EAAT2 activators for Alzheimer’s disease 

Introduction

Alzheimer’s disease (AD) is the most common neurodegenerative disease, a most devastating disorder of the human mind and the major cause of dementia (1-3,106-108). Worldwide, 50 million people have dementia, estimated to reach 150 million in 2050, of which 70% is due to AD.
AD is diagnosed every 3 seconds, and detected first by slowly progressing and irreversible memory and mind problems, followed by remarkable changes in behavior and personality, and in the end, loss of self. AD survival time is 5-10 years. Family history of dementia, advanced or old age, are the only major risk factors of developing AD. These are the risks we cannot do anything about. Other risks are many, such as cardiovascular and cerebrovascular diseases, depression, diabetes, head trauma, obesity, psychiatric symtoms, stroke, and the APOE4 gene. 1% of AD is inherited and caused by dominant mutations in the APP, PS1 or PS2 genes [4-6].
AD is the most expensive disease there is in healthcare, consuming 1% of global economy. In the US, we spend $1 billion a day to look after 6.2 million people with AD living at home and long-term care facilities. In 2019, National Institutes of Health (NIH) funded AD research with $2.2 billion [7-10].
According to the amyloid hypothesis, AD begins in the brain with Aβ peptides accumulation, aggregation, and amyloid formation (15,16). Yet, in clinical trial studies, reducing Aβ peptides production and brain amyloid did not slow cognitive decline or improve daily living of AD patients. Similarly, preventive trials in cognitively unimpaired people at high risk, or genetically destined, of developing AD have failed to slow cognitive decline [17-24,125]. The results of these studies are against the amyloid hypothesis of AD etiology, its origin and disease mechanisms. The amyloid hypothesis is too good to be true [25], “too big to fail” [130], as Rudy Castellani and Mark Smith said in 2011.

Synaptic glutamate signaling

Glutamate is an amino acid and also the major synaptic signaling molecule of neurons in the brain, essential in cognition, learning and memory formation [26,27]. Glutamate is a non-essential amino acid, that is, all cells except neurons can synthesize glutamate de novo. Neurons make glutamate from glutamine (delivered by astrocytes), and store it at 60 mM in synaptic vesicles at nerve terminals.
Brain has 86 billion neurons and 85 billion glia cells (astrocytes, oligodendrocytes and microglia cells). Neurons are connected by synapses, a 200-300 nm wide and 20-40 nm high gap, maintained by extracellular matrix (ECM) and covered by astrocyte cell membrane processes. Brain has 1,000 trillion synapses [28-31]. One neuron connects, synapses to one or few neurons, which can be connected, synapsed by 40,000–140,000 neurons. As an ensamble, astrocytes form non-overlapping ‘territories’, which can cover 0.1-2 million synapses [32,33].
Intracellular calcium Ca[2+] concentration is 10–20 nM and extracellular 1–2 mM. This 100,000-fold concentration difference generates the transmembrane electrochemical Ca[2+] power gradient for synaptic glutamate signaling and other events in neurotransmission [34-36]. When the neuron AP (action potential) arrives at the axon terminal, it opens voltage-gated membrane channels for Ca[2+] inflow down the concentration gradient, which generates the Ca[2+] concentration-spike that initiates membrane fusion reaction of synaptic vesicles with the terminal membrane, and glutamate release into the synapse (synaptic cleft). All this happens in a few milliseconds, and increases synaptic glutamate concentration from 25 nM to 1–5 mM, a 100,000-fold increase. In the synapse (in one μm3 volume) with diffusivity of 0.46μm2/ms, it takes one μs for glutamate to reach and activate the postsynaptic membrane glutamate AMPA, NMDA and KA receptors. which are Ca[2+] channels. Here, the Ca[2+] inflow depolarizes the membrane potential (makes it more positive-inside), which generates EPSP (excitatory postsynaptic potential), which, however, fades away as it moves toward the nerve cell body, the soma. Nevertheless, if the EPSPs generated at several (hundreds and thousands) synapses arrive in soma at the same time, summation of them can cause enough membrane potential change to generate a real AP at the axon hillock [37].
In long lasting effects in the postsynaptic neuron, Ca[2+] inflow leads to calcium signaling [ref], in which Ca[2+] first binds and activates CaM (calmodulin), which then binds and activates calcium-regulated CaMKII kinase (phosphorylase) and CaN (calcineurin) phosphatase, which in turn regulate the activity of their target proteins in a variety of events, such as synapse remodeling by MMP (matrix metalloproteinase) enzymes, important in learning, memory formation and cognition [38]. Of note, CaN is the only calcium-regulated phosphatase in neurons. In astrocytes, CaN inhibits EAAT2 expression.
Glutamate is neurotoxic
In 1957, Lucas and Newhouse showed that high blood glutamate level caused mice to lose their sight due to retinal cell death [39]. Twelve years later, John Olney reported that mice treated with monosodium glutamate (MSG) showed neuron cell loss in brain areas not protected by the blood–brain-barrier, developed obesity and other disturbances, a phenomenon he called excitotoxicity [40]. In 1989, Rosenberg and Aizenman demonstrated “Hundred-`fold increase in neuronal vulnerability to glutamate toxicity in astrocyte-poor cultures of rat cerebral cortex” [41]. In 1992, Rosenberg and colleagues did a simple experiment with an astrocyte-rich neuron cell culture without sodium in the culture medium. Astrocyte glutamate uptake was impaired, to the extent, that they were no longer able to protect neurons from dying of glutamate caused excitotoxicity [42].

