EG (cas 23,313–21‐5) was obtained from Weikeqi (Sichuan, China), CP‐94253 (4‐{5‐propoxypyrrolo [3,2‐b]pyridin‐3‐yl}‐1.2,3,6‐tetrahydropyridine, catalog # SML0588), DOI (1‐(4‐iodo‐2,5‐dimethoxyphenyl)propan‐2‐amine, catalog #D101), dopamine (4‐(2‐Aminoethyl)benzene‐1,2‐diol, catalog #H8502), and caffeine from Sigma‐Aldrich, ramelteon (N‐[2‐[(8S)‐2.6,7,8‐tetrahydro‐1H‐cyclopenta [e][1]benzoxol‐8‐yl]ethyl]propanamide) from MedChemExpress (catalog # HYA0014), and dihydroergotamine ((2,4, 7R)‐N‐[(1S,2S,4R, 7S)‐7‐benzyl‐2‐hydroxy‐4‐methyl‐5,8‐dioxo‐3‐oxa‐6,9‐diazatricyclo[7.3.0.0  ^∧  {2.6}]dodecan‐4‐yl]‐6‐methyl‐6,11‐diazatetracyclo [7.6.1.0^∧{2.7}.0 ^∧ {12.16}]hexadeca‐1 (16),9,12,14‐tetraene‐4‐carboxamide)and SB 224289 ([4‐[2‐methyl‐4‐(5‐methyl‐1,2,4‐oxadiazol‐3‐yl)phenyl]phenyl]‐(1′‐methylspiro [6,7‐dihydro‐2H‐furo [2,3‐f]indole‐3.4′‐piperidine]‐5‐yl)methanone) from APEx BIO (catalog #B3459 and B6641, respectively). OxB (RSGPPGLQGRAQRLLQASGNHAAGILTM‐NH2), Aβ40 (rat:DAEFGHDSGFEVRHQKLVFFAEDVGSNKGAIIGLMVGGVV; human: DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVV) and Aβ42 (rat: DAEFGHDSGFEVRHQKLVFFAEDVGSNKGAIIGLMVGGVVIA; human: DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVVIA) were synthesized by Shangting (Shanghai,China). The Flp‐In™ T‐REx™ 293 cells, Lipofectamine®2000 transfection reagent, and cell culture materials were from Invitrogen (Thermo Fisher Scientific Inc., United States). The anti‐VSV‐G antibody (catalog #V5507) was obtained from Sigma‐Aldrich, and the anti‐ERK1/2‐MAP kinase (catalog # 9102S) and anti‐phospho‐ERK1/2‐MAP kinase (catalog # 9101S) antibodies were from Cell Signaling Technology (Nottingham,United Kingdom).

The VSV-G-mGluR5-5-HT_1B, VSV-G-mGluR5-5HT_2A, D₂, MT₂, and OX₂ constructs were established by inserting human 5-HT_1B cDNA into a pcDNA5/FRT/TO vector (Invitrogen) using the In-Fusion® PCR Cloning System (Clontech, United States) as previously described (Xu et al., 2012). Constructs were introduced into HEK293T cell lines for preliminary transient transfection studies and then into Flp-In™ T-REx™ 293 cells to generate inducible stable cell lines (Ward et al., 2011).

The cells were co-transfected with the pOG44 plasmid and the desired protein cDNA in pcDNA5/FRT/TO in a 9:1 ratio using Lipofectamine®2000 to generate Flp-In™ T-REx™ 293 cells that can inducibly express the indicated constructs (Ward et al., 2011). After 48 h, the medium was supplemented with 200 ng ml⁻¹ hygromycin to select stable transfected cells. The cell pools were tested for inducible expression by adding 100 ng ml⁻¹ doxycycline for 12 h, followed by western blot analysis for phosphorylation of extracellular regulated ERKs corresponding to GPCR activation and VSV-G protein expression.

HEK293T cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 0.292 g⋅ L⁻¹ l-glutamine, whereas PC12 and SH-SY5Y cells were maintained in PRIM 1640 medium and 10% (v/v) fetal bovine serum (FBS) at 37°C in a 5% CO₂ humidified atmosphere. The cells were transfected with the indicated constructs using Lipofectamine®2000.

