Apical and basolateral pools of proteinase-activated receptor-2 direct distinct signaling events in the intestinal epithelium

Abstract

Studies suggest that there are two distinct pools of proteinase-activated receptor-2 (PAR₂) present in intestinal epithelial cells: an apical pool accessible from the lumen, and a basolateral pool accessible from the interstitial space and blood. Although introduction of PAR₂ agonists such as 2-furoyl-LIGRL-O-NH₂ (2fAP) to the intestinal lumen can activate PAR₂, the presence of accessible apical PAR₂ has not been definitively shown. Furthermore, some studies have suggested that basolateral PAR₂ responses in the intestinal epithelium are mediated indirectly by neuropeptides released from enteric nerve fibers, rather than by intestinal PAR₂ itself. Here we identified accessible pools of both apical and basolateral PAR₂ in cultured Caco2-BBe monolayers and in mouse ileum. Activation of basolateral PAR₂ transiently increased short-circuit current by activating electrogenic Cl⁻ secretion, promoted dephosphorylation of the actin filament-severing protein, cofilin, and activated the transcription factor, AP-1, whereas apical PAR₂ did not. In contrast, both pools of PAR₂ activated extracellular signal-regulated kinase 1/2 (ERK1/2) via temporally and mechanistically distinct pathways. Apical PAR₂ promoted a rapid, biphasic PLCβ/Ca²⁺/PKC-dependent ERK1/2 activation, resulting in nuclear localization, whereas basolateral PAR₂ promoted delayed ERK1/2 activation which was predominantly restricted to the cytosol, involving both PLCβ/Ca²⁺ and β-arrestin-dependent pathways. These results suggest that the outcome of PAR₂ activation is dependent on the specific receptor pool that is activated, allowing for fine-tuning of the physiological responses to different agonists.

proteinase-activated receptor-2 (PAR₂) is a G protein-coupled receptor (GPCR) that is activated via cleavage of its NH₂ terminus by serine proteinases including trypsin, mast cell tryptase, kallikreins, membrane-type serine proteinase 1, and the tissue factor VIIa-Xa complex (5, 14, 17, 29, 37, 48). Proteolytic cleavage of the NH₂-terminal domain exposes a tethered ligand that binds and activates the receptor itself. PAR₂ is reported to trigger a wide variety of cellular responses (e.g., proliferation, chemotaxis, ion transport, epithelial barrier function, and tumor cell metastasis) in a cell type-specific manner (30, 40). In vitro, peptides corresponding to the tethered ligand (SLIGRL/KV-NH₂) and synthetic analogs of tethered ligand peptides, such as 2-furoyl-LIGRL-O-NH₂ (2fAP) can activate PAR₂ and trigger downstream signals in the absence of proteolytic cleavage (2, 30, 43). Previous studies have established that activation of PAR₂ can evoke both pro- and antiinflammatory responses in the intestine, including vasodilation, smooth muscle relaxation, cytokine upregulation, immune cell recruitment, and increased nociception and hyperalgesia (7, 8, 12, 23, 31–33).

Here we examined the roles of apical and basolateral PAR₂ in intestinal epithelial cells. Studies that used PAR₂-specific antibodies or conducted intracolonic administration of fluorescently conjugated 2fAP suggested that intestinal epithelial cells (both in vivo and in various cell culture lines) express PAR₂ on both the apical and basolateral surfaces (14, 30, 37). Furthermore, expression of basolateral PAR₂ at the perijunctional margin of human colonic mucosa was reported to be increased in patients with Crohn's disease, and its expression correlated with a neutrophil-mediated decrease in transepithelial resistance (12). However, the accessibility of each receptor pool to agonists and the corresponding downstream signaling and physiological responses, to either basolateral or apical receptor activation, have not yet been clarified. Studies suggest that basolateral PAR₂ can be activated by mast cell tryptase or proteinases released by recruited leukocytes (14, 33, 50), while apical receptor can be activated by trypsin (either pancreatic or secreted by the enterocytes) or by luminal proteinases released by invading pathogens (12, 18, 28, 33, 34, 37). Many of these studies suggest that inflammatory responses in the intestine can be induced by luminal introduction of PAR₂-activating peptides and by serosal exposure to mast cell tryptase. While the specific cellular responses such as ion secretion and changes in transepithelial electrical resistance (R_T) have been reported in response to basolaterally administered agonist in cultured cells (19, 41, 45, 51–53), little is known about signaling via apical PAR₂. Furthermore, the possibility that there are distinctions in signaling between the apical and basolateral receptor populations has not yet been addressed. This is a particularly pertinent question given the recent evidence that PAR₂ is capable of signaling through multiple independent pathways, involving G protein-dependent and β-arrestin-dependent events, leading to a diverse array of physiological responses (44, 54, 55, 57, 58). Here we examine the following: 1) whether apical and basolateral PAR₂ are accessible to labeled 2fAP in polarized intestinal epithelial monolayers (Caco2-BBe); 2) whether there are temporal and mechanistic distinctions in apical and basolateral PAR₂ signaling; and 3) whether either pool uses β-arrestin-dependent signaling pathways.

