N -(Heteroaryl)thiophene sulfonamides as angiotensin AT2 receptor ligands

C T Two series of N -(heteroaryl)thiophene sulfonamides, encompassing either a methylene imidazole group or a tert - butylimidazolylacetyl group in the meta position of the benzene ring, have been synthesized. An AT 2 R selective ligand with a K i of 42 nM was identified in the first series and in the second series, six AT 2 R selective ligands with significantly improved binding affinities and K i values of < 5 nM were discovered. The binding modes to AT 2 R were explored by docking calculations combined with molecular dynamics simulations. Although some of the high affinity ligands exhibited fair stability in human liver microsomes, comparable to that observed with C21 undergoing clinical trials, most ligands displayed a very low metabolic stability with t ½ of less than 10 min in human liver microsomes. The most promising ligand, with an AT 2 R K i value of 4.9 nM and with intermediate stability in human hepatocytes (t ½ = 77 min) caused a concentration-dependent vasorelaxation of pre-contracted mouse aorta.


Introduction
Angiotensin II (Ang II) exerts its actions predominantly through activation of two receptor subtypes; the AT1 receptor (AT 1 R) and the AT2 receptor (AT 2 R).Stimulation of AT 1 R results in hypertension and affects the fluid/electrolyte balance.There are three classes of antihypertensive drugs available on the market that are acting by interfering with the renin− angiotensin system (RAS).Inhibitors of the metalloprotease angiotensin converting enzyme (ACE) [1,2] and the aspartyl protease renin [3] both suppress the formation of Ang II from its precursor angiotensinogen.Antagonism of AT 1 R by angiotensin receptor blockers (ARBs) i.e. the sartans is another established strategy to control blood pressure in hypertensive patients.The first ARB losartan [4,5] was introduced onto the market in 1995.Recently, receptors for a variety of components in RAS have been examined as potential targets for drug intervention [6] and the AT2 receptor emerged as a promising new target for drugs [5,[7][8][9][10].It is notable that AT 2 R frequently mediates opposite responses to those observed after stimulation of AT 1 R [11,12].Furthermore, in contrast to AT 1 R, the receptor is sparsely expressed in healthy subjects but importantly, AT 2 R is strongly up-regulated as result of various types of tissue damage [13][14][15] i.e. neuronal injury [16] and vascular injury [17] and after myocardial infarction [18][19][20] and brain ischemia [21,22].
A large number of selective and potent peptide and non-peptide AT 2 R agonists have been reported [23][24][25][26].The first drug-like AT 2 R agonist C21 (1) that was discovered in our laboratories [27], has been studied extensively and demonstrated advantageous responses in several experimental disease models [9,24,25].Recently structurally very close analogues of C21 were reported [28,29].Most AT 2 R selective agonists comprise an imidazole heterocycle although there are exceptions [30].Notably, migration of the imidazolylmethyl group from the para to meta position converted the AT 2 R agonist 1 into an AT 2 R ligand, C38 (2) exhibiting antagonistic properties in a neurite outgrowth assay, vide infra.(Fig. 1) [31,32].Thus, pretreatment of NG108-15 cells expressing angiotensin AT 2 R, with 2 suppressed the neurite outgrowth that had been induced by the AT 2 R agonists 1 or by the endogenous Ang II, similarly to the well-established angiotensin AT 2 R antagonist, PD123319.Both Ang II and 1 induced a Rap1-dependent activation of p42/p44(mapk), whereas preincubation of cells with 2 inhibited p42/p44(mapk) and Rap1 activation induced by Ang II [32].On the contrary, we recently reported, that 2 acts as a partial agonist in an assay relying on iNOS-derived NO release in M1 phenotypic macrophages [33].Furthermore, we disclosed that an extension of the linker by incorporation of a carbonyl group can provide ligands with high affinity to AT 2 R, e.g. the tert-butyl group substituted ligand 3 [34].Today, both agonists and antagonists to AT 2 R are assessed in the clinic as potential new classes of drugs.EMA401, shown in preclinical studies to exert analgesic properties in neuropathic pain [35][36][37] is a structural analogue of the old prototype AT 2 R antagonists PD-123319 (EMA 200) and PD-126055 (EMA400) and the S-enantiomer of the latter [35,37,38].The analgesic efficacy and safety of EMA401 in patients with postherpetic neuralgia and painful diabetic neuropathy were examined in Phase II studies [39].The chemical structure of EMA401 is very different from the structure of 2. The impact of structural modifications of the AT 2 R ligand 2 (C38) on AT 2 R affinity has not been well established [31,40,41].Furthermore, very little data on the metabolic stability of 2 and related analogous have yet been disclosed [34,40,41].
We herein report a) the synthesis, b) the stability in liver microsomes, c) a proposed binding mode at the AT2 receptor, and d) the binding affinities of two classes of AT 2 R ligands, the N-(heteroaryl) thiophene sulfonamides 5-11 and 13-21, structurally related to 2 and 3, respectively.

