Potent Agonists of the Protease Activated Receptor 2 (PAR₂)

Abstract

Novel peptidomimetic pharmacophores to PAR₂ were designed based on the known activating peptide SLIGRL-NH₂. A set of 15 analogues was evaluated with a model cell line (16HBE14o-) that highly expresses PAR₂. Cells exposed to the PAR₂ activating peptide with N-terminal 2-furoyl modification (2-furoyl-LIGRLO-NH₂) initiated increases in intracellular calcium concentration ([Ca²⁺]_i EC₅₀ = 0.84 μM) and in vitro physiological responses as measured by the xCELLigence real time cell analyzer (RTCA EC₅₀ = 138 nM). We discovered two selective PAR₂ agonists with comparable potency: compound 1 (2-aminothiazol-4-yl; Ca²⁺ EC₅₀ = 1.77 μM, RTCA EC₅₀ = 142 nM) and compound 2 (6-aminonicotinyl; Ca²⁺ EC₅₀ = 2.60 μM, RTCA EC₅₀ = 311 nM). Unlike the previously described agonist, these novel agonists are devoid of the metabolically unstable 2-furoyl modification and thus provide potential advantages for PAR₂ peptide design for in vitro and in vivo studies. The novel compounds described herein also serve as a starting point for structure–activity relationship (SAR) design and are, for the first time, evaluated via a unique high throughput in vitro physiological assay. Together these will lead to discovery of more potent agonists and antagonists of PAR₂.

INTRODUCTION

Protease-activated receptors (PARs) 1–4 are a family of four G-protein-coupled receptors (GPCRs) that are self-activated by a tethered sequence exposed by proteolysis of the extracellular domain.¹ The four members of the GPCR family of PARs are expressed on a wide variety of cell types. PARs are activated by proteolysis in response to endogenous and exogenous proteases and can contribute to both cellular homeostasis and pathology.^1,2 In the case of PAR₂, the exposed peptides such as SLIGVK-NH₂ (human) or SLIGRL-NH₂ (murine) remain tethered on the receptor and activate an orthosteric binding site located on second loop of the receptor.¹ The different N-termini of the PARs present substrates for a variety of proteases that create selective activation mechanisms for signal transduction. This enzyme-linked self-activation is limited to PARs among GPCRs.³ There are a variety of enzymes that can expose the tethered ligand on PAR, but a key difference between PAR₂ and the other PARs is the lack of activation by thrombin.⁴ As an obvious consequence of its activation mechanism, PAR₂ is associated with pathologies with a strong protease release. The involvement of PAR₂ in inflammatory diseases such as arthritis, lung inflammation (asthma), inflammatory bowel disease, sepsis, and pain disorders makes PAR₂ an attractive target for drug intervention.

Significant tools used to study PARs are small peptides or peptidomimetics that mimic the ligand binding properties of the tethered ligand exposed by proteolysis of the N-terminus from the natural receptor (reviewed in ref 1b): SFLLRN-NH₂ activates PAR₁; human-derived SLIGKV-NH₂ and murine-derived SLIGRL-NH₂ activate PAR₂; GYPQVN-NH₂ activates PAR₄. However, these short peptides activate cognate receptors with lower potency compared with native tethered ligands, which require lower binding energy as a consequence of the covalent link to the receptor. Also, untethered activating peptides have the potential to react across PARs. For example, the PAR₁ activating peptide SFLLRN-NH₂ can also activate PAR₂.⁵ In lieu of this caveat, activating peptides/peptidomimetics provide useful tools for establishment of SAR and rational drug design because they limit off-target effects that are often a complication of natural protease activation and allow for a more potent and specific activation of individual PARs. Indeed, the most potent PAR₂ agonist reported to date was developed by modifying a known activating peptide sequence.⁶ The N-terminal serine of SLIGRL-NH₂ was substituted with 2-furoyl, and the resulting analogue 2-furoyl-LIGRLO-NH₂ is 10–25 times more efficient at increasing [Ca²⁺]_i in human and rat PAR₂ expressing cells.^6,7 The development of 2-furoyl-LIGRLO-NH₂ significantly moved research toward a better understanding of PAR₂ agonist activity and established a platform for rational peptidomimetic design. Such peptides, peptidomimetics, and other small molecules⁸ have been demonstrated to have increased specificity and affinity for PAR₂ and have been used to both activate PAR₂ and, in high concentrations, desensitize cells to subsequent PAR₂ responses in model cells.^6,7,9 Besides therapeutic intervention, agonists of PAR₂ can serve as tools for SAR and for defining the cellular and molecular signaling mechanisms that may contribute to pathology.

