Geniposide and its iridoid analogs exhibit antinociception by acting at the spinal GLP-1 receptors
Abstract
We recently discovered that the activation of the spinal glucagon-like peptide-1 receptors (GLP-1Rs) by the peptidic agonist exenatide produced antinociception in chronic pain. We suggested that the spinal GLP-1Rs are a potential target molecule for the management of chronic pain. This study evaluated the antinociceptive activities of geniposide, a presumed small molecule GLP-1R agonist. Geniposide pro- duced concentration-dependent, complete protection against hydrogen peroxide-induced oxidative damage in PC12 and HEK293 cells expressing rat and human GLP-1Rs, but not in HEK293T cells that do not express GLP-1Rs. The orthosteric GLP-1R antagonist exendin(9-39) right-shifted the concentration eresponse curve of geniposide without changing the maximal protection, with identical pA2 values in both cell lines. Subcutaneous and oral geniposide dose-dependently blocked the formalin-induced tonic response but not the acute flinching response. Subcutaneous and oral geniposide had maximum inhi- bition of 72% and 68%, and ED50s of 13.1 and 52.7 mg/kg, respectively. Seven days of multidaily subcu- taneous geniposide and exenatide injections did not induce antinociceptive tolerance. Intrathecal geniposide induced dose-dependent antinociception, which was completely prevented by spinal exendin(9-39), siRNA/GLP-1R and cyclic AMP/PKA pathway inhibitors. The geniposide iridoid analogs geniposidic acid, genipin methyl ether, 1,10-anhydrogenipin, loganin and catalpol effectively inhibited hydrogen peroxide-induced oxidative damage and formalin pain in an exendin(9-39)-reversible manner. Our results suggest that geniposide and its iridoid analogs produce antinociception during persistent pain by activating the spinal GLP-1Rs and that the iridoids represented by geniposide are orthosteric agonists of GLP-1Rs that function similarly in humans and rats and presumably act at the same binding site as exendin(9-39).
1. Introduction
The endogenous incretin glucagon-like peptide-1 (GLP-1) and exogenous exenatide have biological actions in the pancreas, such as stimulating glucose-dependent insulin synthesis and secretion, decreasing glucagon levels and notably altering b-cell proliferation and apoptosis (Baggio and Drucker, 2007; Li et al., 2003). They also have therapeutic value in treating type 2 diabetes mellitus (DeFronzo et al., 2005; Triplitt and Defronzo, 2006). These actions are mediated by activating the GLP-1 receptors (GLP-1Rs), which are members of the class B1 family of seven-transmembrane- spanning, heterotrimeric G-protein coupled receptors (GPCRs) and whose signal transduction is mainly effected through the cal- cium and cyclic AMP/protein kinase A (PKA) signaling pathways (Holst, 2007; Ramos et al., 2008; Roger et al., 2011). In humans and rodents, single, structurally identical GLP-1Rs are expressed in the pancreatic islets (Tornehave et al., 2008), lungs, brain, cardiovas- cular system, dorsal root ganglia and spinal cord, among others (Bullock et al., 1996; Gong et al., 2014a; Hanson et al., 1998; Himeno et al., 2011; Kim et al., 2013; Teramoto et al., 2011).
We recently discovered that GLP-1Rs are specifically expressed in spinal dorsal horn microglial cells and are up-regulated during microglial proliferation and hypertrophy after peripheral nerve injury. The intrathecal administration of the GLP-1R peptidic ago- nists exenatide and GLP-1(7-36) produced specific, potent anti- nociception of 60e90% during formalin-induced tonic pain, painful peripheral neuropathy, bone cancer-induced mechanical allodynia and diabetic pain. Our results suggested that the activation of the spinal GLP-1Rs by peptidic agonists leads to specific anti- nociception in chronic pain hypersensitivity states, notably re- fractory neuropathic pain, cancer pain and painful diabetic neuropathy (Gong et al., 2014a). The GLP-1R antinociception hy- pothesis can be confirmed by the application of small molecule GLP-1R agonists. Small molecule agonists with different chemical structures to exenatide and GLP-1 may be helpful for excluding the possibility that the antinociception induced by exenatide and GLP-1 is not related to the activation of the GLP-1Rs. Favorable results will also be useful for the development of orally-available analgesics that can easily penetrate the central nervous system.
The GLP-1Rs belong to a class B1 family of GPCRs that interact with peptidic ligands by a hypothesized two-step mechanism. The extracellular N-terminal ectodomain interacts with the C-terminal residues of cognate ligands and positions the N-terminus of the ligand to interact with the critical determinants in the receptor’s transmembrane regions, leading to full activation and signal transduction (Castro et al., 2005; Hoare, 2005; Laburthe et al., 2007). Small molecule compounds may not have sufficient mass or diameter to react with both the N-terminal ectodomains and the receptor transmembrane regions, and therefore possibly interact with only one binding site. It would be interesting to know whether small molecules produce antinociception by activating the GLP-1Rs with a full intrinsic efficacy. As peptidic exenatide and GLP-1 are not orally-active, the development of orally-available, non-peptidic GLP-1R agonists has been an active, although as yet unfruitful, area of research for both anti-diabetic drugs and analgesics (Wang et al., 2010; Willard et al., 2012). As exenatide and GLP-1 cannot readily cross the blood brain barrier, orally-available GLP-1R agonists with the ability to penetrate the central nervous system, leading to more specific, efficacious analgesics, must be developed.
