Development and Evaluation of Small Peptidomimetic Ligands to Protease-Activated Receptor-2 (PAR2) through the Use of Lipid Tethering

Protease-activated receptor-2 (PAR2) is a G-Protein Coupled Receptor (GPCR) activated by proteolytic cleavage to expose an attached, tethered ligand (SLIGRL). We evaluated the ability for lipid-tethered-peptidomimetics to activate PAR2 with in vitro physiological and Ca2+ signaling assays to determine minimal components necessary for potent, specific and full PAR2 activation. A known PAR2 activating compound containing a hexadecyl (Hdc) lipid via three polyethylene glycol (PEG) linkers (2at-LIGRL-PEG 3-Hdc) provided a potent agonist starting point (physiological EC50 = 1.4 nM; 95% CI: 1.2–2.3 nM). In a set of truncated analogs, 2at-LIGR-PEG 3-Hdc retained potency (EC50 = 2.1 nM; 1.3–3.4 nM) with improved selectivity for PAR2 over Mas1 related G-protein coupled receptor type C11, a GPCR that can be activated by the PAR2 peptide agonist, SLIGRL-NH2. 2at-LIG-PEG 3-Hdc was the smallest full PAR2 agonist, albeit with a reduced EC50 (46 nM; 20–100 nM). 2at-LI-PEG 3-Hdc retained specific activity for PAR2 with reduced EC50 (310 nM; 260–360 nM) but displayed partial PAR2 activation in both physiological and Ca2+ signaling assays. Further truncation (2at-L-PEG 3-Hdc and 2at-PEG 3-Hdc) eliminated in vitro activity. When used in vivo, full and partial PAR2 in vitro agonists evoked mechanical hypersensitivity at a 15 pmole dose while 2at-L-PEG 3-Hdc lacked efficacy. Minimum peptidomimetic PAR2 agonists were developed with known heterocycle substitutes for Ser1 (isoxazole or aminothiazoyl) and cyclohexylalanine (Cha) as a substitute for Leu2. Both heterocycle-tetrapeptide and heterocycle-dipeptides displayed PAR2 specificity, however, only the heterocycle-tetrapeptides displayed full PAR2 agonism. Using the lipid-tethered-peptidomimetic approach we have developed novel structure activity relationships for PAR2 that allows for selective probing of PAR2 function across a broad range of physiological systems.


Introduction
Protease-activated receptors (PARs) are a sub-family of Gprotein coupled receptors (GPCRs) that have a unique mode of activation. PARs contain an embedded ligand that is exposed following proteolytic cleavage of the extracellular oriented NH 2 terminus [1]. The different N-termini of the PARs present substrates for a variety of proteases that create selective activation (or inactivation) mechanisms for signal transduction [2,3,4]. The most common, diffusionally limited ''tethered ligand'' uncovered following trypsin-like serine protease activity of PAR 2 [exposing SLIGKV (human) or SLIGRL (rodent)] serves as a potent agonist to the receptor. As an obvious consequence of its activation mechanism, PAR 2 is associated with pathologies that have a strong protease release, including inflammatory related diseases such as arthritis, asthma, inflammatory bowel disease, sepsis, and pain disorders [1,2,4]. Stimulation of PAR 2 in pain-sensing primary sensory neurons (nociceptors) leads to the sensitization of a variety of receptors including the noxious heat and capsaicin receptor TRPV1 [5,6,7]. This sensitization of sensory neuronal channels underlies thermal [7,8,9] or mechanical hypersensitivity [8,10,11] elicited by activation of PAR 2 . The involvement of PAR 2 in pain and other pathologies makes it a prime target for drug discovery. Importantly, PAR 2 has been associated with itch based partly on data obtained using the relatively potent PAR 2 signaling peptide, SLIGRL-NH 2 . It is now clear that this peptide also stimulates an additional GPCR, Mas1 related G-protein coupled receptor type C11 (MrgprC11), and this receptor is responsible for the pruritic properties of SLIGRL-NH 2 [12]. Therefore, assessing the selectivity of PAR 2 ligands against receptors that are selectively expressed in sensory ganglia (e.g., MrgprC11; [13,14]) is critical to developing selective probes for PAR 2 .
Small peptides or peptidomimetics that mimic the ligand binding properties of the tethered ligand exposed by proteolysis of the N-terminus of the receptor have been used to directly activate PARs [2,15,16,17]. Activating peptides (e.g., SLIGKV-NH 2 and SLIGRL-NH 2 ) and peptidomimetics (e.g., 2-furoyl-LIGRLO-NH 2 [18] and 2at-LIGRL-NH 2 [19]) have provided useful tools for establishment of structure-activity relationships (SAR) and rational drug design because they limit off-target effects that are often a complication of natural protease activation. Early SAR studies suggested that the minimal peptide sequence required for PAR 2 activation is a pentamer (either SLIGR-NH 2 or the less potent LIGRL-NH 2 [17,20]). More recently, heterocycle-dipeptide mimetics have been shown to retain PAR 2 activity [21]. However, full characterization of these shortened compounds has been hindered by a lack of assays sufficiently sensitive to evaluate full concentration responses. Commonly used assays require high concentrations (. 50 mM) that potentially limit PAR 2 -selectivity or prevent full solubility for preferred Ca 2+ activation studies [21]. It is now evident that a variety of GPCRs, including PAR 2 , can elicit signaling pathway-specific activation with distinct physiological responses [4,22,23,24,25,26]. A means to establish better evaluation of the minimal peptidomimetic structure required for full PAR 2 activation would benefit PAR 2 ligand discovery efforts.
