A Small Molecule Mimetic of the Humanin Peptide as a Candidate for

Nov 21, 2017 - In F11 neurohybrid cells transfected with amyloid precursor protein (APP) with the familial AD mutant V642I-APP, there is significant n...
3 downloads 3 Views 3MB Size
Download PDF
Letter Cite This: ACS Chem. Neurosci. 2018, 9, 462−468

pubs.acs.org/chemneuro

A Small Molecule Mimetic of the Humanin Peptide as a Candidate for Modulating NMDA-Induced Neurotoxicity Mohammad Parvez Alam,† Tina Bilousova,† Patricia Spilman,† Kanagasabai Vadivel,‡ Dongsheng Bai,† Chris J. Elias,† Denis Evseenko,§ and Varghese John*,† †

Drug Discovery Laboratory, Department of Neurology, UCLA, Los Angeles, California 90095, United States Department of Orthopedic Surgery, DGSOM, UCLA, Los Angeles, California 90095, United States § Department of Orthopedic Surgery, University of Southern California (USC), Los Angeles, California 90033, United States ‡

S Supporting Information *

ABSTRACT: Humanin (HN), a 24-amino acid bioactive peptide, has been shown to increase cell survival of neurons after exposure to Aβ and NMDAinduced toxicity and thus could be beneficial in the treatment of Alzheimer’s disease (AD). The neuroprotection by HN is reported to be primarily through its agonist binding properties to the gp130 receptor. However, the peptidic nature of HN presents challenges in its development as a therapeutic for AD. We report here for the first time the elucidation of the binding site of Humanin (HN) peptide to the gp130 receptor extracellular domain through modeling and the synthesis of small molecule mimetics that interact with the HN binding site on the gp130 receptor and provide protection against NMDA-induced neurotoxicity in primary hippocampal neurons. A brain permeable small molecule mimetic was identified through exploratory medicinal chemistry using microfluidic flow chemistry to facilitate the synthesis of new analogues for screening and SAR optimization. KEYWORDS: Humanin, gp130 receptor, NMDA, excitotoxicity, primary neurons, microfluidics

A

transfected with amyloid precursor protein (APP) with the familial AD mutant V642I-APP, there is significant neuronal cell death in vitro. Treatment with HN protects against this cell death, but this protection is eradicated if the F11 cells are also treated with a gp130 neutralizing antibody. In addition, the study shows that activation of STAT3 is essential for the HN activity, thus treatment of the F11 cells with HN resulted in upregulation of the tyrosine phosphorylation of STAT3 (pTyr705-STAT3).7 Immunofluorescence (IF) labeling in aged Tg2576 AD model mice showed that pSTAT3 levels were significantly reduced in the dentate gyrus (DG) and CA1 brain regions of vehicle-treated mice while treatment intranasally with the HN derivative peptide Colivelin (CLN) restored pSTAT3 levels completely up to levels seen in nontransgenic mice.8 Furthermore, HNG has been shown to act through the gp130/IL6ST receptor complex to activate STAT3, AKT and ERK1/2 signaling pathways.9 These results support the reports which suggest that an increase in gp130 signaling may be of benefit in AD.10 Others have suggested that the anti-AD therapeutic benefits of HN could be through interaction with residues (17−28) of monomeric Aβ that

lzheimer’s disease (AD) is the most prevalent age-related dementia, and currently approved therapeutics provide only temporary symptomatic relief. Therefore, new approaches to therapeutic development are urgently needed. Neuronal cell death is a key feature of AD pathology, and factors leading to neuronal loss are many and not completely elucidated.1 In 2001, Hashimoto et al. identified Humanin (HN) a 24amino acid peptide from occipital lobe of post-mortem AD patient brain tissue and showed that it protected neurons from amyloid-beta (Aβ)-related toxicity.2 Muzumdar and co-workers reported a decrease in the endogenous HN plasma level with age, suggesting that decreased HN levels may be linked to cognitive decline during aging.3 The potent HN analogue HNG, with a substitution of serine-14 to glycine, showed neuroprotection against Aβ1−42-induced toxicity in primary neurons at nanomolar levels.2 HN exerts its function through binding to both extracellular receptors and intracellular binding partners.4 The proposed extracellular binding partners include G-protein coupled receptors, formyl peptide receptor like (FPRL)-1 and 2;5,6 and a trimeric receptor complex of gp130, ciliary neurotrophic factor receptor α (CNTFR), and the IL-27 receptor subunit, WSX-1.7 Hashimoto and co-workers reported that HN confers its protection against Aβ1−42 neurotoxicity through agonism of gp130 and that it is required for HN-mediated protection against Aβ1−42-induced death. In F11 neurohybrid cells © 2017 American Chemical Society

