Address correspondence and reprint requests to Biljana Georgievska, AstraZeneca R&D, Innovative Medicines CNS & Pain, Södertälje, Sweden. E-mail: firstname.lastname@example.org
Abnormal tau phosphorylation resulting in detachment of tau from microtubules and aggregation are critical events in neuronal dysfunction, degeneration, and neurofibrillary pathology seen in Alzheimer's disease. Glycogen synthase kinase-3β (GSK3β) is a key target for drug discovery in the treatment of Alzheimer's disease and related tauopathies because of its potential to abnormally phosphorylate proteins and contribute to synaptic degeneration. We report the discovery of AZD1080, a potent and selective GSK3 inhibitor that demonstrates peripheral target engagement in Phase 1 clinical studies. AZD1080 inhibits tau phosphorylation in cells expressing human tau and in intact rat brain. Interestingly, subchronic but not acute administration with AZD1080 reverses MK-801-induced deficits, measured by long-term potentiation in hippocampal slices and in a cognitive test in mice, suggesting that reversal of synaptic plasticity deficits in dysfunctional systems requires longer term modifications of proteins downstream of GSK3β signaling. The inhibitory pattern on tau phosphorylation reveals a prolonged pharmacodynamic effect predicting less frequent dosing in humans. Consistent with the preclinical data, in multiple ascending dose studies in healthy volunteers, a prolonged suppression of glycogen synthase activity was observed in blood mononuclear cells providing evidence of peripheral target engagement with a selective GSK3 inhibitor in humans.
Alzheimer's disease (AD) is a severely debilitating disease, which affects more than 35 million patients worldwide (Querfurth and LaFerla 2010; Alzheimer's Association 2011). The onset is insidious and as the disease progresses, there is an impairment of learning abilities, disorientation and decline in language function. The pathological hallmarks include extracellular amyloid plaques and the presence of intra-neuronal neurofibrillary tangles. Synaptic loss and neurodegeneration is evident early resulting in an increased rate of brain atrophy.
Glycogen synthase kinase-3β (GSK3β) is a highly conserved serine/threonine kinase that has highest abundance in the brain during development and is localized primarily in neurons (Takahashi et al. 1994). Although GSK3 has two major isoforms, α and β, identifying an isoform-specific GSK3 inhibitor is challenging since the two isoforms are 98% identical within the ATP pocket of the catalytic domain (Woodgett 1991). GSK3 is the major kinase which phosphorylates the microtubule binding protein tau (Hanger et al. 1992; Ishiguro et al. 1993; Lovestone et al. 1994) and GSK3β has been shown to co-localize with neurofibrillary tangles in post-mortem AD brain (Yamaguchi et al. 1996; Imahori and Uchida 1997; Pei et al. 1999). Abnormal hyperphosphorylation of tau leads to microtubule dysfunction and is believed to promote formation of paired helical filaments, a key component of neurofibrillary tangles (Ballatore et al. 2007). Paired helical filaments are found in neuronal cell bodies and apical dendrites and are implicated as a causative agent in a number of neurodegenerative tauopathies. Consistent with this, over-expression of GSK3β in transgenic mice has been reported to cause tau hyperphosphorylation, neurodegeneration, and behavioral deficits (Lucas et al. 2001; Hernández et al. 2002).
In addition, GSK3β has been implicated to have a role in synaptic plasticity in neuronal networks because of its influence on diverse substrates involved in signaling pathways (Hooper et al. 2008; Peineau et al. 2008). Inhibition of GSK3β facilitates induction of long-term potentiation (LTP) and blocks induction of long-term depression (Hooper et al. 2007; Peineau et al. 2007, 2009), two major forms of long-lasting plasticity in the brain triggered by the synaptic activation of NMDA receptors. LTP is characterized by increased synaptic efficacy and is thought to be one of the neurophysiological correlates of learning and memory (Citri and Malenka 2008). One memory paradigm often used in behavioral testing of rodents is contextual fear conditioning (CFC). Impairments in fear conditioning can be induced by NMDA receptor blockade (using, e.g., low doses of MK-801), thus disrupting a key molecular component in memory formation (Bardgett et al. 2003).
Given the wide range of pathological and functional roles of GSK3β in the brain, it is important to identify a brain permeable small molecule GSK3 inhibitor that can be tested in the clinic for its therapeutic potential in the treatment of AD and related tauopathies. We hereby report the discovery of a potent and selective orally active, brain permeable GSK3 inhibitor, AZD1080, a clinical candidate that progressed into Phase 1 clinical trials in man.
Materials and methods
Co-crystallization of AZD1080 (2-Hydroxy-3-[5-[(morpholin-4-yl)methyl]pyridin-2-yl]-1H-indole-5-carbonitrile citrate) with GSK3β was performed as previously described (Bhat et al. 2003).