Lipidation of peptide receptor agonists has been used to increase their potency via a variety of mechanisms [27]. Because of the naturally tethered ligands in PAR 2 , we hypothesized that lipidation of peptide and peptidomimetic agonists could provide a membrane bound tether to better mimic the natural receptor activation and thus increase their potencies [28]. Modification of the potent PAR 2 peptidomimetic agonists 2at-LIGRL-NH 2 and 2at-LIGRLO-NH 2 with polyethylene glycol (PEG) spacers and a hexadecyl (Hdc) or a palmitoyl (Pam) group (e.g., 2at-LIGRL-PEG 3 -Hdc or 2at-LIGRLO(PEG 3 -Pam)-NH 2 ) improves ligand potency to the low nanomolar range without sacrificing specificity to PAR 2 as demonstrated in cell lines or in cells isolated from PAR 2 wild type vs. PAR 2 -/mice [28]. Because of this increased potency, we hypothesized that this synthetic tethered ligand (STL) approach could be used to more closely examine SAR of peptidomimetics in an effort to better understand the minimal components necessary to specifically activate PAR 2 . In this report, we used the STL approach coupled with real time cell analysis (RTCA) and digital Ca 2+ imaging microscopy to evaluate 14 compounds. We describe six STL compounds consisting of full or truncated parent peptidomimetic (2at-LIGRL-NH 2 ) linked to three PEGs and one Hdc and evaluate their potencies, efficacies and specificities at PAR 2 , including screening against MrgprC11, to determine a minimal sequence necessary for specific activation of PAR 2 in vitro and in vivo. Moreover, we used a parallel approach to fully evaluate potency of heterocycle di-and tetra-peptide mimetics using Ser 1 and Leu 2 substitutions known to activate PAR 2 [19,21]. These findings identify a minimal structure required for specific full and/or partial activation of PAR 2 and thus, further elucidate highly potent and specific probes to examine the function of this receptor in vitro and in vivo.
Solid Phase Synthesis (Figure 1, steps ii-iii) The aliquot of secondary amide resin from the previous step (10 mmol) was swollen in DCM, washed with tetrahydrofuran-DCM, and the Fmoc-PEG was coupled via symmetrical anhydride (6 equiv of N a -Fmoc-PEG and 3 equivalents of DIC in tetrahydrofuran-DCM) overnight. An on-resin test using Bromophenol Blue was used for qualitative and continuous monitoring of reaction progress. Fmoc group was removed with 10% piperidine in DMF (2 min + 20 min). The resin was washed with DMF (3X), DCM (3X), 0.2 M HOBt in DMF (2X), and finally with DMF (2X). Fmoc-PEG and following Fmoc-protected amino acid were coupled using pre-activated 0.3 M HOBt ester in DMF (3 equiv of N a -Fmoc amino acid or Fmoc-PEG, 3 equivalents of HOBt and 3 equivalents of DIC) monitored by Bromophenol Blue test. To avoid deletion sequences and slower coupling rate in longer sequences, the double coupling was performed at all steps with 3 equivalents of amino acid or Fmoc-PEG, 3 equivalents of HBTU and 6 equivalents of DIEA in DMF. Wherever beads still tested Kaiser positive, a third coupling was performed using the symmetric anhydride method (2 equivalent of amino acid and 1 equivalents of DIC in DCM). Any unreacted NH 2 groups on the resin thereafter were capped using an excess of 50% acetic anhydride in pyridine for 5 min. When the coupling reaction was finished, the resin was washed with DMF, and the same procedure was repeated for the next amino acid until all amino acids were coupled. 2-aminothiazole-4-carboxylic acid and 5-isoxazole-carboxylic acid were attached to the resin as symmetrical anhydride (6 equivalents of acid and 3 equivalents of DIC in DCM-DMF).
Cleavage of Ligand from the Resin (Figure 1, step iv) A cleavage cocktail (10 mL per 1 g of resin) of TFA (91%), water (3%), triisopropylsilane (3%), and thioanisole (3%) was injected into the resin and stirred for 4 hr at room temperature. The crude ligand was isolated from the resin by filtration, the filtrate was reduced to low volume by evaporation using a stream of nitrogen, and the ligand was precipitated in ice-cold diethyl ether, washed several times with ether, dried, dissolved in water and lyophilized to give off-white solid powders that were stored at 220uC until purified. The crude compound was purified by sizeexclusion chromatography and preparative HPLC.

