Development of highly potent protease-activated receptor 2 agonists via synthetic lipid tethering

Abstract

Protease-activated receptor-2 (PAR₂) is a G-protein coupled receptor (GPCR) associated with a variety of pathologies. However, the therapeutic potential of PAR₂ is limited by a lack of potent and specific ligands. Following proteolytic cleavage, PAR₂ is activated through a tethered ligand. Hence, we reasoned that lipidation of peptidomimetic ligands could promote membrane targeting and thus significantly improve potency and constructed a series of synthetic tethered ligands (STLs). STLs contained a peptidomimetic PAR₂ agonist (2-aminothiazol-4-yl-LIGRL-NH₂) bound to a palmitoyl group (Pam) via polyethylene glycol (PEG) linkers. In a high-throughput physiological assay, these STL agonists displayed EC₅₀ values as low as 1.47 nM, representing a ∼200 fold improvement over the untethered parent ligand. Similarly, these STL agonists were potent activators of signaling pathways associated with PAR₂: EC₅₀ for Ca²⁺ response as low as 3.95 nM; EC₅₀ for MAPK response as low as 9.49 nM. Moreover, STLs demonstrated significant improvement in potency in vivo, evoking mechanical allodynia with an EC₅₀ of 14.4 pmol. STLs failed to elicit responses in PAR₂^−/− cells at agonist concentrations of >300-fold their EC₅₀ values. Our results demonstrate that the STL approach is a powerful tool for increasing ligand potency at PAR₂ and represent opportunities for drug development at other protease activated receptors and across GPCRs.—Flynn, A. N., Hoffman, J., Tillu, D. V., Sherwood, C. L., Zhang, Z., Patek, R., Asiedu, M. N. K., Vagner, J., Price, T. J., Boitano, S. Development of highly potent protease-activated receptor 2 agonists via synthetic lipid tethering.

Our current understanding of G-protein-coupled receptor (GPCR) functional pharmacology and physiology, as well as the development of therapeutics, has been greatly aided by the discovery and improvement of highly potent and specific ligands to these receptors (1–3). Approaches to the development of these compounds have largely been based on high-throughput screening of libraries using either functional (2) or in silico (4) screens. However, such approaches have not been successful in developing drugs for many GPCRs, including those in the protease-activated receptor (PAR) family. Tethering of peptides with lipid moieties has been used to express and target genetically encoded toxins for cell-type specific manipulation of systems (5), to deliver and target peptides that activate GPCRs (6, 7) or interfere with GPCR/G-protein interactions (8), and to allow peptide access to the intracellular leaflet for membrane targeted enzymes (e.g., kinases; ref. 9). Although membrane targeting is a recurring theme of lipidation, alternative factors can alter the pharmacological properties and ultimately, biological activity of receptor interaction (10). Despite the proliferation of biologically useful lipidation/membrane-tethering approaches and the presence of a naturally tethered ligand on activation of PARs, this technique has not been used to enhance the potency of peptide or peptidomimetic ligands at PARs. (Fig. 1).

Figure 1.

Activation of PAR₂ by protease and STL. PAR₂ is a GPCR with an extended amino terminus that contains an embedded tethered ligand (red) in the extracellular environment. A, B) On protease activation (A), the ligand is uncovered and interacts with the receptor to activate cell signaling and physiology (B). C) STLs are proposed to insert into the membrane (olive triangle) and with appropriate linker groups (magenta) attached to a high potency ligand (red) activate the receptor independent of protease.

Members of the PAR family of GPCRs can be activated by endogenous or exogenous proteases that cleave at the extracellular NH₂ terminus of the receptor exposing a tethered peptide sequence, which serves as a ligand to the receptor (11, 12). PAR₂ has gained much attention as a potential pharmacological target for allergic asthma (13–15), cancer (16), inflammation (17, 18), and pain (19–22), but the development of highly potent ligands at this receptor has proven to be an elusive goal. To date, the most potent ligands are near the micromolar range and structure activity relationship (SAR) data at the receptor is limited (12, 23–30). Based on the nature of the tethered endogenous ligand for PAR₂, we reasoned that lipidation that could result in membrane targeting of peptidomimetic ligands might afford an opportunity to develop highly potent and specific ligands at PAR₂. The discovery of such a set of ligands, and the elucidation of techniques that would allow for their improvement, would greatly enhance our understanding of PAR₂ physiology and pharmacology and open new vistas for the development of therapeutics targeting PAR₂.

Herein we describe the first high-potency PAR₂ agonists developed using a synthetic lipid-tethering approach (Fig. 1C) and demonstrate their efficacy in the low-nanomolar range (i.e., EC₅₀ 1–3 nM). Using existing PAR₂ agonists, we found that adding a synthetic membrane tether improves potency by up to 300-fold for a diverse set of ligands and that this enhancement in potency occurred across physiological and signaling assays. Critically, our synthetic tethered ligand (STL) approach offers exceptional specificity; these ligands have no effect in cells derived from PAR₂-knockout (PAR₂^−/−) mice at >300-fold their EC₅₀ values obtained in PAR₂ expressing cells. Therefore, we have created the first set of truly highly potent and specific PAR₂ agonists. These findings will have important implications for the further development of therapeutics targeting PAR₂, GPCRs, or other receptors.