Although 2-furoyl-LIGRLO-NH₂ has been useful in many PAR₂ cellular and molecular studies, the furane ring is not considered as a safe structural element for drug design.¹⁰ Furane-containing drugs (including the structurally similar 2-acetyl furane and at least eight other drugs like the diuretic drug furosemide) can be metabolically activated by cytochrome P450 to form electrophilic intermediates (γ-ketocarboxylic acids) by oxidative opening of the furane heterocycle. These reactive metabolites have been shown to bind covalently to liver proteins and cause hepatotoxicity in vivo. Because furane-based structures present a potential risk for hepatotoxicity, substitutes for the furane in 2-furoyl-LIGRLO-NH₂ may provide a more optimal drug design paradigm.

Despite the potential for drug development from known peptide/peptidomimetic PAR₂ agonists as starting points for drug discovery (e.g., SLIGVK-NH₂, SLIGRL-NH₂ and 2-furoyl-LIGRLO-NH₂), there are limited reports on SAR^5,8b,11 and a limited selection of small molecule compounds that can be used as tools to understand the contribution of PAR₂ to cellular and tissue function. We hypothesized that the 2-furoyl residue could be substituted by other metabolically stable heterocycles at the N-terminus of the peptide. Consistent with this hypothesis, Barry et al.^8b has presented a small compound library based on a X-LIGRLI-NH₂ hexapeptide motif. Results using a Ca²⁺ mobilization cellular assay to evaluate the X-LIGRLI-NH₂ library showed that (i) no heterocyclic substitution was better than 2-furoyl-LIGRLI-NH₂, (ii) aliphatic residues lead to less Ca²⁺ mobilization, (iii) aromatic residues were required, but electron donating effects of the heteroatom can potentiate the activity (2-furoyl-LIGRLI-NH₂, EC₅₀ = 0.16 μM, compared to 4-phenyl-LIGRLI-NH₂, EC₅₀ = 0.8 μM, and (iv) steric effects can diminish activity (2-furoyl-LIGRLI-NH₂, EC₅₀ = 0.16 μM, compared to 2-benzofuranyl-LIGRLI-NH₂, EC₅₀ = 0.25 μM; phenyl-LIGRLI-NH₂, EC₅₀ = 0.8 μM, compared to 4-biphenyl-LIGRLI-NH₂, EC₅₀ = 15.8 μM). These results supported our hypothesis that a metabolically stable modification of 2-furoyl-LIGRLO-NH₂ can be found. We focused our initial design on truncation and conformational constraint of the peptide chain in an effort to move toward small molecule drug design for PAR₂. Here, we present potential PAR₂ ligands based on the shortened pentapeptide sequence X-LIGRL-NH₂.

RESULTS

Synthesis

Compound peptidomimetics were synthesized using standard Fmoc chemistry on Rink amide resin.¹² Compounds were cleaved off the resin with TFA-scavenger cocktails and purified by PR-HPLC and/or size-exclusion chromatography. All compounds were 95%+ pure from analytical HPLC and expected MS results.