Geniposide (Fig. 1) is one of the main iridoid glycosides of Gardenia jasminoides (Hu et al., 2005), which has been widely used in Chinese traditional medicine for its homeostatic, antiphlogistic, antinociceptive and antipyretic effects and indications. Geniposide can induce neuronal differentiation and protection against oxida- tive stress and can regulate insulin secretion (Guo et al., 2012; Liu et al., 2006, 2009). Its insulin regulation ability is blocked by the GLP-1R antagonist exendin(9-39) (Guo et al., 2012). The potency, efficacy and activation mode of geniposide’s activation of the GLP- 1Rs has not yet been characterized. Many structurally related iri- doid glycosides are found in plants (Ghule and Yeole, 2012; Jeong et al., 2012; Park et al., 2011a, 2011b; Sundaram et al., 2012; Xu et al., 2011). Understanding the structureeactivity relationship of geniposide may suggest whether the iridoids represent a novel core structure for GLP-1R activation.
We studied the effects of geniposide and a series of structurally related iridoids in GLP-1R-sensitive in vitro and in vivo assays and compared them to exenatide. We tested the protective effect of geniposide against hydrogen peroxide-induced oxidative damage in PC12 cells that express rat GLP-1Rs, HEK293 cells that stably express human GLP-1Rs and HEK293T cells that do not express GLP-1Rs. This hydrogen peroxide assay is known to be GLP-1R- sensitive (Liu et al., 2009; Oeseburg et al., 2010).
We examined the antinociceptive effect of geniposide using a formalin test. Formalin induces an acute flinching response, rep- resenting an acute nociceptive reflex response without central sensitization, and a tonic flinching response, which represents persistent pain involving central sensitization (Coderre et al., 1990; Dickenson and Sullivan, 1987). The primary active sites for geni- poside antinociception and its ability to develop antinociceptive tolerance were also determined.
We explored the blocking effects of exendin(9-39), the GLP-1R gene silencer siRNA/GLP-1R and cyclic AMP/PKA signaling pathway inhibitors on spinal geniposide antinociception. We compared the activity of six iridoid analogs of geniposide, geni- posidic acid, genipin, genipin methyl ether, 1,10-anhydrogenipin, loganin and catalpol (Fig. 1), in the hydrogen peroxide and formalin tests to geniposide. Our results indicate that orally-available geni- poside and its iridoid analogs produce antinociception entirely via the activation of the spinal GLP-1Rs and suggest that the iridoids are novel orthosteric agonists of rat and human GLP-1Rs that possibly act at the same binding site as exendin(9-39).
2. Materials and methods
2.1. Drugs
Geniposide, geniposidic acid and genipin were purchased from the Linchuan Zhixin Biotechnology Co. (Jiangxi, China). Loganin was purchased from Fluka Chemicals (Shanghai, China) and catapol from the Chengdu Purifa Technology Development Co. (Sichuan, China). 1,10-Anhydrogenipin was synthesized by dis- solving genipin in methylene chloride with first triphenylphosphine and then dii- sopropyl azodicarboxylate, according to a previous report (Zhang et al., 2006). Genipin methyl ether was obtained by the methanolysis of geniposide in the
presence of sulfuric acid (Yu et al., 2008). The structures of 1,10-anhydrogenipin and genipin methyl ether were confirmed by NMR spectroscopy with a purity of more than 95%. Morphine hydrochloride, exenatide and exendin(9-39) were obtained from Northeast Pharmaceuticals (Shenyang, China), Kaijie Bio-Pharmaceutical Co. (Chengdu, China) and Shanghai TASH Biotechnology Co. (Shanghai, China), respectively. 2′,5′-Dideoxyadenosine (DDA) and N-[2-(p-bromocinnamylamino)ethyl]-5-isoquinolinesulfonamide (H-89) were purchased from Santa Cruz Biotech (Santa Cruz, CA, USA). All drugs and reagents were freshly dissolved in normal saline so- lution, except 1,10-anhydrogenipin and genipin methyl ether, which were dissolved in 25% dimethyl sulfoxide (DMSO) in saline.
