Distinct mechanisms of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyrimidine resistance revealed by transcriptome mapping in mouse striatum

Abstract

The etiology of idiopathic Parkinson’s disease is thought to involve interplay between environmental factors and predisposing genetic traits, although the identification of genetic risk factors remain elusive. The neurotoxicant, 1-methyl-4-phenyl-1,2,3,6- tetrahydropyrimidine (MPTP) produces parkinsonian-like symptoms and pathology in mice and humans. As sensitivity to MPTP is genetically determined in mice this provides an opportunity to identify genes and biological mechanisms that modify the response to an exogenous agent that produces a Parkinson’s disease-like condition. MPTP primarily targets dopaminergic nerve terminals in the striatum and elicits changes in striatal gene expression. Therefore, we used Affymetrix® and real-time PCR technology to characterize temporal mRNA changes in striatum in response to MPTP in genetically MPTP-sensitive, C57BL/6J, and MPTP-resistant Swiss Webster and Bax−/− mice. We identified three phases of mRNA expression changes composed of largely distinct gene sets. An early response (5 hours) occurred in all strains of mice and multiple brain regions. In contrast, intermediate (24 hours) and late (72 hours) phases were striatum specific and much reduced in Swiss Webster, indicating these genes contribute and/or are responsive to MPTP-induced pathology. However, Bax−/− mice have robust intermediate responses. We propose a model in which the acute entry of MPP+ into dopaminergic nerve terminals damages them but is insufficient per se to kill the neurons. Rather, we suggest that the compromised nerve terminals elicit longer lasting transcriptional responses in surrounding cells involving production of molecules that feedback on the terminals to cause additional damage that results in cell death. In Swiss Webster, resistance lies upstream in the cascade of events triggered by MPTP and uncouples the acute events elicited by MPTP from the damaging secondary responses. In contrast, in Bax−/− mice resistance lies downstream in the cascade and suggests enhanced tolerance to the secondary insult rather than its attenuation.

Introduction

Parkinson’s disease (PD) is a progressive neurodegenerative disorder characterized by the loss of dopaminergic (DA) neurons in the substantia nigra pars compacta (SNpc). Although the etiology of idiopathic PD is unclear, it may originate from interplay between environmental agents (Landrigan et al., 2005, Brown et al., 2006) and predisposing genetic traits (Huang et al., 2004, Benmoyal-Segal and Soreq, 2006, Farrer, 2006, Wood-Kaczmar et al., 2006). However, the identification of specific genetic risk factors for idiopathic PD remains elusive.

The most widely studied experimental system of PD is the murine 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) model (Dauer and Przedborski, 2003, Smeyne and Jackson-Lewis, 2005). Although it does not recapitulate PD in its entirety, it reproduces several of its cardinal features including loss of DA neurons in the SNpc (Dauer and Przedborski, 2003, Przedborski and Vila, 2003, Smeyne and Jackson-Lewis, 2005). Moreover, as MPTP sensitivity in mice is genetically determined (Sundstrom et al., 1987, German et al., 1996, Hamre et al., 1999, Vila et al., 2001), it provides an opportunity to identify potential genetic risk factors for PD.

Several lines of evidence suggest that the primary site of injury in both PD and the MPTP model is the SNpc nerve terminals in the striatum (Bradbury et al., 1986, Herkenham et al., 1991, Nurmi et al., 2001, Rinne et al., 2001). Thus the initial event is damage of the synaptic terminals in the striatum, followed by retrograde degeneration and cell death (Bradbury et al., 1986, Eberling et al., 1997). This mechanism is particularly relevant for MPTP toxicity as its active metabolite, 1-methyl-4-phenylpyridinium (MPP+) selectively accumulates in DA nerve terminals in striatum via uptake through the dopamine transporter (Chiba et al., 1985, Javitch et al., 1985, Gainetdinov et al., 1997).

