUrd peroxidase mouse antibody. Absorbance was measured at 450 nm (Titer plate Multiskan MS). Results were expressed as proliferation index calculated as follows:
Neutralization studies
PMNs (2 × 106/mL, 500 μL/well) and A549 cells were preincubated with α1-anti-trypsin (200 μg/mL) or several neutralizing antibodies directed against human (a) LFA-1 (10 μg/mL), (b) ICAM-1 (10 μg/mL), (c) TNFα (1 μg/mL), and (d) IL-1α (1 μg/mL) and with irrelevant control goat polyclonal antibodies (1-10 μg/mL) for 1 h at 37°C. PMNs were then fixed in 1% paraformaldehyde, washed in PBS, resuspended in complete medium, and added to the A549 monolayer for 24 h. The number of adherent cells was assessed using the N-acetyl-β-d-hexosaminidase assay. Results are expressed as percentage of inhibition of detachment of A549 cells defined as follows:
Immunocytochemical studies
A549 cells were resuspended at a density of 2.105/mL in complete medium and cultured on a four-well treated glass slide LabTek Chamber Slide System (500 μL/well) for 4 h at 37°C in humidified 5% CO2/95% air and exposed to 1% paraformaldehyde-fixed PMN (500 μL/well at 2.106/mL). After a 17-h incubation period, the supernatant containing desquamated cells was gently discarded, and adherent A549 cells were carefully washed once with PBS and fixed with acetone or 4% formaldehyde/PBS solution for TNFR type 1 or ICAM-1 immunostaining, respectively. The plastic medium chamber and gasket were removed. Samples were blocked in 3% hydrogen peroxide/PBS and in 10% rabbit serum solution/PBS before incubation with the primary antibody for 1 h at room temperature. Standard avidin-biotin immunoperoxidase methods with diaminobenzine as the chromogen were used for detection. To test the specificity of immunostaining, antibodies were replaced by isotype-matched controls.
Clinical samples
To study aerogenous spread of adenocarcinoma, we conducted a search of the tumor registry database maintained by the service of pulmonary and respiratory critical care medicine to identify all consecutive patients with adenocarcinoma having BAC features, presenting with a pneumonia-like consolidation, who had undergone surgical resection from January 1990 to January 2003 (3). All 25 samples were reviewed by the same lung pathologist (M. Antoine) and classified according to 2004 revised WHO/International Association for the Study of Lung Cancer diagnostic criteria (5). Pathologic tumor-node-metastasis (TNM) staging (pTNM) was done according to the current International TNM Classification System for Lung Cancer (18). Clinical, radiologic, and pathologic findings and pTNM staging at diagnosis are given in Table 1 . Patients received no preoperative chemotherapy that could interfere with tissue analysis. According to the national guidelines, biological samples collected at diagnosis could be used for research purpose after diagnosis has been done. Each patient signed an approval form for research use of samples. No samples were collected specifically for this study. Data were analyzed anonymously. The mean age of the patients was 62.9 ± 2.4 years (range, 31-81 years). Follow-up data were recorded until death. The cutoff date for survival analysis was September 2005. Survival time was defined from diagnosis to death or to the cutoff date.
Clinical and radiologic characteristics and pathologic classification of patients with adenocarcinoma with BAC features (N = 25)
Bronchioloalveolar lavage samples. Bronchioloalveolar lavage was used as a diagnostic procedure in 17 cases. Briefly, 200 mL sterile saline in four 50-mL portions were infused into the radiologically abnormal segment or lobe. Fluid was recovered by gentle suction and then pooled and filtered through sterile gauze. Total and differential cell counts were done on cytospin preparations stained with May-Grunwald-Giemsa. Presence of tumor cells was recorded. No samples were collected specifically for this study, and data were analyzed anonymously.
Tissue samples. Tumoral and distant normal pulmonary tissues were obtained from lobectomy (n = 15), pneumonectomy (n = 9), and surgical biopsy (n = 1). All tissues were both routinely fixed in 10% formaldehyde and embedded in paraffin, frozen in liquid nitrogen, and stored at −80°C. Sections were stained with H&E according to a standard protocol.