Astrocyte glutamate transporter EAAT2

Humans have five glutamate transporters, also called excitatory amino acid transporter (EAAT), transmembrane pro teins, which take glutamate from the outside to the inside of cells. Glutamate transporters differ in their tissue and cell distribution, sub-cellular level of expression, and glutamate uptake kinetics [43-45].
EAAT2 is made of the same three proteins of 573 amino acids, encoded by 11 axons of 34 Mb long gene on chromosome 11. The proteins are put together as they are being synthesized. There are several splice and exon-skipping variants of EAAT2 [130,131]. EAAT2 has eight transmembrane (TM) doq12mains, two helical hairpins HP1 and HP2, a trimerization domain and the glutamate transport domain (made of TM7, HP2 and TM8).
EAAT2 is 36% identical in amino acid sequence with the glutamate transporter homologue, aspartate transporter GltPh from Pyrococcus horikoshii archaebacterium. Most of the amino acids indicated in glutamate (aspartate) binding and transport are conserved. In the crystal structure, GltPh appears as trimer [132,133].
1% of brain protein is EAAT2, which covers 95% of synaptic glutamate uptake. Astrocytes have most of the EAAT2 protein. By electron microscope studies, 90% of EAAT2 in the brain is found on the perisynaptic astrocyte membrane processes covering synapses, at the density of 8,500/μm2, or 25,000 per synapse [26,46].
Without signaling, synaptic glutamate concentration is 25 nM, and with signaling 1–5 mM, a 100,000-fold increase. As soon as the glutamate signaling starts, it is stopped in 0.1–2 ms by astrocyte EAAT2, which takes up and clears glutamate from the synapses [47,48]. This effectively prevents extended glutamate signaling and excitotoxicity, which can lead to synapse loss and neuron cell death (Fig.1).
As seen in the movies made by Ryan et al. [50], using high-speed atomic force microscope, GltPh works like a 3-lift elevator, each lift going down and up (by isomerization as it is called), independently and randomly the membrane 15 nm, a third of the membrane span. Going down takes 50 ms, while going up happens in 0.5 ms [49]. If the EAAT2 elevator is as slow as GltPh (which we don’t know yet), it immediately raises the question: how can astrocytes and EAAT2 handle all that synaptic glutamate signaling in neurotransmission with 1 ms precision? It seems that, just by binding glutamate, EAAT2 can easily clear synaptic glutamate (Olga Boedker, pers.com), even at the rapid firing rate of 100 Hz [120]. Indeed, as 1% of brain protein, and 25,000 copies/synapse on astrocyte cell membrane covering synapses, EAAT2 is exactly where it should be.

Increasing EAAT2 expression with drugs

Astrocyte EAAT2 expression is regulated by several signaling systems acting on gene transcription and transcript splicing, mRNA processing and translation, protein trafficking and membrane targeting, and finally, EAAT2 activity [43-45,51,52]. A number of diverse drugs can increase EAAT2 expression, such as ceftriaxone (CEF), LDN/OSU-0212320, rapamycin, raloxifene and riluzole.
CEF is a potent wide-spectrum β-lactam antibiotic [116,117], which increases EAAT2 expression by gene transcription via the nuclear factor kappa-B (NFκB) signaling pathway. NFκB binds and activates the EAAT2 gene promoter at the -272 position [118]. Raloxifene is a selective estrogen receptor modulator (SERM) used in invasive breast cancer treatment, and in the management of osteoporosis [158,159]. Raloxifene increases EAAT2 expression by gene transcription via many signaling systems involving EGFR, MAPK, PI3K/Akt, Src, CREB and NFκB [160]. LDN/OSU-0212320 increases EAAT2 expression by mRNA translation
These, and many other drugs regulating EAAT2 expression, are excellently reviewed and discussed by Andréia Fontana [52,131]. Interestingly enough, parawixin-1, a neuroprotective compound isolated from Parawixia bistriata spider venom, activates EAAT2 in liposomes for glutamate uptake [53]. Parawixin-1 is EAAT2 specific, it increases glutamate uptake 70% by COS-7 cells expressing EAAT2 but not by COS-7 cells expressing EAAT1 or EAAT3. Recently, Fontana and colleagues have found, after a virtual screening of two million chemical compounds, three drugs that increase glutamate uptake by astrocytes, and EAAT2 liposomes. In astrocyte-neuron cell culture, a drug called GT949 protected neurons from dying of glutamate caused excitotoxicity [54].

