Main

PAF 1-O-alkyl-2-acetyl-sn glycero-3-phosphocholine is a phospholipid intra- and intercellular mediator implicated in the pathology of intestinal diseases such as NEC (1) and IBD (2). It has been shown that PAF levels are elevated in infants with NEC (3). In rats, exogenous PAF given intravenously results in ischemic bowel necrosis (4). Furthermore, studies using a rat model of NEC have shown that enteral administration of PAFR antagonists reduces the incidence of NEC (5, 6). Recently, elevated PAF levels have been found in stool samples of patients with IBD (2). Correlating with these findings, decreased levels of the PAF degrading enzyme, PAF-AH, have been observed in the distal ileum and serum samples of patients suffering from Crohn's disease (7). In addition, lexipafant, a PAFR antagonist significantly reduced mucosal damage in an experimental model of colitis (8).

Once activated, PAF initiates production of other inflammatory mediators such as tumor necrosis factor-α, prostaglandins, thromboxane, and complement. It is unknown, however, whether the pathologic consequences of elevated PAF levels in the intestine are only the result of the initiation of an inflammatory cascade, or whether PAF itself may have a direct effect on intestinal epithelial cells. Other inflammatory mediators have been shown to alter critical hallmark functions of the intestinal epithelium, e.g. ion transport (9, 10). Studies of PAF on tissue explants have also implied PAF induced ion transport. Notably, isolated rat jejunum (11) as well as human, rabbit, and rat colon have all demonstrated PAF-stimulated Cl secretion (1214).

However, despite extensive investigation of the effects of PAF on tissue and organ systems, specific effects at the intestinal epithelial cell level are poorly understood. It is difficult to extrapolate a direct PAF effect on epithelial cells from studies on tissue specimens because tissue samples incorporate multiple cell types with many interactions. PAF has been shown to stimulate Cl secretion in human and rat colonic explants, but these effects were blocked by indomethacin and by a thromboxane A2 synthase inhibitor, implying a significant role for intermediary mediators (12, 13). Thus, it is unclear at present whether PAF-induced transepithelial Cl transport in intact intestine represents a direct interaction of PAF with intestinal epithelial cells.

PAFR are abundantly expressed in the epithelial cells of the small intestine and colon, as well as in cultured epithelial cells lines (15). To examine the effect of PAF on ion transport, specifically in intestinal epithelial cells, we mounted HT29-CL19A human colonic adenocarcinoma cell monolayers in Ussing chambers and measured PAF-induced transepithelial ion transport. Specifically, we determined that PAF directly stimulates transcellular active ion transport, documented PAFR expression and apical polarity in epithelial cell monolayers, and identified Cl as the charge carrier for PAF-induced ion transport.

MATERIALS AND METHODS

Materials.

DIDS was obtained from Calbiochem (San Diego, CA, U.S.A.). PAF, glybenclamide, amphotericin B, 8-(4-Chlorophenylthio) adenosine 3′: 5′-cyclic monophosphate (cpt-cAMP), indomethacin, and the MAb to the secretory piece of the human polymeric immunoglobulin receptor were obtained from Sigma Chemical (St. Louis, MO, U.S.A.). CV-6209 and octylonium bromide were from Biomol (Plymouth Meeting, PA, U.S.A.). DPC was obtained from Aldrich (St. Louis, MO, U.S.A.). The polyclonal antibody to human PAFR was from Cayman Chemical (Ann Arbor, MI, U.S.A.), anti-rabbit IgG-FITC was from DAKO (Carpinteria, CA, U.S.A.), and streptavidin-HRP conjugate was from Pierce Chemical (Rockford, IL, U.S.A.). Carbamyl PAF was purchased from Alexis Corporation (San Diego, CA, U.S.A.). Human recombinant PAF-AH was a generous gift from Dr. Gary Peterman (ICOS Corporation, Seattle, WA, U.S.A.).

Epithelial cell culture.