MATERIALS AND METHODS

PAR₂ ligands

PAR₂ agonists were prepared by semimanual solid-phase peptide synthesis (31, 32) performed in fritted syringes using a Domino manual synthesizer obtained from Torviq (Niles, MI, USA) and described in detail in Supplemental Data. The ornithine (Orn) subset was prepared on the Rink resin, loaded with orthogonally protected (9H-fluoren-9-ylmethoxy)carbonyl-Orn-(allyloxycarbonyl) [Fmoc-Orn(Aloc)]. Following t-butyloxycarbonyl (Boc)-protected 2-aminothiazol-4-yl-LIGRL-NH₂ (2-at-LIGRL) assembly on the N^α terminus of Orn, the resin intermediate was split. Resin aliquots were placed in separated reactors, the Aloc at the N^ε amino group was removed, and the synthesis continued to complete linker lipid polyethylene glycol (PEG)_N- palmitoyl (Pam) (Supplemental Fig. S1). The subset lacking Orn was prepared using a standard linear solid-phase scheme on a Pam-secondary amine resin prepared from aldehyde resin by reductive alkylation. The completeness of first Fmoc-PEG coupling to the secondary amine resin was closely monitored as a potentially difficult coupling reaction. The final products were cleaved off the resin, purified by high-performance liquid chromatography (HLPC), and characterized by electrospray ionization–mass spectrometry (ESI-MS) and/or matrix assisted laser desorption ionization–time of flight (MALDI-TOF). Full details are available in Supplemental Data.

Epithelial cell line cultures

16HBE14o- cells are a SV40-transformed human bronchial epithelial (HBE) cell line (33) and were obtained through the California Pacific Medical Center Research Institute (San Francisco, CA, USA). Kirsten virus-transformed normal rat kidney (kNRK) cells, which display limited PAR₂ prior to transfection, were a kind gift from Dr. Nigel Bunnett (University of California, San Francisco, CA, USA) and grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% FBS, penicillin, streptomycin, and hygromycin (150 μg/ml) at 37°C in a 5% CO₂ atmosphere. Both cell lines were maintained and expanded as described previously (27). Briefly, cell lines were expanded in tissue culture flasks prior to transfer to cultureware for specific experiments. Flasks (16HBE14o-) and cultureware (16HBE14o- and kNRK) were coated initially with a matrix coating solution [88% Laboratory of Human Carcinogenesis (LHC) 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 h at 37°C, after which the coating solution was removed and allowed to dry for ≥1 h. 16HBE14o- cells were plated onto the matrix-coated cultureware at a concentration of 1 × 10⁵ cells/cm². 16HBE14o- or kNRK cells were cultured in appropriate growth medium [Eagle's minimum essential medium (MEM) or DMEM/F12, respectively] supplemented with 10% FBS, 2 mM glutamax, penicillin, and streptomycin at 37°C in a 5% CO₂ atmosphere. Medium was replaced every other day until the cells reached confluence (5–7 d). Cells were then transferred to appropriate matrix-coated substrate for analyses [96-well E-plates lined with gold electrodes for real-time cell analyzer (RTCA); glass coverslips for Ca²⁺ imaging; fluorescent protected 96-well plates for in-cell Western (ICW); 6 or 24 well plates for Western blot].

Mouse tracheal epithelial (MTE) cultures

Isolation and primary culture of MTE cells was adapted from previously reported procedures (34, 35). Mice were anesthetized with isoflurane and euthanized by cervical dislocation. Tracheas were removed, washed in phosphate-buffered saline (PBS) for 5 min at room temperature, cut lengthwise, and then transferred to collection medium (1:1 mixture of DMEM and Ham's F12 with 1% penicillin-streptomycin) at 37°C. Tracheae were then incubated at 37°C for 120 min in dissociation medium (44 mM NaHCO₃, 54 mM KCL, 110 mM NaCl, 0.9 mM NaH₂PO₄, 0.25 μM FeN₃O₉, 1 μM sodium pyruvate, and 42 μM phenol red, pH 7.5; supplemented with 1% penicillin-streptomycin and 1.4 mg/ml Pronase). Enzymatic digestion was stopped by adding 20% FBS to the dissociation media. Epithelial cells were dissociated by gentle agitation followed by physical removal from the tracheas. Epithelial cells were then centrifuged at 1000 rpm 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 centrifuged at 1000 rpm for 5 min at room temperature. MTE cells were resuspended in full culture medium (1:1 mixture of DMEM and Ham's F-12 supplemented with 1% penicillin-streptomycin, 5% FBS; 15 mM HEPES, 26.1 mM sodium bicarbonate, 4 mM l-glutamine, 10 μg/ml insulin, 5 μg/ml transferrin, 25 ng/ml epidermal growth factor, and 30 μg/ml bovine pituitary extract) seeded onto 96-well E-plates coated with collagen/fibronectin/BSA matrix or glass coverslips coated with raised collagen (36) at 37°C with 5% CO₂. MTE cells were fed every other day in full culture medium.

High-throughput in vitro physiological response

16HBE14o- (or MTE) cells were grown on matrix-coated E-plates (Roche Applied Science, Indianapolis, IN, USA). Impedance from each well was measured overnight at 37°C in a 5% CO₂ atmosphere using the xCELLigence RTCA (Roche Applied Science) system to ensure proper growth to 95% confluence (less for MTE) in each well. On the following day, growth medium was replaced with 100 μl of prewarmed Hanks' balanced salt solution (HBSS; 37°C). Background readings were obtained for an additional 30 min while the machine and buffer were allowed to reach room temperature and CO₂. Agonists (including controls published in ref. 27), diluted in HBSS at 2× final concentration, were added to the E-plate wells in 100-μl aliquots for a final volume of 200 μl. Relative impedance, expressed as the cell index was measured every 30 s for 4 h. According to the manufacturer's definition, cell index = Z_i − Z₀/15 Ω, where Z_i is impedance at a given time point during the experiment (i.e., postagonist addition), and Z₀ is impedance before the addition of agonist. Thus, a loss of adhesion would generate a lower cell index; an increase in cell adhesion, which is typically seen with GPCR/G_q activation, would result in an increase in the cell index. Individual traces of the cell index over time represent the average of 4 experiments from a single E-plate normalized at the point of agonist addition and against untreated cells to correct for cell proliferation according to the manufacturer's instructions. This method also allows for the direct comparison of the cell index across individual wells within the high-throughput experiment. Individual traces were assayed for peak changes, and dose-response curves were fitted using GraphPad Prism software (GraphPad, San Diego, CA, USA). Derivation of EC₅₀ values is shown in Supplemental Fig. S2.