MATERIAL AND METHODS

All chemicals were from Sigma unless otherwise stated. Antibodies and final dilutions for Western blot (WB) and immunofluorescence (IF) analysis were as follows: from BD Biosciences, mouse monoclonal anti-cofilin (1:1,000 for WB) and mouse monoclonal anti-zona occludin 1 (ZO-1) (1:500 IF); from Cell Signaling, rabbit anti-phospho (Ser3)-cofilin (1:1,000 for WB); rabbit anti-phospho (Thr202/Tyr204)-ERK (1:1,000 for WB, 1:500 for IF); and mouse anti-ERK1/2 (1:1,000 for WB). Rabbit anti-PAR₂ 9717 (1:300 for IF), raised against a peptide corresponding to the COOH terminus of mouse PAR₂ (SVKTSY) and characterized previously (14), was obtained from the Center for Ulcer Research and Education (University of California Los Angeles). Alexa Fluor 488-labeled phalloidin (1:20 IF) was from Invitrogen. Activating peptides 2fAP (2-furoyl-LIGRL-ornithine-NH₂), rhodamine-labeled 2fAP and reverse 2fAP (2-furoyl-LRGIL-ornithine-NH₂) were synthesized by Genemed (South San Francisco, CA). Pharmacological inhibitors were used as follows: Ca²⁺ chelator BAPTA-AM (Sigma) was prepared in dimethyl formamide, diluted to 10 mM in DMSO and used at 30 μM final concentration; PLC inhibitor U73122 (Tocris) was solubilized in chloroform, diluted to 100 mM in DMSO immediately before use at 1 μM final concentration; PKC inhibitor GF109203X (Tocris) was solubilized at 25 mM in DMSO and used at 1 μM final concentration. Dominant-negative β-arrestin-1, corresponding to amino acids 319–418, was a gift from Dr. Robert Lefkowitz (Duke University Medical School, Durham, NC).

Transfection and cell lines.

A human intestinal colon carcinoma cell line, selected for brush border formation (Caco2-BBe), was obtained from Dr. David Lo (University of California, Riverside) and grown in advanced Dulbecco's modified Eagle's medium, 15 mM HEPES, and 10% fetal calf serum. Transient transfections were performed on 95% confluent cells using FuGENE6 (Roche), and experiments were performed between 48 and 72 h after transfection. For other experiments, Caco2-BBe cells stably transfected with a luciferase reporter construct in which luciferase transcription is controlled by three AP-1 binding sites and a TATA box cloned upstream from the luciferase gene (42). To accomplish this, the AP-1-luciferase reporter was cotransfected into Caco2-BBe cells along with pRSVneo. The transfected cells were cultured in medium containing 400 μg/ml G418, and neomycin-resistant clones were pooled to create a cell line stably transfected with the reporter. The transfection and assays of luciferase activity were performed as described previously (39).

PAR₂ labeling with rhodamine-2fAP.

Caco2-BBe cells were seeded on permeable supports (Costar Snapwell, 24 well, 0.4 μm pore) at 2 × 10⁴ cells per well and grown for 10–14 days to confluence. After the monolayers attained a resistance of 750 Ω/cm² (Epithelial Voltohmmeter, WPI, Sarasota, FL), they were exposed to 1 μM rhodamine-labeled 2fAP or RAP at 4°C for 1 h, or at 4°C for 1 h then 37°C for 1 h. The cells were rinsed three times with ice-cold sterile phosphate-buffered saline solution (PBS) to remove unbound label, and then fixed in neutral buffered formalin (Fisher) on ice for 1 h. Fixed cells were labeled with Alexa Fluor 488-phalloidin.

Protein analysis and Western blotting.

Caco2-BBe cells were seeded on permeable supports (Costar Snapwell, 6 well, 0.4 μm pore) at 5 × 10⁴ cells per well, then starved of serum for 24 h. Where indicated, inhibitors were added to both apical and basolateral compartments at the concentrations indicated above for 10 min. Vehicle controls were preincubated with DMSO alone. Monolayers were exposed to 1 μM 2fAP on the apical or basolateral surface for 0–60 min at 37°C, then solubilized in Laemmli sample buffer [62.5 mM Tris·HCl (pH 6.8), 2% (wt/vol) sodium dodecyl sulfate, 50 mM dithiothreitol, and 0.01% (wt/vol) bromophenol blue]. Protein was analyzed by SDS-PAGE (15% for cofilin, 10% for ERK1/2) followed by WB. Blots were imaged using the LICOR Odyssey imaging system, and LICOR software was used to calculate integrated intensities of bands. Images of WB were assembled using Adobe Photoshop 5.0 and Adobe Illustrator CS4. Some gels were spliced to eliminate blank lanes or lanes containing samples unrelated to the figure.

Electrical measurements.

The short-circuit current (I_sc) and R_T across confluent Caco2-BBe monolayers grown on permeable supports (Costar Snapwell, 6 well, 0.4 μm pore) were measured using a conventional Ussing chamber technique, as described previously (3). Snapwell inserts were incubated in chambers (EM-CSYS-2, Physiologic Instruments) containing Tyrode solution (140 mM NaCl, 4 mM KCl, 1 mM MgCl₂, 1.25 mM CaCl₂, 12 mM glucose, and 10 mM Na-HEPES, pH 7.4) mixed continuously by gas (100% O₂) lift. Chambers were maintained at 37°C by heated water jackets. Once the resting I_sc stabilized (10 min), the changes in I_sc and R_T after sequential addition of 1 μM of 2fAP to the luminal and serosal compartments were recorded.

Microscopy.