The siRNA targeting the GLP-1Rs and its nonspecific oligonucleotide were syn- thesized by GenePharma Co. (Shanghai, China), following Yin et al. (2010). The 19 nucleotide duplex and two unpaired nucleotides overhanging 3′ ends were as follows. GLP-1R: 5′-GUA UCU CUA CGA GGA CGA GUU-3’/5′-CUC GUC CUC GUA GAG AUA CUU-3′; nonspecific oligonucleotide control: 5′-UUC UCC GAA CGU GUC ACG UUU-3’/5′-ACG UGA CAC GUU CGG AGA AUU-3′. To formulate the siRNAs, 25 kDa linear polyetherimide (PEI) (Polysciences, Niles Illinois, PA, USA) was dissolved in 5% dextrose diethyl pyrocarbonate-treated water (pH 7.4). 1 mg RNA was dissolved in 1.5 mg PEI in a PEI:RNA ratio of six equivalents of PEI nitrogen per RNA phosphate, forming RNAepolymer complexes at room temperature for 20 min (Chen et al., 2012).
2.2. Cell culture and the determination of cell viability
PC12 cells expressing GLP-1Rs originally derived from the rat adrenal medulla pheochromocytoma (Greene and Tischler, 1976; Perry et al., 2002) were purchased from the Shanghai Institute for Cell Biology Cell Bank (Shanghai, China) and were routinely cultured in high-glucose Dulbecco’s modified Eagle’s medium (DMEM) containing 5% fetal bovine serum, 10% horse serum, 100 U/ml penicillin and 100 mg/ L streptomycin. HEK293 cells that stably express human GLP-1Rs (a gift from HD Biosciences, Shanghai, China) and parent HEK293T cells that do not express GLP-1Rs (purchased from Applied StemCell, Inc., CA, USA) were cultured in high-glucose DMEM containing 10% fetal bovine serum, 2.0 mM L-glutamine, 100 mg/ml hygromycin B, 100 U/ml penicillin and 100 mg/L streptomycin. Cells were cultured at 37 ◦C in a humidified atmosphere of 95% and 5% CO2 and were passed by trypsinization every two to three days.
The cultured cells were grown on 96-well plates at a density of 1.6 × 104 cells/ well. Twenty-four hours after incubation, once the cells were attached, hydrogen peroxide (500 or 800 mM, final concentration) was added to the PC12 cells for 15 min or to the HEK293 and HEK293T cells for 5 min. The cells were washed, then treated with iridoids and exenatide at different concentrations in the absence or presence of different concentrations of exendin(9-39) and cultured again for 12 h. The MTT assay was conducted to determine the cell viability. The cells in 96-well plates were rinsed with phosphate buffered saline. 0.5 mg/ml MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-
diphenyl tetrazolium bromide, from Amresco, Solon, OH, USA) was added to each well. The microplate was incubated at 37 ◦C for 4 h. The medium with MTT was removed and 200 ml DMSO was added to each well. The plate was shaken on a microplate shaker to dissolve the blue MTT-formazan. The absorbance was read at 570 nm against a reference wavelength of 630 nm on a microplate reader (Multiskan MK3, Thermo Labsystems, Vantaa, Finland). The hydrogen peroxide-induced cell viability was expressed as a percentage of normal cell viability in the absence of hydrogen peroxide.
2.3. Western blotting
The lumbar enlargement region of the spinal cord (consisting of approximately the L3eL5 segment) were dissected from rats after sacrifice and homogenized in radioimmunoprecipitation (RIPA) buffer containing the protease inhibitor phenyl- methylsulfonyl fluoride (Beyotime Institute of Biotechnology, Jiangsu, China). The protein concentration of the RIPA lysates was determined using a standard bicin- choninic acid protein assay (Beyotime Institute of Biotechnology, Jiangsu, China).RIPA lysate (20 mg of protein) was separated using gel electrophoresis on a 12% polyacrylamide gel and transferred to a polyvinylidene fluoride membrane (Milli- pore, Bedford, MA, USA). The membrane was incubated with the GLP-1R antibody (ab39072, 1:2000, rabbit polyclonal, Abcam, Cambridge, UK) and b-actin antibody (1:2000, rabbit polyclonal, Protein Tech Group), then was incubated with a sec- ondary antibody (IRDye 800 conjugated affinity purified goat anti-rabbit IgG; Rockland). The resulting protein bands were visualized with the Odyssey Infrared Imaging system from Li-Cor Biosciences. The band intensity was quantified using a computer-assisted image analysis program (Image J Software, National Institutes of Health, Bethesda, MD, USA). The GLP-1R/b-actin band intensity ratio was calculated to quantify the relative protein expression levels, controlling sampling errors.
2.4. Immunofluorescence staining
Immunofluorescence labeling of the GLP-1Rs were performed on cultured PC12, HEK293 and HEK293T cells and was observed using a confocal microscope (TCS SP8, Leica Microsystems). The cells were placed in six-well plates (5 × 104 cells/well) with poly-L-lysine coated coverslips at the bottom, fixed in 4% paraformaldehyde and incubated in 10% goat serum (vol/vol) and 0.5% Triton-X 100 (vol/vol) in phosphate buffered saline for 1 h. The cells were then incubated with the GLP-1R antibody for 24 h at 4 ◦C. The GLP-1R antibody was tested for the optimal dilution and nonspecific staining. The GLP-1R band was visualized with the Alexa 555- conjugated goat anti-rabbit secondary antibody (1:200, Invitrogen). DAPI (4′,6- diamidino-2-phenylindole, Sigma) staining was also used to determine the posi- tion of the cell nuclei.
