Identification of Novel Allosteric Modulators of Glutamate Transporter

Nov 15, 2017 - Further characterization of the two best ranking EAAT2 activators for efficacy, potency, and selectivity for glutamate over monoamine t...
1 downloads 24 Views 2MB Size
Subscriber access provided by READING UNIV

Article

Identification of novel allosteric modulators of glutamate transporter EAAT2 Sandhya Kortagere, Ole Valente Mortensen, Jingsheng Xia, William Lester, Yuhong Fang, Yellamelli V. V. Srikanth, Joseph M. Salvino, and Andreia Cristina Karklin Fontana ACS Chem. Neurosci., Just Accepted Manuscript • DOI: 10.1021/acschemneuro.7b00308 • Publication Date (Web): 15 Nov 2017 Downloaded from http://pubs.acs.org on November 16, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Chemical Neuroscience is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

Identification of novel allosteric modulators of glutamate transporter EAAT2

Sandhya Kortagere†, Ole V. Mortensen‡, Jingsheng Xia‡, William Lester&, Yuhong Fang&, Yellamelli Srikanth‡, Joseph M. Salvino‡ and Andréia C. K. Fontana‡* † Department of Microbiology and Immunology, Centers for Molecular Parasitology, Virology and Translational Neuroscience, Institute for Molecular Medicine, Drexel University College of Medicine, Philadelphia, PA 19129, United States ‡

Department of Pharmacology and Physiology, Drexel University College of Medicine,

Philadelphia, PA, 19102, United States &

Analytical Chemistry, Division of Pre-Clinical Innovation (DPI), NCATS, National

Institute of Health, Rockville, MD 20850, United States

Corresponding author: *Andréia C. K. Fontana Department of Pharmacology and Physiology, Drexel University College of Medicine, Philadelphia, PA, 19102, United States Tel: +1 215-762-4399. E-mail: [email protected] ORCID: 0000-0002-4791-8746

1 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 61

ABSTRACT Dysfunction of excitatory amino acid transporters (EAATs) has been implicated in the pathogenesis of various neurological disorders, such as stroke, brain trauma, epilepsyand neurodegenerative diseases among others. EAAT2 is the main subtype responsible for glutamate clearance in the brain, having a key role in regulating transmission and preventing excitotoxicity. Therefore, compounds that increase the expression or activity of EAAT2 have therapeutic potential for neuroprotection. Previous studies identified molecular determinants for EAAT2 transport stimulation in a structural domain that lies at the interface of the rigid trimerization domain and the central substrate binding transport domain. In this work, a hybrid structure based approach was applied, based on this molecular domain, to create a high-resolution pharmacophore. Subsequently, a virtual screening of a library of small molecules was performed; identifying ten hit molecules that interact at the proposed domain. Among these, three compounds were determined to be activators, four were inhibitors and three had no effect on EAAT2-mediated transport in vitro. Further characterization of the two best ranking EAAT2 activators for efficacy, potency and selectivity for glutamate over monoamine transporters subtypes and NMDA receptors, and efficacy in cultured astrocytes is demonstrated. Mutagenesis studies suggest that the EAAT2 activators interact with residues forming the interface between the trimerization and the transport domains. These compounds enhance the glutamate translocation rate, with no effect on substrate interaction, suggesting an allosteric mechanism. The identification of these novel positive allosteric modulators of EAAT2 offer an innovative approach for the development of therapies based on glutamate transport enhancement.

2 ACS Paragon Plus Environment

Page 3 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

Keywords: EAAT2, EAAT2 activators, glutamate transporter, glutamate uptake, hybrid structure based method, positive allosteric modulation, transport enhancement.

3 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 61

INTRODUCTION Glutamate is the major excitatory amino acid neurotransmitter in the mammalian central nervous system (CNS) and is critical for normal brain function including cognition, memory, learning, developmental plasticity, and long-term potentiation 1-3. The termination of glutamate neurotransmission is achieved by rapid uptake of the released glutamate by presynaptic and astrocytic sodium-dependent transporters 46

. This is a process that involves co-transport of glutamate and three sodium ions,

followed by the counter transport of potassium 7-9. Five structurally distinct subtypes of Na+-dependent glutamate transporters have been cloned to date, including (rat/human homolog): GLAST/EAAT1, GLT-1/EAAT2, EAAC1/EAAT3, EAAT4 and EAAT5 4. The most abundant subtype EAAT2 /GLT-1 10 is expressed throughout the brain, in the spinal cord, primarily in astrocytes, and also in neurons and oligodendrocytes, and accounts for approximately 95 % of the total glutamate transport activity and 1 % of total brain protein in the CNS 11-15. Consequently, it plays a central role in the maintenance of extracellular glutamate homeostasis and preventing excitotoxicity 16-19. The failure of proper removal of glutamate by astrocytes from the synapses results in a pathological process termed glutamate excitotoxicity. Excitotoxicity is characterized by a sustained elevation of extracellular glutamate levels and excessive activation of postsynaptic glutamate receptors resulting in Ca2+ influx 20 and activation of a cascade of phospholipases, endonucleases, and proteases such as calpain that can lead to apoptotic or necrotic cell death 21. Excitotoxicity plays a central role in secondary damage following acute pathologies, such as traumatic brain injury (TBI) 22-28, hyperexcitability, seizures and epilepsy 29-32, stroke 17, 33-36, cerebral and retinal ischemia 37, 38;