HT29-CL19A cells originate from the subcloning of a human colonic adenocarcinoma cell line and display a highly differentiated phenotype, unlike the original HT29 cells (16). Cells were cultured in Dulbecco's modified Eagle media with 10% fetal bovine serum, penicillin 50 U/mL, and streptomycin 50 μg/mL. Cells were grown at 37°C in a 5% CO2 atmosphere. Media was replaced every 48 h, and cells were passaged every 7 d. For Ussing chamber experiments, 200 μL of cell suspension containing 5 × 105 cells was plated on top of six-well Snapwell filters (12-mm diameter, 0.4-μm pore size, Corning Life Sciences, Acton, MA, U.S.A.). Culturing HT29-CL19A cells for 8–11 d resulted in the formation of a polarized epithelial monolayer with well-developed mucosal and serosal membranes. HT29-CL19A monolayers were studied when resistances were >250 Ω/cm2.

Isc and Rte measurements.

After 8–11 d of incubation, current and resistance measurements were obtained using an Ussing chamber system equipped with digital data acquisition software that also allows software control of clamp electronics (Physiologic Instruments, San Diego, CA, U.S.A.). A blank filter was initially inserted to make adjustments for fluid resistance and voltage asymmetry, and then filters with cell monolayers were placed between chambers filled with 3 mL of buffer solution. The standard physiologic buffer used contained 116 mM NaCl, 5.4 mM KCl, 0.4 mM MgCl2, 1.8 mM CaCl2, 5.5 mM Glucose, 26 mM NaHCO3, and 0.9 mM NaH2PO4. Isc was recorded in voltage clamp mode. Rte was evaluated using 10-mV pulses every 20 s. Isc recordings presented in the figures are representations of single readings and are the result of digital data acquisition of actual current before the voltage pulse; therefore, the current spikes elicited by the voltage pulses are not displayed. When transport was evaluated under asymmetric buffer ion conditions, compensation for bridges and fluid resistance was performed under identical asymmetric ion conditions. Isc and Rte were recorded and analyzed using the Acquire and Analyze program, (Physiologic Instruments). A positive current was defined as the movement of negative ions from the serosal to the mucosal compartment or the flow of positive ions in the opposite direction.

Identification of PAFR mRNA using Reverse-Transcriptase PCR and restriction digestion mapping.

Total RNA was extracted by RNA-extracting agent STAT-60 (Tel Test B, Friendswood, TX, U.S.A.), according to the manufacturer's instructions. RNA integrity was verified by electrophoresis in 1% agarose gel. For each sample, 3 μg total RNA was reverse transcribed using Moloney-murine leukemia virus reverse transcriptase (Invitrogen, Carlsbad, CA, U.S.A.). cDNA was amplified by 35 cycles of PCR with AmpliTaq DNA polymerase using sense 5′-TATAACCGCTTCCAGGCAGT-3′ and antisense 5′-GAAACAGTAGATAACAGGGTC-3′ primers, respectively (17). The size of the predicted PAFR PCR product is 549 bp. After PCR, the PAFR PCR products were digested with PuvII or Ava II restriction enzymes at 37°C for 60 min and were separated by electrophoresis on a 1.2% agarose gel containing 1:10,000 diluted SYBR Green I (QIAGEN, Valencia, CA, U.S.A.).

Cell-surface biotinylation, immunoprecipitation, and streptavidin blotting of PAFR.

Cell-surface biotinylation was performed using cells grown on Snapwell membranes as described for Ussing chamber experiments. Cells were washed in ice-cold PBSCM. Sulfo-NHS-LC-biotin (1 mM) in PBSCM was then added to either the mucosal or serosal chamber. After incubation for 30 min on ice, cells were washed in ice-cold PBSCM, and blocked with 100 mM glycine in PBSCM for 30 min on ice. For immunoprecipitation, cells were lysed for 30 min at 4°C in buffer containing 20 mM HEPES, 150 mM NaCl, 1 mM EDTA, and 1% Nonidet P-40 (pH 7.4) supplemented with 10 μg/mL aprotinin, 10 μg/mL leupeptin, and 10 μg/mL 4-(2-aminoethyl)-benzenesulfonyl fluoride. Lysates were then centrifuged at 16,000 ×g for 5 min, and supernatants were supplemented with 1/4 volume of 5× RIPA buffer [250 mM Tris (pH 7.5), 120 mM sodium deoxycholate, 600 mM NaCl, 5% Triton X-100 (vol./vol.) and 0.5% SDS]. After adding 0.5 μg anti-PAFR polyclonal antibody, 10 μL of anti-pIGR containing ascites fluid (basolateral biotinylation positive control), or an equal concentration of rabbit/mouse nonimmune IgG (negative control), lysates were incubated at 4°C for 1 h on a rotating shaker, followed by the addition of 20 μL protein G agarose (50% slurry) and incubation for an additional 1 h. Beads were collected with centrifugation at 1000 ×g for 2 min, and washed three times using 1× RIPA. Pellets were resuspended in equal volumes of reducing SDS sample buffer [60 mM Tris (pH 6.8), 2% SDS, 5% β-mercapto-ethanol, 0.1% bromphenol blue, 50% glycerol], heated for 5 min to 95°C, resolved on reducing SDS-PAGE, then transferred to polyvinylidene fluoride membranes (Bio-Rad, Cambridge, MA, U.S.A.). Membranes were overlaid with streptavidin–horse radish peroxidase conjugates (100 ng/mL), incubated for 1 h, then washed three times with 0.5% Tween 20 in Tris-buffered saline. Blots were visualized using the ECL Plus chemiluminescence reagent (Amersham Pharmacia Biotech, Oakville, ON, Canada) and a Storm 860 Phosphor Imager (Molecular Dynamics, Sunnyvale, CA, U.S.A.).