Confluent Caco2-BBe monolayers were fixed and imaged as described previously (26). Serial sections were taken on Zeiss LSM-510 using ×63 or ×100 objectives. Three-dimensional reconstructions of the images were done using Imaris 6.3 software.

Immunolabeling of PAR₂ in ileum.

Four-month-old female wild-type C57Bl6 mice (Jackson Labs) and PAR₂^−/− mice in a C57Bl6 background (from Dr. Robin Plevin, University of Strathclyde, Glasgow, UK) were used for immunofluorescent labeling. All animal protocols were approved by the Institutional Animal Care and Use Committee of University of California, Riverside. Mice were euthanized, and short segments of mouse ileum were fixed in ice-cold formalin for 5 h, infiltrated with cryoprotectant (30% sucrose in PBS) overnight, and frozen in OTC medium (Triangle Biomedical Sciences) at −35°C. Sections of 10 μm thickness were cut on a cryostat microtome (Microm) and mounted on polylysine-coated glass slides (Fisher Superfrost Plus). Sections were incubated sequentially with blocking solution (1 h), primary antibody (overnight at 4°C), and secondary antibodies conjugated to Alexa Fluor-488 and/or -546 (Molecular Probes). Antibody-9717 against mouse PAR₂ proved suitable for use with undenatured sections but not with sections subjected to antigen unmasking with either heat or SDS. When colabeling with mouse monoclonal antibody against ZO-1, endogenous mouse IgG was blocked by preincubation in goat anti-mouse IgG for 1 h. Confocal images were acquired with a Zeiss LSM-510 microscope and assembled using Adobe Photoshop.

Barbed end labeling.

Caco2 monolayers plated on collagen-coated Transwell inserts, or mouse ileum sections were prepared as described above and treated with 2fAP for 5 min. 2fAP administration was restricted to the basolateral surface of Caco2 monolayers as described above. It was not possible to restrict 2fAP administration to the basolateral surface of mouse ileum, so tissue sections were bathed in 2fAP containing solution. Media were immediately exchanged for actin monomer buffer [20 mM HEPES, pH 7.5, 138 mM KCl, 4 mM MgCl₂, 3 mM EGTA, 0.2 mg/ml of saponin, 1 mM ATP, 1% BSA, and 0.3 μM rhodamine-labeled G-actin (from Cytoskeleton, Boulder, CO)], and cells were incubated for 1 min to allow incorporation of labeled monomers into cells. Cells/tissue were washed in 0.1 M glycine-PBS for 10 min, then fixed, stained, and imaged as described in Microscopy.

Data and statistical analysis.

All graphs and statistical analysis were performed using Prism or Microsoft Excel 2003. All experiments were performed a minimum of three times. Phosphoprotein levels were normalized to total protein levels before calculation of fold change with respect to untreated controls. One-way ANOVA and two-tailed Tukey t-tests were used to determine statistical significance of and significant differences between values under different conditions.

RESULTS

Expression of accessible apical and basolateral PAR₂ in polarized Caco2-BBe monolayers.

Serine proteinases released by neutrophils and mast cells can activate PAR₂ in the intestinal epithelium evoking inflammation, and administration of PAR₂ agonists to the serosal side of epithelial monolayers promotes Ca²⁺ mobilization and ion secretion (15, 19, 33), suggesting that basolateral PAR₂ is also accessible to agonists in the interstitial space and blood. Studies have also shown that PAR₂ activating peptides administered intracolonically can bind PAR₂ and trigger inflammatory responses (7, 8, 30), suggesting apical PAR₂ is expressed on the cell surface and accessible to peptide agonist. However, it is formally possible that process of administering the peptide compromised the epithelium, allowing access of peptide agonists to the serosa. Thus, we examined the surface accessibility of apical and basolateral PAR₂ in polarized colonic epithelial cell monolayers (Caco2-BBe) grown on permeabilized supports. The integrity of the cell monolayer was confirmed by measuring R_T; only monolayers with R_T > 750 Ω/cm² were used in subsequent experiments. Rhodamine-labeled PAR₂ agonist peptide (2-furoyl-LIGRL, Rh-2fAP) was administered to either the apical or the basolateral compartment for 1 h, and cells were held at 0°C to prevent receptor trafficking, after which they were fixed and stained with antibody to ZO-1 to mark tight junctions. Confocal fluorescence microscopy revealed that apically administered Rh-2fAP bound the surface of the epithelial monolayers (Fig. 1, A and B), whereas an inactive control peptide (2-furoyl-LRGIL, Rh-RAP) did not, indicating that the signal is not due to nonspecific binding of labeled peptides (Fig. 1, C and D). No significant Rh-2fAP was detected in the lower chamber after apical administration, as determined by measurement of absorbance of media in a microplate reader (absorbance = 546 nm), confirming that no leakage across the monolayers occurred. Three-dimensional reconstruction of Z sections revealed that apical PAR₂ was localized in the same plane as the tight junctions, consistent with their presence on the apical surface (Fig. 1I). Since PAR₂ agonist peptides promote removal of receptor from the surface by clathrin-mediated receptor endocytosis (22), we next investigated whether the labeled peptide was internalized along with PAR₂. After monolayers were warmed to 37°C, Rh-2fAP (Fig. 1, E and F), but not Rh-RAP (Fig. 1, G and H), was observed in vesicles below the region demarcated by the ZO-1 (Fig. 1J). Basolaterally administered Rh-2fAP labeled PAR₂ along the basolateral membrane, a pool of receptor not labeled by apically applied agonist (Fig. 2, A–D, I). When cells were warmed to 37°C, bound Rh-2fAP peptide was internalized (Fig. 2, E–H, J). Interestingly, when the cells were warmed to 37°C in the presence of Rh-2fAP, ZO-1 redistributed from the tight junctions to an intracellular pool below the apical plane (Table 1). This effect on ZO-1 was not observed with the inactive control peptide (Fig. 2). We conclude that both apical and basolateral pools of PAR₂ exist, and both are expressed on the cell surface as they are both accessible to agonist. Furthermore, while basolateral PAR₂ activation promotes redistribution of tight junction proteins from the membrane to the cytoplasm, apical administration does not.