Shortly after its administration MPTP elicits the dumping of dopamine from SNpc nerve endings in striatum (Jackson-Lewis et al., 1995). This event coincides with the induction of several immediate-early genes (Duchemin et al., 1992, Smith et al., 1997, Perez-Otano et al., 1998, Agani et al., 2000, Chen et al., 2001). By 24 hours, depletion of striatal tyrosine hydroxylase (Kuhn et al., 2003, Sriram et al., 2004) and damage to DA synapses (Linder et al., 1995), coincide with induction of several genes, including hemeoxygenase-1 (Hmox1) (Fernandez-Gonzalez et al., 2000) and cytokines/chemokines (Nagatsu et al., 2000, Hebert et al., 2003, Sriram et al., 2004, Shen et al., 2005, Sriram et al., 2006, Pattarini et al., 2007). Subsequent neuronal death is progressive between 12 hours (Jackson-Lewis et al., 1995, Kuhn et al., 2003) and 7 days (Boyd et al., 2007). Based on this evidence, these early changes in gene expression could contribute to neuronal demise.

We hypothesized that sensitivity to MPTP may reside in genes expressed in striatum and their identification may point to genetic risk factors for PD.

Experimental Procedures

Animals and Experiments

Female C57BL/6J and B6.129X1-Bax^tm1Sjk/J (Bax +/−) mice of both genders were purchased from Jackson Laboratories (Bar Harbor, Maine, USA). Bax+/− mice were bred in-house and intercrossed to obtain homozygous knockout animals (Bax−/−) and wildtype littermates (Bax+/+). The genotype for Bax mice was performed by Transnetyx (Cordova, TN, USA). Female SWR mice were purchased from Jackson Laboratories and Harlan (Indianapolis, IN, USA). Animals were housed in micro-isolator units with a 12 hour light-dark schedule and constant temperature. Further details of the Materials and Methods are provided in Pattarini et al. (2007). MPTP was administered by IP injections at the dosage and schedule specified in the Results section. Animals were sacrificed at different time points after the first dose of MPTP and brain regions were dissected and immediately frozen on dry ice to preserve RNA integrity. Samples were stored at −80°C. All studies were approved by the St. Jude Children's Research Hospital Animal Care and Use Committee (ACUC) and were conducted in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals (NIH Publication No. 80-23, revised 1996). Efforts were made to minimize the number of animals involved in each experiment.

RNA Isolation

Total RNA was extracted with TRIzol® Reagent from Invitrogen (Carlsbad, California, USA) according to manufacturer instructions. Briefly, 1 ml TRIzol® was added to frozen samples and immediately homogenized. Samples were mixed with 200 µl of chloroform and centrifuged for 15 min at 12,000 × g at 4 °C. The aqueous phase was transferred to a new vial, mixed with 50 µg of glycogen from Roche Applied Science (Penzberg, Germany) and 500 µl of isopropanol and centrifuged (10 min at 12,000 × g, 4 °C) to precipitate total RNA. The pellet was washed with ice cold 70% ethanol and air dried for 10 min. Total RNA was resuspended in RNase-free water and it was checked for integrity by agarose gel electrophoresis. Samples that appeared degraded were discarded.