Evaluation of shedding cells in tumor tissue
To better understand the pathologic substrate of adenocarcinoma aerogenous spreading, we examined all formalin-fixed, paraffin-embedded tissue sections from the 25 surgical pieces for shedding of tumor cells using two scores at ×50 magnification (see Supplementary Fig. S1) and expressed as the mean ± SEM of all sections analyzed per patient [nine sections per patient; range, 3-17; i.e., (a) a shedding score defined as the surface area of isolated desquamated cells/total tumor area per section (19) and (b) a micropapillary score defined as the presence of micropapillary components per section quantified as 0 for undetectable, 1 for <10%, 2 for between 10% and 30%, and 3 for >30% of the tumor surface; ref. 20]. Micropapillary components are characterized by small papillary tufts lying freely within alveolar spaces. These small nests contained no fibrovascular connective tissue cores (19, 21). Scores were evaluated independently by two investigators (M. Antoine and N. Rabbe) with a final consensus.
Immunohistochemical studies
Immunostaining for neutrophil elastase was done on paraffinized embedded tissue. Three-micrometer sections were deparaffinized, rehydrated, and washed with PBS. Immunostaining was processed on an automated instrument (Ventana Nexes, Ventana Medical Systems) using an indirect biotin-avidin system, the Ventana basic 3,3′-diaminobenzidine detection kit (Ventana Medical Systems), according to the manufacturer's instructions. Isotype-matched antibody was used as a negative control. Staining worked for formalin-fixed paraffinized-embedded tissue and not for the eight samples for which Bouin's fixative had been used.
Immunostaining for TNFR type 1 or ICAM-1 was done on frozen tissue (n = 8) as described previously for cytochemical studies, except that frozen tumor samples were embedded in freezing medium, cryotome sectioned (5 μm thick), and evaluated by routine H&E staining.
Statistical analysis
For quantitative variables, comparisons were made using the ANOVA (more than two pairwise comparisons) or Mann-Whitney nonparametric test (two pairwise comparisons) and correlation studies using Spearman's ρ coefficient. Neutrophil alveolitis was coded as dichotomous (high or low level) and the cutoff value was the median of distribution. The threshold of significance was set at P < 0.05. Data were presented as mean ± SE and processed with StatView and Survival Tools 5.0 software (Abacus Concepts).
Univariate Cox regression analysis and Kaplan-Meier analysis were used to evaluate the effect of tumor shedding (shedding score, micropapillary score, and presence of desquamated tumor cells in bronchioloalveolar lavage) on mortality of patients with adenocarcinoma with BAC features.
Results
Tumor cell detachment in vitro was induced by coculture with PMN. Exposure of an A549 monolayer to increasing doses of PMN led to dose-dependent detachment of A549 cells, as measured by the hexosaminidase enzyme assay on adherent cells (Fig. 1A ). Detachment of A549 cells was significantly detectable starting since a ratio of PMN to A549 cells as low as 2:1 and was achieved for a 30:1 ratio. Exposure of the A549 monolayer to other inflammatory cells, such as peripheral blood mononuclear cells or alveolar macrophages, for 24 h did not lead to detachment of A549 even for a ratio of inflammatory cells to A549 of 30:1 (data not shown). Exposure of A549 to one million PMNs for 24 h led to time-dependent detachment of A549 cells significantly detectable after 12 h (Fig. 1B).
Time- and dose-dependent detachment of A549 cells induced by PMNs. Monolayer of A549 cells was exposed to PMN 0 to 3 × 106 per well (i.e., PMN/A549 ratio, 0-30:1) for 24 h (A) and 1 × 106 fresh PMNs per well for a 0- to 24-h incubation period (B). Supernatant was gently removed and the number of adherent cells was assessed using the N-acetyl-β-d-hexosaminidase assay as described in Materials and Methods. Absorbance was measured at 450 nm. Points, mean of 10 different experiments, with four different wells per condition (percentage of shedding cells); bars, SEM.
Detached tumor cells were viable and able to proliferate after readherence. Detached A549 cells were able to once again adhere to plastic wells. Viability of A549 cells was assessed using Annexin V-Fluos binding and neutral red staining. Eighty-seven ± 2.3% of detached A549 cells, after coculture with PMNs, were negative for Annexin V and propidium iodide staining, as were 96.6 ± 1.3% of A549 cells cultured alone (n = 3; Fig. 2A ). However, the intensity of Annexin V binding in detached A549 cells was higher than in A549 cells cultured alone, as seen in Fig. 2A. Using neutral red assay, 98 ± 3% (n = 3) of the detached A549 cells were stained (Fig. 2B). These findings suggested that almost all A549 cells were viable after detachment induced by PMN. The proliferation index according to 24-h BrdUrd uptake was similar for adherent and detached cells (BrdUrd proliferation index; mean (n = 3) ± SEM: 54.6% ± 6.5 versus 53.1% ± 5.8, respectively; nonsignificant), suggesting that detachment did not modify the proliferative activity of A549 cells measured after readherence.