The GSK3β, Cdk2, and Cdk5 Ki's were determined using scintillation proximity assays and kinetic analyses as described earlier (Bhat et al. 2003; Berg et al. 2012). The GSK3α assay was performed as described for the GSK3β assay, but utilizing recombinant human GSK3α obtained from University of Dundee (Scotland, UK). The KM value of ATP used to calculate the Ki value for GSKα was 10 μmol/L. Inhibition of Cdk1 was performed by Upstate Ltd (Dundee, UK), according to a method previously described (Davies et al. 2000). The KM value of ATP used to calculate the Ki value was 51 μmol/L. Erk2 activity was determined according to the manufacturer's description using an Ser/Thr kinase SPA kit (GE Healthcare, Uppsala, Sweden, RPNQ 0200), p42 MAPK kinase (NEB, Ipswich, MA, USA; P6080S, 20 U/well) , and biotinylated MBP (Upstate #13-111). The KM value of ATP used to calculate the Ki value was 71 μmol/L.
The selectivity of AZD1080 (at a concentration of 10 μM) was tested against 24 protein kinases at University of Dundee. The enzymatic activity was measured in the presence of 0.1 mM ATP. Off target selectivity was also evaluated at MDS Pharma Services (Taipei, Taiwan) on the following targets: adenosine A1, A2A, A2B, adrenergic α1A, α1B, α1D, α2A, α2B, β1, β2, and norepinephrine transporter, bradykinin B1 and B2, calcium channel type L, benzothiazepine and dihydropyridine, calcium channel type N, dopamine D1, D2L, D3, D4 and dopamine transporter, endothelin ETA and ETB, epidermal growth factor, estrogen ERα, GABAA agonist and benzodiazepine site, GABAB and GABA transporter, glucocorticoid, glutamate kainite, NMDA agonist, glycine and phencyclidine site, histamine H1, H2 and H3, imidazoline I2, interleukin IL-1α, leukotriene B4 and D4, muscarinic M1, M2 and M3, neuropeptide Y1 and Y2, nicotinic, opiate δ, κ and μ, phorbol ester, platelet activating factor, potassium channel (KATP), purinergic P2X and P2Y, serotonin 5-HT1A, 5-HT2A, 5-HT3 and transporter, sigma σ1 and σ2, sodium channel site 2, tachykinin NK1, and testosterone.
Inhibition of tau phosphorylation in vitro and in vivo
3T3 fibroblasts engineered to stably express 4-repeat human tau were used and detection of total tau and phosphorylated tau (P-Ser396) by Western blotting was performed as previously described (Bhat et al. 2003).
All animal procedures were conducted in accordance with the European Communities Council Directive of 24 November 1986 (86/609/EEC) and according to Stockholm Animal Research Ethical Committee guidelines and approved by the Stockholm South Animal Research Ethical Committee. All animal studies are reported according to the ARRIVE guidelines. A total of 58 young (11–12 days old) male and female Sprague–Dawley rats (Taconic, Denmark) were kept with their mothers in conventional housing and fed standard rodent chew and tap water ad libitum. The rats received AZD1080 (3 or 10 μmol/kg) or vehicle (water with 0.5% ascorbic acid, 0.01% EDTA, pH 2.0) via oral gavage (6 mL/kg). At 1, 2, 3, 6 or 24 h after administration the rats were killed by decapitation and the hippocampi dissected out and snap frozen on dry ice. The right hippocampus was taken for determining the concentration of AZD1080 (see below). The left hippocampus was sonicated in lysis buffer (6 μL/mg tissue) consisting of Cell Extraction Buffer (Life Technologies, Paisley, UK), HALT phosphatase inhibitor cocktail (Nordic Biolabs, Täby, Sweden) and Complete Protease inhibitor cocktail tablet (Roche Diagnostics Scandinavia AB, Bromma, Sweden) and centrifuged at 14 000 g for 20 min. Measurement of phosphorylated and total tau in the supernatants was performed by sandwich immunoassay using Gyrolab Workstation in CD-format (Bioaffy 200, Gyros AB, Uppsala, Sweden). Antibodies against phosphorylated tau (P-Thr231; AT180; Thermo Fisher, Rockford, IL, USA) and total tau (Tau5; Abcam, Cambridge, UK) were used. The ratio of P-Thr231 tau to total tau was calculated and percent inhibition versus vehicle-treated animals was plotted.
Blood was sampled from the rats after decapitation into pre-chilled EDTA microtainer tubes. Plasma was prepared by centrifugation for 10 min at approximately 3000 g at +4°C and the samples were stored frozen at −70°C until bioanalysis of AZD1080.
