gem-Difluorobisarylic derivatives: design, synthesis and anti-inflammatory effect

Introduction New fluorinated diaryl ethers and bisarylic ketones were designed and evaluated for their anti-inflammatory effects in primary macrophages. Methods The synthesis of the designed molecules started from easily accessible and versatile gem-difluoro propargylic derivatives. The desired aromatic systems were obtained using Diels–Alder/aromatization sequences and this was followed by Pd-catalyzed coupling reactions and, when required, final functionalization steps. Both direct inhibitory effects on cyclooxygenase-1 or -2 activities, protein expression of cyclooxygenase-2 and nitric oxide synthase-II and the production of prostaglandin E2, the pro-inflammatory nitric oxide and interleukin-6 were evaluated in primary murine bone marrow-derived macrophages in response to lipopolysaccharide. Docking of the designed molecules in cyclooxygenase-1 or -2 was performed. Results Only fluorinated compounds exerted anti-inflammatory activities by lowering the secretion of interleukin-6, nitric oxide, and prostaglandin E2, and decreasing the protein expression of inducible nitric oxide synthase and cyclooxygenase-2 in mouse primary macrophages exposed to lipopolysaccharide, as well as cyclooxygenase activity for some inhibitors with different efficiencies depending on the R-groups. Docking observation suggested an inhibitory role of cyclooxygenase-1 or -2 for compounds A3, A4 and A5 in addition to their capacity to inhibit nitrite, interleukin-6, and nitric oxide synthase-II and cyclooxygenase-2 expression. Conclusion The new fluorinated diaryl ethers and bisarylic ketones have anti-inflammatory effects in macrophages. These fluorinated compounds have improved potential anti-inflammatory properties due to the fluorine residues in the bioactive molecules.


Introduction
Diaryl ethers are key scaffolds present in many natural or synthetic organic molecules, which are often used in medicinal chemistry [1]. Fenoprofen for instance is one of the synthetic diarylethers [2] with nonsteroidal anti-inflammatory, analgesic and antirheumatic effects [3]. More precisely, it is a derivative of 2-aryl propanoic acids, which is an important class of nonsteroidal antiinflammatory drugs including flurbiprofen, ibuprofen, naproxen and fenoprofen.
Moreover, benzophenone analogues, such as ketoprofen, recently have been reported also as potent antiinflammatory agents by inhibiting prostaglandin (PG) production [4,5]. It has been shown that benzoylphenyl acetic acid for instance has anti-inflammatory activity by decreasing the volume of paw edema in treated rats [6].
On the other hand, the introduction of fluorine into organic molecules may cause profound pharmacological Ayoub et al. BMC Chemistry (2019) 13:124 effects by improving the activity and selectivity of the bioactive molecules [7]. The utility of fluorine in the design of drugs results mainly from its ability to modify some functional activities, such as increasing lipophilicity [8] and extending its bioavailability [9]. Moreover, carbon forms stronger bond with fluorine (CF) n , with a higher oxidative and thermal stability than a carbon-hydrogen bond [10]. The CF 2 unit for instance is generally considered as a bioisostere of the oxygen atom or of a carbonyl group [11].
We therefore synthesized new gem-difluorobisarylic derivatives and evaluated their anti-inflammatory effects. We first investigated their effects on PGE 2 production in mouse primary macrophages in response to lipopolysaccharide (LPS) and their anti-cyclooxygenase (COX)-1 and -2 activities. We next studied their effects on the production of the pro-inflammatory nitric oxide (NO) and interleukin (IL)-6 and the expression of NO synthase-II (NOS-II) and COX-2.

Synthesis of bisarylic derivatives
Based on our previous study [12], five bisarylic compounds A1 to A5 were designed as indicated in Scheme 1.
In our strategy, the gem-difluoro unit has been chosen as a mimic of either the ether oxygen (fenoprofen series) or of a carbonyl group (ketoprofen series). First, two phenylpropionic acid derivatives A1 (as a non-fluorinated reference) and the corresponding gem-difluoro derivative A2 were proposed as analogues of fenoprofen and ketoprofen. The comparison of the inhibitory activities of compounds A1 and A2 would allow establishing the impact of fluorine atom on the efficiency of these compounds. On the other hand, three other derivatives A3, A4, and A5 were designed as simplified benzoic acidtype derivatives, with three different substituents in meta position on the second aromatic ring (Scheme 1).

