Neutrophil Elastase Acts as a Biased Agonist for Proteinase-activated Receptor-2 (PAR2)*

Human neutrophil proteinases (elastase, proteinase-3, and cathepsin-G) are released at sites of acute inflammation. We hypothesized that these inflammation-associated proteinases can affect cell signaling by targeting proteinase-activated receptor-2 (PAR2). The PAR family of G protein-coupled receptors is triggered by a unique mechanism involving the proteolytic unmasking of an N-terminal self-activating tethered ligand (TL). Proteinases can either activate PAR signaling by unmasking the TL sequence or disarm the receptor for subsequent enzyme activation by cleaving downstream from the TL sequence. We found that none of neutrophil elastase, cathepsin-G, and proteinase-3 can activate Gq-coupled PAR2 calcium signaling; but all of these proteinases can disarm PAR2, releasing the N-terminal TL sequence, thereby preventing Gq-coupled PAR2 signaling by trypsin. Interestingly, elastase (but neither cathepsin-G nor proteinase-3) causes a TL-independent PAR2-mediated activation of MAPK that, unlike the canonical trypsin activation, does not involve either receptor internalization or recruitment of β-arrestin. Cleavage of synthetic peptides derived from the extracellular N terminus of PAR2, downstream of the TL sequence, demonstrated distinct proteolytic sites for all three neutrophil-derived enzymes. We conclude that in inflammation, neutrophil proteinases can modulate PAR2 signaling by preventing/disarming the Gq/calcium signal pathway and, via elastase, can selectively activate the p44/42 MAPK pathway. Our data illustrate a new mode of PAR regulation that involves biased PAR2 signaling by neutrophil elastase and a disarming/silencing effect of cathepsin-G and proteinase-3.

Neutrophil infiltration represents a rapid innate immune defense mechanism that accompanies acute inflammation.
Although one main neutrophil function is to target invading microorganisms and to respond to damaged tissue, there are many host cell responses triggered by the released neutrophil serine proteinases. In inflammation, apart from their bactericidal actions, the serine proteinases trigger a variety of effects ranging from cytokine, kinin, and growth factor generation to the clustering of integrins (1). PARs 3 belong to a novel fourmember family of G-protein-coupled receptors that are proteolytically activated by the unmasking of an N-terminal tethered ligand (TL) sequence by serine proteinases (2)(3)(4). Of note, many proteinases can silence/disarm PARs by cleaving downstream of the TL sequence, so as to prevent further activation by proteinases, but not by synthetic PAR-activating peptides with sequences derived from the TL (2,5). In previous work, we found that neutrophil elastase (NE) and cathepsin-G (CG) can block G q -mediated calcium signaling by trypsin-activated proteinase-activated receptor-2 (PAR 2 ), presumably by removing the enzyme-targeted tethered ligand sequence (6,7). PAR 2 activation by serine proteinases is widely considered to cause a proinflammatory response in various pathological conditions, in part via a neurogenic mechanism, and thus its disarming by neutrophil proteinases could be considered as an anti-inflammatory effect. There are also intriguing data that assign a protective function to PAR 2 activation in certain inflammatory conditions (8 -11). The mechanism(s) for this potentially biphasic action of PAR 2 is not known, but it may relate to the ability of PAR 2 to exhibit biased signaling either via a G q -coupled calcium signal or via G i /G 12/13 -coupled p44/42 MAPK signals (12). Our recent report showed that proteolytically revealed mutated rat TL sequences and one of the PAR 2 -activating peptides we tested were able to trigger such biased agonism, activating MAPK but not calcium signaling pathways, without either receptor internalization or the recruitment of arrestins (12). Evidence for biased signaling by PAR 2 also comes from the observation that a novel peptidomimetic PAR 2 antagonist could inhibit intracellular calcium release induced by a PAR 2 agonist but failed to inhibit ERK signaling (13). It is, however, not known if endogenous agonists can trigger biased sig-naling through PAR 2 . In trying to uncover if such biased signaling might occur in vivo, we hypothesized that neutrophil proteinases that can cleave the PAR 2 N terminus but do not trigger PAR 2 calcium signaling might nonetheless activate MAPK to act as endogenous biased agonists for PAR 2 . We show here that neutrophil proteinases (NE, CG, and proteinase-3 (PR3)) can all disarm PAR 2 -dependent calcium responses to trypsin, whereas NE (but neither CG nor PR3) concomitantly activates MAPK signaling. We show further that all three neutrophil proteinases can target the PAR 2 N terminus to remove the TL but via cleavage at distinct cleavage sites. We propose that under inflammatory conditions, this novel TL-independent signaling via PAR 2 , biased toward the MAPK pathway, may underlie differential PAR 2 -dependent responses in inflammation.