Statistics.

Results are presented as mean values ± SEM. For statistical analysis, t test or ANOVA with Tukey's multiple comparison tests was used. Significance was defined as p < 0.05. Isc values were analyzed and are presented as changes from the values measured immediately before the respective treatment, or ΔIsc.

RESULTS

Comparison of PAF- and cAMP-induced Isc.

HT29-CL19A monolayers exhibited a 7.9 ± 0.2 μA/cm2 baseline Isc. Monolayers were treated with PAF (10 μM) in sequence with cpt-cAMP (300 μM), a cell-permeable cAMP analogue. The concentration of cpt-cAMP was chosen based on previous dose response data (not shown) to determine whether PAF elicits additional current after maximum activation by cAMP, the best-characterized regulator of ion transport in this cell line (18, 19). The order was varied to see whether there was any cumulative effect (Fig. 1a). PAF (10 μM) alone stimulated a ΔIsc to a maximum of 22.0 ± 2.8 μA/cm2, followed by decay to a stable 4.5 ± 0.7 μA/cm2 ΔIsc, which persisted for >30 min. PAF added after maximally activated cAMP-induced current resulted in an additional 17.7 ± 3.3 μA/cm2 current. A chloride channel blocker, glybenclamide (500 μM), added to the mucosal side returned Isc to the baseline level, suggesting that Cl is the principal charge carrier in these effects (n = 6–10).

Figure 1
figure 1

Regulation of Isc by PAF in HT29-CL19A cells. HT29-CL19A cells were mounted in Ussing chambers and Isc was recorded. Panel a shows representative current traces of Isc after sequential addition of either 10 μM PAF followed by 300 μM cAMP (thick line) or 300 μM cAMP followed by 10 μM PAF (thin line) and finished with the addition of 500 μM glybenclamide. Arrows indicate the time of treatments. Panel b depicts ΔIsc (max) as a function of the PAF concentration applied to filter-grown HT29-CL19A cells mounted in Ussing chambers. Curve was fitted to the data points using the Michaelis-Menten equation. Each data point represents mean ± SEM for n = 4–10 experiments. Calculated Ks = 46 μM, R2 = 0.91.

The dose dependence of PAF-stimulated Isc.

Previously published reports on PAF-mediated changes in various cell types demonstrate two very distinct ranges of dose responses. Effective doses in the 1–100 μM range have been observed primarily in epithelial tissues (2023), whereas other cell lines exhibit several orders of magnitude higher sensitivity to PAF, in the 10–500 nM range (2426). To correlate our observations with these earlier studies, we evaluated the dose response relationship of PAF-elicited Isc in HT29-CL19A cells. PAF regulated ion transport at doses ranging from 0.1 to 200 μM in this cell line (Fig. 1b). Vehicle alone (ethanol) had no effect on Isc (not shown). Curve fitting to the results of dose response observations using the Michaelis-Menten equation revealed a Ks for PAF-induced Isc of 46 μM (R2 = 0.91).

Identification of the charge carrier responsible for PAF currents.