Analytical Evaluation
The purity of products was checked by analytical Reverse Phase-HPLC using a Waters Alliance 2695 Separation Model with a Waters 2487 dual wavelength detector (220 and 280 nm) on a reverse phase column (Waters Symmetry C18, 4.6675 mm, 3.5 mm). Compounds were eluted with a linear gradient of aqueous CH 3 CN/0.1% CF 3 CO 2 H at a flow rate of 1.0 mL/ min. Purification of ligands was achieved on a Waters 600 HPLC using a reverse phase column (Vydac C18, 15-20 mm, 226250 mm). Peptides were eluted with a linear gradient of CH 3 CN/0.1% CF 3 CO 2 H at a flow rate of 5.0 mL/min. Separation was monitored at 230 and 280 nm. Size exclusion chromatography was performed on a borosilicate glass column (2.66250 mm, Sigma, St. Louis, MO) filled with medium sized Sephadex G-25 or G-10. The compounds were eluted with an isocratic flow of 1.0 M aqueous acetic acid. The pure compounds were dissolved in deionized water or dimethylsulfoxide at approximately 1 mM concentrations. Structures were characterized by ESI (Finnigan, Thermoquest Liquid Chromatography-Quadruplet ion trap instrument) or MALDI-TOF (Bruker Reflex-III) with a-cyanocinnamic acid as a matrix). For internal calibration an appropriate mixture of standard peptides was used with an average resolution of 8,000-9,000. High resolution mass measurements were carried out on a Bruker Ultraflex MALDI TOF-TOF and an Apex Qh Fourier Transformation-Ion Cyclotron Resonance (9.4 T) high resolution instrument.
Tissue culture 16HBE14o-cells, a SV40 transformed human bronchial epithelial cell line [29], were obtained through the California Pacific Medical Center Research Institute (San Francisco, CA, USA). Cells were maintained and expanded as previously described [19]. Briefly, cell lines were expanded in tissue culture flasks prior to transfer to 96 well E-plates (Roche) for experiments. Flasks and 96 well E-plates were coated initially with a matrix coating solution (88% Lechner and LaVeck basal medium, 10% bovine serum albumin (BSA; from 1 mg/ml stock), 1% bovine collagen type I (from 2.9 mg/ml stock), and 1% human fibronectin (from 1 mg/ml stock solution) and incubated for 2 hr at 37uC, after which the coating solution was removed and allowed to dry for at least 1 hr. 16HBE14o-cells were plated at a concentration of 1610 5 cells/cm 2 and grown in Eagle's Minimal Essential Medium supplemented with 10% Fetal Bovine Serum (FBS), 2 mM glutamax, penicillin and streptomycin (growth medium) at 37uC in a 5% CO 2 atmosphere. Growth medium was replaced every other day until the cells reached confluence (5-7 days). Cells were then transferred to 96 well E-plates for RTCA, or collagen/fibronectin/BSA coated glass coverslips for Ca 2+ imaging experiments.
Primary mouse tracheal epithelial (MTE) cells were cultured as described [28]. Briefly, mouse tracheas were removed, washed in phosphate-buffered saline for 5 min at room temperature, cut lengthwise, and transferred to collection medium [1:1 mixture of Dulbecco's Modified Eagle Medium (DMEM) and Ham's F12 with 1% penicillin-streptomycin) at 37uC. Tracheae were then incubated at 37uC for 2 hr in dissociation medium (44 mM NaHCO 3 , 54 mM KCl, 110 mM NaCl, 0.9 mM NaH 2 PO 4 , 0.25 mM FeN 3 O 9 , 1 mM sodium pyruvate, and 42 mM phenol red, pH 7.5; supplemented with 1% penicillin-streptomycin and 1.4 mg/ml pronase). Enzymatic digestion was stopped by adding 20% FBS. Epithelial cells were gently scraped from the tracheas, centrifuged at 100 x g for 5 min at room temperature. Cell pellets were washed in base culture medium (1:1 mixture of DMEM and Ham's F12 with 1% penicillin-streptomycin and 5% FBS) and  CHO cells were cultured in DMEM supplemented with 10% FBS and 1% penicillin and streptomycin at 37uC in a 5% CO 2 atmosphere. One day before transfection, CHO cells were plated in a 60 mm cell culture dish at a concentration of 1610 5 cells/cm 2 and grown without antibiotics. An MrgprC11 cDNA in pcDNA3.1 vector was transfected into CHO cells using Lipofectamine TM 2000 (Invitrogen) prior to transfer to coverslips for experiments. Coverslips were coated with 0.1 mg/ml poly-Dlysine (from 2 mg/ml stock) and incubated for 1 hr at room temperature, after which the coating solution was removed and the coverslips were washed twice with double distilled water. CHO cells were seeded on coverslips at a concentration of 1610 5 cells/ cm 2 6 hr after transfection. Transfected cells were incubated for 24 hr prior to Ca 2+ imaging.
In vitro physiological response screening 16HBE14o-cells on E-plates in growth medium and in a 37uC, 5% CO 2 incubator were monitored for the establishment for relative impedance overnight every 15 min using the xCELLigence TM Real Time Cell Analyzer (RTCA, Roche) [19,28,30]. When cells reached baseline impedance the next day, and prior to the experiment, the RTCA was moved to room air and temperature where full growth medium was replaced with 100mL modified Hank's Balanced Saline Solution (HBSS) pre-warmed to 37uC. The RTCA was then allowed to come to room temperature (45-60 min) prior to ligand addition. Each well was then supplemented with 100 mL HBSS containing appropriate ligands to measure concentration response ranges in quadruplicate. Additional wells were used for vehicle controls. Relative impedance in each well was monitored every 30 sec over 4 hr. Peak responses, defined as the maximal change in Normalized Cell Index, were used to define maximal response concentrations and physiological EC 50 s for each ligand.
