Neutrophil Elastase Activates Protease-activated Receptor-2 (PAR2) and Transient Receptor Potential Vanilloid 4 (TRPV4) to Cause Inflammation and Pain*

Background: Proteases cleave protease-activated receptor-2 (PAR2), which activates transient receptor potential (TRP) ion channels to cause inflammation and pain. Results: Neutrophil elastase cleaves PAR2, resulting in Gαs-mediated cAMP formation, transient receptor potential vanilloid 4 (TRPV4) activation, and sensitization of nociceptive neurons, inflammation, and pain. Conclusion: Elastase causes PAR2- and TRPV4-mediated inflammation and pain. Significance: PARs and TRP channels mediate responses to diverse proteases. Proteases that cleave protease-activated receptor-2 (PAR2) at Arg36↓Ser37 reveal a tethered ligand that binds to the cleaved receptor. PAR2 activates transient receptor potential (TRP) channels of nociceptive neurons to induce neurogenic inflammation and pain. Although proteases that cleave PAR2 at non-canonical sites can trigger distinct signaling cascades, the functional importance of the PAR2-biased agonism is uncertain. We investigated whether neutrophil elastase, a biased agonist of PAR2, causes inflammation and pain by activating PAR2 and TRP vanilloid 4 (TRPV4). Elastase cleaved human PAR2 at Ala66↓Ser67 and Ser67↓Val68. Elastase stimulated PAR2-dependent cAMP accumulation and ERK1/2 activation, but not Ca2+ mobilization, in KNRK cells. Elastase induced PAR2 coupling to Gαs but not Gαq in HEK293 cells. Although elastase did not promote recruitment of G protein-coupled receptor kinase-2 (GRK2) or β-arrestin to PAR2, consistent with its inability to promote receptor endocytosis, elastase did stimulate GRK6 recruitment. Elastase caused PAR2-dependent sensitization of TRPV4 currents in Xenopus laevis oocytes by adenylyl cyclase- and protein kinase A (PKA)-dependent mechanisms. Elastase stimulated PAR2-dependent cAMP formation and ERK1/2 phosphorylation, and a PAR2- and TRPV4-mediated influx of extracellular Ca2+ in mouse nociceptors. Adenylyl cyclase and PKA-mediated elastase-induced activation of TRPV4 and hyperexcitability of nociceptors. Intraplantar injection of elastase to mice caused edema and mechanical hyperalgesia by PAR2- and TRPV4-mediated mechanisms. Thus, the elastase-biased agonism of PAR2 causes Gαs-dependent activation of adenylyl cyclase and PKA, which activates TRPV4 and sensitizes nociceptors to cause inflammation and pain. Our results identify a novel mechanism of elastase-induced activation of TRPV4 and expand the role of PAR2 as a mediator of protease-driven inflammation and pain.

Although pancreatic trypsins effectively activate PAR 2 , the identity of the proteases that activate PAR 2 under patho-physiological conditions is uncertain. In particular, given the widespread distribution of PAR 2 and the restricted expression of trypsin I/II to the pancreas, where activity is tightly controlled by endogenous inhibitors, the proteases that activate PAR 2 in tissues other than the pancreas and intestine remain to be identified. Other proteases that can cleave PAR 2 at Arg 36 2Ser 37 include trypsin IV (mesotrypsin) (17,18), tryptase (19,20), coagulation factors VIIa and Xa (21), acrosin (22), granzyme A (23), membrane-type serine protease 1 or matriptase (24), TMPRSS2 (25), and kallikrein 2, 4, 5, 6, and 14 (26 -29). These proteases would be expected to trigger the same canonical signaling events as trypsin I/II, leading to similar physiological outcomes. Alternatively, proteases that cleave PAR 2 at different sites can either remove the trypsin-revealed tethered ligand domain and thereby prevent activation by the canonical mechanism (i.e. disarm the receptor), or can activate distinct signaling mechanisms (i.e. biased agonism). Such proteases include cathepsin S, which cleaves at Gly 41 2Lys 42 (30) and Glu 56 2Thr 57 (31), neutrophil elastase, which cleaves at Ser 67 2Val 68 (32), cathepsin G, which cleaves at Phe 64 2Ser 65 (32), and proteinase 3, which cleaves at Val 61 2Asp 62 (32). However, the precise molecular mechanisms and the physiological consequences of biased protease signaling are poorly defined.
