Development of Photoactivatable Allosteric Modulators for the Chemokine Receptor CXCR3
Introduction
Investigation of structural determinants of the interactions be- tween small molecules and G protein-coupled receptors (GPCRs) is crucial to understand the binding modes, mecha- nisms of action, and to facilitate structure-based drug discov- ery.[1]
Chemical crosslinking with photoactivatable ligands (photoaffinity labeling; PAL) is attracting tremendous interest as a biochemical tool for structural elucidation of the binding pocket of a protein.[2] It was introduced by Frank Westheimer in the early 1960s.[3]
PAL can be applied to investigate ligand– receptor interactions, to identify the binding pocket, or to dis- cover unknown biological targets.
In PAL, the photoactivatable ligand is first incubated with its receptor to form a reversible (noncovalent) ligand–receptor complex in the binding site due to its intrinsic affinity for the receptor.
Upon UV radiation, a highly reactive intermediate capable of crosslinking to the nearby fragment of the receptor is released that binds the ligand covalently to the receptor. The labeled seg- ment of the receptor can then be identified by subsequent mass spectrometry analysis.
A typical photoactivatable ligand contains a photoreactive group (photophore), a known target-specific reversible ligand that directs the probe to the binding site of the protein, and a reporter tag (radioactive, chromophore, fluorescent, or im- munoreactive group) for the detection of the crosslinked prod- uct.[4]
Importantly, the selected photophore and the reporter tag should not significantly impair binding. The most widely used photoreactive groups are arylazides, 3-(trifluoromethyl)-3- phenyldiazirines and benzophenones, which elicit nitrenes, car- benes or benzophenone-derived diradicals, respectively, when exposed to UV light at the appropriate wavelengths.[5,6]
Each photophore has its own advantages and disadvantag- es. Diazirines and benzophenones are activated at a longer wavelength (~ 360 nm), which prevents protein damage con- trasting to azides, which need to be irradiated at a shorter wavelength (~ 250 nm).[7]
Photoactivated benzophenones are less reactive than carbenes and can be used for PAL even in the presence of water molecules in the binding pocket.[8] Al- though the small size and favorable activation wavelength im- plies that diazirines are optimal photophores, their use is limit- ed by the fact that they can undergo undesired side reactions during photolysis, which can, for example, result in the forma- tion of a diazo isomer.[9]
Recently, the PAL approach was suc- cessfully used to identify the binding pocket of many GPCR li- gands such as the inverse agonist T140 on the CXCR4,[10] sub- stance P on neurokinin 1,[11] and propofol on the GABAA recep- tor.[12]
In our study, we explored the possibility to covalently label the binding pocket of 8-azaquinazolinone derivatives, which were identified as promising allosteric modulators of the che- mokine receptor CXCR3 with enhanced properties like signal- ing bias and probe-dependence.[13,14] CXCR3 is a member of the group A or rhodopsin-like GPCRs.[15,16]
It directs activated T cells to the sites of inflammation when bound to its chemo- kines CXCL9, CXCL10 or CXCL11.[16] Because of its involvement in several inflammatory and autoimmune diseases, such as ul- cerative colitis,[17] multiple sclerosis,[18] atherosclerosis,[19] rheu- matoid arthritis,[20] type 1 diabetes,[21] chronic obstructive pul- monary disease[22] and asthma,[23] CXCR3 is a promising thera- peutic target for their treatment.
Currently, the structure-based drug design of agents target- ing CXCR3 is hampered by the lack of structural information describing in detail the interactions between an allosteric ligand and the receptor.
Some limited site-directed mutagene- sis studies have begun to propose the amino acid residues in- volved in the binding of low-molecular-weight compounds to CXCR3.[14,24] Due to limitations of the mutagenesis experiments, obtained results need to be supported by further structural studies.
In order to develop allosteric ligands that would enable PAL at CXCR3, we designed, synthesized and biologically character- ized 8-azaquinazolinone derivatives carrying various types of photoreactive moieties. Compound 10 containing a 3-trifluoro- methyl-3-phenyldiazirine moiety was identified as promising tool for further PAL studies at CXCR3.
Results and Discussion
Synthesis
We have synthesized a series of photoactivatable ligands with different photophores (azide, aliphatic diazirine, trifluoro-meth- ylaryldiazirine or benzophenone) attached to the 2,3-disubsti- tuted 8-azaquinazolinone scaffold known from our lead com- pounds AMG 487 and ‘cold’ (non-tritiated) RAMX3 (Figure 2).
