Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Exposure to elevated ambient particulate air pollution (particulate diameter  10 µm) is associated with increased morbidity and mortality, particularly in people with pre-existing lung disease (1, 2). In animal models, the pulmonary effects of inhaled environmental particles include epithelial remodeling (3) and inflammation (4). In vitro, particles can induce proinflammatory cytokine mRNA and/or protein production (e.g., macrophage inflammatory protein [MIP]-2, tumor necrosis factor [TNF]-, interleukin [IL]-8, IL-6) in alveolar macrophage (AM) (7) and epithelial cells (11), depending on particle composition (e.g., transition metal, lipopolysaccharide [LPS] components) and cell culture conditions (e.g., adherent versus nonadherent, LPS-primed).

A limitation of these in vitro studies is that they examined the behavior of AMs or epithelial cells separately. However, in situ, these two cell types encounter inhaled particles simultaneously and may interact to coordinate the biologic response. Recreating AM-epithelial interaction in vitro might therefore provide a more accurate model of the in situ milieu than studying each cell type in isolation. The potential significance of AM-epithelial interaction is suggested by studies in which co-cultures of monocytes with bronchial epithelial cells showed amplified nuclear factor (NF)-B activation, and proinflammatory cytokine production upon soot particle exposure (14, 15).

To expand on these observations with regards to AM-epithelial interactions, we investigated whether co-culture of alveolar epithelial cells with AMs affected the pro-inflammatory response to particles. We tested a panel of particles of different composition (titanium dioxide [TiO₂], -quartz [SiO₂], residual oil fly ash [ROFA], and urban air particles [UAP]). TiO₂ is relatively inert in vivo and in vitro (16), SiO₂ carries reactive oxygen species on its surface (19), ROFA contains soluble transition metals (20), and UAP is a sample of air particles of heterogenous composition and size (21). Our objectives were: (i) to test the hypothesis that particles elicit TNF- and MIP-2 release in AM-epithelial co-culture to levels greater than that produced by each cell type alone, and (ii) to investigate the nature of AM-epithelial interactions that enable increased cytokine response to particles. We specifically assayed for TNF- and MIP-2 because they are important proinflammatory mediators and are elevated by particles in rodents (22).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cells

The rat alveolar type II epithelial cell line RLE-6TN (RLE) established by Driscoll and coworkers (23) was generously provided by B. Mossman (University of Vermont, Burlington, VT). Rat fetal lung fibroblasts (RFL) were purchased from ATCC (Manassas, VA). AMs were freshly isolated by bronchoalveolar lavage (BAL) from female CD rats 12-16 wk old (Harlan Sprague Dawley, Indianapolis, IN).

Cell Co-Culture

RLE were grown to confluence with Dulbecco's modified Eagle's medium (DMEM)/F12 supplemented with 7% NCS (both from Life Technologies, Rockville, MD). RLE cells were serum deprived in 0.5% NCS for 24 h before co-culture with AMs to synchronize them in growth arrest. RFL were grown to confluence in F12K (Mediatech, Inc., Herndon, VA) media supplemented with 20% fetal bovine serum (FBS) (Gemini Bioproducts, Woodland, CA) and also serum-deprived with 0.5% FBS for 24 h before co-culture with AMs.

Rat AMs were freshly obtained by BAL and suspended in either DMEM/F12 with 0.5% NCS for RLE co-culture or F12K with 0.5% FBS for RFL co-culture. AMs were allowed to settle on top of the RLE or RFL monolayer at an AM:RLE or AM:RFL ratio of 1:1.4 (10⁵ AM:1.4 × 10⁵ RLE/RFL) for 1 h before particle exposure. This AM:substrate ratio was chosen after pilot studies which identified this number of macrophages as optimal for detection of cytokine release. Microscopy showed that the AMs appeared to adhere, but not completely cover, the underlying cells. This allowed direct interaction of particles with both the AMs and epithelial cells.