A total of 161 male C57BL/6 mice (Taconic, Denmark), 8–12 weeks of age, were used. The animals were kept in conventional housing (3–5 mice per cage) and fed standard rodent chew and tap water ad libitum. Typically 9–12 mice were included in each experimental group and 2–4 mice in the satellite groups (for determination of compound exposure in plasma and brain, see below). AZD1080 (4.0 or 15 μmol/kg) or vehicle (water with 0.5% ascorbic acid, 0.01% EDTA, pH 2.0) was administered by oral gavage (10 mL/kg) acutely or subchronically (twice daily) for 3 days. The training trial was performed at 1.5, 3, or 5 h after final administration with AZD1080. To disrupt learning, the mice received subcutaneous administration of MK-801 (0.1 or 0.15 mg/kg; (+)-MK.801 hydrogen maleate, Sigma-Aldrich, St Louis, MO, USA) or vehicle (saline) 30 min before the training trial.
CFC was performed using a two-chamber apparatus with a gridded floor, controlled by software from the manufacturer (TSE, Bad Homburg, Germany). During the training trial, the animal was allowed to explore the illuminated chamber for 180 s after which a foot shock (0.7 mA, 2 s duration) was delivered through the bars of the grid. The animal was left in the chamber for an additional 30 s after the foot shock. At 24 h after the training trial the animal was returned to the same chamber (test trial) and videotaped. The freezing behavior (expressed as% freezing during a 180 s period) was scored manually afterward by an investigator blinded to the treatment. Freezing was defined as lack of movement except that required for respiration.
Blood was withdrawn from anesthetized satellite mice by cardiac puncture and transferred into pre-chilled EDTA microtainer tubes. Plasma was prepared according to above. The left brain hemisphere was dissected, weighed and frozen on dry ice. The plasma and brain samples were stored frozen at −70°C until bioanalysis of AZD1080.
Ex vivo LTP in hippocampus
Animals and drug treatment
A total of 55 male C57BL/6 mice (Taconic, Denmark), 10–12 weeks of age, were used. Mice were dosed twice daily for 3 days via oral gavage (10 mL/kg) with either vehicle (0.5% ascorbic acid + 0.01% EDTA in water) or AZD1080 (15 μmol/kg). Individual mice were killed on the day of experimentation at 2 h following their final dose.
Slice preparation and maintenance
For the LTP studies, standard brain slice preparations were employed and a SliceMaster multi-station electrophysiology recording platform (Scientifica Ltd., East Sussex, UK) was used. Briefly, individual mice were anesthetized with isoflurane and decapitated. Transverse hippocampal slices (400 μm) were harvested and allowed to equilibrate in artificial cerebrospinal fluid (aCSF) This should be abbreviated in the text. at 30°C for at least 1 h before recording. aCSF was composed of the following (in mM): 130 NaCl; 3.5 KCl; 1.25 NaH2PO4; 24 NaHCO3; 10 Glu; 2.5 CaCl2; 1 MgCl2. Slices were immersed in aCSF which was delivered at a rate of ~2 mL/min from gravity-controlled reservoirs, and warmed within the slice chamber to 30–32°C. To most precisely replicate the in vivo conditions of the cognition experiments, AZD1080 (50 nM) was included in the aCSF for all slices taken from AZD1080-treated mice throughout the ex vivo experiments. Recording electrodes were pulled from borosilicate glass (TW150-4, WPI, Sarasota, FL, USA) and filled with NaCl (2 mM). The recording microelectrodes were placed in stratum radiatum among the apical dendrites of CA1 pyramidal cells. A concentric bipolar stimulating electrode (WPI) was placed in the Schaffer collateral fibers where they emerge from CA3. Input/output relationships were obtained in each experiment and stimulus intensity was adjusted to produce fEPSP amplitudes approximately 50% of maximum. Baseline data were collected for 10–15 min at a stimulation frequency of 0.03 Hz.
Conditioning protocol & LTP induction
The concentration of MK-801 used in these experiments was chosen to impair LTP induction. Previous work has reported an IC50 of 130 μM in rat hippocampus against LTP (Frankiewicz et al. 1996; Coan et al. 1997). However, this value was observed only after at least 6 h incubation time. Long incubation time is necessary in slices for two reasons: (i) MK-801 has relatively low affinity for the NMDA receptor (Kemp et al. 1987) and (ii) the reduced degree of NMDA receptor activity in brain slice preparations results in a slow development of receptor blockade by the open channel blocker MK-801. To facilitate activity-dependent block of NMDA receptors by MK-801, slices underwent a minimal tetanus protocol (MTP) before LTP induction. The MTP was designed to activate NMDA receptors to a level sufficient to facilitate rapid channel block by MK-801, but to remain below the level necessary to directly induce long-term plasticity. Once a stable baseline was established, MK-801 (300–500 nM) was added to the perfusion for 10 min. Slices were exposed to the MTP protocol (stimulated four times at 5-min intervals with one train of five pulses delivered at 200 Hz). Following the MTP, stimulation was returned to the monitoring level of 0.033 Hz for at least 10 min before LTP induction. LTP was induced using 100 Hz stimulation delivered for 1 s. Stimulus duration was increased to 5 msec during the 100 Hz train. Following the LTP protocol, recording resumed at 0.033 Hz and MK-801 was removed from the perfusion solution. Peak fEPSP amplitude values for 5 min prior to 100 Hz stimulation were averaged to provide the control measurement and peak fEPSP amplitude values from 40 to 45 min post-100 Hz stimulation were averaged to provide the amount of potentiation. LTP was defined as the percentage increase in potential of the post-100 Hz measurement compared to the control value. Peak amplitude of the potential was captured by Spike2 software during data acquisition for analysis offline.