Synthetic procedures
All these molecules were synthesized from bromo intermediates B (Scheme 2, Table 1) and were tested for their anti-inflammatory activity.

Synthesis of key intermediates 7 and 10
Addition of the lithium salt of compound 2 to 3-bromobenzaldehyde 3 at low temperature (− 80 °C) gave propargyl alcohol 4 in 70% yield. After oxidation with Jones reagent, propargylic ketone 5 was isolated in 80% yield. Then, Diels-Alder reaction and DDQ aromatization provided the intermediate 7 with a good yield for both steps (Scheme 3).
After treatment of ketone 5 by DAST, compound 8 was obtained in 71% yield. Similarly, Diels-Alder reaction and DDQ aromatization proceeded well by giving the fluorinated intermediate 10 with excellent yields (Scheme 4).

Preparation of compounds A1 and A2
Starting from the key scaffolds 7 and 10, Suzuki-Miyaura couplings, with boronic acid, afforded biphenyl type compounds 11 and 12 in 92 and 94% yield, respectively.
Then, hydroboration to 13 and 14, followed by oxidation with Jones reagent led to the desired analogues A1 and A2 in good yields (Scheme 5).

Preparation of compounds A3, A4, and A5
Starting from intermediates 10 and 12, reduction with LiBEt 3 H furnished alcohols 15 and 16 respectively, then oxidation by Jones reagent gave the desired acid A3. However, in the case of 16, an unexpected cleavage of the double bond occurred, affording acid A4. Using the same gem-difluoro intermediate 12, catalytic hydrogenation to 17, followed by reduction and oxidation afforded the desired derivative A5 in good yields (Scheme 6).

Biological activities
We investigated the effects of these derivatives on inflammation in bone marrow-derived macrophages (BMDM) by first evaluating their capacity to decrease LPS-dependent increase of PGE 2 secretion and COX-2 expression. Compound A1 (non-fluorinated) and compound A2 (fluorinated) effects were compared to evaluate the importance of the fluorine atom. Only compound A2 inhibited significantly in a dose-dependent manner the Scheme 2 Retrosynthetic analysis for the preparation of compounds A secretion of PGE 2 (Fig. 1a) (IC 50 = 16.5 ± 8.9 µM) with no effect on COX-2 expression (Fig. 1c) supporting the importance of fluorine in inhibiting PGE 2 production.
In parallel, we compared the inhibitory effects of compounds A3, A4 and A5, which are all fluorinated but present differences in R1 group (  the bromine intermediate obtained in the synthetic reaction of compound A5. Compound A4 is a ketone intermediate obtained unexpectedly with good yield during the synthesis of compound A5. Compounds A3, A4 and A5 have the carboxyl group attached to the benzene ring in the ortho position relative to CF 2 group. Figure 1b showed a dose response effect of these derivatives on PGE 2 secretion, in which compounds A4 and A5 significantly decreased PGE 2 secretion at 25 and 50 µM with IC 50 of 28.1 ± 22.8 and 22.4 ± 21.5 µM, respectively. Compound A3 did not show a strong inhibition at similar concentrations. Under these conditions, only compounds A4 and A5 significantly downregulated COX-2 expression (Fig. 1d). Further analysis showed a dose-dependent inhibitory effect on COX-2 expression for compounds A4 and A5 ( Fig. 1e and f, respectively). Thus, the nature of R groups in compounds A4 and A5 is important for their inhibitory effect on COX-2 expression and consequently PGE 2 production. We next addressed the question whether COX activity was inhibited. We performed COX-1 activity using Human Embryonic kidney (HEK)-293 cells stably overexpressing COX-1. Cells were treated with all compounds at 10 and 50 µM and PGE 2 was measured after the addition of arachidonic acid (AA). The results showed that compound A5 had the maximal inhibitory effect on COX-1 activity with more than 80% inhibition at 50 µM ( Fig. 2a) with an IC 50 of 5.2 µM. COX-2 activity was also assessed on BMDM treated for 30 min with aspirin to inhibit basal COX activity prior to the addition of 10 ng/mL LPS for 24 h which induces COX-2. These cells were then treated with 10 and 50 µM of derivatives and further incubated with AA. PGE 2 production revealed that compound A5 inhibited strongly COX-2 activity with an IC 50 of 13.3 µM, whereas moderate effect was observed for compounds A2, A3 and A4 (Fig. 2b). Indeed, the assay used for COX-2 activity cannot exclude an effect on mPGES-1.
In parallel, we assessed the effect of these compounds on the production of IL-6 and NO measured by its breakdown product nitrite. Figure 3 showed a dose response inhibition of compounds A2 (Fig. 3a, c), A4 and A5 (Fig. 3b, d) for both IL-6 and NO secretion. IC 50 are presented in Table 2 and were significant for compounds A4 and A5.
NOS-II is the inducible form of nitric oxide synthase, and is responsible for the production of the measured NO. For this, NOS-II protein expression was analyzed in LPS-stimulated BMDM, treated with 50 µM of bisarylic derivatives for 24 h. Results revealed that the fluorinated compounds A2, A4 and A5 inhibited NOS-II expression in parallel to NO production ( Fig. 3e, f ).