We investigated the mechanisms and patho-physiological outcomes of neutrophil elastase-induced biased agonism of PAR 2 . Elastase is one of the major proteases released from infiltrating neutrophils in inflamed tissues. Given its high circulating concentration (up to 1 M) and long half-life (6 -8 h) during inflammation, elastase has been proposed as a target for antiinflammatory therapy (33). Elastase is a biased agonist of both PAR 1 and PAR 2 , but by distinctly different mechanisms. Elastase cleaves PAR 1 at Leu 45 2Arg 46 , distal to the thrombin cleavage site, which reveals a tethered ligand domain (RNPND-KYEPF-NH 2 ) that activates G␣ i/o -mediated ERK signaling (34). Elastase cleaves PAR 2 at Ser 67 2Val 68 , distal to the trypsin cleavage site, which activates PAR 2 by a mechanism that does not involve exposure of a tethered ligand domain (32). How-ever, the functional importance of the elastase-biased agonism of PAR 2 is uncertain.
Given the important proinflammatory and pro-nociceptive actions of elastase, PAR 2 , and TRP channels, we investigated whether the elastase-biased agonism of PAR 2 activates TRPV4 and causes inflammation and pain. Our results reveal that elastase-activated PAR 2 robustly couples to G␣ s , leading to a PKA-dependent activation of TRPV4 and hypersensitivity of nociceptive neurons, and PAR 2 -and TRPV4-mediated inflammatory edema and mechanical hyperalgesia.
Materials-Human sputum elastase was from SERVA Electrophoresis GmbH ( Generation of cDNA Constructs, and Cell Culture-Generation of human PAR 2 and human TRPV4 constructs for expression in X. laevis oocytes have been described (38). PAR 2 constructs for expression in mammalian cells have been described (9,15). Human embryonic kidney (HEK) 293 cells and sarcoma virus-transformed rat kidney epithelial (KNRK) cells were maintained in DMEM with 10% fetal bovine serum (FBS) and 1% penicillin and streptomycin. Generation and maintenance of HEK293 and KNRK cells stably expressing human PAR 2 constructs have been described (12,15,39,40).
Elastase Cleavage of N-terminal PAR 2 Fragments-Peptides corresponding to N-terminal fragments of human PAR 2 (100 M) were incubated with elastase (10 units/ml (390 nM)) in Hanks' balanced salt solution, pH 7.4, for 1, 15, or 60 min at 37°C. Reactions were quenched with an equal volume of 50% acetonitrile and 0.1% trifluoroacetic acid in H 2 O. The reaction products were separated by reverse phase high pressure liquid chromatography using a Phenomenex Luna 3-m C8 column (100 Å, 100 ϫ 2 mm) with a gradient of 0 to 60% acetonitrile in 0.05% trifluoroacetic acid over 10 min. Products were identified by mass spectrometry using a Shimadzu LCMS 2000.
Two-electrode Voltage Clamp Studies of X. laevis Oocytes-Oocytes were obtained from adult X. laevis as described (38). Defolliculated stage V-VI oocytes were injected (Nanoject II automatic injector, Drummond) with 0.5 ng of TRPV4 cRNA alone, 10 ng of PAR 2 cRNA alone, or both TRPV4 and PAR 2 cRNA. Oocytes were studied 2 days after injection using the two-electrode voltage-clamp technique as described (37,38,44). A Ca 2ϩ -free solution was used to prevent activation of endogenous Ca 2ϩ -activated chloride channels by TRPV4-mediated Ca 2ϩ influx and to delay a Ca 2ϩ -induced decay of TRPV4 current (31,45). Oocytes were voltage-clamped at Ϫ60 mV. Oocytes were incubated with trypsin (2.48 units/ml (8 nM)), elastase (1 units/ml (3 M)), or vehicle, and were then challenged with the TRPV4 agonist GSK1016790A (50 nM) and the TRPV4 antagonist HC067047 (100 nM).
Mechanical Hyperalgesia and Edema in Mice-For behavioral assessments, C57BL/6 wild-type, PAR 2 Ϫ/Ϫ or Trpv4 Ϫ/Ϫ mice were acclimatized to the experimental room, restraint apparatus, and investigator for 2-h periods on 2 successive days before experiments, and the investigator was blinded to the experimental treatments. von Frey filaments were used to determine mechanical pain response as described (31). An increase in the filament stiffness required to induce paw withdrawal indicates mechanical analgesia, whereas a decrease in the filament stiffness required to induce withdrawal indicates mechanical hyperalgesia. To assess inflammatory edema of the paw, hind paw thickness was measured using digital calipers before and after treatments (46). Baseline von Frey scores were taken the day before the experiments. To examine the effects of elastase, mice were sedated with 5% isoflurane and elastase (100 units/ml (3.9 M), 10 l) or vehicle (0.9% NaCl, 10 l) was injected subcutaneously into the plantar surface of one hind paw. Mechanical hyperalgesia and edema were measured between 30 and 240 min after intraplantar injections.
Statistical Analyses-Results are expressed as mean Ϯ S.E. Differences between two groups were examined using unpaired t-tests. Differences between multiple groups were examined using an analysis of variance and a Bonferroni's or Dunnett's post hoc test. A p value Ͻ0.05 was considered to be significant.