The structure of 5-HT_1B, access number 5V54 (Name: Crystal structure of 5-HT_1B receptor in complex with methiothepin), was obtained from the PDB database (http://www.rcsb.org/pdb/home/home.do). The 3D structures and properties of CP-94253 and EG were obtained from the PubChem database (https://pubchem.ncbi.nlm.nih.gov). The free molecular docking software AutoDock Tools-1.5.6 was employed, and the Autodock 4 (Department of Molecular Biology of Scripps Research Institute, La Jolla, CA and Department of Cognitive Science of University of California, San Diego, La Jolla, CA) was used for flexible docking, as previously reviewed and discussed (Beuming et al., 2015; Kontoyianni, 2017).

Cells were washed once in cold phosphate-buffered saline (PBS) (120 mM NaCl, 25 mM KCl, 10 mM Na₂HPO₄, and 3 mM KH₂PO₄, pH 7.4) and harvested with ice-cold radioimmunoprecipitation assay (RIPA) buffer (50 mM HEPES, 150 mM NaCl, 1% Triton X-100, and 0.5% sodium deoxycholate, 10 mM NaF, 5 mM EDTA, 10 mM NaH₂PO₄, and 5% ethylene glycol pH 7.4) supplemented with a protease inhibitor cocktail, as previously described. After heating the samples at 37°C for 5 min, the cell lysates were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The proteins were electrophoretically transferred onto a nitrocellulose membrane, which was then blocked (5% fat-free milk powder in PBS containing 0.1% Tween-20) at room temperature on a rotating shaker for 1 h, washed twice in tris-buffered saline (TBS, pH 7.4) containing 0.1% Tween 20, and incubated with the appropriate primary antibody for 2 h, followed by incubation with a secondary antibody for 1 h at room temperature. Subsequently, the proteins of interest were visualized using the ECL Chemiluminescence System (Santa Cruz Biotechnology) according to the manufacturer’s instructions. Protein levels were quantified by band intensity using the Quantity One 1D Analysis Software.

Intact Flp-In™ T-REx™ 293 cells induced to express mGluR5-VSV-G-5-HT_1B were treated with EG, dihydroergotamine, CP-94253, or the vehicle at the indicated concentrations and used for subsequent studies. In the DARTS experiment, cell lysates were subjected to pronase digestion at room temperature for 3 min, then loaded for western blotting to detect the 5-HT_1B protein. In the CETSA experiment, the cells were scraped off and lysed via alternate freezing and thawing thrice with liquid nitrogen, then heated to varying temperatures for 5 min. The 5-HT_1B protein was detected with western blotting. In both experiments, GAPDH was used as the loading control for quantification.

The 5-HT_1B FRET sensor-expressing cells were placed in a microscope chamber containing physiological HEPES-buffered saline solution (130 mM NaCl, 5 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, 20 mM HEPES, and 10 mM d-glucose, pH 7.4). The cells were then imaged using an inverted Nikon TE2000-E microscope (Nikon Instruments) equipped with a 40× (numerical aperture = 1.3) oil immersion Fluor lens, as previously described (Xu et al., 2012). The monochromator was set at 427 nm/bandwidth (BW) 5 and 504 nm/BW 5 nm to visualize the surface located cyan fluorescent protein (CFP) and FlAsH separately. FRET and donor emission images were recorded simultaneously using a Quadview 2 (QV2) image splitting device (Photometrics, United Kingdom), coupled to a CoolSnap-HQ2 camera connected to the microscope. The FRET and donor signals were detected simultaneously using the following Chroma (Brattleboro, VT) ET dichroic and emitter series mounted in the QV2 cube: ET t505LPXR dichroic, ET535/30 nm, and ET 632/60 nm. Using the streaming capability of the multiple dimensional wavelength acquisition modules of MetaMorph, the ligand-induced changes in intramolecular FRET were recorded directly in the computer’s hard drive at 40 ms intervals during excitation with 427 nm light. The Cool Snap-HQ2 camera was operated in the 14-bit mode, and exposure time, 8 × 8 binning, and camera gain, were kept constant for all streaming experiments. Computerized control of all electronic hardware and camera streaming acquisition was achieved using the MetaMorph software (version 7.7.5 Molecular Devices, Sunnydale, CA). A peristaltic pump, operated at a 5 ml/min flow rate, was used to rapidly add or remove test ligands to or from the imaging chamber. The ratiometric quantification of intramolecular charge changes was then calculated as the average 535 nm emission intensity divided by the average 470 nm emission intensity. The FRET ratiometric was set to 1.0 at the onset of each experiment and plotted over time.