EXPERIMENTAL PROCEDURES
Chemicals and Other Reagents-Culture medium (DMEM and DMEM/F-12) and fetal bovine serum (FBS) were from Invitrogen. Fluo-4/AM no wash calcium indicator dye and the ReASH fluorescent probe were from Invitrogen. Fugene-6 transfection reagent was obtained from Roche Applied Science. Calcium ionophore A23187, porcine trypsin (catalogue no. T-7418; ϳ14,900 units/mg), collagen type IV from human placenta, insulin/transferrin/sodium selenite, hydrocortisone, 5-triiodo-L-thyronine sodium salt, PGE-1, and EGF were from Sigma. A maximum specific activity of 20,000 units/mg was used to calculate the approximate molar concentration of trypsin in the incubation medium (1 unit/ml, ϳ2 nM). The neutrophil enzymes, neutrophil elastase, cathepsin-G, and proteinase 3, purified from human sputum, were obtained from Elastin Products (Owensville, MO) with specific activities of 875, 3.1, and 95 units/mg, respectively. No trace trypsin-like activity was detectable in the enzymes as confirmed by us with a 96-well plate-based enzyme activity assay using the fluorogenic substrate, t-butoxycarbonyl-QAR-AMC (Bachem, Torrence, CA). The maximum concentrations of enzymes in our experiments were 1 M at 3 units/ml for NE and PR3 and 4 M at 300 milliunits/ml for CG. The NE inhibitor, fragment of elafin, was from Anaspec (Fremont, CA). NE activity was measured using the fluorogenic substrate methyl-O-succinyl-Ala-Ala-Pro-Val-AMC (elastase substrate V, EMD Chemicals, Gibbstown, NJ). The PAR-activating peptide, SLIGRL-NH 2 , and the rat PAR 2derived sequence ITGKGAPVEPGFSVDEFSASVLTGKLT (D27) as well as the TL-containing sequences (TL sequence underlined) for human PAR 1 (NATLDPRSFLLRNPNDKYE), PAR 2 (GTNRSSKGKSLIGKVDGTSHV), and PAR 4 (GDDSTP-SILPAPRGYPGQV) used as enzyme substrates were synthesized by the Faculty of Medicine Peptide Synthesis facility at the University of Calgary with purity verified by amino acid analysis and mass spectral analysis.
Cell Culture-KNRK and HEK cells were cultured as described previously (12). Briefly, cells were grown to confluence in DMEM supplemented with 10% FBS and 1% antibiotic/ antimycotic. Cells that were transfected with either the rat PAR 2 -encoding pCDNA3.1 vector or the empty pCDNA3.1 vector were grown in DMEM supplemented with 600 g/ml G418. Confluent cultures were dissociated in PBS containing 1 mM EDTA and subcultured as appropriate. Mouse embryonic fibroblasts from wild type (MEFwt) and ␤-arrestin-1/2 double knock-out animals (MEF␤arrDKO) were a gift from Dr. Robert Lefkowitz and have been described previously (14,15).
Primary human proximal tubular (HPT) cells were isolated as described before (16). Briefly, normal renal cortical tissue obtained from an adult kidney undergoing nephrectomy was dissected, minced, digested with collagenase IV, and passed through a 75-m mesh. Filtered cells were washed in Hanks' balanced salt solution and resuspended in hormonally defined medium and plated in a Petri dish. After 30 min, the non-adherent HPT cells were transferred to a collagen-coated plate and cultured in DMEM/F-12 (Invitrogen) supplemented with 1% FBS, 1% penicillin/streptomycin, 125 ng/ml prostaglandin E1, 25 ng/ml epidermal growth factor, 1.8 g/ml L-thyroxine, 3.38 ng/ml hydrocortisone, and 2.5 mg/ml insulin/transferrin/ sodium selenite supplement. Experiments involving the use of human tissue were approved by the Conjoint Health Research Ethics Committee at the University of Calgary.