Results
Elastase Cleaves PAR 2 at Ala 66 2Ser 67 and Ser 67 2Val 68 -To identify the sites at which elastase cleaves human PAR 2 , we incubated elastase (10 units/ml (390 nM)) with two synthetic peptides spanning the N terminus of PAR 2 including trypsin cleavage site (Arg 36 2Ser 37 ), cathepsin S cleavage site (Glu 56 2Thr 57 ), and the remaining extracellular domain up to the first transmembrane domain (residues 31-75) (Fig. 1A). Elastase rapidly cleaved PAR 2 peptide 2 (residues 61-75) but not PAR 2 peptide 1 (residues 31-60) (Fig. 1, B-D). We observed ϳ50% degradation of peptide 2 at 15 min and almost complete degradation after 60 min (Fig. 1D). The products that were identified by mass spectrometry are consistent with elastase cleaving human PAR 2 at Ala 66 2Ser 67 and Ser 67 2Val 68 (Fig. 1B). This finding is consistent with a previous report in which a lower concentration of elastase was used to cleave PAR 2 fragments (32). Elastase Stimulates PAR 2 -dependent cAMP Accumulation but Not Ca 2ϩ Mobilization-Elastase cleavage of PAR 2 at Ser 67 2Val 68 stimulates G␣ 12/13 -and Rho kinase-dependent phosphorylation of ERK1/2 (32). Whether elastase activates other signaling pathways is uncertain. Cathepsin S, a lysosomal protease that is also a biased agonist of PAR 2 , stimulates G␣ smediated cAMP accumulation (31). To investigate whether elastase also stimulates PAR 2 -dependent cAMP activation, we compared responses of KNRK cells expressing empty vector control (KNRK-VC) or human PAR 2 (KNRK-hPAR 2 ). As previously reported, trypsin stimulated a concentration-dependent increase in [Ca 2ϩ ] i in KNRK-PAR 2 cells but not KNRK-VC cells (12,31) (Fig. 2, A and B). Conversely, elastase had no effect on [Ca 2ϩ ] i at any concentration in both cell lines (Fig. 2, A and B).
Elastase Induces PAR 2 Coupling to G␣ s but Not G␣ q -cAMP formation may due to direct activation of G␣ s , which activates adenylyl cyclase, or a result of indirect activation of other signaling pathways, such as the G q pathway via activation of Ca 2ϩ -sensitive adenylyl cyclases (47). As an indicator of direct activation of G proteins, we used BRET to examine the change in proximity between PAR 2 and heterotrimeric G proteins in HEK293 cells (31). We transiently expressed PAR 2 -RLuc8 with G␤ 1 , G␥2-Venus and different G␣ proteins, including G␣ s and G␣ q . Changes in BRET ratios between PAR 2 -RLuc8 and G␥2-Venus were monitored in cells treated with trypsin or elastase. Trypsin (10 units/ml (34.7 nM)) stimulated and increased the BRET signal in cells expressing either G␣ q or G␣ s (Fig. 3, A and  B). In contrast, elastase (10 units/ml (390 nM)) only induced a change in BRET in cells expressing G␣ s but not G␣ q (Fig. 3, A  and B). These findings are consistent with our signaling data showing that trypsin-activated PAR 2 mobilizes Ca 2ϩ and generates cAMP, whereas elastase-activated PAR 2 generates cAMP but does not mobilize Ca 2ϩ . They suggest that elastase activates PAR 2 -dependent G␣ s signaling that leads to cAMP formation. Notably, in G␣ s -expressing cells, whereas trypsin increased BRET, elastase decreased BRET. This observation suggests that trypsin-activated PAR 2 adopts a different conformation from elastase-activated PAR 2 relative to G proteins. The G␥ 2 -Venus fluorescence was similar in all experiments, suggesting comparable levels of expression (Fig. 3C).
Previous studies using immunofluorescence and confocal microscopy suggest that whereas trypsin-activated PAR 2 undergoes endocytosis, elastase-activated PAR 2 remains at the plasma membrane (32,49). To quantitatively assess PAR 2 trafficking at the plasma membrane, we measured the BRET signal between PAR 2 -RLuc8 and two proteins that reside at the plasma membrane: KRas-Venus, which is present in cholesterol-independent microdomains (50), and RIT-Venus, which is uniformly distributed. Trypsin (10 units/ml (34.7 nM)) but not elastase (10 units/ml (390 nM)) induced a rapid decrease in BRET between PAR 2 -RLuc8 and both RIT-Venus and KRas-Venus (Fig. 4, D and E). These results are consistent with the observation that trypsin-but not elastase-activated PAR 2 couples to ␤-arrestins and internalizes.
