|Home | About | Journals | Submit | Contact Us | Français|
Glutamate is the predominant excitatory amino acid neurotransmitter in the mammalian central nervous system (CNS). Glutamate transporter EAAT2 /GLT-1 is the physiologically dominant astroglial protein that inactivates synaptic glutamate. Previous studies have shown that EAAT2 dysfunction leads to excessive extracellular glutamate and may contribute to various neurological disorders including amyotrophic lateral sclerosis (ALS). The recent discovery of the neuroprotective properties of ceftriaxone, a beta lactam antibiotic, suggested that increasing EAAT2 /GLT-1 gene expression might be beneficial in ALS and other neurological/psychiatric disorders by augmenting astrocytic glutamate uptake. Here we report our efforts to develop a new screening assay for identifying compounds that activate EAAT2 gene expression. We generated fetal derived-human immortalized astroglial cells that are stably expressing a firefly luciferase reporter under the control of the human EAAT2 promoter. When screening a library of 1040 FDA approved compounds and natural products, we identified harmine, a naturally occurring beta-carboline alkaloid, as one of the top hits for activating the EAAT2 promoter. We further tested harmine in our in vitro cell culture systems and confirmed its ability to increase EAAT2/GLT1 gene expression and functional glutamate uptake activity. We next tested its efficacy in both wild type animals and in an ALS animal model of disease and demonstrated that harmine effectively increased GLT-1 protein and glutamate transporter activity in vivo. Our studies provide potential novel neurotherapeutics by modulating the activity of glutamate transporters via gene activation.
Glutamate is the predominant excitatory amino acid (EAA) neurotransmitter in the mammalian CNS. It activates ligand-gated ion channels that are named after their agonists N-methyl-D-aspartate (NMDA), amino-3-hydroxy-5- methyl-4-isoxazolepropionate (AMPA), kainate and G-protein-coupled metabotropic receptors. Extracellular accumulation of EAAs and excessive activation of EAA receptors contribute to neuronal cell death observed both in acute insults to the CNS and in chronic neurodegenerative diseases, including ALS, Huntington’s disease, Alzheimer disease and behavioral disorders such as depression (Gegelashvili et al., 2001).
Low extracellular levels of glutamate are maintained by transport into neurons and astrocytes. Five distinct glutamate transporters have been cloned that allow for sodium-dependent high-affinity glutamate transport, designated GLT1/EAAT2, EAAC1/EAAT3, GLAST/EAAT1, EAAT4, and EAAT5 (Danbolt, 2001). The EAAT (excitatory amino acid transporter) nomenclature refers to the human transporter species while GLT refers to the rodent transporter homologue of EAAT2. Expression of EAAT1 and EAAT2 is generally restricted to astroglia; other EAATs are neuronal. The astroglial transporter EAAT2 is the dominant transporter in brain and spinal cord, accounting for up to 95% of all activity (Danbolt, 2001).
Excess levels of glutamate in the cerebral spinal fluid (CSF) (10 fold increase) are found in up to 40% of ALS patients (Rothstein et al., 1990). Reduced functional transport of glutamate has been observed in postmortem tissue from patients with ALS (Rothstein et al., 1992). This is due to a selective loss of EAAT2, whose immunoreactivity is dramatically reduced in ALS (Rothstein et al., 1995). At least 40% (and as much as 75%) of all sporadic ALS patients appear to have defects in glutamate transport and excessive extracellular glutamate levels (Spreux-Varoquaux et al., 2002). SOD1G93A transgenic ALS rats also show reduced synaptosomal glutamate uptake activity as well as increased extracellular levels of glutamate and reduced clearance of glutamate (Dunlop et al., 2003; Howland et al., 2002). All SOD1 mutant rodent animal models of ALS show reduced levels of GLT1 (Dunlop et al., 2003; Howland et al., 2002). Alterations of glutamate transport protein expression have also been reported in other neurodegenerative and demyelinating diseases including Huntington’s disease, Alzheimer’s disease, and multiple sclerosis.
Genetic overexpression of the EAAT2/GLT1 protein is protective in vitro and delays disease in vivo (Guo et al., 2003). A recent unbiased small molecule screen employing low throughput spinal cord tissue slices revealed that beta lactam antibiotics, including ceftriaxone, increase GLT1 /EAAT2 expression, protect against neural injury and delay disease in ALS mice (Rothstein et al., 2005).