A549 cells were viable after detachment, as assessed by Annexin V-Fluos expression and the neutral red test. A, after a 24-h incubation period with fresh PMNs, detached (medium) and adherent (bottom) A549 cells were evaluated for apoptosis by Annexin V and propidium iodide labeling as described in Materials and Methods. Top, A549 treated with Doxorubine (2 μg/mL) as positive controls for apoptosis. Among detached cells, A549 cells were distinguished from PMNs with gating set on forward scatter and side scatter. Both adherent and detached A549 cells were low Annexin V-FITC signals and low propidium iodide signals. Representative of three experiments. B, monolayers of A549 cells were exposed to 1 × 106 fresh PMNs per well for a 24-h incubation period, and then supernatant-containing detached cells were harvested and incubated with neutral red dye reagent as described in Materials and Methods. A549 (arrow) and PMNs (arrowhead) were easily distinguished from each other by their cytologic features. Representative of three experiments.
The effects of PMNs on cell detachment and viability differed between cancer cell lines and normal immortalized epithelial cell lines. We next examined the effects of PMNs on LED 03-04 cells of a cancer cell line established from a patient with adenocarcinoma with BAC features and on cells of the BEAS2B human bronchial epithelial cell line transformed with the SV40 T antigen. Similar detachment and viability were seen following PMN coculture with LED 03-04 cells and A549 cells, whereas results obtained with BEAS2B cells were different. Percentage detachment induced by PMN on BEAS2B cells was higher than on cancer cells, whereas the percentage viability of detached BEAS2B cells was lower (Fig. 3 ). These findings suggested that cancer cells were more resistant to detachment and to death induced by detachment than were immortalized normal epithelial cells.
Effect of PMN on human cancer cells and immortalized bronchial epithelial cells. A, monolayers of A549, primary LED 03-04, and BEAS2B cells were exposed for 24 h to 0 to 1 × 106 fresh PMNs per well (i.e., PMN/A549 ratio, 0-10:1). Supernatant was gently removed and the number of adherent cells was assessed using the N-acetyl-β-d-hexosaminidase assay as described in Materials and Methods. Absorbance was measured at 450 nm. Points, mean of five different experiments, with four different wells per condition (percentage of shedding cells); bars, SEM. *, P < 0.05 with the ANOVA test. B, viability of shedding cells from A549, primary LED 03-04, and BEAS2B after exposure of the monolayer to 3 × 106 fresh PMNs per well for 24 h was assessed with neutral red dye reagent as described in Materials and Methods. Columns, mean of three different experiments (percentage of viable epithelial cells); bars, SEM.
Tumor cell detachment required cell-to-cell contact with PMNs. Exposure of the A549 monolayer to three million PMNs for 24 h in a Transwell culture system prevented shedding of A549 cells. Exposure to a 24-h PMN conditioned medium had a marginal effect on tumor cell detachment compared with exposure to PMN themselves (Fig. 4 ). These findings indicated that detachment of A549 cells required cell-to-cell contact with PMNs. Paraformaldehyde-fixed PMNs also enabled A549 detachment, suggesting that the molecules involved were constitutively expressed on the PMN membrane (Fig. 4).
Tumor cell detachment induced by PMN required cell-to-cell contact. A monolayer of A549 cells was exposed to 3 × 106 PMNs per well, fresh, fixed in paraformaldehyde 1%, or placed in BD Falcon cell culture inserts (Transwell) and to PMN conditioned medium (CM). The number of adherent cells was assessed using the N-acetyl-β-d-hexosaminidase assay as described in Materials and Methods. Absorbance was measured at 450 nm. Columns, mean of three different experiments, with four different wells per condition (percentage of detached cells); bars, SEM.