Preparation of Samples for Microarray Analysis

Technical procedures for microarray analysis, including quality control of RNA, labeling, hybridization and scanning of the arrays were performed by the Hartwell Center for Bioinformatics & Biotechnology (HC) at St. Jude Children's Research Hospital (SJCRH) according to standard operating procedures for Affymetrix protocols (GeneChip® Expression Analysis manual, Affymetrix, Santa Clara CA, USA). Prior to their use, RNA sample integrity was analyzed with the 2100 Bioanalyzer Lab-on-a-chip system (Agilent Technologies, Santa Clara, CA, USA). Total RNA samples that were not degraded were labeled using the Gene Chip IVT Labeling Kit (Affymetrix) according to manufacturer instructions. Briefly, as a quality control of the labeling process, samples were first spiked with the GeneChip® Poly-A RNA Control Kit (Affymetrix) that contains mRNA (not present in eukaryotic cells) for the following B. subtilis genes: lys (1:100,000 ratio of copy number), phe (1:50,000), thr (1:25,000) and dap (1:7,500). Samples were then used to prepare the 1^st strand cDNA using the One-Cycle cDNA Synthesis Kit (Affymetrix) containing SuperScript II followed by the 2^nd strand cDNA synthesis with T4 DNA polymerase. cDNA was cleaned using cDNA Cleanup Spin Column (Affymetrix), and biotin-labeled cRNA was prepared using the Gene Chip IVT Labeling Kit (Affymetrix). Labeled cRNA was purified with Cleanup Spin Column (Affymetrix), quantified, fragmented and spiked with biotin-labeled cRNA for bioB (1,5 pM), bioC (5 pM), bioD (25 pM) and Crex (100 pM) (GeneChip® Eukaryotic Hybridization Control Kit, Affymetrix). This procedure allowed us to asses both the linearity of detection and the lowest accurately detectable concentration (1.5 pmol). Samples were loaded onto the Affymetrix® Mouse Genome 430 2.0 Arrays (Affymetrix) previously washed with 1X hybridization buffer (100 mM MES, 1M Na⁺, 20 mM EDTA, 0.01% Tween-20 pH 6.6) and hybridized overnight (16 hours) at 37 °C. Arrays were washed and stained with streptavidin conjugated to phycoerthyrin, using the automated GeneChip® Fluidics Station 400 (Affymetrix) and scanned to produce an image file with the GeneArray™ scanner (Affymetrix). Total RNA from each animal was loaded onto individual Affymetrix microarray chips. Experimental reproducibility can be estimated by comparing columns within a figure as well as between corresponding columns in Figure 1, Figure 3, Figure 4, Figure 6, and Figure 8.

Figure 1. Temporal profile of mRNA responses to MPTP in the striatum of C57BL/6J mice.

Animals were injected with either MPTP (20 mg/kg) or saline (control) every two hours for a total of 4 injections. Striata were harvested at 5 (3 injections), 24 and 72 hours. Total RNA was hybridized to the Affymetrix Mouse Genome 430 2.0 array and results analyzed as described in Materials and Methods. Hierarchical cluster analysis reveals three discrete cohorts of probesets modulated at each time point. Only a small fraction of probesets are regulated at multiple time points. Each vertical column represents a single mouse treated as indicated in the figure and each horizontal row is an individual probeset. Probesets that are upregulated in treated compared to control mice appear in red, those that are downregulated appear in green. The relative log2 (ratio) is reflected by the intensity of the color. A total of 47 animals, 23 treated with saline and 24 treated with MPTP, were used. The data reveals three general phases of gene expression changes (early, intermediate and late), that are constituted by relatively unique clusters of probesets.

Figure 3. Hierarchical cluster analysis reveals differences between the early (5 hr) and intermediate (24 hr) MPTP-induced responses in striatum, cortex and cerebellum of C57BL/6J mice.

Animals were injected with either MPTP (20 mg/kg) or saline (control) every two hours for a total of 4 injections and striatum, cortex and cerebellum harvested at 5 (3 injections) and 24 hours. Total RNA was hybridized to the Affymetrix Mouse Genome 430 2.0 array and results analyzed as described in Materials and Methods. In this hierarchical cluster analysis each vertical column represents a single mouse treated as indicated on top of the figure and each horizontal row is an individual probeset. Probesets that are upregulated in treated compared to control mice appear in red, those that are downregulated appear in green. The relative log2 (ratio) is reflected by the intensity of the color. A total of 10 animals, 4 treated with saline and 6 treated with MPTP, were used. The visual pattern generated by the hierarchical cluster analysis program depends on the number and type of samples used for the analysis. Therefore, similar time points, may display visually different patterns in different figures although the genes within the clusters are identical. Whereas the early phase occurs to various extents in all three brain regions, the intermediate phase is restricted to the striatum.

Figure 4. The MPTP-induced intermediate phase, but not the early phase differ between MPTP-sensitive (C57BL/6J) and MPTP-resistant (SWR) strains of mice.