Ca²⁺ signaling measurements

Intracellular calcium ion concentration ([Ca²⁺]_i) was measured using digital imaging microscopy, as described previously (27). A typical experiment consisted of 20 s of recording of cells in balanced buffer to determine resting [Ca²⁺]_i, followed by a 10 s wash to introduce ligand or control (including agonists from ref. 27) and an additional 2 min 30 s of monitoring for changes in [Ca²⁺]_i for PAR₂ activation experiments, or up to 10 min for subsequent washes required for desensitization experiments. Briefly, coverslip cultures were washed with HBSS and loaded for 40–45 min in 5 μM Fura-2-AM in HBSS. Fura-2 fluorescence was observed on an Olympus IX70 microscope with an ×40 oil-immersion objective (Olympus, Tokyo, Japan) after alternating excitation between 340 and 380 nm by a 75-W xenon lamp linked to a Delta Ram V illuminator (PTI, Birmingham, NJ, USA) and a gel optic line. [Ca²⁺]_i for each individual cell in the field of view was calculated by ratiometric analysis of Fura-2 fluorescence using equations published previously (37). Individual ratios were calculated every 0.6 s throughout the 3-min experiments and every second throughout the 10-min experiments. [Ca²⁺]_i traces over time were average [Ca²⁺]_i of all cells within a field of view (∼80–100) and are representative ≥4 experiments. Ca²⁺ response graphs are percentage of cells that increase Ca²⁺ above 200 nM within the 3-min experiment. Ca²⁺ dose responses are calculated from Ca²⁺ response experiments using GraphPad Prism software.

Detection of mitogen-activated protein kinase (MAPK) signaling

ICW studies were performed as described previously (27, 38). Briefly, cells were exposed to a dose response of each agonist (including controls from ref. 27), from 3 μM to 0.3 nM, for 5 min at room temperature. Cells were then fixed, incubated in primary (rabbit anti-pERK; Invitrogen), secondary antibodies (Cell Signaling Technologies, Danvers, MA, USA), and nuclear stain 1,5-bis{[2-(di-methylamino) ethyl]amino}-4,8- dihydroxyanthracene-9,10-dione (DRAQ5; Cell Signaling Technologies) and imaged on an Odyssey Scanner (LI-COR, Lincoln, NE, USA) as per manufacturers' instructions. For Western blot experiments, 16HBE14o- cells were plated on matrix-coated 6-well plates. Once cells reached confluence (5–7 d), the culture medium was removed, and the cells were incubated with each agonist diluted in HBSS for 5 min. After the agonist was removed, cells were lysed in a modified radioimmune precipitation assay buffer (50 mM Tris, pH 7.4; 150 mM NaCl;, 1 mM EDTA, pH 8.0; and 1% Triton X-100). Lysates were sonicated for 10 min at 4°C, centrifuged for 15 min at 14,000 g to pellet nuclei and debris, and stored at 80°C until blot analysis. Extracellular signal-regulated kinase 1/2 (ERK1/2), both activated phosphorylated ERK (pERK; antibodies from Cell Signaling Technology) and nonactivated (total ERK; antibodies from Cell Signaling Technology), were assayed by standard SDS-PAGE. Band densities were measured with ImageJ (U.S. National Institutes of Health, Bethesda, MD, USA), and pERK was standardized to total ERK to assess changes in ERK activity.

Assessment of mechanical allodynia

Compounds were injected into the plantar surface of the hindpaw in a total volume of 25 μl 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, USA) with the up-down method (39).

Statistics

All statistical analyses were evaluated with GraphPad software. Multivariate comparisons were done with a 2-way ANOVA with Tukey's or Bonferroni multiple comparison posttest as appropriate for the individual experiment. Pairwise comparisons were done with a 2-tailed Student's t test. A value of P < 0.05 was used to establish a significant difference between samples. Data are presented as means ± sem unless otherwise noted.

RESULTS

Discovery of highly potent STLs at PAR₂ using a high-throughput physiological assay