Cloning and Transfection-The plasmid encoding the rat PAR 2 receptor (wt-rPAR 2 ) was constructed as described previously (17). Permanent KNRK cell lines expressing PAR 2 and pCDNA3.1 have been described previously (17). HEK-293 cells, used for the PAR 2 -biarsenical fluorochrome binding domain (BAB) construct cleavage experiments and the bioluminescence resonance energy transfer (BRET) experiments, were transiently transfected with the pcDNA3.1 ϩ vector encoding the human PAR 2 receptor with the N-terminal BAB tag or the pcDNA3.1 vector encoding the human PAR 2 gene with a c-terminal YFP tag, using the Fugene 6 (Roche Applied Science) transfection reagent (12). Transiently transfected cells were assayed 48 h post-transfection.
Measurement of Trypsin and NE Activity-NE activity was measured using the substrate methyl-O-succinyl-Ala-Ala-Pro-Val-AMC (elastase substrate V, EMD Chemicals). 50 l of 200 milliunits/ml elastase in Hanks' balanced salt solution containing 10 mM HEPES, pH 7.0, 1.5 mM MgCl 2 , 1.5 mM CaCl 2 , and 0.2% Nonidet P-40 was incubated with 50 l of 200 M elastasespecific substrate methyl-O-succinyl-Ala-Ala-Pro-Val-AMC (elastase substrate V, EMD Chemicals). The fluorescence from the free AMC as an index of proteolytic activity was measured every 10 min for 60 min using a Victor X4 fluorescent plate reader (PerkinElmer Life Sciences). In order to confirm inhibition of NE by elafin, NE and elafin were preincubated for 5 min at room temperature prior to measuring NE activity as described above. Trypsin activity was measured using the substrate t-butoxycarbonyl-QAR-AMC with similar conditions as for NE. The effect of neutrophil elastase on trypsin activity was measured by preincubating the two enzymes for 5 min at 37°C prior to measuring trypsin activity.
Calcium Signaling-KNRK cells or HPT cells were incubated for 30 min with Fluo-4/AM no wash calcium indicator and assayed for calcium signaling responses as described previously (12) on a victor X4 fluorescent plate reader (PerkinElmer Life Sciences) or an Aminco Bowman fluorescence spectrophotometer (Thermo Life Science). Receptor disarming was monitored essentially as described previously (18), with minor modifications. Briefly, PAR 2 -expressing KNRK cells or HPT cells were first exposed or not to 0.3-3 units/ml of the three neutrophil proteinases for 20 min before assaying responses to increasing concentrations of trypsin or SLIGRL-NH 2 .
p44/42 MAPK Signaling-Western blot detection of p44/42 MAPK was performed essentially as described (12). In brief, KNRK cells that had been transfected with a PAR 2 -expressing construct (PAR 2 -KNRK) or empty vector (pcDNA3-KNRK) were placed in DMEM with no serum for 12 h and stimulated with either trypsin, SLIGRL-NH 2 , NE, CG, or PR3 for 10 min. Cells were then placed on ice, and agonists were removed prior to lysis with cold lysis buffer. Protein samples were boiled for 10 min in Laemmli buffer and resolved on SDS-polyacrylamide gels. Activation of p44/42 MAPK was monitored by immunoblotting with phospho-p44/42 MAPK-specific antibodies and visualizing on an Eastman Kodak Co. Imagestation, and the increases in p44/42 MAPK phosphorylation were quantified relative to the total p44/42 MAPK or ␤-actin signal detected in the same samples. Densitometry analysis of multiple Western blot images was done to quantify the percentage increase over base line of p44/42 MAPK activation by the different proteinases. MAPK signaling in HPT cells was assessed as described above for KNRK cells. For p44/42 MAPK studies in MEFs, cells were treated as described for KNRK cells except that after incubation with agonists, cells were washed with ice-cold PBS and lysed directly in 1ϫ Laemmli sample buffer. Phospho-p44/42 MAPK and total p44/42 MAPK antibodies were visualized on Western blots with Alexa 680 and IR800-coupled secondary antibodies using the LICOR Odyssey Imaging system.