Fig. 1.

Labeling of apical proteinase-activated receptor-2 (PAR₂) with rhodamine-2-furoyl-LIGRL-O-NH₂ (Rh-2fAP) in polarized Caco2-BBe monolayers. A–D: Caco2-BBe monolayers were labeled with apically administered Rh-2fAP (A and B) or a rhodamine-labeled negative control peptide, 2-furoyl-LRGIL (Rh-RAP) (C and D). Cells were fixed and stained with antibody for tight junction protein zona occludin-1 (ZO-1) to demarcate the apical surface. E–H: to monitor internalization, Rh-2fAP (E and F) and Rh-RAP (G and H)-labeled cells were warmed to 37°C for 1 h. I and J: three-dimensional reconstruction of the microscopy images via Imaris software further shows labeling of apical PAR₂ with Rh-2fAP on the apical membrane surface (I) and PAR₂-induced internalization at 37°C (J).

Fig. 2.

Labeling of basolateral PAR₂ with Rh-2fAP in polarized Caco2-BBe cells. A–D: Caco2-BBe monolayers were labeled with basolaterally applied Rh-2fAP (A and B) and Rh-RAP (C and D) for 1 h at 4°C, fixed, and stained with anti-ZO-1 as described for Fig. 1. E–H: internalization of bound Rh-2fAP was monitored by warming cells, treated with Rh-2fAP (E and F) or Rh-RAP (G and H), to 37°C for 1 h. I and J: three-dimensional reconstruction of images of cells on ice (I) and at 37°C (J), via Imaris software.

Table 1.

Redistribution of zona occludin-1 by PAR₂

Treatment	ZO-1 Redistribution∗
RAP apical	492 ± 51
2fAP apical	364 ± 52
RAP basolateral	498 ± 92
2fAP basolateral	955 ± 100†‡

Values are means ± SE. Caco2-Bbe monolayers, treated with rhodamine 2-furoyl-LIGRL-O-NH₂ (2fAP) and maintained on ice or warmed to 37°C for 1 h, as described in Figs. 1 and 2, were analyzed by confocal microscopy, and three-dimensional images were reconstructed using Imaris software. The numbers of zona occludin-1 (ZO-1) fluorescent vesicles, >50 nM, were quantified, and the numbers of vesicles located 5 μm below the apical surface are presented. Measurements were taken from 10 images from 2 independent experiments. PAR₂, proteinase-activated receptor 2; RAP, 2-furoyl-LRGIL.

^∗Number of ZO-1-positive fluorescent spheres ≥5 μm below apical surface.

^†Significantly greater than RAP treated, P = 0.01;

^‡significantly greater than 2fAP apical, P = 0.005.

These experiments using cultured cells suggest that PAR₂ could be present at both the apical and basolateral surfaces in vivo. To evaluate this possibility, we examined the distribution of PAR₂ in mouse terminal ileum by confocal immunofluorescence microscopy. We observed labeling primarily at the brush border of villus enterocytes, and weaker labeling of the lateral membrane and submucosal nerve plexi (Fig. 3, left), confirming previous reports (16). No such labeling was observed in identically processed segments of ileum from PAR₂ knockout mice, confirming the specificity of antibody-9717 (Fig. 3, right).

Fig. 3.

Labeling of PAR₂ in proximal ileum. Isolated mouse proximal ileum from wild-type mice (left) and PAR₂^−/− mice (right) was fixed, sectioned, and stained with PAR₂ antibody 9717 (green) and TOPRO3 (blue) for detection of cell nuclei, and phalloidin (red) for detection of cell cytoskeleton and microvilli. Bottom: higher magnifications of boxed regions at top. (Arrows indicate apical and basolateral PAR₂.)

Basolateral PAR₂ activation evokes electrogenic Cl⁻ secretion in Caco2-BBe.

Studies on intestinal mucosa have demonstrated that basolateral, but not apical, activation of PAR₂ promotes Cl⁻ secretion in distal colonic mucosa and cultured cell lines (19, 51). We first confirmed this result in Caco2-BBe monolayers. As shown in Fig. 4, serosal addition of 2fAP evoked a transient I_sc response, whereas luminal addition did not. The I_sc response triggered by basolateral PAR₂ peaked after 40 s and averaged 14.5 ± 2.50 μA/cm² (Fig. 4C). Luminal 2fAP elicited a small but significant increase in R_T after a brief (∼1.5 min) delay, whereas serosal 2fAP produced a reduction in R_T that was (as expected) temporally correlated with the induced secretory current (Fig. 4B). We conclude that, in contrast to basolateral PAR₂, apical PAR₂ is not coupled to the Cl⁻ secretory mechanism.

