Activation of ion secretion via proteinase-activated receptor-2 in human colon

Marcus Mall, Tanja Gonska, Jörg Thomas, Stephanie Hirtz, Rainer Schreiber, Karl Kunzelmann


Proteinase-activated receptor (PAR) type 2 (PAR-2) has been shown to mediate ion secretion in cultured epithelial cells and rat jejunum. With the use of a microUssing chamber, we demonstrate the role of PAR-2 for ion transport in native human colonic mucosa obtained from 30 normal individuals and 11 cystic fibrosis (CF) patients. Trypsin induced Cl secretion when added to the basolateral but not luminal side of normal epithelia. Activation of Cl secretion by trypsin was inhibited by indomethacin and was further increased by cAMP in normal tissues but was not present in CF colon, indicating the requirement of luminal CF transmembrane conductance regulator. Effects of trypsin were largely reduced by low Cl, by basolateral bumetanide, and in the presence of barium or clotrimazole, but not by tetrodotoxin. Furthermore, trypsin-induced secretion was inhibited by the Ca2+-ATPase inhibitor cyclopiazonic acid and in low-Ca2+ buffer. The effects of trypsin were almost abolished by trypsin inhibitor. Thrombin, an activator of PAR types 1, 3, and 4, had no effects on equivalent short-circuit currents. The presence of PAR-2 in human colon epithelium was confirmed by RT-PCR and additional experiments with PAR-2-activating peptide. PAR-2-mediated intestinal electrolyte secretion by release of mast cell tryptase and potentiation of PAR-2 expression by tumor necrosis factor-α may contribute to the hypersecretion observed in inflammatory processes such as chronic inflammatory bowel disease.

  • protease-activated receptors
  • ion transport
  • trypsin
  • cystic fibrosis transmembrane conductance regulator
  • inflammatory bowel disease

serine proteases have recently been demonstrated to act as signaling molecules (12). They regulate cells by specifically cleaving and activating members of a new family of protein-activated receptors (PARs). Apart from the three types of thrombin receptors, PAR-1, -3, and -4, the subtype PAR-2 has been identified as a receptor for trypsin and mast cell tryptase (12, 27). PARs are G protein-coupled receptors that are activated either by soluble ligands that reversibly bind to the receptor or by irreversible cleavage by a protease. In the latter case, a serine protease will cleave the receptor at the extracellular NH2-terminal end and a tethered ligand, which is now exposed, becomes available and binds to the receptor (12). PAR-2 was cloned initially from a mouse genomic library and subsequently from human kidney cDNA (5,27). The receptor shows a wide organ distribution and is highly expressed in pancreas, kidney, colon, liver, small intestine, and airways (5, 7, 10). Because little is known about the function of PAR-2 in the human colon, we examined in the present study the effects of trypsin and PAR-2-activating peptide (AP) in human rectal biopsies.

The function and physiological significance of PAR-2 is still poorly understood. However, recent studies suggested that PAR-2 might participate in the control of ion transport in gastrointestinal epithelia. Thus PAR-2 has been implicated in activating ion transport in cultured dog pancreatic duct cells (26). These studies suggested an activation of luminal Cl and basolateral K+ channels in pancreatic epithelial cells due to stimulation of basolaterally located PAR-2. Such an activation of ion secretion may promote clearance of toxins and debris from the pancreatic duct. Thus pancreatic trypsin may activate PAR-2, which are located in the pancreatic duct under both physiological and pathophysiological conditions. PAR-2 have also been identified in both luminal and basolateral membranes of epithelial cells in the small intestine (21). Expression of these receptors has been demonstrated by Northern blot analysis and immunohistochemistry (10, 21), and previous data obtained from rat jejunal mucosa suggested activation of ion transport by trypsin (33). On the basis of considerable differences in agonist concentrations and potency profiles for activation of short-circuit currents, the authors concluded that a receptor different than PAR-2 is activated by trypsin in the rat intestine (33).

However, the nature of ion conductances that are activated by stimulation of intestinal PARs and the responsible intracellular second-messenger pathways are still largely unknown. In the present study, we explored these questions by using a modified Ussing-chamber technique, which allows measurement of ion transport in small human rectal biopsies. The results presented here indicate pronounced activation of ion transport in the human colon by trypsin. Electrolyte secretion could be elicited repeatedly by trypsin in the same mucosal biopsy. It is thus very likely that activation of PARs, present in basolateral membranes of human colonic epithelial cells, contributes to the hypersecretion found in inflammatory processes such as chronic inflammatory bowel disease (13).



Ussing-chamber measurements were performed on rectal mucosa biopsies obtained from 30 normal individuals and 11 cystic fibrosis (CF) patients. All CF subjects presented with pancreatic insufficiency and chronic lung disease and fulfilled the diagnostic criteria of CF, including elevated sweat tests (30). Genotyping by DNA analysis of all CF patients showed that five were homozygous and six were heterozygous for ΔF508-CF transmembrane conductance regulator (CFTR; allelic frequency of 72%). Testing of an additional panel of the 19 most prevalent CFTR mutations among the Caucasian population in Europe, including G542X, N1303K, 1717–1 G>T, W1282X, G551D, R553X, R1162X, R334W, R117H, 621+1G>T, 3849+10kbC>T, 3659delC, 1078delT, R347P, A445E, S1251N, ΔI507, 2183AA>G, and E60X (ELUCIGENE CF20; AstraZeneca Diagnostics) failed to identify the second disease causing mutation in six CF patients. Small superficial tissue biopsies were obtained by rectoscopy and forceps biopsy performed at the University Children's Hospital Freiburg. The study was approved by the ethical committee, and the patients had given written informed consent. For children under the age of 18 yr, parents obtained detailed information and gave signed informed consent.