Several membrane-bound molecules were involved in tumor cell detachment. To identify membrane-bound molecules involved in the detachment of epithelial cells, we did neutralization experiments using antibodies or inhibitors of several candidate molecules. Inhibition of elastase by its inhibitor α-1 anti-trypsin decreased the detachment of tumor cells (Fig. 5 ), whereas inhibition of matrix metalloproteinases with tissue inhibitor of metalloproteinases 1 had no effect (data not shown). The blockade of adhesion molecules LFA-1 and ICAM-1 with specific antibodies decreased the detachment of epithelial cells compared with the isotype control (Fig. 5). Neutralization of the membrane-bound cytokines TNFα and IL-1α using specific antibodies (Fig. 5) or their recombinant soluble receptors (recombinant human soluble TNFR inhibitor and IL-1α receptor inhibitor, respectively; data not shown) also decreased the detachment of tumor cells. Neutralization of other cytokines (IL-6, IL-1β, and hepatocyte growth factor) with specific antibodies had no effect on detachment (data not shown).
Membrane-bound molecules involved in tumor cell detachment induced by PMNs. Effect of neutralizing antibodies and inhibitors on tumor cell detachment. PMNs (2 × 106/mL, 500 μL/well) and A549 cells (1 × 105 per well) were preincubated with several neutralizing antibodies (Ab) directed against human LFA-1, ICAM-1, TNFα, IL-1α, or irrelevant control goat polyclonal antibodies (IgG) and with α1-anti-trypsin (a1AT) for 1 h. PMNs were fixed in 1% paraformaldehyde before being added to the A549 monolayer. The number of adherent cells was assessed using the N-acetyl-β-d-hexosaminidase assay as described in Materials and Methods. Columns, mean of five experiments, with four different wells per condition (percentage of A549 detachment inhibition); bars, SE.
ICAM-1, TNFR, and elastase were expressed in vivo. As shown by immunocytochemistry analysis, expression of TNFR and ICAM-1 was negative in resting A549 tumor cells in vitro but was induced by contact with PMNs (Fig. 6A ). Expression was observed only for scattered cells and was cytoplasmic for TNFR and ICAM-1. These findings suggest that induction of expression of TNFR and ICAM-1 preceded tumor cell detachment. To extend these findings in vivo to patients, we did immunohistochemistry analysis on tumor tissue from the eight patients for whom frozen tissue was available (Fig. 6B); we showed that tumor cells expressed ICAM-1 in all samples examined and TNFR in all except one. As observed in vitro, TNFR expression was cytoplasmic. ICAM-1 expression was localized on the cytoplasmic membrane, mostly at the apical pole, for adherent cells and was cytoplasmic for desquamated cells. This suggested that expression of ICAM-1 was different according whether the cell was polarized or not. Finally, in 17 of the 25 tissue specimens that were able to be evaluated for elastase staining, numerous elastase-positive PMNs were in close contact with detached tumor cells (Fig. 6B). The IL-1αR antibody was not commercially available for immunohistochemistry.
Membrane-bound molecules involved in tumor shedding were expressed on tumor tissue. A, TNFR and ICAM-1 expression on A549 cells. Immunocytochemical staining for TNFR and ICAM-1 on A549 adherent cells after they were exposed (PMN+) or not (PMN−) for a 17-h incubation period to 1 × 106 per well paraformaldehyde-fixed PMN. Representative of three different experiments. Magnification, ×200. B, tumor cells expressed TNFR and ICAM-1, whereas PMN expressed elastase in tumor tissue as assessed by immunohistochemical studies. Representative of eight experiments. Magnification, ×200.
Tumor shedding was associated with neutrophil alveolitis and shorter survival. We used 25 tissues and 17 matched bronchioloalveolar lavages from 25 patients with adenocarcinoma with BAC features (Table 1) to determine the presence of tumor shedding in vivo. In tumor tissue, shedding was detected in all samples, with a median shedding score of 42% (range, 4-95%; n = 25). Micropapillary clusters were detected in 23 of the 25 tumor tissue samples, with a median micropapillary score of 1.40 (range, 0-2.1; n = 25), and tumor cells in 7 of the 17 bronchioloalveolar lavages. The shedding score was higher for patients in whom tumor cells were detected by bronchioloalveolar lavage (P = 0.011, Mann-Whitney nonparametric test; Fig. 7A ). Furthermore, the micropapillary score was significantly correlated with the shedding score (P = 0.021; ρ = 0.628; Fig. 7B) and was positively associated with the PMN count in bronchioloalveolar lavage (P = 0.051, Mann-Whitney nonparametric test; Fig. 7C). Taken together, these findings suggested that during the tumor cell shedding process, shedding resulted in micropapillary formation in alveolar lumen and could thus be indirectly evaluated by bronchioloalveolar lavage.
