Gene Expression-Based Screen for Parkinson's Disease Identifies

Jun 7, 2016 - Program in Neuroscience, Harvard Medical School, Boston, ... Department of Science and Mathematics, National Technical Institute for the...
0 downloads 4 Views 3MB Size
Letter pubs.acs.org/chemneuro

Gene Expression-Based Screen for Parkinson’s Disease Identifies GW8510 as a Neuroprotective Agent Nivanthika K. Wimalasena,†,‡,§ Viet Q. Le,⊥ Kandatege Wimalasena,∥ Stuart L. Schreiber,†,∇ and Rakesh Karmacharya*,†,‡,¶ †

Center for the Science of Therapeutics, Broad Institute of Harvard and MIT, Cambridge, Massachusetts 02142, United States Center for Experimental Drugs and Diagnostics, Psychiatric and Neurodevelopmental Genetics Unit, Center for Human Genetic Research, Harvard Medical School and Massachusetts General Hospital, Boston, Massachusetts 02114, United States § Program in Neuroscience, Harvard Medical School, Boston, Massachusetts 02115, United States ⊥ Department of Science and Mathematics, National Technical Institute for the Deaf, Rochester Institute of Technology, Rochester, New York 14623, United States ∥ Department of Chemistry, Wichita State University, Wichita, Kansas 67260, United States ¶ Schizophrenia and Bipolar Disorder Program, Harvard Medical School and McLean Hospital, Belmont, Massachusetts 02478, United States ∇ Howard Hughes Medical Institute, Department of Chemistry and Chemical Biology, Harvard University, Cambridge, Massachusetts 02138, United States ‡

S Supporting Information *

ABSTRACT: We carried out a gene expression-based in silico screen in order to identify small molecules with gene-expression profiles that are anticorrelated with a gene-expression profile for Parkinson’s disease (PD). We identified the cyclin-dependent kinase 2/5 (CDK2/5) inhibitor GW8510 as our most significant hit and characterized its effects in rodent MN9D cells and in human neuronal cells derived from induced pluripotent stem cells. GW8510 demonstrated neuroprotective ability in MN9D cells in the presence of 1-methyl-4-phenylpyridium (MPP+), a widely used neurotoxin model for Parkinson’s disease. In order to delineate the nature and extent of GW8510’s neuroprotective properties, we studied GW8510 in human neuronal cells in the context of various mechanisms of cellular stress. We found that GW8510 was protective against small-molecule mitochondrial and endoplasmic reticulum stressors. Our findings illustrate an approach to using small-molecule gene expression libraries to identify compounds with therapeutic potential in human diseases. KEYWORDS: GW8510, Parkinson’s disease, neuroprotection, induced pluripotent stem cells, iPSC, neural progenitor cells

P

phenotypes presents significant challenges for in vitro phenotypic screens. Therefore, a potential alternative approach to drug discovery utilizes gene expression-based highthroughput screening (GE-HTS) to discover compounds that modulate disease-related gene-expression patterns.9,10 Such GE-HTS screens have been used successfully to identify small molecules that modulate gene-expression signatures of interest, that is, the identification of small molecules that induce differentiation of acute myeloid leukemia cells,9 identification of cytosine arabinoside as a modulator of EWS/FLI in EWI sarcoma,10 and identification of aurintricarboxylic acid as a

arkinson’s disease (PD) is neurological disorder characterized by motor deficits, difficulties with coordination, and cognitive decline.1 The prevalence of PD in the population rises with age and is expected to double by 2030.2,3 PD causes dopaminergic neuronal death in the substantia nigra, a region in the midbrain, as well as α-synuclein aggregation into plaques.4 A number of genes have been implicated in familial cases of PD.5,6 However, most PD cases are sporadic and arise through a combination of unknown genetic and environmental factors.5 There is significant variability in efficacy of current treatments, and many patients do not have a robust response to current medications.7 Because known genes only account for about 10% of PD cases and the etiology of sporadic cases remains unclear, attempts to identify new therapeutic leads using a target-based screen is not feasible.8 In addition, lack of reliable and valid © 2016 American Chemical Society

Received: March 9, 2016 Accepted: June 7, 2016 Published: June 7, 2016 857

DOI: 10.1021/acschemneuro.6b00076 ACS Chem. Neurosci. 2016, 7, 857−863

Letter

ACS Chemical Neuroscience

Table 1. (A) Top 10 Small Molecule Hits in CMap That Were Anticorrelated with the Input PD Signature Profilea and (B) Top 20 Small Molecule Hits Sorted by p-Value after the Scores for All Experiments for a Given Compound Were Combined To Generate a Mean Score

a

The dose and the cell type that the data were generated from are shown in the table as well. The occurrence of our top hit GW8510 is highlighted in red.

gene-expression profiles with existing small-molecule geneexpression databases to identify such compounds. We undertook a study to identify small molecules that had gene-expression profiles that were anticorrelated to PD geneexpression profiles, with the rationale that normalizing aberrant gene expression to baseline levels may have a therapeutic effect. We curated a gene-expression profile for PD based on published studies and used that profile to query the Connectivity Map.13 Among the hits, GW8510 had the most highly significant anticorrelation with our compiled profile for PD. GW8510 is a 3′-substituted indolone and is known to inhibit cyclin-dependent kinases 2 and 5 (CKD2 and -5), which regulate the cell cycle.16,17 Based on the strong anticorrelation score and low p-value of GW8510, we undertook cellular studies to investigate its neuroprotective properties. GW8510 was protective against the immortalized mouse MN9D cell line in the presence of the neurotoxin MPP+, which is often used as a surrogate in vitro cellular PD model.18 We then examined potential neuroprotective effects of GW8510 in human neuronal cells derived from induced pluripotent stem cells