Ussing-chamber experiments.

Rectal tissue biopsies were immediately stored in an ice-cold buffer solution of the following composition (in mM): 127 NaCl, 5 KCl, 5d-glucose, 1 MgCl2, 5 Na pyruvate, 10 HEPES, and 1.25 CaCl2 and 10 g/l albumin. Tissues were mounted into a perfused microUssing chamber with a circular aperture of 0.95 mm2 as described previously (24). In brief, the luminal and basolateral sides of the epithelium were perfused continuously at a rate of 10 ml/min (chamber volume, 1 ml). The bath solution had the following composition (in mM): 145 NaCl, 0.4 KH2PO4, 1.6 K2HPO4, 5d-glucose, 1 MgCl2, and 1.3 Ca-gluconate, and pH was adjusted to 7.4. For low Cl solutions, Cl was reduced to 5 mM by equimolar replacement of 142 mM NaCl by Na+ gluconate, and 4 mM Ca gluconate was added to compensate for the Ca2+-chelating effects of gluconate. Bath solutions were heated by a water jacket to 37°C. Experiments were carried out under open-circuit conditions. Transepithelial resistance (R te) was determined by applying short (1 s) current pulses (ΔI = 0.5 μA), and the corresponding changes in transepithelial voltage (Vte; ΔVte) and basal Vte were recorded continuously. Values for the Vte were referred to the serosal side of the epithelium. Voltage deflections obtained under conditions without the mucosa present (ΔV′te) were subtracted from those obtained in the presence of the tissues.R te was calculated according to Ohm's law [R te = (ΔVte − ΔV′te)/ΔI]. The equivalent short-circuit current (I eq) was determined from Vte and R te, i.e.,I eq = Vte/R te. Tissues were allowed to equilibrate for 30 min before basal bioelectric properties were taken.

Cell culture, RNA isolation, and RT-PCR.

HT29 and T84 colonic carcinoma cells were grown in culture as described previously (18). In brief, cells were grown in Dulbecco's modified Eagle's medium with (in mM) 10 Na+-Hepes buffer, 4 l-glutamine (with 0.04 g/l penicillin), 0.09 streptomycin, and 100 newborn calf serum in 5% CO2. Total RNA was isolated from superficial biopsies of rectal mucosa and from the human colonic cell lines HT29and T84 using RNeasy spin columns (Quiagen, Hilden, Germany), as described previously (22), and was reverse transcribed at 37°C for 1 h using random primer and RT (Superscript RT, Life Technologies). The size of the expected 543-bp fragment of PAR-2 was amplified by PCR using the sense primer 5′-GTGTTTGTGGTGGGTTTGCC-3′ and antisense primer 5′-CATCAGCACATAGGCAGAGG-3′ (94°C for 60 s, 35 cycles of 94°C for 1 min, 57°C for 30 s, 72°C for 60 s). PCR products were visualized by loading an 8-μl sample on a 0.9% agarose gel using a 123-bp marker as a standard. The PCR product was subcloned into pBluescript SK (−) vector and sequenced using Thermo Sequenase I (Pharmacia) and a 373A DNA sequencer (Applied Biosystem).

Compounds and statistics.

Amiloride, bumetanide, indomethacin, cyclopiazonic acid (CPA) IBMX, forskolin, TTX, trypsin (bovine, 9,820 U/mg protein), trypsin inhibitor (type III-O chicken egg white; 1 mg will inhibit 1.1 mg trypsin with an activity of 10,000 BAEE U/mg protein), and thrombin (bovine, 56 U/mg protein) were all obtained from Sigma (Deisenhofen, Germany). PAR-2-AP (SLIGRL-NH2) corresponding to the tethered ligand of mouse PAR-2 and the reverse peptide (RP; LRGILS-NH2) were synthesized by solid-phase methods and purification by high-pressure liquid chromatography (Big Biotech, Freiburg, Germany). All used chemicals were of highest grade of purity available. From some individuals, transepithelial measurements were performed on more than one tissue sample. When multiple samples were studied by the same protocol, data were averaged to obtain a single mean value for each individual subject. Continuous bilateral bath perfusion allowed to perform consecutive measurements under different experimental conditions on the same tissue, and all experiments were performed in a paired fashion, in which each tissue served as its own internal control. Data for transepithelial measurements are shown as original recordings or as means ± SE and are generally reported as peak responses of Vte and I sc(n = number of subjects). Statistical analysis was performed using paired Student's t-test. Data obtained from CF and non-CF tissues were compared by unpaired Student'st-test. P values <0.05 were accepted to indicate statistical significance.