An array of STLs based on the potent peptidomimetic ligand 2-at-LIGRL-NH₂ (compound 1; Figs. 2 and 3; ref. 27) targeting PAR₂ was created. The first 6 STLs contained 2-at-LIGRL-NH₂, an Orn amino acid, varying PEG (n=0–5) linkers and a Pam membrane tether (compounds 2–7; Supplemental Figs. S1 and S2 and Fig. 3). We used a human bronchial airway epithelial cell line (16HBE14o-) that highly expresses PAR₂ with limited PAR₁, PAR₃, or PAR₄ expression (13) to initially screen PAR₂ drug potency. The natural adherence of these cells lends well to the high-throughput RTCA assay system that can examine in vitro physiological reactions to PAR₂ activation by detecting changes in cell impedance (27, 40, 41). In vitro physiological response traces (Fig. 2A), calculated dose-response curves (Fig. 2B), and corresponding EC₅₀ values (Fig. 2C) for compounds 2–7 are shown (development of EC₅₀ values from an RTCA experiment is outlined in Supplemental Fig S2). The addition of Orn and Pam did not result in a significant change in potency [compound 2; EC₅₀ 326 nM, 95% confidence interval (CI) 285–373 nM] when compared with the parent peptidomimetic (compound 1; EC₅₀ 314 nM, 95% CI 244–403 nM; ref. 27). However, a significant increase in potency was obtained by increasing the distance between the parent ligand peptidomimetic and the Pam tether with PEG moieties (compounds 3–5, PEG_N; N=1–3). 2-at-LIGRLO(PEG₁-Pam)-NH₂ (compound 3), with a single PEG moiety, displayed an EC₅₀ of 12.7 nM (95% CI 8.9–18.2 nM). An additional increase in potency was obtained by subsequent addition of PEG linkers and a maximum potency was observed with 2 or 3 PEG linkers (compound 4, 2-at-LIGRLO(PEG₂-Pam)-NH₂: EC₅₀ 1.84 nM, 95% CI 1.62–2.08 nM; and compound 5, 2-at-LIGRLO(PEG₃-Pam)-NH₂: EC₅₀ 1.47 nM, 95% CI 1.19–1.81 nM). Further elongation of the linker slightly reduced STL potency (Compound 6, 2-at-LIGRLO(PEG₄-Pam)-NH₂: EC₅₀ 3.73 nM, 95% CI 2.13–6.53 nM; and compound 7, 2-at-LIGRLO(PEG₅-Pam)-NH₂: EC₅₀ 6.34 nM, 95% CI 3.22–12.5 nM). An STL with a serine headgroup (compound 8, SLIGRLO(PEG₃-Pam)-NH₂; Fig. 3) was also tested for potency (Fig. 2D). SLIGRLO(PEG₃-Pam)-NH₂ displayed a higher EC₅₀ (13.1 nM; 95% CI 7.4–23.1 nM) than the matching STL with the 2-aminothiazole headgroup, verifying the utility of the 2-aminothiazole substitution in agonist development (27).

Figure 2.

PAR₂ STLs are highly potent agonists, as measured by in vitro RTCA assay. 16HBE14o- cells were monitored as for physiological activity over time following dose-response STL exposure (also see Supplemental Fig. S2). A) Dose-response traces of STLs with Orn (i.e., 2-at-LIGRLO(PEG_N-Pam)-NH₂; N=0–5). B) Calculated dose-response curves. C) Top: EC₅₀ values with 95% confidence intervals (CIs) corresponding to dose-response curves in panel B. Maximum potency in Orn-containing STLs is obtained with 2–3 PEG linkers, with a slight decrease as PEG linkers increase. Bottom: calculated EC₅₀ values with 95% CI for STLs designed without Orn (2-at-LIGRL-PEG_N-Pam; N=0–3) and for the SLIGRLO(PEG₃-Pam)-NH₂ STL. D) Dose-response curves corresponding to EC₅₀ and 95% CI values in bottom part of panel C. In the absence of Orn, no PEG is needed to initially improve potency over the parent peptide, and there is an increase in potency up to 3 PEG linkers. STL construction using the native tethered ligand sequence also confers a potency advantage over the peptidomimetic agonists (i.e., 2-at-LIGRL-NH₂). Traces in panel A are averages of 4 experiments; sem values were excluded to improve clarity. Data on 2-at-LIGRL-NH₂ are from ref. 27.

Figure 3.

Synthetic tethered ligand structures and potency measurements across assays. Compound numbers, compound names, structures, and EC₅₀ values (with 95% CIs) across a variety of physiological and signaling assays are shown for comparison. PEG, polyethylene glycol; Pam, palmitoyl; Cho, cholyl (cholic acid residue); Lau, lauroyl; Myr, myristoyl. Ca²⁺ indicates percentage of cells displaying a Ca²⁺ response, analyzed using digital imaging microscropy; allodynia indicates in vivo allodynia response using calibrated von Frey filaments with the up-down method. For EC₅₀ values labeled not detected, assays were run, but responses were not observed up to 10 μM, 50 μm (Ca²⁺), or 3 nmol (allodynia). Asterisks indicate data from ref. 27; control experiments conducted for the present work did not alter published EC₅₀ values.

In the above compounds, a side chain of Orn was used as an attachment point for the PEG_N-Pam addition. To better understand whether Orn largely served as a linker amino acid or whether it has a specific effect on STL potency, the RTCA experiments were repeated with STLs constructed without Orn; i.e., 2-at-LIGRL-PEG_N-Pam (compounds 9–12; Figs. 2C, D; and 3). Surprisingly, the addition of the Pam group in the absence of Orn or a PEG linker was sufficient to increase agonist potency when compared to the parent compound (compound 9, 2-at-LIGRL-Pam: EC₅₀ 27.9 nM, 95% CI 21.3–36.6 nM). Similar to the Orn-containing compounds, the addition of additional PEG linkers improved potency, with 2 or 3 PEG moieties providing maximum potency (compound 10, 2-at-LIGRL-PEG₁-Pam: EC₅₀ 7.59 nM, 95% CI 6.30–9.14 nM; compound 11, 2-at-LIGRL-PEG₂-Pam: EC₅₀ 2.80 nM, 95% CI 1.76–4.46 nM; compound 12, 2-at-LIGRL-PEG₃-Pam: EC₅₀ 1.72 nM; 95% CI 1.30–2.28 nM).