To investigate more thoroughly the charge carrier associated with PAF-induced currents, we influenced the driving forces for chloride in our system, and evaluated the effects of chloride channel inhibitors on the current. Low-chloride (7.6 mM) buffer was prepared by substituting sodium gluconate for NaCl in the standard physiologic buffer. The effect of PAF on Isc was measured with or without permeabilization of the basolateral membrane with 100 μM amphotericin B for 30 min before the addition of other agents, and in the presence of a chloride gradient across the monolayer. The chloride gradient was created by placing low Cl buffer in either the serosal or the mucosal chamber, while maintaining physiologic Cl concentration in the opposing chamber. PAF added in the presence of serosal amphotericin and a serosal to mucosal Cl gradient resulted in a 30.5 ± 2.4 μA/cm2 ΔIsc, whereas reversal of the gradient in the presence of serosal amphotericin resulted in a reversed current −13.5 ± 1.7 μA/cm2 (Fig. 2a). To ensure that the PAF-induced current was not due to changes in paracellular conductance, the effect of PAF on Isc was also evaluated without permeabilization of the basolateral membrane. As depicted by the dashed line in Figure 2a, PAF did not elicit any detectable change in Isc (−0.47 ± 0.05 μA/cm2, n = 6) when low Cl buffer was placed in the serosal chamber without permeabilization of the basolateral membrane, indicating that PAF-induced current is due to transcellular, and not paracellular ion transport. The Cl channel blockers glybenclamide (500 μM), DPC (500 μM), and DIDS (600 μM) blocked PAF-induced stable ΔIsc by 81 ± 8%, 80 ± 9%, and 80.6 ± 9.2%, respectively (Fig. 2b, all significant at p < 0.05, n = 4). All three inhibitors were added only in the mucosal chamber, with simultaneous addition of solvent to the opposing chamber to avoid an effect on basolateral transporters. Solvents added to both sides had no significant effect, and none of the channel blockers affected baseline currents (data not shown).

Figure 2
figure 2

Identification of the charge carrier for PAF-stimulated Isc. Filter-grown HT29-CL19A cells were mounted in Ussing chambers and Isc was recorded. Panel a shows Isc tracings from a representative experiment comparing PAF stimulated Isc after permeabilization of the serosal plasma membrane in the presence of a serosal to mucosal Cl gradient across the monolayer (thick line) or a mucosal to serosal gradient (thin line). Dashed line illustrates a lack of PAF effect in the presence of a mucosal to serosal Cl gradient when the basolateral membrane was not permeabilized by amphotericin B. Arrow indicates time of PAF addition. Panel b shows representative Isc tracings depicting the effect of the Cl channel blockers DPC (500 μM), glybenclamide (500 μM), and DIDS (600 μM) on PAF-induced ΔIsc after stabilization of current. Arrow indicates time of addition of inhibitors.

PAF stimulates Isc directly and specifically in HT29-CL19A cells.

PAF can be metabolized in various ways, and its by-products may affect ion transport. HT29-CL19A monolayers were treated with the PAF analogue carbamyl PAF (27) or with PAF in the presence of the degrading enzyme PAF-AH (28) to confirm a direct and specific effect of PAF on Isc. Carbamyl PAF at 10 μM elicited 27.6 ± 2.6 μA/cm2 ΔIsc, whereas including PAF-AH in the chamber before adding PAF blocked PAF-induced ΔIsc to 2.6 ± 1.1 μA/cm2 (p < 0.05, n = 4).

The stimulation of Cl transport by PAF is not mediated by prostaglandin synthesis.

Earlier studies indicating that PAF applied to the serosal side of intestinal explants stimulates Cl transport found that this stimulatory effect required prostaglandin synthesis, and, therefore, was blocked by indomethacin. Although the stimulation of transepithelial Cl transport by PAF exhibited very different characteristics, to further differentiate the direct mucosal activation of Isc by PAF from the previously described serosal indirect effect, we evaluated the potential role of prostaglandin synthesis. After pretreatment and the subsequent continuous presence of 10 μM indomethacin, both the time course and magnitude of PAF-stimulated Isc was indistinguishable from Isc in the absence of indomethacin (Fig. 3, a and b). Interestingly, indomethacin caused a small but significant increase in the cAMP-stimulated Isc.