Cells stably expressing inducible GPCRs were placed in 6-well plates and allowed to grow overnight. Expression of the constructs was induced by adding 10 ng ml⁻¹ doxycycline for 12 h and rendered quiescent via serum starvation for 12 h, then stimulated for 5 min by the indicated agonist using FBS as the positive control and deionized water as the negative control. Cells were then placed on ice and harvested with ice-cold RIPA buffer. Phosphorylation of ERK1/2 MAP kinases was detected via western blotting using a phospho-ERK1/2-specific antibody. The nitrocellulose membranes were subsequently stripped of immunoglobulins and re-probed using an anti-ERK1/2 antibody to assess the protein loading equivalence.

Total RNA was isolated from PC12 or worm model cells treated as previously indicated, using TRIzol (Ambion Life Technologies), and the mRNA was reversed-transcribed using the RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher, catalog #K1621) according to the manufacturer’s instructions. The primer sequences for quantitative PCR were as follows: TNF-α F (5′-CCC​TCA​CAC​TCA​GAT​CAT​CTT​CT-3′), TNF-α R (3′-GCT​ACG​ACG​TGG​GCT​ACA​G-5′), β-Actin F (5′-GAT​TAC​TGC​TCT​GGC​TCC​TAG​C-3′), and β-Actin R (3′-GAC​TCA​TCG​TAC​TCC​TGC​TTG​C-5′). Previously described primers were used for the serotonin receptor subtypes (Koizumi and Nakajima, 2014) and the worm model (Bellier et al., 2009; Sahu et al., 2012). Three independent experiments were performed on triplicate samples. PCR amplification was performed using the SYBR Green PCR Master Mix Kit (Thermo Fisher). All quantifications were normalized to GAPDH levels.

The PC12 cells were seeded in a 96-well plate at a 5 × 10³ density; 12 h after seeding, CP-94253 or EG was added with or without SB 224289. After 24 h of seeding, the amyloid peptides were added, and after 18 h, the cells were subjected to CCK-8 (catalog #C0037, Beyotime, China), MTT (catalog #M1025, Solarbio, Beijing, China), and LDH (catalog # BC0680, Solarbio, Beijing, China) assays following the manufacturers’ instructions.

The C. elegans CL14176 strain was synchronized to the L1 phase and allocated to 160 μL S media with the appropriate treatment drug concentrations. The worms were cultured at 16°C for 24 h and transferred to 25°C, at which point they were activated to express amyloid peptides that would result in paralysis (Dostal and Link, 2010). At 38 h, the worms were counted for the paralysis phenotype. Each group contained approximately 60 worms. The positive control group was set by adding 6.27 mM caffeine.

Variables were compared between non-treated and treated groups using Student’s t-tests. Statistical differences in protein expression, cell viability, and relative mRNA levels between two groups were analyzed using a One-way Analysis of variance (ANOVA). Data were analyzed using GraphPad Prism 6.01 software (GraphPad Software, La Jolla, CA, United States). Data are expressed as mean ± S.D. of values from at least three independent experiments. A value of p < 0.05 was considered statistically significant.

Molecular docking is widely used to identify specific molecular targets, such as GPCRs (Beuming et al., 2015; Kontoyianni, 2017). We used this technique to determine whether EG binds to the human GPCR 5-HT_1B. The binding affinity of EG to 5-HT_1B was similar to that of the agonist CP-94253, as the binding energies were −8.09 and −8.48 kcal mol⁻¹, respectively. EG (Figures 1D,E) docked into the same 5-HT_1B pocket as CP-94253 (Figures 1A,B). In addition, EG (Figure 1F) interacted with 5-HT_1B similarly to the agonist CP-94253 (Figure 1C). The specific interaction parameters are shown in Supplementary Figure S1. The results suggested that EG might target 5-HT_1B and act similarly to CP-94253 as a ligand of this receptor.