Received: September 8, 2017 Accepted: November 21, 2017 Published: November 21, 2017 462

DOI: 10.1021/acschemneuro.7b00350 ACS Chem. Neurosci. 2018, 9, 462−468

Letter

ACS Chemical Neuroscience

Figure 1. Predicted binding site for HN, compound 2 and SC144 in the extracellular domains D4-D6 of gp130. (A) HN, compound 2 and SC144 are predicted to bind to the interface region between extracellular domains 4 and 5 of gp130. The extracellular domains D4-D6 of gp130 and HN (green) are shown in the model. Compound 2 (magenta) and SC144 (cyan) are shown in stick representation. (B) Molecular surface representation of the gp130 extracellular domains 4−6 are shown with their electrostatic potential. (C, D) The gp130 residues involved in interaction with compound 2 (magenta) and HN (green) are shown, respectively. Carbon atoms in the D4 domain are colored light brown, while the carbon atoms in the D5 domain are shown in cyan. The residues within 5 Å radius of compound 2 or HN are shown in the models.

therapeutic development in neurodegenerative disorders such as AD.

results in prevention of Aβ oligomerization and provides neuroprotection.11,12 Recent studies have shown that HN can rescue neurons from NMDA-mediated toxicity in a manner similar to the NMDA receptor antagonist MK801, but the mechanism of the rescue by HN is not through antagonism of the NMDA receptor.13 Excitotoxicity through NMDA receptor overactivation is associated with disease progression in AD,14 therefore over the years there has been significant drug discovery efforts to identify new NMDA receptor antagonists for treatment of AD and other neurological disorders such as Parkinson’s disease, cerebral ischemia, and stroke wherein excitotoxicity is thought to play a role. However, such antagonists have been associated with significant adverse effects and neurotoxicity. A partial antagonist of the NMDA receptor called memantine (Namenda) was approved for AD as a symptomatic therapy over a decade ago but provides only limited benefit.15 We show here, for the first time, that HN acts to rescue the NMDA-mediated toxicity in primary hippocampal neurons through its interaction with the domains-4 and -5 (D4-D5) of the gp130 receptor. Our discovery was based on computational modeling of HN peptide and the extracellular domains of gp130. Interestingly, through screening and exploratory medicinal chemistry approaches we identified a small molecule series that also binds to gp130 receptor in the same binding site between domains D4-D5 as HN. The small molecule mimetics, like HN, rescue neurons from NMDA-mediated toxicity. Furthermore, our modeling analysis shows that a known antagonist, SC144,16 of gp130 also binds to the same HN binding site and reverses the neuroprotection conferred by these mimetics providing experimental evidence of their direct interaction with gp130 pathway. Thus, these small molecules are HN mimetics revealing a new mechanism for suppressing NMDA receptor overactivation and excitotoxicity for new

■

RESULTS AND DISCUSSION Our discovery efforts for small molecule HN mimetic began with a screening hit (RCGD)-423 (compound 1), which was found to be a regulator of cartilage growth and differentiation. This compound was identified by our collaborator at UCLA through an HTS effort using the 200,000 UCLA screening library, to find compounds that increase articular chondrocyte activation and facilitate joint repair. Compound 1 was shown to bind gp130 initiating a signaling cascade comprising activation of the JAK-STAT3 pathway, phosphorylation of STAT3 (pSTAT3), and ultimately regulation of gene expression in chondrocytes (Shkhyan et al. submitted manuscript) and hair follicle stem cells.17 Given that HN has been shown to elicit its neuroprotective effects through gp130 receptor agonism,7 we wanted to use compound 1 as a starting lead to conduct exploratory medicinal chemistry to find gp130 interacting HN mimetics. We rationally designed and synthesized compounds 2−8 and evaluated their protective effects against NMDAinduced neurotoxicity. The ClusPro docking server was used to model the interactions between HN and the extracellular domains of gp130;18 this also revealed the HN binding domains D4-D5 as a potential high-affinity binding sites for the HN peptide in the gp130 extracellular region (Figure 1). In addition, for compounds 1−8 we performed modeling studies using the Swissdock program; these again revealed domains D4-D5 as a potential high-affinity binding site in the gp130 extracellular region (Figure 1, Supporting Information, Figures S1−S7). The binding free energy of HN and compound 2 to gp130 D4D6 were determined to be −46.99 (±9.65) and −18.61 (±2.30) kcal/mol, respectively using the MM/PBSA method implemented in AMBER (see the Supporting Information). 463