Particle Exposure

The cells were exposed to a panel of particles (TiO₂, SiO₂, ROFA, or UAP) at 1-50 µg/ml in a 1 ml volume. TiO₂ (~ 1 µm diameter) and SiO₂ (~ 1 µm diameter) were obtained from J. Brain (24) while ROFA was provided by J. Godleski (25). UAP was standard SRM 1649 collected in Washington, DC and purchased from the National Bureau of Standards (Washington, DC). Stock suspensions of all the particles in DMEM/F12 + 0.5% NCS were sonicated for 1 min before introduction to the cells. Cells were separately stimulated with 50 ng/ml LPS (Escherichia coli serotype 0127-B8; Sigma Chemical Co., St. Louis, MO) as a positive control. After 24 h exposure, the supernatants were collected and stored at 20°C for TNF- and MIP-2 analyses. Visual inspection of the co-cultures with light microscopy revealed that the particles associated with both AMs and RLE/ RFL cells, indicating that the AMs did not out-compete the substrate cells for particle access. In a subset of experiments, the contribution of particle-adsorbed LPS on cytokine release was determined by pretreating the sonicated particles for 10 min with 10 µg/ml recombinant endotoxin neutralizing protein (rENP; Associates of Cape Cod, Falmouth, MA) before use.

Cytokine Assays

TNF- was measured with a fluorescent microplate bioassay as previously described (17). This bioassay exploits the cytotoxic sensitivity of the fibroblast cell line WEHI 164 clone 13 to TNF-. Cell death is proportional to TNF- concentration and is quantified by nuclear uptake of propidium iodide (Sigma). MIP-2 was measured by sandwich enzyme-linked immunosorbent assay as previously described (26).

Particle Uptake

To distinguish between RLE and AMs in the studies of the effect of co-culture on AMs particle uptake, AMs were loaded with 0.5 µM Cell Tracker Green (CTG; Molecular Probes Inc., Eugene, OR) for 15 min at 37°C before they were cultured alone or co-cultured with the RLE. After 24 h incubation with particles, the monocultured and co-cultured cells were harvested from the cell culture wells with 0.1% trypsin (DIFCO Laboratories, Detroit, MI) in balanced saline solution and resuspended in DMEM/F12 containing 7% NCS. Particle uptake (surface association and intracellular localization) was assessed by measuring the right angle scatter of light (RAS) of viable cells as they passed through the beam of a 15 mW 488 nm emitting air-cooled argon laser (Cyonics Ltd., Sunnyvale, CA) in a Coulter ELITE flow cytometer (Coulter Corp., Miami, FL) (27). RAS is an index of cell granularity and increases in proportion to particle uptake in a particle-specific manner. The RAS of AMs was distinguished from the RAS of RLE by electronically gating the measurements on fluorescent green cells only. In some experiments, the viability of AMs and RLE cells harvested after the 24 h period of culture with particles was determined by propidium iodide exclusion and was typically  80%. No differences in cell viability were apparent between AM monocultures and co-cultures.

Cell-Cell Contact

To investigate the role of AM-RLE cell-cell contact in cytokine release by the co-cultured cells, AMs were physically separated from the RLE monolayer by plating them inside transwell inserts with 0.1 µm diameter pores (Corning Costar, Cambridge, MA) at the same AM:RLE ratio of 1:1.4. Final media volume was 600 µl outside the insert and 100 µl inside the insert. After 1 h, the particles were introduced to the cells under one of the following conditions: inside the insert, outside the insert, or into both compartments. Particle load was adjusted accordingly for the reduced volume in these experiments such that particle mass was identical to the experiments without inserts. Under all three particle exposure conditions, the supernatants inside and outside the insert were collected after 24 h and pooled for cytokine analyses.

MIP-2 Immunostaining In Vitro

RLE cells were grown to confluence in 90 × 90 mm wells of Lab Tek chamber slides (Nalge Nunc Intl., Napierville, IL) and serum-deprived with 0.5% NCS-supplemented DMEM/F12 for 24 h. AMs were added to the RLE at a 1:1.4 AM:RLE ratio to a volume of 400 µl/well. After 1 h, 100 µl of 230 µg/ml SiO₂ or UAP (final mass 23 µg/well) suspended with 1:200 brefeldin A (Golgi Plug, final dilution 1:1000; BD PharMingen, San Diego, CA) in 0.5% NCS-supplemented DMEM/F12 was added to the cells. The cells were incubated with the particles for 6 h at 37°C. The supernatant was removed and the slides dried and stored frozen at 20°C.