Figure 3
figure 3

Prostaglandin synthesis is not required for the induction of Cl transport by PAF from the mucosal side of the epithelial monolayer. Filter-grown HT29-CL19A monolayers were mounted in Ussing chambers. Panel a depicts a representative Isc trace in which cells were treated with 10 μM indomethacin to block prostaglandin synthesis, Isc was recorded and PAF (20 μM) was added followed by cAMP (300 μM) as indicated by arrows. Panel b illustrates the mean ± SEM change of Isc elicited by indomethacin treated, and untreated HT29-CL19A monolayers (n = 6). *p < 0.05.

The effect of PAFR antagonists on PAF-induced Isc.

Given that the sensitivity of PAF-dependent activation of ion transport in epithelial cells is strikingly different from the activation of PAFR in endometrial cells, T and B lymphocytes, and macrophages, we evaluated the effect of PAFR antagonists to confirm a specific PAFR-mediated effect. CV6209 (20 μM) and octylonium bromide (40 μM) almost completely abolished the stimulation of Isc by 20μM PAF (Fig. 4a). PAF alone induced 26.4 ± 3.0 μA/cm2 ΔIsc (n = 10), whereas pretreatment with CV6209 or octylonium bromide reduced PAF-induced ΔIsc to 4.0 ± 1.3 μA/cm2 and 4.7 ± 2.6 μA/cm2, respectively (p < 0.01). Neither receptor antagonist had a significant effect on cAMP-induced ion transport in HT29-CL19A cells, indicating a specific inhibitory effect on PAF-induced ion transport.

Figure 4
figure 4

Analysis of specificity and sidedness of PAF-induced Isc. Filter-grown HT29-CL19A monolayers were mounted in Ussing chambers. Panel a depicts a representative Isc trace in which cells were treated with 20 μM CV6209 or 40 μM octylonium bromide to desensitize PAFR, then Isc was recorded and PAF (20 μM) was added followed by cAMP (300 μM). Panel b: HT29-CL19A cells were mounted in Ussing chambers, and Isc was recorded. Arrow indicates the addition of 10 μM PAF into either the serosal chamber (thin line) or mucosal chamber (thick line).

PAF stimulates Isc only when added to the mucosal surface of HT29-CL19A monolayers.

To determine the polarity of PAF-induced Isc, PAF was added to either the mucosal or serosal chamber and resulting Isc changes were recorded and quantified. When PAF (10 μM) was only added to the mucosal side of the monolayer, mean ΔIsc was 35.0 ± 5.2 μA/cm2, whereas PAF (10 μM) added only to the serosal chamber resulted in no current (ΔIsc = 0.2 ± 0.1 μA/cm2, Fig. 4b).

Analysis of PAFR expression and polarity.

We next tested whether HT29-CL19A cells express the same PAFR identified in other cell types. Total RNA was isolated from cultured cells and analyzed by reverse-transcriptase (RT)-PCR using primers designed to hybridize with the known human PAFR. The authenticity of PCR products was verified using restriction digest mapping. Figure 5a illustrates that HT29-CL19A cells constitutively express PAFR transcripts as shown by amplification of the predicted 549 bp fragment by RT-PCR. The fragment was cut by Pvu II (417 bp and 132 bp), and Ava II (427 bp and 122 bp), as predicted based on the published sequence (29).

Figure 5
figure 5

Analysis of PAFR expression and polarity in HT29-CL19A cells. Panel a illustrates the results of RT-PCR and restriction digest mapping analysis of PAFR mRNA in HT29-CL19A cells. Total RNA was extracted from the HT29-CL19A cell line and amplified by RT-PCR for PAFR mRNA. After PCR, PAFR PCR products were digested with PuvII or Ava II restriction enzymes and separated by electrophoresis on a 1.2% agarose gel. The 549-bp band in the undigested sample, the 417-bp/132-bp fragments by Pvu II digestion and the 427-bp/122-bp fragments by Ava II digestion correspond to the sizes predicted based on the published sequence. Panel b: Domain-selective biotinylation, immunoprecipitation, and streptavidin blotting of PAFR. Filter-grown HT29-CL19A cells were biotinylated either from the mucosal or serosal direction, and then cell lysates were immunoprecipitated using a PAFR antibody and blotted with streptavidin. The image depicts a representative biotin-streptavidin blot of immunoprecipitated PAFR and corresponding unbiotinylated and/or nonimmune controls as indicated. The arrow points to the characteristic 68-kD PAFR band only observed in cells with domain-selective biotinylation of the mucosal surface. Panel c: Verification of successful biotinylation of basolateral proteins. Immunoprecipitation performed using a monoclonal anti pIGR. The characteristic 110-kD band was observed only in samples biotinylated from the serosal side. This experiment was repeated three times with similar results.