To verify the interaction of EG with the 5-HT_1B receptor, we applied DARTS and CETSA, previously reported methods (Chang et al., 2016), for identifying target proteins of natural products without chemical modifications. Unlike that with the single target protein, the conformation and stabilization of the compound-target complex are altered in either DARTS or CETSA, and the interaction between the compound and target protein can be visualized via western blotting (Lomenick et al., 2009; Molina et al., 2013; Chang et al., 2016). In this study, DARTS was performed on the whole-cell lysate of cells stably induced to express 5-HT_1B. The proteins were digested with pronase, whereas EG, CP-94253, and dihydroergotamine (a 5-HT_1B agonist, FDA-approved for migraine) were used to prevent target depletion. The 5-HT_1B protein in the band protected by EG (10⁻⁶ M) was enriched approximately 2-fold compared with that in the vehicle, and EG showed a similar protective effect with dihydroergotamine at a low concentration (10⁻⁸ M). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was resistant to pronase under this condition and served as a loading indicator (Figures 2A,B). The 5-HT_1B protein protected by EG, compared with CP-94253, from digestion is shown in Supplementary Figure S2A. CETSA was performed using intact cells induced to express 5-HT_1B. The dihydroergotamine- or vehicle-treated cells were heated to varying temperatures as indicated. The 5-HT_1B protein in the soluble fractions was separated from the precipitated destabilized protein and detected with western blotting (Figure 2C). Figure 2D shows ligand-target interaction plotted against temperature to display the shifts. The results revealed a physical interaction between the agonist dihydroergotamine and its target protein 5-HT_1B in intact cells. Additionally, EG-induced thermodynamic stabilization of 5-HT_1B was observed (Figures 2E,F). The 5-HT_1B protein thermodynamically stabilized by EG, compared with CP-94253, is shown in Supplementary Figure S2B.

Intramolecular FRET sensors have been used as real-time optical tools for GPCR ligand binding (Xu et al., 2012; Lohse et al., 2008). We successfully produced effective 5-HT_1B sensor constructs using previously described methods (Xu et al., 2012). The full-length human 5-HT_1B receptor was added to CFP in-frame with the C-terminal tail (Figures 3A,B), and the 12-amino acid sequence (FLNCCPGCCMEP) containing the fluorescein arsenical hairpin binder via tetra-cysteine (FlAsH) labeling sequence (CCPGCC) was inserted into the third intracellular loop of the receptor. In addition, it contained the VSV-G peptide epitope within the extracellular N-terminal domain (Figure 3B). Functionality of the 5-HT_1B sensor has been previously assessed in a study on ERK1/2 MAP kinase phosphorylation (p-ERK) (Xu et al., 2012). The full-length coding sequence of the human 5-HT_1B receptor without alterations to the receptor’s intracellular segments was also cloned into the pcDNA5/FRT/TO vector described previously (Xu et al., 2012) and in the Methods section. Then, these constructs were transfected into Flp-In™ T-REx™ 293 cells to generate stable cell lines. In these cells, the doxycycline-induced expression of either the 5-HT_1B sensor or the full-length human 5-HT_1B was examined using an anti-VSV antibody. The 5-HT_1B sensor performed effectively in response to the selective 5-HT_1B agonist CP-94253, compared with the full-length human 5-HT_1B, at the p-ERK level (Figure 3C). The 5-HT_1B sensor was effectively delivered to the cell surface, on which its normal functions are based (Figure 3D). These results showed that the sensor was functionally similar and virtually identical to the wild-type receptor.