Here we report our effort to employ a new drug discovery approach by developing a cell-based luciferase reporter screening system for identifying EAAT2 promoter activators. By screening a library of up to 1040 FDA approved compounds and natural products, we identified harmine, a naturally occurring beta-carboline alkaloid, as one of the top hits for turning on the EAAT2 promoter. We demonstrated that harmine effectively increased GLT-1 expression both in vitro and in vivo. Our studies provide a potential new neurotherapeutic strategy by modulating the activity of glutamate transporters via gene activation.
We obtained a human immortalized astroglial-fetal derived cell (HIA) line from Dr. Ahmet Hoke at Johns Hopkins University (Hoke et al., manuscript in preparation). To be able to use this line in our screening assay, we stably transfected HIA cells with a plasmid encoding firefly luciferase reporter under the control of a short fragment of the human EAAT2 promoter (2.5 kb) as described previously (Rothstein et al., 2005). The obtained cell line (CC4 line) was then validated to see if it is suitable for our screening assay, through testing its reproducibility and calculating its S/N ratio and Z’ factor. CC4 cells were seeded on 96-well plates and treated with different compounds. After 3-days incubation, cells were lysed in plates and luciferase activity was measured with Fluostar OPTIMA plate reader (BMG Labtech, NC), following manuals from Promega.
Primary astroglial cells were cultured from postnatal 2-3days mouse pups. Cortices were dissected out and dissociated with papain and subsequently cultured on collagen-coated T75 flask in DMEM containing 10 % fetal bovine serum (FBS). At DIV14 astrogial cells were seeded into collagen-coated 6-well plates at a concentration of 7×105 cells/well. Primary cortical neurons were isolated from cortices of E16 embryonic mice. After dissociated with papain, 1 million neurons were seeded per well on the top of the confluent astrocytes. Co-cultures were first maintained in Neurobasal medium supplemented with 5% FBS and 2% B-27. After 4 days half of the medium was changed into serum free Neurobasal medium supplemented with 2% B27. Cells were treated with various compounds at DIV5 for 3-6 days accordingly.
We obtained a human neural stem cell line (HSC) from Dr. Clive Svendsen at Wisconsin Medicine (Svendsen et al., 1998). HSC spheres were maintained in maintenance medium (DMEM/F-12 supplemented with 1% N2, 0.1% LIF and 20ng/ml EGF). To differentiate the stem cells into astrocytes, HSC spheres were dissociated with acutase (Chemicon) and seeded on Poly-L-Ornithine /Laminin coated 12-well plates. After being cultured in DMEM/F-12 supplemented with 10% fetal bovine serum for at least 7 days, cells were treated with various compounds.
Total RNA from cultured cells or tissues were isolated by using RNA EASY kit (Qiagen) which was then reverse transcribed using High Capacity cDNA Archive kit (Applied Biosystems). Real-time PCR was performed using iCycler iQ Real-Time Detection System from BioRad.
Cultured cells were directly lysed with ice-cold lysis buffer (20mM This-HCl, pH 7.4, 140mM NaCl, 1mM EDTA, 10% glycerol, 1% Triton-X and 0.1% SDS). Tissues harvested from mice were first homogenized in ice-cold lysis buffer with glass homogenizers, then sonicated, followed by centrifugation at 13,000 × g for 15min at 4°C. The protein concentrations of the supernatants were quantified with DC protein assay kit (BioRad, CA). 5-10 μg of total proteins per lane were loaded on precasting 10% polyacrylamide gel (BioRad) for SDS-PAGE. After being transfered onto PVDF membrane, immunoblots were probed with primary antibodies: anti-GLT-1 (1:2000), anti-GLAST (1:200), anti-GFAP (1:5000, Chemcon), and anti-Actin (1:5000, Sigma); and subsequently incubated with HRP conjugated secondary antibody (1:10,000, GE Healthcare). Immunoblots were detected with SuperSignal West Pico chemiluminescent substrates (Thermo Scientific) and visualized by VersaDoc system (BioRad). Intensities of bands were determined by ImageJ software.