Fig. 4.

Activation of basolateral PAR₂ induces an increase in short-circuit current (ΔI_sc) and a decrease in transepithelial resistance (R_T). Caco2-BBe monolayers were grown to confluence on Snapwell filters with 0.4-μm pores, mounted in a Ussing chamber, and allowed to equilibrate for 30 min. A change in current was evoked every 20 s with a 2-mV bipolar pulse to monitor transepithelial resistance. Short-circuit current was recorded after addition of 2fAP (1 μM), first to the apical (luminal) side, which generated no response, and then to the basolateral (serosal) side. A: representative tracing of the change in short-circuit current over time. B: transepithelial resistance was calculated for the bracketed regions in A, and mean values from 3 independent experiments were determined: 1, resting values; 2, values after apical PAR₂ addition; 3, values after basolateral PAR₂ addition; 4, values after transient short-circuit changes returned to baseline. *P < 0.05. C: graph depicting mean ± SE ΔI_sc over time after either serosal (○) or luminal (●) 2fAP treatment from 3 separate experiments.

Apical and basolateral PAR₂ promote temporally and mechanistically distinct pathways of ERK1/2 activation.

Studies dissecting the molecular mechanism of PAR₂-stimulated ion transport in colonic epithelial cells revealed a complex mechanism involving Gαq/Ca²⁺-dependent activation of selective PKC isoforms, transactivation of EGF receptor (EGFR), ERK1/2 activation leading to prostaglandin E2 (PGE₂) release and subsequent activation of the Gαs/cAMP-coupled PGE₂ receptor (51–53). EGFR transactivation and subsequent activation of ERK1/2 is a mechanism shared by other Ca²⁺-dependent Cl⁻ secretory pathways in the intestine (36) and PAR₂ is capable of activating ERK1/2 by multiple mechanisms, leading to distinct cellular responses (21, 25, 33, 38, 44). Whether all of these pathways exist sequentially or represent independent mechanisms for promoting Cl⁻ secretion is not clear. We first investigated whether both receptor pools were capable of activating ERK1/2 and whether they utilized the same signaling pathways to do so. After introduction of 2fAP to the apical or basolateral side of Caco2-BBe monolayers grown on permeabilized supports (Fig. 5, A and B, respectively), we observed a temporal difference in the activation of ERK1/2, as determined by Western blot analysis with anti-phospho-ERK. Apical administration of 2fAP promoted a biphasic increase in ERK1/2 phosphorylation, peaking at 5 min (4.86 ± 0.70-fold over baseline), decreasing at 30 min, and peaking again at 60 min (3.49 ± 0.36-fold over baseline). In contrast, basolateral 2fAP promoted a slow and prolonged ERK1/2 activation reaching 6.20 ± 0.96-fold change over baseline by 60 min. We conclude that both receptor pools are capable of signaling to the ERK1/2 pathway.

Fig. 5.

Temporally distinct ERK1/2 activation by apical and basolateral PAR₂. A and B: polarized Caco2-BBe monolayers were treated with 2fAP at either the apical (A) or the basolateral (B) surface for 0–60 min. Representative Western blot analyses performed with anti-phospho-ERK1/2 (pERK, top) and anti-total ERK2 (tERK, bottom) and imaged using the LICOR-Odyssey 2 color Infrared Imaging system are shown. Integrated intensities of each band were determined, pERK levels were normalized to tERK, and fold changes in normalized pERK in treated versus untreated samples were determined. C: bar graph depicting mean ± SE fold increase over baseline in normalized pERK. Baseline is defined as normalized pERK from untreated cells. Statistically significant increases in ERK1/2 activation were determined by ANOVA and differences between experimental groups were determined by two-tailed t-tests: *P < 0.04, **P < 0.005, n = 4.

To determine which signaling pathways were utilized by each receptor pool, we repeated ERK1/2 phosphorylation assays in the presence of inhibitors the classical Gαq signaling pathway which results in activation of PLCβ, generation of inositol triphosphate and diacylglycerol, mobilization of intracellular Ca²⁺, and activation of classical PKC isoforms. We used three different inhibitors of this pathway: PLCβ inhibitor U73122, calcium chelator BAPTA-AM, and the broad-spectrum PKC inhibitor GF109203X. ERK1/2 activation by apical PAR₂ was abolished by BAPTA-AM, U73122, and GF109203X. In contrast, basolateral ERK1/2 activation was insensitive to GF109203X inhibition but was inhibited by BAPTA-AM and U73122 by ∼50% (Fig. 6, A–C). Thus, both basolateral and apical PAR₂ can activate ERK1/2 through Gαq/Ca²⁺-mediated pathways, but only the apical PAR₂ pathway requires PKC and relies more heavily on the Ca²⁺ and PLCβ. PAR₂ can also promote ERK1/2 activation via G protein-independent pathways involving β-arrestins (21, 38, 47). To examine the contribution of β-arrestins to basolateral and apical PAR₂ signaling, cells were transfected with a dominant-negative mutant of β-arrestin-1 consisting of the clathrin-binding domain (amino acids 319–418). Transfection of dominant-negative β-arrestin (DNarr) inhibited ERK1/2 activation by basolaterally, but not apically, administered 2fAP, suggesting that only the basolateral PAR₂ pool required input from this pathway (Fig. 6D). A small increase in baseline pERK was also observed in resting DNarr-transfected cells, which may be due to the ability of β-arrestins to promote desensitization and internalization of many GPCRs present in epithelial cells, some of which may contribute to the baseline levels of ERK1/2. We conclude that PAR₂ uses multiple mechanisms for ERK1/2 activation in intestinal epithelial cells, and that the apical PAR₂ pool requires PLCβ, Ca²⁺, and PKC, while the basolateral PAR₂ requires PLCβ, Ca²⁺, and β-arrestins.