4 ACS Paragon Plus Environment

Page 5 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

and chronic pathologies, such as Amyotrophic lateral sclerosis (ALS) 39, 40, Alzheimer’s disease 41, Huntington’s disease 42, 43, neuropathic pain 44 and HIV-associated neurocognitive disorder (HAND) 45, 46. Also, ischemic events in humans and in animals lead to an acute and sustained increase in extracellular glutamate concentrations 23, 24, 47

which suggest a lack of proper clearance by glutamate transporters. In fact,

dysfunctional glutamate transporters are often the initiating event or part of the cascade leading to brain injury 17, 19. Reductions in EAAT2 activity result in increased predisposition for seizures and susceptibility to damage due to ischemia 12, 48 49. Knockout of the EAAT2 gene resulted in exacerbated damage compared to their wildtype counterparts following cerebral injury in mice 12, and controlled cortical impact in rats 50. Additionally, a transgenic approach for EAAT2 overexpression with double transgenic mice created from crossing an ALS mouse model to a mouse model overexpressing EAAT2 resulted in animals that displays delayed grip strength decline, motor neuron loss and increased life expectancy, further validating this target 51. Overall, these studies emphasize the therapeutic relevance of genetic and pharmacological regulation of glutamate transporter function to control glutamate levels and excitotoxicity in several neurological conditions/diseases. Pharmacological manipulation of EAAT2 expression or function is of significant interest from a therapeutic point of view, and it is a well-established target for the development of therapeutic agents to prevent excitotoxicity52-57. In this sense, several transcriptional or translational up regulators of EAAT2, such as GPI-104658, ceftriaxone59-65, harmine66 and pyridazine derivatives67, 68, display neuroprotective effects based on their ability to

5 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 61

prevent glutamate excitotoxicity and inhibit neuronal cell death. This is a promising approach but more relevant for chronic conditions or prophylactic use 69. Positive allosteric modulation (PAM) of EAAT2/GLT-1 with pharmacological compounds represents a relatively novel and desirable approach to the treatment of conditions/diseases involving glutamate excitotoxicity. We hypothesize this will enable acute interventions and a potentially much safer and effective therapy, not likely to produce a severe side effect profile 70. In this regard, a compound purified from Parawixia bistriata spider venom, referred to as Parawixin1, stimulates glutamate uptake in rat synaptosomes, protects retinal tissue from ischemic damage 71, and acts directly and selectively on the glutamate transporter EAAT2 53. Thus, this spider venom provides proof of principle for the development of drugs that act through direct stimulation of EAAT2 function. In a previous study, this venom was employed as a tool to further understand structural and mechanistic aspects of the transporter enhancement 72. Furthermore, the elucidation of several crystal structures of a bacterial homologue, GltPh, have greatly facilitated our understanding of the mechanistic aspects of the transport cycle 73-75. The three-dimensional protein structures revealed that the transporter can be divided into two domains. The C-terminal part forms a transport domain containing the substrate binding sites formed by two hairpin loops, and surrounding the central core transport domain is parts of the N-terminal part of the protein, forming a peripheral rigid scaffold, designated the trimerization domain. This trimerization domain provides the interaction interface between the three subunits of the trimer and is believed to facilitate the elevator-like movement of the inner transport domain 75 (Figure 1). Structural elements that are critical for transport stimulation were

6 ACS Paragon Plus Environment

Page 7 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

resolved and determined to involve residues residing at the interface between the transport and the trimerization domains 72. In the present work, this knowledge derived from the structural elements was employed in a hybrid structure based (HSB) approach, using the crystal structure of the homolog transporter GltPh. A virtual screening identified ten novel small molecules that interact within this unique region of the EAAT2. These compounds were characterized for their selectivity to EAAT2 and their ability to allosterically activate or inhibit the function of EAAT2.

RESULTS AND DISCUSSION Previous studies have discovered a natural compound that enhanced the function of EAAT2 and it had neuroprotective properties53, 71. The molecular determinants of the transport enhancement in EAAT2 were determined with a mutagenesis guided study72. In the present work, based on this unique structural information, we successfully used the HSB approach to identify novel compounds that interact with this region.