2.5. Animals
Male Swiss mice (20e25 g) and Wistar rats (180e250 g) were purchased from the Shanghai Experimental Animal Institute for Biological Sciences (Shanghai, China) and housed in a temperature- and humidity-controlled environment on a 12 h light/dark cycle (lights on at 7:00 AM) with food and water ad libitum. The animals were acclimated to the laboratory environment for three to five days before undergoing surgery or entering the study. All of the experiments were performed in accordance with the Animal Care and Welfare Committee of Shanghai Jiao Tong University School of Pharmacy and followed the animal care guidelines of the Na- tional Institutes of Health. Every effort was made to reduce the number of animals used, to minimize their suffering and to use alternatives to in vivo techniques were they existed. The experimental study groups for behavior testing were assigned randomly without the researcher’s knowledge.
2.6. Intrathecal catheterization and injection in rats
The catheterization and injection of the rats was conducted intrathecally as described by Gong et al. (2011a) and intraventricularly as described by Lu et al. (2012).
2.7. Mouse and rat formalin test
The rat formalin test was performed as previously described (Gong et al., 2011a, 2014b). The rats were acclimated individually to the observation cage for 30 min prior to receiving a subcutaneous injection of 50 mL 5% formalin in 0.9% saline on the dorsal side of the left hindpaw. The resulting nociceptive behavior was manually quantified by counting the number of times that the formalin-injected paw flinched in 1 min epochs. Measurements were taken at 10 min intervals beginning imme- diately after the formalin injection and ending 90 min later.
The mouse formalin test was performed as previously described (Gong et al., 2014b, 2012). 10 ml 5% formalin was subcutaneously injected to the dorsal side of the right hindpaw and the mouse was immediately placed in a transparent obser- vation box. The duration of nociceptive licking was manually quantified from zero to five minutes after the injection, representing the acute nociceptive response, and 20e40 min after the injection, representing the tonic pain.
3. Results
3.1. Activation of rat and human GLP-1Rs via the exendin(9-39)- sensitive site
The activation of the GLP-1Rs is known to protect against hydrogen peroxide-induced oxidative damage (Liu et al., 2007; Oeseburg et al., 2010). We evaluated the protective effects of gen- iposide and exenatide in PC12 cells that express GLP-1Rs to identify whether geniposide can directly activate the GLP-1Rs (Perry et al., 2002). We confirmed the presence of GLP-1Rs in the PC12 cells by immunostaining (Fig. 2A). Treatment with exenatide (up to 3 × 10—8 M) or geniposide (up to 3 × 10—3 M) in PC12 cells was not associated with any cellular dysfunction. Exenatide completely protected the PC12 cells from hydrogen peroxide-induced oxidative damage in a concentration-dependent manner (EC50 = 3.8 × 10—9 M). Geniposide produced the same concentration-dependent protection (EC50 = 1.7 × 10—4 M), with an EC50 approximately 45,000-fold less potent than exenatide (Table 1 and Fig. 2B). Treatment with geniposidic acid, genipin methyl ether, 1,10-anhydrogenipin, loganin or catalpol at up to 3 × 10—3 M in PC12 cells was not associated with any cellular dysfunction. However, 10—4 M genipin caused significant cell death, probably due to its protein cross-linking activity, whereas its dehydrated derivative 1,10-anhydrogenipin lacked this activity (Zhang et al., 2006). All of the compounds except genipin were therefore tested further. Geniposidic acid, genipin methyl ether, 1,10-anhydrogenipin, loga- nin and catalpol exhibited similar concentration-dependent pro- tection against hydrogen peroxide-induced viability loss (Fig. 2C), with EC50 values ranging from 6.0 × 10—5 to 2.4 × 10—4 M (Table 1).
In a recent study, we investigated the protective effects of two other iridoid compounds, shanzhiside methylester and 8-O-acetyl shanzhiside methylester (Fig. 1) (Zhu et al., 2014). Their ED50 values of 9.3 × 10—5 M and 12 × 10—5 M are also listed in Table 1.