The defrosted AM+RLE cells were fixed for 10 min in ice-cold paraformaldehyde followed by 10 min of ice-cold methanol. Cells were incubated in a humid chamber overnight at 4°C with 10 µg/ml goat anti-rat MIP-2 (Santa Cruz Biotechnology, Santa Cruz, CA), then treated with 1:200 biotinylated horse anti-goat immunoglobulin G (Vector, Burlingame, CA) for 1 h at 25°C followed by 1:100 avidin-biotin-peroxidase complex (Vector) for another 1 h at 25°C.

Cell Adhesion Blockers

In experiments testing whether anti-CD54 and anti-CD18 blocked AM-RLE binding, RLE and AMs were pretreated for 10 min with IgG₁, anti-CD54, and/or anti-CD18 (BD PharMingen) at 10 times their final concentrations before co-culture. These mAbs were certified by the company to be azide-free and to contain low endotoxin ( 0.01 ng/µg protein). In experiments testing the ability of heparin (Elkins-Sinn Inc., Cherry Hill, NJ) or RGD peptide (Peptite 2000; Telios Pharmaceutical Inc., San Diego, CA) to block AM-RLE binding, each cell type was pretreated with the blockers at their final concentrations for 10 min before co-culture.

Statistics

Dose-response relationships were tested by analysis of variance followed by Fisher's PLSD post-hoc test to correct for multiple comparisons. All other comparisons were made with unpaired t tests. P < 0.05 was considered significant.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Cytokine Responses to Particles

TNF- and MIP-2 release from RLE, AMs, and co-cultured AM+RLE were evaluated in response to a 24-h exposure to TiO₂, SiO₂, ROFA, or UAP. Stimulation by 50 ng/ml LPS was used a positive control. In RLE alone, TNF- signals after particle or LPS stimulation were below the detection limit of 25 pg/ml. In AMs alone, the TNF- release after particle stimulation was small and only achieved statistical significance with 25 µg/ml UAP (Figure 1A; UAP versus control: 100 ± 2.6 versus 11 ± 0.8 pg/ml; P < 0.05, n = 4-7). Co-culturing AMs with RLE synergistically enhanced basal TNF- levels (P < 0.01); the mean AM+RLE level was 642 pg/ml, which was an order of magnitude greater than the sum of the mean basal AMs level (11 pg/ml) and the mean basal RLE level (0.038 pg/ml). The signal was further potentiated by each particle at specific concentrations. SiO₂ ( 25 µg/ml; P < 0.05, n = 4-7) elicited the highest responses while UAP ( 12.5 µg/ml; P < 0.05, n = 4-7) elicited responses even at a low concentration. ROFA elicited a response only at 25 µg/ml (P < 0.05, n = 4-7). TiO₂ at 50 µg/ml also generated a signal above baseline (P < 0.05, n = 4-7). LPS stimulated equal TNF- release from AMs and AM+RLE.

View larger version (20K): [in this window] [in a new window]   Figure 1.   Particle-induced increases in TNF- and MIP-2 in RLE monoculture (squares), AM monoculture (circles), and AM+RLE co-culture (triangles). AMs were allowed to settle on a monolayer of adherent RLE cells at a 1:1.4 ratio (105 AM: 1.4 × 105 RLE). The cells were exposed to particles for 24 h and the supernatants were assayed for TNF- and MIP-2. LPS was used as a positive control. (A) TNF-. (B) MIP-2. Data are mean ± SE; n = 4-7. *P < 0.05 for AM+RLE particle versus baseline. #P < 0.05 for LPS: AMs versus AM+RLE.

MIP-2 showed patterns similar to those of TNF- under basal conditions and in response to particles (Figure 1B). MIP-2 signals after particle or LPS stimulation of RLE were below the detection limits of the assay. In AMs alone, only UAP  25 µg/ml elicited a significant MIP-2 signal (25 µg/ml UAP versus control: 0.76 ± 0.08 versus 0.55 ± 0.04 ng/ml; P < 0.05, n = 4-7). AM+RLE co-culture (8.58 ng/ml) resulted in synergistically higher basal MIP-2 levels than the sum of RLE (0.15 ng/ml) and AM (0.55 ng/ml). Certain concentrations of particles further augmented MIP-2 levels in AM+RLE co-culture. As found with TNF- release, SiO₂ ( 12.5 µg/ml; P < 0.01, n = 4-7) and UAP ( 12.5 µg/ml; P < 0.01, n = 4-7) were the most bioactive of the particles. ROFA ( 25 µg/ml; P < 0.05, n = 4-7) was more bioactive than TiO₂ (50 µg/ml; P < 0.05, n = 4-7). LPS-induced MIP-2 release from AM+RLE was double the release from AMs alone (P < 0.0001, n = 7).