EAAT2 in AD

In the transgenic APP/PS1 mice, a model of AD, CEF increased EAAT2 expression in the hippocampus and significantly slowed cognitive decline, as measured in the Morris water maze test. CEF also increased the expression of glutamine synthetase (GS) and system N glutamine transporter-1 (SN1), essential in the glutamine-glutamate recycling between astrocytes and neurons, and astrocyte delivery of glutamine to neurons. Most interestingly, CEF had none of these effects if EAAT2 was inhibited with dihydrokainate [55], therefore also indicating that EAAT2 activity is somehow linked to GS and SN1 expression.
In the APPswe,Ind mouse model of AD (which express 40% less EAAT2 in the brain), LDN/OSU-0212320 improved “cognitive functions, restored synaptic integrity, and reduced amyloid plaques”. Even after stopping the drug treatment, the effects were observed for one month, which prompted the authors to write: “EAAT2 is a potential disease modifier with therapeutic potential for AD” [56].
When these mice were crossed with transgenic mice expressing 2-fold more EAAT2, brain EAAT2 expression was normalized in the crossed mice, to the effect that they also showed improved “cognitive functions, restored synaptic integrity, and reduced amyloid plaques” [58].
When the APPswe/PS1dE9 mice, one more model of AD, were crossed with transgenic mice having only one EAAT2 gene, the crossed mice had increased spatial memory problems at 6 months and behavioral disorders at 9 months. These results suggest that astrocyte impairment in synaptic glutamate uptake (due to reduced EAAT2 expression) enhances AD progression [57].
Conditional deletion of brain EAAT2 in mice causes early problems in short-term and long-term memory, and in spatial reference learning [59]. Remarkably, EAAT2 deficiency also results in impaired Innate and adaptive immune systems, and altered expression of genes associated with inflammation and synaptic function, similar to those observed in AD and the aging human brain [59].
In a 1996 study by Masliah et al [60], of midfrontal cortex of postmortem cognitively unimpaired and AD brains were compared in the amount of EAAT2 (as measured by [3H]aspartate binding), synaptophysin, and spectrin degradation products. AD brains had 30% less EAAT2, 48% less synaptophysin, and more spectrin degradation products. These data suggest that decreased EAAT2 activity in AD is associated with increased synaptic damage and neurodegeneration.
As found by Jacob et al [61], studying postmortem brains, EAAT1 and EAAT2 gene and protein expression was already reduced in the early stages of AD development, in hippocampus and gyrus frontalis medialis, and was particularly obvious near the amyloid plaques. In later stages, synaptic glutamate receptor KA (GluK4) subunit was upregulated, and AMPA (GluA4) and NMDA (GluN1A) receptor subunits were downregulated.
The loss of synapses in hippocampus and neocortex is an early event in the development of AD, and the best correlate in cognitive decline [119-121]. Scheff and colleagues [122] have estimated the number of synapses, by electron microscope, in the outer cell layer of the dentate gyrus in the (age-matched) postmortem brains of individuals with mild to moderate AD (early AD, eAD), mild cognitive impairment (MCI), or no cognitive impairment (NCI). Individuals in the eAD group had significantly fewer synapses than individuals in the MCI and NCI groups. 75% in the MCI group had fewer synapses compared to the NCI group. Remarkably, synaptic numbers correlated with the individual’s Mini-Mental State Examination (MMSE) score, but showed no correlation with APOE genotype or Braak-staging of AD. Indeed, as the authors put it “This study supports the concept that synapse loss is an early event in the disease process and suggests that MCI may be a transition stage between NCI and eAD with synaptic loss a structural correlate involved in cognitive decline” [122].
The above studies are just a few examples supporting the idea that impaired astrocyte glutamate uptake, extended synaptic glutamate signaling and excitotoxicity underlie the first signs and symptoms of AD development.
Török and colleagues [123] have constructed an ultrafast glutamate sensor (igloo), a genetically engineered green fluorescent protein to image glutamate at synapses of choice. In a rat hippocampal tissue slice culture stimulated at 100 Hz, iGluu was fast enough to resolve individual glutamate release events, that is, in every 10 ms. This finding clearly shows how rapidly glutamate can be cleared from the synapses ex vivo.
The transgenic Q175 mouse is a model of Huntington disease. Q175 mice express less EAAT2 around the corticostriatal nerve terminals, and as shown by glutamate imaging with IGluu, synaptic glutamate clearance was slow. Treatment of wild type mice with the EAAT2 inhibitor TFB-TBOA mimicked the delayed synaptic glutamate clearance seen in the Q175 mice. As the authors write “The resuls provide a positive answer to the hitherto unresolved question of whether neurodegeneration (e.g., Huntington’s disease) associates with a glutamate uptake” and then suggest “. . . astrocytic Glu transport remains a promising target for therapeutic intervention . . .” [124].
HIV, the virus that causes AIDS, provides the best proof for impaired astrocyte synaptic glutamate uptake leading to neurodegeneration and development of dementia. HIV also causes HIV-associated neurocognitive disorder (HAND), also called HIV-associated dementia (HAD) [62,63]. In the brain, HIV infects astrocytes and microglia cells, but not neurons or oligodendrocytes. In astrocytes, HIV envelope glycoprotein gp120 inhibits EAAT2 gene transcription and glutamate uptake [64-66]. Increasing EAAT2 expression with CEF protects against HIV-associated neurotoxicity [65]. Of note, as reported by Togas and colleagues in 1994, transgenic mice expressing gp120 in brain astrocytes develop neurodegeneration [67]. Transgenic HIV Vpr mice, a model of HAND, express less EAAT2 in the brain [68]. HIV-infected people on CART (combination antiretroviral therapy), which includes two HIV proteinase inhibitors, amprenavir (APV) and lopinavir (LPV), often experience cognitive and behavioral problems [69]. Vivithanaporn and colleagues [70], have shown APV and LPV inhibit EAAT2 expression and gluta mate uptake by astrocytes in cell culture. APV or LPV treated HIV Vpr mice also show enhanced impairment in learning and memory formation, associated with reduced EAAT2 expression in the brain.
A most important note to consider when studying postmortem brains to understand cognitive, neurologic and psychiatric brain disorders is the following. Transcriptomic and histologic analyses have shown the postmortem (PM) brain has a uniquely different gene expression pattern and cell morphologies from the antemortem (AM) brain, or fresh brain, such as the brain tissue resected at epilepsy or b rain cancer surgery. In 2021, Jeff Loeb and colleagues [157] documented their extensive studies on AM and PM brains, in which they found the PM cortex had reduced expression of genes associated with neuron cell activity, increased expression of genes associated with astrocyte and microglia cell activity, while the expression of house keeping genes was the same.
It usually takes up to 24 hours before the PM brain is prop erly preserved. When Jeff and colleagues did an experiment of “simulated death”, as they affectionally called it, and studied gene expression over time in freshly isolated epileptic cortex tissue kept at 24C for 24 hours, they found the same changes emerging in gene expression, similar to those found in the PM brain. Histologic examination of the epileptic cortex tissue over time revealed ongoing degeneration of neurons, while astrocytes and microglial cells were outgrowing their membrane processes (similar to that seen in neuroinflammation). These observations seem to rule out the possibility of the changes being (epileptic) disease-related, in contrast to being changes dictated by time after death.
Without saying, these profound and rapid time‐dependent changes in gene expression and brain cells after death cast doubt on studying PM brains to understand cognitive, neurologic and psychiatric brain disorders.