To test whether geniposide can activate human GLP-1Rs, the protective effects of exenatide and geniposide were examined in HEK293 cells that stably expressed human GLP-1Rs. The expression of GLP-1Rs in the HEK293 cells was again confirmed with specific immunostaining (Fig. 2D). Treatment with exenatide (up to 3 × 10—8 M) or geniposide (up to 3 × 10—3 M) in HEK293 cells was not associated with any cellular dysfunction. Both exenatide (EC50 = 2.8 × 10—9 M) and geniposide (EC50 = 1.3 × 10—4 M) pro- duced concentration-dependent complete protection against hydrogen peroxide-induced oxidative damage, at a ratio of approximately 1:46,000 (Table 1 and Fig. 2E). These results suggest that there is no difference in geniposide and exenatide’s ability to activate GLP-1Rs in rats or in humans. We tested geniposide and its iridoid analogs in HEK293T cells that do not express GLP-1Rs (http://webserver.mbi.ufl.edu/wshaw/293.html) to confirm that their protective effects against hydrogen peroxide-induced oxida- tive damage are via the activation of the GLP-1Rs. There was no specific GLP-1R immunostaining in the HEK293T cells (Fig. 2F). As shown in Fig. 2G, treatment with exenatide (30 nM) did not have any protective effect against hydrogen peroxide oxidative damage in HEK293T cells. Conversely, treatment with nicotinamide (3 mM) exhibited approximately 60% protection, in agreement with pre- vious findings (Wilkinson et al., 2008). Geniposide and its iridoid analogs at 1 mM had none or only residual protective effects, which were probably nonspecific.
3.2. Antinociception occurs via actions at the spinal cord with no induction of antinociceptive tolerance
We tested the antinociceptive effects of exenatide and genipo- side in mice. Five groups of normal mice received a subcutaneous administration of saline (10 ml/kg) or exenatide (10, 30, 100 or 300 mg/kg) 30 min before a formalin challenge. The subcutaneous injection of formalin in control mice produced a biphasic licking response consisting of an acute phase (within 5 min after the formalin injection) and a tonic phase (20e40 min after the injec- tion). Compared to the saline control, the subcutaneous injection of exenatide dose-dependently prevented the formalin-induced tonic flinching response, with maximum inhibition at 100 mg/kg or less, but did not block the formalin-induced acute nociceptive response up to 300 mg/kg (Fig. 4A). A doseeresponse analysis of exenatide by best fit showed that the Emax was 82.0% and the ED50 was 22.7 mg/kg (5.4 nmol/kg, 95% confidence limits: 21.4e24.0 mg/kg) (Fig. 4B).
Six groups of mice received a subcutaneous administration of saline (10 ml/kg) or geniposide (3, 10, 30, 100 or 300 mg/kg) 30 min before a formalin challenge. The subcutaneous geniposide dose- dependently prevented the formalin-induced tonic flinching response, but did not block the formalin-induced acute nociception (Fig. 4C). A doseeresponse analysis showed that the Emax was 71.5% and the ED50 was 13.1 mg/kg (33.8 mmol/kg, 95% confidence limits: 9.0e19.1 mg/kg) (Fig. 4B). The ED50 dose ratio for subcutaneous geniposide vs. exenatide in mole base was roughly 6200:1.
Six groups of fasting mice (12 h) received an oral gavage of saline (10 ml/kg) or geniposide (10, 30, 100, 300 or 1000 mg/kg) 1 h before a formalin challenge. The oral gavage of geniposide dose- dependently prevented the formalin-induced tonic response but not the acute flinching response (Fig. 4D), with an Emax of 58% and an ED50 of 52.7 mg/kg (95% confidence limits: 50.1e55.4 mg/kg) (Fig. 4B). The apparent oral bioavailability was 25% that of the subcutaneous route. No apparent sedative or motor side effects were observed after the oral gavage or subcutaneous administra- tion of geniposide or exenatide.
The antinociceptive effects of geniposide via direct injection to the spinal cord, brain and peripheral afferent nerve terminals were tested. Six groups of rats chronically implanted with intrathecal cannulas received an intrathecal single injection of saline (10 ml) or geniposide (3, 10, 30, 100 or 300 mg) 30 min before a formalin in- jection. The paw subcutaneous injection of formalin in control rats produced a characteristic biphasic flinching response consisting of an initial, rapidly decaying acute phase (within 10 min after the formalin injection) followed by a slowly rising, long-lived (10e 90 min) tonic phase. Although up to 300 mg geniposide did not block the flinching response in the acute phase, it did prevent the formalin-induced tonic flinching response in a dose-dependent manner (Fig. 4E). The areas under the flinching response curve from 10 to 90 min (AUC10e90 min) were calculated. A doseeresponse analysis showed that the Emax value was 57.0% and the ED50 value was 17.3 mg (44.5 nmol, 95% confidence limits: 12.5e23.8 mg) (Fig. 4F). In our previous study, the ED50 value for exenatide spinal analgesia in the formalin-induced tonic pain was 2.5 ng (0.6 pmol) (Gong et al., 2014a). The ED50 dose ratio for spinal geniposide vs. exenatide in a mole base was approximately 74,000:1. No apparent sedative or motor side effects were observed after the intrathecal injection of geniposide.
Two groups of rats chronically implanted with intraventricular cannulas received an intraventricular injection of 5 ml saline or 300 mg geniposide 30 min before a formalin injection. Compared to the saline control, 300 mg intraventricular geniposide did not significantly block either the formalin-induced tonic response or the acute flinching response (Fig. 4G). Two groups of rats received a paw subcutaneous injection of either formalin or 300 mg geniposide dissolved in formalin. The paw co-injection of geniposide did not effectively produce antinociception (Fig. 4H). No apparent sedative or motor side effects were observed after the intraventricular or local injection of geniposide. These results indicate that the spinal cord is a primary site responsible for geniposide-induced anti- nociception rather than a supraspinal or peripheral site.