Fig. 6.

Distinct mechanisms of ERK1/2 activation by apical and basolateral PAR₂. A–C: polarized Caco2-BBe monolayers were treated with 2fAP at either the apical or the basolateral surface for 0–60 min after pretreatment with vehicle (DMSO) or with the following inhibitors: Ca²⁺-chelating agent BAPTA-AM (A), PLCβ inhibitor U73122 (B), or PKC inhibitor GF109203X (C). Top: representative Western blots. Bottom: bar graphs depicting mean ± SE fold change over baseline in normalized phospho-ERK. Apical administration is indicated by “A” and basolateral administration is indicated by “B” after the time for each blot. Baseline is defined as normalized pERK observed in vehicle-pretreated cells without 2fAP addition. D: monolayers were transfected with green fluorescent protein (GFP)-tagged dominant-negative β-arrestin-1 (DNARR-GFP) or GFP alone, and PAR₂-stimulated ERK1/2 phosphorylation was determined as described above. Bar graph depicts mean ± SE fold increase in ERK1/2 phosphorylation over baseline, which is defined as that observed in untreated, GFP-transfected cells. For all graphs, statistically significant increases in ERK1/2 phosphorylation over baseline or between bracketed groups (inhibitor vs. control treated) were determined by two-tailed t-tests: *P < 0.05, **P < 0.02, ***P < 0.005, n = 4. Lanes in A and B were spliced to remove blank lanes (A) and duplicate samples (B) as indicated by white spaces. E: monolayers were untreated (Unt) or treated with apically (AP) and basolaterally (BL) administered 2fAP for 60 min, fixed, and stained with anti-pERK and phalloidin. Arrows indicate nuclear pERK. Arrowheads indicate membrane pERK. F: Caco2-BBe cells were cultured until they were confluent monolayers on Snapwell supports in serum-containing medium. Cells were washed with PBS and transferred to serum-free medium for 16 h. The cells were then treated with PBS vehicle (control; CNT), 1 μM 2fAP apical, 1 μM 2fAP basolateral, or 500 nM phorbol 12-myristate 13-acetate (PMA) apical and basolateral. Six hours later, cell extracts were prepared and assayed for luciferase activity, expressed in relative light units (RLU)_. Each bar is the mean ± SE of results obtained with 4 different Snapwells. *Significantly higher than control, as determined by two-tailed t-tests, P < 0.05.

Previous studies have demonstrated that β-arrestin-dependent ERK1/2 activation by PAR₂ and other GPCRs results in sequestration of the activated ERK1/2 at the membrane, where it plays a role in actin cytoskeletal reorganization, rather than transcriptional and proliferative effects associated with nuclear ERK1/2 localization (20, 21, 49, 56). Thus, we predicted that differential activation of ERK1/2 observed with apical versus basolateral PAR₂ activation might also be associated with different localization patterns. Caco2-BBe monolayers were treated with 2fAP, administered either apically or basolaterally, for 60 min, fixed, and stained with anti-phospho-ERK1/2 and phalloidin (to visualize cell membrane). Upon apical PAR₂ activation, phosphorylated ERK1/2 was predominantly nuclear, while after basolateral addition, phosphorylated ERK1/2 was mostly retained at the membrane and diffusely localized throughout the cytoplasm (Fig. 6E). Thus, consistent with our previous observations, β-arrestin-dependent activation of ERK1/2 was associated with cytosolic/membrane sequestration. To test the possibility that basolateral and apical PAR₂ activation might differentially affect transcription, we examined the effect of 2fAP on activation of the transcription factor AP-1 in monolayers of Caco2-BBe, stably transfected with a reporter construct in which luciferase transcription is controlled by a promoter element consisting of three AP-1 binding sites and a TATA box (42). AP-1 is activated by several signaling pathways including ERK1/2 and JNK (46). Interestingly, activation of basolateral but not apical PAR₂ resulted in induction of the AP-1-luciferase reporter (Fig. 6F). Thus, the transcriptional effects resulting from ERK1/2 activation are likely to differ depending on whether apical or basolateral PAR₂ is activated.

Basolateral PAR₂ activation results in activation of cofilin, while apical PAR₂ activation does not.