PAR-2 are expressed in native human colonic epithelium and are activated by trypsin.

RNA was prepared from superficial biopsies of rectal mucosa and from the human colonic cell lines HT29 and T84, and RT-PCR was performed using primers specific to the sequence of human PAR-2. A 543-bp fragment was obtained, which was verified as a PAR-2 fragment by subsequent cloning and sequencing (Fig.1). This result confirmed expression of PAR-2 for distal colon and rectum, which has been detected recently by immunocytochemistry (10) and Northern blot analysis in rat colon (21).

Fig. 1.

RT-PCR analysis of RNA obtained from a rectal biopsy (A) and from human colonic cell lines HT29 and T84. A 543-bp fragment of proteinase-activated receptor type 2 (PAR-2) was obtained in all 3 preparations after reverse transcription of total RNA (+) but not without RT (−) and subsequent PCR. Sequencing of the fragment confirmed amplification of human PAR-2.

First, we examined a possible role of PAR in regulation of ion transport in native human colonic tissues obtained from normal individuals. Baseline bioelectric properties were assessed after an equilibration period of 30 min in perfused microUssing chambers and are summarized in Table 1. When trypsin (1 μM) was added to the basolateral side of rectal tissues, we observed a transient increase in lumen-negative Vte andI eq, with peak responses of −0.8 ± 0.2 mV and −51.5 ± 19.5 μA/cm2 (n = 6), respectively. As the trypsin-induced lumen-negative responses obtained under basal conditions could reflect either an increase in cation absorption or anion secretion, subsequent experiments were performed in the presence of amiloride to inhibit electrogenic Na+absorption. Amiloride (10 μM, luminal) significantly inhibited Vte and I eq by 0.6 ± 0.1 mV and 26.2 ± 3.3 μA/cm2 (n = 30), respectively. In the presence of amiloride, trypsin was added to either the luminal or basolateral side of normal rectal biopsies. Basolateral perfusion with trypsin (1 μM) induced a transient increase in lumen-negative Vte (ΔVte = −2.3 ± 0.7 mV) and I eq (ΔIeq = −112.6 ± 35.8 μA/cm2) and significantly decreasedR te by −1.1 ± 0.4 Ωcm2(n = 6), whereas luminal trypsin had no effect on transepithelial ion transport (ΔVte = 0.0 ± 0.0 mV; ΔI eq = −0.7 ± 0.7 μA/cm2; n = 6; Fig.2, A and B).

View this table:
Table 1.

Bioelectrical properties of rectal biopsies from normal individuals and CF subjects under basal conditions, after treatment with amiloride (10 μM, luminal), indomethacin (10 μM, basolateral), and after cAMP-dependent stimulation with IBMX (100 μM) and Fors (10 μM)

Fig. 2.

Effect of trypsin (Tryp; 1 μM) on ion transport in human rectal mucosa from normal individuals. All experiments were performed in the presence of amiloride (10 μM) to block electrogenic Na+ absorption. Original experiment (A) and summary of Tryp-activated equivalent short-circuit current (I eq; B) showing that basolateral (bl) but not luminal (lu) addition of Tryp induced a transient lumen-negative Cl secretory response. C: when tissues were continuously perfused with different concentrations of Tryp (10 and 100 nM and 1 μM), the initial increase inI eq (peak) was followed by inhibition ofI eq (plateau) at Tryp concentrations ≥100 nM.D and E: concentration-response curve for the effect of Tryp bl on anion secretion was determined in tissues prestimulated with IBMX/forskolin (100 μM/1 μM) and resulted in an EC50 in the range of 20 nM. Transepithelial resistance (R te) was determined continuously from the transepithelial voltage (Vte) downward deflections obtained by pulsed current injection. Time gaps between recordings were 30 min. *Statistical significance for the effect of Tryp on basalI eq (paired t-test); #significant difference of the effect of bl vs. luminal Tryp (pairedt-test).

After basolateral trypsin was added, Vte andI eq typically returned to control values within 3–5 min and the perfusion was switched back to control buffer. In some tissues, we observed a further reduction in Vte andI eq below control values that was sustained for 10–30 min after trypsin washout. To address the possibility of trypsin-mediated inhibition of anion secretion in the late phase of the response, we performed experiments in which trypsin was added for 20 min. As shown in Fig. 2 C, 1 μM trypsin induced an initial increase followed by sustained reductions of Vte(ΔVte = 0.9 ± 0.1 mV) andI eq (ΔIeq = 40.3 ± 5.8 μA/cm2; n = 5) below baseline values. To examine whether inhibition of anion secretion in the late phase of the response was concentration dependent, we performed similar experiments with lower concentrations of trypsin (10 and 100 nM; basolateral). Whereas 100 nM trypsin elicited qualtitatively similar responses, we did not detect sustained inhibition of ion transport by 10 nM trypsin (Fig. 2 C). According to these data, basolateral but not luminal trypsin induced transient activation and, at higher concentrations, sustained inhibition of anion secretion.