Shedding in vivo in tumor tissue. A, shedding and micropapillary scores were correlated in vivo. Dot, a tumor sample (n = 25). P < 0.05 was considered significant with the ρ Spearman's correlation test. B, bronchioloalveolar lavage with the presence of desquamated tumor cells (tumor cells +) was associated with a higher shedding score in tumor tissue than bronchioloalveolar lavage without the presence (tumor cells −). P < 0.05 was considered significant with the Mann-Whitney nonparametric test. C, tumor tissues with a high micropapillary score have higher neutrophil alveolitis in bronchioloalveolar lavage than tumor tissues with a low micropapillary score. P = 0.051 with the Mann-Whitney nonparametric test. D, shedding cell percentage was a factor in poor survival of patients. Differences in survival are presented according to the median of the shedding score. P = 0.03 with univariate Cox regression analysis and P = 0.08 with the Kaplan-Meier test.
To investigate whether the shedding of tumor cells had prognostic significance in patients with adenocarcinoma with BAC features, it was included in univariate Cox regression analysis. None of the patients were lost to follow-up. Seventeen patients had died by the cutoff date for this analysis. The median survival time after diagnosis was 40 months (range, 0.5-179.5). The shedding cell percentage was a significant factor in shorter survival (P = 0.034, univariate Cox analysis). Patients with a high shedding percentage (>median of the distribution, 42%) were 2.4 times more likely to die (relative risk, 2.4; 95% confidence interval, 0.85-6.61; Fig. 7D).
Discussion
In the study reported herein, we found that neutrophils induce detachment of tumor cells by signals through cell-to-cell contact and that detached tumor cells are still viable and able to proliferate in vitro. Several membrane-bound molecules, such as adhesion molecules, cytokines, and enzymes are involved in this phenomenon. They are constitutively expressed at the cell surface of neutrophils and are inducible on tumor cells by PMN contact. In a highly homogeneous population of patients with adenocarcinoma with BAC features, substantial tumor cell shedding measured on tumor tissue was associated with the presence of desquamated tumor cells in bronchioloalveolar lavage. Tumor cell shedding on tumor tissue was associated with more severe neutrophil alveolitis in bronchioloalveolar lavage and was a factor of poor outcome. Tumor shedding induced by neutrophils may be an important event in adenocarcinoma progression.
The harmful effect of neutrophils on pneumocytes has been incriminated in the destruction of lung parenchyma seen in emphysema and in mediating lung injury in adult respiratory distress syndrome and fibrosing alveolitis. In that research area, detachment of lung epithelial cells induced by neutrophils has been examined previously, using tumor or transformed epithelial cells but also primary small airway epithelial cells (22–25). According to the different coculture conditions used (resting neutrophils or not, small or large numbers of neutrophils, incubation period, granule extracts, postsecretory conditioned medium, culture inserts, etc.), detachment of epithelial cells may have been related to soluble mediators secreted by neutrophils or to direct cell-to-cell contact between neutrophils and epithelial cells. We chose to work with resting neutrophils at low doses and for a relatively long period of coculture, as these are the most appropriate for simulating in vivo conditions of adenocarcinoma with BAC features. We observed much stronger detachment through cell-to-cell contact than through soluble mediators, in contrast with most previous reports studying acute lung injury. In the latter situation, large numbers of neutrophils are recruited in a very short time, often by the presence Gram-negative bacteria toxins (lipopolysaccharide). Under these conditions, stimulated neutrophils, at high doses and for a relatively short period of coculture, induced detachment related to soluble mediators, such as secreted elastase (23, 25).
In the present study, paraformaldehyde-fixed PMN induced detachment of tumor epithelial cells, showing that protease activity was not required by neutrophils. We identified several molecules involved in the detachment of tumor epithelial cells that were adhesion molecules (ICAM-1 and LFA-1), transmembrane cytokines (TNFα and IL-1α), and membrane-bound elastase. The experiments of neutralization with paraformaldehyde-fixed PMN showed that LFA-1, TNFα receptor, IL-1α, and neutrophil elastase were constitutively expressed on neutrophils. ICAM-1 and TNFα were inducible on tumor cells after neutrophil contact. Further studies are needed to better understand the cascade of events implicating these molecules in cell detachment, to consider them as potential therapeutic targets for blocking aerogenous spread. However, interactions between these membrane-bound molecules at the epithelial and neutrophil cell surfaces could be similar to the ones observed between B and T cell during antigen presentation. We were not able to find neutrophils and/or epithelial tumor cells from knockout animals for gene encoding the membrane-bound molecules involved in tumor cell detachment.