To demonstrate the contribution of the Pam group to STL potency, we first evaluated three alternative and biologically relevant molecules attached to a 2-at-LIGRLO(PEG₂)-NH₂ (choline) or 2-at-LIGRL-PEG₂ agonist/linker sequences [lauroyl (Lau) and myristoyl (Myr) groups; compounds 13–15; Figs. 3 and 4). All of these tethers displayed reduced potency when compared with similar STLs containing a Pam tether. Finally, scrambled peptide STLs [compound 16, ISRLLG-PEG₃-Pam, Fig. 4; compound 17, ISRLLGO(PEG₃-Pam)-NH₂, Fig. 3] did not elicit physiological responses. In summary, we found 2-at-LIGRLO(PEG_N-Pam)-NH₂ or 2-at-LIGRL-PEG_N-Pam STLs, with N=2 or 3, to be the most potent PAR₂ STLs. These compounds were followed in potency by a group of ligands with additional PEG linkers, 2-at-LIGRLO(PEG₄-Pam)-NH₂, and a third group of potent ligands: 2-at-LIGRLO(PEG_N-Pam)-NH₂ (N=1, 5), 2-at-LIGRL-PEG_N-Pam (N=0 or 1), and SLIGRLO(PEG₃-Pam)-NH₂. The scrambled sequence STLs [ISRLLG-PEG_N-Pam or ISRLLGO(PEG_N-Pam)-NH₂] and the O(PEG₃-Pam)-NH₂ (not shown) controls did not demonstrate any efficacy.

Figure 4.

In vitro physiological analyses of membrane tethers for STLs. 16HBE14o- cells were assayed as described in Fig. 2. Dose responses to STLs with a base structure 2-at-LIGRL-PEG₂-x, where x = Pam, Lau, or Myr groups; or 2-at-LIGRLO(PEG₂-y), where y = cholic acid (Cho), were constructed along with an STL that contained a scrambled peptide sequence (ISRLLG-PEG₃-Pam). Ai–v) Individual traces of STLs: compound 11, 2-at-LIGRL-PEG₂-Pam (Ai); compound 13, 2-at-LIGRLO(PEG₂-Cho)-NH₂ (Aii); compound 14, 2-at-LIGRL-PEG₂-Lau (Aii); compound 15, 2-at-LIGRL-PEG₂-Myr (Aiv); compound 16, ISRLLG-PEG₃-Pam (Av). Note differences in concentration ranges applied during experiments. B) Comparison between dose responses of the STLs using different membrane tethers. The STL containing Pam as the membrane tether was the most potent PAR₂ agonist. In contrast, the STL with a Pam tether containing a scrambled peptide sequence displayed a limited response, and thus serves as an excellent control. (see text and Fig. 3 for EC₅₀ values).

PAR₂ STLs activate major signaling pathways associated with PAR₂ activation with dramatically enhanced potency

The RTCA analyses above provided an in vitro physiological assay to compare ligand activity. Activation of PAR₂ at the primary trypsin site stimulates two downstream signaling pathways: Ca²⁺ and MAPK (23). To verify that STL ligands activated primary signaling pathways, we first used digital imaging microscopy to assay the effects of the two most potent STL ligands [2-at-LIGRLO(PEG₂-Pam)-NH₂ and 2-at-LIGRLO(PEG₃-Pam)-NH₂] and a lower-potency STL [2-at-LIGRLO(Pam)-NH₂) on Ca²⁺ signaling in model 16HBE14o- cells; e.g., ref.27]. Representative experiments using 8 nM 2-at-LIGRLO(PEG₂-Pam)-NH₂ (Fig. 5A–D), 8 nM 2-at-LIGRLO(PEG₃-Pam)-NH₂ (Fig. 5E–H), or 10 μM 2-at-LIGRLO(Pam)-NH₂ (Fig. 5I–L) demonstrate high-potency activation of Ca²⁺ signaling. In each case, cells displayed initial increases in [Ca²⁺]_i ∼30 s after addition of the agonist, with most cells responding by 60 s. However, a significant difference was found in the concentrations of STL required for this full response that was dependent on the presence or absence of an appropriate PEG linker (i.e., 8 nM vs. 10 μM). To better understand this difference, we conducted dose-response experiments (3 nM–30 μM) for each of the STLs tested (Fig. 5M; see Supplemental Fig. S2 for dose-response curve construction). Both 2-at-LIGRLO(PEG₂-Pam)-NH₂ (EC₅₀ 4.04 nM, 95% CI 3.34–4.88 nM) and 2-at-LIGRLO(PEG₃-Pam)-NH₂ (EC₅₀ 3.95 nM, 95% CI 3.43–4.55 nM) displayed relative EC₅₀ values representing a >300-fold improvement from the parent peptide, 2-at-LIGRL-NH₂ (EC₅₀ 1.82 μM, 95% CI 1.47–2.25 μM; ref. 27). In contrast, 2-at-LIGRLO(Pam)-NH₂ did not display a significant increase in potency (EC₅₀ 1.12 μM, 95% CI, 830 nM–1.52 μM). The relative rank order of potency among these compounds, 2-at-LIGRLO(PEG₂-Pam)-NH₂ = 2-at-LIGRLO(PEG₃-Pam)-NH₂ ≫ 2-at-LIGRLO(Pam)-NH₂ = 2-at-LIGRL-NH₂ agrees with that found in the in vitro physiological assay above.

Figure 5.

PAR₂ STLs potently activate Ca²⁺ and MAPK signaling in model epithelial cells. [Ca²⁺]_i was monitored in 16HBE14o- cells over time and in response to STLs. A–L) [Ca²⁺]_i at time 0 9 (A, E, I) and changes at 30 s (B, F, J), 60 s (C, G, K), and 90 s (D, H, L) following addition of 8 nM 2-at-LIGRLO(PEG₃-Pam)-NH₂ (A–D), 8 nM 2-at-LIGRLO(PEG₂-Pam)-NH₂ (E–H), and 10 μM 2-at-LIGRLO(Pam)-NH₂ (I–L). Concentrations were chosen to illustrate full Ca²⁺ responses for each STL. Addition of the STL induced rapid increases in [Ca²⁺]_i that began within 30 s of drug application, peaked between 45–90 s, and initiated a return to baseline near 90 s. In each image, white lines approximate cell borders and [Ca²⁺]_i is depicted in the color bar shown in panels A, E, I. M) Dose-response curves generated for each STL. A clear shift in potency is apparent when 2 or 3 PEG moieties are included in the STL (see text and Fig. 3 for EC₅₀ values). N) MAPK signaling in 16HBE14o- cells was monitored using high-throughput ICW (left) and traditional Western blot (representative blot at right) analyses. Dose-response curves based on the percentage maximal pERK/DRAQ5 are graphed for 2-at-LIGRLO(PEG₃-Pam)-NH₂, 2-at-LIGRLO(PEG₂-Pam)-NH₂ and 2-at-LIGRLO(Pam)-NH₂. All three agonists activated MAPK signaling with nanomolar potency; however, STLs with 2 or 3 PEG linkers were more potent (see text and Fig. 3 for EC₅₀ values). In agreement with in vitro physiological signaling and Ca²⁺ signaling is a clear shift in potency when 2 or 3 PEG moieties were included in the STL (see text for EC₅₀ values). For each dose of agonist, 3 ≥ n ≥ 10; labels in panel N correspond with curves in panels M, N.