C57BL/6J and SWR mice were injected with either MPTP (20 mg/kg) or saline (control) every two hours for a total of 4 injections. Striatum was harvested at 5 (3 injections) and 24 hours, and total RNA was isolated and hybridized onto the Affymetrix Mouse Genome 430 2.0 array. The expression profile of the probesets previously identified in C57BL/6J mice (Figure 1 and Supplemental table 1) was compared to a new cohort of MPTP-sensitive (C57BL/6J) mice as well as to a cohort of MPTP-resistant (SWR) mice at 5 and 24 hours after treatment to generate this heat map. Each vertical column represents a single mouse treated as indicated on top of the figure and each horizontal row is an individual probeset. Probesets that are upregulated in treated compared to control mice appear in red, those that are downregulated appear in green. The relative log2 (ratio) is reflected by the intensity of the color. A total of 19 C57BL/6J and 24 SWR mice (9 C57BL/6J and 11 SWR treated with saline; 10 C57BL/6J and 13 SWR treated with MPTP) were used to perform the experiments described in this figure. The visual pattern generated by the hierarchical cluster analysis program depends on the number and type of samples used for the analysis. Therefore, similar time points, may display visually different patterns in different figures although the genes within the clusters are identical. Whereas the early phase is indistinguishable in MPTP-sensitive (C57BL/6J) and MPTP-resistant (SWR) strains of mice, the intermediate phase is attenuated in resistant SWR mice.

Figure 6. Hierarchical cluster analysis of basal gene expression differences in striata of MPTP-sensitive (C57BL/6J) and MPTP-resistant (SWR) mice.

Striatum from untreated animals of both MPTP-sensitive (C57BL/6J) and MPTP-resistant (SWR) mice were harvested and RNA extracted and hybridized onto the Affymetrix Mouse Genome 430 2.0 array to identify genes with differential expression between the two strains. Each vertical column represents a single mouse of the strain indicated on top of the figure (a total of 18 animals, 9 for each strain, were used) and each horizontal row is an individual probeset. Probesets for one strain that are elevated with respect to the other strain appear in red whereas those that are reduced appear in green. The relative log2 (ratio) is reflected by the intensity of the color. 364 probesets are differentially expressed between sensitive (C57BL/6J) and resistant (SWR) strains of mice.

Figure 8. The intermediate response in Bax knockout mice does not differ from that in wild type littermates.

Bax null mice on a C57BL/6J background and wild type littermates were injected with either MPTP (20 mg/kg) or saline (control) every two hours for a total of 4 injections. Striatum was harvested at 24 hours, and total RNA isolated and hybridized onto the Affymetrix Mouse Genome 430 2.0 array. The expression profile of the probesets previously identified in C57BL/6J mice (Figure 1 and Supplemental table 1) was compared in a cohort of Bax−/− and wild type littermates 24 hours after treatment with MPTP to generate this heat map. Each vertical column represents a single mouse treated as indicated on top of the figure and each horizontal row is an individual probeset. Probesets that are upregulated in treated compared to control mice appear in red, those that are downregulated appear in green. The relative log2 (ratio) is reflected by the intensity of the color. A total of 20 animals (3 female and 3 male wild type and 2 female and 2 male Bax−/− mice were treated with saline and 5 males of each genotype were treated with MPTP) were used to perform these experiments. The transcriptional response did not differ between the two genotypes, although Bax−/− mice are resistant to MPTP.

Analysis of Microarrays

The microarrays used in this study (Mouse Genome 430 2.0 Arrays, Affymetrix) contain 45,101 probesets, representing 39,000 transcripts and variants, and they are currently the most comprehensive genechip array available for the mouse. Scanned images were analyzed with the Gene Chip Operating Software (GCOSv1.2, Affymetrix). Assessment of probeset present/absent calls was made using the Single Array Analysis method in GCOS using the statistical algorithm with default analysis parameters (http://www.affymetrix.com/support/technical/whitepapers/sadd_whitepaper.pdf). Probeset signal values were scaled by global methods to a target value of 500.