To study the effects of EG on the 5-HT_1B receptor, we imaged the intramolecular rearrangement of the 5-HT_1B sensor in response to EG, compared with CP-94253, by measuring emission from CFP and FlAsH. The normalized basal FRET signal in cells induced to express the 5-HT_1B sensor decreased immediately upon the addition of CP-94253, then rapidly restored agonist withdrawal in a concentration-dependent manner. In contrast, addition of the 5-HT_1B antagonist SB 224289 (10⁻⁶ M) did not produce a significant alteration in sensor response beyond the vehicle effect, and, as anticipated, blocked the effect of CP-94253 (Figure 4A). This was similar to the results obtained using EG, although the FRET signal changed in the opposite direction to that of CP-94253 (Figure 4B). Moreover, the variability of FRET changes from EG was approximately 60% of CP-94253 (Figure 4C), illustrating that structure changes in the receptor triggered by binding of CP-94253 and EG were distinct in amplitude and distance, at least between the third intracellular loop and C-terminal, where CFP and FlAsH are located, respectively. 5-HT_1B has been previously shown to be internalized from the cell surface via agonist binding in a time-dependent manner (Janoshazi et al., 2007). Our findings revealed that EG induced 5-HT_1B internalization, similarly to CP-94253 (Figure 4D). These results demonstrated that EG could be a 5-HT_1B agonist.

To further study the EG target, we used stable cell lines that were induced to overexpress serotonergic-related GPCRs, including the serotonin receptors 1B and 2A (5-HT_1B and 5-HT_2A), dopamine receptor 2 (D₂), melatonin receptor 2 (MT₂), and orexin receptor 2 (OX₂). In the stable cell lines that over-expressed 5-HT_1B, EG promoted the phosphorylation of ERK1/2 MAP kinases in a concentration-dependent manner. This activation through the 5-HT_1B receptor-mediated pathway was similar to the 5-HT_1B selective agonist, i.e., the positive control CP-94253 (Figure 5A). In addition, the 5-HT_1B selective antagonist SB224289 at 10⁻⁷ M inhibited the effect of the concentration-dependent promotion of the ERK1/2 MAP kinase phosphorylation by EG and CP-94253 activity (Figure 5B). However, no response to the EG of other 5-HT-related GPCRs activities, such as 5-HT_2A, D₂, MT_2, or OX_2, was observed (Figures 5C–F). These results showed that EG exhibited agonistic activity toward 5-HT_1B and could selectively activate the 5-HT_1B-mediated signaling pathway.

Aβ-induced neuronal toxicity is an important pathology in AD, and there are two principal variants of amyloid peptides in humans, of which Aβ₄₂ is more toxic than Aβ₄₀ (Hamley, 2012). EG and its derivatives exhibit neural protection activity and can be used in AD treatment (Lin et al., 2015). We applied Aβ₄₀ and Aβ₄₂ to PC12 (with rat Aβ) and SH-SY5Y (with human Aβ) cell lines, respectively. The cells were subsequently assessed using the Cell Counting Kit-8 (CCK-8) for viability assay. Cell survival decreased with increased Aβ concentration in both PC12 (Figure 6A) and SH-SY5Y cells (Supplementary Figure S3). No significant differences in viability were observed in the treatments with <5 μM Aβ₄₂ or <10 μM Aβ₄₀. PC12 cell death was observed at 10 μM Aβ₄₂ and reached approximately 18% at 20 μM, whereas cell death in the Aβ₄₀ treatment group began at 20 μM. At 50 μM, 47% of the Aβ₄₂-treated and 52% of the Aβ₄₀-treated cells survived. These results suggest that Aβ peptides dose-dependently induce neuronal toxicity, and support the hypothesis that Aβ₄₂ is more toxic than Aβ₄₀. To determine the effects of EG on neural cells, we pre-treated PC12 cells with or without EG, followed by exposure to the amyloid peptides Aβ₄₀ and Aβ₄₂, respectively. The results showed that EG could alleviate the reduction in Aβ₄₀-and Aβ₄₂-induced cell viability by approximately 8.22 and 8.64%, respectively (Figure 6B).

As EG could selectively activate 5-HT_1B signaling and alleviate Aβ-induced mortality, we further investigated the underlying mechanisms. Activation of 5-HT_1B by its agonist reduces the levels of pro-inflammatory TNF-α in rat nerve cells (Khalilzadeh et al., 2018). The inflammatory cascade is one of the most important processes in Aβ-induced toxicity, and the Aβ oligomer-elevated TNF-α signaling is a cause of memory loss in mice (Lourenco et al., 2013); thus, stimulation of 5-HT_1B may relieve Aβ-induced TNF-α signaling. The two amyloid peptides Aβ₄₂ and Aβ₄₀ induced the death of PC12 cells to a similar extent (Figure 6A). Therefore, we conducted the following experiments with Aβ₄₂ because it exhibited neuronal toxicity at a lower concentration than Aβ₄₀.