The use of primary cultured MTE cells required different treatments and resulted in reduced overall signaling. Briefly, MTE cells were transferred to E-plates in minimal culture medium (100 mL/well of 1:1 mixture of DMEM and Ham's F-12, 1% penicillin-streptomycin, 3.6 mM sodium bicarbonate, 4 mM Lglutamine) and allowed to adhere for 4 hr. At that time each well was supplemented with 2x concentration of agonist in minimal culture medium. Relative impedance in each well was monitored every 30 sec for up to 2 hr.
In vitro Ca 2+ Imaging 16HBE14o-cells or CHO cells were loaded with fura 2acetomethoxyl ester (CalBiochem or Molecular Probes) for 30 min at room temperature. Cells were washed with HBSS and allowed to sit for at least 20 min prior to digital imaging. For activation and desensitization assay experiments using 16HBE14o-cells, [Ca 2+ ] i was measured as previously described [19]. Experiments consisted of 20 sec of recording of cells in HBSS to determine resting [Ca 2+ ] i, followed by a 10 sec wash to introduce ligand and up to 10 min of recording for ligand washes required for desensitization experiments. Briefly, fura-2 fluorescence was observed on an Olympus IX70 microscope with a 40X oil immersion objective after alternating excitation between 340 and 380 nm by a 75 W Xenon lamp linked to a Delta Ram V illuminator (PTI) and a gel optic line. Intracellular Ca 2+ concentration ([Ca 2+ ] i ) for each individual cell in the field of view was calculated by ratiometric analysis of fura-2 fluorescence using equations originally published in [31]. Individual ratios were calculated every sec throughout the experiments.

In vivo mechanical sensitivity
Male ICR mice (Harlan) or PAR 2 -/mice and their wild type littermates on a C57Bl/6 background weighing 25-30 grams were used for these studies. Animal protocols were approved by the Institutional Animal Care and Use Committee of The University of Arizona. Compounds were injected into the plantar surface of the hindpaw in a total volume of 25 mL using a 31-gauge needle. Compounds were diluted using sterile saline. Mechanical thresholds were determined using calibrated von Frey filaments (Stoelting Co, Wood Dale, IN) with the up-down method [32]. The experimenter was always blinded to the treatment conditions and animals were randomized such that animals in a single experimental group were never all housed together.

Statistical Analysis
All statistical analyses were evaluated with GraphPad software (San Diego, CA). Multivariate comparisons were done with a twoway ANOVA with Tukey's or Bonferroni multiple comparison post-test as appropriate for the individual experiment. Pair-wise comparisons were done with a two-tailed Student's t-test. A value of p,0.05 was used to establish a significant difference between samples. Data in Figures are graphed 6 Standard Error of the Mean (SEM) unless otherwise noted.

Results
Determination of a minimal peptidomimetic structure needed for PAR 2 activation using synthetic tethered ligands (STLs) STL construction. Truncated analogs of 2at-LIGRL-PEG 3 -Hdc, compounds 1-6, were synthesized using standard Fmoc chemistry on aldehyde amino methyl resin as described in Figure 1 and in detail in Flynn, et al [28]. Briefly, compounds assembled on the solid support were cleaved from the resin with TFA-scavenger cocktails and purified by Reverse Phase-HPLC and/or size-exclusion chromatography. All compounds gave . 95% analytical HPLC and expected MS (data not shown).
Specificity of PAR 2 agonists. Although the RTCA experiments using 16HBE14o-cells provide a highly sensitive physiological assay that encompasses various cell signaling responses to an agonist, it is inherently limited in detecting receptor specificity. To further evaluate specificity of known and novel STLs, we first compared RTCA responses to compounds 1-4 using primary cultured mouse tracheal epithelial (MTE) cells from wild type and PAR 2 -/mice ( [28]; Figure 5). The effective Cell Index for peak concentration response for each compound 1-4 was reduced in MTE cultures when compared to the 16HBE14o-cells. More importantly, compounds 1-4 all required PAR 2 expression to establish RTCA responses. Also similar to the 16HBE14o-RTCA traces, compound 4 displayed a reduced peak response in the PAR 2 -expressing MTE. To demonstrate signaling competence in both cultures, stimulation with the PAR 2 independent agonist ATP resulted in similar RTCA responses in both wild type and PAR 2 -/primary MTE cultures. Because [Ca 2+ ] i changes are a primary outcome following PAR 2 activation, we further tested for PAR 2 specificity using a Ca 2+ desensitization assay with the known PAR 2 agonist, 2at-LIGRL-NH 2 [19,33]. Using digital imaging microscopy, we first evaluated minimal ligand concentrations that would induce 90-100% activation in 16HBE14o-cells within a 5 min experiment ( Figure 6). Sample traces of average [Ca 2+ ] i changes plotted over time for compounds 1-4 and the parent peptidomimetic, 2at-LIGRL-NH 2 are consistent with RTCA recordings. Compounds 1 and 2 were highly potent ligands (15 nM) whereas compound 3 (300 nM) required higher concentrations to elicit the full Ca 2+ response, albeit with a significant delay in the time required to reach threshold [Ca 2+ ] i when compared with compounds 1 and 2 ( Figure 6F, G). Compound 4 required even higher concentrations (2 mM) to achieve threshold [Ca 2+ ] i changes. Even at this heightened concentration compound 4 displayed a significant drop in average peak [Ca 2+ ] i as well as lack of return to baseline [Ca 2+ ] i within the 5 min experiment ( Figure 6D).