We then used domain-selective biotinylation to detect PAFR present on either the apical or basolateral plasma membrane. On streptavidin blots, the characteristic PAFR 68-kD band (15, 30) was detected from lysates of cells that were exposed to the biotinylating agent on their apical surface (Fig. 5b). The correct-size protein band was not detected from cell lysates that were not exposed to the biotinylating reagent, or when nonimmune IgG was used instead of the anti-PAFR antibody. In addition, this band was not detected from cells biotinylated from the basolateral side; only a faint lower molecular weight band was detected that was also present in the nonimmune control. Polymeric IGR was used as a positive control to confirm biotinylation of basolateral proteins across the supporting membrane (Fig. 5c).

DISCUSSION

Vectorial ion transport is one of the hallmark functions of intestinal epithelium. Na+ and Cl are the two solutes transported in the largest quantities, and disordered Cl secretion is the basis for several diseases. Exaggerated Cl secretion causes the severe secretory diarrhea associated with Vibrio cholerae and Escherichia coli toxins. Conversely, when Cl secretion is impaired, as in cystic fibrosis, an intestinal obstruction such as meconium ileus can result (31). In addition, Cl secretion has been thought to be a participant in innate mucosal defense mechanisms against bacteria (32).

A number of inflammatory mediators have been shown to regulate Cl transport either by regulating the activity of ion channels, or by regulating their gene expression. PAF, a phospholipid mediator that has been implicated in the pathology of NEC, IBD, and asthma (2, 3, 33), has been shown to activate Cl transport in both intestinal explants and airway epithelial cells. To date, direct effects of PAF on ion transport of epithelial monolayers have been demonstrated only in airway epithelial cells (20). The direct effect by PAF on airway epithelial cells has been demonstrated at micromolar concentrations and elicited from the mucosal side of the monolayers. The effect on intestinal transport has been thought to be mediated only indirectly via the stimulated production of mediators such as prostaglandins (12, 21) and thromboxanes (13). The indirect effect of PAF on intestinal explants has only been observed from the serosal side at nanomolar concentrations. Merendino et al.(15) have previously demonstrated PAFR expression in the intestinal epithelium, and several studies have documented increased stool PAF content in patients with IBD compared with disease-free controls. Given the high-level expression of PAFR in the intestinal epithelium, we hypothesized that, in addition to previously documented indirect effects, PAF would have a direct effect on intestinal epithelial cells.

The purpose of our study was to characterize the role of PAF in the direct regulation of intestinal epithelial cell ion transport. We performed our experiments in the HT29-CL19A cell line; a well-described transformed human colonic crypt cell model (16, 34). Our results demonstrate that PAF stimulates Cl transport across monolayers of HT29-CL19A cells. PAF-induced Isc followed the direction of a Cl gradient when basolateral plasma membranes were permeabilized, and was inhibited by the Cl channel blockers glybenclamide, DPC, and DIDS. On the other hand, without the permeabilization of the basolateral plasma membrane and in the presence of a mucosal to serosal Cl gradient, PAF did not elicit any current, excluding the possibility that increased paracellular permeability is responsible for the observed current. All three Cl channel blockers inhibited PAF-induced Isc, even when only added to the mucosal compartment and in cells with permeabilized basolateral membranes, indicating that this inhibition was not a result of interference with basolateral loading mechanisms. The elicited PAF Isc was not affected by indomethacin, confirming a direct effect not requiring prostaglandin synthesis. Thus, our results indicate that PAF-induced Isc results from the opening of a chloride channel in the apical cell membrane, and represents vectorial Cl transport in a serosal to mucosal direction. Several Cl channels have been identified in intestinal cells and the precise identification of the ion channel opened after activation by PAF was beyond the scope of this study. Further studies using techniques such as patch clamping and single-channel analysis may be helpful in identifying the specific Cl channel opened by PAF.