Measurement of PAR 2 "Disarming" by Neutrophil Proteinases Using Biarsenical Fluorochrome Binding Motif-tagged PAR 2 -
The BAB, containing a tetracysteine motif (19), AFLNCCPGC-CMEP, was inserted by introducing corresponding oligonucleotides just downstream of the signal peptide of human PAR 2 (Gln 27 ). The PAR 2 DNA constructs containing the N-terminal BAB domain and a C-terminal YFP fluorochrome tag were cloned in pCDNA3.1 ϩ at the restriction enzyme sites between EcoRI and XbaI (PAR 2 -BAB-YFP). HEK-293 cells obtained from ATCC (Manassas, VA) were transfected with the PAR 2 -BAB-YFP construct with the FuGENE6 transfection reagent (Roche Applied Bioscience). The expression level of the expressed receptor was quantified using a fluorescence plate reader (Victor 4X, PerkinElmer Life Sciences), and confocal microscopy was used to monitor the receptor location by visualizing the receptor YFP tag. To detect proteolytic release of the N-terminal PAR 2 domain, PAR 2 -BAB-YFP cells were labeled with a 0.2 M concentration of the biarsenical fluorochrome ReAsH (using the TC-TC-ReAsH II in-cell tetracysteine tag detection kit (Invitrogen)) for 30 min followed by washing with the included BAL buffer. Labeled cells were then incubated with proteinases for 30 min in HEPES-buffered saline, pH 7.4, containing 1.5 mM CaCl 2 . Supernatants were removed from the cells and briefly centrifuged, and fluorescence measurements were done using a Victor X4 fluorescent plate reader, with an excitation wavelength of 540 nm and an emission wavelength monitored at 615 nm to detect release of the peptide containing the receptor BAB tetracysteine motif. Fluorescence readings were also obtained from the cell monolayer to detect loss or not of the BAB tetracysteine motif from the cell surface.
Neutrophil Proteinase Cleavage of Synthetic PAR 2 -derived Peptides-Cleavage of the PAR 2 -derived synthetic peptides spanning the TL regions (human PAR sequences; NATLDPRS-FLLRNPNDKYE (PAR 1 ), GTNRSSKGKSLIGKVDGTSHV (PAR2), and GDDSTPSILPAPRGYPGQV (PAR 4 )) or encompassing the region downstream of the PAR 2 TL region (rat PAR 2 sequence ITGKGAPVEPGFSVDEFSASVLTGKLT, highly homologous with the comparable human sequence, VTGKGVTVETVFSVDEFSASVLTGKLT) was done as described previously for trypsin IV and tissue kallikreins (20,21). Peptides (100 M in a total volume of 150 l) were incubated with neutrophil proteinases (0.7 unit/ml) for 15 min at 37°C. Reactions were stopped by adding 150 l of ice-cold 0.1% trifluoroacetic acid (TFA) in water. Samples were fractionated by reverse-phase high performance liquid chromatography (5-40% acetonitrile gradient in 0.1% TFA over 30 min at a flow rate of 1 ml/min), and eluted peptides were analyzed by MALDI mass spectrometry. Neutrophil proteinase cleavage of peptides spanning the TL regions of PAR 1 (NATLDPRSFLLRNPND-KYE) and PAR 4 (GDDSTPSILPAPRGYPGQV) were also assessed as described above for the PAR 2 peptides.