Molecular docking
We finally carried out model analysis of the inhibitors with ovine COX-1 [13] and murine COX-2 [14], to examine how these compounds dock with the active sites of the enzymes and to determine the amino acids involved in the interaction with the compounds. Ibuprofen docked into the hydrophobic cavity of COX-2 formed by Arg121, Tyr356, Ser354, Leu353, Val350 and Tyr349, where the carboxyl group of ibuprofen interacts with Arg121 and Tyr356 by a salt bridge and a hydrogen bond. The compounds A3, A4 and A5 were docked near Arg121, similarly to ibuprofen. Compounds A3 and A4 showed interaction with Tyr356 (Fig. 4). The binding scores of compounds A3, A4 and A5 (− 7.7 kcal/mol, − 7.7 kcal/ mol and − 7.5 kcal/mol respectively) are comparable to ibuprofen (Table 3). Furthermore, the difluoromethyl group present in these compounds, which introduces a strong electrostatic field in this hydrophobic pocket, would be in favor the interaction with Arg121. Compounds A1 and A2, even though they occupy the same active pocket, interact with Arg121 through the carboxylate group with less binding energy, have a bulky side chains that negatively would affect the stability of these molecules in the hydrophobic pocket. For COX-1, ibuprofen docked into the hydrophobic pocket composed of the amino acids Arg120, Tyr355, Ser353, Leu352, Val349, Tyr348, Val116, Leu531, Ser530, Ala527, Gly526 and Ile523. All compounds docked in the same active hydrophobic pocket of COX-1. Only Arg120 interacts with the carboxylate group by a salt bridge (Fig. 5).
Similarly, to ibuprofen, compounds A2, A3, A4 and A5 showed a moderate binding energy compared to ibuprofen (− 7 kcal/mol and − 7.8 kcal/mol respectively, Table 3) whereas compound A1 showed the lowest binding energy, which is compatible with biological activities. More analyses are required to fully understand the key role of the fluorine atoms on the biological activity of these molecules. However, to explain these results, it is  . 4 Two-dimensional pose of compounds A1 to A5 and Ibuprofen inside the binding pocket of mouse COX-2 as crystallized by [13]. Ligand-receptor interactions as highlighted by Maestro (Shrodinger, LLC). Ligands are represented in stick, and amino acids within the binding pocket are labeled. An arrow represents the H-bonds between an amino acid and ligand groups. A line shows a potential salt bridge between two charged groups possible that the bulky and lipophilic CF 2 group could fit better in the pocket of these proteins than the carbonyl of ketoprofen or the oxygen atom of fenoprofen. Further, in the case of compounds A3, A4 and A5 it can also increase the acidity of the CO 2 H in ortho position.