Inhibition of glycogen synthase phosphorylation in peripheral blood mononuclear cells
Animals and drug treatment
A total of 71 adult male Sprague–Dawley rats (250–300 g; Taconic, Denmark) were used. The rats received an acute dose of AZD1080 (1, 3 or 10 μmol/kg) or vehicle (water with 0.5% ascorbic acid, 0.01% EDTA, pH 2.0) via oral gavage (dosing volume 5 mL/kg). At 1, 2, 3, 6, or 24 h after administration the rats were anesthetized and blood, from abdominal aorta, was sampled in heparin microtainer tubes. Peripheral blood mononuclear cells (PBMC) were isolated from the blood samples as described below. Separate blood samples were obtained for plasma processing and subsequent bioanalysis.
Phase 1 clinical study in healthy volunteers
The pharmacokinetic and pharmacodynamic profile of AZD1080 after oral multiple ascending doses was studied in a single center, double-blind, placebo-controlled, randomized study in healthy male and non-fertile female volunteers. The study was approved by the independent Institutional Review Board/Research Ethics Committee: Regional Ethics Committee in Stockholm, Sweden. The study was performed in accordance with the ethical principles that have their origin in the Declaration of Helsinki and informed consent was obtained from all subjects prior to initiation of the study.
Middle-aged subjects (45–64 years) received 15 mg AZD1080 (n = 6 subjects) or placebo (n = 6 subjects) once daily for 14 days. On day 1 and 8 following start of treatment blood samples were collected at the following time-points: 30 min pre-dose, 0 min, 20 min, 40 min, 1, 2, 4, 6, 10, and 24-h post-dose (for analysis of peripheral biomarkers and concentrations of AZD1080). PBMC were isolated from the blood and separate blood samples were obtained for plasma processing and subsequent bioanalysis.
Determination of phosphorylated and total glycogen synthase in PBMC
PBMC were isolated from rat and human blood using Accuspin Histopaque tubes (Sigma-Aldrich) according to instructions by the manufacturer. Briefly, the tubes were centrifuged at 1000 g for 15 min at 20°C. The content of the upper chamber was transferred to polypropylene tubes, the tubes were filled with phosphate-buffered saline (PBS) and centrifuged for 250 g for 10 min at 4°C. The supernatant was discarded, the pellet was mixed with 2 mL water and the tubes were filled with 12 mL PBS. The tubes were centrifuged at 250 g for 10 min at 4°C and the supernatant was discarded. The pellet was dissolved in 0.5 mL PBS and the suspension was transferred to 1.0 mL Nunc cryotubes. The tubes were centrifuged at 250 g for 10 min and the supernatant was completely aspirated. Lysis buffer (Cell Extraction Buffer, 1x P2714 and 1 mM PhenylMethaneSulfonylFluoride) was added to each pellet (300 μL). The pellets were incubated on ice for 30 min (mixed every 10 min) and the suspension was transferred to eppendorf tubes and centrifuged at 14 000 g for 10 min at +4°C. The supernatant was aliquoted and stored frozen at −70°C until further analysis. Rat PBMC lysates were diluted 2-fold for phosphorylated GS (P-GS) and 10-fold for total GS (T-GS) measurements using ELISA. Human PBMC lysates were diluted 20-fold for P-GS and 80-fold for T-GS. Determination of P-GS [pSpS641/645 epitope] and T-GS in PBMC lysates was determined using commercial ELISA kits (Biosource), according to the manufacturer's recommendations.
Bioanalysis of AZD1080
Determination of the total concentration of AZD1080 in plasma and brain homogenates was performed by ultrafiltration, followed by reversed-phase liquid chromatography and electrospray tandem mass spectrometry. AZD1080 determinations were performed in the range of 0.001–1.024 μmol/L. The concentration of AZD1080 in the brain (density = 1 g/mL) was determined without correction for possible blood.