To test whether trypsin-activated receptors are localized on epithelial cells or on subepithelial neuronal fibers, we performed experiments in which neurotransmitter release from the enteric nervous system was blocked by TTX. Perfusion with TTX (1 μM, basolateral) for 20 min had no effect on basal Vte and I eq(ΔVte = −0.1 ± 0.1 mV; ΔI eq = −4.5 ± 1.7 μA/cm2; n = 3). Importantly, TTX pretreatment had no effect on trypsin-mediated ion transport (trypsin peak response before TTX: ΔVte = −0.7 ± 0.3 mV; ΔI eq = −36.3 ± 15.6 μA/cm2 vs. trypsin peak response after TTX: ΔVte = −0.8 ± 0.4 mV; ΔI eq = −36.4 ± 12.6 μA/cm2; n = 3). These results indicate that PARs are localized on the basolateral membrane of colonocytes. As shown previously, basal and stimulated anion secretion in human colon can vary considerably after equilibration of tissues in Ussing chambers (25), most likely reflecting differences in the basal level of secretagogues released from individual native tissues. To control for the variability in endogenous autocrine or paracrine stimulation, the following experiments were performed in tissues prestimulated with IBMX (100 μM; basolateral) and forskolin (1 μM; basolateral). A concentration-response relationship was obtained for the effects of basolateral trypsin when added in concentrations ranging from 0.1 to 1 μM. EC50 was 21 nM (Fig. 2, Dand E). As shown in Fig. 2 D, trypsin was able to activate short-circuit currents repeatedly in the same tissue when a recovery period of 20–30 min was allowed. These results point out to a highly specific action of trypsin on basolateral protease-activated receptors, probably PAR-2.

We obtained further evidence indicating that trypsin acts on PAR-2 by applying a PAR-2-specific AP (SLIGRL-NH2). AP corresponds to the tethered ligand of cleaved PAR-2 and therefore does not require enzymatic cleavage for receptor activation. As summarized in Fig. 3, AP at concentrations as low as 10 μM induced a significant Cl secretory response when applied to the basolateral side (ΔVte = −0.2 ± 0.0 mV; ΔI eq = −13.6 ± 3.0 μA/cm2; n = 6) of colonic epithelia. High concentrations of AP (100 μM) elicited larger responses (ΔVte = −0.4 ± 0.1 mV; ΔI eq = −30.7 ± 4.5 μA/cm2; n = 4), and luminal application of AP was without effects on I eq (data not shown). In contrast, a peptide made from a reverse amino acid sequence (RP, LRGILS-NH2) had no effect on transepithelial ion transport (Fig. 3). Therefore, these data suggest specific stimulation of electrolyte transport by activation of basolateral PAR-2.

Fig. 3.

Effects of PAR-2-activating peptide (AP) and the reverse peptide (RP) on ion transport in human normal rectal mucosa. Original recording (A) and summary of peptide-inducedI eq (B) demonstrate that AP (10 and 100 μM; bl) but not RP (10 and 100 μM; bl) induced a lumen-negative Cl secretory response. All experiments were performed in the presence of amiloride (10 μM) and IBMX/forskolin (Fors; 100 μM/1 μM). *Statistical significance for the effect of peptide on basal I eq (paired t-test). #Significant difference of the effect of RP compared with AP (unpairedt-test); §significant difference for the effect of peptides at a concentration of 100 μM (filled bars) compared with 10 μM (open bars).

Cl secretion is not activated by thrombin and is inhibited by trypsin inhibitor.

PAR activation requires specific enzymatic cleavage by trypsin or tryptase, which releases a tethered ligand that activates its own receptor (12). It has been shown that trypsin-mediated activation of PAR-2 on epithelial cells can be abolished by preincubation of trypsin with trypsin inhibitor (26). To investigate whether the effects of trypsin on ion transport in human rectal biopsies could be antagonized by trypsin inhibitor, we compared the effects of trypsin in the absence and presence of trypsin inhibitor (1:1). After trypsin was pretreated with the inhibitor, the effects of trypsin onI eq were attenuated significantly from −56.8 ± 11.6 to −11.8 ± 5.6 μA/cm2(ΔVte was reduced from −1.0 ± 0.1 to −0.2 ± 0.1 mV; n = 4; Fig. 4,A and B). Trypsin has been shown to act on several PARs in the rank order PAR-2 ≥ PAR-4 > PAR-1 = PAR-3, whereas thrombin is a strong activator of PAR-1, PAR-3, and PAR-4 (12). To further characterize the PAR involved in trypsin-induced ion transport in the human distal colon, the effects of both trypsin and thrombin were compared. As depicted in Fig. 4,C and D, thrombin (1 μM, basolateral) failed to induce Cl secretion, whereas trypsin activated lumen-negative responses of Vte andI eq in paired experiments. Together, the present experiments strongly suggest that trypsin acts on basolateral PAR-2 in the human distal colon.

Fig. 4.