In silico screening was successful at identifying novel EAAT2 allosteric modulators Molecular dynamics simulations of the modeled EAAT2 in 1-palmitoyl-2-oleoylsn-glycero-3-phosphocholine (POPC) membrane suggested the presence of an allosteric pocket in the structural region identified in the previous study72 that is located at a structural interface between the transport and the surrounding trimerization domain

7 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 61

(Figure 2A), a region that is important for facilitating conformational changes during the translocation cycle. This region is distal from the central substrate binding region formed by Hairpins 1 and 274, suggesting that compounds interacting at this site will have an allosteric mode of action. From the molecular dynamics simulation, we identified five residues namely, M86, D485, W472, K299 and L295 that lined the allosteric pocket. Using these five residues we designed a five point three dimensional “receptor pharmacophore” (Figure 2B) for virtual screening of our assembled libraries. The virtual screening approach identified ten molecules that were evaluated in dose-response assays for glutamate uptake in COS-7 cells expressing EAAT2 (Table 1). A scifinder substructure search confirmed the novelty of the compound series. A broader substructure search also confirmed the lack of any obvious potent off-target biological activity for this series. Among the ten molecules, three compounds (GT949, GT951 and GT939) were found to be PAMs of EAAT2, while four molecules (GT729, GT835, GT938 and GT922) were inhibitors (negative allosteric modulators, NAMs) and three molecules (GT867, GT988 and GT996) had no effect. In this study we have focused on the further characterization of compounds GT949 and GT951, as proof of concept to demonstrate and characterize positive allosteric modulation of EAAT2. It is intriguing to note that the NAMs shown in table 1 are significantly less potent than the PAMs. We speculate this could be a result of positive cooperativity between the substrate glutamate and the PAMs for activating uptake. In studies of allosteric potentiation of GPCRs, where this concept is much more developed and strong tools exists, a possibly related phenomenon has been described76. For some allosteric modulators of GPCRs it was found that allosteric compounds have different affinities for 8 ACS Paragon Plus Environment

Page 9 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

the active and inactive states of the receptor. In alignment with this idea, we speculate that both PAMs and NAMs of EAAT2 are of relatively low affinity but in the presence of the substrate glutamate the affinity of PAMs is significantly increased to enable uptake augmentation. This cooperativity is different, or maybe absent for NAMs, and consequently results in much lower affinity for inhibiting uptake. If this mechanism is true for the EAAT2 PAMs, the affinity of the PAMs should be much lower in the absence of glutamate (corresponding to the inactive state), and enhanced by glutamate (corresponding to the active state) cooperatively. Unfortunately, a radioligand is not currently available to perform assays for measuring the binding affinity of the PAMs in presence and absence of glutamate to confirm this hypothesis. Recently, the crystal structure of the human EAAT1 was solved in complex with substrate L-Aspartate, competitive inhibitor TBOA and allosteric modulator UCPH-10177. A structural superposition of our modeled EAAT2 onto this crystal structure of EAAT1 suggested that although both these proteins share the GltPh-like domain architecture, they have significant differences in the arrangement of the helices (Figure 3) due to large insertions or deletions in EAAT1. Interestingly, UCPH-101 binds towards the inner opening of the membrane and at a site that is distinctly different from the allosteric site that bind PAMs such as GT949 which is towards the outer opening of the membrane and there are no shared residues between the two sites. Both these allosteric sites are proximal to the inhibitor binding site but on diagonally opposite ends and may exist simultaneously in EAAT2 but the UCPH-101 crevice in EAAT2 may be smaller than that observed in EAAT1. The substrate binding pocket is structurally conserved between EAAT1 and EAAT2.

9 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 61

Novel compounds stimulate EAAT2-mediated glutamate transport with high potency and selectivity for EAAT2, with no effects on EAAT1 or EAAT3 We have focused on the series exemplified by the potent hit compounds GT949 and GT951, which were evaluated in a dose-response assays on glutamate uptake in COS-7 cells transfected with EAAT1, 2 or 3. In this assay, GT949 and GT951 enhanced glutamate transport with EC50 values of 0.26 ± 0.03 nM and 0.8 ± 0.3 nM respectively (Figure 4). The increase in rate acceleration for glutamate removal by EAAT2 was ~70% for both compounds, which is comparable to what it was previously observed for Parawixin1 ~70% 53, 72. GT949 and GT951 also demonstrated selectivity to EAAT2 and had no effect on glutamate activity mediated by EAAT1 or EAAT3 (Figure 3).

Compounds are non-competitive positive allosteric stimulators of EAAT2 GT949 and GT951 were also tested for their effect on glutamate uptake kinetics in EAAT2 transfected cells. Both compounds enhanced glutamate transport in a noncompetitive fashion, with an increase in Vmax of about 47% and 75% for GT949 and GT951, respectively (Figure 5). KM was not statistically different between the groups (One-way ANOVA followed by Dunnett's post-hoc test comparing to vehicle). This strongly indicates that they work through positive allosteric modulation.

10 ACS Paragon Plus Environment

Page 11 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

Potency of the enantiomers is remarkably different The docking studies suggested differences in potency and binding between the R and S enantiomers of GT949. Since GT949 was initially tested as a racemic mixture of the enantiomers, GT949 was subject to chiral separation (see Supplemental Materials). Here we show that the GT949A enantiomer is 22- fold more potent than GT949B enantiomer and 6- fold more potent than the racemic mixture (Figure 6). This provides compelling evidence that these molecules are interacting in the chiral environment of the allosteric binding site.