Synthesis
The limited number of analogues of C38 (2) that so far have been reported exhibit relatively poor binding affinities [34,40,41].In efforts to improve the latter series of compounds we aim at identifying compounds comprising the meta substitution pattern of 2 and that demonstrate both high affinities to the AT2 receptor and favorable in vitro ADME properties.Such compounds could conceivably be suitable for further examination in various animal models.We explored C38 analogues encompassing various N-heteroaryl sulfonamide groups as replacement of the alkoxycarbonyl group.
So far, essentially all AT 2 R ligands containing biaryl scaffolds are encompassing either a tetrazole ring (although most commonly found among AT 1 R ligands, e.g.losartan) or an alkoxycarbonyl sulfonamide group (O-alkyl sulfonyl carbamate) (e.g. 1 and 2) as carboxylic acid bioisosteres attached to the aryl or thiophene ring [24,25].
The primary sulfonamide 4, obtained by a procedure previously described by our group [40], was proposed as a suitable starting material for the preparation of novel N-heteroaryl sulfonamides.Two different synthetic approaches were envisioned; a) a transition metal catalyzed coupling of 4 with a heteroaryl bromide or b) a nucleophilic aromatic substitution (SNAr) between 4 and e.g. a heteroaryl chloride or fluoride.The 2-thiazolyl sulfonamide 5 was synthesized in 19 % yield by a copper catalyzed coupling of 4 and 2-bromothiazole.A series of compounds (6)(7)(8)(9)(10)(11) was synthesized in rather low yields (2-30 %) by SNAr reactions from 4, employing potassium carbonate and various 2-chloropyrimidines and other chlorinated heteroaromatics with heating in DMF (Scheme 1).
We recently reported that a 2-(tert-butyl)-1H-imidazole-1-yl)acetyl) phenyl group rather than a imidazolylmethyl group such as in the AT 2 R derivative 2 attached to the meta position rendered very potent and selective AT 2 R ligands [34].For the preparation of a series of N-(heteroaryl)thiophenesulfonamides, all comprising the tert-butylimidazolyl group, the previously synthesized sulfonamide 12 [34] was used as starting material.The syntheses of 13-21 were subsequently conducted applying either copper-catalyzed N-arylations of the sulfonamide with the appropriate heteroarylbromide (A) or alternatively the SNAr reaction (B, Scheme 2).Careful optimization of reaction temperature and heating times using microwave (MW) heating [42] and sealed reaction vials was carried out to improve the conversion of starting material 12 regardless of whether methods A and B were used, see Supplementary data.
For the more electron-deficient 2-chloropyrimidines, the SNAr strategy (B) was preferred with cleaner reaction mixtures (entries 15, 17, 18 and 20).However, for entry 15, nucleophilic attack also at the 5fluoro position was observed as a side product.The products synthesized by copper catalysis were observed to decompose in solution in the presence of the catalyst and air.Analysis of the decomposition products suggest that the α-position of the ketone had been oxidized, resulting in the formation of α-keto acids and, subsequently, carboxylic acids.
Similar reactivity has been observed for α-bromo ketones [43].This decomposition could, however, be minimized efficiently in later work by adding an NH 4 Cl (aq.) washing to the work-up procedure in order to remove the Cu species prior to chromatography.