Comparison of the effects of Tryp in the absence or presence of Tryp inhibitor and thrombin on ion transport in human rectal mucosa from normal subjects [non-cystic fibrosis (CF)]. Original recording (A) and summary of Tryp-activatedI eq (B) showing that Tryp-induced Cl secretion was largely reduced when Tryp (1 μM, bl) was preincubated with Tryp inhibitor (1:1). The original recording (C) and the summary of agonist-inducedI eq (D) show that thrombin failed to activate Cl secretion. All experiments were performed in the presence of amiloride (10 μM) and IBMX/Fors (100 μM/1 μM). Time gaps between recordings were 30 min. *Statistical significance for the effect of Tryp on basal I eq (pairedt-test); #significant difference of the effect of Tryp pretreated with inhibitor or thrombin compared with Tryp (pairedt-test).

Trypsin-induced Cl secretion depends on cAMP-activation and requires luminal CFTR Cl channels.

We previously demonstrated that Ca2+-mediated Cl secretion induced by cholinergic stimulation of human distal colon relies on functional CFTR as the luminal Clchannel (24). However, previous results obtained on cultured cells from the kidney collecting duct and pancreatic duct suggest activation of Ca2+-dependent Clchannels by stimulation of PAR-2 (2, 26). Given these results, we asked whether trypsin may activate otherwise dormant non-CFTR Cl channels in human rectal mucosa. To that end, we examined the effects of trypsin in the absence and presence of cAMP-dependent stimulation and compared the responses obtained in tissues from normal individuals and CF patients. The effects of trypsin were compared with the effects of carbachol (100 μM, basolateral) in a strictly paired fashion under three different conditions:1) under basal conditions, 2) after perfusion with indomethacin, and 3) after cAMP-dependent stimulation with IBMX (100 μM) and forskolin (1 μM, both basolateral). As shown in Fig. 5 A, trypsin-induced anion secretion required coactivation by cAMP. Under baseline conditions with variable CFTR activity, trypsin induced a lumen-negative secretory response of −66.6 ± 22.8 μA/cm2 (ΔVte = −1.3 ± 0.5 mV;n = 10). Treatment with the cyclooxygenase inhibitor indomethacin inhibits prostaglandin synthesis, including the formation of PGE2, which has been identified as a major endogenous agonist of cAMP-dependent Cl secretion in human colon. Indomethacin (10 μM, basolateral, 60 min) pretreatment abolished trypsin-induced ion transport almost completely (ΔVte = −0.1 ± 0.1 mV; ΔI eq = −4.8 ± 2.2 μA/cm2; n = 10), whereas subsequent activation with IBMX and forskolin induced a sustained secretory response and resulted in a significant increase in trypsin-activated anion secretion (ΔVte = −1.4 ± 0.4 mV; ΔI eq = −99.8 ± 29.5 μA/cm2; n = 10) (Fig. 5, A andC). Ion-substitution experiments showed that cAMP and trypsin-induced secretory responses were carried by Cl. In the presence of IBMX and forskolin, replacement of extracellular Cl by gluconate (bilateral equimolar replacement of 142 mM Cl by gluconate; 5 mM Cl remaining) resulted in a significant inhibition of Vte andI eq (ΔVte = 1.5 ± 0.2 mV; ΔI eq = 46.9 ± 15.0 μA/cm2; n = 4). Furthermore, trypsin-mediated secretion was almost abolished in the presence of low Cl buffer (trypsin response under normal Cl: ΔVte = −2.6 ± 0.3 mV; ΔI eq = −80.0 ± 23.8 μA/cm2 vs. trypsin response under low Cl: ΔVte = −0.2 ± 0.1 mV; ΔI eq = −2.2 ± 0.3 μA/cm2; n = 4), demonstrating that PAR-2 activation induces Cl secretion in normal human colon. In contrast, in tissues obtained from CF patients (n = 10), trypsin failed to induce Cl secretion under all experimental conditions. Instead, trypsin activated a lumen-positive K+ secretory response, which was +12.2 ± 4.5 μA/cm2 (ΔVte = 0.4 ± 0.1 mV;n = 10) in the presence of indomethacin. These experiments demonstrate that trypsin-induced Cl secretion depends on the presence of functional CFTR as the luminal Cl channel. Although the magnitude of cholinergic secretion in both normal and CF tissues was significantly larger compared with trypsin-induced secretion, both agonists demonstrated a very similar behavior. This observation may suggest that both agonists share a common intracellular signal-transduction pathway.

Fig. 5.

Effects of Tryp (1 μM, bl) and carbachol (CCH) (100 μM, bl) on ion transport under basal conditions and in the absence (indomethacin, 10 μM, bl) and presence (IBMX, 100 μM and Fors 1 μM, bl) of cAMP-dependent activation of rectal biopsies from normal and CF subjects. All experiments were performed in the presence of amiloride (10 μM, luminal). A: in normal tissues (non-CF), Tryp and CCH induced a transient Cl secretory response under basal conditions that was further increased after cAMP stimulation of Cl secretion. In the presence of indomethacin (60 min), Cl secretion was abolished and CCH induced a reversed lumen-positive K+ secretory response.A: in CF tissues, Tryp- and CCH-induced Clsecretion was defective, and both agonists activated transient K+ secretion, which did not depend on cAMP activation. Time gaps between recordings were 40 min. Note that the magnification was changed during the course of the original recordings. C: summary of Tryp- and CCH-induced I eq obtained from experiments as shown in A (non-CF) and B(CF). Mean values ± SE. *Statistical significance for the effects of Tryp and CCH. #Significant difference for the effects of Tryp vs. CCH, **significant difference for Tryp- and CCH-inducedI eq compared with basal conditions, $Significant difference for Tryp- and CCH-induced I eqcompared with indomethacin (paired t-test). §Significant difference compared with non-CF (unpairedt-test).