Particle-Specific Potentiation of Cytokine Release in Co-Culture

Previous reports have shown that a significant fraction of TNF- and MIP-2 signals stimulated by ambient particles in AMs are attributable to adsorbed LPS on the particles (7- 10). In our experiments, the markedly higher MIP-2 response to LPS in AM+RLE than in AMs alone suggested that co-cultures were more sensitive to LPS. It was therefore possible that the potentiated cytokine release to particles, especially to UAP, in co-culture was due to particle-adsorbed LPS. To investigate the contribution of adsorbed LPS to the enhanced cytokine release to particles in co-culture, adsorbed LPS was neutralized by pretreating the particles with rENP before delivery to co-cultured AM+ RLE. The particle concentrations used in this and subsequent experiments were selected for maximal bioactivity (based on results shown in Figure 1). As shown in Figure 2, TNF- and MIP-2 responses to LPS were abolished in the presence of rENP, confirming the efficacy of this reagent to neutralize LPS. The TNF- and MIP-2 responses to SiO₂ also significantly decreased with rENP treatment, indicating that rENP was also effective in neutralizing particle-adsorbed LPS. rENP-treated SiO₂ stimulated significantly higher TNF- than baseline but was unable to generate a MIP-2 signal above baseline, indicating a dependency of the MIP-2 signal on adsorbed LPS for SiO₂. Notably, rENP treatment had no effect on the UAP-stimulated TNF- or MIP-2 responses, which remained significantly higher than baseline. Thus, in contrast to AMs alone, TNF- and MIP-2 responses to UAP in AM+RLE co-culture were independent of adsorbed LPS. Similarly, rENP had no effect on TNF- or MIP-2 elicited by TiO₂ or ROFA.

View larger version (16K): [in this window] [in a new window]   Figure 2.   Effect of rENP on TNF- and MIP-2 responses to particles in AM+RLE co-culture. AMs and RLE were incubated together at 1:1.4 ratio as in Figure 1. Particles were pretreated for 10 min with 10 µg/ml rENP (solid bars) or vehicle (open bars) before introduction to the cells. For each particle, concentrations were used according to maximal bioactivity with minimal cytotoxicity. Supernatants were collected after 24 h exposure with particles. (A) TNF-. (B) MIP-2. Data shown are mean ± SD from one of two experiments with similar results; n = 4-6 per sample. *P < 0.05 for AM+RLE particle versus baseline. #P < 0.05 for vehicle versus rENP.

AM-Epithelial Specific Potentiation of Particle-Induced Cytokines

To determine if the amplified TNF- and MIP-2 responses to particles were specific to AM-epithelial interactions, rat AMs were co-cultured with rat fetal lung fibroblasts (RFL) instead of RLE. TNF- or MIP-2 produced by RFL alone in response to LPS 50 ng/ml, TiO₂ 50 µg/ml, SiO₂ 50 µg/ml, ROFA 25 µg/ml, or UAP 50 µg/ml were below the assay detection limits (n = 2). In contrast to results with AM+ epithelial co-culture, TNF- and MIP-2 produced by AM+ fibroblast co-culture were similar to levels produced by AMs alone in response to the same stimuli (Table 1, n = 2).

Particle Uptake

It was possible that the interaction with the RLE in co-culture influenced the AMs to ingest more particles, resulting in amplified cytokine secretion that merely reflects increased AMs particle load. To determine whether the enhanced cytokine signals in co-culture were associated with altered particle uptake by AMs, cells were exposed to particles as usual and uptake was assessed by measuring the RAS of CTG-labeled AMs. Labeling the AMs with green fluorescent CTG allowed discrimination from RLE cells upon analysis of mixtures of the two cell types by flow cytometry. TNF- and MIP-2 elicited by particles or LPS in AMs or AM+RLE were unaffected by CTG labeling (data not shown, n = 3). RAS increases with particles were the same in CTG-labeled AMs cultured alone or with RLE (Figure 3, n = 3). The amplified cytokine responses to particles in co-culture were unrelated to particle load of the AMs.