One of the major pathways activated by β-arrestins downstream of PAR₂ is the dephosphorylation and activation of the actin filament-severing protein, cofilin (57, 58). Cofilin promotes rapid actin reorganization by severing existing actin filaments and creating free barbed ends to which actin monomers can spontaneously add. Since only basolateral PAR₂ required β-arrestins for ERK1/2 activation, we examined whether it also resulted in cofilin dephosphorylation. Consistent with the findings above, basolaterally administered 2fAP induced cofilin dephosphorylation after 5 and 60 min (Fig. 7, A and C) while apical 2fAP did not (Fig. 7, B and C). As was previously demonstrated, PAR₂-stimulated cofilin activation was β-arrestin dependent, as transfection of dominant-negative β-arrestin blocked cofilin dephosphorylation in response to basolateral PAR₂ activation (Fig. 7D). A major effect of cofilin actin filament-severing activity is the creation of actin free barbed ends, which promotes remodeling of the actin cytoskeleton. Barbed end formation can be monitored in cells by observing incorporation of labeled actin monomers into lightly permeabilized cells after agonist treatment. Actin monomers will not incorporate into stable filaments; thus they can be used to monitor sites of filament severing. Consistent with cofilin activation, we observe that actin monomers incorporate into cells upon basolateral, but not apical, PAR₂ activation in both cultured Caco2-BBe monolayers (Fig. 7E). Similar barbed incorporation, colocalizing with the tight junction marker ZO-1, was observed upon 2fAP treatment in isolated intestinal mucosa (Fig. 7F). We conclude that basolateral PAR₂ is capable of triggering β-arrestin-dependent and G protein-dependent signals, while apical PAR₂ does not trigger β-arrestin-dependent signaling.

Fig. 7.

Basolateral PAR₂ promotes cofilin phosphorylation. A and B: Caco2-BBe monolayers were treated as described in Fig. 4, and Western blots were analyzed with anti-phospho-cofilin (pcofilin, top) an anti-total-cofilin (tcofilin, bottom). Representative Western blots from cells activated with basolateral (A) or apical (B) 2fAP are shown. C: fold change over untreated was determined as described for phospho-ERK, and mean ± SE fold change over baseline is depicted in the bar graphs. Statistically significant results between treated and untreated cells are indicated or between apical or basolateral phospho-cofilin levels (*P < 0.05). D: monolayers were transfected with GFP-tagged dominant-negative β-arrestin-1 corresponding to amino acids 319–418, and cofilin dephosphorylation was determined as described above. Samples were run in triplicate, and images were spliced to remove repeated samples. E: Caco-2BBe monolayers were treated with basolateral 2fAP for 5 min, after which cells were lightly permeabilized and free actin barbed ends were labeled with rhodamine-actin monomers (red). Cells were fixed and stained with phalloidin (green) to highlight all cellular actin filaments, and images were collected with a Zeiss LSM510 confocal microscope, ×100 objective. Insets: side views of barbed end labeling at tight junctions are shown. Arrows indicate apical surface. F: intestinal mucosa was treated with or without 2fAP for 5 min, after which rhodamine actin labeling was performed as described in D. Tissue was fixed, frozen sections (10 μm each) were prepared and stained with phalloidin, and images were collected as described above.

DISCUSSION

Studies on PAR₂ in the intestine have suggested important roles for both apically and basolaterally accessible receptor. For example, luminal introduction of 2fAP to mice colon triggered inflammation, while upon injury and pathogen invasion, mast cells release tryptase to induce inflammation, which presumably involves basolateral and serosal PAR₂ (8, 12, 13, 23, 24, 30, 32, 35). The studies described here demonstrate the existence of spatially and temporally distinct signaling pathways in colonic epithelial cells by apical and basolateral pools of PAR₂. Both pools of receptor are accessible to agonist activation and promote activation of ERK1/2; however, the mechanisms of ERK1/2 activation differ between apical and basolateral receptors, with only basolateral PAR₂ utilizing a β-arrestin-dependent signaling component. Furthermore, only basolateral agonist promoted activation of the actin filament-severing protein, cofilin, and generation of free actin barbed ends, and redistribution of ZO-1 from the membrane, all processes that might be involved in modulation of tight junctions (Fig. 8). The use of different signaling mechanisms by apically and basolaterally exposed PAR₂ may allow for a distinct set of cellular events in response to serine proteinases present in the lumen or those released from immune cells present in the serosa at the basolateral surface.

Fig. 8.

Model for PAR₂ signaling from apical and basolateral surfaces. Apical and basolateral PAR₂ would be accessible to different proteolytic agonists, leading to distinct responses. Activation of both PAR₂ pools leads to activation of traditional Gαq (which leads to phosphatidylinositol 4,5-bisphosphate hydrolysis into inositol triphosphate and diacylglycerol), mobilization of intracellular Ca²⁺, and activation of PKC. ERK1/2 activation by basolateral PAR₂ is mediated by both the Gαq-dependent pathway and a β-arrestin-dependent pathway but is independent of PKC; the majority of ERK1/2 activated by this pathway does not translocate to the nucleus. PAR₂ activated at the apical surface activates ERK1/2 and requires typical Gαq-dependent signals such as Ca²⁺ and PKC, but it does not require β-arrestins. Basolateral but not apical PAR₂ is capable of promoting AP-1 activation, which may occur indirectly via phosphorylation of a cytoplasmic protein kinase, such as p90^Rsk. Transcriptional targets of apical PAR₂ have not yet been identified but likely involve nuclear ERK1/2. Basolateral but not apical PAR₂ stimulated short-circuit current changes, which have been reported to involve Ca²⁺, ERK1/2, and transactivation of EGF receptor. Basolateral but not apical PAR₂ also activates cofilin via a β-arrestin-dependent-pathway, leading to generation of free actin barbed ends at tight junctions (TJ).