For desensitization studies, 16HBE14o-cells were first exposed to a high concentration of 2at-LIGRL-NH 2 to effectively eliminate PAR 2 based signaling prior to application of compounds 1-4. Thus, any increase in [Ca 2+ ] i in response to compounds 1-4 would indicate a response that was not specific to PAR 2 . When 16HBE14o-cells were desensitized with 50 mM 2at-LIGRL-NH 2 , a second wash with 50 mM 2at-LIGRL-NH 2 did not result in an increase of [Ca 2+ ] i ( Figure 7A-D). Subsequent treatment with any of the four compounds also did not result in measurable changes in [Ca 2+ ] i . This loss of response was not caused by loss of Ca 2+ signaling itself, as ATP remained an effective agonist following desensitization of PAR 2 . We further tested the agonist ability of these novel PAR 2 agonists by using 10 fold the fully activating concentration for each compound to desensitize 16HBE14o-responses to 10 mM 2at-LIGRL-NH 2 ( Figure 7E- It has been demonstrated that MrgprC11 can be activated by the PAR 2 peptide agonist SLIGRL-NH 2 , (EC 50 = 10 mM) however, the Leu 6 -truncated peptide, SLIGR-NH 2 , lost MrgprC11 signaling capacity while retaining PAR 2 activity [12]. To examine PAR 2 /MrgprC11 selectivity, we evaluated Ca 2+ responses in MrgprC11 transfected CHO cells [12] with the parent peptidomimetic, 2at-LIGR-NH 2 and the most potent STL compounds (1, 2) from this study (Figure 8). When 2at-LIGRL-NH 2 was applied to MrgprC11 transfected CHO cells at very high concentration (10 mM), a modest Ca 2+ response was observed (10% of cells in the field of view); no response was observed at 1 mM. In contrast, 10 mM of the full length PAR 2 -STL, compound 1, induced a robust Ca 2+ response (88 6 19%) in the MrgprC11 transfected cells. This Ca 2+ response decreased profoundly at concentrations of compound 1 tested at 100 times higher than the RTCA EC 50 of , 1 nM (20% response at 1 mM, 10% response at 100 nM, 2.5% response at 10 nM, 0% at 1 nM). In contrast, the potent PAR 2 agonist compound 2 (2at-LIGR-PEG 3 -Hdc) induced limited Ca 2+ responses in the MrgprC11 transfected cells even at the highest concentrations tested (20% at 10 mM, 15% at 1 mM, 0.5% at 100 nM). All three compounds tested displayed selectivity for PAR 2 versus MrgprC11 with 2at-LIGRL-NH 2 and compound 2, 2at-LIGR-PEG 3 -Hdc, displaying minimal MrgprC11 activity at concentrations up to 10 mM.
In vivo efficacy. Stimulation of PAR 2 in vivo is known to promote mechanical sensitization reflected by a mechanical hypersensitivity response in the von Frey test [8,10,11]. Compounds 1-5 were individually injected into the hindpaw of mice following evaluation of baseline mechanical sensitivity and mechanical thresholds were evaluated at 1 and 3 hr post-injection. Based on the EC 50 s of parent compounds [19,28,32] we utilized a dose of 15 pmoles for experimentation. Consistent with in vitro findings, compounds 1-4 evoked mechanical hypersensitivity at 1 and 3 hr following injection ( Figure 9A-D) whereas compound 5 lacked activity ( Figure 9E). We did not note any itch response following injection of any tested compound. To determine specificity of these ligands, we tested the parent compound for mechanical hypersensitivity in wild type (WT, C57Bl/6 background) and PAR 2 -/mice (C57Bl/6 background). Compound 1 (15 pmoles) stimulated mechanical hypersensitivity in WT mice but failed to do so in PAR 2 -/mice ( Figure 9F). Therefore, these compounds are specific agonists at PAR 2 in vivo with the minimal peptide sequence in vivo matching the in vitro activity.