AD drugs

Acetylcholine esterase inhibitors (donepezil, galantamine, rivastigmine) or memantine (which inhibits NMDA receptor and synaptic glutamate signaling) do not slow cognitive decline in AD progression, only provide temporary relief from the signs and symptoms, and even if that, lose their efficacy in a year [REF],
It is fair to say the amyloid cascade hypothesis of AD, formulated in 1992 by Hardy and Higgins [15], has generated significant research interest in APP (amyloid precursor protein) and Aβ peptides to make them perhaps the most studied proteins in biology, comparable to the proteins and enzymes involved in the blood clotting cascade mechanism. At the same time, however, the amyloid hypothesis has almost singularly misguided for 30 years AD research and clinical trial studies to discover and develop disease-modifying drugs, with efficacy and safety, to treat and prevent AD. Reducing Aβ peptides production with small molecule drug inhibitors of β- or γ-secretase, or clearing amyloid from the brain with anti-Aβ antibodies, did not slow cognitive decline in AD patients, or people at high risk of developing AD, due to carrying the APOE4 gene or having elevated brain amyloid (as imaged with PET scans). Similarly, anti-Aβ antibodies did not help people destined to AD, due to the inherited dominant APP, PS1 or PS2 gene mutations.
Most alarmingly, often times the drugs only harmed the study participants, by causing significant health problems and enhanced cognitive decline. One can only imagine the study participants, their family members and other caregivers volunteering for the trials, their high hopes all but dashed on the long and winding road to AD prevention and cure. Prevention is the only cure.
Whatever, the amyloid hypothesis is not dead yet, as exemplified with the recent resurrection of aducanumab [132], a human monoclonal antibody, which binds to Aβ oligomers and fibrils, and clears amyloid from the brain [133,134]. On June 7, 2021, the US Federal Drug Administration (FDA) approved aducanumab (branded as ADUHELM by Biogen) in the treatment of AD, as the first new AD drug in 18 years (after memantine in 2003). This FDA drug approval immediately generated considerable media attention, fierce debate in biomedical journals, and two Congressional hearings, amid concerns of the unusual business contacts between Biogen and FDA during the approval process, and whether ADUHELM in fact helps in slowing cognitive decline [135]. Moreover, two major hospital systems, Cleveland Clinic and Mount Sinai, have announced their decision of not administering ADUHELM to their patients. After reviewing the Biogen data on aducanumab trials, also Department of Veterans Administration (VA), serving 9.2 million war veterans, declined the use of ADUHELM for their patients.