We next compared activation of MAPK pathways using a dose-response (0.3 nM– 3 μM) of each STL in a high-throughput ICW assay (Fig. 5N; ref. 27). Consistent with the RTCA and Ca²⁺ signaling assays, 2-at-LIGRLO(PEG₃-Pam)-NH₂ (EC₅₀ 9.49 nM, 95% CI, 1.94–46.4 nM) and 2-at-LIGRLO(PEG₂-Pam)-NH₂ (EC₅₀ 10.9 nM, 95% CI 3.85–31.3 nM) were both significantly more potent than 2-at-LIGRLO(Pam)-NH₂ (EC₅₀ 331 nM, 95% CI, 148–741 nM) and 2-at-LIGRL-NH₂ (EC₅₀ 458 nM, 95% CI, 204–1.03 μM; ref. 27). These results were supported by similar experiments using traditional Western blotting. In summary, the signaling pathway analyses further demonstrate that STLs are acting as potent and full PAR₂ agonists that mimic protease activation of PAR₂ at its primary trypsin target site.

PAR₂ STLs are highly specific ligands

To ensure that the robust increase in potency observed above did not come at the cost of specificity, we first performed a series of PAR₂-desensitization assays in 16HBE14o- cells (24, 27, 28, 42, 43). Because it was consistently the most potent STL across assays, our specificity experiments were focused on 2-at-LIGRLO(PEG₃-Pam)-NH₂ ligand. Typical population experiments showing average [Ca²⁺]_i changes in response to 10 nM 2-at-LIGRLO(PEG₃-Pam)-NH₂ (Fig. 6A) were used as baseline activation measures. In separate experiments, 10 nM STL was added after first applying two desensitizing doses (50 μM) of the PAR₂ peptidomimetic, 2-furoyl-LIGRLO-NH₂ (2-f-LIGRLO-NH₂; Fig. 6B (24). Subsequent application of 10 nM 2-at-LIGRLO(PEG₃-Pam)-NH₂ did not elicit a Ca²⁺ response. However, addition of ATP did demonstrate a fully functional, PAR₂-independent, Ca²⁺ response. Reciprocal desensitizing experiments, using 100 nM 2-at-LIGRLO(PEG₃-Pam)-NH₂ to desensitize cells, prevented Ca²⁺ response by 10 μM 2-f-LIGRLO-NH₂ (Fig. 6C). In additional experiments using thrombin activation and desensitization, we ruled out nonspecific responses at other PAR family members (Supplemental Fig. S3).

Figure 6.

STLs require PAR₂ expression for response. Drug desensitization and genetic modification of PAR₂ both limit STL signaling in model and primary cultured cells. A–C) Representative population graphs of the average change in [Ca²⁺]_i in 16HBE14o- cells in response to STLs are plotted over time and shown with sem. A) 16HBE14o- cells maintained a baseline [Ca²⁺]_i until application of 10 nM 2-at-LIGRLO(PEG₃-Pam)-NH₂ (at 140 s) induced a robust increase in average [Ca²⁺]_i that began to return to baseline within 60 s. B) Application of 2-f-LIGRLO-NH₂ at 25 s fully desensitized the 16HBE14o- cells to PAR₂ agonists, as evidenced by a lack of Ca²⁺response following further exposure to 2-f-LIGRLO-NH₂ at 300 s. Subsequent application of 2-at-LIGRLO(PEG₃-Pam)-NH₂ did not elicit a Ca²⁺response. Ca²⁺ response to 2.5 μM ATP demonstrated that Ca²⁺ signaling remained intact, and lack of response to 2-at-LIGRLO(PEG₃-Pam)-NH₂ was PAR₂ specific. C) Application of 100 nM 2-at-LIGRLO(PEG₃-Pam)-NH₂ similarly desensitized 16HBE14o- cells, as evidenced by a lack of response to 2-f-LIGRLO-NH₂, without compromising ATP-induced Ca²⁺ signaling. D) 2-at-LIGRLO(PEG₃-Palm)-NH₂ (60 and 100 nM) was applied to either kNRK cells transfected with human PAR₂ (kNRK-PAR₂) or a control plasmid (kNRK) and monitored for Ca²⁺ response for 3 min. While the kNRK-PAR₂ cells responded with robust Ca²⁺ signaling, little Ca²⁺ response representing background PAR₂ expression in the nontransfected control cells was found. E, F) Primary cultured mouse tracheal epithelial cells from wild-type or PAR₂^−/− mice were exposed to 2-at-LIGRLO(PEG₃-Pam)-NH₂ and monitored for Ca²⁺ response (E) or a physiological response using the RTCA (F). In both cases, full responses were observed in wild-type cells and minimal responses in PAR₂^−/− cells, despite increases of 2-at-LIGRLO(PEG₃-Pam)-NH₂ concentrations of 100- and 30–fold (E, F, respectively). A minimum of n = 4 was performed for each dose and cell type; data were analyzed using a Student's t test. *P < 0.01; **P < 0.005.