Agonist Design

Our goal was to design potent and selective agonists of PAR₂ that contain a metabolically stable substitution of the furane ring at the N-terminus. Furthermore, we strived to modify the peptide portion of the agonist into a peptidomimetic structure to increase stability for systemic applications. In contrast to a previous reported library used to help guide our design,^8b we used the LIGRL-NH₂ pentapeptide instead of the more active but longer hexapeptides X-LIGRL-NH₂.

Ca²⁺ Mobilization Assay

A set of 15 compounds based on the X-LIGRL-NH₂ sequence were first screened in model epithelial cells (16HBE14o-) by monitoring changes in [Ca²⁺]_i in response to an initial compound dose of 100 μM (based on activity of native SLIGRL-NH₂ peptide activation). [Ca²⁺]_i was determined for 80–100 cells using the Ca²⁺-sensitive fluorescent dye fura-2 and digital imaging microscropy.¹³ To quantify the response, cells were determined to be “activated” if their [Ca²⁺]_i reached 200 nM or more (from a resting value of 50–75 nM) in a 3 min time period. The percentage of activated cells for each experiment was determined. Compounds 11–16 lacked activity in the Ca²⁺ mobilization assay (0% activated cells) at 100 μM dose and were eliminated from further screening. Compounds 5–10 activated cells at doses of 100 μM but failed to activate cells at doses of 2.5 μM (for details see Supporting Information Table S1). Compounds 1–4 displayed activated Ca²⁺ responses in a significant percentage of cells (>5%) at doses of ≤2.5 μM. EC₅₀ values were calculated for these four agonists. As a point of reference, SLIGRL-NH₂ has an approximate EC₅₀ of >40 μM in this assay (data not shown).

Because there are a variety of mechanisms that can lead to [Ca²⁺]_i mobilization in epithelial cells, we examined the specificity of the novel compounds for PAR₂. We redeployed the Ca²⁺ mobilization assay in either low expressing PAR₂ HeLa cells and HeLa cells transfected with human PAR₂ (HeLa-PAR₂) or low expressing kNRK and cells transfected with human PAR₂ (kNRK PAR₂). We found that 2-furoyl-LIGRLO-NH₂ and the novel compounds 1 and 2 were able to quickly stimulate increases in [Ca²⁺]_i in HeLa-PAR₂ cells (not shown) or kNRK-PAR₂ cells (Figure 1). In the absence of PAR₂ overexpression, agonist-induced [Ca²⁺]_i changes were minimized and representative of background PAR₂ expression in these cell types (Figure 1). From these data we conclude that compounds 1 and 2 represent novel, potent, and selective full agonists at PAR₂.

Figure 1.

Ca²⁺ signaling induced by novel agonists in the presence or absence of PAR₂: typical experiments showing the average [Ca²⁺]_i (±SEM) plotted over time for all cells (~90) in a single experiment following agonist exposure. In each experiment 5 μM agonist was added (denoted by line) to PAR₂ expressing kNRK cells (top) or plasmid expressing control kNRK cells (bottom). In the PAR₂-transfected cells, agonists induce rapid and robust changes in [Ca²⁺]_i. However, in the absence of PAR₂ overexpression, [Ca²⁺]_i response is minimal. Similar traces are observed with PAR₂-transfected/nontransfected HeLa cells (not shown).

In Vitro Physiological Response Assay

Novel agonists 1 and 2 with Ca²⁺ mobilization assay EC₅₀ values approaching that determined for 2-furoyl-LIGRLO-NH₂ were evaluated along with 2-furoyl-LIGRLO-NH₂ in an in vitro physiological response assay using the xCELLigence real time cell analyzer (RTCA, Table 1).¹⁴ The RTCA records changes in impedance (reported as a cell index) over a prolonged time course in a noninvasive system. These readings represent a physiological output that reflects the combined cellular effects of multiple signaling pathways such as those activated by PAR₂.^1,4 Each agonist was applied to the cells over an appropriate dose range and cell index was monitored every 15 s for 2 h. We found that agonist activity was similar among the three compounds, with 2-furoyl-LIGRLO-NH₂ and compound 1 displaying nearly identical RTCA EC₅₀ values and compound 2 displaying less potent activity (Figure 2).