The position for the photoreactive groups was selected based on the structure–activity relationship studies of quinazolinone- derived antagonists of CXCR3, which showed that the modifi- cations of the given fragment of the molecule are well tolerat- ed by the receptor and do not result in a significant loss of af- finity.[14, 25–27]
The synthesis of the racemic starting material 1 was carried out by applying a protocol described by Johnson et al.[26] and pound 2 with 4-fluoro-3-(trifluoromethyl) phenylacetic acid yielded compound 5. Compound 6 was prepared by tert-butyl- oxycarbonyl (Boc) deprotection of compound 5 and subse- quent amide coupling with 3-(3-methyl-3H-diazirin-3-yl)propa- noic acid.
To introduce the biotin affinity tag, secondary amine 2 was acylated with fluorenylmethyloxycarbonyl (Fmoc)-pro- tected 15-amino-4,7,10,13-tetraoxapenta-decanoic acid fol- lowed by Fmoc deprotection and acylation with (+)-biotin N-hydroxysuccinimide ester.
The Boc protecting group was removed and the resulting piperidine was then coupled with 3- (3-methyl-3H-diazirin-3-yl)propanoic acid to furnish the final product 9 (Scheme 3).
Deprotection of compound 5 (prepared from 1 as shown in Schemes 1 and 3) and treatment with 4-[3-(trifluoromethyl)-3H- diazirin-3-yl] benzyl bromide in the presence of potassium car- bonate led to the formation of the photoactivatable com- pound 10 (Scheme 4).
Compound 12 was obtained by reaction of the starting material 1 with 4-[3-(trifluoromethyl)-3H-diazir- in-3-yl]benzyl bromide and subsequent acylation of 11 with in situ-activated 4-fluoro-3-(trifluoromethyl)phenylacetic acid.
Deprotection of 5 and acylation of piperidine 13 with Boc-4- benzoyl-l-phenylalanine in the presence of bromotripyrrolidi- nophosphonium hexafluorophosphate (PyBroP) generated amide 14. Compound 15 was obtained through Boc deprotec- tion of its precursor compound 14 (Scheme 5).
Photolabile aromatic azide 24 (Scheme 6; see also Support- ing Information) was prepared in four steps from 20 (analogue of 1). Mono-trichloroethyl chloroformate (Troc)-protected 1,4- phenylenediamine was used in the assembly of the 8-azaqui- nazolinone heterocyclic system from 2-aminonicotinic acid and N-Boc-d-alanine under treatment with isobutyl chloroformate and N-methyl morpholine.
Primary amine 20 was afforded after removal of the Boc protecting group. The Troc group was removed under reductive conditions, and the liberated aromat- ic amine was converted to azide via an SNAr reaction by treat- ment with tert-butyl nitrite and trimethylsilyl azide.
Pharmacological characterization
To validate the affinity and intrinsic activity of the novel com- pounds, we carried out radioligand displacement and cAMP accumulation assays.
The radioligand displacement assay was performed using a radiolabeled negative allosteric modulator of CXCR3, [3H] RAMX3, which was synthesized in our group for the deter- mination of affinities of structurally similar analogues pre- sumed to occupy the same binding pocket.[27]
The binding assay was performed on membrane preparations of HEK 293T cells transiently expressing CXCR3. All the photoactivatable ligands exhibited micro- to nanomolar affinity towards the CXCR3 receptor. Compound 12 (pKi = 8.47) exhibited the best affinity towards CXCR3 receptor followed by compound 10 (pKi = 7.93), which is comparable to the reference compound, ‘cold’ RAMX3 (cRAMX3) (pKi = 8.85).[27]
In general, the diazirine-derived compounds (6, 10 and 12) displayed better affinities than their benzophenone counterparts, 14 and 15 (one-way ANOVA followed by Tukey’s multiple comparison test: 6 vs 14: not significant; 6 vs 15: p < 0.001; 10 vs 14: p < 0.05 ; 10 vs 15 : p < 0.001; 12 vs 14: p < 0.001; 12 vs 15 : p < 0.001), which indicates that the steric demand of the bulky benzophenone group was not fully tolerated in the CXCR3 binding pocket. The introduction of the polyethylene glycol (PEG) linker and biotin, as present in compound 9, led to a strong decrease in affinity, also. Conclusions To facilitate the explorations of the allosteric binding pocket at the CXCR3 receptor, we have synthesized photoactivatable compounds based