To further illustrate specificity, we used PAR₂^−/− cell cultures. In model epithelial cell lines, kNRK cells with or without PAR₂ expression, 2-at-LIGRLO(PEG₃-Pam)-NH₂ stimulated significant Ca²⁺ responses in PAR₂-transfected kNRK cells at two different doses (100 and 60 nM), with minimal stimulation of control vector-transfected kNRK cells (Fig. 6D). Similarly, we examined Ca²⁺ responses in primary cultured MTE cells from wild-type or PAR₂^−/− mice (Fig. 6E). MTE cells cultured from PAR₂-expressing animals elicited Ca²⁺ responses after exposure to 100 nM 2-at-LIGRLO(PEG₃-Pam)-NH₂, while MTE cells cultured from PAR₂^−/− animals did not respond to a 100-fold increase in concentration (10 μM). Both wild-type and PAR₂^−/− MTE cell cultures elicited full Ca²⁺ responses when challenged with ATP, demonstrating intact Ca²⁺ signaling. We also examined wild-type and PAR₂^−/− cultured MTE cells in the RTCA in vitro physiological assay (Fig. 6F). Consistent with a loss of signaling, wild-type cells displayed a full response to the 2-at-LIGRLO(PEG₃-Pam)-NH₂ agonist with a minimal dose (30 nM), while cells cultured from PAR₂^−/− mice displayed minimal response at 20-fold higher concentrations (600 nM). These experiments show that model and primary cultured cells require PAR₂ expression for both signaling and in vitro physiological responses to the newly developed STL agonists.

2-at-LIGRLO(PEG₃-Pam)-NH₂ potently stimulates PAR₂ in vivo

The above studies demonstrate that PAR₂ STLs are highly potent and specific ligands in vitro, but to aid in the understanding of PAR₂ pharmacology and physiology, these compounds must also have utility in vivo. To this end, we asked whether 2-at-LIGRLO(PEG₃-Pam)-NH₂ would evoke mechanical allodynia in vivo (Fig. 7). Multiple previous studies have shown that PAR₂-activating peptides evoke allodynia in mice (19, 22, 44). Intraplantar injection of the STL induced allodynia lasting for 4–9 d depending on dose (Fig. 7A) with an EC₅₀ of 14.4 pmol (95% CI 7.9–26.1 pmol). To ensure specificity of the response, we repeated the experiments with an STL that contained a scrambled sequence (ISRLLGO(PEG₃-Pam)-NH₂). In direct comparison experiments using 3 nmol of each compound (Fig. 7B), only the PAR₂-specific STL induced allodynia. Hence, PAR₂ STLs are potent and specific tools for in vivo use.

Figure 7.

PAR₂ STLs potently evoke allodynia in vivo. A) Mice were injected with increasing doses of 2-at-LIGRLO(PEG₃-Pam)-NH₂, and mechanical thresholds were measured daily for 11 d. Inset: area under the curve analysis revealed a dose-dependent increase in allodynia magnitude and duration with 2-at-LIGRLO(PEG₃-Pam)-NH₂. B) A scrambled STL peptide [ISRLLGO(PEG₃-Pam)-NH₂] failed to evoke mechanical allodynia at the highest dose tested for PAR₂ STL. Data were analyzed using 2-way ANOVA with Bonferroni posttest; n = 6 in all experiments. ***P < 0.001.

DISCUSSION

The primary finding of the present work is the discovery of the first set of agonists at PAR₂ with high potency in the low-nanomolar range. This enhancement in potency was achieved using a lipid-tethering approach based on the hypothesis that mimicking endogenous activation mechanisms of the receptor would lead to an enhancement in pharmacological activity (e.g., Fig. 1). These agonists improve on potency across signaling pathways engaged by PAR₂ by ≥45-fold (MAPK) and up to 450-fold (Ca²⁺), without any apparent cost in terms of specificity. We propose that this approach will be a powerful tool for expanding the pharmacological repertoire of PAR₂ ligands and that it may be a suitable approach for enhancing potency across GPCR families, such as the PARs.

Tethering of ligands to the cell membrane has been used in a number of contexts to generate useful pharmacological tools. Perhaps the most common approach is myristoylation of pseudosubstrate inhibitors of membrane-localized kinases, such as protein kinase C (PKC), to allow access to the intracellular membrane leaflet for achievement of enzyme inhibition (9). More recently, genetically encoded toxins or other ligands containing membrane-tethering signal sequences have been used to selectively express ligands in cell populations or among neuronal circuits for the dissection of physiological pathways (5, 45, 46). Similar approaches have been taken for expressing constitutively active kinases (47, 48). Finally, membrane tethering has also been used to deliver peptides that activate GPCRs directly (e.g., refs. 6, 49) or that interfere with GPCR/G-protein interactions (called pepducins) to disrupt cell signaling (17, 50, 51). Peducins have been used for PAR₂, in part because traditional efforts in PAR₂ drug discovery have failed to achieve highly potent agonists or antagonists (17, 51). Using lipid-tethering technology, we have overcome the agonist part of this problem, creating the first generation of highly potent PAR₂ agonists.