Elastase Evokes PAR 2 -dependent Sensitization of TRPV4 Currents-After activation by canonical agonists such as trypsin, or biased agonists such as cathepsin S, PAR 2 activates and sensitizes TRPV4, which contributes to neurogenic inflammation and mechanical hyperalgesia (8,31). To evaluate whether elastase-activated PAR 2 sensitizes TRPV4, we expressed TRPV4 alone or both PAR 2 and TRPV4 together in X. laevis oocytes, and measured whole cell currents using the two-electrode voltage-clamp technique. Preincubation of oocytes expressing PAR 2 and TRPV4 with elastase (1 units/ml (3 M)) for 5 min resulted in a 6-fold increase in the response to the TRPV4-selective agonist GSK1016790A (50 nM) compared with vehicle (Fig. 5, A and B, second and fourth columns). In contrast, elastase did not affect the response to GSK1016790A in oocytes expressing TRPV4 but not PAR 2 (Fig. 5B, first and  third columns). These results suggest that elastase induces a PAR 2 -dependent sensitization of TRPV4 in X. laevis oocytes, as we have previously described (38). To evaluate the mechanism of this sensitization, we preincubated oocytes with a PKC inhibitor GF109203X (1 M), adenylyl cyclase inhibitor SQ22536 (20 M), or PKA inhibitor PKI (10 M). SQ22536 and PKI prevented elastase-induced and PAR 2 -mediated sensitization of TRPV4, whereas GF109203X partially inhibited sensitization (Fig. 5, A and B, fifth to seventh columns). All three inhibitors had no effect on GSK1016790A-stimulated TRPV4 currents in oocytes expressing TRPV4 alone (data not shown). In control experiments, elastase activation of the epithelial sodium channel was preserved in the presence of protein kinase inhibitors (data not shown). Thus, the inhibitory effect of the kinase inhibitors on the PAR 2 -mediated sensitization of TRPV4 by elastase is not due to an inhibition of enzymatic activity. These results suggest that adenylyl cyclase and PKA are critically important for elastase-stimulated and PAR 2 -dependent sensitization of TRPV4, and are consistent with the capacity of elastase-activated PAR 2 to couple to G␣ s and cAMP accumulation.
Because elastase cleaves PAR 2 distal from the trypsin site, elastase can remove the trypsin-exposed tethered ligand and disarm the receptor to subsequent activation by trypsin (32,49).
Elastase Activates Endogenous PAR 2 and TRPV4 in DRG Neurons-Proteases can activate PAR 2 on nociceptive neurons to stimulate the release of neuropeptides in peripheral tissues, resulting in neurogenic inflammation, and in the dorsal horn of the spinal cord, leading to pain transmission (5, 6). PAR 2 activates TRP channels, which amplify the proinflammatory and pro-nociceptive actions of proteases (8,11,13). We have previously reported that proteases that activate PAR 2 by canonical mechanisms, such as trypsins and tryptase (6,7), and biased mechanisms, including cathepsin S (31), can signal to nociceptors by activating PAR 2 and TRP channels. To determine whether elastase similarly signals to nociceptors, we assessed cAMP accumulation, ERK1/2 activation, and [Ca 2ϩ ] i in mouse DRG neurons in short term culture.
In DRG from wild-type mice, trypsin (10 units/ml) and elastase (1 or 10 units/ml) stimulated ERK1/2 activation (Fig. 6A). Elastase robustly stimulated ERK1/2 activation to a similar degree at both 1 (39 nM) and 10 units/ml (390 nM). Elastase (10 units/ml) did not stimulate ERK1/2 activation in DRG from Par 2 Ϫ/Ϫ mice, but instead inhibited ERK1/2 activation by mechanisms that remain to be elucidated (Fig. 6B). Similarly, trypsin and elastase (both 10 units/ml) stimulated cAMP accumulation in DRG from wild-type mice (Fig. 6C), and the stimulatory effect of elastase was not observed in DRG from Par 2 Ϫ/Ϫ mice (Fig. 6D). These results suggest that elastase stimulates ERK1/2 activation and cAMP accumulation in DRG by a PAR 2 -dependent process, consistent with our current observations and with published studies showing similar stimulator actions of elastase in KNRK-PAR 2 but not KNRK-VC cells (32).