0.5 μM L-glutamate (cold:radioactive=99:1) and 0.3 μCi L-[3H]glutamate per sample (PerkinElmer) was used for measuring glutamate uptake with cultured cells; while 0.5μM L-glutamate (cold:radioactive=99:1) and 0.075 μCi L-[3H]glutamate per sample was used for measuring glutamate uptake with crude synaptosomes prepared from mice cortical tissues. For glutamate uptake assay with cultured cells, cells were first washed and pre-incubated at RT for 10 min in Na+ buffer (5mM Tris-HCl, pH 7.2, 10mM HEPES, 140mM NaCl, 2.5mM KCl, 1.2mM CaCl2, 1.2mM MgCl2, 1.2mM K2HPO4, and 10mM D-glucose) in the presence or absence of various glutamate transporter inhibitors. Glutamate uptake reaction was started by incubating cells for 5 minutes at 37°C in Na+ buffer containing 0.5 μM L-glutamate and 0.3 μCi L-[3H]glutamate per sample, followed by rapid washing twice with ice-cold Na+-free assay buffer (5mM Tris-HCl, pH 7.2, 10mM HEPES, 140mM Choline-Cl, 2.5mM KCl, 1.2mM CaCl2, 1.2mM MgCl2, 1.2mM K2HPO4, and 10mM D-glucose). Cells were then lysed with 0.1N NaOH solution and radioactivity was measured using a scintillation counter. For glutamate uptake assay with crude synaptosomes, mice cortical tissues were first homogenized in ice-cold 0.32M sucrose and then centrifuged at 800 × g for 10min. Supernatants were subsequently centrifuged at 20,000 × g for 20min at 4°C. Pellets which contain crude synaptosomes were then washed in 0.32M sucrose, centrifuged and re-suspended in ice-cold 0.32M sucrose. Glutamate uptake reactions were started by adding crude synaptosomes in Na+ buffer containing 0.5 μM L-glutamate and 0.075 μCi L-[3H]glutamate per sample and incubated for 3 minutes at 37°C, and stopped by rapidly adding four volumes of ice-cold Na+-free assay buffer. Radioactive crude synaptosomes were then collected on filter paper using a Brandel Harvester and washed with Na+-free assay buffer. Each piece of filter paper containing radioactive crude synaptosomes was transferred into scintillation vials and radioactivity was measured using a scintillation counter. Protein concentrations were measured by Bradford method.
Six to eight week old SODG93A mice were ordered from Jackson Lab. Wild-type C57/B6 mice (8 weeks) were ordered from Charles River. The care and treatment of animals in all procedures strictly followed the NIH guide for the Care and Use of Laboratory Animals and Guidelines for the use of animals in neuroscience research and the Johns Hopkins University IACUC. To perform pharmacokinetics studies, 8 week old wild-type C57/B6 mice were injected intraperitoneal with 30mg/kg harmine at time zero. At designated time points three animals were sacrificed and plasma, cortical and spinal cord tissues were collected. Tissues were homogenized in 2mM SDS solution. Plasma and homogenate were sent out to a commercial research organization and analyzed using LC/MS/MS detection method.
A reporter cell line (CC4) was generated by stably expressing a firefly luciferase reporter under the control of a 2.5kb human EAAT2 (GLT-1) promoter fragment in human immortalized astroglial-fetal derived cells (Rothstein et al., 2005; Yang et al., 2009). Compared to other reporter systems, luminescence based assays are most sensitive therefore suitable to develop a screening assay that can be converted into a high throughput screening (HTS) assay to screen large chemical libraries (100,000s of compounds). We reasoned that after long-term (i.e., days) exposure of CC4 to various compounds, we could assess the compounds’ efficacy on increasing EAAT2 promoter activity by measuring luciferase activity. We validated the suitability of the CC4 cell line for our drug screening assay by testing its reproducibility as well as by assessing drug screening assay parameters such as signal-to-noise ratio (S/N) and Z’ factor (Figure 1A). We concluded that the CC4 cell-based luminescent assay is a robust assay with high reproducibility, a reasonable S/N (close to 11) and Z’ factor (close to 0.6). We therefore started to screen a library of 1040 drugs and nutritionals that are FDA approved in the USA (US drug collection; MicroSource Discovery Systems, Inc.). Each drug was tested at a 10μM concentration and was examined in triplicate. A representation of a 96-well plate from the screening is illustrated in Figure 1B. From this screen, we identified a series of compounds that were capable of increasing luciferase activity, presumably due to the enhancement of EAAT2 promoter activity (Table 1). A naturally occurring beta-carboline alkaloid, harmine, was identified as one of the most effective compounds. It was subjected to a dose response curve analysis (0.001-30μM) resulting in an EC50 value of 3.2μM (Figure 1C). High concentrations of harmine (≥ 30μM) were toxic to our cultured cells (see single data point in dose response curve). We therefore calculated the EC50 value using the 10μM response as our maximal value.