To better characterize differences among these heterocycletetrapeptides and heterocycle-dipeptides, we constructed companion STLs with two polyethylene glycol groups and a hexadecyl group (i.e., PEG 2 -Hdc attached to the C-terminus) and tested them for in vitro physiological responses with RTCA ( Figure 11, Figure 3). Compounds 11 (2at-Cha-IGR-PEG 2 -Hdc) and 12 (5io-Cha-IGR-PEG 2 -Hdc) displayed RTCA EC 50 s in the nM range (EC 50 = 16 nM, 95% CI: 12-20 nM and EC 50 = 6.8 nM, 95% CI: 4.1-11 nM, respectively). Similar to the pattern observed above, truncation to the heterocycle-dipeptide STL, compounds 13 (2at-Cha-I-PEG 2 -Hdc) and 14 (5io-Cha-I-PEG 2 -Hdc), resulted in less potent agonists (13: EC 50 = 86 nM, 95% CI: 59-130 nM and 14: EC 50 = 43 nM, 95% CI: 28-65 nM). Additionally, both compounds 13 and 14 displayed a delayed onset of response that was most prominent at submaximal concentrations. Further, compound 13 clearly did not attain peak Normalized Cell Index responses observed by other compounds in this group. A beneficial outcome of increased sensitivity using the STL construction and Figure 10. In vitro physiological responses of 16HBE14o-cells following addition of agonist compounds 7-10. Each of the top four panels (A-D) represents physiological response to agonist compounds as described for Figure 2. Concentrations for each experiment (at right of plots) show concentration responses that include supramaximal (black dashed lines) and maximal (solid line) responses. Compound 7: 2at-Cha-IGR-NH 2 , compound 8: 5io-Cha-IGR-NH 2 , compound 9: 2at-Cha-I-NH 2 , and compound 10: 5io-Cha-I-NH 2 all display rapid RTCA responses. However, compounds 7, 9, and 10, all exhibit reduced peak Normalized Cell Index responses. Concentration response curves developed from RTCA using the peak response within the 4 hr experiment are shown in the bottom panel. Compounds 7-10 display activity consistent with previously described heterocycle-pentapeptides PAR 2 agonists [19,28]. EC 50 s for each compound are shown in Figure 3. doi:10.1371/journal.pone.0099140.g010 RTCA analysis was the separation of potency when comparing compounds that only differed in their respective heterocycle head group. Notably, when Cha was substituted for Leu 2 the isoxazole containing compounds displayed an increased potency over the aminothiazoyl containing compounds in their peptidomimetic form (compounds 7-10; Figure 10E, Figure 3) that was only clearly separable when tested in their STL form (compounds 11-14; Figure 11E, Figure 3).
We further characterized compounds 11-14, using the Ca 2+ signaling assays. Typical Ca 2+ traces with average [Ca 2+ ] i changes (85-110 cells) plotted over time for compounds 11-14 are shown ( Figure 12A-D). Concentrations for each compound were established by their ability to elicit 80-100% activation of 16HBE14o-cells above threshold ( Figure 12E), and were consistent with the RTCA data in that the heterocycle-tetrapeptide STL constructions required lower concentrations than the heterocycle-dipeptide STLs. However, only compound 12 (5io-Cha-IGR-PEG 2 -Hdc) elicited Ca 2+ traces consistent with a full PAR 2 agonist. Both of the heterocycle-Cha-I-PEG 2 -Hdc compounds (13 and 14) could not consistently activate .95% of the cells in the 5 min experiment ( Figure 12E). Examination of Ca 2+ signaling data showed that compounds 11, 13 and 14 all exhibited a significantly delayed time to Ca 2+ threshold following ligand application, and the heterocycle-dipeptide compounds also exhibited a reduced peak [Ca 2+ ] i change ( Figure 12F-G).
Compounds 11-14 were subjected to Ca 2+ desensitization assays to test for specificity of response. In desensitization assays using 50 mM 2-at-LIGRL-NH 2 as the specific PAR 2 ligand to desensitize 16HBE14o-cells, none of compounds 11-14 induced significant Ca 2+ signaling ( Figure 13A-D), consistent with PAR 2 specificity for each of these compounds. Also as above, compounds 11-14 were used at 10x maximal Ca 2+ signaling response concentrations to assay their ability to desensitization PAR 2 responses in 16HBE14o-cells to 10 mM 2at-LIGRL-NH 2 . Although compounds 11 and 12 were able to desensitize Ca 2+ responses in these assays, compounds 13 and 14 did not completely desensitize Ca 2+ responses to 10 mM 2at-LIGRL-NH 2 ( Figure 13E-H). Examination of average Ca 2+ responses following application of high concentrations of compounds 13 and 14 suggested only partial activation of the 16HBE14o-cells ( Figure 13I).

Discussion
We have used a high sensitivity in vitro physiological assay combined with synthetic tethered-ligand (STL) approach to evaluate distinct protease-activated receptor-2 (PAR 2 ) ligand structure activity relationships (SAR). First, using the RTCA physiological assay, we were able to present a minimal peptide sequence required for full and partial PAR 2 activation and fully characterize EC 50 s for these truncated compounds. The use of this minimal peptide sequence both in vitro and in vivo opens new avenues for drug discovery and probing of physiological function at this receptor. Second, we were able to optimize SAR for PAR 2 ligands with differing, high activity heterocycle (5-isoxazol and 2aminothiazoyl) substitution of Ser 1 , paired with amino acid sequences naturally occurring in PAR 2 or the previously used cyclohexylalanine (Cha) substitution for Leu 2 . Such discovery, which is facilitated by the STL approach, is ideal to evaluate otherwise minimally potent and/or questionably selective compounds for PAR 2 and thus, provide a solid backbone for drug discovery. Finally, we provide detail on PAR 2 specificity of these compounds, an important point considering the recently discovered pharmacological similarity between PAR 2 and MrgprC11.