Drug discovery and development is not easy

Pharmacodynamics and kinetics of drugs as measured by adsorption, distribution, metabolism, excretion and toxicity (ADMET) can tell if the drugs are good or bad. Naturally, of course, ADMET cannot be the same for all the people. As for toxicity, however, the major problem is the heart. Half of the drugs in human use can cause adverse drug reaction (ADR) with heart involvement due to drug ‘off-targeting’ NCX, the transmembrane sodium-calcium exchanger protein on the cardiac muscle cell membrane. As it is, ADMET and ADR kill 95% of drugs in clinical development. In the medicalized US society ADR is the number four cause of death.
Accordingly, as the second step to find EAAT2 drugs (Specific Aim 2), we screen-out the drugs acting on NCX, using an assay, which targets NCX reconstituted in liposome membrane and measures NCX activation or inhibition with red fluorescent Oxonol VI light emission, as described in experimental detail below (2.1. Screening-out EAAT2 drugs acting on NCX)
Discovery and development of drugs for brain disorders has an additional hurdle to pass-by, the blood-brain-barrier (3B), which effectively prevents brain entry of drugs. P-glycoprotein, also called MDR1 (multidrug resistance-1), is a transmembrane protein powered by ATP that ‘reverses’ drug entrance to cells, a major problem also in cancer chemotherapy [37]. Therefore, as the third step to find EAAT2 drugs (Specific Aim 3), we only continue studying drugs that can cross 3B in cell culture models.

In perspective

As Hans-Christian Danbolt once said “Like glutamate itself, glutamate transporters are somehow involved in almost all aspects of normal and abnormal brain activity” [26]. A number of recent papers and reviews indicate EAAT2 as a novel target for drug discovery and clinical development in a variety of cognitive, neurologic, psychiatric and neurodevelopmental brain disorders, such as AD, ADHD, alcoholism, ALS, autism spectrum disorder, bipolar disorder, depression, drug addiction, epilepsy, glioblastoma, Huntington disease, migraine, chronic pain, Parkinson disease, schizophrenia, ischemic stroke, and essential tremor [52,71-95,113].
As discussed above, Fontana and colleagues have discovered and studied two drugs, parawixin-1 and GT949, which bind and active EAAT2 in glutamate uptake and, in astrocyte-neuron cell culture, protect neurons against glutamate excitotoxicity [52,54]. It is a good guess, there are many more drugs to be found, which can do the same, and do it with a better efficacy, specify, safety and brain entry.
Here, as the first step to find such drugs, we describe a simple assay, which targets EAAT2 reconstituted in liposome membrane, and measures glutamate uptake with the red light emitted by Oxonol VI molecule, as described in experimental detail below (see 1.2. Liposome glutamate assay). The assay is blind, not structure-based, assumes nothing about the glutamate transport mechanism, and targets nothing but EAAT2. The assay is easy to do, fast, and cost-effective. Adapted in a high-throughput screening (HTS) format, thousands of drugs can be screened in a rather short time.
The power of this assay is that it only finds drugs that can bind and activate EAAT2 in glutamate uptake. This feature should significantly limit ‘off-targeting’ and possible harmful effects of drugs, the main hurdles and rate-limiting steps in discovering and developing AD drugs for human use. In contrast, in a drug screen targeting EAAT2 on astrocytes in cell culture, it would not be possible to find drugs nowhere near with the same precision of action. This is because drugs screened on astrocytes for enhanced glutamate uptake can do it, not just by acting on EAAT2, but in many other ways by increasing gene expression or mRNA translation of EAAT2 (see above). Such mechanisms cannot be EAAT2 specific. No gene is an island.
In addition, glutamate uptake by EAAT2 in liposomes works the same way as it does in astrocytes in cell culture, as measured by similar vMax and Km values for glutamate uptake [53]. This suggests a self-autonomous mechanism of action of EAAT2, with no additional cellular components involved, and it also suggests that drugs activating EAAT2 in liposomes act similarly on EAAT2 in astrocytes.

EXPERIMENTAL

Specific Aim 1. EAAT2 drug screening: the liposome glutamate assay
Here, we describe a simple assay to find drugs that can activate EAAT2 in glutamate uptake. The assay targets EAAT2 reconstituted in lipid vesicle (liposome) membrane, and measures glutamate uptake with the red fluorescent Oxonol VI dye (Fig.2). Oxonol VI is negatively-charged, barely lipid soluble molecule, with 19,000 fold difference in concentration in water over lipid [96]. Oxonol VI diffuses in and out of liposomes, and when it binds to the lipid membrane, its excitation and emission wavelengths increase 10-15 nm. This ‘red shift’ makes it possible to measure the lipid-bound Oxonol VI in liposomes [96].
EAAT2 glutamate uptake is powered by the Na[+] concentration difference across the membrane, 140 mM outside and 14 mM inside of the cell. EAAT2 is an electrogenic transporter in which three Na[+] ions and one H[+] ‘proton’ ion and one glutamate molecule are co-transported inside the cell [98]. Accordingly, drugs that activate EAAT2 in liposome glutamate uptake lead to liposomes becoming more positive-inside, to the effect, that more Oxonol VI accumulates inside the liposomes, binds to the lipid membrane, and emits more 660 nm red light (when excited with 580 nm yellow light). In other words, the simple idea of the liposome glutamate assay is this: the more glutamate in, the more red light out.