View larger version (14K): [in this window] [in a new window]   Figure 3.   Effect of co-culture on particle uptake by AMs. AMs were labeled with Cell Tracker Green before incubating with RLE as in Figure 1. AM+RLE were treated with particles for 24 h. For each particle, concentrations were used according to maximal bioactivity with minimal cytotoxicity. The AMs and RLE cells were detached from the plate with 0.1% trypsin in the presence of Ca2+, then analyzed by flow cytometry. Particle uptake by the AMs was analyzed by measuring increases in RAS after electronic gating on green fluorescent cells. Units are arbitrarily defined by calibration of the flow cytometer. Data are mean ± SE, n = 3. P > 0.05 for each particle AM (open bars) versus AM+RLE (solid bars).

Cell-Cell Contact

To evaluate the requirement of AM-RLE contact for the amplified cytokine responses of co-cultured cells, the two cell types were physically separated by culturing the AMs in transwell inserts suspended above the RLE monolayer. Particles were delivered exclusively inside the insert, exclusively outside the insert, or 50% into each compartment. In a cell-free preparation, TNF- infused inside the insert equilibrated with the medium outside the insert by > 80% after 24 h (data not shown). Figure 4 shows the results from particles delivered inside the insert. Separating the RLE and AMs co-culture with the insert significantly decreased basal and particle-stimulated TNF- release (P < 0.05, n = 3). Similar decreases were obtained in AM+ RLE co-cultures when particles were delivered outside the insert (to RLE alone) or to both compartments (data not shown, n = 2). In the absence of RLE, AM monoculture produced negligible TNF- and MIP-2 signals under all three particle delivery conditions (Figure 4A and data not shown, n = 2). TNF- generated by SiO₂ or UAP in the separated co-cultures was the same as seen in AMs alone. The TNF- generated by TiO₂ and ROFA by the separated cultures was also substantially diminished, but a statistically significant increased residual release was still observed (P < 0.05, n = 3). MIP-2 released by AMs alone was unaffected by plating the AMs inside the insert or on the bottom of the culture well (Figure 4B). In co-culture, separation with the insert significantly decreased basal and TiO₂-stimulated MIP-2 release in AM+RLE (P < 0.05, n = 3). Most notably, separation abolished the enhanced MIP-2 release seen in contact co-cultures treated with SiO₂, ROFA, or UAP.

View larger version (19K): [in this window] [in a new window]   Figure 4.   Effect of physical separation of RLE and AMs on TNF- and MIP-2 responses to particles in co-culture. AMs were allowed to settle on a monolayer of RLE or inside transwell inserts suspended above the RLE. In the wells containing inserts, particles were delivered inside the insert. For each particle, concentrations were used according to maximal bioactivity with minimal cytotoxicity. Supernatants were collected for cytokine assay after 24 h of particle incubation. (A) TNF-. The responses of AMs alone or AM+RLE to LPS were unaffected by physical separation. (B) MIP-2. The reductions in LPS-induced MIP-2 in AMs alone (P = 0.1) and in basal (P = 0.08) and LPS-induced (P = 0.1) MIP-2 in the separated co-cultures were not statistically significant. Data are mean ± SE, n = 3. *P < 0.05 comparing AM+RLE with or without insert. Open bars, AM; lightly striped bars, AM+insert; solid bars, AM+RLE; dark striped bars, AM+RLE+insert.