BRET-based Detection of ␤-Arrestin-1 Interaction with PAR 2 -PAR 2 was conjugated with YFP at the C terminus (PAR 2 -YFP), and BRET to Renilla luciferase-tagged ␤-arrestin-1 (Rluc-␤-arr1) was determined essentially as described previously (22). HEK-293 cells were used instead of the KNRK cell line, because the HEK cells provide for a much greater transfection efficiency that is required for the BRET studies. In brief, HEK-293 cells were transiently transfected with 1 g of PAR 2 -YFP construct along with 0.1 g of the Rluc-␤-arr1 construct. Cells were plated in white 96-well culture plates (PerkinElmer Life Sciences), and interactions between the receptors and ␤-arrestin-1 were detected by measuring BRET at timed intervals over 60 min following the addition of 5 M coelenterazine (Nanolight Technology, Pinetop, AZ) on a Mithras fluorescence plate reader (Berthold, Mandel Scientific (Guelph, Canada)) in luminescence mode using the appropriate filters.
Monitoring PAR 2 Internalization-Using the approach described in more detail elsewhere (12), HEK cells were plated on glass bottom Petri dishes (MatTek Corp., Ashland, MA) and transiently transfected with 1 g of PAR 2 -YFP construct. After 48 h, the medium in the Petri dish was replaced with serum-free medium, and the cells were treated with trypsin (10 nM), SLI-GRL-NH 2 (10 M), or NE (3 units/ml) for 30 min at 37°C. After fixing cells with 4% formaldehyde, the cellular localization of the receptor tag YFP signal was detected using an Olympus FV1000 confocal system on an Olympus IX70 microscope with the Fluoview system software. Internalization of cell surface receptor was quantified morphometrically by counting the number of intracellular fluorescent speckles per cell in the images, indicative of receptor internalization to endocytic vesicles. Speckles in all cells were counted in a randomly selected representative ϫ40 image. The average number of speckles per cell was calculated, and the observation was repeated for comparable fields for cells observed in three independently conducted experiments.

Neutrophil Proteinases Target Distinct Cleavage Sites and
Disarm PAR 2 by Cleaving the Receptor N Terminus Downstream of the TL Sequence-As we anticipated from our previous work (6, 7) and as shown in Fig. 1A (representative data for NE) and Fig. 1B, pretreatment of PAR 2 -expressing cells with NE, CG, or PR3 abrogated subsequent trypsin-mediated calcium signaling, without affecting the response to the PAR-activating peptide, SLIGRL-NH 2 . The abrogation of the ability of trypsin to activate PAR 2 was not due to neutrophil proteinase digestion of trypsin itself as confirmed by experiments where preincubating trypsin with NE (as opposed to its target cells) did not affect either its ability to cause calcium signaling or to cleave a fluorogenic substrate (supplemental Fig. S2). This loss of PAR 2 sensitivity to trypsin activation upon pretreating cells with the neutrophil proteinases can be explained by the ability of all three enzymes to cleave and release the N-terminal domain of PAR 2 from the cell surface (Fig. 2), making the TL unavailable for trypsin-mediated unmasking. This N-terminal cleavage was verified by detection of the fluorescently labeled receptor N-terminal fragment containing a biarsenical fluorochrome binding motif in the cell supernatant after enzyme treatment. To assess similarities or differences in the cleavage of the PAR 2 N-terminal extracellular sequence by the three neutrophil proteinases, we used a synthetic 27-mer peptide substrate representing the rat PAR 2 sequence downstream of the TL and just C-terminal to a series of three prolines that would segregate this sequence from the upstream TL moiety (Fig. 3). HPLC separation and mass spectral identification of the cleavage products showed that each of the enzymes cleaved at a distinct site in this peptide with predominant cleavage observed: at Ser 68 -Val 69 for NE, at Phe 65 -Ser 66 for CG, and at Val 62 -Asp 63 for PR3 (Fig. 3).
In contrast with the cleavage of the PAR 2 27-mer that represents a PAR 2 sequence downstream from the TL, we failed to see any cleavage of the peptide corresponding to the human PAR 2 N-terminal region spanning the TL cleavage activation site, GTNRSSKGK2SLIGKVDGTSHV (where the downward arrow denotes the cleavage site that unmasks the TL) (data not shown). These studies indicated that the neutrophil-derived proteinases are unable to reveal the PAR 2 -activating TL and agree completely with previous proteomic work demonstrating the inability of neutrophil proteinases to cleave sequences representing the TL of PAR 2 (23). Thus, neutrophil enzymes disarm PAR 2 by removing the N-terminal TL receptor-activating sequence, whereas the retention of calcium signaling in response to SLIGRL-NH 2 after treatment with neutrophil proteinases showed that the extracellular receptor loops involved in signaling remained functional.