Fig. 1.
An AT 1 R antagonist (losartan), an AT 2 R agonist (1), an AT 2 R ligand (2), an AT 2 R ligand with a carbonyl methylene linker (3) and generic structures of new AT 2 R ligands.Scheme 1.General synthesis of N-heteroaryl sulfonamides from 4.

Results and discussion
The AT 2 R binding affinity and the stability in human and mouse liver microsomes of the N-(butoxycarbonyl)thiophene sulfonamide 2, 3, and the N-(heteroaryl)thiophene sulfonamides 5-11 are shown in Fig. 2. The corresponding data derived from N-(butoxycarbonyl)thiophene sulfonamide 22 [31] and the N-(heteroaryl)thiophene sulfonamides 13-21 are depicted in Fig. 3.
Regarding the affinity to the AT2 receptor, 2 exhibited a K i value of 19 nM [31] in a radioligand assay based on replacement of [125I]Ang II from AT2 receptors in pig uterus membrane, as previously described [44,45].The data presented herein on the AT 2 R affinities of 2, 5-22 are derived from a binding assay using membrane preparations from a human cell line, HEK-293 transfected with and overexpressing human AT 2 R (HEK293-hAT 2 R) where replacement of the non-selective AT 2 R/AT 1 R ligand sarile ([ 125 I][Sar 1 ,Ile 8 ]-angiotensin II) rather than the less metabolically stable Ang II (Eurofins Cerep SA, France) (Figs. 2  and 3) was used.The prototype AT 2 R ligand 2 demonstrated a 10-fold lower affinity, K i = 270 nM, in the HEK-293 cell assay using the peptide ligand sarile, that like Ang II, acts as an AT 2 R agonist [46].
In this series of 1,3-substituted compounds, we also replaced the butoxycarbonyl group of 2 with a series of nitrogen atom containing heterocycles.A replacement with a thiazole, a heterocycle found in the antibacterial drug sulfathiazole, furnished 5, which is less potent than as an AT 2 R binder (2: K i = 270 nM versus 5: K i = 650 nM) but with pronounced increased stability in human liver microsomes, HLM (t ½ = 61 min) as compared to 2 (t ½ = 10 min).Attachment of various pyrimidines provided the compounds 6-10.These pyrimidine compounds, with the exception of 8 (t ½ = 43 min), were all metabolically unstable in the human liver microsomes and none of them displayed a significantly improved affinity to the AT2 receptor compared to 2. Compound exhibited a K i of 1100 nM.The bromopyridine derivative 10, on the contrary exhibits a 7-fold higher AT 2 R affinity (K i = 42 nM) than 2, but the stability in the human liver microsomes remains unsatisfying (t ½ = 14 min) although a fair stability is registered in human hepatocytes (t ½ = 64 min), see Experimental section.Regarding the Caco-2 cell monolayer permeability of 10, the apical to basolateral apparent permeability of 17.3 × 10 − 6 cm/s was revealed, see Experimental section.Replacement of the thiazole ring of 5 for a benzothiazole system provided 11, that is a less potent AT 2 R binder than 2, but like the thiazole 5, the compound demonstrates a high stability in the human liver microsome assay (t ½ = 56 min).In summary, regarding this series of compounds the bromopyrimidine 10 exhibits the highest affinity to the AT2 receptor and a fair apparent cell permeability but an unsatisfying stability in human liver microsomes.
The previously prepared meta-substituted imidazolylacetylphenylthiophene sulfonamide 22 [31], exhibited a K i value of 280 nM and is in essence equipotent to 2 (Fig. 3).Noteworthy, the corresponding benzimidazole analogue of 22 had previously been selected for functionality studies and found to act as an AT 2 R antagonist [31].A comparison of the stability of 22 and 2 revealed that the stability in human liver microsomes was improved (22: t ½ = 30 min versus 2: t ½ = 10 min).For comparison the AT 2 R antagonist EMA401 and the AT 2 R agonist (C21), that both are evaluated in clinic, exhibited in the same human liver microsome assay t ½ values of 38 min and 22 min, respectively.