Cell Adhesion Molecules Involved in Potentiated Co-Culture Responses

To investigate the mechanism of contact-dependent cytokine potentiation in AMRLE co-culture, RLE and AMs were treated with inhibitors against candidate cell adhesion molecules. We examined integrins because they mediate cell-cell and cell-extracellular matrix (ECM) binding (28, 29) as well as heparin targets, because heparin is a nonspecific anti- inflammatory agent that modulates immune cell adhesion (30). Table 2 shows the effect of heparin, RGD peptide (against 1/3 integrins), and anti-CD18/anti-CD54 (against 2 integrins and one of its ligands) on TNF- and MIP-2 secretion by AM+RLE in response to 50 µg/ml UAP. Similar results were obtained using a range of inhibitor concentrations (heparin: 1, 10, 100 U/ml; RGD: 5, 10, 25 µg/ml; anti-CD18/anti-CD54: 10, 20 µg/ml) (data not shown). None of these agents affected TNF- or MIP-2 release by UAP in AM+RLE. Similarly, none of these agents affected TNF- or MIP-2 release by UAP in AMs alone, although RGD attenuated the TNF- response to LPS in a dose-dependent manner (data not shown). Anti-CD18 and anti-CD54 delivered together increased basal TNF- production in AM+RLE but did not block enhanced TNF- release in response to particles. Thus, the amplified basal or particle-induced cytokine responses seen in co-culture could not be blocked by these inhibitors of 1/2/3-integrins or heparin-binding proteins.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

We found that co-culture of AMs with epithelial cells results in synergistically amplified basal and particle-induced proinflammatory cytokine release. The amplification occurred with all the particles tested. The amplified responses were specific to AM-epithelial interactions as AM-fibroblast co-culture failed to potentiate basal or particle-induced cytokine levels. To elucidate the mechanisms underlying the potentiated cytokine response to particles in co-culture, we examined whether co-culture altered particle uptake by AMs and/or whether the potentiation required physical cell-cell contact. Particle uptake by AMs was similar in AM+RLE co-culture and in AM monoculture so the amplified cytokine responses to particles in co-culture were not due to increased AM activation by increased particle uptake. By contrast, physical separation of AMs from RLE markedly attenuated or completely abolished the synergistic increase in basal and particle-induced TNF- and MIP-2 signals in co-culture. To identify the cell surface molecules that might mediate this contact-dependent intercellular communication, we blocked several candidate molecules: 2 integrins, 1/3 integrins, and heparin-binding proteins. Individual blockade of these different molecules had no effect on cytokine release in co-culture, suggesting that none of these are involved, or that multiple classes of molecules are simultaneously involved.

In vitro studies are valuable for understanding the mechanisms of particle health effects, but most studies so far have focused on single cell types. However, a number of cell types in the lung may interact to coordinate biologic responses. To more accurately model in vivo events in vitro, it may be useful to include interactions among relevant cell types. To this end, we examined AMs and alveolar epithelial cells in co-culture because these are two cell types that interact in the lung and come in direct contact with inhaled particles. Our data demonstrate that proinflammatory outcomes are altered in co-culture compared with monoculture, indicating the importance of considering cell-cell interactions when investigating biologic mechanisms of air pollution effects in vitro.

An additional finding was that basal production of TNF- and MIP-2 increased synergistically with AM+RLE co-culture above basal production by AMs alone. It is unclear whether this elevated baseline signifies that the AMs are slightly activated in the presence of epithelial cells in vitro or whether the elevated baseline is the normal, unactivated state of the AMs in a milieu that more closely resembles their in vivo milieu.

We attempted to identify the source of MIP-2 in AM+RLE co-culture because both RLE and AMs express mRNA for this cytokine upon LPS or SiO₂ exposure (11). When quiescent, LPS-, SiO₂,-, or UAP-exposed AM+RLE cells were immunostained for MIP-2, both cell types showed positive staining under all four conditions (data not shown). However, relative production by the two cell types under the different conditions could not be determined due to technical limitations. To detect MIP-2 inside the cell that produced it, protein secretion has to be blocked with agents such as brefeldin A (Golgi Plug). Under our experimental conditions, this reagent may have obscured differences among the different treatment groups because any MIP-2 that is produced accumulates within the cell and so will give a positive signal with the anti-MIP-2 antibody. Insofar as both cell types stained positively for MIP-2 under baseline and stimulated conditions, we concluded that the secreted MIP-2 was derived from both the RLE and AMs. We did not examine which cell type was the source of TNF-. Although primary type II cells have been reported to produce TNF- upon LPS stimulation in vitro (31), RLE cells did not generate a detectable TNF- signal to LPS in our experiments, indicating that they are likely a poor source of TNF-. By contrast, the AMs responded robustly to LPS in our studies and are well-documented to produce TNF- abundantly. We assumed that even if the RLE were producing TNF- below our detection limits, the majority of the TNF- signal in co-culture would be attributable to the AMs.