Trypsin-induced Cl secretion requires basolateral Cl uptake and activation of K+ channels.

In the presence of amiloride, lumen-negative responses of Vte and I eq are caused by anion secretion. Due to the lack of highly specific inhibitors of CFTR Cl channels, we used bumetanide (100 μM, basolateral), an inhibitor of the Na+-K+-2 Clcotransporter, to block basolateral Cl uptake. As shown in Fig. 6, both cAMP-activated and trypsin-induced anion secretion was almost completely abolished by bumetanide, indicating that trypsin activated transepithelial Cl secretion.

Fig. 6.

Effect of bumetanide (100 μM, bl) and Ba2+(5 mM, bl) on Tryp (1 μM, bl)-activated Cl secretion in the presence of cAMP stimulation (IBMX 100 μM and Fors 1 μM). As shown in the original recording (A) and the summary (B), bumetanide inhibited cAMP- and Tryp-activated Cl secretion in normal tissues (non-CF). Similar results were obtained when tissues were perfused with Ba2+ from the bl side (C and D). All experiments were performed in the presence of amiloride (Amil; 10 μM, luminal). Time gaps between recordings were 20 min. Mean values ± SE. *Statistical significance for the effect of Tryp; #statistical significance for the effect of bumetanide and Ba2+; $significant difference for the effect of Tryp in the absence and presence of bumetanide or Ba2+.

We have shown previously that cholinergic stimulation of Cl secretion depends on basolateral Ca2+-activated K+ channels (24). Here we examined the effect of K+-channel inhibitors (BaCl2 and clotrimazole) on trypsin-induced secretion in human distal colon. BaCl2(5 mM, basolateral) significantly inhibited the sustained Cl secretory response in the presence of cAMP stimulation (IBMX/forskolin). Furthermore, the trypsin (1 μM; basolateral)-induced Cl secretory response was significantly reduced from −148.5 ± 4.3 to −62.1 ± 2.7 μA/cm2 (ΔVte was reduced from −2.5 ± 0.6 to −1.2 ± 0.4 mV; n = 7) (Fig. 6,C and D). Similar observations were made using clotrimazole as a more specific inhibitor of Ca2+-dependent K+ channels. In the presence of IBMX and forskolin, clotrimazole (30 μM) inhibited I eq from −26.7 ± 8.2 to −9.2 ± 2.5 μA/cm2 and abolished the trypsin-induced response almost completely from −100.5 ± 37.4 to −3.9 ± 2.5 μA/cm2(ΔVte was reduced from −1.7 ± 0.6 to −0.5 ± 0.2 mV; n = 3). These results indicate that trypsin-induced Cl secretion requires coactivation of basolateral K+ channels.

PAR-mediated ion transport requires an increase in intracellular Ca2+.

Activation of PAR-2 by trypsin has been shown to increase intracellular Ca2+ and generation of PGE2 (12). The present results suggested that ion transport activated by stimulation of PAR-2 is caused by increase in intracellular Ca2+. To further confirm the role of Ca2+ as the mediator of PAR-2 effects in human colon, tissues were treated for 20 min with the Ca2+-ATPase inhibitor CPA (50 μM, both sides). Inhibition of ATPase-dependent Ca2+ reuptake into endoplasmic stores is expected to cause a transient increase in intracellular Ca2+, followed by inhibition of agonist-induced store release. Accordingly, addition of CPA induced an initial Cl secretory response of −62.8 ± 13.9 μA/cm2 (ΔVte = −0.6 ± 0.1 mV;n = 5). After 20 min of CPA treatment, a stable plateau was reached, which was not different from precontrol values. As shown in Fig. 7, A and B, the effect of trypsin was almost completely abolished in the presence of CPA. Similar to the trypsin response, cholinergic stimulation by carbachol (CCH) was also inhibited in the presence of CPA (Fig.7, C and B). To address the role of extracellular Ca2+ in PAR-2-mediated secretion, we compared the effect of trypsin in low (1 μM) and normal Ca2+ (1.3 mM) bath solutions. Bilateral perfusion with 1 μM Ca2+ had no effect on Vte or I eq. In the presence of low-Ca2+ buffer, the effect of trypsin was almost abolished (trypsin response under normal Ca2+: ΔVte = −1.2 ± 0.2 mV; ΔI eq = −56.5 ± 7.9 μA/cm2 vs. trypsin response under low Ca2+: ΔVte = −0.3 ± 0.1 mV; ΔI eq = −23.2 ± 11.8 μA/cm2; n = 3). Similar observations were obtained when tissues were stimulated with CCH in normal compared with low Ca2+ solution (data not shown). These data suggest that activation of PAR by trypsin and stimulation of Clsecretion in the human colon depends on intracellular Ca2+signaling by endoplasmic stores release and extracellular Ca2+ entry.

Fig. 7.