DOI: 10.1021/acschemneuro.7b00350 ACS Chem. Neurosci. 2018, 9, 462−468

Letter

ACS Chemical Neuroscience

Table 1. Schematic for the Microfluidic Reactor Syntheses of Compounds 1−8 and Effect of Small Molecule HN Mimetics on NMDA-Triggered LDH release in Primary Hippocampal Neurons

reactor without the use of a catalyst which is unprecedented (Table 1). We use a flow-chemistry approach that we have previously described,22 for our synthesis. Microfluidic flow chemistry facilitated rapid syntheses of the new analogues. This is a green chemistry approach with a cleaner reaction profile, smaller carbon footprint, higher yields, and quick scalability with minimum reaction optimization enabling us to quickly gain SAR insights into the optimal gp130 binding molecules for neuroprotection. The reactions are rapid, proceeding to completion in a few minutes and products are obtained in good yields, in milligram quantities. As a result of these efforts, we report the synthesis of new chemical entities (NCEs) such as N-(2,4-difluorophenyl)-4-(pyridin-3-yl)thiazol-2-amine (compound 6). With a number of analogues in hand, we next determined the ability of the small molecules to confer neuroprotection against NMDA-induced toxicity (Table 1). In this study, the cytotoxicity of NMDA was quantified by measuring the activity of lactate dehydrogenase (LDH) released from primary cultured hippocampal neurons into the media and confirmed by living cell density analysis using calcein staining. Our results showed that NMDA treatment (100 μM) for 2.5 h caused cytotoxicity in the primary neurons in vitro and

Compound 2 remained bound to the gp130 D4D6 domain during the 50 ns molecular dynamic simulations similar to that seen with the larger 24 amino acid peptide HN and which has more interactions to the receptor due to its relatively larger size than 2. For the small molecule mimetic 2 this implies good affinity of compound 2 for the gp130 receptor at the D4-D5 domain; the calculated binding energy for compound 2 is ∼2.5fold lower than that of HN. The known gp130 antagonist, SC144, was also modeled in parallel to investigate its binding specificity (Supporting Information, Figure S8). SC144 was found to have high-affinity binding for the same site in the D4D5 domains. Thus, modeling reveals that the compounds 1−8, the gp130 antagonist SC144 and HN all bound to domains (D4-D5) of the gp130 receptor. Compound 1 and analogues belong to the 2-aminothiazoles family of compounds. While many methods have been reported to synthesize 2-aminothiazoles, including solution, solid phase synthesis, a tandem one-pot aqueous phase synthesis,19 and microwave promoted synthesis,20 our method for 2-aminothiazoles preparation involves Hantzsch condensation21 using 2-bromo-1-phenylethan-1-one (and its derivatives) and 1-(4bromophenyl)thiourea (and its derivatives) in a microfluidic 464

DOI: 10.1021/acschemneuro.7b00350 ACS Chem. Neurosci. 2018, 9, 462−468

Letter

ACS Chemical Neuroscience

Figure 2. (A) Effect of cotreatment of gp130 antagonist, SC144, MK801 and compounds 2 and 8 on NMDA-triggered LDH release in cultured hippocampal neurons. NMDA increased LDH release (bar 2), and both MK801 and compound 2 rescued the LDH increase (bars 3 and 4, respectively). Antagonist SC144 prevented compound 2 rescue (bar 5). Compound 8 also rescued LDH and SC144 prevented this rescue (bars 6 and 7, respectively). Each bar represents mean ± SEM of four independent observations. Statistical significance was at P < 0.05. Effect of compound 2 on density of hippocampal neurons under NMDA-induced toxicity. Fluorescent microscopy was used to show representative calcein staining, and cellular density of each group. (B) Control (DMSO-only) density is shown. (C) A clear reduction of neuronal density is seen as a result of NMDAinduced toxicity. (D) With MK801 pretreatment, the density of NMDA-treated neurons was similar to the DMSO-only control. (E) Pretreatment with compound 2 also prevented NMDA-induced toxicity and the density of living neurons was similar to DMSO-only control.