To investigate whether elastase cleavage of PAR 2 leads to activation of TRP channels, we measured [Ca 2ϩ ] i in small to medium diameter neurons. Elastase (10 units/ml) stimulated a rapid and sustained increase in [Ca 2ϩ ] i in neurons from wildtype mice (Fig. 7A). Responses to elastase were detected in ϳ50% of these neurons, similar to the proportion of neurons responding to trypsin (Fig. 7B). The magnitude of response to elastase was markedly diminished in neurons from Par 2 Ϫ/Ϫ mice (Fig. 7C) and Trpv4 Ϫ/Ϫ mice (Fig. 7D). Fewer neurons from Par 2 Ϫ/Ϫ mice and Trpv4 Ϫ/Ϫ mice responded to elastase (Fig. 7E). Removal of extracellular Ca 2ϩ ions strongly inhibited the proportion of elastase-responsive neurons in wild-type mice (Fig. 7F). Our results suggest that elastase activates PAR 2 FIGURE 5. Elastase-induced sensitization of TRPV4 and disarming of PAR 2 . A and B, X. laevis oocytes co-expressing TRPV4 and PAR 2 or expressing TRPV4 alone were preincubated with vehicle or elastase for 5 min. Elastasetreated oocytes were treated with vehicle or inhibitors of adenylyl cyclase (SQ22536), PKA (PKI), or PKC (GF109203X). Oocytes were treated with the TRPV4 agonist GSK1016790A (GSK) and the TRPV4 antagonist HC067047 (HC) to study TRPV4 currents. A, representative traces from oocytes co-expressing TRPV4 and PAR 2 . B, mean ⌬I GSK1016790A values of pooled data from oocytes expressing TRPV4 alone or TRPV4 and PAR 2 . C, oocytes expressing PAR 2 were preincubated with vehicle (veh) or elastase for 5 min, washed (W), and trypsinevoked whole cell currents were measured. Columns represent mean ⌬I trypsin values. n indicates number of individual oocytes measured. N indicates the number of batches of oocytes. ***, p Ͻ 0.001, unpaired t test; § § §, p Ͻ 0.0001, unpaired t test.   (F and H), and one-way analysis of variance (E and G) compared with vehicle or wild-type groups. on nociceptors, which in turn activates TRPV4, allowing the influx of extracellular Ca 2ϩ ions. The residual responses in mice lacking PAR 2 may be attributable to activation of other PARs, including PAR 1 , which is also expressed by nociceptors (51) and which responds to elastase (34), or due to elastase activation of ion channels, such as the epithelial sodium channel (52,53). Residual responses in TRPV4-deficient mice may be due to PAR 2 -dependent activation of other TRPs, such as TRPV1 or TRAPA1 (7,11).
To characterize the signaling mechanism underlying elastase-stimulated and PAR 2 -dependent activation of TRPV4, we pretreated neurons with inhibitors of adenylyl cyclase, PKA, PKC, or Rho kinase. The adenylyl cyclase inhibitor SQ22536 (20 M), the PKA inhibitor PKI (10 M), and the Rho kinase inhibitor Y27362 (10 M) all reduced the percentage of neurons responsive to elastase, whereas the PKC inhibitor GF109203X (1 M) had no effect (Fig. 7G). These results suggest that elastase-cleaved PAR 2 activates TRPV4 in nociceptors by G␣ s -, adenylyl cyclase-, and PKA-dependent mechanisms, as well as by a Rho kinase-dependent process. They are consistent with the role of adenylyl cyclase and PKA in mediating elastase-stimulated and PAR 2 -dependent sensitization of TRPV4 in X. laevis oocytes. Our findings also agree with the reports that elastaseactivated PAR 2 stimulates Rho kinase (32), and that Rho kinase contributes to PAR 2 -mediated TRPV4 sensitization in oocytes (38). Preincubation of elastase with its specific inhibitor elafin (10 M, 30 min) abolished elastase stimulation of [Ca 2ϩ ] i in nociceptors, indicating a requirement for proteolytic activity (Fig. 7H).
Elastase Evokes PKA-dependent, and PKC-and Rho Kinaseindependent Hyperexcitability of Nociceptive Neurons-Hypersensitivity of nociceptors can lead to exacerbated pain transmission. We have previously shown that canonical agonists, such as trypsin and tryptase (54,55) and biased agonists, such as cathepsin S (31), cause hyperexcitability of DRG neurons from mice. To directly determine whether elastase induces hyperexcitability, we made patch clamp recordings from small diameter DRG neurons in short term culture. We assessed excitability by measuring the rheobase (minimum current required to fire a single action potential) and the action potential discharge frequency at twice rheobase. Preincubation with elastase (10 units/ml (390 nM), 1 h) resulted in a Ͼ50% reduction in rheobase, which is indicative of hyperexcitability (Fig. 8, A and B). Elastase-evoked hyperexcitability was abolished by the PKA inhibitor PKI (10 M), but unaffected by the PKC inhibitor GF109203X (1 M) or the Rho kinase inhibitor Y-27632 (10 M) (Fig. 8, A and B). Elastase did not affect the frequency of action potential discharge at the current to twice rheobase (data not shown). Our results suggest that elastase causes hyperexcitability of nociceptive neurons by a PKA-dependent process, consistent with elastase stimulation of PAR 2 -dependent cAMP accumulation in DRG.