Effect of the Ca2+-ATPase inhibitor cyclopianzonic acid (CPA; 50 μM, both sides) on Tryp- and CCH-induced Cl secretion in the presence of cAMP stimulation (IBMX 100 μM and Fors 1 μM). As shown in the original recording (A) and summaries (B), CPA induced a transient Cl secretory response. After 20 min of incubation with CPA, the effect of Tryp (1 μM, bl) was largely inhibited. Similar results were obtained when the effect of CCH (100 μM) was examined before and after CPA incubation (C and D). All experiments were performed in the presence of Amil (10 μM, luminal). Time gaps between recordings were 30 min. Mean values ± SE. *Statistical significance for the effect of Tryp and CCH; #statistical significance for the CPA peak response; $significant difference for the effect of Tryp and CCH in the absence and presence of CPA.


Trypsin can act as a signal molecule that specifically regulates cells by cleaving and activating PAR-2 (12). Trypsin has previously been shown to regulate ion transport in cultured epithelial cells from the pancreatic duct and kidney collecting duct by acting on PAR-2 located on the basolateral side of the epithelium (2,26). Whereas a physiological role of PAR-2 detected in cultured kidney cells remains obscure, a protective function during pancreatitis was claimed for PAR-2 expressed on the basolateral side of pancreatic duct cells. A recent report elucidated the role of PAR-2 in regulating salivary and pancreatic exocrine secretion in vivo in rats and in mice (20). A similar protective role of PAR-2 has been suggested for the airways (7): trypsin/trypsinogen is released from the airways and binds to PAR-2 coexpressed in airway epithelial cells, which then leads to paracrine release of a cyclooxygenase product, most likely PGE2, which induces bronchorelaxation (7). This study also analyzed the time course of recovery for responsiveness to PAR-2 activation after trypsin desensitization. It was found that a second trypsin response could be induced as early as 15 min after the initial application of trypsin and that trypsin-mediated relaxation recovered almost completely after 45 min. Resensitization could be blocked by brefeldin A and cycloheximide, indicating a fast de novo synthesis and trafficking of preformed receptors. The present results suggest that PAR-2 have a similarly high turnover rate in human distal colonic epithelium, because we were able to induce Cl secretion repeatedly only 20–40 min after the last application of trypsin. Protease-activated receptors have also been examined in cultured human bronchial epithelial cells. It was found that stimulation of PAR-2 transiently increased Cl secretion, which was followed by a sustained inhibition of amiloride-sensitive short-circuit currents (11). It is therefore likely that PAR-2 play a role in regulation of Na+ absorption and Cl secretion in human airways.

A basolateral PAR has been implicated in trypsin-mediated activation of ion transport in rat jejunum (33). From differences in the potency profile of the originally described AP (SLIGRL-NH2) and an alternative PAR-2-selective AP (LIGRLO-NH2), the authors suggested that a receptor different from PAR-2 is activated by trypsin in rat intestine. However, such a receptor has not yet been identified. It cannot be ruled out that 1) different PAR-2 subtypes exits that demonstrate different potency profiles for these synthetic agonists or that 2) the binding affinity of PAR-2 for different peptides is modulated by coexpression of other associated membrane proteins in the native epithelium. PAR-2 expression was detected in both basolateral and luminal membranes of enterocytes of the rat jejunum in another study (21). Stimulation of cultured enterocytes by basolaterally applied trypsin or PAR-2-activating peptide induced a release of arachidonic acid, generation of inositol 1,4,5-trisphosphate, and production and secretion of PGE2 (21). Thus pancreatic trypsin in concentrations usually present in the lumen of the jejunum may be able to activate PAR-2 located in either luminal or basolateral membranes of enterocytes (14) The data shown in the present study demonstrate that basolateral PAR-2 are responsible for trypsin-mediated activation of Cl secretion in human distal colon. Because trypsin had no effect on ion transport when applied from the luminal side, the receptors mediating trypsin-activated ion secretion are strictly located on the basolateral side of the epithelium. Differences in luminal and basolateral PAR-2 expression in small intestine and distal colon may indicate different physiological functions but could also be due to species differences. Importantly, pretreatment of tissues with TTX had no effect on trypsin-mediated Cl secretory responses, indicating that PAR-2 is expressed on the basolateral membrane of epithelial cells rather than on subepithelial structures, e.g., neuronal fibers. To assess trypsin-mediated Cl secretion, experiments were generally performed in the presence of amiloride to inhibit electrogenic Na+ absorption. However, we made quantitatively similar observations when trypsin was added under basal conditions in the absence of amiloride. Although it cannot be excluded that the increase in Vte and I eqobserved in the absence of amiloride is related to activation of cation absorption, the lumen-negative trypsin response in the absence of amiloride would also be in agreement with anion secretion under physiological conditions. This is well conceivable, because colonic crypts are composed of functionally distinct compartments, with Cl secretion predominantly taking place at the crypt basis and Na+ absorption at the surface epithelium (15). Experiments using trypsin concentrations ≥100 nM indicate that that PAR-2 activation may actually inhibit anion secretion in the continuous presence of the agonist. Alternatively, the transient decrease in Vte and I eqmay be due to activation of transepithelial K+ secretion. Furthermore, this observation could be caused by more complex mechanisms; e.g., endogenous PAR-2 activation in native tissues could contribute to basal and stimulated ion transport, and addition of trypsin at high concentrations may lead to inactivation of these PAR-2 by cleaveage and/or internalization. However, the trypsin concentrations (≥100 nM) required to induce inhibition of Vte and I eq in the plateau phase of the trypsin response were much higher than the EC50(21 nM). Therefore, inhibition of ion transport by PAR-2 appears unlikely under physiological conditions.