A replacement of the butoxycarbonyl group of 22 for a thiazole ring in combination with attachment of a tert-butyl group to the imidazole ring, to reduce CYP-mediated metabolism, provided 13 with a K i value of 96 nM.Compared to the thiazole derivative 5, that was considerably more metabolically stable than the butoxycarbonyl derivative 2, the thiazole compound 13 exhibited a very low metabolic stability in HLM (t ½ = 6.7 min).The pyrimidine sulfonamide analogues in general exhibited good AT 2 R binding affinities.Thus, 14 displayed a high affinity to the AT2 receptor and a K i value of 4.9 nM.The bromo Scheme 2. Synthesis of N-heteroaryl sulfonamides from 12. Fig. 2. The AT 2 R affinity and the stability of 2, 3, 5-11 (t ½ in min) in human liver microsomes (HLM) and in mouse liver microsomes (MLM) are shown.compound 17 with a K i of 1.5 nM decomposed very rapidly in the human liver microsome assay.Further elaborations with various substituents at the pyrimidine revealed that both a chloro atom and a fluoro atom delivered very potent binders with K i of 3.3 nM ( 15) and K i = 1.7 nM (16), respectively.While 16 and 17 decomposed in a few minutes in HLM, the fluoro derivative 15 was more stable in this microsome assay.Compound 14 and 15, both potent AT 2 R binders, exhibited the highest stability in microsomes of the sulfonamide ligands assessed in this report.Introduction of an electron withdrawing trifluoromethyl group rendered a similarly potent AT 2 R ligand ( 18), but a pyridine nucleus rather than the pyrimidine resulted a 10-fold decrease in affinity, cf 18 and 19.The introduction of a nitrile group resulted in a compound 20, equally potent to 17, but replacement of the pyrimidine heterocycle in 20 with a benzene ring to provide 21 was deleterious for activity.Furthermore, the metabolic stability of 18-21 in the two liver microsome assays employed were very low.Regarding the selectivity of the compounds studied herein, they exhibited high AT 2 R/AT 1 R ratios and <40 % binding at 10 μM at AT 1 R (saralasin), see Experimental section.
The apparent permeabilities (P app ) across a human intestinal epithelial Caco-2 cell monolayer of the sulfonamide 14 with intermediate stability in human hepatocytes (t ½ = 77 min) and the AT 2 R antagonist 2 (C38) with a somewhat lower stability in human hepatocytes (t ½ = 30 min) were next compared.Compound 14 exhibited a apical to basolateral apparent permeability of 4.9 × 10 − 6 cm/s and 2 a corresponding value 3.4 × 10 − 6 cm/s.Efflux ratios of 77 for compound 14 and 7 for 2 were measured.The former clinical candidate, the AT 2 R antagonist EMA401 with a AT 2 R K i of 5.6 nM in our binding assay demonstrates a poorer apparent permeabilitiy (1.8 × 10 − 6 cm/s) and an efflux ratio of 32 in the in vitro models applied herein [34], but nevertheless it exhibits a good oral bioavalability (33 %) [38].
Ligand 14, exhibiting a high receptor selectivity (AT 2 R, K i = 4.9 nM and AT 1 R, K i > 3 μM) was selected for a functional study and a comparison with 1 (C21).Isolated aortic preparations from mice were previously used to determine the vascular effects of 1 [34,47] Compound C21 evoked relaxation in mouse aorta that was blocked by the nitric oxide synthase inhibitor L-NAME, suggesting that the effect of 1 was mediated by nitric oxide [47].We report that ligand 14 caused a similar concentration-depending vasorelaxation as C21 although the physiological mechanism behind this observation is not fully clarified (Fig. 4).