Elastase Evokes Inflammation and Mechanical Hyperalgesia by Activating PAR 2 and TRPV4 -Both canonical agonists, such as trypsin and tryptase, and biased agonists, such as cathepsin S, can activate PAR 2 and TRP channels at the periphery terminals of primary nociceptive neurons, which causes neurogenic inflammation and hyperalgesia (5-8, 31). We investigated whether elastase, as a biased agonist for PAR 2 , induces inflammation and pain by activating PAR 2 and TRPV4. We administered elastase (100 units/ml (3.9 M), 10 l) to mice by intraplantar injection, and measured mechanical pain using calibrated von Frey filaments applied to the plantar surface of the paw, and assessed inflammatory edema by measuring paw thickness with calipers. In wild-type mice, elastase induced marked mechanical hyperalgesia that was sustained for at least 4 h (Fig. 9A). Elastase also induced a Ͼ3-fold increase in paw thickness that declined after 1 h but was sustained for at least 4 h (Fig. 9B). Deletion of Par2 attenuated elastase-evoked mechanical hyperalgesia and inflammatory edema. Deletion of Trpv4 attenuated elastase-induced inflammation to a similar degree as deletion of Par2. However, Trpv4 deletion caused a greater reduction in elastase-induced hyperalgesia than Par2 deletion. Thus, both PAR 2 and TRPV4 contribute to elastaseevoked inflammation and pain. The residual pain and inflammation observed in Par 2 Ϫ/Ϫ and Trpv4 Ϫ/Ϫ mice may be attributable to elastase activation of other PARs or ion channels.

Discussion
We report that neutrophil elastase is a biased agonist of PAR 2 that stimulates G␣ s -mediated cAMP accumulation and ERK1/2 activation in cell lines expressing human PAR 2 and in DRG neurons from mice. In X. laevis oocytes and mouse nociceptive neurons, elastase activation of PAR 2 sensitizes and activates TRPV4 by adenylyl cyclase-and PKA-dependent mechanisms. Elastase causes PKA-dependent hyperexcitability of nociceptors, and evokes PAR 2 -and TRPV4-mediated mechanical hyperalgesia and inflammation. Elastase Is a Biased Agonist of PAR 2 That Stimulates a Distinct Signaling Profile-We observed that elastase-activated PAR 2 generates a signaling profile that is distinctly different from that evoked by trypsin. By using BRET to examine the proximity of PAR 2 to G␥ in HEK293 cells, we found that elastase stimulated G␣ s -but not G␣ q -dependent BRET, whereas trypsin stimulated a G␣ s -and G␣ q -dependent BRET. These results suggest that whereas elastase-activated PAR 2 couples to G␣ s alone, trypsin-activated PAR 2 couples to G␣ s and G␣ q . Our BRET results are consistent with the observations that elastase stimulates accumulation of cAMP but not mobilization of intracellular Ca 2ϩ in KNRK-PAR 2 cells, whereas trypsin stimulates both cAMP accumulation and Ca 2ϩ mobilization. These responses are PAR 2 -dependent, because elastase and trypsin did not affect cAMP or Ca 2ϩ levels in KNRK-VC cells. Our results confirm the inability of elastase to mobilize intracellular Ca 2ϩ in KNRK-PAR 2 cells (32). Elastase also stimulates a PAR 2dependent activation of ERK1/2 in KNRK-PAR 2 cells that is blocked by a Rho kinase inhibitor, suggesting a G␣ 12/13 -mediated mechanism (32). Considered together, our findings are consistent with the proposal that elastase is a biased agonist of PAR 2 that stimulates PAR 2 -coupling to G␣ s , accumulation of cAMP, and activation of ERK1/2, but which is unable to evoke G␣ q -mediated Ca 2ϩ signaling. In this regard, elastase resembles cathepsin S, another PAR 2 biased agonist that selectively stimulates coupling to G␣ s , cAMP accumulation, but not Ca 2ϩ mobilization (31).
The structural basis that underlies the different signaling properties of trypsin-, elastase-, and cathepsin S-cleaved PAR 2 remains to be determined. However, because these proteases cleave PAR 2 at distinct sites, and promote activation by different mechanisms, differences in receptor conformation are likely to underlie these divergent signaling mechanisms. Whereas trypsin cleaves human PAR 2 at Arg 36 2Ser 37 to reveal the tethered ligand SLIGKV (4), and cathepsin S cleaves at Glu 56 2Thr 57 to expose the tethered ligand TVFSVDEFSA, elastase acts distal to both the trypsin and cathepsin S sites at Ser 67 2Val 68 (32). Given the close proximity of the elastase site to the first transmembrane domain, it is not surprising that elastase activates PAR 2 by a mechanism that does not involve exposure of a tethered ligand (32). The observation in G␣ sexpressing cells that elastase and trypsin induced opposite changes in BRET between PAR 2 -RLuc8 and G␥-Venus suggests that elastase-cleaved PAR 2 adopts a different conformation from trypsin-cleaved PAR 2 relative to G␥. BRET analysis also suggest that cathepsin S-cleaved PAR 2 adopts a different conformation from the trypsin-cleaved receptor (31). Whether these differences underlie the biased agonism of elastase and cathepsin S remains to be determined.
