Human conjunctival epithelial cell responses to platelet-activating factor (PAF): signal transduction and release of proinflammatory cytokines

Isolation and culture of human conjunctival epithelial cells

The procedures for isolating and culturing human conjunctival epithelial cells (HCECs) from 21 post-mortem human donors were as previously described with minor modifications [18,19]. Briefly, human conjunctival tissue was aseptically dissected (within 8–12 h of death) and transported from eye banks in Dexol^® or Optisol^® preservation medium (Chiron Ophthalmics, Irvine, CA) in ice. Tissue from different donors was kept separate and rinsed in phosphate buffered saline (PBS). It was then treated with dispase (Collaborative Research, Bedford, MA) at 10 U/ml in 50% Hanks buffered salts and keratinocyte basal medium (KGM; Clonetics, San Diego, CA), containing 0.05 mM calcium, at 4 °C for 24–48 h. Complete KGM was prepared by adding 30 µg/ml of bovine pituitary extract, 0.5 µg/ml hydrocortisone, 0.05 µg/ml amphotericin B and 50 µg/ml gentamicin, 5 µg/ml insulin, 10 µg/ml transferrin, and 0.05 mM calcium chloride. After incubation with dispase, the slightly yellow conjunctival epithelium was removed with a scalpel from the white connective tissue. The epithelium was gently dissociated to generate individual cells and washed in Dulbecco’s modified Eagle medium (DMEM) containing 10% fetal bovine serum using a centrifugation (280x g/10 min)/re-suspension procedure. The cell pellet was re-suspended in a low calcium (0.05 mM) KGM, plated into T75 flasks (collagen type IV coated), and incubated at 37 °C under a humidified atmosphere of 95% air and 5% CO₂. The medium was changed a day later and then changed every two days thereafter. The cells became confluent in approximately 10 days at which point they were subcultured (passage 1; P1) by rinsing with PBS, incubating in dispase for about 20 min until the cells detached, washing in DMEM by centrifugation, and then plating on collagen-coated 24 well culture plates (Corning/Costar, Cambridge, MA). All subsequent experiments for PI turnover and intracellular calcium mobilization were performed with the P1 cells. For the [Ca²⁺]_i experiments, the cells were cultured on sterilized #0 glass coverslips (Biophysica Technologies Inc., Sparks, MD) for three to seven days.

PAF-induced PI turnover

Phosphoinositide (PI) turnover assays were conducted as follows. Briefly, HCECs were incubated with [³H]myo-inositol (1 µCi/0.5 ml; 15–17 Ci/mmol; Amersham, Arlington Heights, IL) in DMEM for 24 h at 37 °C. After this time, the medium was aspirated and the cells were rinsed with 1 ml of warm DMEM and then exposed to PAF or methyl carbamyl (mc)PAF (10 pM–10 µM; Biomol Inc., Plymouth Meeting, PA) in DMEM that contained 10 mM LiCl for 60 min at 37 °C to stimulate the production and accumulation of [³H]inositol phosphates ([³H]IPs) [16,17]. To determine the potencies of some commercially available PAF antagonists (Biomol Inc.), the latter drugs were added to the cells 30 min before the addition of PAF. The medium was aspirated at the end of the incubation, and the assay stopped with the addition of 1 ml of ice-cold formic acid (0.1 M). After about 15 min, 0.9 ml of the cell lysates were transferred to Econo-columns^® (Bio-Rad, Richmond, CA), which contained 1 ml AG1X8 ion-exchange resin in the formate form. The columns were washed with deionized water (10 ml) to remove the free [³H]myo-inositol, which was discarded, and the water-soluble [³H]IPs were then eluted from the columns using 4 ml of ammonium formate (1.2 M) made in 0.1 M formic acid. A water-accepting scintillation fluid (15 ml) was then added to the eluates, and the [³H]IPs were quantified by liquid scintillation spectrometry on a beta counter at about 50% efficiency. The data obtained from these studies were analyzed using a nonlinear, iterative, sigmoidal curve-fitting computer program [16,17] and the appropriate means±SEM calculated for results from several independent experiments and from several cultures of HCECs from different donors. The PAF antagonists used in these and other studies described below included the following: 1-o-hexadecyl-2-o-acetyl –sn-glycero-3-phospho (N,N,N-trimethyl) hexanolamine (abbreviated to hexanolamine); methyl 2-(phenylthio)ethyl-1,4-dihydro-2,4,6-trimethyl-pyridine-3,5-dicsrboxylate (PCA-42481); rac-3-(N-octadecylcarbomoyl)-2-methoxy) propyl-(2-thiazolioethyl) phosphate (CV-3988); and (+/−)-cis-3,5-dimethyl-2-(3-pyridyl)thiazolidin-4-one HCl (SM-10661).