Modelling
After independent docking explorations in GLIDE, followed by molecular dynamics (MD) refinement, a binding pose was identified for ligand 14.Fig. 5 depicts the proposed binding mode for the very potent compound (14) that is compared to binding mode of 2 (C38).A model of the binding mode of the prototype AT 2 R ligand 2 when binding to the receptor as deduced from docking redefined by molecular dynamic simulations was recently suggested [41].The sulfonamide group of the high affinity ligand 14 is anchored via salt-bridge interactions with R182 4.64 and K215 5.42 and a hydrogen bond between the sulfonamide N and T125 3.33 (the Ballesteros-Weinstein generic amino acid numbering scheme is indicated as superscript [48]).The phenyl ring of 14 is lining towards W100 2.60 and L124 3.32 , allowing the tert-butyl imidazole substituent to accommodate within a hydrophobic cluster composed by residues Y51 1.39 , Y103 2.64 , Y104 2.65 , Y108 2.69 , P301 7.36 and I304 7.39 .The side chain carbonyl group that is essential for high affinity to AT 2 R is bridging to R182 4.64 and Y103 2.64 via a water molecule as shown in Fig. 5.The isobutyl group at the thiophene ring is located in a deeper region of the transmembrane cavity, defined by residues L124 3.32 , M127 3.36 , W269 6.48 , F272 6.51 and F308 7.43 .Finally, the pyrimidine ring is located between transmembrane helices TM3-TM5 defined by the residues I211 5.38 , K215 5.42 , T178 4.60 and F129 3.37 .A hydrogen bond is established between one of the pyrimidine nitrogens and a water molecule bridging to T178 4.60 .The introduction of the carbonyl group of the ketone function of 14 significantly improves the affinity of the   The binding mode proposed for 14 explains to some extent briefly the SAR for the 13-21 series, as depicted in Fig. 5, although to accurately establish binding modes of e.g. the high affinity ligands 17 and 18 more detailed modeling with the individual ligands binding to the flexible receptor must be performed.

Conclusion
The N-(alkoxycarbonyl)thiophene sulfonamide function, has been replaced by N-(heteroaryl)thiophene sulfonamide groups.Two series of ligands were synthesized and assessed regarding binding affinity and stability in liver microsomes.AT 2 R selective ligands, one exhibiting a K i of 42 nM (10), were identified in the first series of compounds comprising a methylene imidazole group in the meta position of the biaryl scaffold.In the second series, AT 2 R selective ligands with a tertbutylimidazolylacetyl group attached to the meta position of the scaffold were found to demonstrate significantly improved binding affinities.Hence, six compounds (14-18 and 20) with K i values of <5 nM were identified.We report that ligand 14, with an AT 2 R K i value of 4.9 nM and with intermediate stability in human hepatocytes (t ½ = 77 min) but low stability in both human and mice microsomes, caused a concentration-dependent vasorelaxation of pre-contracted mouse aorta.A binding mode of compound 14 was obtained by automated docking combined with MD refinement, which explains the SAR of the series.

Chemistry
All solvents and chemicals were used as purchased without further purification.Microwave heating was performed in a Biotage singlemode microwave reactor producing controlled irradiation at 2450 MHz with a power of 0-400 W. The reaction temperature was determined and controlled using the built-in online IR-sensor.Microwave mediated reactions were performed in septum sealed Biotage vials.Analytical TLC was performed on silica gel 60 F-254 plates and visualized with UV light (λ = 254 nm).Automated flash column chromatography was performed on Biotage Isolera or Grace Reveleris X2 instruments using commercial silica cartridges.Analytical HPLC/ESI-MS was performed using UV detection (214, 254 and 280 nm) and electrospray ionization (ESI) MS on a C18 column (50 × 3.0 mm, 2.6 μm particle size, 100 Å pore size) with gradients of acetonitrile in 0.05 % aqueous HCOOH as mobile phase at a flow rate of 1.5 mL/min.High resolution molecular masses (HRMS) were determined on a mass spectrometer equipped with an ESI source and 7-T hybrid linear ion trap (LTQ).Nuclear magnetic resonance (NMR) spectra ( 1 H, 13 C and 19 F NMR) were recorded at 400, 101 and 376 MHz respectively.Chemical shifts (δ) are reported in ppm referenced to trimethylsilane via residual solvent signals. 13C NMR and 19 F NMR spectra were recorded in proton decoupling mode and 19 F NMR spectra were indirectly referenced to CFCl 3 using the solvent lock signal.Coupling constants J are reported in hertz (Hz).All final compounds were ≥95 % pure as determined by HPLC (UV at 254 nm) and NMR.