Data were analyzed using Prism 5.0 software (GraphPad, San Diego, CA, USA) or Microcal Origin 8.0 (Originlab, Northhampton, MA, USA). The results are expressed as mean ± SEM. Statistical analysis was performed using one-way anova followed by Bonferroni's multiple comparison test (selected pairs) where p < 0.05 was considered significant.
Discovery of the GSK3 inhibitor AZD1080
Using innovative structure-based design approaches, we have discovered AZD1080, a potent and selective small molecule inhibitor of GSK3 (Molecular weight: 334 g/mole; solubility 24 μg/mL in water, Fig. 1a). To verify the binding mode, we generated a high resolution X-ray crystal structure (Fig. 1b). AZD1080 binds in the ATP pocket of GSK3β where the oxindole ring system forms hydrogen bonds to the back-bone of the ATP pocket. The cyano group is directed toward the conserved salt bridge and the solubilizing morpholine ring is pointed out toward the solvent area. The high resolution X-ray crystal structure confirms that AZD1080 inhibits GSK3β by binding within the ATP pocket.
AZD1080 is potent and selective in vitro
The potency of AZD1080 to inhibit GSK3β and its closely related kinases was evaluated in vitro. AZD1080 inhibited recombinant human GSK3α and GSK3β with a pKi of 8.2 (6.9 nM) and 7.5 (31 nM), respectively, and subsequently we refer to AZD1080 as a GSK3 inhibitor. AZD1080 showed selectivity against cdk2 (pKi = 5.9; 1150 nM; 37-fold), cdk5 (pKi = 6.4; 429 nM; 14-fold), cdk1 (pKi = 5.7; 1980 nM; 64-fold) and Erk2 (pKi< 5; > 10 μM; > 323-fold). AZD1080 (at 10 μM) was also evaluated for pan-kinase selectivity and showed good overall selectivity versus 23 kinases (Fig. 1c), as well as against 65 different receptors, enzymes and ion channels in MDS Pharma screen (< 50% effect at 10 μM AZD1080; see 'Materials and methods' for details).
AZD1080 inhibits human tau phosphorylation in cells
AZD1080 was initially tested in Caco-2 cells and exhibited high permeability values (17 × 10−6 cm/s) indicating that the compound readily penetrates into cells. To study GSK3β-mediated inhibition of tau phosphorylation (P-Ser396) in vitro, 3T3 fibroblasts engineered to stably express 4-repeat human tau were used. Concentration-dependent inhibition of tau phosphorylation was observed for AZD1080 (IC50 = 324 nM; Fig. 1d) and the non-selective reference GSK3 inhibitor LiCl (IC50 = 1.5 mM; Fig. 1d) indicating that AZD1080 was several orders of magnitude more potent than LiCl. The maximal inhibitory effect on P-Ser396 was observed at 1-2 h and the effect remained for at least 8 h (Fig. 1e). These results indicate that AZD1080 has the ability to inhibit phosphorylation of human tau protein in cells.
Central target engagement with AZD1080 in rats
To test whether AZD1080 has the potential to penetrate the blood–brain barrier, it was initially tested in a bovine endothelial blood–brain barrier cell assay and results indicated that AZD1080 exhibited high blood–brain barrier permeability values (8 × 10−3 cm/min). In addition, pharmacokinetic analysis in blood after oral administration revealed that AZD1080 had a good oral bioavailability in rats (15–24%) with a half-life of 7.1 h, making this compound attractive for further in vivo testing. To determine whether AZD1080 could inhibit tau phosphorylation in brain after oral administration, we employed an in vivo pharmacokinetic-pharmacodynamic (PK/PD) model in which tau phosphorylation is increased at a developmental period corresponding to increased dendritic pruning in the rat brain (Takahashi et al. 2000). AZD1080 or vehicle was given acutely via oral gavage to young rats. In time-response studies, a maximal inhibitory effect of 38 ± 2% and 48 ± 2% (3 and 10 μmol/kg, respectively) on tau phosphorylation (P-Thr231 epitope) was achieved in hippocampus 6 h after administration of AZD1080, compared to vehicle control (Fig. 2). The inhibitory effect was long-lasting and still pronounced at 24 h after oral treatment with AZD1080 at 3 μmol/kg (15 ± 7%; Fig. 2a) and 10 μmol/kg (30 ± 3%; Fig. 2b). In the contralateral hippocampus, maximal concentrations of AZD1080 were observed 1–2 h after oral dosing at 3 μmol/kg (~400 nM; Fig. 2a) and 2–3 h at 10 μmol/kg (~1200 nM; Fig. 2b). In contrast, the maximal inhibitory effect of AZD1080 on tau phosphorylation was observed at 6 h after dose, suggesting a delayed pharmacodynamic response. The brain/plasma exposure ratio for AZD1080 was estimated to 0.5–0.8 at peak concentrations, demonstrating good brain permeability in vivo. Overall, the studies in young rats demonstrate that AZD1080 is efficacious at inhibiting tau phosphorylation at the P-Thr231 epitope, displaying both time- and dose-dependent pharmacodynamics in brain.