Table 1.

Ca²⁺ Mobilization and in Vitro Physiological (RTCA) Assays

		[Ca2+]i measurements
	R-CO-Leu-Ile-Gly-Arg-Leu-NH2a	% response 100 μM	% response 10 μM	% response 2.5 μM	EC50 (μM)	xRTCAb EC50 (nM)
	2-furoyl-Leu-Ile-Gly-Arg-Leu-Orn(Aloc)-NH2	100 ± 0	97.8 ± 1.0	93.6 ± 3.3	0.84 ± 0.08	138 ± 13
1	2-aminothiazol-4-yl	100 ± 0	96.2 ± 1.4	67.5 ± 7.5	1.77 ± 0.24	142 ± 18
2	6-aminonicotinyl	100 ± 0	93 ± 2.0	55 ± 11.6	2.60 ± 0.32	311 ± 26
3	6-aminopyridin-2-yl	99.3 ± 0.7	94.4 ± 2.6	27.1 ± 6.9	3.50 ± 0.36	ND
4	4-aminophenyl	93.0 ± 0.3	86.9 ± 4.1	16.5 ± 5.3	6.59 ± 1.7	ND

^aThe structures do not contain C-terminal Orn as 2-furoyl-Leu-Ile-Gly-Arg-Leu-Orn(Aloc)-NH₂.

^bEC₅₀ = concentration of compound that was able to generate 50% maximal intracellular activity. Values are expressed ± SEM (N ≥ 2 for 100 μM; N ≥ 4 for all others in Ca²⁺ assay; data are expressed % response ± SEM; N ≥ 8 in the RTCA assay). ND = not determined.

Figure 2.

Physiological signaling induced by novel agonists. Dose response curves were calculated based on area under the curve for 2 h of agonist exposure and calculated from ½ log dose steps from 10 μM down to 10 nM for each agonist.

DISCUSSION

To date, there has not been a comprehensive SAR study on N-terminal modifications of short pentapeptide sequences (e.g., X-LIGRL-NH₂) built as agonists to PAR₂. Therefore, we have designed and synthesized a set of 15 analogues to investigate these N-terminal modifications and their ability to activate PAR₂ (Scheme 1). The analogues were first evaluated using a Ca²⁺ mobilization screen. This initial screen indicated a subset of N-terminal modifications that approached the potency of the furan analogue 2-furoyl-LIGRLO-NH₂ with modifications that have structural similarity to be used in further SAR.

Scheme 1.

Solid-Phase Synthesis of Compound 1 and Structures of PAR-2 Agonists^a

^a (a) (i) Fmoc/^tBu synthetic strategy; (ii) 10% piperidine in DMF, 2 + 20 minutes; (iii) HBTU coupling of 2-aminothiazole-4-carboxylic acid;

(b) (i) 91% TFA, 3% thioanisole, 3% 1,2-ethanedithiol, 3% water, 4 h; (ii) ether extraction; (iii) HPLC and SEC purifications.