NE Activates PAR 2 -dependent p44/42 MAPK Signaling, whereas CG and PR3 Do Not-Because we previously observed that PAR 2 can selectively couple to the p44/42 MAPK pathway without triggering calcium release (12), we wondered if the neutrophil proteinases might act as endogenous biased agonists at PAR 2 , selectively activating MAPK but not calcium transients. Following activation of KNRK cells stably expressing rat-PAR 2 (KNRK-PAR 2 ) with NE for 10 min (but not with either CG or PR3) we indeed observed a robust p44/42 MAPK signaling with an EC 50 for NE of about 0.7 unit/ml (Fig. 4). This response was not observed in empty vector-transfected KNRK cells (Fig. 4C).  Supernatants were removed from the cells and briefly centrifuged, and fluorescence was measured using a Victor X4 fluorescent plate reader, with an excitation wavelength of 540 nm and an emission wavelength monitored at 615 nm. Fluorescence readings were obtained from the culture supernatants to detect release of the peptide containing the receptor BAB tetracysteine motif and from the cell monolayer to detect loss or not of the BAB tetracysteine motif from the cell surface (mean Ϯ S.E. (error bars), n ϭ 3). *, significant increase in detectable released fluorescence in assay buffer compared with untreated sample (NT); #, significant decrease in detectable cell surface fluorescence compared with untreated sample (p Ͻ 0.05).
Despite being able to cleave the PAR 2 N terminus to prevent trypsin signaling, neither CG nor PR3 was able to activate MAPK signaling in PAR 2 -transfected KNRK cells (Fig. 4A).
Given that short synthetic peptides corresponding to the trypsin-revealed N terminus of PAR 2 can activate the receptor, we synthesized peptides derived from the N-terminal sequence of PAR 2 exposed by CG (SASVLTGKLTTVFL-NH 2 ) and NE (VLTGKLTTVFL-NH 2 ) (Fig. 3). In contrast with the trypsinrevealed TL-derived peptide (SLIGRL-NH 2 ), the NE/CG-revealed TL peptides (10 -100 M) were unable to trigger either a calcium signal or MAPK signal in PAR 2 -transfected KNRK cells (supplemental Fig. S1).
NE Disarms Calcium Signaling and Activated MAPK Signaling in HPT Cells That Endogenously Express PAR2-In order to confirm the existence of NE-dependent biased signaling in cells that endogenously express PAR2, we have examined the effect of NE on calcium and MAPK signaling in HPT cells that naturally express high levels of PAR2 (24). Treatment of HPT cells with NE did not stimulate a calcium signal but significantly attenuated subsequent trypsin-stimulated calcium transients (Fig. 5, A and B). In contrast, treatment of HPT cells with NE for 5 min stimulated robust p44/42 MAPK signaling that declined but remained higher than base line at 15 and 30 min post-treatment (Fig. 5, C and D). Rho 2  6). In order to confirm the specificity of NE-stimulated MAPK activation, we monitored the ability of NE to activate MAPK following incubation with the specific NE inhibitor elafin (supplemental Fig. S3). We found that the NE-stimulated MAPK activation was completely inhibited by 10 M elafin and that this inhibition was dependent on the elafin concentration, with less inhibition at lower concentrations (Fig. 7).