An alternative approach to assaying minimal peptide structures is the use of single Alanine substitutions (Ala-scan) in SLIGRL-NH 2 and Ca 2+ activation assays to assess PAR 2 activation [15,16,34]. Collectively, the Ala-scan studies revealed the importance of Ser 1 and Leu 2 for peptide-induced activation of PAR 2 with full loss of activity when the Leu 2 was substituted with Ala across all assays. The effects of Ser 1 substitution with Ala, however, was dependent on the cellular assay with one group demonstrating near complete loss of activity in PAR 2 expressing kNRK cells [34], another showing significant shift in activity in PAR 2 expressing oocytes [15], and the third demonstrating only a slight loss of activity in transfected mouse embryonic fibroblasts [16]. Other substitutions were again consistent across assays, where Ala substitutions of Ile 3 and Arg 5 decreased PAR 2 activation while substitutions at Gly 4 and Leu 6 did not appreciably alter potency. When multiple Ala substitutions were made to activating peptides, it was shown that SLAAAA-NH 2 could not activate Ca 2+ signaling in kNRK cells [34]. Through our STL-truncation approach coupled with the sensitive, in vitro physiological responses of RTCA, we found minimal changes in PAR 2 activation between the parent compound 1 (2at-LIGRL-PEG 3 -Hdc) when Leu 6 was removed (compound 2), successive reductions in potency following removal of Arg 5 -Leu 6 (3) and Ile 4 -Arg 5 -Leu 6 (4) and a complete loss of potency following removal of Ile 3 -Gly 4 -Arg 5 -Leu 6 (5) or Leu 2 -Ile 3 -Gly 4 -Arg 5 -Leu 6 (6). The minimal activating sequence both in vitro and in vivo required Leu 2 and Ile 3 in addition to the heterocycle substitute for Ser 1 (compound 4, 2at-LI-PEG 3 -Hdc). Interestingly, when Ala substitutions were introduced into the receptor and activity uncovered by trypsin activation, the naturally tethered SLAAAA sequence was sufficient for PAR 2 activation, albeit a less than full cellular response [34]. This minimal activation could not be duplicated using the STL approach, where compound 5 (2at-L-PEG 3 -Hdc), was inactive both in vitro and in vivo. It is possible that loss of activity in 5, could be caused by absence of a peptide backbone or a lost interaction with the side chain of Ile 3 that may be required in the absence of trypsin cleavage of the receptor. The importance of a peptide backbone is apparent when comparing RTCA activity from compounds 3 (2at-LIG-PEG 3 -Hdc) and 4 (2at-LI-PEG 3 -Hdc). The relatively high potency of compound 3 (EC 50 = 46 nM) was achieved by retention of the Gly 3 , amino acid without any side chain. Compound 4, however, displayed significantly reduced potency in addition to a delay in time to peak and a reduction in peak Normalized Cell Index. Subsequent reductions in the ability for compound 4 to fully activate Ca 2+ signaling suggest that activation by this minimal sequence results in only partial agonism of PAR 2 . PAR 2 activation is traditionally monitored by Ca 2+ response following Gq activation and subsequent Ca 2+ responses (e.g., [18,19,20,21,28]). However, it is well accepted that activation of PAR 2 by native proteases or peptidomimetics can result in the recruitment of a variety of G-Proteins and multiple signaling pathways [1,2]. The RTCA approach used herein to screen PAR 2 agonists relies on the cellular physiological response that is resultant of the various signaling pathways activated by the candidate drug [30]. Response patterns to individual compounds are reflective of the signaling pathways activated and as such, have been used to classify GPCR ligands into subgroups [35]. Compounds 1-4 tested in these studies displayed RTCA responses consistent with PAR 2 drugs that elicit both Ca 2+ and MAPK signaling [19,28], and do not appear to invoke ''biased signaling'' via PAR 2 [4,25,26,36]. Comparison of RTCA responses from primary cultured mouse tracheal epithelial (MTE) cells obtained from wild type or PAR 2 -/mice successfully demonstrated the need for PAR 2 expression to invoke physiological responses to these compounds. Traditional ''desensitization'' studies using Ca 2+ signaling responses confirmed PAR 2 specificity of truncated analogues. Extension of the traditional desensitization studies using high concentrations of the newly designed STLs as the agent to desensitize PAR 2 to a known specific peptidomimetic agonist, 2at-LIGRL-NH 2 allowed for further understanding of compound/ PAR 2 SAR. For example, the inability of compound 4 to fully desensitize cells at these heightened concentrations is in agreement with the RTCA results that suggest partial agonism by this selective PAR 2 agonist. The prototypical peptide activator for PAR 2 , SLIGRL-NH 2 , has recently been shown to contribute to the itch response via an alternative GPCR known to be expressed selectively in sensory neurons, MrgprC11 [12,37]. Although this receptor is not expressed in 16HBE14o-or MTE cells, and thus not a contributor to the in vitro results, activation of this GPCR in in vivo experiments could profoundly affect specificity in pain/itch pathways. Application of the parent STL (compound 1, 2at-LIGRL-PEG 3 -Hdc) to MrgprC11 transfected CHO cells resulted in a robust Ca 2+ response, however, this required . 5,000-fold the RTCA EC 50 concentration. Compound 2, with a truncated Leu 6 resulted in limited activity at MrgprC11 at 10 mM and no activity at 1 mM. From these experiments we conclude that retention of the Arg 5 -Leu 6 is preferred for MrgprC11 activation by our STL compounds. Concentrations required to activate Ca 2+ responses in transfected MrgprC11 cells demonstrate at least several hundred fold selectivity for PAR 2 over MrgprC11 by the STL compounds. Finally, the lack of response by 2at-LIGRL-NH 2 at the EC 50 concentration for SLIGRL-NH 2 suggests that the Ser 1 substitution confers selectivity for PAR 2 over MrgprC11 in the absence of tethering. Therefore, the approach taken herein has identified highly potent and selective compounds that can be utilized to selectively probe the function of PAR 2 in sensory biology.