All of our observations suggest that in HT29-CL19A cells, PAF activates Cl transport via a specific receptor-mediated effect. We confirmed the presence of PAFR mRNA in HT29-CL19A cells by RT-PCR and the PAFR protein by immunoprecipitation and streptavidin blotting of the biotinylated receptor. The effect of PAF on Isc was mimicked by carbamyl-PAF, a nonmetabolizable analogue of PAF, was abolished by PAF-AH, a PAF degrading enzyme, and was inhibited by the PAFR antagonists CV6209 and octylonium bromide. The PAFR antagonists did not have an effect on cAMP-induced Isc, verifying the specificity of their effect. Furthermore, the observed apical polarity of the PAFR protein was consistent with the functional effect, i.e. PAF activated Isc only from the mucosal chamber.

Yet, as previously mentioned, significant differences exist between PAF-induced effects in epithelial cells compared with PAFR activation in other cell types. We observed stimulation of ion transport by PAF at micromolar concentrations in cultured intestinal epithelial cells similar to that seen in cultured bronchial epithelial cells (20), but different from the high-affinity activation identified in Chinese hamster ovary cells transfected with PAFR, endometrial cells, T and B lymphocytes, primary hippocampal neurons, macrophages (2426, 35), and the indirect serosal effects observed in rat and human tissue explants (12, 13). The mechanism reducing the affinity of PAFR to its ligand in the apical plasma membranes of intestinal and airway epithelial cells compared with other nonepithelial cells is unknown. Experimental evidence suggests that G proteins or membrane lipids can modify the affinity of PAF to PAFR. When the receptor is bound to heterotrimeric G proteins, it exhibits high PAF affinity, but this affinity is reduced when the G proteins dissociate (36). Also, treatment of PAFR-expressing cells with polyunsaturated fatty acids that alter the lipid composition of extracted membranes changes the binding affinity of PAF to PAFR (37). Therefore, either the profile of associated heterotrimeric G proteins or the surrounding lipids could influence the affinity of PAFR to PAF in different cell types. Alternatively, a second, low-affinity binding site might exist on the PAFR that could explain the effects observed exclusively in cultured epithelial cells.

The in vivo relevance of low-affinity PAFR on the intestinal mucosal surface may be controversial. Although there is no information available regarding the exact concentrations of PAF at the mucosal surfaces of intestinal epithelial cells, it is likely that relatively high concentrations might exist, especially under pathologic conditions. The two key enzymes involved in the biosynthesis of PAF, the secretory form of type II phospholipase A2 and PAF acetyltransferase, have both been shown to be expressed by intestinal epithelial cells (38, 39). Type II phospholipase A2 is expressed in Paneth cells, and its expression is increased in ulcerative colitis (38). Indeed, cultured colonic biopsies from ulcerative colitis patients have been shown to produce PAF (40) at 0.8 nmol/g(tissue)/h rates. In another study, tissue concentrations of PAF in micromolar concentrations were found in mucosal scrapings of biopsy specimens from patients with ulcerative colitis (41). Escherichia coli and other bacteria that are commonly found in the gastrointestinal tract are also extremely capable of producing PAF (42, 43). PAF can be readily detected in stool from patients with ulcerative colitis and Crohn's disease (2). Considering that the PAF-degrading enzyme PAF acetylhydrolase is present in the intestinal lumen, any detectable PAF in the stool suggests potentially much higher concentrations upstream in the intestinal tract. Therefore, a low-affinity PAFR in the luminal plasma membrane of epithelial cells might indicate an adaptation specifically designed to be activated only under potentially high levels of PAF that might exist under pathologic conditions in the intestinal lumen.

In conclusion, PAF directly and specifically stimulates Cl transport in intestinal epithelial cells through the opening of an apical chloride channel. This direct stimulation of Cl transport can be elicited only by mucosal application of PAF, and the polarity of the functional effect correlates with our observation that PAFR is expressed in the apical domain of polarized HT29-CL19A cells. PAF-induced Cl transport may represent a physiologic function, or may be related to PAF-induced mucosal injury. Further studies of the direct interactions of epithelial cells with PAF leading to a better understanding of the physiologic and pathologic roles of PAF in mucosal homeostasis may elucidate mechanisms of intestinal injury resulting from altered PAF metabolism.