on a 2,3-disubstituted-8-azaquinazolinone scaffold. The compounds were characterized with [3H]RAMX3 radioligand binding and cAMP accumulation assays. Com- pounds 6, 10 and 12 exhibited nanomolar affinity towards the CXCR3 receptor. Except the biotin labeled compound 9, which had no effect on CXCR3, all test compounds were found to be negative allosteric modulators of CXCR3. Moreover, we carried out a radioligand displacement assay to probe irreversible in- teraction of the compounds with the target receptor. The high- est degree of covalent binding to CXCR3 receptor was ob- served for photoactivatable compound 10 bearing a 3-trifluor- omethyl-3-phenyldiazirine group. This led to a reduction of radioligand binding by 80 % as evidenced by [3H] RAMX3 radio- ligand displacement assay. Compound 10 could thus be used as a promising chemical tool for further exploration of the allo- steric binding site of CXCR3 by mass spectrometry. Experimental Section Biology Cell culture and transfection: Human embryonic kidney (HEK) 293T cells were cultured in a 100 mm cell culture plate in Dulbecco’s modified Eagle medium/nutrient mixture F-12 (DMEM/F-12) supplemented with 10 % fetal bovine serum (FBS), 2 mm l-glutamine, and 1% penicillin–streptomycin, and incubated at 378C in a humid atmosphere with 5% CO2. At 50–70 % confluency, the cells were transiently transfected with 2 mg of CXCR3 cDNA using TransIT-293 transfection reagent (Mirus Corporation) and were harvested 48 h after transfection. CXCR3 HEK membrane preparations: 48 h post transfection, cells were washed with phosphate-buffered saline (PBS) twice and har- vested using a scraper. Afterwards, cells were treated with Tris EDTA buffer (10 mm Tris, 0.5 mm EDTA, 5 mm KCl, 140 mm NaCl, pH 7.4), and harvested using a cell scraper. Cells were pelleted by centrifuge at 1100 g for 8 min at 4 8C, resuspended in Tris-EDTA-MgCl2 buffer (50 mm Tris, 5 mm EDTA, 1.5 mm CaCl2, 5 mm MgCl2, 5 mm KCl, 120 mm NaCl, pH 7.4) and lysed with an Ultra-Turrax. After centrifugation at 50000 g for 18 min at 4 8C, the membranes were resuspended in the binding buffer (50 mm Tris, 1 mm EDTA, 5 mm MgCl2) and subsequently homogenized with a glass-Teflon homogenizer (20 strokes). The homogenized membranes were shock-frozen in liquid nitrogen and stored at 808C. The protein concentration was determined by using the Lowry method [34] with bovine serum albumin (BSA) as a standard. Radioligand displacement assay: Receptor binding studies were per- formed on membrane preparations of HEK 293T cells expressing the human CXCR3 receptor. The tritium-labeled RAMX3 (specific activity: 80.4 Ci mmol—1) at a concentration of 1.5 nm was used for the assays.[27] To determine unspecific binding, NBI-74330 (5 mm) was used. The assays were carried out in 96-well plates at a protein concentration of 30 mg mL—1 in a total volume of 200 mL. The incu- bation buffer contained 20 mm HEPES, 10 mm MgCl2, 100 mm NaCl and 0.1 % BSA (pH 7.4). After incubating for 1h at 378C, the binding was stopped by filtration through Whatman GF/B filters using a 96-channel cell harvester (Brandel, Unterföhring). The filters were rinsed five times with ice cold Tris-NaCl buffer. After drying for 3h at 608C, filters were sealed with melt-on scintillator sheets Melti-Lex G/HS, and the trapped radioactivity was measured in a micro- plate scintillation counter (Micro Beta Trilux scintillator). Three to five experiments per compound were performed with each con- centration in triplicate. cAMP accumulation assay: HEK 293T cells transiently transfected with 2 mg of human CXCR3 and 2 mg of the biosensor CAMYEL using TransIT-293 transfection reagent from Mirus 293 (Mirus Bio LLC, Madison, USA) were used for the determination of cAMP level. 