Potential anti-neuroinflammatory NF-кB inhibitors based on 3,4-dihydronaphthalen-1(2H)-one derivatives

Abstract Nuclear factor kappa B (NF-кB) inhibition represents a new therapeutic strategy for the treatment of neuroinflammatory diseases. In this study, a series of 3,4-dihydronaphthalen-1(2H)-one (DHN; 6a-n, 7a-c) derivatives were synthesised and characterised by NMR and HRMS. We assessed the toxicity and anti-neuroinflammatory properties of these compounds and found that 6m showed the greatest anti-neuroinflammatory properties, with relatively low toxicity. Specifically, 6m significantly reduced reactive oxygen species production, down-regulated the expression of NOD-like receptor pyrin domain-containing protein 3 (NLRP3), apoptosis-associated speck-like protein containing a CARD (ASC), and caspase-1 and prevented lipopolysaccharide-stimulated BV2 microglia cells polarisation towards an M1 phenotype. Furthermore, 6m significantly decreased IκBα and NF-кB p65 phosphorylation, thus inhibiting the NF-кB signalling pathway. This suggests that 6m may be explored as a functional anti-neuroinflammatory agent for the treatment of inflammatory diseases in the central nervous system, such as multiple sclerosis, traumatic brain injury, stroke and spinal cord injury.


Introduction
Activated microglia-mediated inflammatory responses play an important role in the pathological development of inflammatory neurodegenerative diseases in the central nervous system (CNS) 1,2 . During inflammatory neurodegenerative diseases in CNS, the resident microglia become activated and polarised to a pro-inflammatory M1 phenotype 3 . It has been reported that pro-inflammatory cytokines (tumor necrosis factor (TNF-a), interleukin (IL)-6, IL-1b) secreted from M1 microglia increase blood-brain barrier (BBB) permeability by activating the nuclear factor kappa B (NF-rB) signalling pathway [4][5][6][7][8] . Concomitantly, the disrupted BBB promotes glial activation to exacerbate CNS inflammation 9 . In addition, activated microglia can produce reactive oxygen species (ROS), which may indirectly induce neuroinflammation by activating NF-jB 10 . Moreover, the production of ROS can activate the NOD-like receptor pyrin domain-containing protein 3 (NLRP3) inflammasome formation through NF-jB signalling pathway [11][12][13] , which induces the polarisation of M1 microglia and the secretion of pro-inflammatory cytokines to exacerbate neuroinflammation of CNS [14][15][16][17] . Therefore, developing an NF-rB inhibitor with anti-neuroinflammatory properties and low toxicity is a potential therapeutic strategy for treating inflammatory neurodegenerative CNS diseases 18 . 3,4-Dihydronaphthalen-1(2H)-one (DHN) derivatives with antitumor and anti-inflammatory activities have been investigated as novel modulators of allergic and inflammatory responses 19,20 and as potential inhibitors of retinoic acid (RA)-metabolizing enzymes for the treatment of skin conditions and cancer. Recently, several benzylidene-substituted DHN derivatives were reported as potential Bcl-2 inhibitors 21 . Additionally, 4-amino derivatives of DHN were developed as anti-inflammatory agents that stabilise mast cells 22 . Furthermore, dihydronaphthalene-1,4-dione derivatives inhibit the acetyltransferases ATase1 and ATase2 in neurons and glial cells, and therefore may serve as novel agents to prevent Alzheimer's disease and dementia 23 . However, DHN derivatives have rarely been developed as anti-neuroinflammatory drugs. In this study, a series of new benzylidene-substituted DHN derivatives were designed and synthesised through Claisen-Schmidt condensation reactions, then evaluated for the anti-neuroinflammatory activities.
Synthesis of 6a-n: general procedure 7-Methoxy-3,4-dihydronaphthalen-1(2H)-one (denoted as 5) was obtained according to the literature 24 . 5 (0.53 g, 3.0 mmol) and aromatic aldehyde (3.0 mmol) were dissolved in 10 ml of methanol. After 3.0 ml 20% NaOH solution were added, the mixture was stirred for 3-5 h at ambient temperature (monitored by TLC). The solvents were then removed by pouring, and the residues were purified on a silica gel by column chromatography using petroleum ether/EtOAc (2:1, v/v) as the eluent to produce light yellow powders 6a-n.