Elastase cleavage of PAR 2 was first reported in vitro in Escherichia coli expressing the N-terminal domain of the receptor (56). Because elastase cleaves PAR 2 distal to the trypsin cleavage site, elastase pretreatment can prevent subsequent activation by trypsin, and thereby "disarm" the receptor (49,56). We found that pretreatment with elastase prevented trypsin signals in oocytes, which confirms this mechanism. Cathepsin S can similarly disarm PAR 2 in both mammalian cells and oocytes (31).
Regulation and Trafficking of Elastase-activated PAR 2 -After activation by trypsin, PAR 2 is phosphorylated by second messenger kinases and GRKs, and associates with ␤-arrestins, which mediate receptor desensitization and endocytosis (15,39). We observed that trypsin stimulated a large and sustained increase in BRET between PAR 2 and GRK2 and ␤-arrestins, consistent with GRK2 and ␤-arrestin recruitment. This recruitment coincided with a decreased BRET between PAR 2 and the resident plasma membrane proteins RIT and K-Ras, suggesting trypsin-induced endocytosis of PAR 2 . Conversely, we observed that elastase-activated PAR 2 was unable to recruit GRK2, interact with ␤-arrestins, or undergo endocytosis. The lack of GRK2 recruitment by elastase-activated PAR 2 is consistent with the lack of ␤-arrestin recruitment, which is known to rely on GRKmediated phosphorylation of agonist-occupied receptors (57). Our findings support a previous report that elastase-activated PAR 2 is unable to recruit ␤-arrestins (32). Cathepsin S-activated PAR 2 also fails to recruit ␤-arrestins and does not internalize (31). Further studies are required to define the mechanisms that regulate signaling of elastase-and cathepsin S-activated PAR 2 .
The inability of elastase to promote the recruitment of GRK2 and ␤-arrestins to PAR 2 and to stimulate receptor endocytosis FIGURE 9. Elastase-evoked and PAR 2 -and TRPV4-dependent inflammation and pain. Elastase or vehicle was administered by intraplantar injection to wild-type, PAR 2 Ϫ/Ϫ , and Trpv4 Ϫ/Ϫ mice. A, mechanical hyperalgesia was measured using von Frey filaments. B, paw edema was assessed by measurement of paw thickness with calipers. *, p Ͻ 0.05; ****, p Ͻ 0.0001 two-way analysis of variance compared with vehicle. n ϭ 3-12 mice.
is likely to affect signaling of elastase-activated PAR 2 . Trypsinactivated PAR 2 undergoes ␤-arrestin-dependent endocytosis in KNRK-PAR 2 cells. ␤-Arrestins assemble an endosomal signalosome comprising PAR 2 , Raf, and MEKK that is necessary for activation and cytosolic retention of ERK1/2 (15). A mutant PAR 2 that is unable to associate with ␤-arrestins and internalize activates nuclear ERK1/2 by a mechanism that involves transactivation of the epidermal growth factor receptor. Because elastase-cleaved PAR 2 neither recruits ␤-arrestins nor internalizes, it may only activate nuclear ERK1/2 but not cytosolic ERK1/2. Further experiments will be required to investigate the mechanisms by which elastase-activated PAR 2 stimulates ERK1/2, and to determine the importance of PAR 2 trafficking for protease-evoked inflammation and pain.
Because elastase was unable to stimulate GRK2 translocation to PAR 2 at the plasma membrane, we investigated the possibility that elastase alters the potential association of PAR 2 with plasma membrane-localized GRKs. Of 7 known GRKs, GRK4, GRK5, and GRK6 are primarily localized to plasma membrane (58,59), and GRK6 is expressed in spinal cord and enteric neurons that also express PAR 2 (60 -62). Deletion of GRK6 leads to an increase in capsaicin-induced post-colitis hyperalgesia, suggesting that GRK6, like PAR 2 , contributes to post-inflammatory pain (62). Here we show that both trypsin and elastase lead to a decrease in BRET between PAR 2 and GRK6. These data suggests that GRK6 contributes to PAR 2 regulation or signaling in response to proteases that activate the receptor by canonical and biased mechanisms. Because GRK6 is primarily located at the plasma membrane, a decrease in the BRET signal may reflect PAR 2 moving away from the membrane. However, this possibility is unlikely because elastase does not cause PAR 2 internalization, demonstrated by both immunofluorescence (32) and our BRET analysis of the association of PAR 2 with plasma membrane-resident proteins.