Compounds stimulate glutamate transport in cultured astrocytes GT949 and GT951 were tested for effects on glutamate uptake mediated by cultured astrocytes (Figure 7). Based on our initial glutamate uptake assay in COS-7 cells GT996 was chosen as a negative control. Raw data (DPM) demonstrated that the background, obtained in the presence of TBOA, represented ~4.7% of the specific signal (Figure 7A). Dose response curves of GT949 (racemic mixture and separated enantiomers, figure 7B) show that potencies vary from 1 ± 0.07 nM (racemic mixture), 0.5 ± 0.04 nM (enantiomer A) and 15 ± 1.3 nM (enantiomer B). The efficacy of augmentation was ~58 and 50% for the racemic mixture and B enantiomer, while the A enantiomer was more efficacious, resulting in an 81% increase in transport. For GT951, we observed an EC50 value of 0.3 ± 0.2 nM and efficacy of uptake enhancement of ~27% (Figure 6C). Finally, we confirmed that GT996 was inactive at modulating glutamate transport in cultured astrocytes (Figure 6D). These data confirms that the compounds are active in native environment of EAAT2, cultured glia and astrocytes. 11 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 61

We observed a difference in potencies of the compounds between the overexpressing heterologous cell lines and the primary glial cultures. The racemates and the separated enantiomers are much less potent in glial cells than in COS-7 cells, although maintaining the trend that the A enantiomer is more potent than B. Differences in potencies can be related to cell-specific factors such as post-translational modifications of EAAT2 that impact the structure of the allosteric site and/or differences in cellular proteins that interact with and regulate EAAT2. To understand the contribution of the various EAAT subtypes to glutamate uptake in cultured glial cells, we performed uptake inhibition assay with selective inhibitors of the various EAATs (supplemental Figure 1). The data suggests that glutamate uptake in these cultures is mediated at least in part by EAAT2 (shown by partial inhibition by WAY 213613, a selective EAAT2 inhibitor), but also EAAT1 (shown by partial inhibition by UCPH-101, a selective EAAT1 inhibitor) and possibly EAAT3 (since not all of the uptake was inhibited by the combination of EAAT1 and 2 inhibitors). Therefore, we conclude that EAAT2 transporter mediates glutamate uptake in our astrocyte cultures, in addition to other EAAT transporters. We also performed immunoblotting to further support the pharmacological studies and the results validated the presence of GLT-1/EAAT2 in the astrocyte cultures (supplemental Figure 2). In these assays, we also demonstrated the presence of neurons as indicated by the microtubule associated protein MAP-2 marker. According to previous studies, pure astrocyte cultures express only GLAST 78. However, transporter activity can be regulated in different ways, including gene expression, transporter protein targeting and trafficking, and posttranslational modifications of the transporter 12 ACS Paragon Plus Environment

Page 13 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

protein. In this regard, some studies have suggested that added glutamate and/or growth factors can induce expression of astroglial glutamate transporters GLT1 and GLAST 79-81, and others suggest that, in co-cultures of astrocytes with neurons, expression of GLT-1 is induced in astrocytes, while expression of GLAST is slightly augmented 82, 83. Therefore, our glia culture also contains neurons, which could be responsible for inducing the expression of EAAT2.

Compounds have low affinity inhibitory or no effect on monoamine transporters and do not modulate the activity of NMDA receptor in primary cultures Dose response curves of GT949 and GT951 demonstrated lack of modulation of the activity of human serotonin (hSERT), noradrenaline (hNET) and dopamine (hDAT) transporters at highest concentration tested (IC50> 1mM, not shown). Application of 100 µM NMDA with 3 µM glycine in the absence of Mg2+ in neuronal cultures induced a sustained calcium response. This response was drastically attenuated by 40 µM AP-V, a specific NMDA receptor antagonist, suggesting that NMDA-induced calcium response was mediated by NMDA receptors. Vehicle, GT949, or GT951 had no effects on NMDA-induced calcium response, indicating that the compounds do not modulate the activity of NMDA receptors (Figure 8). The lack of inhibition of NMDA receptors suggests that these compounds may be devoid of the serious side effects that have limited the translation of EAAT2 activators to the clinic 84, 85. Taken together, these initial selectivity evaluations indicate that the compounds are selective for glutamate transporter EAAT2 over transporters EAAT1 and EAAT3 and monoamine transporters, and do not affect NMDA receptor activity. 13 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 61