A previous study reported on PAR 2 activation using peptidomimetic derivatives of the first three amino acids of the natural tethered ligand for PAR 2 (e.g., Ser 1 -Leu 2 -Ile 3 -NH 2 ) at relatively high concentrations (50 mM) and demonstrated partial PAR 2 activation using Ca 2+ signaling assays in HEK293 cells [21]. Significantly, one compound from this group, 5io-Cha-Ile-NH 2 (published as compound 9 in [21] and compound 10 in this report) had an estimated EC 50 similar to the peptide activator SLIGRLI-NH 2 [21]. However, the authors noted that lack of solubility of 5io-Cha-I-NH 2 at high concentrations (100 mM) prevented full EC 50 determination in their assay. Based on the minimal differences in potency observed in compounds 1 and 2 above, we took advantage of the sensitivity of the RTCA and STL approach to better evaluate EC 50 s of the heterocycle-dipeptides along with longer heterocycle-tetrapeptides. This new group included Ser 1 substitute heterocycles 2-aminothiazoyl and 5isoxazol with the Leu 2 substitute cyclohexylalanine (Cha) and in combination with Ile 3 or Ile 3 -Gly 4 -Arg 5 terminated with an amino group. We found that compounds 7-10 all elicited RTCA responses in 16HBE14o-cells, however, only compound 8 (5io-Cha-IGR-NH 2 ) elicited a traditional rapid and robust RTCA response typical of full and specific PAR 2 agonists (e.g., compounds 1-3 herein; [19,28]). Although not as potent as compounds 1 and 2 above, heterocycle-tetrapeptide STLs were highly potent activators of 16HBE14o-cells, with RTCA EC 50 s of 16 nM (11) and 6.8 nM (12), and significantly more potent than their corresponding heterocycle-dipeptide STLs (13: EC 50 = 86 nM; 14: EC 50 = 43 nM). Direct comparison of 5isoxazoyl heterocycle with 2-aminothiazoyl heterocycle substitutions resulted in an ,2 fold decrease in RTCA EC 50 s in both the heterocycle-tetrapeptide and heterocycle-dipeptide STLs. Substitution of Leu 2 with Cha reduced potency in the heterocycletetrapeptides (compare compounds 2 and 11), whereas the same substitution increased potency in the heterocycle-dipeptide construct (compare compounds 4 and 13). Although these latter comparisons are tempered by differences in PEG 2 (compounds 9-12) vs. PEG 3 spacers (compounds 1-4), such spacer differences using 2at-LIGRL-and 2at-LIGRLO-as parent groups in STLs did not alter potency across assays [28], and thus, the different PEG spacers likely do not alter these conclusions. Ca 2+ desensitization assays using 16HBE14o-cells confirmed specificity of the compounds 11-14 for PAR 2 . However, high concentrations of compounds 13 and 14 could not fully activate Ca 2+ response nor were they effective at desensitizing 16HBE14o-cells from activation by 10 mM 2-at-LIGRL-NH 2 . These data suggest that the heterocycle-tetrapeptide STLs fully and specifically activate PAR 2 , whereas the heterocycle-dipeptide STLs are PAR 2 specific, yet partial agonists.
The use of lipid tethering combined with RTCA and supplemented with traditional Ca 2+ signaling analysis allowed for more robust and interpretable SAR for PAR 2 , including smaller structural nuances that provide an efficient vehicle for future drug development. A strength of this sensitive, tethered ligand approach is the ability to test peptidomimetic ligands in a form that better mimics the natural activation of proteaseactivated receptors that results in a significant increase in potency. For example, RTCA allowed for separation of potency of peptidomimetic compounds (e.g., 4.5 fold differences in RTCA EC 50 ranging from 240 nM to 1.1 mM among compounds 7-10). The increased potency of STL derivatives also allowed for accurate Ca 2+ signaling studies and confirmation of partial agonism without non-specific effects associated with using high concentrations of newly developed untethered ligands that can obscure SAR. It is interesting that the partial RTCA and Ca 2+ signaling agonist compound 4 (2at-LI-PEG 3 -Hdc) gave a similar response to the full agonists compounds in our in vivo assays. These data provide evidence that full agonists (or by analogy, full antagonists) to PAR 2 may not be needed for full effects in vivo. It is accepted that our STL approach increases hydrophobicity in the ligand. Although increased hydrophobicity has traditionally been considered as a negative for building drugs, more recently lipidation of peptides has been recognized as a viable avenue for drug discovery [27]. In closing, we propose that the STL approach will continue to lead to the discovery of high potency peptidomimetics and small molecules as this technique can better identify contrasts between compounds and the resulting higher quality SAR will be enriched with otherwise undetectable structures which may contribute to high fidelity design.