AZD1080 reverses cognitive deficits in mice
In addition to the substrate tau, GSK3β has diverse effects on multiple signaling pathways including glutamatergic signaling in the brain. Therefore, we examined the potential of AZD1080 to reverse cognitive deficits using a CFC paradigm in mice where memory is disrupted by a low dose of the NMDA receptor antagonist MK-801. Animals were subjected to an adverse event (electric foot shock) in a defined context (training trial) and the conditioned fear response was evaluated 24 h later (test trial). MK-801 treatment given 30 min prior to the training trial significantly disrupted the freezing response in the test trial compared to vehicle-treated mice in all experiments (MK-801 vs. vehicle, p < 0.05 in Fig. 3a and p < 0.001 in Fig. 3b and c). Acute oral treatment with AZD1080 (15 μmol/kg) 3 h prior to the training trial (i.e., 2.5 h prior to MK-801 administration) was not able to prevent the MK-801-induced deficits in CFC (Fig. 3a). In contrast, subchronic (3 days) oral treatment with AZD1080 at 4 or 15 μmol/kg significantly blocked the MK-801-induced memory deficit (AZD1080 vs. MK-801, p < 0.05 at 4 μmol/kg and p < 0.01 at 15 μmol/kg, Fig. 3b), raising the hypothesis that longer treatment may be required to prime the synapses to function effectively. Subsequently, the duration of effect was explored with extended time-points to include training at 1.5 and 5 h after last dose of AZD1080 (4 μmol/kg). AZD1080 significantly prevented MK-801-induced impairments at both time-points (AZD1080 vs. MK-801, p < 0.01 at 1.5 h and p < 0.05 at 5 h, Fig. 3c), suggesting the effects are of long duration. The average concentration of AZD1080 measured in the brain at 3 h after acute administration (15 μmol/kg) was 36 ± 2 nM. In the subchronic experiments, the concentrations of AZD1080 in the brain were 19 ± 3 nM (4 μmol/kg) and 61 ± 10 nM (15 μmol/kg), all measured 3 h after last dose. The measured compound concentrations were around the in vitro IC50 of AZD1080 (31 nM) required for inhibition of GSK3β. No detectable concentrations of AZD1080 were observed at 5 h after last dose (4 μmol/kg), which is consistent with the PK/PD model in young rats suggesting a prolonged pharmacodynamic effect.
AZD1080 rescues dysfunctional synapses
To further explore the mechanism underlying the observed cognitive effects, we examined the potential of AZD1080 to have modulatory effects on both basal synaptic transmission and induction of LTP. The effect of GSK3 inhibition on synaptic transmission at Schaffer collateral synapses was tested in slices taken from mice dosed with AZD1080 (15 μmol/kg) for 3 days, that is, a dose corresponding to where pro-cognitive effects were seen in the CFC model. Acute NMDA receptor inhibition with the open channel blocker MK-801, when paired with a series of conditioning stimuli, produced a rapidly developing (< 30 min) impairment of LTP induction (Fig. 4a). The MTP had no significant (p = 0.48) effect on LTP induction (122 ± 6% of control amplitude, n = 16), when compared to slices that underwent the LTP induction protocol alone (116 ± 6% of control amplitude, n = 11). Acute administration of MK-801, when paired with the MTP protocol, was sufficient to impair Schaffer collateral LTP. Care was taken to achieve a MK-801 concentration that was the minimal sufficient to impair LTP induction. At a concentration of 300 nM, MK-801 did not disrupt LTP induction (120 ± 4% of control amplitude, n = 11), when compared with slices not treated with MK-801 (122 ± 6%, n = 16). However, 500 nM MK-801 produced a significant impairment of LTP induction when compared to vehicle controls (107 ± 3%, n = 14 vs. 122 ± 6%, n = 16 for controls, p < 0.05; Fig. 4b). A concentration of 500 nM was therefore used for all subsequent challenge experiments.
Subchronic in vivo exposure to AZD1080 alone had no significant effect on ex vivo LTP induction when compared to vehicle controls (AZD1080 vs. vehicle; Fig. 4b). Subchronic exposure to AZD1080 did, however, prevent disruption of LTP induction caused by an acute challenge with MK-801 (107 ± 3%, n = 14 MK-801 treated slices vs. 121 ± 5%, n = 18 slices treated with both MK-801 and AZD1080; Fig. 4b), raising the possibility that AZD1080 has a protective effect in dysfunctional systems. Taken together, the CFC and LTP experiments suggest that subchronic, but not acute, treatment with AZD1080 is able to prevent memory deficits in vivo, most likely by modulation of proteins downstream of GSK3β signaling pathways involved in synaptic function.