Having established that the particles stimulated cytokine production in co-culture, we investigated the potential contribution of trace LPS adsorbed to the particles on the cytokine responses by pretreating the particles with rENP. Only the cytokine responses to SiO₂ were attenuated by rENP treatment. The amplified TNF- and MIP-2 responses to TiO₂, ROFA, and UAP in co-culture were due to properties specific to the particles and not to adsorbed LPS. This finding was particularly intriguing for UAP since a considerable portion of the TNF- and MIP-2 signals stimulated by ambient air particles in AMs in vitro have been reported to be due adsorbed LPS (7). The absence of a LPS contribution to the UAP stimulated cytokine responses in AM+RLE co-culture suggests that intercellular interactions sensitized the cells to respond to other components of UAP.

To investigate the nature of the interaction between the RLE and AMs that gave rise to the amplified cytokine responses in co-culture, the RLE and AMs were physically separated by plating the AMs inside transwell inserts instead of directly on top of the RLE. This allowed the discrimination of the relative importance of soluble mediators versus cell-cell contact on the cytokine responses. Separation abolished the amplified basal and particle-induced TNF- signal in co-culture, regardless of whether the AMs or RLE or both cell types were stimulated with particles. Basal MIP-2 signal was also attenuated, but remained significantly higher than AMs alone. The particles also failed to generate a MIP-2 signal above baseline when the RLE and AMs were separated. These data demonstrated that AM-RLE contact was critical for the amplified TNF- levels at rest and in response to particles. Contact was required for the particle-potentiated MIP-2 signal, but only partially necessary for the amplified basal MIP-2 signal in co-culture. One or more soluble mediators appears to contribute to elevated basal MIP-2. It is unlikely that TNF- plays this role because basal MIP-2 levels were elevated in spite of negligible basal TNF- levels when the co-cultured AMs and RLE were separated by the insert. The effect of TNF- on MIP-2 release by AM or RLE monoculture was not further investigated. The differences in TNF- and MIP-2 outcomes suggest that the mechanisms leading to the amplification of each cytokine differ. Intercellular contact was also necessary for the alterations reported in another co-culture model in which ECM synthesis in response to coal dust required contact between AMs and type II epithelial cells (32). Cell-cell contact thus appears to be important for the effects seen in co-culture.

Because AM-RLE contact was necessary for the enhanced cytokine responses to particles in our system, we sought to identify the cell surface adhesion molecules that might be involved in the potentiation. Integrins and heparin-binding proteins were potential candidates because integrins mediate cell-cell/ECM binding and are coupled to many signal transduction pathways that control cell function (28, 29), whereas heparin is a broad anti-inflammatory agent that modulates immune cell migration (30). None of the inhibitors tested affected the TNF- or MIP-2 responses to UAP in AM+RLE. The potential efficacy of these agents at the concentrations used was evident from other assays. Anti-CD18 and anti-CD54 antibodies labeled AMs and RLE, respectively, by immunofluorescence and flow cytometry (data not shown). Heparin attenuated basal and UAP-induced TNF- in AMs alone whereas RGD peptide attenuated LPS-induced TNF- in AMs alone (data not shown). These outcomes indicated that the agents had the potential to affect cytokine levels in co-culture. Delivering anti-CD18 and anti-CD54 together enhanced basal TNF- and MIP-2 in AM+RLE co-culture without concomitant effects on the particle responses. These results were unexpected, but they were not directly relevant to the identification of the mediator of cytokine amplification to particles, so we did not investigate them further. Although 1/3 integrins, CD18 (2 integrins)/CD54, and heparin-binding proteins on their own did not mediate the amplified cytokine responses to UAP in AM-RLE co-culture, it is possible that one or more of these adhesion molecules may be involved in conjunction with each other or other molecules.

A noteworthy limitation of our studies is that RLE cells are a reasonable but imperfect model of the alveolar epithelium. These are immortalized cells, so they may behave differently from primary type II cells. In addition, a monolayer of type II cells misrepresents the alveolar surface that AMs would encounter in situ, where 96% of the surface area is covered by type I cells even though 60% of the total epithelial cells are type II (33). Nevertheless, the data suggest that (i) intercellular interactions between alveolar epithelial cells and AMs are important determinants of the proinflammatory response to particles in the lung, and (ii) AM-epithelial co-culture may be a useful in vitro model for the study of the biologic effects of air particles.