novel modulator of the gene-expression signature related to platelet derived growth factor receptor (PDGFR) activation.11 Since in vitro screening methods with GE-HTS require significant resources, we used published disease-related geneexpression profiles to interrogate a database that contained small-molecule gene-expression profiles.12 The Connectivity Map (CMap) contains detailed gene-expression data for thousands of bioactive small molecules generated from multiple experiments in a number of human cell lines.13 While the CMap database does not contain data from all cell types, it provides a good starting point to compile candidate small molecules that have gene-expression profiles of interest. Such an in silico approach has been used with gene-expression profiles of treated primary acute myelogenous leukemia (AML) cells to discover the small molecules celastrol and 4-hydroxy-2nonenal that proved to be effective in targeting AML stem cells.14 In the search for novel neuroprotective agents, there have been interesting discoveries with high-throughput screening of small molecule libraries using reporter assays.15 However, there have not been any studies that have combined the disease 858

DOI: 10.1021/acschemneuro.6b00076 ACS Chem. Neurosci. 2016, 7, 857−863

Letter

ACS Chemical Neuroscience

Figure 1. Effect of GW8510 on viability of MN9D cells in the presence of MPP+. (A) Cell viability as measured by MTT assay with varying concentrations of MPP+ in the presence or absence of 50 μM GW8510. (B) Cell viability as measured by MTT assay for cells in the presence of 150 μM MPP+ with varying concentrations of GW8510. The control lane was not treated with MPP+ or GW8510. (C) Depiction of percent change in viability with increasing concentration of GW8510 compared with the baseline viability in the presence of MPP+ alone. Error bars show standard error of the mean; * indicates P < 0.05, ** indicates P < 0.01, and *** indicates P < 0.005.

(iPSCs). GW8510 protected against MPP+ toxicity in a dosedependent manner, with protective effects at low nanomolar concentrations in human neuronal cells. We also studied a number of mitochondrial and endoplasmic reticulum (ER) stressors in order to delineate the nature and extent of GW8510s neuroprotective effects. These studies show that GW8510 was neuroprotective against MPP+ as well as mitochondrial and ER stressors in human iPSC-derived neural progenitor cells.

that were significantly up- or down-regulated in PD patients compared with healthy controls. We have listed the annotated genes and the direction and magnitude of the changes seen in the PD tissue in Supplementary Table 1. We included down regulation of DRD1−5 to optimize our query PD profile, given the known loss of dopaminergic neurons in PD. We used this curated PD gene-expression profile to query the Connectivity Map12,13 which calculated the correlations for gene expression profiles from small molecule experiments in the database in relation to the input profile. We rank-ordered the top 10 results based on their anticorrelation values (Table 1A). Other compounds of note were H-7, oxolinic acid, and hyoscyamine. Since the database contained results from multiple experiments of the same compound (Supplementary Table 1), we also averaged the scores of all the experiments for a given compound and rank-ordered the compounds by p-value (Table 1B). GW8510 had the lowest p-value and highest anticorrelation, with a correlation of coefficient of −0.925 (Table 1B). The GW8510 data in the Connectivity Map had been collated from four separate GW8510 experiments carried out at a concentration of 10 μM in two different cell lines: PC3, a human prostate cancer line, and MCF7, a human breast cancer cell line. In PC3 cells the anticorrelation scores from two experiments were −1 and −0.976, while in MCF7 cells the scores were −0.927 and −0.799 (Table 1A). While these results



RESULTS AND DISCUSSION We sought to identify small molecules that could potentially modulate the disease biology underlying PD. We took an approach of interrogating existing gene-expression data sets to identify small molecules that could modulate a biological signature of interest. In this case, we wanted to identify small molecules that had gene-expression profiles that were anticorrelated to PD gene-expression profiles. We tested our top candidate molecule in cellular assays in mouse and human neuronal cells against a number of cellular stressors to delineate the nature and extent of its neuroprotective properties. We collated a disease gene-expression profile by combining genes that were significantly up- or down-regulated in four published PD gene-expression studies.19−22 These studies, which included experiments in blood, prefrontal cortex, substantia nigra, and putamen, reported sets of 15−20 genes 859

DOI: 10.1021/acschemneuro.6b00076 ACS Chem. Neurosci. 2016, 7, 857−863

Letter

ACS Chemical Neuroscience are limited by the choice of cell line and dosage, they suggest that GW8510 may modulate the PD gene-expression profile. Second, the p-value of this anticorrelation was very significant (Table 1B), suggesting a strong link between the input PD profile and the experimentally measured GW8510 profile. Based on the hypothesis that a compound that modulates gene expression in the opposite direction of the gene expression changes observed in PD would be of therapeutic interest, we decided to further characterize GW8510 in cellular assays with neuronal cells. Since PD involves the death of dopaminergic neurons, we initially undertook studies in the MN9D cell line, a rodent dopaminergic cell line, in the presence of the neurotoxin MPP+, which is often used as a surrogate in vitro cellular disease model for PD.18 MN9D cells were incubated for 15 h with varying concentrations of MPP+ in the presence or absence of 50 μM GW8510, and cell viability was measured using the MTT assay.23 We found that preincubation with 50 μM GW8510 for 20 min significantly protected MN9D cells from toxicity induced by different concentrations of MPP+ (Figure 1A). We also found that GW8510 increased cell viability in a dosedependent manner in MN9D cells exposed to MPP+ (Figure 1B). We found that GW8510 accorded maximal protection at a dose of 25 μM, with a highly significant p-value of