[Ca²⁺]_i mobilization

Intracellular calcium mobilization studies were performed as follows. Briefly, HCECs (passage 1) were cultured on sterilized glass coverslips (#0; Biophysica Technologies Inc.) for three to seven days. On the day of experiment, the cells were loaded with fura-2 acetoxymethyl ester (Fura-2/AM; Molecular Probes Inc., Eugene, OR) by incubating at room temperature for 30 min with Buffer A (125 mM NaC1, 5 mM KC1, 1.8 mM CaC1₂, 2 mM MgC1₂, 0.5 mM NaH₂PO₄, 5 mM NaHCO₃, 10 mM glucose, 0.1% BSA, 5 mM Fura-2/AM, 10 mM HEPES, pH 7.2). After the incubation, the coverslip was rinsed twice with Buffer B (Buffer A without BSA or Fura-2/AM) and mounted in an incubation chamber on the mechanical stage of a Nikon Diaphot microscope (Nikon, Garden City, NY). The chamber was filled with 2 ml of Buffer B at room temperature during the experiment. At times indicated, PAF and/or antagonists were added in volumes of 20 µl and were removed by rinsing with 5 rinses of 2 ml each time of Buffer B. Calcium ratio fluorometry was performed using a DeltaScan-4000 ratio fluorescence system (Photon Technology International, South Brunswick, NJ) [20]. The fluorescent dye, Fura-2, within the cells was excited by alternating between 340 nm and 380 nm excitation wavelengths, and the intensity of the intracellular emission fluorescence at 510 nm for each excitation wavelength was monitored in real time by a SIT microscope video camera (model C2400; Hamamatsu Hamamatsu City, Japan). The intracellular Ca²⁺ concentration ([Ca²⁺]_i) of each cell was calculated from the intensity ratio of fluorescence at the two excitation wavelengths according to the equation of Grynkiewicz et al. [21] and the appropriate mean±SEM calculated for results from several independent experiments using fresh cells from new donors. In some experiments, when the exact concentration of calcium was not critical to the interpretation of the data, only the intensity ratio was presented. In such cases, a higher ratio indicates higher [Ca²⁺]_i.

PAF-induced cytokine release

The release of cytokines and their quantification were performed using specific ELISAs as previously described [18]. Briefly, for each experiment, the cells from an individual donor were trypsinized as above and subcultured overnight at a density of 2×10⁴ viable cells/ml per well using fibronectin (FNC)-coated 24 well Primaria^® plates (Becton Dickerson Labware, Lincoln Park, NJ). Duplicate or triplicate cultures were set up for each experimental group. The following morning, the medium was replaced with a fresh KGM medium (deficient in hydrocortisone) with or without PAF, and the cultures were incubated at 37 °C for an additional 24 h unless otherwise specified. Supernatants were collected at various times during the stimulation period for time-course experiments, centrifuged at 200x g for 3 min, transferred into separate tubes, and stored frozen at -20 °C. When the effects of PAF antagonists were investigated, they were added to the cells simultaneously with PAF. Cell monolayers were lysed by freeze–thaw cycling at -70 °C followed by sonication on ice. The lysates were clarified by centrifugation as described and stored at -20 °C. The cell culture supernatants were analyzed for IL-6, IL-8, and GM-CSF by ELISA (Quantikine Kits™; R&D Systems, Minneapolis, MN) as directed by the manufacturer. The optical density readings of the resultant color change were measured at 450 nm (540 nm reference) using a Bio-Tek™ microplate reader (Bio-Tek Instruments, Winooski, VT).