The linkers used in our STLs include the amino acid Orn and varying numbers of PEG moieties. The development of the most used specific agonist at PAR₂, 2-f-LIGRLO-NH₂ was created with an Orn group as a means for easy attachment of probes, such as fluorescent makers (24), whereas PEG groups were chosen based on previous drug design experience spacing multivalent ligands (31, 52). An individual PEG linker, 20 atoms in length, provides a structure that can be highly randomized in presentation yet largely ignored in terms of receptor interaction (53). The observation that an Orn moiety does not provide similar potency improvements in STL compared to PEG (i.e., compare compounds 2, 9, 10) suggests that choice of linker is important in developing the STL. The increase in potency observed as PEG linkers increased from 0–3 also suggests an optimal distance is required between agonist presentation and membrane tethering. The choice of membrane tether also played a significant role in STLs potency (Fig. 4). Comparison of membrane tethers using three saturated carbon chains (Lau, Myr, or Pam) and choline (linked via Orn) resulted in clear differences in potency in the in vitro physiological screen, with the Pam addition working best. Interestingly, the Lau group, when combined with the PEG₂ linker, displayed equal potency, and the choline attached compound displayed reduced potency, when compared with the peptidomimetic parent compound, 2-at-LIGRL-NH₂ (Fig. 3). While these modifications limit presentation of the ligand to PAR₂ that could be due to changes in membrane tethering, alternative explanations are possible, and direct experiments that test this hypothesis have not been performed at this time.

We anticipate that our STL approach will yield significant insight into PAR₂ pharmacology for several reasons. First, and most notably, the increase in potencies observed over previously described PAR₂ ligands are greatly enhanced across assays. This suggests that even very low potency PAR₂ non-STL pharmacophores can be recognized in high-throughput assays if ligands are attached to an optimized synthetic tether, which we have described here. Hence, this approach creates a major opportunity to renew screening efforts at PAR₂ using the STL approach for the identification of novel PAR₂ ligands. Second, the lack of highly potent PAR₂ ligands has limited discovery efforts aimed at PAR₂. For instance, although a number of agonists and a small number of antagonists have been found (12, 25), no traditional positive or negative allosteric modulators have been identified, and signaling pathway specific ligands (biased signaling) are only now being recognized (54–56). Third, the high-throughput RTCA screening protocol used as the primary screen in this study has been shown to be sensitive enough to detect differential signaling following ligand binding of the β₂ adrenergic receptor in a HEK cell model (40). Similarly, we can detect differential signaling from our 16HBE14o- cells when individual Ca²⁺ or MAPK signaling pathways are blocked with BAPTA or U0216, respectively (data not shown). However, all of the STL ligands tested herein displayed similar activation patterns on the RTCA to full agonists developed previously and, thus, without indication of biased signaling at PAR₂. Testing of the most potent STLs (e.g., Fig. 5) confirmed increases in both Ca²⁺ signaling and MAPK signaling following PAR₂ activation. We anticipate that STL chemistry combined with high-throughput and sensitive screening has the potential to greatly enhance research in these areas. Finally, this approach has already led to one important insight: peptidomimetic STLs at PAR₂ have increased potency (∼10-fold in in vitro physiological assay) compared to the STL endogenous ligand (SLIGRL-NH₂; compare compounds 5 and 8). This finding demonstrates that the peptidomimetic ligand approach is a bona fide improvement on engaging the agonist pharmacophore of the receptor. Further development of SAR along these lines may lead to the development of even more potent agonists and further insight into antagonist SAR at this receptor.

A major rationale for developing highly potent and specific agonists at PAR₂ is to gain better insight into the role of PAR₂ in disease. PAR₂^−/− mice have led to important insights into the role of PAR₂ in inflammation (14, 57, 58) and pain (19, 21, 22, 59, 60), and the recent discovery of pepducin G protein uncouplers at PAR₂ (17, 51) have further established PAR₂ as an important drug target. We show a clear improvement in PAR₂ potency in vitro using the STL method. This finding is paralleled by similar enhancements in in vivo activity. Previous studies have used ∼15 nmol doses of SLIGRL (22, 61, 62) to evoke thermal and mechanical allodynia in mice and rats. We show here that doses as low as 8 pmol are sufficient to evoke mechanical allodynia using a highly potent PAR₂ STL. Notably, a synthetic tethered scrambled peptide elicited no response at the maximum dose tested for the STL (3 nmol). Hence, our newly generated PAR₂ STLs are highly potent and selective tools for probing PAR₂ function both in vitro and in vivo.

The STL approach described here is applicable to ligand discovery efforts at other GPCRs. It is known that lipidation can alter peptide activation of receptors through a variety of means, including metabolic stability, membrane binding, membrane permeability, receptor binding, and bioavailability (10). A wide range of GPCRs are activated by peptides; peptidomimetic strategies have been broadly utilized to generate novel, potent, and specific ligands at these receptors. However, a large number of peptide-activated GPCRs, including the PARs, still remain with relatively undeveloped pharmacology. Because PARs are naturally activated by tethered ligands, coupling traditional peptidomimetic drug discovery efforts with the STL approach can be utilized to enhance the potency of existing ligands and the discovery of new ligands with novel pharmacological properties. While the feasibility of this approach for the development of therapeutics is limited, proliferation of pharmacological tools at GPCRs is an important step toward development of therapeutics and the STL approach may be utilized to achieve that goal. In addition, although we believe the STL approach will best be utilized in identifying novel small molecules that can be subsequently altered for final drug design, the glucagon-like peptide 1 (GLP-1) analog and clinical drug liraglutide (63) provides a recent example where palmitoylation and corresponding linker regions are an important component of drug function (64). Thus, the possibility of using specific and potent STLs as drugs themselves is not without merit (10).

In summary, we have discovered the first single-digit nanomolar potency agonists at PAR₂. These compounds are effective across signaling pathways activated by the receptor in vitro, engage the receptor with high potency in vivo, and are exquisitely selective. These ligands will be useful for probing PAR₂ function in a variety of physiological contexts. Moreover, the STL approach described here has the potential to greatly enhance drug discovery efforts aimed at other PARs and potentially a wide range of other GPCRs.