triggered release of LDH (Table 1 and Figure 2A). The LDH level in the NMDA treatment group was about 26% higher than that in the DMSO control. MK801, a noncompetitive antagonist of NMDA receptor, was used as a positive control in these studies.23 As shown in Table 1 and Figure 2A, the addition of MK801 significantly reduced, by 43%, the release of LDH from NMDA-treated neurons. We evaluated different batches of HN and HNG from different vendors in our assay, and only one of them (Tocris Biosciences, cat# 5154, SI) was active and exhibited protection against NMDA-induced neurotoxicity (Figure S9). Neither HN nor the compounds by itself had any effect on LDH levels from neurons which were not treated with NMDA. However, the release of LDH was significantly inhibited in NMDA-treated neurons when treated with the HN mimetics. Neurons were pretreated with compounds for 16 h before exposure to NMDA. The 2-aminothiazole derivatives showed varied neuroprotective activity depending on the aryl substitution on either side of the central aminothiazole core. As shown in Table 1, the levels of NMDA induced LDH release was inhibited by 20% upon treatment with our initial screening hit 1 while greater inhibition of the LDH release was seen with the HN mimetic analogues 2−8 at 10 μM. The levels of NMDA induced LDH release was inhibited by 21% on treatment with compound 3, the pyridine analogue of compound 2. While compound 4 decreased the LDH levels by 32%, compound 5 with two fluoro groups at the ortho and para position inhibited LDH release by 43%, similar to the activity of MK801. Compound 6 is the pyridine analogue of compound 5, and it inhibited NMDA induced LDH release similar to what observed for compound 3. Compound 7 with the phenethyl group showed similar inhibition as compound 3 and compound 6. Introduction of fluoro group at the para position in both the rings, compound 8, showed slightly worse inhibition of LDH release as compared to when the fluoro group was just present only in one ring.

Compound 2 was one of the best, and its neuroprotective activity was similar to that of MK801. The known gp130 antagonist, SC144, treatment reversed the neuroprotection provided by compounds 2 or 8 against NMDA-induced neurotoxicity (Figure 2A). Compound 2 was assessed for its downstream signaling based on its effects on STAT3, AKT and ERK activation as previously reported upon HNG treatment in HEK293 cells, SH-SY5Y cells and primary hippocampal neurons.9 For our experiments we used the potent HN analogue HNG1724 in SHSY5Y cells and primary hippocampal neurons. Treatment of the cells with HNG17 and compound 2 (10 μM) in serum free media were performed for varying time periods as indicated in Figure S11. In SH-SY5Y cells, compound 2 treatment showed a 2-fold increase (as measured using the ratio of densitometry values for the p-STAT3 and STAT3 bands) in phosphorylation of STAT3 within 10 min at its regulatory Tyr705 site (Figure S11A); this is similar to that seen with HNG17 at 5 min. However, in primary hippocampal neuronal cultures, the pSTAT3 levels were below levels of detection both for HNG17 and compound 2 at all time points (Figure S11B); this could be due lower cross reactivity of the p-STAT3 antibody for the mouse protein. Compound 2 treatment showed a 9-fold increase in phosphorylation of AKT at its regulatory Thr308 site in the serum free media condition in SH-SY5Y cells (Figure S11A) after 15 min incubation and a 4-fold increase in primary cortical neurons for the same time (Figure S11B). These increases in p-AKT were greater than that seen with HNG17 in these cells. Compound 2 treatment caused a rapid 2-fold increase in phosphorylation of ERK1/2 at its regulatory Thr202/ Tyr204 site in the serum free media condition in SH-SY5Y cells (Figure S11A) and 1.2-fold in primary cortical neurons (Figure S11B); similar increases were seen with HNG17 in the SHSY5Y cells and less so in the primary neuronal cells. In SHSY5Y cells, cotreatment of compound 2 with gp130 antagonist, 465