Luminescence based screening assays may lead to false positive hits that non-specifically interfere with the luciferase activity rather than increasing luciferase expression via promoter activation (Auld et al., 2008). In order to test the specificity of harmine, we performed the assay in cells expressing luciferase driven by a gene promoter that is not related to glutamate transporters (Monocarboxylate Transporter 4). We found that harmine specifically increased the EAAT2 promoter driven luciferase but not the unrelated control (Figure 2A). Parallel experiments demonstrated that actinomycin, a blocker for transcriptional activation, completely abolished the effect of harmine (Figure 2B). This indicated that harmine increased luciferase activity of CC4 cells through transcriptional regulation. Last, we performed reverse transcription-polymerase (RT)- quantitative PCR and found that harmine increased luciferase mRNA after 24-hour treatment in CC4 cells (Figure 2C). These data supported our hypothesis that harmine acts as a transcriptional activator of EAAT2.
We next validated harmine in our primary cortical neuron-astrocyte co-culture system to determine whether harmine is capable of increasing endogenous GLT-1 mRNA. Harmine efficiently increased GLT-1 mRNA after 48-hour treatment (Figure 3A). Unexpectedly, harmine also significantly increased the expression of GLAST, another astrocytic glutamate transporter (Figure 3B). When other genes were examined, including Glial fibrilary acidic protein (GFAP), Neuron-specific class III beta-tubulin (Tuj1) and neuronal glutamate transporter (EAAC1), an increase in their gene expression (data not shown) was not seen. The specific induction of harmine on both GLT-1 and GLAST suggests that there might be a common transcriptional regulation of the key proteins in glial glutamate uptake.
Mouse GLT-1 and GLAST promoter sequences are not 100% homologous to the human gene and the primary screen with CC4 cells only tested a partial fragment of the EAAT2 promoter. It was unclear if harmine could show activity on the human full-length EAAT2 and/or EAAT1 promoters. We then tested its effect in human stem cell derived astrocytes, employing a human neural stem cell. This is not an immortalized cell line, but a human neural stem cell line maintained as a neurosphere that has the full potential to either differentiate into neurons or astrocytes (Svendsen et al., 1998). We dissociated the neurospheres and cultured the cells with a specific medium to differentiate them into astrocytes. We then treated those astrocytes with harmine and showed a significant induction of both human EAAT1 (GLAST) and EAAT2 (GLT-1) expression after 48 hour treatment (Figure 3C & D).
We next asked if harmine is capable of increasing endogenous GLT-1 and GLAST proteins in our primary mouse cortex co-culture system. Indeed, we demonstrated that harmine significantly increased both endogenous GLT-1 and GLAST protein after a 6-day treatment (Figure 4A) while other astroglial proteins like GFAP or neuronal protein Tuj1 were unaffected by drug treatment.
To verify that the harmine-induced newly formed protein is functional, we performed a glutamate uptake assay in our cortical co-culture system. As shown in Figure 4B, harmine dose dependently increased Na+ dependent glutamate transporter activity. To determine which of the glutamate transporter subtypes contribute to the measured glutamate uptake we performed the uptake assays in the presence of subtype specific glutamate transporter blockers. TBO-A, a non-specific glutamate transporter blocker, inhibits all Na+ dependent glutamate uptake, while DHK specifically blocks GLT-1-dependent glutamate uptake. Subtraction of specific from non-specific uptake revealed that harmine increased both GLT-1 and GLAST dependent uptake (Figure 4C).