Data analyses

Functional data (dpm/well; pmol/well, etc.) were analyzed using a nonlinear, iterative, sigmoidal curve-fitting computer program (Origin®; OriginLab Corp., Northamptin, MA) [16,17]. Data from several experiments from different donors were combined and represented as the percentages of the maximal responses/inhibitions to provide an overall summary of the key pharmacological data. The molar potency (EC₅₀) values represented the concentration of PAF required to produce 50% of the maximal response. The molar antagonist potency (IC₅₀) represented the concentration of the PAF antagonists to cause 50% inhibition of the maximal functional response induced by PAF. Equilibrium drug inhibition constants (drug dissociation constants; K_is) derived from the functional assays represented the molar drug concentrations required to produce 50% inhibition of the response while the pK_i values represented the –log of this parameter. The equilibrium antagonist potency (K_i) values were calculated according to the equation [16,18]:

PI turnover studies

PAF (100 nM) maximally stimulated PI turnover in primary HCECs by 2.3±0.02 fold (n=21) above basal levels and with a potency (EC₅₀) of 5.9±1.7 nM (n=4; Figure 1A). For example, the [³H]IPs generated by 100 nM PAF were 10,160±1,254 dpm (n=21) while the basal control levels of [³H]IPs were 4,373±606 dpm (n=21). This represents a 2.3 fold stimulation above basal levels. However, the PI response diminished significantly relative to the maximal response when the concentration of PAF was increased to 1–10 µM (Figure 1A).

The effects of PAF (100 nM) on PI turnover were potently and concentration-dependently blocked by the PAF antagonists, hexanolamine (IC₅₀=694 nM; K_i=38 nM; -log K_i [pK_i]=7.4±0.12), PCA-42481 (IC₅₀=897 nM; K_i=50 nM; pK_i=7.4±0.23), CV-3988 (IC₅₀=13.2 µM; K_i=771 nM; pK_i=6.1±0.21), and SM-10661 (IC₅₀=14 µM; K_i=789 nM; pK_i=6.1±0.01 [n=3 independent experiments for each antagonist]; Figure 1B).

[Ca²⁺]_i mobilization studies

PAF was observed to stimulate [Ca²⁺]_i mobilization in most cells studied. However, the cellular responses varied for 10 nM PAF, ranging from no response (Figure 2A), a rapid response within a few seconds followed by response decay (Figure 2B), a rapid response followed by sustained [Ca²⁺]_i mobilization (Figure 2C), and a rapid response followed by oscillatory behavior (Figure 2D). The predominant type of response, however, was that shown in Figure 2C. The [Ca²⁺]_i mobilized in 76 cells peaked after treatment with 10 nM of PAF or mcPAF, which was then followed by a sustained elevated plateau level lasting at least 15 min (Figure 2C; Figure 3A). During the current studies, the mean resting [Ca²⁺]_i in primary HCECs was found to be 57±8 nM (mean±SEM, n=23), and this increased to different levels depending on the cell studied and the concentration of PAF added to the incubation medium.

PAF or mcPAF caused an increase in [Ca²⁺]_i in most primary HCECs in a concentration-dependent manner (Figure 3B). A detectable increase in [Ca²⁺]_i was observed when 0.1 nM of PAF was added to the HCECs, maximal responses to PAF in PI (Figure 1A) and Ca2⁺ (Figure 3B) assays were induced by 100 nM PAF from a concentration-response viewpoint. Cumulative concentration-response curves were constructed for [Ca²⁺]_i mobilization by plotting the peak responses versus the corresponding PAF concentrations (e.g., Figure 3C). The mean EC₅₀ value for PAF was calculated to be 0.81 nM (-log EC₅₀ [pEC₅₀]=9.09±0.18, n=14). The mean peak [Ca²⁺]_i mobilized by 10 nM PAF was 622±82 nM (n=23), and the mean peak [Ca²⁺]_i mobilized by 10 nM mcPAF was 792±140 nM (n=4). These potency values correlated well with the actions of PAF on PI turnover described above. Since the effects of PAF and mcPAF on [Ca²⁺]_i mobilization in these cells were essentially identical, data obtained for the two compounds were pooled and analyzed together in the following studies.