Neutrophils have been linked to carcinogenesis for the past several years. Initially, carcinogenesis induced by neutrophils was evaluated by Weitzman et al. (26). They showed that a mouse fibroblast cell line was converted to form tumors in nude mice after coculture with neutrophils. Several other experiments supported the hypothesis that tumor-infiltrating neutrophils have the ability to induce mutations in tumor cells through the production of reactive oxygen or nitrogen species (27, 28). Others have reported that neutrophils could accelerate the malignancy of tumor cells, especially by inducing a metastatic phenotype (29), or that they might directly induce neovascularization through the release of vascular endothelial growth factor and IL-8 (30). Finally, studies showed that neutrophils influenced cancer progression by favoring tumor cell motility or invasiveness. We already showed that tumor-infiltrating neutrophils recovered from patients with adenocarcinoma with BAC features released mitogenic and scattering factors, such as hepatocyte growth factor, favoring cancer cell migration (12). In breast cancer, contact between breast cancer cells and PMNs induced the release of stored oncostatin M from neutrophil granules necessary for tumor invasiveness; the mechanism of such contact is under investigation (31). In the present study, neutrophils induced cancer cell desquamation in vitro with no cytotoxicity on desquamated cancer cells, as shown by Annexin V-Fluos expression and neutral red staining, nor was there any loss of proliferative potential, as shown by the BrdUrd proliferative index. We hypothesized that in vivo, desquamated cancer cells are able to readhere, proliferate, or migrate to neighboring alveolar areas.
To study in vivo the aerogenous spread of lung adenocarcinoma, we collected tumor tissue from patients with adenocarcinoma with BAC features presenting with parenchyma consolidation and who had undergone surgical resection from January 1990 to January 2003 (3). Micropapillary components, desquamated tumor cells in tumor alveolar spaces, and desquamated tumor cells recovered by bronchioloalveolar lavage were detected in 92%, 100%, and 41% of patients, respectively. Carcinomas with micropapillary components have been described in several organs with gland/air lumen or cavities, including breast, urinary bladder, ovary, and lung. Micropapillary components have been detected in 40% of lung adenocarcinomas (19, 20). The very high rate of detection observed in our population may indicate that it is a marker of aerogenous progression. In a series of 344 patients with early-stage lung adenocarcinoma, a micropapillary component was associated with the occurrence of intrapulmonary metastasis and with a nonsmoking status, two factors characteristic of aerogenous spread of adenocarcinoma (19).
Desquamated cancer cells in the alveolar space of tumor tissue was quantified as a “shedding score” and was significantly associated with poor survival using univariate Cox regression analysis. A trend toward significance was observed only when using the log-rank test when a “shedding score” was dichotomized according to the median of distribution. The discrepancy observed between the two tests was probably related to the small size of our population. Tumor cells desquamated in bronchioloalveolar lavage was a factor of shorter survival in a nonsurgical series of 52 patients with adenocarcinoma with BAC features, and a micropapillary pattern has also been related to poorer outcome in adenocarcinoma (12, 19). All these findings suggest that tumor cell desquamation, whatever the test used to measure it, should be considered as a factor for poor prognosis in lung adenocarcinoma.
One of the most common genetic alterations in adenocarcinoma with BAC features is activating mutations of K-Ras (32, 33). Several mouse models of lung adenocarcinoma generated by K-Ras activation are characterized by aerogenous spreading and by abundant infiltration of tumor tissue by macrophages and neutrophils (7, 8, 34). Interestingly, Sparmann et al. (35) have shown that Ras activation induces secretion of IL-8 (CXCL8), eliciting a local inflammatory reaction with neutrophils critical for neovascularization and sustained tumor growth. The next step, upon which we are currently working, is to better characterize the role of Ras activation in aerogenous spread and neutrophil infiltration in human lung adenocarcinoma.
Footnotes
-
Grant support: La Ligue contre le Cancer-Comité de Paris (RS06/75-103), Rectorat de l'académie de Paris-Chancellerie des Universités (Legs Poix 8 juin 2006), and Cancéropole Ile-de-France-Institut National du Cancer (R04088LL Allocation RMA04016LLA).
-
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
-
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).
- Received October 23, 2006.
- Revision received January 25, 2007.
- Accepted March 22, 2007.
References
Volume 13, Issue 12