We found that aliphatic modifications such as diglycolyl (15), 1-piperidinethyl (14), and 2-cyanomethyl (10) were ineffective (structures are depicted in Scheme 1). Heteroaromatic substituents also did not guarantee PAR₂ activity as documented for indole derivatives (11 and 13) or 2-hydroxypyridinyl (12). However, a subset of heteroaromatic substituents (5–9) elicited [Ca²⁺]_i activation at the highest doses tested (100 μM) but failed at doses shown to be effective for highest activity compounds (e.g., 10 or 2.5 μM). In this manner, they were as effective as the native peptide sequence (SLIGRL-NH₂). The most potent agonists created (1–4) share a common structural feature, nitrogen based heterocycle (thiazole or pyridine) with an amino group orientated in the ortho position to the nitrogen (Table 1). 2-aminothiazol-4-yl (1) induced Ca²⁺ mobilization EC₅₀ (1.77 μM) that was higher but not significantly different from that induced by 2-furoyl-LIGRLO-NH (Ca²⁺ EC₅₀ = 0.84 μM). In contrast, 6-aminonicotinyl (Ca²⁺ EC₅₀ = 2.6 μM) (2) and 6-aminopyridin-2-yl (3) (Ca²⁺ EC₅₀ = 3.5 μM) derivatives induced Ca²⁺ mobilization EC₅₀ values that were significantly higher than that observed for 2-furoyl-LIGRLO-NH₂. In 4-aminophenyl (4), the nitrogen atom was removed from the aromatic ring of the nicotinyl residue and the Ca²⁺ mobilization EC₅₀ dropped to 6.5 μM. Other isomers of aminonicotinic acid analogues such as 6-aminopyridin-4-yl (6) and 6-aminopyridin-5-yl (7) or 3-aminotriazine (8) were ineffective at 2.5 μM. From this short peptapeptide SAR we conclude that both the amino group and the aromatic nitrogen contributed to the PAR₂ Ca²⁺ signaling agonistic activity.

Because PAR₂ activation can lead to downstream signaling pathways in addition to Ca²⁺ signaling, we evaluated the most potent PAR₂ agonists on 16HBE14o- cells using RTCA, an in vitro physiological assay. This high-throughput assay detects changes in cell membrane interaction with a tissue culture surface by continually measuring impedance across the tissue culture surface.¹⁴ We selected the two high potency novel agonists (1 and 2) and compared them to the most potent PAR₂ agonist described to date, 2-furoyl-LIGRLO-NH₂. Both novel agonists (1 and 2) stimulated an increase in cell index in a dose-dependent manner and reached their peak responses within 30 min (not shown). Each compound induced the greatest increase in cell index at a dose of 3 μM, which is approximately ¹/₁₀ the concentration required for a maximum response to the native ligand SLIGRL-NH₂ (not shown). Physiological signaling outputs induced by novel agonists (1 and 2) and 2-furoyl-LIGRLO-NH₂ are plotted as dose response curves in Figure 2. The calculated physiological EC₅₀ values are summarized in Table 1. The physiological EC₅₀ of the novel agonist 1 was 142 nM, nearly identical to that of 2-furoyl-LIGRLO-NH₂ (138 nM). A significantly higher physiological EC₅₀ (311 nM) was determined for compound 2. Both agonists (1 and 2) achieved equal E_max values equivalent to that of the 2-furoyl-LIGRLO-NH₂,⁶ indicating that these novel ligands are full agonists at PAR₂. This RTCA analysis demonstrates potent activation of model epithelial cells by known and novel PAR₂ agonists and presents a high-throughput in vitro physiological assay that allows better comparison of full activity of PAR₂ agonists than traditional single pathway analyses (e.g., Ca²⁺ mobilization).

EXPERIMENTAL SECTION

Synthesis

Peptidomimetics were prepared as previously published by solid-phase synthesis as summarized in Scheme 1 on Rink amide TentaGel resin (0.23 mmol/g) using Fmoc/^tBu synthetic strategy and standard DIC and HBTU or HCTU activations.^12c,15 The synthesis was performed in fritted syringes using a Domino manual synthesizer obtained from Torviq (Niles, MI). N-Terminal heterocyclic acids were obtained from Sigma-Aldrich or TCI and coupled using HBTU activation. For compound 4, the Fmoc-protected version (Fmoc-4-aminobenzoic acid) was used and otherwise was acid coupled as purchased. All compounds were fully deprotected and cleaved from the resin by treatment with 91% TFA (3% water, 3% EDT, and 3% TA). After ether extraction of scavengers, compounds were purified by HLPC and/or size-exclusion chromatography (Sephadex G-25, 0.1 M acetic acid) to >95% purity. All compounds were analyzed for purity by analytical HPLC and MS by ESI or MALDI-TOF.