To evaluate TNF-α levels under 5-HT_1B-related conditions, we used quantitative RT-PCR to detect TNF-α mRNA levels in 5-HT_1B over-expressing stable PC12 cells (Supplementary Figure S4B) with Aβ₄₂ and with or without 5-HT_1B activators and blockers (Figure 7A). PC12 cells have endogenous 5-HT_1B as well as 5-HT_2A/3/6 receptor subtypes (Supplementary Figure S4A). We used 5-HT_1B-over-expressing stable PC12 cells to accentuate the effect of 5-HT_1B. Consistent with previous studies (Shrewsbury et al., 2008), our results showed that Aβ₄₂ elevated TNF-α expression, which could be reversed by treatment with the 5-HT_1B agonist CP-94253 and dihydroergotamine and the novel 5-HT_1B activator EG. Moreover, the selective serotonin 5-HT_1B antagonist SB 224289 blocked EG, CP-94253, and dihydroergotamine-mediated TNF-α restraint under Aβ₄₂ conditions. The 5-HT_1B activators alone could decrease TNF-α levels in the 5-HT_1B over-expressed PC12 cells, whereas the antagonist SB 224289 contributed to pro-inflammatory TNF-α signaling and was synergistic with Aβ₄₂ (Figure 7B). These results suggested that EG resorted to the 5-HT_1B pathway to reduce Aβ-induced pro-inflammatory TNF-α signaling.

The Aβ-induced inflammatory response and subsequent neuronal death are considered important events in AD pathology (Hamley, 2012; Unzeta et al., 2016). We found that the death of Aβ₄₂-induced 5-HT_1B over-expressing stable PC12 cells was significantly reversed by the 5-HT_1B agonists and EG, as assessed via the CCK-8 and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays (Figures 7C,D, Supplementary Figure S5A). However, SB 224289 alone did not lead to significant cell death under the indicated conditions (10⁻⁷ M, 42 h, Supplementary Figure S6). EG and the 5-HT_1B agonists reduced the Aβ₄₂-induced cytotoxicity that could be blocked by SB 224289 (Figure 7E, Supplementary Figure S5B).

To further explore the ability of EG to afford protection against amyloid peptides, we tested transgenic C. elegans for resistance to the induced Aβ expression. The C. elegans CL14176 strain (Link et al., 2003), which could be induced to express Aβ by modulating the temperature and leads to paralysis in 2–3 d, was used. This paralysis phenotype could serve as a measurement of Aβ toxicity to assay the effects of compounds protective against Aβ, using the paralysis delay to indicate the suppression of Aβ toxicity in the worm model (Dostal and Link, 2010). The results showed that EG at different concentrations retarded the paralysis time from 6 to 7 h compared with that for the untreated group (∼52 h). The performance of the 200 μM EG group was superior to that of the positive control caffeine group (Supplementary Table S1, Figure 8A). In this model, the survival rate of the EG group at 200 μM was significantly prolonged (Figure 8B). The expression of immune response-related genes that are analogs of the inflammatory effectors in humans, including the nuclear receptor nhr-57, effectors of hypoxia pathway, pqn-5 (prion-like glutamine (Q)/asparagine (N) -5) gene, part of the unfolded protein response pathway that responded to endoplasmic reticulum stress, the collagen gene col-41, and transthyretin-like protein gene ttr-21, between the untreated and EG groups was observed via quantitative RT-PCR (Bellier et al., 2009; Sahu et al., 2012). The EG could significantly reduce the Aβ-induced expression of the immune response-related genes nhr-57, pqn-5, and col-41. However, it failed to suppress ttr-21 expression (Figure 8C), which suggested that EG protected the worm model from AD in a complex manner rather than arbitrarily suppressing all the immune response pathways.