In 1987, Apell and Bersch [96] described in impressive details Oxonol VI as an optical indicator for membrane potentials in lipid vesicles, validating their assay with Na[+]K[+]ATPase reconstituted in liposomes. When ATP was added to activate Na[+]K[+}ATPase to transport three Na[+} ions in and two K[+] ions out the liposomes, thereby depolarizing the membrane potential (more positive-inside), the liposomes emitted more red light (Fig.3). 1% change in the Oxonol VI red light signal equals 1 mV change in membrane potential, which takes 270 ms to record, but can measure changes up to 150-200 mV [96,97]. As Ronald Clarke once put it, Oxonol VI is a slow dye (pers,com).
Of note, in cells, the Na[+] and K[+] ion transport by Na[+}K[+]ATPase goes the other way, Na[+] out and K[+] in (because ATP is made inside the cells), This Na[+] and K[+] ion traffic maintains the cell membrane potential at -70 mV (negative-inside), most essential for neurons in generating AP.
Interestingly, as captured with antibodies, Na[+]K[+]ATPase can bind to EAAT2, and EAAT2 can bind to presenilin PS1 [136,137].

1.1. EAAT2 liposomes

EAAT2 protein, prepared from rat brain homogenates [53], or from the human EAAT2 cDNA-transfected E.coli cultures, purified by affinity chromatography on rabbit polyclonal antibodies against the N-terminal domain of EAAT2 (GeneTex cat.no GTX47746), is reconstituted in liposomes by the dialysis method as described in a very useful detail in the ‘cook-book’ of Apell and Damnjanovic [97]. This requires three days in preparation, but produces reproducible liposomes appropriate for a long-term storage and use (Hans-Jürgen Apell, pers,cod).
However, before we begin our drug screening project with EAAT2 liposomes, we will prepare them by the spin-column method (which takes only a few minutes), as described by Fontana and colleagues [53]. The spin-column method provides a fast way to experiment on the amounts of different ingredients to make the ‘best’ liposomes for EAAT2 drug screening, such as the type of phospholipids, EAAT2 protein/lipid ratio, Na[+] and K[+] concentrations, and pH, to study their effects on glutamate uptake.
Spin-column method. Sephadex G-25 fine, kept overnight in a cholate buffer solution with 140 mM NaCl, 140 mM KCl, 15 mM sodium-phosphate buffer, pH 7.4, and 1% glycerol (the cholate buffer), is packed in 1 ml plastic syringes (plugged with cotton fiber at the bottom) and centrifuged (140g 2 min) to remove the void volume. EAAT2 in chelate buffer (2 mg/ml protein) is centrifuged (10,000g 20 min), and the supernatant (adjusted to 0.1 mg/ml) is mixed with 1.5 vol of cholate buffer with salts and phospholipids of choice, incubated on ice for 15 min, and 0.2 ml applied to the G25 columns, which are then centrifuged (500g 2 min). EAAT2 liposomes, 30-100 nm in diameterwwhich form spontaneously during this spin, are pooled and stored at +4C.The spin column cholate buffer becomes the inside medium of the liposomes. To change the outside medium is easy, EAAT2 liposomes are just suspended in the medium of choice.
Dialysis method. In 0.5 ml Eppendorf tubes, 100 μl of EAAT2 membrane suspension (2 mg/ml protein) are centrifuged at 160K for 15 min. The supernatant is discarded, and then chalet buffer solution is added, pipetted up and down to dissolve the pellet, and adjusted for 2 mg/ml proteintion, and centrifuged as above. Should see three things: a pellet (fragments of undissolved membranes), a faint turbid layer, and a major clear layer above. Pick up both layers, add 100 μl cholate buffer, vortex, transfer into a dialysis bag, immerse in 200 ml of dialysis buffer 1, and leave it for 72 hr on stirrer mixing at +4C. EAAT2 liposomes of 90 nm in diameter [96] are formed spontaneously.
EAAT2 reconstituted in the liposome membrane can be in two orientations, yet only the EAAT2 facing the outside can transport glutamate and Na[+] inside.