Structure function studies: novel compounds GT949 and GT951 engage EAAT2 in a similar fashion as Parawixin1 To provide additional evidence that Parawixin1 and the novel small molecule hits (GT949 and GT951 compounds) are engaging the transporter in a similar fashion, dose response curve of the compounds were performed on eight different EAAT2 mutants: H71S, M86V, L290S, L295A, G298A, K299A, S465L and W472I (Table 2). These mutants were selected by their role in transport-enhancement elicited by the venom and their role in the pharmacophore design in the HSB screening. These structure function studies of structural determinants of the binding pocket were also guided by the molecular modeling and docking studies shown in Figure 9. Specifically, changes in potencies for GT949 on the EAAT2 mutants suggest that residues in TM2, TM5, and TM8 in the putative activation domain (Figure 9A) are important for transport enhancement and that residues M86 (TM2), L295 (TM5), S465 (TM8) and W472 (TM8) are critical for GT949 activity. Molecular docking of GT949 to EAAT2 supports these studies as M86, L295, S465, and W472 all contribute directly to the binding of GT949 (Figure 9B). Mutations such as H71S, L290S, and G298A of residues that are not found to be in direct contact with GT949 in the docking studies also does not significantly affect GT949 modulation of EAAT2 in functional studies. Similar to GT949, docking studies suggest the binding affinity of GT951 in the pocket could be influenced by direct interactions with M86, L295, K299 (TM5), S465, and W472 (Figure 9C). Site directed mutagenesis studies confirmed that potencies for GT951 was indeed affected by mutations of all these residues but not by mutations like H71S, L290S, and G298A. 14 ACS Paragon Plus Environment

Page 15 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

Some differences are to be noted between our findings regarding the novel compounds and the venom. In the studies of the venom, residue S465 was not found to be important for its action. This was taken as evidence that the venom was interacting with an outward facing conformation when residue W472 is in closer vicinity of the other critical residues compared to S465. Here we find both S465 and W472 to be important for the action of both GT949 and GT951. This could suggest that the novel compounds also interact with the inward facing conformation, however, as visualized by docking studies in Figure 9B and C, the novel compounds are rather large molecules and will interact with both residues at the same time when interacting with the outward facing conformation. In summary, both the functional mutagenesis studies and the docking studies found that the novel compounds interact with the interface between the trimerization and transport domain at a binding site shaped by the key residues (M86, L295, K299, and W472). Both the docking and structure function studies also suggest that the novel molecules engage the transporter in an analogous fashion as the spider venom natural product (Table 2). From a mechanistic perspective it is of interest that several recent studies have revealed that residues in close proximity to the allosteric site studied here can affect glutamate translocation rates. Several papers studying biophysical properties of the bacterial GltPh glutamate transporter 86, 87, suggest that the large elevator-like movement of the central transport domain occur rapidly. It was also noted that the transporter cycle occurs through long inactive periods followed by rapid bursts of substrate translocation 86. The same study proposed that the process would involve a

15 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 61

separation of the translocation domain from the rigid scaffold. This idea was further supported by a recent study that replaced residues in the bacterial transporter with the corresponding human residues88 to impart characteristics of the human transporter and this resulted in markedly increased transport domain dynamics with increased rates of substrate transport88. The authors suggested that mutations R276S and M395 (residues that correspond to A362 and R476, respectively, in the human counterparts) favor structurally ‘unlocked’ intermediate states in the transport cycle. Remarkably, these are part of the same structural domain that we showed to be important for the action of the spider venom 72 and that we used to choose the molecular determinants employed in the pharmacophore developed in the present study. Because the compounds and the spider venom interact with this pocket we speculate they could be facilitating transport in a manner similar to the mutations described that unlock intermediate states and transitions between outward- and inwardfacing conformations88. A recent review89 supports the idea that interactions between the interface between the transport and trimerization domains are important determinants of the transport rate. It was proposed that if this interface could be engaged by the binding of a small molecule to a binding site within this interface this might alter the equilibrium between different states or conformations of the transporter and result in altered transition dynamics, thereby reducing or increasing transport, respectively89.

CONCLUSIONS

16 ACS Paragon Plus Environment

Page 17 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

To the best of our knowledge, these studies present the first identification of direct and selective positive allosteric modulators (PAMs) of EAAT2 glutamate transport. We believe this mechanism of action could have several therapeutic applications for treating conditions that involve excitotoxicity. An approach based on direct positive allosteric modulation of EAAT2 activity has several advantages compared with earlier approaches targeting EAAT2 that all have relied on increasing levels of EAAT2 protein expression19, 90. Because positive allosteric modulation is a fast direct process, it does not require synthesis and trafficking of new protein and will consequently have immediate acute effects and not rely on prophylactic pretreatments. Also, concerns about protein upregulation following chronic treatments and resulting potential side effects of the earlier approaches has given rise to concerns. Regarding this issue, we see no effects of GT949 on the expression of GLT-1 expression (supplementary figure 2). Immunoblotting of samples of glia incubated for 24h with or without 1 µM GT949 revealed that GLT-1 expression is unaffected by GT949, 24 h post treatment. Finally, considering that glutamate release is a key early event in excitotoxicity34; increasing glutamate clearance through PAM of EAAT2 is likely to be more efficacious than targeting downstream signaling processes that are activated following excitotoxicity. In conclusion, these novel PAMs of EAAT2 provide a promising entry point for identifying a new class of neuroprotective compounds with clinical potential that function by enhancing the removal of excessive glutamate in the synaptic cleft and preventing glutamate-mediated excitotoxicity. Further investigations into the mechanism of action of this class of compounds should pave the way for studies of their potential for in vivo

17 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 61

neuroprotection in acute conditions such as traumatic brain injury and stroke but also chronic neurodegenerative disorders such as Alzheimer’s, Parkinson’s and ALS. Interestingly, aberrant glutamate signaling has also been implicated in mental health disorders40, 91 and drug use disorders89, 92-95 and EAAT2 PAMs could therefore have therapeutic potential for these conditions as well. However, medicinal chemistry efforts are required to progress therapeutic developments, including optimizing the drug-like properties of the compounds to make them more likely to pass through the blood brain barrier and enter the CNS.