The precise role of epithelial PAR₂ in intestinal inflammation is still unclear, and the identification of separate signaling pathways by different pools of receptor is important for the ultimate understanding of PAR₂-induced inflammation. Intestinal epithelial cells are the first line of defense against the luminal pathogens, providing a barrier against their invasion of the underlying serosal layer. Factors that affect epithelial barrier function can involve multiple components, including alterations in ion and water secretion, modification of tight junction proteins, and reorganization of actin filaments. Studies have also suggested PAR₂ can lower intestinal epithelial integrity by promoting tight junction reorganization, decreasing R_T, and increasing Cl⁻ secretion (6–8, 19, 33, 52, 53).

Intestinal Cl⁻ secretion is regulated by a complex interplay of neurocrine, paracrine, endocrine, and inflammatory signals to provide for optimal luminal fluidity, convective mixing, and flushing in defense against pathogenic intruders (4). Transient Cl⁻ secretory responses are triggered by agonists (such as acetylcholine and PAR₂) that evoke a mobilization of intracellular Ca²⁺ (36, 51), whereas sustained Cl⁻ secretion is observed in response to agonists (such as vasoactive intestinal peptide) that promote the accumulation of intracellular cAMP (39). PAR₂-induced Cl⁻ secretion was shown here and by others to require activation of basolateral receptors (19, 51), which is somewhat puzzling because this process is reported to involve some of the same Gαq-dependent events elicited by apical receptors. One explanation is that the mechanisms leading to changes in I_sc involve interaction of PAR₂ with receptors or effectors present only at the basolateral surface. For example, many reports have suggested that intestinal Cl⁻ secretion involves transactivation of EGFR and subsequent activation of ERK1/2, and others have suggested involvement of secreted factors such as PGE2 and substance P (51). Thus, there is likely an additional layer of complexity beyond the differences in Gαq versus β-arrestin-dependent signaling described here, and differences in the response to Gαq-dependent signals downstream of apical and basolateral receptors may explain why apical PAR₂ does not promote short-circuit current changes.

In the studies described here, 2fAP added at the basolateral surface led to a mild reduction in transepithelial resistance and redistribution of ZO-1 from the tight junctions, events that might be associated with disruption of barrier function. Only basolateral PAR₂ activated the actin filament-severing protein cofilin and promoted actin assembly at tight junctions, processes previously shown to be β-arrestin dependent (57, 58). However, because both apical and basolateral pools of PAR₂ are able to promote activation of ERK1/2, and apical PAR₂ is a more potent activator of early ERK1/2 phosphorylation, this difference is not due to one pool simply being less accessible to activation. Both cofilin activity and ERK1/2-dependent myosin light-chain kinase (MLCK) phosphorylation, have been shown to be important for redistribution of tight junction proteins. Furthermore β-arrestins have been suggested to mediate PAR₂-induced MLCK-phosphorylation (6, 33). The data shown here suggest that activation of basolateral PAR₂ might play a greater role in tight junction reorganization (Fig. 8). Thus, it will be interesting to determine whether this pathway is involved in the regulation of epithelial integrity.

Another distinction between the Gαq/Ca²⁺ and β-arrestin-dependent pathways of ERK1/2 activation by PAR₂ is that the Gαq pathway is associated with gene expression and proliferation, while the β-arrestin-dependent pathway is associated with actin reorganization and chemotaxis (21, 25, 26). Indeed, we observe a primarily nuclear ERK1/2 localization in response to apical PAR₂ activation, while basolaterally activated ERK1/2 was sequestered in the cytoplasm and at the cell membrane. Apical PAR₂ might be exposed to luminal proteases upon damage to the protective mucus layer or released from invading pathogens, and activation of ERK1/2 might generate a transcriptional response appropriate to combating the injury or infection. In contrast, basolateral PAR₂ may respond primarily to proteases present in the bloodstream or released from invading immune and mast cells to promote cytoskeletal reorganization and gene expression/proliferative responses. The two pools of receptors may also promote a distinct set of transcriptional responses. In favor of this hypothesis, basolateral but not apical PAR2 activation resulted in activation of the AP-1 transcription factor. Although cytoplasmic/membrane sequestration of ERK1/2 by β-arrestins is typically associated with a decrease in transcription of genes involved in cell cycle progression, some cytosolic targets of ERK1/2, such as the protein kinase p90^Rsk, can translocate to the nucleus to elicit proliferative responses such as AP-1 activation (9–11). These responses are also associated with prolonged ERK1/2 activation such as elicited by basolateral receptor (11). β-Arrestin-dependent phosphorylation of p90^Rsk by ERK1/2 has been demonstrated for the angiotensin II type 1A receptor (1). Furthermore, the small amount of nuclear ERK1/2 (depicted in Fig. 8, middle cell) may also contribute to the observed activation of AP-1 by phosphorylating Elk-1 (27).

There are also reports of 2fAP applied to the lumen of the intestine increasing bacterial translocation. We have observed that luminal activation of PAR₂, as well as activation of apical PAR₂ in cultured monolayers, increases bacterial uptake into epithelial cells (C. Lau and K. A. DeFea, unpublished observations), which may be another mechanism by which apical PAR₂ contributes to PAR₂-induced inflammation. Future studies addressing the possible differences in gene expression profiles in response to apical versus basolateral PAR₂ activation as well as elucidation of other downstream effects of apical PAR₂ activation are needed to help answer this question.

GRANTS

This work was supported by National Institutes of Health Grant 1R01GM066151 (to K. A. DeFea).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).