Harmine is a harmala alkaloid (Figure 5A) and generally these alkaloids have a wide spectrum of pharmacological actions, including MAO inhibition, binding to benzodiazepine receptors, convulsive or anticonvulsive actions and tremorogenesis (Aricioglu et al., 2003; Glennon et al., 2000; Guan et al., 2001; Husbands et al., 2001; Kim et al., 1997; Lutes et al., 1988; Song et al., 2004). To gain some insight into the specificity of harmine’s ability to up-regulate glutamate transporter gene expression, we tested the activities of anumber of harmine analogs. Harmol and harmane showed ~ 50% and 20% activity respectively compared to harmine, while other derivatives showed no activity (Figure 5B). Figure 5C shows a full dose response curve of harmol using our primary screening assay. We calculated its EC50 value (3.8 μM) using the 10μM response as our maximal value, since higher concentrations of harmol (≥ 30μM) were toxic to our cultured cells. Then we tested harmol (3μM) in our neuron-astrocyte co-culture system found that there was a positive trend towards a GLT-1 upregulation and a significant increase in GLAST expression (Figure 5D).
To confirm BBB penetration, we performed a simple pharmacokinetics study on WT mice. Animals were dosed i.p. with 30mg/kg harmine at time zero and at designated time points three animals were sacrificed. Plasma and brain homogenate were analyzed by LC/MS/MS detection method. Figure 6A shows the time course of harmine concentrations in plasma, spinal cord and cortex. As described previously, harmine penetrates the brain very rapidly, most likely due to its highly lipophilic structure (Zetler et al., 1974). Harmine is eliminated from the brain very quickly, reaching 10% of its maximum levels at about 90min after dosing. Pharmacokinetic parameters differed slightly between cortex and spinal cord with Cmaxcortex = 24377 ng/g and Cmaxspinal cord = 18623 ng/g. As reported before, plasma levels of harmine were lower than tissue levels (about 1/3 at time zero) and rapidly declined over the time course of our measurements (Zetler et al., 1974). An estimation of the molarity of harmine in the brain led to a value of 3μM at the 2 hr time point, which parallels the concentration needed in vitro to achieve mRNA and protein upregulation of glutamate transporters.
To test whether harmine increases protein levels of GLT-1 in vivo we treated SOD1G93A mice. These mice show ALS-like symptoms at around 90 days (disease onset), and they reach disease end-stage typically around 130 days. SOD1G93A mice (10-11 mice per group) were treated from D90 to D100 with either harmine (10mg/kg and 30mg/kg; i.p.) or saline twice a day. At the end, animals were sacrificed and tissues (cortex, spinal cord) were harvested and analyzed by western blot. After normalization to actin, we observed that 10mg/kg of harmine significantly increased GLT-1 protein expression (p< 0.05, t test) in cortex, and lead to a positive trend of protein upregulation in the spinal cord (Figure 6B).
Next, we asked if harmine-induced increase in GLT-1 protein would increase glutamate transporter activity in vivo. We injected wild type mice (10-11 mice per group) twice a day with harmine (10mg/kg, i.p.) or saline for 10 days. At the end, animals were sacrificed and glutamate uptake was measured from cortical tissue homogenates (Figure 6C & D). We demonstrated that harmine significantly increased both Na+ dependent glutamate uptake (which represents both, GLT-1 and GLAST-dependent uptake) and GLT-1 specific glutamate uptake (p<0.05, t test).
The discovery and development of new drugs that could modulate glutamate transporters could be a novel approach to multiple neurological and psychiatric disorders. Dysfunctional glutamate transmission and consequent accumulation of extracellular glutamate has been a target for therapy in ALS, Huntington’s disease, epilepsy, multiple sclerosis as well as depression and schizophrenia. In this paper, we have outlined a new cell-based screening assay for the identification of small molecule compounds that activate gene expression of glutamate transporters, proteins that are known to become dysfunctional and downregulated during ALS disease progression, as well as in other neurological disorders (Gegelashvili et al., 2001) (Lepore et al., 2007; Rothstein, 2009; Rothstein et al., 2005).