NE-cleaved PAR 2 Does Not Recruit ␤-Arrestin and Is
Retained on the Cell Surface-PAR 2 activation by trypsin is followed by rapid receptor phosphorylation, recruitment of ␤-arrestin to the receptor, and subsequent internalization and targeting of the activated PAR 2 either to lysosomes for degradation or to an intracellular scaffold for further signaling via MAPK (12,14,25). A lack of ␤-arrestin recruitment would be expected to result in a different signaling response and differences in receptor trafficking compared with trypsin-activated PAR 2 . In order to determine if NE-cleaved PAR 2 interacts with ␤-arrestin, we transfected cells expressing YFP-tagged PAR 2 along with the Rluc-␤-arr1 and employed an energy transfer (BRET) assay to detect receptor ␤-arrestin-1 interactions after PAR 2 activation. Trypsin and PAR 2 -activating peptide (2f-LIGRLO-NH 2 ) stimulation of PAR 2 -expressing cells resulted in a substantial increase in the BRET ratio, indicating a recruitment of ␤-arrestin-1 to the activated receptor (Fig. 8A). However, following activation of PAR 2 with NE, we failed to observe any increase in the BRET ratio (Fig. 8A). Similar data were FIGURE 5. NE disarms calcium signaling and activated p44/42 MAPK in HPT cells. HPT cells were incubated with the Fluo-4 no wash calcium indicator for 30 min, and trypsin-stimulated calcium signaling was monitored in cells that were preincubated or not with 3 units/ml NE for 20 min. A, top, representative trace showing trypsin signaling in HPT cells. Bottom, representative trace showing the lack of NE-stimulated calcium signaling in HPT cells and the subsequent attenuation of trypsin triggered calcium signaling in these cells (// indicates 20 min). The PAR 2 -AP SLIGRL-NH 2 is still able to stimulate calcium signaling in cells that are disarmed by NE (mean Ϯ S.E. (error bars), n ϭ 3; *, significant decrease in trypsin-stimulated calcium signal in HPT cells that were exposed to NE compared with cells that were not preincubated with NE; p Ͻ 0.05). B, histogram showing trypsin-stimulated calcium signaling in HPT cells that were preincubated or not with 3 units/ml NE as a percentage of calcium stimulated by the calcium ionophore A23187 (mean Ϯ S.E., n ϭ 3). *, significant increase in p44/42 MAPK signal compared with no treatment (NT) (p Ͻ 0.05). obtained when monitoring recruitment of ␤-arrestin-2 (data not shown). The lack of ␤-arrestin recruitment to NE-activated PAR 2 correlated with a lack of internalization of the receptor in cells that had been activated with NE (Fig. 9). As seen for NE, neither CG nor PR3 caused receptor internalization (Fig. 9, lower panels). In contrast, activation of cells with either trypsin or SLIGRL-NH 2 resulted in loss of cell surface receptor expression and increases in intracellular PAR 2 -YFP levels in endosome-like structures (Fig. 9, upper panels).

DISCUSSION
The discovery that serine proteinases can signal to cells by cleaving and activating PARs via a unique tethered ligand mechanism established a novel paradigm for inflammatory cell regulation (26,27). Our new work described here demonstrates for the first time that neutrophil elastase, an endogenously expressed proteinase, can act as a biased agonist at PAR 2 . In an inflamed tissue, increased neutrophil infiltration and release of neutrophil serine proteinases will probably shift the balance of PAR 2 signaling toward this newly identified signaling mechanism and could have important implications for our understanding of how these receptors regulate inflammation.
Our new data broaden considerably, well beyond the trypsinrelated serine proteinase family, the spectrum of proteinases that can cause tissue responses via the PARs, including the possibility of PAR 2 biased signaling (MAPK versus calcium), demonstrated here for elastase but not for the other two neutrophil proteinases. Previous work demonstrating the disarming of PAR 2 by neutrophil and Pseudomonas elastases (6, 28) considered only the silencing of the calcium signal and did not recog-  nize the potential impact on MAPK activation that our new observations reveal. Interestingly, the mechanism of MAPK activation also appears to be distinct from the trypsin-triggered response. We observe that the elastase-activated receptor is impaired in its ability to internalize and fails to interact with ␤-arrestin. ␤-Arrestin interaction is known to be important for PAR 2 internalization (29). Further, ␤-arrestin scaffolds the internalized receptor to Raf-1 and activated ERK-MAPK and ensures cytosolic retention of the activated ERK (30), whereas ERK-MAPK activation by PAR 2 receptors that are unable to interact with ␤-arrestin results in the nuclear translocation of activated ERK. These differences in localization of signaling complexes would suggest that the MAPK signaling pathway triggered by NE, which is independent of ␤-arrestin interaction, would result in ERK activation of nuclear transcription, whereas trypsin-activated PAR 2 ERK-MAPK signaling would target non-nuclear substrates. This possible difference in MAPK signaling will probably impact on the inflammatory response that PAR 2 directs. The inhibition of the NE-triggered MAPK signal by inhibitors of Rho-associated kinase indicate that this response is occurring through the coupling of the receptor to the G␣ 12/13 . The ability of PAR 2 to engage G␣ 12/13 has been reported in a number of other cells (31,32), but the implications of selectively activating this arm of PAR 2 signaling by NE remain to be fully understood.