We investigated the antinociceptive effects of the iridoid com- pounds geniposidic acid, genipin, 1,10-anhydrogenipin, genipin methyl ether, loganin and catalpol via direct intrathecal adminis- tration. Each six groups of rats received an intrathecal injection of
10 ml saline or 25% DMSO in saline (to control for 1,10-anhydrogenipin and genipin methyl ether) or an iridoid (3, 10, 30,100 or 300 mg) 30 min before a formalin injection. Up to 300 mg of any of the iridoid compounds did not inhibit the acute nociceptive response but did prevent the formalin-induced tonic pain in a dose- dependent manner (Fig. 5AeF). No apparent sedative or motor side effects were observed for these iridoids, except for 300 mg genipin, which paralyzed rats 24 h after injection, probably due to its pro- tein cross-linking effect (Zhang et al., 2006). A doseeresponse analysis using the AUC10e90 min showed that the Emax values for the iridoids ranged from 62.8% to 88.6% and that the ED50 values ranged from 11.3 mg to 23.8 mg (33.7e90.2 nmol) (Fig. 5G, Table 1).
Shanzhiside methylester (ED50 = 23.3 mg) and 8-O-acetyl shanzhiside (ED50 = 20.5 mg) were also shown to alleviate the formalin-
induced tonic flinching response in our recent study (Zhu et al., 2014) and are included in Table 1.As the activation of the peripheral GLP-1Rs produces hypogly- cemia and the spinal cord is the primary site for GLP-1R anti- nociception, we examined whether the spinal exenatide and geniposide antinociception observed was due to hypoglycemia. Three groups of intrathecally cannulated rats received an intra- thecal single injection of saline (10 ml), exenatide (300 ng) or gen- iposide (300 mg). Random blood sugar (representing postprandial sugar) levels were measured prior to and 1 h after the intrathecal injection using the glucose oxidase method and the OneTouch®
Ultra® 2 Blood Glucose Meter (Johnson & Johnson Company,California, USA) (Gong et al., 2011b). Compared to the saline control group, neither exenatide nor geniposide induced hypoglycemia in rats (Fig. 5H). Our results rule out the possibility that spinal GLP-1R antinociception is due to hypoglycemia.
Morphine tolerance to analgesia is a common phenomenon and hampers morphine’s clinical usefulness for the treatment of cancer and non-cancer pain. We tested whether a chronic treatment of geniposide induced antinociceptive tolerance. Seven groups of mice received multiple bidaily subcutaneous injections of saline (10 ml/kg, bid), exenatide (100 mg/kg, bid), geniposide (100 mg/kg, bid) and morphine (10 mg/kg, bid) for seven continuous days, fol- lowed by a single subcutaneous injection of saline (10 ml/kg), exenatide (100 mg/kg), geniposide (100 mg/kg) or morphine (5 mg/ kg) on the eighth day, 30 min before a formalin challenge. A single injection of morphine, but not exenatide or geniposide, blocked the formalin-induced acute nociceptive response, whereas morphine, geniposide and exenatide all inhibited the tonic flinching response in mice treated multidaily with saline (Fig. 6A, B). In the tonic phase, multiple bidaily injections of geniposide and exenatide over seven days maintained their antinociceptive efficacies, whereas the same long-term regimen of morphine completely abolished its antinociception (Fig. 6B). These results indicate that geniposide and exenatide, unlike morphine, do not induce antinociceptive toler- ance after chronic treatment.
3.3. Antinociception by the activation of the spinal GLP-1Rs and subsequent cyclic AMP/PKA signaling pathway
To test whether geniposide produced antinociception via the activation of the spinal GLP-1Rs, a GLP-1R gene silencer (siRNA) (Yin et al., 2010) was used in the rat formalin test. Four groups of intrathecally cannulated rats received multidaily intrathecal in- jections of 7.5 mg PEI (control vehicle), 5 mg nonspecific oligonu- cleotide or 5 mg siRNA/GLP-1R for seven continuous days before an intrathecal single injection of saline or geniposide followed by a formalin challenge. The single intrathecal injection of geniposide effectively blocked the formalin-induced tonic pain by 76.7% in PEI- treated rats or 88.0% in nonspecific oligonucleotide-treated rats. The intrathecal injections of siRNA/GLP-1R completely blocked geniposide’s antinociceptive effect (Fig. 7A, B). The rats were sacrificed immediately after the behavior test and spinal homogenates were obtained from lumbar enlargements of the L3eL5 cord for a Western blot. GLP-1R protein expression was significantly reduced by siRNA/GLP-1R (Fig. 7C insert). The GLP-1R/b-actin band intensity ratio was calculated after scanning. Spinal GLP-1R protein expression was reduced 54.0% from the vehicle control and 53.5% from the nonspecific oligonucleotide control by siRNA/GLP-1R (Fig. 7C).