Single-Crystal structure determination of 6m
The 6m single crystals were prepared by slow evaporation of methanol solution under ambient conditions. Suitable single crystal measured at 100 K on a SuperNova diffractometer (Dual, Cu at zero, Mo Ka radiation, k ¼ 0.71073 Å) using an AtlasS2 programmes. Crystal data of 6m: CCDC 1983271 (6m) contains the supplementary crystallographic data for this paper. Copies of the data can be obtained free of charge on application to CCDC, 12 Union Road, Cambridge CB2 1EZ, UK (fax: (þ44)1223-336-033; e-mail: deposit@ccdc.cam.ac.uk).
BV2 microglial cells (ATCC, Manassas, VA, USA) were cultured in 96-well plates then treated with DHN derivatives for 24 h. Next, 10 mL of the CCK-8 solution were added and incubated for 2 h at 37 C, then absorbance at 450 nm was measured using a microplate reader. The survival rate (%) of BV2 microglia treated with compounds (10 mM) was then calculated. Results are the average of three replicates and shown in Table 1.
Screening of anti-neuroinflammatory activities of DHN derivatives by enzyme-linked immunosorbent assay (ELISA) The anti-neuroinflammatory activities of DHN derivatives (6a-n, 7a-c) were screened by enzyme-linked immunosorbent assay (ELISA). Cultured BV2 microglia were pre-treated with or without lipopolysaccharide (LPS) (1.0 mg/mL) for 2 h, then were treated with pyrrolidinedithiocarbamate (PDTC) (30 mM) or DHN derivatives (6a-n, 7a-c) for another 24 h at 37 C. The supernatants were then collected for testing the secreted levels of TNF-a using an ELISA kit (R&D Systems, Madison, WI, USA) according to the manufacturer's instructions. Inhibition rate (%) was calculated as the average of three replicates, with PDTC used as a positive control, and expressed in Table 1. Assessing blood-brain barrier permeability to 6m The ADMET descriptors protocol in Discovery Studio 2017R2 software was used to calculate the BBB penetration of 6m 25,26 . The 3 D minimum energy structure of 6m was generated, and the ADMET model was used to obtain ADMET_PSA_2D, ADMET_AlogP98 values. The ellipses define regions where wellabsorbed compounds are expected to be found. The prediction level was estimated based on the ADMET model. The parameter of BBB function was scored on a scale of 0-3 (0: good penetration; 1: moderate; 2: poor; 3: very poor).

Synthesis and structural characterisation of DHN derivatives
In this study, the synthetic routes to 3,4-dihydronaphthalen-1(2H)one derivatives are shown in Schemes 1 and 2. A key intermediate, 7-methoxy-3,4-dihydronaphthalen-1(2H)-one (5), was prepared with an overall yield of 48% using the three steps based on the method described in the literature 21,24 . First, anisole (1) and succinic anhydride (2) were combined to generate 4-(4-methoxyphenyl)-4-oxobutanoic acid (3) by Lewis acid catalysis (anhydrous AlCl 3 ) with a 92% yield. Secondly, after Wolff-Kishner-Huang-Minlon reduction of 3, 4-phenoxybutanoic acid 4 was generated with an 86% yield. Then, the key intermediate 5 was prepared for cyclisation in the presence of PPA with a lower yield of 65%. Lastly, 5 and several aromatic aldehydes were subjected to Claisen-Schmidt condensation to yield a series of new 3,4-dihydronaphthalen-1(2H)-one derivatives (6a-n). The yields of 6a-n reached approximately 80-91%. During this last step, dry HCl, aqueous NaOH, or other bases can be chosen as the catalyst 27 . Considering its environmental friendliness and availability to laboratories, a 20% NaOH solution was selected as the catalyst (Scheme 1). The structures of 6a-n were characterised by NMR and MS (Supporting Information). Some selected spectral data are discussed below. In the 1 H NMR spectra of our target compounds, the chemical shifts in the range of 7.89-7.65 ppm appear as a singlet attributed to the proton of a,b-unsaturated ketone pharmacophores. All compounds showed two groups of characteristic triplets from the two intra-annular methylene groups at d 3.16-2.81 ppm. The singlets observed in the range of 5.21-5.09 ppm are likely the three protons of 7-methoxy in 3,4dihydronaphthalen-1(2H)-one. In the 13 C NMR spectra, the carbon atom of -C¼O groups appeared in chemical shifts of approximately 187 ppm. Additionally, HRMS spectroscopy data further confirmed the accuracy of the structures of these compounds. In order to investigate the influence of the length of the substituent of 3,4-dihydronaphthalen-1(2H)-one derivatives on neuroinflammation, several benzyloxy-substituted arylaldehydes were selected to synthesise novel 3,4-dihydronaphthalen-1(2H)-one derivatives (7ac). The yields of 7a-c reached approximately 76-79%. All analyses showed the same results as described with 6a-n, with the exception of additional singlets observed in the range of 5.21-5.09 ppm of the 1 H NMR spectra of 7a-c, corresponding to two methylene protons in the benzyloxy group. Corresponding NMR and HRMS data are shown in Supporting Information, which further confirm the accuracy of their structures.
Single crystals of 6m were prepared under ambient conditions, with crystallisation obtained via solvent evaporation in a methanol solution. Single-crystal structure analysis revealed that 6m crystallised in the triclinic space group P-1. The ORTEP diagram is presented in Figure 1. There is only a drug molecule in the asymmetric unit. Compared to the C(2)¼C(11) olefinic bonds, 4methoxy-3-(trifluoromethyl)phenyl and carbonyl groups adopt the E stereochemistry 28,29 . Because of the distorting effect of 3,4-dihydrobenzo[b]oxepin-5(2H)-one, the 7-methoxyphenyl and 4methoxy-3-(trifluoromethyl)phenyl groups are not coplanar with each other, with a dihedral angle of approximately 60.6(2) . This twisted configuration may increase likelihood of interactions with bioactive molecules, for the purposes of creating more potent biological activity 30 .

In vitro toxicity and anti-neuroinflammatory activity analysis
The toxicity against BV2 microglia for all synthesised compounds was tested by the CCK-8 assay. As shown in Table 1, the survival rates of BV2 microglia were above 85% for all synthesised compounds at a concentration of 10 mM. In particular, for 6a, 6c, 6m, and 7a, the survival rates of BV2 microglia can be greater than 95%. These results illustrate that all compounds have no obvious toxic effects on BV2 microglia in vitro. Furthermore, 6m dosedependently decreased the survival rate of BV2 microglia; however, 6m (20 mM) decreased the survival rate to approximately 82% (Supporting Information Figure S1).
TNF-a is a commonly secreted pro-inflammatory cytokine from activated microglia 31,32 . To test the anti-inflammatory activities of our synthesised compounds, TNF-a secretion from LPS-pre-treated BV2 microglia with or without the synthesised compounds was detected by ELISA. PDTC (30 mM) was used as a positive control. As shown in Table 1, following PDTC treatment, TNF-a secretion was inhibited by approximately 58.1% from LPS-pre-treated BV2 microglia, compared to cells treated with LPS alone. When treated with 10 mM of the synthesised compounds, the TNF-a secretion inhibition rates showed variation between each compound. Intriguingly, among these compounds, 6a, 6f, 6 h, 6i, 6m, 6n, and 7 b exhibited significantly higher anti-inflammatory activity than PDTC by inhibiting TNF-a secretion by 68.4%, 69.9%, 67.2%, 72.6%, 80.5%, 65.8%, and 59.5%, respectively.
When analysing the structure-activity relationship, 3,4-dihydronaphthalen-1(2H)-one group is the constant, primary active centre, while the different substitutions on the benzylidene should influence the variation in anti-inflammatory activity. For instance, 6a-c have different numbers of electron-donating groups (-OMe), whereas trimethoxy substitutions result in a loss of anti-inflammatory activity. The anti-inflammatory activities followed the order: 6a > 6b > 6c. Interestingly, a single para-substitution on the benzylidene was sufficient for this activity, such as 6a (para-OMe), 6f (para-F), and 6i (para-CF 3 ). Among the compounds with the same substitution (-F or -CF 3 ), the anti-inflammatory activities followed the order para-substitution (6f and 6i) > meta-substitution (6e and 6 h) > ortho-substitution (6d and 6 g), indicating that the position of substitutions affects the anti-inflammatory activity.
In order to examine the substituent effect of these synthesised compounds, double-substituted 6j-6n, which had substitutions by strong electron-donating -OMe and electron-withdrawing -F and -CF 3 groups, were synthesised then evaluated for anti-inflammatory activity. The results revealed that 6j-6l with ortho-substitutions did not show remarkable anti-inflammatory activity. In contrast, meta-and para-substituted 6m and double-meta-substituted 6n displayed more significant anti-inflammatory activities, with 6m specifically exhibiting the strongest anti-inflammatory activity, revealing that meta-CF 3 and para-OMe substitutions could improve anti-inflammatory activity.
To investigate the influence of the length of the terminal substituent of 3,4-dihydronaphthalen-1(2H)-one derivatives, original arylaldehydes were changed to benzyloxy-substituted arylaldehydes, generating 7a-c. However, these structural changes reduced the anti-inflammatory activity with inhibition rates ranging from 45-60% (Table 1).
In addition, fluorine and trifluoromethyl substitutions play an important role in bioactivity, as their lipophilicity can effectively increase the membrane permeability of drugs 33 . More importantly, the fluorine atom is a strong electronegative group, which can form multiple H-bonds with target proteins 34 . Additionally, stable C-F bonds can improve metabolic stability, and thus prolong the   duration of action in vivo 35 . In this study, we focussed on fluorine and trifluoromethyl substituted DHNs.
Notably, we found that the most effective compound had relatively low toxicity (6m). ADMET properties were estimated using Discovery Studio 2017R2 to calculate the BBB penetration of 6m in the CNS. The results showed the values of ADMET_PSA_2D and ADMET_AlogP98 were 35.16 and 5.172, respectively. The ellipsesdefined regions of 6m can be found in Supporting Information Figure S2. We confirmed that the ADMET_BBB_level value of 6m was at level 0 (good penetrant). This indicates that compound 6m can effectively penetrate the BBB in a suitable therapeutic concentration for further anti-neuroinflammatory studies. Based on the above evidence, 6m (10 mM) was selected for a more-in-depth biological evaluation of anti-neuroinflammatory activity.
6m Decreases ROS production ROS production due to oxidative stress was detected using the fluorescent probe 2,7-DCFH-DA 36 . During oxidative stress, 2,7-DCFH-DA can be oxidised by ROS to dichlorofluorescein (DCF), which can be detected fluorescently by FACS (Figure 2(A)). ROS production was quantified by counting the number of DCF-containing cells. As shown in Figure 2(B), LPS significantly induced ROS production in BV2 microglia, but 6m dose-dependently and significantly inhibited this ROS production. These results suggest that 6m can inhibit oxidative stress in LPS-stimulated BV2 microglia.

6m Reduces expression of inflammasome-related proteins
The activation of NLRP3 leads to the formation of the NLRP3 inflammasome by combining with ASC and caspase-1 37 . This leads to the secretion of pro-inflammatory cytokines (such as TNF-a, IL-1b, IL-18, and IL-6) from activated microglia by activating the NF-jB signalling pathway, thus increasing neuroinflammation [16][17][18][19] , which contributes to neurological disease development and progression in the CNS 38,39 . Moreover, NLRP3 inflammasome inhibition shifts the M1 phenotype to an M2 phenotype by inhibiting the NF-jB signalling pathway 15,40 . As shown in Figure 3(A,B), LPS significantly increased the expression of NLRP3 in BV2 microglia, but this was dose-dependently and significantly reduced by 6m. As a component of the NLRP3 inflammasome, ASC expression showed a similar expression pattern to NLRP3 in LPS-stimulated BV2 microglia cells with or without 6m treatment (Figure 3(A,C)). Caspase-1, another component of the NLRP3 inflammasome, also showed a similar expression pattern to NLRP3 in LPS-stimulated BV2 microglia with or without 6m treatment (Figure 3(A,D)). These results suggest that 6m is a competitive inhibitor for NLRP3 inflammasome activation.
6m Modulates pro-and anti-inflammatory cytokine secretion M1 microglia play a critical role by secreting pro-inflammatory cytokines (TNF-a, IL-1b, IL-18, IL-6) to exacerbate neuroinflammation in the CNS 5 . Conversely, M2 microglia, the other of the two well-established microglia phenotypes, have an anti-inflammatory function by secreting IL-10 15,40 . As shown in Figure 4, LPS significantly increased the secretion of pro-inflammatory cytokines (IL-1b, IL-18, and IL-6), but decreased the secretion of the anti-inflammatory cytokine (IL-10) from BV2 microglia. However, 6m significantly down-regulated the expression of these pro-inflammatory cytokines (IL-1b, IL-18, and IL-6), and increased the secretion of the anti-inflammatory cytokine (IL-10) from these cells. These results indicate that 6m balances the secretion of pro-and anti-inflammatory cytokines to play an anti-inflammatory function.
6m Decreases LPS-stimulated M1 polarisation in BV2 microglia LPS induces M1 polarisation, which can be detected through the expression of M1 phenotype markers (iNOS, CD68) 14 . As shown in Figure 5(A,C), LPS significantly induced the BV2 microglia to polarise to an M1 phenotype indicated by the numbers of iNOS-and CD86-expressing BV2 microglia. However, 6m significantly inhibited this LPS-induced iNOS and CD86 expression. Even though 6m did not significantly up-regulate the number of CD206-and Arg-1expressing cells, the ratio of double-labeled CD206-and Arg-1cells to double-labeled iNOS-and CD86-expressing cells is significantly increased by 6m in LPS-stimulated BV2 microglia ( Figure  5(B,D,E)). Taken together, 6m decreases M1 polarisation of LPSstimulated BV2 microglia to alleviate neuroinflammation.

6m Inhibits NF-rB activation
To further confirm the effects of 6m on inhibiting the NF-rB signalling pathway in LPS-stimulated BV2 microglia, we investigated the expression of IjBa, p-IjBa, NF-jB p65 and p-NF-jB p65 by Western blotting (Figure 6(A)). As shown in Figure 6(B,C,D), LPS significantly induced the phosphorylation of NF-jB p65 and its upstream regulator (IjBa) in BV2 microglia, in accordance with a previous study showing that LPS induces inflammatory cytokine secretion from BV2 microglia via NF-jB signalling pathways 6 . However, 6m significantly and dose-dependently decreased the levels of p-IjBa and p-NF-jB p65 in LPS-stimulated BV2 microglia. Collectively, these results reveal that 6m significantly inhibits the NF-rB signalling pathway following LPS stimulation in BV2 microglia.

Conclusions
In this study, a series of 3,4-dihydronaphthalen-1(2H)-one (DHN; 6a-n, 7a-c) derivatives were synthesised and characterised, then evaluated for their toxicity and anti-neuroinflammatory activity. The meta-CF 3 and para-OMe substituted compound 6m showed the greatest potential bioactivity with lower toxicity and the strongest anti-neuroinflammatory actions. Furthermore, 6m significantly reduced ROS production, NLRP3 inflammasome expression, inflammatory cytokine secretion, and the number of M1 polarised cells following LPS stimulation in BV2 microglia. It also significantly inhibited the activation of the NF-rB signalling pathway. This study reveals that 6m may be developed as a functional anti-neuroinflammatory agent for the treatment of inflammatory neurodegenerative disorders of the CNS, such as multiple sclerosis, Huntington's disease, stroke, and traumatic brain injury.