18 ACS Paragon Plus Environment

Page 19 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

METHODS Materials Radiolabeled substrates, [3H]-glutamic acid (51.1 Ci/mmol), [3H]-dopamine (53.6 Ci/mmol), [3H]-serotonin (28.2 Ci/mmol), [3H]-norepinephrine (14.9 Ci/mmol), were purchased from Perkin Elmer (Boston, MA, USA). DL-TBOA was purchased from Tocris (Bristol, UK). Cell culture media and supplements, including Dulbecco’s modified Eagle’s medium (DMEM) with glucose, Neurobasal-A, fetal bovine serum, fetal calf serum, heat-inactivated horse serum, penicillin/streptomycin, glutamine, L-glutamax, Dulbecco phosphate-buffered saline (D-PBS), L-glutamax-1, B-27 supplement and scintillation fluid were obtained from Thermo Fisher Scientific (Waltham, MA). Transfection reagent TransIT-LT1 was from Mirus Bio LLC (Madison, WI). Reagents for uptake assays and non-radiolabeled substrates were purchased from Sigma-Aldrich (St. Louis, MO).

In silico screening Modeling: a glutamate transporter homologue from Pyrococcus horikoshii GltPh, EAAT2 75, 96 was modeled in a substrate and an inhibitor bound state (pdb codes 2NWL and 2NWW) using the homology modeling program MODELLER software package (ver 9.1) 97, 98. According to an optimized protocol, ten structures of EAAT2 were modeled and one low-energy structure was further refined by energy minimization and constrained MD simulation. The resulting structure was further simulated by embedding it in a POPC membrane patch using the Desmond program (D. E. Shaw Research, New York, NY) with a production run of 30 ns. Optimal positioning of the membrane was

19 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 61

computed using the GltPh structures in orientations of proteins in the membrane (OPM) database. The results were analyzed using in-house trajectory analysis scripts and visualized using VMD program 99. An average three dimensional structure (Figure 2A) was generated using the last 5ns of the production run which was utilized for in silico screening. Hybrid-structure-based (HSB) virtual screening: The HSB method 100 was employed for designing and screening small molecules that could bind to EAAT2 modeled structure. The allosteric pocket to be targeted was derived from analyzing the molecular dynamics simulation trajectories and site directed mutagenesis studies that include H71 [TM2], L290, L295, G298 [TM5], and W472 [TM8] of EAAT2 72. A five-point three dimensional receptor-based pharmacophore (Figure 2B) was designed using residues M86, D485, W472, K299 and L295. The receptor based pharmacophore is central to the hybrid structure based method, wherein the pharmacophore is designed using the residues lining the allosteric site of EAAT2 and is assigned chemical features based on the nature of the residues. The receptor-based pharmacophore consists of pharmacophore elements namely aromatic ring, hydrophobic groups, hydrogen bond donor and acceptor groups assigned to residues W472, L295, M86, D485 and K299 respectively. The resulting pharmacophore was used to perform a virtual screen on a database of 3 million small molecules to obtain high quality hits. The electronic database was assembled by collecting molecules from various combinatorial chemistry libraries derived from several vendors including Asinex Inc, TimTec, Albany molecular research Inc, Enamine Inc, Chemdiv, Natural product libraries, FDA approved molecules library etc. For each molecule in the library, the 20 ACS Paragon Plus Environment

Page 21 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

structure, vendor library id and other molecular information is stored in an oracle database format. The screening was performed using legacy Sybyl 8.1 program with Unity module (Tripos Inc). The hit molecules that resulted from the screening were subjected to a variety of filtering schemes 101-104 including, drug-like properties using the lipinski’s rule, cardio toxicity, blood brain barrier penetration and activation of xenobiotic receptor – PXR that could lead to drug-drug interactions. In silico docking of compounds to allosteric site. Fifty eight hit molecules that passed the filtering criteria were docked to the allosteric site using the docking program GOLD (ver 5.2) 100, 104, 105. To efficiently sample the conformational flexibility of the ligand and the allosteric site residues, 20 independent runs were performed and resulting protein-ligand complexes were ranked using customized scoring schemes 100, 104