Ca²⁺ Mobilization Assay

Model epithelial cells (16HBE14o-, HeLa, kNRK) were grown on matrix coated coverslips for 5–7 days with appropriate growth and selection medium. Cells were loaded with the Ca²⁺ sensitive dye fura-2 and evaluated for intracellular calcium concentration ([Ca²⁺]_i) over 3 min using digital imaging microscopy.^13,16 For calculation of Ca²⁺ mobilization EC₅₀, response was quantified by the percent of cells responding in the field of view with an increase of [Ca²⁺]_i of ≥200 nM. Compounds were initially tested at 100 μM and down to 250 nM, as appropriate (Table 1, Supporting Information Table S2). To demonstrate specificity, the average [Ca²⁺]_i among all cells (±SEM) in the field of view was calculated over time with [Ca²⁺]_i measured every 0.6 s (Figure 1).

In Vitro Physiological Response Assay

16HBE14o- cells were plated onto 96-well E-plates (Roche) and grown in fully supplemented growth medium to 95% confluence overnight at 37 °C, 5% CO₂, and monitored for proliferation using the xCELLigence real time cell analyzer (RTCA; Roche Diagnostics).^14a Prior to the experiment, well cultures were washed and replaced with 100 μL of modified Hank’s balanced saline solution (HBSS) prewarmed to 37 °C and allowed to come to room temperature (30–45 min). Each well was supplemented with 100 μL of HBSS containing appropriate agonists to measure seven concentrations for three agonists in quadruplicate. Additional wells (eight) were used for HBSS controls (0 mM agonists). Concentrations were from 10 μM to 10 nM in ½ log steps. Relative impedance in each well was monitored every 15 s over 2 h. For evaluation purposes, relative impedance at any given time is expressed as a “cell index.” Cell index is defined as (Z_i – Z_o)/(15 Ω), where Z_i is impedance at a given time point during the experiment and Z_o is impedance before the addition of the agonist. Average cell index for each agonist/dose (n = 4) was graphed over time. Physiological EC₅₀ values were calculated from the area under the curve values.

ABBREVIATIONS USED

Aloc

allyloxycarbonyl

Boc

tert-butyloxycarbonyl

BB

bromophenol blue

CH₃CN

acetonitrile

DCM

dichloromethane

DI

deionized

DIPEA

diisopropylethylamine

DMF

N,N-dimethylformamide

DIC

diisopropylcarbodiimide

DMEM

Dulbecco’s modified Eagle medium

Fmoc

fluorenylmethoxycarbonyl

FTICR

Fourier Transform ion cyclotron resonance

ESI-MS

electrospray ionization mass spectrometry

EDT

1,2-ethanedithiol

Et₂O

diethyl ether

HBSS

Hank’s balanced saline solution buffered with 25 mM Hepes

HCTU

O-[1H-(6-chlorobenzotriazol-1-yl)(dimethylamino)ethylene]uranium hexafluorophos phate N-oxide

HEPES

4-(2-hydroxyethyl)-1-piperazineethane-sulfonic acid

HOBt

N-hydroxybenzotriazole

HOCt

6-chloro-1-hydroxybenzotriazole

GPCR

G-protein-coupled receptor

MALDI-TOF

matrix assisted laser desorption ionization time of flight

PAR₂

protease-activated receptor type 2

Pbf

2,2,4,6,7-pentamethyldihydrobenzofuran-5-sulfonyl

SPPS

solid-phase peptide synthesis

RP-HPLC

reverse-phase high performance liquid chromatography

RCTA

xCELLigence real-time cell analyzer

TA

thioanisole

THF

tetrahydrofuran

TIS

triisopropylsilane

TFA

trifluoroacetic acid