1.2. Liposome glutamate assay
To begin our EAAT2 drug screening project, we begin by setting up the liposome glutamate assay with Oxonol VI, as described [97], in a fluorescence cuvette with a light-path of 1 cm, in a fluorescence spectrophotometer with a temperature controlled cuvette holder at +20C equipped with a magnetic stirrer. Oxonol VI is excitated with 580 nm wavelength (slit width 20 nm) and the 660 nm emission (slit width 5 nm) is collected with an integration time of 1s.
On ice, have the following 1-3 ready
EAAT2 liposomes prepared in dialysis buffer 1.
100 μl of 30 μM Oxonol VI solution in ethanol, prepared daily from a stock solution.
40 μl of 250 mM L-glutamate solution. prepared by thawing a frozen Eppendorf tube with 10 μl of 1 M L-glutamate by adding 30 μl of water and vortexing.

For the experiment

Take 50 ml of the measurement buffer, perform sterile filtration into a small glass flask that has been made dust free by blowing it with compressed air before filling.
Add 1 ml of measurement buffer in a fluorescence cuvette together with a stirring magnet, placed into the cuvette holder.
Start recording fluorescence at 660 nm.
Add 1 μl of Oxonol VI solution and wait till you see the fluorescence remains constant.
Add 2.5 μl of EAAT2 liposomes (8 mg/ml lipid) and wait till you see the fluorescence remains constant.
Add 1 μl of L-glutamate and see the fluorescence increasing.
To show the fluorescence measures glutamate uptake by EAAT2, repeat the experiment, and at step 9, add L-glutamate with dihydrokainate (DHK), which inhibits EAAT2 [REF]. Note the difference in fluorescence intensity with or without DHK.
To validate the liposome glutamate assay for EAAT2 drug screening, repeat the experiment, and at step 9, add L-glutamate withGT949, which activates EAAT2 [54], courtesy of Andréia Fontana (Drexel University, Philadelphia, PA, USA), Note the difference in fluorescence intensity with or without GT949. To show GT949 increases glutamate uptake by activating EAAT2, inhibit EAAT2 with DHK.
Experiment with different amounts of GT949 to get an idea for the sensitivity of the liposome glutamate assay in detecting EAAT2 drugs. In other words, how many drug molecules/liposomes are needed to see the red light signal increasing?
In case you want to correlate the red light signal with the number of glutamate molecules in the liposomes, add [3H]glutamate with L-glutamate in step.9 and, after the assay, filter the sample through paper disks to collect the liposomes on the paper, and measure the [3H] activity and calculate the glutamate molecules/liposomes.

1.3. HTS liposome glutamate assay

To set up the stage for HTS (high through-put screening) of EAAT2 drugs, we collaborate with laboratories equipped with such technologies of automation of work done by computing machines.
Once we have found (in step 12. above) the limit of sensitivity of the liposome glutamate assay of EAAT2 drugs, we also need to find the minimal volume in the HTS assay, and the number of EAAT2 liposome s per volume needed to detect the red light. In 96-well plates, the volume is 100-300 μl, in 384-well plates 30-100 μl, and in 1536-well plates 5-15 μl. Which one it is, is easy to find out. However, h`ow to mix the assay components (“stirred but not shaken”) in real time, may take some time to figure out.

1.4. Drugs, chemical compounds and natural products

We begin our EAAT2 drug screening with the 1,040 FDA-approved drugs and nutritionals [99], and the 2,400 chemical compounds in human use assembled at the NCGC (The National Institutes of Health Chemical Genomics Center), the NCGC/NCATS Pharmaceutical Collection (http://tripod.nih.gov/npc), followed by screening a library of 140,000 small molecules [56], courtesy of Chen–Liang Glenn Lin (Ohio State University, Columbus, OH, USA).
Other libraries for our EAAT2 drug screening are Natural Product Activity and Species Source (NPASS), containing 30,927 natural products [100], a 10-digit DNA-encoded library of natural products containing 160 Traditional Chinese Medicines [101], and a high-diversity genetically-encoded combinatorial libraries of high-volume (10^8 – 10^13 members) of small peptide molecules [102].
In streamlining our EAAT2 drug screening, we will experiment on drug pooling, that is, drugs are not tested one by one, but in pools of ten, for example. Even manually, testing 1,040 drugs in pools of ten would be an easy task. As described in step 12. above, in experiments with GT949 (an EAAT2 activating drug), we have estimated the minimum number of drug molecules necessary for detection by the liposome glutamate assay. In control experiments, to rule out possible confounding effects of drug pooling (‘adverse drug interactions’), we spike the pools with GT949.

Specific Aim 2. Screen-out EAAT2 drugs acting on NCX

NCX (sodium-calcium exchanger) is the transmembrane protein, which can transport three Na[+] ions in and one Ca[2+] ion out the cell, when in the forward mode (mood), or the opposite way, when in the inward mode, depending on the Na[+] and Ca[2+] concentrations inside and outside the cell and the membrane potential.
NCX is major ‘off-target’ of drugs, which can cause problems in the heart beat, regulated by membrane potential and exquisitely fine-tuned by NCX activity on the cardiac muscle cell membrane. Accordingly, we set up an assay to find EAAT2 drugs acting on NCX and discard them in further studies.