This assay was designed to be readily translated into a high-throughput screening (HTS) platform to screen large chemical libraries (e.g. 105-106 compounds). A pilot screen was successfully performed with an external HTS facility screening a library of about 30,000 compounds (R. Sattler and J. Rothstein, unpublished observation). In addition, this cell-based assay was used to perform a combinatorial screen of the Microsource library used in this publication (Li et al., manuscript in preparation).
Our laboratory had performed a recent screen for glutamate transporter activators using primary organotypic spinal cord slice cultures (Rothstein et al., 2005). In this screen, we identified numerous beta lactam antibiotics, such as ceftriaxone, as effective small-molecule compounds that upregulated GLT-1 gene expression and observed that it protected against motor neuron cell death. Since primary cultures cannot be used for high throughput screening, we developed the current screening assay and validated it by testing the newly updated Microsource library. A recent report shows the development of a HTS for translational activators of glutamate transporter EAAT2 (Colton et al., 2010). Using a cell-based enzyme-linked immunosorbent assay in a stably transfected astrocyte cell line, Colton and colleagues developed a screening assay to specifically screen small molecule compounds that activate EAAT2 translation, although the active compounds in that assay were not shown to actually increase endogenous protein in vivo. That assay differs from our assay in that we screen for compounds acting on a transcriptional level, which means we screen for the formation of new mRNA followed by new protein synthesis. Studies have shown that mRNA levels of GLT-1 are downregulated even before disease onset (Yang et al., 2009), hence, it will be crucial to boost new mRNA synthesis to overcome the loss of GLT-1/EAAT2 protein during disease.
The use of luciferase-based reporter gene assays has to be taken with care, as some compound structures can lead to false positive hits by stabilizing the luciferase protein instead of modulating gene transcription (Auld et al., 2008). In our study, we focused on only a few of the positive hits we obtained from our screening assay. While most compounds showed specific activity towards the EAAT2 promoter, some compounds with increased luciferase activity failed to be blocked with actinomycin treatment and consequently failed to increase mRNA levels of GLT-1 in vitro. None of these compounds were further tested for transporter activity.
With the newly developed assay, we identified harmine, a naturally occurring small molecule, as an effective compound to induce transcriptional activation of astrocytic glutamate transporters in cultured astroglia- but more significantly, in vivo following repeated administration. Harmine is a beta-carboline alkaloid that was first isolated in 1847 from seeds of Peganum harmala (Syrian rue) and Banisteriopsis caapi, both of which have traditionally been used for ritual and medicinal preparations in the Middle East, Central Asia, and South America (Sourkes, 1999). It is also present in common plant-derived foods and in human tissues (Guan et al., 2001). Known pharmacologic effects of harmine include hallucinogenesis (Sourkes, 1999), convulsive or anticonvulsive actions (Aricioglu et al., 2003), and tremor (Guan et al., 2001; Lutes et al., 1988). Several potential molecular targets for these central pharmacologic effects of harmine have been identified. These include MAO-A (Kim et al., 1997), 5-HT2A (Glennon et al., 2000), imidazoline receptors (I1 and I2 sites) (Husbands et al., 2001), and cyclin-dependent kinases (CDK1, 2, and 5) (Song et al., 2004). To our knowledge, activation of any of these targets does not lead to increased glutamate transporter expression.
Few compounds have been shown to activate increased glutamate transporter-EAAT2/GLT1 protein expression and function in vivo. Our new cellular screen has provided insight into a class of compounds that could serve as a starting point for new GLT1 activators. Given harmine’s CNS side effects and its multi-target activities, harmine itself is not a suitable candidate for immediate clinical applications. However, it may eventually be possible to separate its effects on GLT-1 upregulation and on CNS pharmacology through optimization of its chemical structure. Medicinal chemistry efforts to do so are ongoing, which will hopefully lead to a more potent and selective drug candidate that can be moved through the drug development process reaching from preclinical studies to clinical trials. These trials may include clinical applications such as ALS, but also other neurodegenerative disorders as well as psychiatric diseases.
We thank Ahmet Hoke for HIA cells and Clive Svendsen for HSC line. This work was supported by the Cinque Foundation and the Maryland Stem Cell Research Foundation. The findings from this study are dedicated to the memory of Steven Cinque.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.