Although a number of studies have described the ability of neutrophil-derived enzymes to inactivate PAR 2 -dependent cal-  cium signaling (6, 7), few studies have described the ability of these enzymes to modulate cellular responses through this receptor. A recent study examining the regulation of epithelial barrier permeability has shown that NE and PR3 activation of PAR 2 (and PAR 1 ) on the basolateral aspect of colonic epithelial cells increases barrier permeability (33). The authors proposed that this effect was a mechanism for allowing increased neutrophil transepithelial migration. More recently, in an in vivo mouse model of colitis, it was shown that inhibition of NE by elafin significantly suppresses inflammatory mediators and strengthens the intestinal epithelial barrier functions in colonic tissues from mice as well as in intestinal epithelial cells (34). Given that PAR 2 is believed to be a key player in the pathogenesis of colitis (35,36), it will be interesting to evaluate further the signaling responses and trafficking of the receptor in the colonic epithelial cells to assess the possible relationship to data in this report. Similarly our data in kidney proximal tubular cells, which express an abundance of PAR2, demonstrate that neutrophil infiltration, which is a hallmark of renal inflammation, will shift the balance of signaling via this receptor. This NE-mediated impact on tubular cell signaling in the kidney may trigger an epithelial to mesenchymal transition so as to play a role in renal fibrosis.
The cleavage of the PAR 2 N terminus downstream of the TL-revealing site by the neutrophil proteinases can also be compared with our preliminary evaluation of their cleavage of the human PAR 1 -, PAR 2 -, and PAR 4 -derived sequences spanning the TL domains (supplemental Table 1). NE and PR3 were able to cleave the TL-spanning peptide of PAR 1 at sites that would prevent subsequent activation by thrombin, and thus this novel signaling mechanism that we have uncovered for PAR 2 may reflect a general mechanism for PAR regulation by neutrophil proteinases. There is some evidence to suggest that PAR 1 can exhibit biased agonism. For instance, elastase, which we show cannot unmask the PAR 1 TL sequence in vitro, is nonetheless able to induce apoptosis in human lung epithelial cells via a process that appears to be PAR 1 -mediated (37,38). Further, platelet PAR 1 was recently reported to be cleaved by the metalloproteinase MMP-1 to reveal a "non-canonical" receptor-activating tethered ligand that is distinct from that revealed by thrombin (39). The MMP-1-dependent activation of PAR 1 was shown to bias signaling toward the Rho-GTP and MAPK pathways. In addition, activated protein C has been shown to cause a PAR 1 -mediated increase in endothelial barrier integrity, whereas thrombin activation of PAR 1 decreases endothelial barrier function (40,41). Functional selectivity of the coupling of PAR 1 to different G-protein-mediated responses, with differences in coupling between the receptor activation by thrombin and agonist peptide activation of the receptor, has also been reported (42). Whether neutrophil proteinase-mediated disarming of either PAR 1 or PAR 4 triggers biased signaling by these receptors, as we describe here for PAR 2 , is currently unknown. Nonetheless, the influx of inflammatory neutrophils in response to injury or infection, via the actions of elastase, proteinase-3, and cathepsin-G acting on the PARs, could in principle stimulate a complex set of responses by activating PAR 4 (cathepsin-G) (7, 43), disarming PAR 1 and PAR 2 (cathepsin-G and proteinase-3) (7, 23), and selectively activating a unique arm of the PAR 2 MAPK signaling pathway (elastase). In summary, we identify in neutrophil elastase an endogenously expressed biased enzyme agonist for PAR 2 . We also propose that proteinase-triggered biased signaling via PAR 2 , shown here for the first time for neutrophil elastase, may underpin the distinct responses stimulated by PAR 2 in inflammatory settings.