24 h after transfection, the cells were harvested with Versene (Life technologies) and resuspended in DMEM/F-12 phenol red free medium. The cells were seeded on a 96-well white half-area plate (Grainer-Bio One) with 20000 cells per well in 100 mL DMEM/F-12 phenol red free medium. After incubation overnight at 378C (5 % CO2, 95 % relative humidity), media was aspirated and replaced with 30 mL of Dulbecco’s phosphate-buffered saline (DPBS) with CaCl2 and MgCl2 (Gibco, Life Technologies). The bioluminescence substrate colenterazine h (Promega, Madison, USA) was added after 1h incubation at a final concentration of 5 mm, and the cells were incubated for 5 min in dark. Various concentrations of test and reference compounds (prepared in DPBS (pH 7.4) containing 0.2 % BSA, in the presence of 5 nm CXCL11, 10 mm forskoline, 0.1 % DMSO) were added to a final volume of 50 mL. Bioluminescence was measured after 15 min by using a CLARIOstar microplate reader (BMG, labtech, Ortenburg, Germany). Three experiments per compound were performed with each concentration in triplicate. Photoaffinity labeling and radioligand binding studies : Stock solu- tions of photoactivatable ligands were prepared in DMSO to a con- centration of 10—2 m. Ligands were diluted to lower concentrations with HEPES buffer. Membrane preparations of HEK 293T cells transi- ently transfected with CXCR3 were used for the assay. Photoaffinity labeling was performed using enriched membrane preparations and the photoactivatable compounds. The complex in HEPES buffer was incubated for 1h at 37 8C. After binding, membranes were exposed to photolysis for 45 min on ice, at a wavelength of 366 nm using a UV lamp (8 W). Membranes were then subjected to centrifugation at 4 8C for 15 min at 16000 g. The supernatant was discarded, and the membranes were washed three times with HEPES buffer including 30 min incubation and centrifugation after each wash. The pellet was resuspended with HEPES buffer, and the membranes were used for radioligand binding study. To displace the photoactivatable compounds from the binding pocket, 10 nm [3H] RAMX3 was used. Data analysis: Data were analyzed by nonlinear regression using the algorithms in PRISM 6.0 (GraphPad Software, San Diego, CA, USA). Data were fitted to a sigmoidal curve by nonlinear regression analysis in which the log IC50 value and the Hill coefficient were free parameters. The log IC50 values were transformed into log Ki values according to the equation for the calculation of competition curves as described by Cheng and Prusoff.[31] Chemistry General: Anhydrous solvents and all reagents were of commercial quality and were used without further purification. 4-[3-(Trifluoro- methyl)-3H-diazirin-3-yl]benzyl bromide was purchased from TCI chemicals. High-resolution mass spectrometry (HRMS) experiments were run on JEOLJMS-GC Mate II (EI–MS) and Finnigan Thermo Quest TSQ 7000 (ESI–MS) spectrometers. NMR spectra were ob- tained on Bruker Avance 360 or Bruker Avance 600 spectrometers. Chemical shifts (d) are reported in parts per million (ppm) relative to tetramethylsilane (TMS) as an internal reference in the indicated solvent, and coupling constants (J) are reported in Hertz (Hz). Purification by flash chromatography was performed using silica gel 60; TLC analyses were performed using Merck 60 F254 alumi- num sheets visualized by UV light (254 nm) and KMnO4. Analytical HPLC was performed on an Agilent 1100 HPLC system employing a VWL detector and a Zorbax Eclipse XDB-C18 column (4.6 mmx 150 mm, 5 mm) using the following binary solvent system: CH3OH in 0.1 % aq TFA (trifluoroacetic acid); 10–100 % CH3OH over 15 min, 100 % for 3 min, flow rate of 1.0 mLmin—1, l= 254 nm. The purities of all test compounds were determined by HPLC to be > 95 %.
tert-Butyl-4-[({1-[3-(4-ethoxyphenyl)-4-oxo-3,4-dihydropyri- do[2,3-d]pyrimidin-2-yl]ethyl}amino)methyl]piperidine-1-carbox- ylate (2): A solution of 1 (0.30 g, 0.97 mmol) in 1,2-dichloroethane (16 mL), tert-butyl-4-formylpiperidine-1-carboxylate (0.22 g, 0.94 mmol) was added and stirred for 1h at RT followed by addi- tion of sodium triacetoxyborohydride (0.422 g, 1.934 mmol).
The resulting slurry was stirred for 4h at RT. The reaction mixture was diluted with CH2Cl2 (20 mL) and washed with saturated aq NaHCO3 (3 x 20 mL). The organic layer was dried over anhydrous Na2SO4, fil- tered and concentrated in vacuo. The obtained yellowish crude was purified by silica gel chromatography (3 % MeOH in CH2Cl2) to give the title compound as a white foamy product (0.43 g, 90 %). AMG 487