Previous studies on cultured renal and pancreatic cells suggested activation of lumenal Cl and basolateral K+channels by trypsin along with an increase in intracellular Ca2+ (2, 26). From the present experiments on human native colonic mucosa, there is significant evidence for activation of basolateral Ca2+-dependent K+channels. However, stimulation of PAR-2 did not activate CFTR or alternative Ca2+-dependent Cl channels in distal colonic tissues. This result confirms previous studies demonstrating the absence of Ca2+-activated Cl channels in human and mouse distal colon (16,24, 25). Cholinergic stimulation of Cl secretion in the mammalian colon has been studied in great detail (6, 17,24, 25, 32). Here, we compared the effects of carbachol and trypsin on normal and CF rectal biopsies and found qualitatively identical responses for both agonists, suggesting that CCH and trypsin activate similar ion conductances. Moreover, depleting the endoplasmic reticulum from Ca2+ renders the tissue unresponsive to trypsin. Thus trypsin is likely to act via release of IP3, an increase in intracellular Ca2+, and activation of basolateral Ca2+-dependent K+ channels, which enhances the driving force for luminal Cl secretion.

An important aspect of the present results is the reversibility of the PAR-2 mediated effects. Trypsin cleaves PAR-2 and activates the receptor irreversibly. Resensitization is due to mobilization of large Golgi stores and synthesis of new receptors (4). A very rapid partial recovery was observed in the present study already after 15 min, which is similar to what has been observed in the airways (7). This finding indicates a very rapid turnover of PAR-2 in native tissue and supports their significance in the regulation of ion transport in vivo. PAR-2 is also activated by inflammatory mediators such as tryptase, which is released during mast cell degranulation. In the human gut, mast cells are resident in the mucosa associated lymphatic tissue, where they secrete other proinflammatory cytokines, including tumor necrosis factor-α (TNF-α) (3). TNF-α and interleukin-1 as well as bacterial lipopolysaccharides have been shown to induce a sustained 10-fold increase in PAR-2 expression in endothelial cells (28), supporting participation of PAR-2 in the inflammatory response observed in chronic inflammatory bowel disease. Interestingly, TNF-α has been shown to be an essential mediator in inflammatory bowel disease, and recent clinical trials have shown that treatment with TNF antibody downregulates inflammation successfully in patients with Crohn's disease who did not respond to conventional treatment (1,31). Inflammatory bowel disease, as it occurs in Crohn's disease, is characterized by mast cell infiltration, which forms an essential component of intestinal granuloma (23, 29). Moreover, mast cells have been implicated in affecting ion transport in the human intestine in a previous study, and changes in ion transport have been found in patients with inflammatory bowel disease (9). These results are in parallel with those showing activation of colonic myocytes by mast cell tryptase and consecutive disturbances in colonic motility in Crohn's disease (8). Together with the results from this study demonstrating PAR-2-mediated activation of ion transport in human distal colon, epithelial PAR-2 may contribute to the pathophysiology of inflammatory bowel disease and thus may form a novel pharmacological target.


We thank the CF patients and volunteers who participated in this study. We gratefully thank Dr. P. Greiner (Children's Hospital, Univ. of Freiburg) for performing rectoscopy procedures and Dr. H. H. Seydewitz (Children's Hospital, Univ. of Freiburg) for genotype analysis of CF patients. We further thank Dr. R. C. Boucher (Cystic Fibrosis Research and Treatment Center, Univ. of North Carolina, Chapel Hill) for discussion and review of the manuscript.


  • The study was supported by Deutsche Forschungsgemeinschaft KU-1228/1 and KU-756/4–1, Zentrum Klinische Forschung 1 (A2), Univ. of Freiburg, Cystic Fibrosis Australia, Australian Research Council A00104609 and Mukoviszidose E.V.

  • Present address for M. Mall: Cystic Fibrosis/Pulmonary Research and Treatment Center, School of Medicine, The Univ. of North Carolina at Chapel Hill, 7011 Thurston Bowles Bldg., Chapel Hill, NC 27599-7248.

  • Address for reprint requests and other correspondence: M. Mall, Cystic Fibrosis/Pulmonary Research and Treatment Center, School of Medicine, The Univ. of North Carolina at Chapel Hill, 7011 Thurston Bowles Bldg., Chapel Hill, NC 27599-7248 (E-mail:mmall{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • 10.1152/ajpgi.00137.2001


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 20.
  20. 21.
  21. 22.
  22. 23.
  23. 24.
  24. 25.
  25. 26.
  26. 27.
  27. 28.
  28. 29.
  29. 30.
  30. 31.
  31. 32.
  32. 33.
View Abstract