Elastase Activates Nociceptive DRG Neurons, and Leads to PAR 2 -and TRPV4-dependent Inflammation and Pain-Our results show that elastase stimulates cAMP accumulation and ERK1/2 activation in DRG from wild-type but not Par 2 Ϫ/Ϫ mice, which is consistent with our findings in KNRK-PAR 2 and KNRK-VC cells. However, in marked contrast to observations in KNRK-PAR 2 cells, in which elastase failed to increase [Ca 2ϩ ] i , elastase stimulated a robust increase in [Ca 2ϩ ] i in DRG neurons from wild-type mice. Elastase-evoked Ca 2ϩ signals in neurons from wild-type mice were strongly inhibited by removal of extracellular Ca 2ϩ ions, and also suppressed by deletion of Par2 or Trpv4. These results are consistent with the proposal that elastase-activated PAR 2 triggers the activation of TRPV4, which allows influx of extracellular Ca 2ϩ ions. Residual responses to elastase in Par 2 Ϫ/Ϫ and Trpv4 Ϫ/Ϫ mice may be due to activation of other PARs and TRP channels.
We used pharmacological inhibitors to characterize the mechanism by which elastase-activated PAR 2 may stimulate TRPV4 in DRG neurons. Adenylyl cyclase, PKA, and PKC are the major effectors of G␣ q and G␣ s signaling. To investigate their involvement in elastase-mediated Ca 2ϩ signaling, we treated neurons with inhibitors of adenylyl cyclase, PKA, and PKC. Both adenylyl cyclase and PKA inhibitors attenuated elastase-evoked Ca 2ϩ influx. However, inhibition of PKC did not affect Ca 2ϩ influx. Considered together, these results suggest that in DRG neurons elastase-activated PAR 2 causes an adenylyl cyclase-and PKA-dependent activation of TRPV4. They are consistent with the known involvement of PKA in regulating TRPV4 (63), and our observation that cathepsin S also causes a PKA-mediated activation of TRPV4 in DRG neurons (31). PAR 2 coupling to TRPV4 has been observed in multiple systems with a variety of PAR 2 agonists. However, the mechanism of coupling is agonist-dependent. Trypsin-activated PAR 2 stimulates TRPV4 in sensory neurons by PKC-and tyrosine kinase-dependent processes, and by the generation of arachidonic acid metabolites (8,9). In contrast, diesel exhaust particles can activates the PAR 2 -TRPV4 axis in airway epithelial cells by a G i/o and phosphatidylinositide 3-kinase-dependent mechanism (64).
Elastase also increased the excitability of DRG nociceptors, as revealed by a decrease in the input current required to fire action potentials. Although further studies are required to identify the mechanism of elastase-evoked neuronal sensitization, this sensitization was also suppressed by an inhibitor of PKA but unaffected by a PKC inhibitor, in line with our previous observations of cathepsin S-evoked neuronal hyperexcitability (31).
In X. laevis oocytes co-expressing PAR 2 and TRPV4, we observed that pretreatment with elastase caused a 6-fold increase in GSK1016790A-stimulated current, indicative of TRPV4 sensitization. This effect was not observed in oocytes expressing TRPV4 alone, and is thus PAR 2 -dependent. Inhibitors of adenylyl cyclase and PKA abolished elastase-evoked sensitization of TRPV4, which is consistent with observations in neurons. However, in oocytes and PKC the inhibitor also partially inhibited sensitization, whereas a PKC inhibitor had no effect on elastase-induced sensitization of TRPV4 in DRG neurons. Species differences may account for this discrepancy.
We found that intraplantar injection of elastase leads to sustained mechanical hyperalgesia and inflammatory edema. Deletion of Par 2 partly blocked elastase-mediated inflammation and pain, whereas deletion of Trpv4 strongly inhibited both pain and inflammation. The residual effects of elastase in Par 2 Ϫ/Ϫ and Trpv4 Ϫ/Ϫ mice may be attributable to activation of other PARs and TRP channels that are expressed by nociceptors.
Infiltration of neutrophils is a essential component of the host defense mechanism against invading pathogens during acute inflammation. Proteases such as elastase, proteinase 3, and cathepsin G, which are released from neutrophils during inflammation, mediate killing of invading microorganisms. Our results suggest that proteases such as elastase can also cleave PAR 2 and TRPV4 on sensory nerves to induce the acute neurogenic inflammation and pain, which may also serve as an acute protective mechanism.