DOI: 10.1021/acschemneuro.7b00350 ACS Chem. Neurosci. 2018, 9, 462−468

Letter

ACS Chemical Neuroscience

∼7.5:1 for 10 mg/kg SQ injection. Based on the good brain permeability of compound 2, it is a promising lead candidate to conduct proof-of-concept testing in a mouse model of NMDAinduced toxicity in future studies. In the present study, we elucidate using modeling, the binding site of HN in the extracellular domains of the receptor gp130. Our results show that HN and small molecule mimetics bind to the same site at the interface of the D4-D5 domain and provide neuroprotection from NMDA neurotoxicity in primary hippocampal neurons. The small molecule HN mimetics hold great promise toward developing an orally available bioactive small molecule that confers the benefits of the humanin peptide. HN has shown to have multiple beneficial effects in neurodegeneration, diabetes, and artherosclerosis, but due to its peptidic structure, it presents challenges in its development as a therapeutic.25−27 Small molecule mimetics of HN are attractive agents for therapeutic development both in AD and other neurological disorders where HN has shown beneficial effects such as in traumatic brain injury (TBI),28 stroke,29 Aβ-induced cerebrovascular dementia,30 and amyotrophic lateral sclerosis (ALS).4 Humanin has a number of different cytoprotective and metaboloprotective effects such as cardioprotection against cellular apoptosis during myocardial injury,31 as well as modulation of systemic insulin sensitivity and glucose homeostasis.32 Given the good oral brain permeability of the HN mimetic compound 2, it represents a promising molecular tool for in vivo proof-of-concept testing as a neuroprotective agent in NMDA and Aβ-induced toxicity models. Further efforts will be made to explore SAR and expand our library of small molecule HN mimetics, while evaluating the underlying gp130 signaling mechanism. These agents could lead to a new pharmacological class of therapeutic agents for AD.

SC144, inhibited phosphorylation of STAT3, AKT and ERK1/ 2, this effect is similar to what was reported on treatment with the HN analogue HNG treatment and with the anti-gp130 antibody or the gp130 antagonist SC144.9 This data supports the downstream signaling effects of compound 2 primarily through its interaction with the gp130 receptor. Furthermore, cotreatment with compound 2 and AG490, a JAK inhibitor, greatly reduced the AKT phosphorylation indicating compound 2 acts upstream to JAK. In contrast, ERK1/2 phosphorylation did not change much, as there are alternate pathways involved besides through the trimeric gp130 complex involving ciliary neurotrophic factor receptor α (CNTFR), and the IL-27 receptor subunit, WSX-1, for activation of the ERK pathway.9 These data support compound 2 being a HN mimetic at the gp130 receptor. Calcein (calcein-acetoxymethyl ester; calcein-AM), a highly lipophilic nonfluorescent and cell permeable compound, staining was used to evaluate the density of living primary neurons. Calcein-AM can be converted by intracellular esterases into calcein, an anionic fluorescent form, only labeling viable cells, and thus provides both morphological and functional information about living neuronal cells. NMDA (100 μM, 2.5 h) alone induced a significant decrease in neuronal survival, with more than 50% cell death observed. With MK801 pretreatment, the density of NMDA-treated neurons was rescued to levels better than DMSO-only control, this indicates that the toxicity was induced by NMDA receptor activation (Figures 2B−D and S10). Furthermore, there was no difference in density of living neurons between the compound 2 group and the DMSO-only control group, suggesting that compound 2 could also reduce NMDA-induced toxicity similar to the direct antagonist, MK801. The neuroprotective effects of compound 2 make it a promising candidate for further testing in AD models. Since any potential therapeutic for AD needs to cross the blood-brain barrier (BBB), we further evaluated the brain permeability of compound 2 in mice. Mice were dosed orally at 10 or 30 mg/ kg, or injected subcutaneously (SQ) at 10 mg/kg, and euthanized after 1, 2, 4, 6, and 8 h post dose. We found that at 2 h after SQ delivery at 10 mg/kg the brain Cmax was 161 ng/ g while dosing at 30 mg/kg orally, however, resulted in the brain Cmax of 156 ng/g (0.57 μM) as shown in Figure 3. The brain to plasma ratio for 2 was ∼4:1 for oral 30 mg/kg and

■

METHODS

■

ASSOCIATED CONTENT

General Procedure for Synthesis. The chemical synthesis of compound 2 began from commercially available 2-bromo-1-phenylethan-1-one and 1-(4-fluorophenyl)thiourea. They were premixed in a separate vial and were pumped through a preheated glass microfluidic reactor maintained at 80 °C and 2 bar pressure (Syrris Asia Flow Chemistry Module) at a 500 μL/min flow rate with a one minute residence time in the reactor using Syrris Asia pumps to afford compound 2 from the output stream. Other compounds were synthesized (Table 1) by similarly pumping corresponding starting material through the preheated glass microfluidic reactor. The general procedure for cell testing and pharmacokinetic analysis is provided in the Supporting Information. Statistical Analysis. All the data were expressed as the mean ± SEM. Significant differences were determined by Student’s t tests, and one-way ANOVA followed by Dunnett’s multiple comparisons test using GraphPad Prism 7 software. Values of #