. The protein ligand complexes were scored using a two-tiered scoring scheme. The

first levels of scoring and ranking the complexes was performed using the default Goldscore method available from the Gold docking program. Based on the Goldscore ranking, twenty five best ranking protein-ligand complexes (with twenty five different small molecules) were chosen for customized scoring. The customized scoring scheme is a knowledge based method which is constructed by differentially weighting the positive interactions and penalizing the negative interactions100, 104. In this case study, we utilized the nature of interactions with the pharmacophore elements and residues within 4Å from the center of the pharmacophore to determine the positive and negative interactions. An aromatic stacking interactions such as the arene-arene interactions between GT949 triazole ring and the indole ring of W472 or hydrogen bonded interactions between piperazine of GT949 and T192 or hydrophobic interactions with 21 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 61

hydrophobic residues were treated as positive interactions and those that had conflicting properties were considered negative interactions. Based on the customized scoring, ten best ranking protein-ligand complexes were identified. These ten best ranking molecules were then tested in in vitro assays for functional selectivity and allosteric activity at the EAAT2. Site-Directed Mutagenesis. To explore the role of the amino acid residues used as anchors in the pharmacophore and/or lining the allosteric binding cavity, we produced mutated EAAT2 transporters. Mutations were from wild-type (WT) pCMV5EAAT2 vector 106 to neutral amino acids or to opposite-charged amino acids, to change particular selected residues in a reciprocal manner, using the QuickChange sitedirected mutagenesis kit (Agilent Technologies, Wilmington DE). The mutants generated in this study are: H71S, M86V, L290A, L295A, G298A, K299A, S465L, and W472I. Mutations were verified by DNA sequencing. Chemical synthesis and chiral separation. The top most best ranking molecule GT949 was synthesized using an Azido-Ugi multicomponent reaction 107, a robust reaction that utilizes an amine, an aldehyde, an isonitrile, and trimethylsilyl azide to form the product in one step (details and chiral separation in Supplemental Materials). All other molecules were purchased from combinatorial chemistry library vendors such as Asinex Inc, Enamine Inc and Chemdiv.

Neurotransmitter transporter studies in cell lines Cell Culture and DNA Transfection. COS-7 cells were maintained in DMEM with 10% fetal calf serum at 5% CO2. Transfection with empty vector pCMV-5 was used to 22 ACS Paragon Plus Environment

Page 23 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

control for the level of endogenous uptake of radiolabeled substrate in each experimental condition. The specificity of the compounds effect was investigated in COS-7 cells transiently transfected with glutamate transporter subtypes EAAT1-3 and monoamine transporters hNET (human noradrenaline transporter), hDAT (dopamine transporter) and hSERT (serotonin transporter). For glutamate uptake assays, subconfluent COS-7 cells were transfected with 0.5 µg of plasmid DNA per well using TransIT-LT1 transfection reagent (Mirus Bio LLC, Madison, WI), and plated at a density of ~100,000 cells per well in 24-well tissue culture plates. For monoamines uptake assays, cells were transfected 0.1 µg DNA per well as above and plated at a density of 10,000 cells per well on 96-wells plates. Dose-response assays. Two days after transfection, the cells were washed with room temperature phosphate buffer PBS-CM (2.7 mM KCl; 1.2 mM KH2PO4, 138 mM NaCl; 8.1 mM Na2HPO4, added 0.1 mM CaCl2 and 1 mM MgCl2, pH 7.4) and incubated for 10 min at 37oC with several concentrations of indicated compounds (0.01 –100 nM). Uptake reactions were initiated by the addition of 50 nM 3H-L-glutamate, serotonin, dopamine or noradrenaline, as appropriated. Kinetic assays. Cells were transfected with wild type (WT) EAAT2 or mutants. Two days later uptake reactions were initiated by the addition of unlabeled L-glutamate and 3H-L-glutamate (1–1000 µM, final concentration, 99% unlabeled and 1% labeled). All reactions were carried on for 10 minutes, then uptake was terminated by removal of solution, followed by two washes with PBS-CM. Cells in 24-well plates were lysed with 1% SDS/0.1 M NaOH. Lysate was transferred to scintillation vials containing 3 mL of scintillation fluid and radioactivity was quantified in a scintillation counter LS

23 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 61

6500 (Beckman Coulter, Brea, CA). For 96-well plates, scintillation fluid was added to each well, and the plate was counted in a Wallac M50 microbeta scintillation analyzer.