Significant desensitization of the PAF-induced [Ca²⁺]_i mobilization responses was observed when the cells were treated repeatedly with PAF or mcPAF. Typically, the second treatment with PAF generated only 0%−20% of the response of the first treatment (data not shown). Extensive rinsing of the cells or a long waiting period (up to 30 min) between the agonist additions did not eliminate the desensitization phenomenon. Thus, the same cells were not used for multiple agonist stimulations.

To confirm that the [Ca²⁺]_i mobilizing effect of PAF was mediated by specific PAF receptors on HCECs, PCA-42481 (a PAF antagonist) was used to antagonize the PAF-induced response. Based on its blockade of PAF-activated PI turnover (Figure 1B), 10 µM PCA-42481 was expected to completely block the action of PAF or mcPAF in the HCECs. Indeed, pretreatment of 25 HCECs with 10 µM PCA-42481 followed by the addition of PAF (10 nM) resulted in minimal mobilization of [Ca²⁺]_i (Figure 4A), indicating effective blockade of the PAF receptor and its subsequent coupling to phospholipase C. PCA-42481 effectively and immediately antagonized the PAF-induced [Ca²⁺]_i mobilization when added after the initiation of the PAF-induced response in 14 cells (e.g., Figure 4B,C).

Since initial studies indicated that some HCECs responded to PAF by exhibiting an apparent oscillatory [Ca²⁺]_i mobilization pattern (Figure 2D), a more extensive evaluation of this phenomenon was investigated. Figure 5A depicts the temporal changes in [Ca²⁺]_i levels, which were visualized in pseudocolor, in three HCECs that were in close proximity of each other when exposed to 10 nM PAF. Followed over time, while cells #1 and #2 showed robust cyclic oscillatory increases and decreases in [Ca²⁺]_i (Figure 5B), cell #3 exhibited a relatively small increase in [Ca²⁺]_i at the onset but this declined over time and remained slightly elevated above basal levels (Figure 5B).

PAF-induced cytokine release

In a time-dependent manner, PAF significantly increased the release of IL-6 and IL-8. (Figure 6). For IL-6 and IL-8, the first phase of release was linear up to 8 h and the second phase was linear up to 24 h following the addition of PAF (Figure 6). Quantitative determination of cytokines released after 24 h following the addition of PAF yielded the following combined data from several experiments (pg/ml): basal IL-6=151±33, +PAF (1 µM)=415±78; basal IL-8=479±77, +PAF (1 µM)=1403±220; basal GM-CSF=44±19, +PAF (1 µM)=156±42 (1.4–3.5 fold above basal; all mean±SEM from four to six experiments using HCECs from four to six different donors; Figure 7). The estimated potency of PAF in stimulating the secretion of these cytokines using cells from different donors over the 24 h stimulation period was 108±48 nM (n=6; Figure 7).

It was noticeable that treatment of HCECs with 10 µM PAF for 24 h resulted in significantly reduced cytokine release relative to that induced by 10 nM–1 µM PAF (Figure 7). Trypan blue exclusion experiments revealed that the 24 h exposure of HCECs to 10 µM PAF caused a total loss of cell viability, and thus, this concentration of PAF was toxic over this period of stimulation. However, the mechanism of cell death induced by 10 µM PAF was not studied here. This profile of results matched those pertaining to the effects of PAF at different concentrations on PI hydrolysis in HCECs. High concentrations of PAF produce a lower response, presumably due to the toxic effects of PAF (Figure 1).

The PAF-induced (1 µM) secretion of IL-6, IL-8, and GM-CSF from the HCECs was antagonized by CV-3988 and PCA-42481 in a concentration-dependent (1–10 µM) manner in two separate experiments using HCECs from two different donors (Figure 8). The potencies (IC₅₀s) of CV-3988 and PCA-42481 for blocking the PAF-induced cytokine release from HCECs were 4.6–8.6 µM (p<0.05–0.001).