Stability in liver microsomes
Human and mouse liver microsomes obtained from XenoTech LLC, KS, USA were used to determine the metabolic stability.Metabolic stability was determined in 0.5 mg/mL human or mouse liver microsomes at a compound concentration of 1 μM in 100 mM potassium phosphate buffer (pH 7.4) in a total incubation volume of 500 μL.The reaction was initiated by addition of 1 mM NADPH.At various incubation times, i.e. at 0, 5, 10, 20, 40 and 60 min, a sample was withdrawn from the incubation and the reaction was terminated by addition of cold acetonitrile containing Warfarin as internal standard.The amount of parent compound remaining was analysed by LC-MS/MS.The LC-MS/MS system used was an Acquity UPLC coupled to a triple quadrupole mass spectrometer (Waters), operating in multiple reaction monitoring (MRM) mode with positive or negative electrospray ionization.Mass spectrometric settings were optimized for each compound for one MRM transition.Chromatographic separation was typically done on a C18 Ethylene Bridged Hybrid (BEH) 1.7 μm column using a general gradient of 1 %-90 % of mobile phase consisting of A, 5 % acetonitrile and 0.1 % HCOOH in purified water, and B, 0.1 % HCOOH in 100 % acetonitrile, over a total running time of 2 min.In a few cases, separation was done on a HSS T3 2 × 50mm 2.1 μm column using a mobile phase consisting of A, 0.05 % heptafluorobutyric acid (HFBA) and 0.05 % propionic acid (PA) in water, and B, 0.05 % HFBA and 0.05 % PA in acetonitrile, with a total running time of 2 min.In both cases, the flow rate was set to 0.5 mL/min and 5 μL of the sample was injected.In vitro half-life (t ½ ) and in vitro intrinsic clearance (Cl int ) were calculated using previously published models [50,51].

Metabolic stability in human hepatocytes
Metabolic stability assay was performed in cryopreserved hepatocytes from one healthy donor [52] Thawed hepatocytes were resuspended in pre-warmed Williams medium E. The incubation was carried out on a heater-shaker at 37 • C, using about 350,000 cells/ml and a final compound concentration of 1 μM.Samples were withdrawn at different time-points and the reaction was quenched by adding ice-cold acetonitrile containing 50 nM Warfarin as internal control.Samples were centrifuged at 3000 rpm for 15 min and before analysis of parent compound remaining by LC-MS/MS as described above.

Caco-2 cell permeability assay
Permeability measurements were performed as previously described [53].Caco-2 cell monolayers (passage 94-105) were grown on permeable filter supports and used for transport studies on day 21 after seeding.Prior to start of the experiment, culture medium was replaced with preheated Hank's Balanced Salt Solution (HBSS) buffered with HEPES to pH 7.4.Apparent permeability was measured, at a compound concentration of 5 μM, in both apical-to-basolateral (A-B) and basolateral-to-apical (B-A) direction at pH 7.4 using a shaking speed of 500 rpm.Immediately after the start of the experiment, a sample was removed from the donor compartment (C0), and subsequent samples were taken from the receiver compartment at 15, 30 and 60 min.At the end of the experiment (t = 60 min), a sample was removed from the donor chamber (Cf) for mass balance calculation.Each experimental condition was assayed in triplicate.Compound concentration was determined by LC-MS/MS as described above.

Radioligand binding assay
All synthesized ligands were evaluated in a radioligand binding assay (Eurofins Cerep SA, France) by replacing [ 125 I][Sar 1 Ile 8 ]-angiotensin II from HEK-293 cells expressing human AT 2 R. [Sar 1 Ile 8 ]-angiotensin II (Sarile) acts as a nonselective AT 2 R agonist [46].The affinity was determined using a seven-point dose-response curve and the measurements were performed in duplicates, the AT 2 R selective antagonist PD-123,319 was used as reference compound.The compounds were also evaluated for inhibition of [ 125 I][Sar 1 Ile 8 ]-angiotensin II binding to human AT 1 R expressed in HEK-293 cells.For AT 1 R the percent inhibition was determined at 10 μM, in duplicates, with the endogenous ligand (angiotensin II) used as reference.