We used the specific orthosteric antagonist exendin(9-39) to verify the action site of geniposide and the other iridoids. Three groups of intrathecally cannulated rats received an intrathecal in- jection of 10 ml saline, 100 mg geniposide or 2 mg exendin(9-39) + 100 mg geniposide 30 min before a formalin challenge. A previous study showed that 2 mg exendin(9-39) did not alter the formalin-induced flinch response, but completely prevented the antinociceptive effect of exenatide and GLP-1(7-36) in the tonic phase (Gong et al., 2014a). As exhibited in Fig. 7D and E, the intrathecal geniposide injection significantly inhibited the formalin-induced tonic pain by 57%, but did not alter the acute nociceptive response. The intrathecal co-injection of 2 mg exendin(9-39) completely blocked geniposide’s antinociceptive effect. Our results further support the spinal cord as the primary site and suggest that the activation of the spinal GLP-1Rs is responsible for geniposide’s antinociceptive effect.
Eleven groups of intrathecally cannulated rats received an intrathecal injection of 10 ml saline or 100 mg geniposidic acid, genipin, genipin ether, 1,10-anhydrogenipin, loganin or catalpol in the absence or presence of 2 mg exendin(9-39) 30 min before a formalin injection. The intrathecal injection of 100 mg of any of the iridoids markedly inhibited formalin-induced tonic pain by approximately 60%, but did not alter the acute nociceptive response. The intrathecal injection of 2 mg exendin(9-39) completely blocked the iridoids’ antinociceptive effects (Fig. 8Ae C). The antinociceptive effects of shanzhiside methylester and 8-O- acetyl shanzhiside methylester were also demonstrated to be completely blocked by exendin(9-39) (Zhu et al., 2014).
The recognition of GLP-1 by its cognate receptors in the pancreatic islets activates a classical signaling cascade through stimulatory G-protein and adenylyl cyclases (Ramos et al., 2008; Roger et al., 2011). Cellular levels of cyclic AMP increase and PKA activates its downstream effectors, resulting in insulin secretion (Holst, 2007). We tested whether the cyclic AMP/PKA signaling pathway is also involved in geniposide-induced antinociception, by using the adenylyl cyclase inhibitor DDA (Ramos et al., 2008) and the PKA inhibitor H89 (N-[2-(p-bromocinnamylamino)ethyl]-5- isoquinolinesulfonamide) (Li et al., 2010). Four groups of rats received an intrathecal injection of 10 ml saline or 1 mg DDA. Ten minutes later, they received a second intrathecal injection of 10 ml saline or 100 mg geniposide, followed 30 min later by a formalin challenge. The intrathecal DDA did not alter either the formalin- induced acute nociceptive response or the tonic pain but completely prevented geniposide’s antinociceptive effect in the tonic phase (Fig. 9A, B).
Four groups of rats received an intrathecal injection of 10 ml saline or 10 mg H89. After 10 min, they received a second intrathecal injection of 10 ml saline or 100 mg geniposide, followed 30 min later by a formalin challenge. The intrathecal injection of H89 did not alter the formalin-induced acute nociception or tonic pain, but did completely prevent geniposide’s antinociceptive effect in the tonic phase (Fig. 9C, D). These results suggest that spinal geniposide antinociception is mediated by the activation of the cyclic AMP/PKA signaling pathway secondary to GLP-1R activation, which is consistent with GLP-1 action in insulin release and glucose regu- lation (Holst, 2007).
4. Discussion
The therapeutic success of peptidic GLP-1R agonists for the treatment of type 2 diabetes mellitus has inspired discovery efforts aimed at developing orally-available small molecule GLP-1R ago- nists. Although the GLP-1R is a member of the structurally complex class B1 family of GPCRs, there have been increasing reports of structurally diverse small molecule orthosteric and allosteric ago- nists with intrinsic efficacy for the GLP-1Rs (Wang et al., 2010; Willard et al., 2012). It has been reported that substituted cyclo- butanes such as Boc5 and BW4-24 (Chen et al., 2007; Liu et al., 2012) and pyrimidines such as Compound B (Sloop et al., 2010) are orthosteric agonists, whereas 6,7-dichloroquinoxalines (Irwin et al., 2010; Teng et al., 2007) and Compound 2 (Knudsen et al., 2007) are allosteric agonists or regulators. This study adds our new attempts to this list. We demonstrate that geniposide is similar but less potent than exenatide. It reverses hydrogen peroxide-induced oxidative damage in both rat and human GLP-1R-expressing PC12 and HEK293 cells, but not in HEK293T cells that do not express GLP- 1Rs. Exendin(9-39) competitively blocks geniposide’s protective effect with similar pA2 values to those when competing with exe- natide. Geniposide 100% reverses hydrogen peroxide oxidative damage and its maximal spinal antinociception of formalin pain is between 60 and 80%, which identical to the action of exenatide and GLP-1(7-36) (Gong et al., 2014a). Our results confirm and extend the findings of Guo et al. (2012) by demonstrating that geniposide ap- pears to be a reversible orthosteric agonist of GLP-1Rs and possibly acts at the same binding site as exendin(9-39) and exenatide. Our results also suggest that full intrinsic efficacy may not be required for GLP-1R activation that simultaneously binds both the presumed extracellular N-terminal ectodomain and the determinant in the receptor transmembrane region.