Conclusion
In conclusion, five bisarylic derivatives were prepared and tested in comparison with fenoprofen. This type of compounds is endowed with certain anti-inflammatory activities in mouse primary macrophages with a significant difference between the fluorinated analogues and the non-fluorinated one, showing the importance of the CF 2 group. All fluorinated derivatives blocked PGE 2 , nitrite and IL-6 production in activated macrophages.
Derivatives A4 and A5 showed additional strong inhibition of COX-2 and NOS-II expression. In addition, derivatives A3, A4 and A5 showed better anti-inflammatory activities than the other derivatives, with compound A5 having COX-1 and COX-2 direct inhibitory activities. Molecular docking of the compounds COX-1 and COX-2 are in support of the biological activity.

Chemistry experimental part
Reactions were carried out as described previously and monitored as described by 19 F NMR and by thin-layer chromatography (TLC) [15]. Yields refer to chromatographically and spectroscopically ( 1 H, 13 C, and 19 F NMR) homogeneous materials. Nuclear magnetic resonance (NMR) spectra have been recorded as previously described [15]. Mass spectral analyses have been performed at the Centre Régional de Mesures Physiques de l'Ouest (CRMPO) in Rennes (France).

Fig. 5
Two-Dimensional pose of compounds A1 to A5 and Ibuprofen inside the binding pocket of human COX-1 as crystallized by [14]. Ligand receptor interactions were evaluated for COX-1 as described in legend for Fig. 4 in anhydrous THF (20 mL). After stirring for additional 20 min at the same temperature, TMSCl (7 mL, 55.00 mmol, 2.5 equiv.) was added dropwise to the reaction mixture that was stirred for 1 h at − 80 °C and then left to rise at room temperature while continuous stirring for additional 2 h. The mixture was treated with concentrated solution of NH 4 Cl, extracted with ethyl acetate (3 times), dried over Na 2 SO 4 and concentrated by evaporating the solvent. Alcohol 4 was isolated over silica gel by column chromatography.

Synthesis of methyl 4-(3-bromophenyl)-4-oxobut-2-ynoate 5
To alcohol 4 (2.1 g, 7.46 mmol) in acetone (18 mL) was added dropwise under magnetic stirring at room temperature, a concentrated (5.4 M) solution of Jones reagent until disappearance of the starting material (TLC analysis). After addition of isopropanol (5 equiv.), the reaction mixture was filtered, and the filtrate was extracted with ethyl acetate. The combined organic phases were dried over Na 2 SO 4 , filtered and concentrated in vacuum. After purification by chromatography on silica gel, ketone 5 was obtained.
The reaction mixture was filtered on silica gel and the residues were washed with ethyl acetate. The organic phase was concentrated in vacuo and compounds 7 and 10 were isolated by chromatography on silica gel.
After purification by flash chromatography on silica gel, alcohols 15, 16 and 18 were isolated.

Synthesis of methyl 2-(difluoro(3-isopropylphenyl) methyl)-4,5-dimethylbenzoate 17
To a solution of 12 (498 mg, 1.51 mmol) in AcOEt (15 mL), was added 50 mg of palladium-charcoal catalyst (10%). The mixture was stirred at room temperature under hydrogen atmosphere. After 2 h, it was filtered and compound 17 was obtained, after purification on silica gel. The physicochemical properties and the spectral data of intermediates 4-18 are presented in the Tables 4 and 5, respectively and in Tables 6 and 7 for the synthesized bisarylic derivatives A1 to A5.

Synthesis of gem-difluorobisarylic derivatives A1, A2, A3, A4 and A5
To alcohol in acetone was added dropwise under magnetic stirring at room temperature, a concentrated (5.4 M) solution of Jones reagent until disappearance of the starting material (TLC analysis). After addition of isopropanol (5 equiv.), the reaction mixture was filtered, and the filtrate was extracted with ethyl acetate. The combined organic phases were dried over Na 2 SO 4 , filtered and concentrated in vacuum. After purification by chromatography on silica gel, carboxylic acid was obtained.