Vasorelaxation
For vasorelaxation studies, thoracic aorta obtained from male FVB/N mice were pre-contracted to 30-40 % maximal contraction to thromboxane AT2 receptor agonist U46619, after which a concentration response curve was obtained for C21 or ligand 14 in separate tissues, as described previously [34].

Modelling experiments
The crystal structure of the active-like human AT 2 R was retrieved from the Protein Data Bank (PDB code 5UNG with antagonist L-161,638 [54,55].and was subject to preparation and minor modifications with the Schrödinger suite (Schrödinger Release 2017-3, Schrödinger, LSS, New York, NY, 2017), including (i) deletion of the engineered B562RIL protein (fused to the truncated N-terminus); (ii) addition of protons, assessment of the rotamers for Asn/Gln/His residues, and protonated state for titratable residues, resulting in all Asp, Gln, Lys, and Arg residues assigned to their default charged state and all His modelled as neutral with the proton on Nδ; (iii) addition of missing side chains, modelling the most probable conformer based on additional crystal structures of AT2 and the related AT1 receptor.

Ligand docking
Ligands from Figs. 2 and 3 were built and their 3D conformation were optimized using the Maestro graphical interface and the LigPrep utility from the Schrödinger suite (Schrödinger Release 2017-3: Maestro, Schrödinger, LSS, New York, NY, 2017; Schrödinger Release 2017-3: LigPrep, Schrödinger, LSS, New York, NY, 2017).This method also allowed determination of their most probable protonation state at physiological pH, with a net negative charge localized on the sulfonylcarbamate group in all cases.Docking was performed with Glide SP using default settings (Schrödinger Release 2017-3: Glide, Schrödinger, LSS, New York, NY, 2017) [56][57][58].The docking grid centroid was placed taken as reference the coordinates of the co-crystallized ligand L-161,638, and expanding the cubic grid box was set to 30 Å on each dimensions.The selection of poses was done on the basis of a double criterion, combining the highest possible scoring while looking for the consensus among all ligands in the series.

Membrane insertion and molecular dynamics equilibration
Each Ligand receptor complex obtained in the previous stage was subject to an MD equilibration following the PyMedDyn protocol, as implemented in a GPCR-ModSim web server [59,60].Briefly, the receptor-ligand complex was inserted in a pre-equilibrated membrane consisting of 1-palmitoyl-2-oleoyl phosphatidylcholine (POPC) lipids, with the transmembrane (TM) bundle aligned to its vertical axis.The simulation box was created with a hexagonal-prism geometry, which was soaked with bulk water and energy-minimized using the OPLS-AA force field for proteins and ligands, combined with the Berger parameters for the lipids [59,[61][62][63].It follows a molecular dynamics equilibration using periodic boundary conditions (PBC) and the NPT ensemble with the GROMACS simulation package [61].
The first phase consists of 2.5 ns with a gradual release of harmonic restraints on protein (and ligand) heavy atoms.The second phase consists of free MD for another 2.5 ns, except for weak distance restraints between 24 pairs of interacting residues corresponding to conserved positions within the TM bundle of class-A GPCRs with a structural role [60,64].The final snapshot was energy minimized and retained for analysis and figures.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 3 .
Fig. 3.The AT 2 R affinity (K i ) and the stability of 13-22 (t ½ in min) in human liver microsomes (HLM) and in mouse liver microsomes (MLM) are shown.

Fig. 5 .
Fig. 5. Ligands 2 (C38) and 14 in complex with the AT 2 R, with the main polar interactions maintained after MD equilibration depicted in dashed lines.The Nterminal, EL3 and parts of TM6-TM7 of the AT 2 R are not shown for better clarity.