This and our previous study (Zhu et al., 2014) tested the iridoid analogs of geniposide, geniposidic acid, genipin methyl ether, 1,10- anhydrogenipin, loganin, catalpol shanzhiside methylester and 8- O-acetyl shanzhiside methylester, using GLP-1R sensitivity assays. The in vitro results and the in vivo antinociceptive effects reveal that all of these compounds similarly activate the GLP-1Rs. If genipo- side’s C-1 glycoside moiety is removed, resulting in genipin (Fig. 1), the antinociceptive activity remains the same. Genipin-induced GLP-1R activation is also not significantly altered by replacing the free C1-hydroxyl with a methyl group, forming genipin methyl ether, or by dehydration, forming 1,10-anhydrogenipin. Geniposide-induced GLP-1R activity is not significantly altered by the removal of the free C4-carboxyl group, forming the C4- decarboxylated iridoid catalpol. Although a detailed structure- activity relationship cannot be extrapolated from our results due to the small number of compounds and the low activities, the re- sults do suggest that the iridoid structure represented by genipo- side and genipin analogs is a novel core structure for activating GLP-Rs. They can be chemically modified further to obtain iri- doids with increased activities.
We previously demonstrated that the activation of the spinal GLP-1Rs by peptidic agonists leads to specific antinociception in chronic pain hypersensitivity states. Spinal GLP-1Rs may represent a promising potential therapeutic target molecule for the man- agement of chronic pain (Gong et al., 2014a). Geniposide and its iridoid analogs produce antinociception via the same mechanism. Geniposide and its iridoid analogs dose-dependently block formalin-induced tonic pain but not acute nociceptive responses. Our previous results also showed that other two iridoid analogs, shanzhiside and 8-O-acetyl shanzhiside methylester, exhibited specific antinociception in peripheral nerve injury- and bone cancer-induced mechanical allodynia and in formalin-induced tonic pain (Zhu et al., 2014). The antinociceptive effects of geni- poside and its iridoid analogs are produced via the spinal cord, rather than the brain ventral area or peripheral nociceptors. The antinociceptive effects of geniposide and its iridoid analogs are completely prevented by exendin(9-39), the GLP-1R silencer siRNA/ GLP-1R and the cyclic AMP/PKA signaling pathway inhibitors (DDA and H89). Geniposide’s in vivo spinal antinociception appears to be correlated to its GLP-1R-mediated in vitro activity, as supported by the similarity of the potency ratios between geniposide and exe- natide’s activation of the GLP-1Rs in the cell protection assay (6500:1) and during direct spinal antinociception (7400:1). The assessment of geniposide and the other iridoids in PC12 and HEK293 cells suggests a direct interaction between these com- pounds and the spinal GLP-1Rs, rather than via an interaction with a signal transduction pathway after the activation of the GLP-1Rs. Our study using geniposide and its iridoid analogs with chemical structures distinct from exenatide and GLP-1(7-36) further vali- dates the hypothesis that the activation of the spinal GLP-1Rs produces specific antinociception.
Exenatide and GLP-1(7-36) are oligopeptides and must be administered parenterally. The ED50 value for the subcutaneous injection of exenatide to produce antinociception is 22.7 mg/kg, roughly 38-folds higher than the 0.6 mg/kg required for its pe- ripheral blood sugar lowering effect (Gong et al., 2011b). The gen- iposide vs. exenatide potency ratio for spinal antinociception is 74,000:1, whereas the ratio for subcutaneous injection is 6200:1. The more than 10-fold-reduction in the potency of exenatide antinociception indicates that it does not easily reach its spinal target molecule and that intrathecal or epidural administration may be more desirable. We therefore evaluated the feasibility of geniposide and its iridoid analogs producing antinociception via peripheral routes. Although geniposide and its iridoid analogs are less potent GLP-1R agonists in vitro, their subcutaneous injection (ED50: 13.1 mg/kg) and oral gavage (ED50: 52.7 mg/kg with an apparent oral bioavailability of 25% of the subcutaneous route) produce a fairly potent antinociception. This is clinically significant, given that gabapentin, a widely-prescribed pain medicine for chronic neuropathic pain and painful diabetic neuropathy, is dosed in a range of 100e200 mg/kg for rats and mice and 900e1200 mg for human patients. Geniposide appears to activate rat and human GLP-1Rs similarly and long-term exposure to geniposide after chronic treatment does not induce antinociceptive tolerance. Our data therefore suggests that orally-available geniposide and its iridoid analogs are good analgesic leads for further development and for chemical modifications to increase their potency.