Supporting information
Experimental details and characterization data of new compounds with copies of 1 H, 13 C and 19 F NMR spectra are presented in the supplementary section, Additional file 1.

Evaluation of inflammation in macrophages
C57BL/6J male mice (20-25 g, 8 week-old) were obtained from Charles River (Ecully, France) and the animal facility of the American University of Beirut. Mice were housed 5 per cage in temperature-and humidity-controlled rooms, kept on a 12-h light-dark cycle, and provided with standard food and water ad lib and with enrichment Table 4 The physicochemical properties of intermediates 4-18   On the day of the procedure, 2-3 mice were euthanatized after 3 min exposure to carbon dioxide. BMDM were isolated as previously described and were plated at 0.8 million cells per well [16]. Flow cytometry analysis was performed using F4/80 -APC antibody (BioLegend 123115) and showed 90% macrophages. BMDM were then treated for 24 h with different concentrations of the bisarylic derivative compounds for 30 min prior to the addition of 10 ng/mL LPS. DMSO concentration did not exceed 0.4% with no effect. The supernatants were assessed for IL-6, PGE 2 and nitrite, the stable derivative of NO. Cells were washed with PBS and lysed in RIPA buffer containing inhibitors of protease. Total protein concentration was determined using DC protein assay (Bio-Rad 500-0115) with BSA as standard. IL-6, nitrite and PGE 2 were measured as described previously [17]. Western blot of NOS-II and COX-2 was performed as previously described [17][18][19]. 10 µg of total protein was assessed. The primary antibodies were developed and characterized as previously described: for COX-2, mouse monoclonal antibody anti-COX-2 (clone COX-214, 1/5000) [20]; for NOS-II, rabbit polyclonal antibody anti-NOS-II (dilution1/2000) [21], and mouse β-actin (dilution 1/10,000) (Sigma-Aldrich A5441). Clarity ™ western ECL substrate (Bio-Rad 170-5061) was used according to the manufacturer's instructions to reveal positive bands visualized using Bio-Rad ChemiDoc.

COX-1 and COX-2 activities
For COX-1 activity, human embryonic kidney (HEK)-293 cells (ATCC CRL-1573, Manassas, VA USA) stably overexpressing human recombinant COX-1 were used [17]. Cells were treated with compounds A1 to A5 for 45 min in Hanks buffer and then 10 µM arachidonic acid (AA) were added for 30 min. PGE 2 was measured and corresponded to the breakdown metabolism of PGH 2 and PGE 2 .
For COX-2 activity, BMDM were treated with 10 μM acetylsalicylic acid (ASA) for 30 min to irreversibly inhibit COX-1 and then washed and treated with LPS 10 ng/mL for 24 h. BMDM were treated with compounds A1 to A5 for 45 min in Hanks buffer, pH 7.4 containing 1 mg/mL BSA prior to the addition of 10 µM of AA for 30 min. Supernatants were collected and PGE 2 was determined.

Toxicity assay
WST-1 assay was used to determine the toxicity of the synthesized compounds (Cell proliferation WST-1 assay, Sigma-Aldrich 5015944001). Briefly, macrophages (50,000 cells per well) were plated in a 96 well plate in RPMI culture medium containing 10% FBS and grown for 24 h. Cells (in triplicates) were treated at 25 and 50 μM of the tested compounds. Culture medium without cells and cells without treatment were used as control. Results were expressed as percentage of cells without treatment. All compounds showed 95% viability at 50 μM.

Molecular docking Target and small molecule preparation
All small molecules (A1 to A5) were built using Openbabel chemical toolbox (PMID: 21982300) and subsequent low energy 3D conformations were generated using Frog2 (PMID: 20444874). Protonation state corresponded to a pH of 7. The 3D structure of the ovine COX-1 (PDBID: