Cell culture reagents used were Penicillin-Streptomycin, RPMI 1640 obtained from Cambrex (Belgium) and fetal bovine serum from Life Technologies Ltd (UK). The TLR ligands used were chloroform extracted Escherichia coli (E.coli) lipopolysaccharide (LPS) and resiquimod (R-848) from Invivogen (USA). Flagellin (purified) and Pam₃cys-ser(lys)₄.3HCl (Pam3) were from Alexis (UK). Mianserin hydrochloride was purchased from Sequoia Research Products (Pangbourne, UK). The structural derivatives of mianserin were synthesized by Oxygen Healthcare Research Pvt. Ltd. (Ahmedabad, India), the synthesis steps are provided in the Supplemental Information. Macrophage colony stimulating factor (M-CSF) was purchased from eBioscience (USA). Dimethyl sulfoxide (DMSO) and Hank's balanced salt solution (HBSS) without calium chloride and magnesium sulfate were purchased from (Sigma, UK). Percoll Plus was purchased from GE Healthcare (Bucks, UK), PBS citrate from Paris Anticorps (France) and lympholyte-H from CedarLane (Ontario, Canada).

Mianserin is a tetracyclic antidepressant whose primary activity is as an antagonist/inverse agonist of a range of aminergic GPCRs. Literature reports include 5-HT, α-adrenergic and histamine H1 receptor activity, but mianserin is also reported to inhibit noradrenaline uptake (Table 1). To explore the requirement of 5-HT receptor binding for the immunomodulatory activity of mianserin, compounds were designed to bias activity in favor of mianserin's anti-inflammatory effects. In our previous study, mianserin was observed to inhibit endosomal TLR induced cytokine production (10). As TLRs are not known to preferentially bind simple amines, the structure was modified in ways that were predicted to be unlikely to bind to the monoamine GPCRs. Although a wide range of mianserin derivatives have previously been described, all retain the N-methyl group, which is one of the primary pharmacophore elements for aminergic receptor binding (16). It has long been hypothesized that central nervous system drugs share common binding features, principally aromatic groups and a basic amine. In the case of MN-1 the secondary amine was acetylated. For the other compounds, the methyl group was removed, the basic nitrogen retained and a polar group attached via an alkyl linker (MN-2, 3, 4, 5, and 6). The intention was to retain inhibition of TLR-induced cytokines but greatly reduce activity at 5-HT receptors (Figure 1).

Leukocyte cones from blood donors, were purchased from NHS Blood and Transplant (Tooting, UK). Written informed consent was obtained from the donors by NHS Blood and Transplant and the study was approved by Brighton and Sussex Medical School Research Governance and Ethics Committee. Peripheral blood mononuclear cells (PBMCs) were isolated from the cones by Lympholyte-H centrifugation. Briefly, cells were diluted to 40 mls with HBSS, then layered onto two tubes each containing 20 mls of Lympholyte-H and centrifuged for 25 min at 900 × g. The PBMC layer was then collected and washed twice in HBSS by centrifugation at 350 × g for 10 min. PBMCs were cultured at 37°C in a cell culture incubator with 5% CO₂ and 21% O₂. Cells were plated at 2 × 10⁵ cells/well in 96-well tissue culture plates (Corning, Amsterdam, The Netherlands) in Roswell Park Memorial Institute (RPMI) 1640 media containing 5% (v/v) FBS and 100 U/ml penicillin/streptomycin or used for the isolation of monocytes. Peripheral blood monocytes were obtained by density centrifugation of PBMCs on an iso-osmotic Percoll gradient. Iso-osmotoic percoll was achieved by adding 1 vol 1.5 M NaCl to 9 vol of Percoll Plus. The gradient was then prepared by mixing 1:1 v/v isosmotic percoll with 1x phosphate buffered saline (PBS) citrate. Briefly, cells were diluted to 20 mls with HBSS, then layered onto a tube containing 20 mls of iso-osmotic Percoll and centrifuged for 15 min at 900 × g. The monocyte layer was then collected and washed twice in HBSS by centrifugation at 350 × g for 10 min.

Immediately after isolation, monocytes were cultured at 37°C in a cell culture incubator with 5% CO₂ and 21% O₂. Cells were plated at 2 × 10⁵ cells/well in 96-well tissue culture plate in RPMI 1640 containing 5% (v/v) FBS and 100 U/ml penicillin/streptomycin. Macrophages were derived from monocytes after differentiation for 4 days in RPMI containing 5% (v/v) FCS, 100 U/ml penicillin-streptomycin) and 100 ng/ml M-CSF. Macrophages were then pre-incubated with media alone, MN-1 or DMSO (vehicle control) for 30 min, then stimulated with 10 ng/ml LPS, 100 ng/ml PAM3cys, 100 ng/ml Flagellin or 1 μg/ml R-848 for 6 h after which the supernatants were collected.

RA synovial membrane cells were isolated as previously described (17, 18), from joint synovium removed from donors during elective joint replacement surgery. All patients gave written informed consent and the study was approved by the South East Coast—Brighton and Sussex research ethics committee (10/H1107/27). Immediately after isolation cells were cultured at 4 × 10⁴ cells per well in a 384 well tissue culture plate (Corning, Amsterdam, The Netherlands) in RPMI containing 5% (v/v) FBS and 100 units/ml Penicillin/streptomycin. Cells were then incubated in the media alone or media containing a range of concentrations of MN-1 for 24 h. Cell viability was assessed after all experiments by the 3-[4,5 dimethylthiazol-2-yl]-2,5-diphenyl-tetrazolium bromide (MTT) assay (Sigma) (19). Briefly, cells were incubated with 100 μl of 0.5 mg/ml MTT in complete cell culture medium for 4 h at 37°C in a cell culture incubator with 5% CO₂ and 21% O₂ to allow formazan crystal formation in metabolically active cells. To dissolve the crystals, 100 μl of MTT lysis buffer (10% SDS and 0.01 M HCl) was then added for 3 h or until all crystals had dissolved. Absorbance was read at 570 nM on a Biotek synergy HT spectrophotometric ELISA plate reader and analyzed using the associated Gen5 software (Biotek, Bedfordshire, UK).

Sandwich ELISAs were employed to measure tumor necrosis factor (TNF) (BD Pharmingen, UK) and interleukin (IL)-1β (R&D systems, Abingdon, UK); TNF capture antibody (551220), TNF biotinylated antibody (554511), IL-1β capture antibody (MAB601), and biotinylated IL-1β antibody (BAF201) were used according to the manufacturer's instructions. Recombinant TNF and IL-1β were purchased from PeproTech (London, UK). Streptavidin HRP (DY998) was used at 1:400 dilution and 3,3′,5,5′-Tetramethylbenzidine microwell peroxidase substrate kit (KPL Inc., USA) was used according to the manufacturer's instructions. All incubation steps were for 1 h at room temperature, apart from the incubation of the standards and samples which was overnight at 4°C. Color formation was stopped using 0.16 M sulfuric acid. Absorbance was read on a Biotek synergy HT spectrophotometric ELISA plate reader and analyzed using the associated Gen5 software.

The affinity of MN and MN-1 toward 5-HT receptors was determined by Eurofins Scientific France (Poitiers, France) using radioligand binding techniques similar to those described by Peroutka et al. (20). A non-selective serotonin receptor binding assay was chosen due to mianserin having activity at several of the 5-HT receptors (Table 1). Briefly, membrane homogenates of cerebral cortex (170 μg protein) were incubated for 60 min at 37°C with 2.5 nM [³H]serotonin in the absence or presence of 10 μM mianserin or MN-1 in a buffer containing 50 mM Tris-HCl (pH 7.4), 4 mM CaCl₂, 10 μM pargyline and 1 g/l ascorbic acid. Nonspecific binding is determined in the presence of 10 μM serotonin.

Following incubation, the samples were filtered rapidly under vacuum through glass fiber filters (GF/B, Packard, Perkin Elmer, UK) pre-soaked with 0.1% bovine serum albumin and rinsed several times with ice-cold 50 mM Tris-HCl using a 96-sample cell harvester (Unifilter, Packard, Perkin Elmer, UK). The filters are dried then counted for radioactivity in a scintillation counter (Topcount, Packard, Perkin Elmer, UK) using a scintillation cocktail (Microscint 0, Packard, Perkin Elmer, UK).

Mean, standard deviation (SD), standard error of the mean (SEM), and statistical significance were calculated using GraphPad version 3 (GraphPad Software Inc., USA). For statistical analysis of multiple means a one way ANOVA with a 95% confidence interval, followed by a Dunnett's multiple comparisons test was used. For the comparison of 2 means, a two tailed t-test of paired data was used with a 95% confidence interval. SEM was used for pooled experimental data whilst SD was used in graphs showing representative experiments. ^****p < 0.0001, ^**p < 0.01, and ^*p < 0.05.