Glutamate transporter studies in astrocytes All animal experiments were treated according to the guidelines for use of animals in research approved by the Drexel University Institutional Animal Care and Use Committee (IACUC) and are in agreement with the U.S. Public Health Service Policy on Humane Care and Use of Laboratory Animals. Astrocyte preparation. Glia was prepared and cultured according a previous study 108, with modifications. Briefly, cerebral cortices from 2-4 day old Holtzman rat pups were dissected under sterile conditions and placed in 60 mm dishes containing dissection medium (in mM: Glucose 16, Sucrose 22, NaCl 135, KCl 5, Na2HPO4 1, KH2PO4 0.22, HEPES 10, pH 7.4, Osmolarity 310+10 mOsm). Tissue was minced with curved scissors, digested in 0.25% trypsin for 15 min. Trypsin action was ended by transferring tissue pieces to another vial with dissection medium. Tissue was then repeatedly passed through a serological plastic pipette until dissociated by trituration in presence of 60 µg/mL DNase. Cells were pelleted by centrifugation for 15 minutes at 280 g, and resuspended in glia plating medium (90% DMEM, 10% FBS and 50 µg/mL gentamicin) and transferred to culture flasks to a 37°C incubator (5-10% CO2). After growth for 10 days in vitro (DIV), cells were detach with 0.05% trypsin, centrifuged and plated at the density of 10,000 cells/well in poly-lysine coated 96-well plates. Plates are grown for 14 DIV before uptake assays.

24 ACS Paragon Plus Environment

Page 25 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

Uptake assays. Assays were performed as described 109. Briefly, using an Elx50® Biotek plate washer (Winooski, VT, USA), cells are washed in PBS-CM buffer. For dose response assays, vehicle and several concentrations of the compounds (0.01 nM- 1 µM) were added and incubated for 10 min at 37oC. Uptake assays are initiated by addition of 50 nM L- [3H]-glutamate, and incubation is carried on for 10 min at room temperature. Non-specific uptake was obtained in presence of 10 µM DL-TBOA. For kinetic assays, cells are washed in PBS-CM buffer and pre-incubated in the presence of either vehicle or a specific concentration of the compound (EC50), uptake reactions are initiated by the addition of unlabeled L-glutamate and 3H-L-glutamate (1– 1000 µM, final concentration, 99% unlabeled and 1% labeled). Incubation was carried on for 10 min at room temperature. Non-specific uptake was also obtained in presence of DL-TBOA. Reactions were finished by washing the plates twice with PBS-CM and the addition of 100 µL of scintillation fluid to each well. Radioactivity was counted in a Microplate Scintillation and Luminescence Counter (Wallac, Shelton, CT, USA).

Determination of the effect of novel compounds on glutamate receptors To determine whether the compounds have effects on the function of NMDA receptors, calcium imaging recordings from cultured cortical neurons was performed as described previously 110.

Cultured neurons preparation. Primary cortical neurons were obtained from the cerebral cortex of neonatal (P1 or P2) C57BL/6 mice, plated at density of 3,000 cells /12 25 ACS Paragon Plus Environment

ACS Chemical Neuroscience

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 61

mm round coverslip and cultured in medium containing Neurobasal A, fetal calf serum (2%), heat-inactivated horse serum (2%), L-glutamax-1 (0.2 mM), and B-27 supplement (2%). Cells were maintained at 37°C in a humidified atmosphere containing 5% CO2. The calcium imaging experiments were conducted on DIV 1.

Calcium imaging. Recordings were performed in living neurons, as described 111. Briefly, cultured neurons were loaded with a fura-2 calcium dye in Tyrode’s solution containing (in mM) 140 NaCl, 5 KCl, 2 CaCl2, 1 MgCl2, 10 HEPES, 5.6 glucose. Coverslips were mounted in imaging chamber (Model RC-25, Warner Instruments, Hamden, CT) and continuously perfused with Tyrode’s solution containing 0.5 µM tetrodotoxin. Images were acquired at 3-s intervals at room temperature (20 to 22°C) using an Olympus inverted microscope equipped with a CCD camera (Hamamatsu ORCA-03G, Japan). The fluorescence ratio was determined as the fluorescence intensities excited at 340 and 380 nm with background subtracted. 100 µM NMDA was applied in presence of 3 µM glycine and in the absence of Mg2+. The NMDA receptor antagonist AP-V (40 µM) was applied to certify that the answer was specifically NMDAinduced calcium response. Vehicle, GT949, or GT951 were applied and calcium changes recorded.

Calcium imaging. Recordings were performed in living neurons, as described 111. Briefly, cultured neurons were loaded with a fura-2 calcium dye in Hank's Balanced Salt Solution (HBSS). 100 µM NMDA was applied in presence of 3 µM glycine and in the absence of Mg2+. The NMDA receptor antagonist AP-V (40 µM) was applied to certify 26 ACS Paragon Plus Environment

Page 27 of 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

that the answer was specifically NMDA-induced calcium response. Vehicle, GT949, or GT951 were applied and calcium currents recorded.

Data analysis All data were analyzed using GraphPad Prism version 5.03 for Windows (GraphPad Software, La Jolla, CA). Dose-response assays were analyzed by non-linear regression for calculation of EC50s or IC50s and assessment of efficacy of the compounds. Graphs represent average ± SEM of at least three independent experiments performed in triplicate, and normalized to percentage of control (vehicle). Michaelis-Menten kinetics was assumed for calculation of Km and Vmax. Statistical significance was assessed using One-way analysis of variance (ANOVA) followed by Dunnett’s or Bonferroni multiple-comparisons posthoc tests with vehicle (for analysis of the effect of compounds) as control (* p