Peripheral target engagement in rat and in human
Given the overall drug like profile of the compound, the pre-clinical data and completion of additional toxicity studies, AZD1080 was progressed into Phase 1 clinical studies. We therefore determined the ability of the GSK3 inhibitor AZD1080 to engage the target both in rodents and in humans by measuring inhibition of glycogen synthase (GS) phosphorylation in blood lymphocytes as a peripheral biomarker. The effect on phosphorylated GS was related to the total levels of GS. First, the effect of AZD1080 on peripheral GSK3 inhibition was validated pre-clinically in dose- and time-response studies in adult rats. Acute oral treatment with AZD1080 at 1–10 μmol/kg produced a dose-dependent reduction of the phosphorylated to total GS ratio, as measured in PBMC isolated at different time-points after dosing. A mean maximal inhibitory effect of 49 ± 2% (compared to vehicle) was observed at the highest dose (10 μmol/kg) at 2 h after dosing (Fig. 5a). The effect remained at 6 h and was back to baseline by 24 h. Peak plasma concentrations of AZD1080 were obtained at 1 h at all doses tested (147 ± 4 nM at 1 μmol/kg; 150 ± 3 nM at 3 μmol/kg; 464 ± 21 nM at 10 μmol/kg).
The PK/PD profile of AZD1080 was also studied in healthy volunteers after oral multiple ascending doses. AZD1080 demonstrated an acceptable safety and tolerability profile in healthy middle-aged male and non-fertile female volunteers after administration for 14 days of up to 15 mg once daily. Steady-state exposure was reached within 5 days and the half-life of AZD1080 was approximately 20 h. There was limited accumulation of AZD1080 in plasma and there was no apparent effect of age on the pharmacokinetic profile. AZD1080 inhibited peripheral GSK3 activity, measured as the ratio of phosphorylated to total GS in lymphocytes on Day 1 (after first dose), with a mean maximal reduction of 23 ± 4% at 6 h after dosing with 15 mg AZD1080 (Fig. 5b). The effect was maintained at 10 h and returned to baseline by 24 h. Similar reductions in GS phosphorylation were observed on Day 8 (average 20% at 2–10 h after dose; data not shown). Maximal plasma concentrations in human were on average 86 ± 16 nM at 1 h following dosing with 15 mg AZD1080. Similar to the inhibition of tau phosphorylation in the hippocampus in young rats, there is a delay in the pharmacodynamic response measured in human PBMC. These results indicate that exposure of AZD1080 in humans at the doses tested have the ability to inhibit the GSK3 enzyme in the periphery. This is the first evidence reported of peripheral target engagement in humans for a selective GSK3 inhibitor.
Given the diverse roles of GSK3 in both brain pathological conditions and in dysfunctional neurons, it is important to identify a brain permeable GSK3 inhibitor that can be tested in the clinic for its therapeutic potential in the treatment of AD and related tauopathies. Using structure-based design, we have discovered a clinical candidate, AZD1080, which is a novel and potent small molecule, orally bioavailable and brain permeable selective GSK3 inhibitor. Based on the overall pharmacological profile reported in this manuscript and its drug like properties, AZD1080 was progressed into Phase 1 clinical trials in man.
The fundamental hypothesis in testing a GSK3 inhibitor in disease state resides on the fact that GSK3β abnormally phosphorylates tau, contributing to paired helical filament formation, as well as affecting other substrates (MAP1B, cAMP-response element binding protein, PS1, etc.) which can contribute to cytoskeletal architecture deficits, leading to neuritic dystrophy (Bhat and Budd 2002; Peineau et al. 2008; Kremer et al. 2011). Since active GSK3 triggers signal transduction events that participate in cell death and loss of neuronal and synaptic plasticity, part of the AD pathology could result from an abnormal increase in GSK3 expression and activity (Bhat and Budd 2002; Bhat et al. 2004). We have shown that AZD1080 specifically binds to and inhibits GSK3β within the ATP pocket of the catalytic domain. The crystal structure shows that the inhibitor binds through three hydrogen bonds, to the backbone atoms of Val-135 (both the amide N and the carbonyl O), a residue located in the hinge/linker region alongside of the ATP-binding pocket of the enzyme. The crystal structure of GSK3β in complex with AZD1080 reveals the interactions formed providing invaluable information for the design of future lines of GSK3 inhibitors.