As we had previously demonstrated an anti-inflammatory effect of mianserin on endosomal TLR signaling, a human PBMC assay was used to screen for activity of the structural derivatives against TLR8 activation. R-848 was used as a ligand for TLR8; although this is a dual ligand for TLR7, it is thought to act solely through TLR8 in human monocytes and macrophages as these cells do not produce TNF in response to TLR7 ligands (10). At a set concentration of 80 μM only mianserin, MN-1 and MN-3 were able to significantly inhibit the production of TNF upon stimulation with the TLR7/8 ligand R-848 (Figure 2A). However, MN-3 only produced a weak inhibition that was variable between donors. Cell viability was assessed using the MTT assay and found to be unaffected (Figure 2B). The concentration of 80 μM was chosen as a sub-optimal inhibitory concentration for the initial screening, to permit detection of compounds with improved inhibitory activity. This had been determined during a previous study where 100 μM mianserin produced an almost complete inhibition of TLR8 (10). As MN-1 produced the strongest inhibitory activity, it was chosen for further study. A dose response curve was generated from TLR8 activated human PBMCs using a quarter log titration to compare the EC₅₀ of MN-1 with mianserin. DMSO was used as a loading control at an equivalent amount to the 100 μM dose of MN-1. When the data was pooled from three independent experiments using blood from three separate donors, the calculated EC₅₀ for mianserin and MN-1 was 37.91 μM and 24.09 μM respectively (Figure 2C). Dose response curves were not generated for the other compounds, as this would have required using significantly elevated concentrations leading to cell toxicity.

To determine if MN-1 was still capable of binding 5-HT receptors at levels comparable to that of mianserin, the affinity of MN-1 for 5-HT receptors (non-selective) was evaluated in rat cerebral cortex in a radioligand binding assay. A non-selective serotonin receptor binding assay was used due to mianserin being reported to have activity at several 5-HT receptors (Table 1). The activity of mianserin at 5-HT receptors is also reported to be in the nM range, so a concentration of 10 μM mianserin and MN-1 was chosen for this assay to ensure these inhibitors would be in excess of what is required to block 5-HT binding (21). The rat cerebral cortex is a useful model for these experiments, as mianserin has a similar pKi on human and rat 5-HT receptors (Table 1). At 10 μM mianserin inhibited [³H] serotonin binding to 5-HT receptors with an average inhibition of 94.5%, whereas MN-1 showed a significantly decreased effect with an average inhibition of only 30% (Table 2).

In our previous study, mianserin selectively inhibited the endosomal TLRs but not those expressed at the cell surface in primary human cells (10). To assess if this selectivity for endosomal TLR inhibition was retained by MN-1, primary human M-CSF macrophages were stimulated with Pam3 (TLR1/2), LPS (TLR4), and flagellin (TLR5) to activate cell surface TLRs and R-848 to activate TLR8 localized in the endosome. TNF production induced by the cell surface TLRs was not inhibited by MN-1 but a dose dependent inhibition of TLR8 induced TNF production was observed (Figure 3A). This was significant when data was pooled from 3 separate donors (Figure 3B). Cell viability was assessed using an MTT assay and found to be unaffected (Figure 3C).

In addition to the ability to inhibit endosomal TLRs, we have previously published that mianserin could also inhibit spontaneous cytokine production in human rheumatoid synovial membrane cultures (10). These cultures spontaneously release cytokines and matrix metalloproteases without the need for exogenous stimulation and were used in the initial studies that identified anti-TNF as a potential therapeutic agent for rheumatoid arthritis (17). Incubation of MN-1 showed a dose dependent inhibition of the spontaneous release of TNF and IL-1 (Figure 4A). DMSO was used as a vehicle control at an equivalent amount to the 100 μM dose of MN-1. Cell viability was assessed using the MTT assay and found to be unaffected at all concentrations (Figure 4B). Data pooled from individual donors showed a significant inhibition of TNF by 46.6% ± 5.1 and IL-1 by 82.2% ± 4.2 in the MN-1 treated samples (Figure 4C).

SS has filed a patent application (WO/2008/090334). Use of 5HT Receptor Antagonists for Treating Arthritis. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.