2.1. Drug assay targeting NCX reconstituted in liposomes

In the forward mode, NCX transports three Na[+] in and one Ca{2+] ion out, which make the membrane potential more positive inside. In the inward mode, NCX transports the Na[+] and Ca[2+] ions in the opposite way and makes the membrane potential more negative inside. These changes in membrane potential changes in NCX liposomes can be measured with red fluorescent Oxonol VI light emission, as described in the liposome glutamate assay
In control experiments, to validate the assay, we use ? To activate, and ? to inhibit NCX.
Specific Aim 3. Screen EAAT2 drugs for brain entry in cell culture models of blood-brain-barrier
Brain entry of drugs is the most common problem and rate-limiting step in developing drugs for brain diseases [109-111]. So, even if we have, hopefully, discovered many EAAT2 drugs, they are useless if they can not cross the blood-brain-barrier (3B) and activate astrocyte EAAT2 in synaptic glutamate uptake. As it is, the brain entry of drugs can be improved by chemical modifications (while retaining their activity), but this is not easy to do time and cost-effectively. Similarly, brain targeting of drugs in nanoparticle formulations takes time and many trials and errors to do it right [104,112].
It is much easier, faster and cost-effective to screen the EAAT2 drugs for brain entry in 3-dimensional cell culture models of 3B, and after that, continue studying only the drugs passing this test for their efficacy, specificity and safety in glutamate uptake in astrocyte cell culture.
We will setup the 3B cell culture model, as developed by Patrick and colleagues [105]. In the screening for the 3B crossing drugs, we capture them with the liposome glutamate assay. Importantly, usin g this assay, we also confirm that, after crossing the 3B, the drugs still retain their efficacy as EAAT2 drugs increasing glutamate uptake.
In control experiments, to validate this 3B model, we use gleev ec, which does not enter the brain [114], and temozolomide, which does enter the brain [115].

Specific Aim 3. EAAT2 drugs for neuroprotection

The 3B crossing EAAT2 drugs are next studied in astrocyte cell culture for their efficacy, specificity and safety in glutamate uptake. Drugs passing this test are then studied in astrocyte-neuron culture for their efficacy in protecting neurons against glutamate excitotoxicity.

3.1. Astrocyte glutamate uptake

As described by Falcucci and colleagues [54], primary rat cortical astrocytes are prepared and cultured for 2 hours in the presence of L-glutamate, [3H]-L-glutamate and the EAAT2 drugs, one by one, and then harvested to measure [3H] activity for glutamate uptake. In control experiments, we activate and inhibit EAAT2 with GT949 and DHK, respectively, to validate the assay and to show the drugs work by activating EAAT2.
In order to streamline this assay, we experiment on measuring astrocyte glutamate uptake with Oxonol VI. If this simple assay works, we will then set up the assay in HTS format.
Next, the EAAT2 drugs passing this test are characterized by astrocyte RNA sequencing [103] for their effect on gene expression, and to have an idea of the events in astrocytes caused by EAAT2 activation or ‘off-targeting’ by the drugs. If need be, the events are studied in detail for any possible harmful outcomes, in order to stop further development of such EAAT2 drugs.

3.2. Protection against glutamate excitotoxicity

Here, to conclude our screening project of discovering Alzheimer drugs targeting EAAT2, we only study drugs that have passed the rigorous tests of quality for brain entry, efficacy and safety in activating astrocyte glutamate uptake, and study them in astrocyte-neuron cell culture for their efficacy in protecting neurons against glutamate excitotoxicity.
As described [54], cortical astrocytes and neurons are prepared from embryonic (E17) brains of Holtzman rats and plated at a density of 35,000 cells/coverslip in 12-well plates, and cultured for two weeks. For the neuroprotection experiments, EAAT2 drugs are added with L-glutamate (10 or 100 μM), and after 24 hours, the cell cultures are processed for immunolabeling with EAAT2 or GFAP antibodies for astrocytes, and with MAP2 antibody to asses the survival of neurons. The antibodies are visualized with secondary antibodies of different colors. To validate this cell culture assay of neuroprotection, we activate EAAT2 with GT949, as described [54]. In control experiments, to prove the neuroprotective EAAT2 drugs work by increasing astrocyte glutamate uptake, we inhibit EAAT2 with DHK.

CONCLUSION

In the near future, we look forward in studying the EAAT2 drugs (passing all the tests in the lab for efficient, specific and safe neuroprotection) for their safety and therapeutic efficacy in mouse models of AD. Also, without saying, we hope the EAAT2 drugs can soon provide much needed light in the prevention and treatment of AD and other cognitive, neurologic, psychiatric and neurodevelopmental brain disorders. Fiat lux.

Acknowledgements

We thank and much appreciate Hans-Jürgen Apell for his continued interest and help in developing the concept of the liposome glutamate assay for EAAT2 drug screening, careful reading of this research proposal, as well as his meticulous text editing the semantics and syntax of our writing.