The importance of AZD1080 as a specific GSK3 inhibitor was validated by its capacity to interfere with tau hyperphosphorylation, a pathological event believed to play an important role in AD (Ballatore et al. 2007). In cells and in rodent brain in vivo, AZD1080 is extremely efficacious at inhibiting tau phosphorylation, thereby addressing the fundamental hypothesis in a preclinical setting. AZD1080 inhibited tau phosphorylation in cells over-expressing tau protein in a dose-dependent manner. More importantly, AZD1080 is a brain permeable small molecule which has favorable oral bioavailability and pharmacokinetic profile. The PK/PD analysis suggests that peak exposure in the hippocampus is within 1 h while the effect on tau phosphorylation inhibition peaks at 6 h and the effect remain up to 24 h. This suggests that a shorter frequency of dosing regimen may be required in the clinic. The advantage of such a finding is that larger safety margins could potentially be derived. The reason for this prolonged effect is unclear but one can speculate that it is because of the tight regulation of phosphorylation and dephosphorylation events on the tau protein.
GSK3 inhibitors have also been reported to influence cognitive processes under certain conditions, specifically in impaired systems (Hooper et al. 2007; Onishi et al. 2011). We have used the CFC behavioral model to demonstrate that freezing is impaired by MK-801 in vehicle-dosed animals, but reverted to normal in mice sub-chronically dosed with AZD1080. Similarly, this study reports that AZD1080 prevents disruption of LTP induction caused by acute treatment with the NMDA receptor antagonist, MK-801. However, it must also be noted that while these experiments provide some mechanistic evidence that agrees with the hypothesis underlying the CFC model (i.e., that MK-801 interferes with learning by disrupting LTP), these experiments do not provide conclusive evidence that LTP preserved in the presence of MK-801 is a direct effect of GSK3 inhibition. The length of exposure to AZD1080 required before reversal of MK-801 inhibition might suggest an effect secondary or tertiary to GSK3 inhibition. These experiments also demonstrated that AZD1080 applied to slices obtained from non-compromised animals had no significant effect on LTP. Interestingly, in GSK3β over-expressing mice treatment with the non-specific GSK3 inhibitor lithium restored LTP, whereas lithium did not improve LTP in wild-type mice (Hooper et al. 2007). This suggests that GSK3 manipulations may produce more apparent results in a compromised system. This is further supported by findings that another GSK3 inhibitor (CT-99021) was able to reverse amyloid-beta-induced deficits in LTP (Jo et al. 2011). Based on this and our studies, it is tempting to speculate that GSK3 inhibition could play a pivotal role in attenuating the downstream detrimental effects of signaling pathways activated by multiple stimuli relevant to Alzheimer's disease.
Together with the finding from Hooper et al. (2007) which showed that LTP could still be induced using standard protocols in mice over-expressing GSK3β (albeit at lower levels compared to wild-type mice), the present results suggest that GSK3 is a modulator of LTP (i.e., useful), not a mediator of LTP (i.e., required) (Citri and Malenka 2008). In an intact system functioning as an integrated whole, the complex mechanisms that regulate LTP may be able to compensate for the loss of one modulatory element, whereas in an impaired system the role of a modulator may become evident. These studies suggest that AZD1080 reverses synaptic plasticity and functional deficits in a dysfynctional neuronal system and the efficacious effect is likely because of modification of pathways downstream of GSK3.
To our knowledge, the ability of a selective GSK3 inhibitor to engage the target both pre-clinically and in humans has not been previously reported. Although lithium has been evaluated for effects on GSK3β activity in human PBMC (Li et al. 2007) it should be noted that lithium is a non-specific GSK3β inhibitor (Kremer et al. 2011). To provide valuable information for dose setting and increase the confidence in translating the pre-clinical effects to humans, we have employed a GS activity assay in PBMC. We have shown that AZD1080 exposures that result in peripheral inhibition of GS activity in rodent PBMC correlate well with that observed in the Phase 1 multiple ascending dose study. These results indicate for the first time that a selective GSK3 inhibitor such as AZD1080 has the ability to inhibit the GSK3 enzyme in humans. Since sampling of cerebrospinal fluid was not performed in the Phase 1 study, it remains to be determined whether these exposures have the ability to inhibit tau phosphorylation in human brain. If these effects are translatable from rodent to humans, then AZD1080 should inhibit tau phosphorylation in the brain at the doses used in the Phase 1 multiple ascending dose study.
In summary, we have discovered a novel GSK3 inhibitor and demonstrated pharmacological validation by its capacity to interfere with tau phosphorylation and reverse synaptic plasticity deficits; events that are believed to play an important role in the pathogenesis of AD. We report for the first time the pharmacological profile of the clinical candidate AZD1080 and human peripheral target engagement in Phase 1 multiple ascending dose studies. Based on the pre-clinical profile, AZD1080 may have both disease-modifying and symptomatic potential in the treatment of AD and related tauopathies.
The authors thank Yafeng Xue and Mats Ormö for contributing to the crystallography experiments, Katherine Spence for the LTP studies, Angelica Hesselgren and Sara Selenius for the cognition studies and Gunilla Martinsson for the ELISA analyses.