Array analysis was performed using Spotfire® DecisionSite 8.2 from TIBCO Software Inc. (Palo alto, California, USA). The following is a brief description of the microarray data analysis procedure. First, probesets that are “Absent” across all samples were excluded (McClintick and Edenberg, 2006). The remaining probeset signals were variance-stabilized by addition of a small constant value equal to half of the average background signal (Rocke and Durbin, 2003). Variance-adjusted signals were log2-transformed and used in the Student's t-test (two groups) or the ANOVA method (>2 groups) to identify differences in probeset expression. Probesets that satisfied the thresholds for false discovery rate < 0.05 (Benjamini and Hochberg, 1995) and fold-change (≥1.5 or ≤0.667) were selected. To identify patterns of co-regulated gene expression, the log2-transformed signals were normalized across samples to a mean of zero and a standard deviation of one (Z-score). This procedure enables comparison of changes within the same relative magnitude. Normalized signals were analyzed by an agglomerative hierarchical clustering algorithm using the Euclidean distance and UPGMA (unweighted average) methods (Quackenbush, 2001, Butte, 2002).

Gene Set Enrichment Analysis

In addition to identifying the differentially expressed genes with an arbitrary cutoff from t-test followed by multiple test correction, we also compared treated samples with untreated ones at each time point using all the probesets on the array with the permutation approach. We used the R-version of a publicly available program, Gene Set Enrichment Analysis (GSEA) (Mootha et al., 2003, Subramanian et al., 2005). GSEA is a computational method that determines whether an a priori defined set of genes shows statistically significant, concordant differences between two biological states. We used gene sets for canonical pathways compiled by Ingenuity Pathway Analysis (http://www.ingenuity.com) for pathway analysis and motif gene sets from the Molecular Signature Database (Subramanian et al., 2005) for transcription factor analysis. Motif gene sets contain genes that share a cis-regulatory motif that is conserved across the human, mouse, rat and dog genomes. The motifs are catalogued in Xie et al (2005) and represent known or likely regulatory elements in promoters and 3’-UTRs. Only results with a value of false discovery rate (q) ≤ 0.25 were considered.

Validation of Microarray Data by Quantitative RT-PCR

Total RNA was reverse transcribed using TaqMan® reverse transcription reagents from Applied Biosystems (Foster City, California, USA). Primers and probes for real-time PCR (qRT-PCR) were designed with Primer Express Software version 1.5 (Applied Biosystems) and synthesized by the HC. Real time PCR was performed using TaqMan® PCR Core Reagent Kit (Applied Biosystem), using the ABI Prism 7900HT system (Applied Biosystem). Absolute quantification was performed using standard curves for each gene of interest. Primers and probes used for qRT-PCR are listed in Table 1.

Table 1. List of primers and probes for qRT-PCR.

Gene Name	mRNA ID #	Forward Primer	Probe	Reverse Primer
Agxt2l1	NM_027907	5’-CCCTGGCGCTGAGGG-3’	5’-Fam-CACATCGGGCCCTCGTGCAA-BHQ-3’	5’-TGGGATCCGCGGCA-3’
Apod	NM_007470	5’-AACGTCTCAGAGCCAGCCA-3’	5’-Fam-AAGTCCAGTTCTTCCCGTTGATGCCA-BHQ-3’	5’-AGGATCCAGTAGGGTGCCG-3’
Cdkn1a	NM_007669	5’-TTCCGCACAGGAGCAAACT-3’	5’-Fam-CCGTTGTCTCTTCGGTCCCGTGG-BHQ-3’	5’-CGGCGCAACTGCTCACT-3’
Comt	NM_007744	5’-AGAGAAGGAGTGGGCCATGA-3’	5’-Fam-CGTGGGTGACGCAAAAGGCCAA-BHQ-3’	5’-CCGAATCACTGCATCCATGA-3’
Edg3	NM_010101	5’-AACCAGCCCAGATGCGC-3’	5’-Fam-TTGCAGAACGAGAGCCTATTTTCAACACTCTTC-BHQ-3’	5’-GCGTGCAGGCCCGAC-3’
Nfkbia	NM_010907	5’-CAGCTCACGGAGGACGGA-3’	5’-Fam-ACTCGTTCCTGCACTTGGCAATCATCC-BHQ-3’	5’-ATGGTCAGCGGCTTCTCTTC-3’
Fosb	NM_008036	5’-CCAGAGCCAGGCCTAGAAGAC-3’	5’-Fam-TCGCAGAGAGCGGAACAAGC-BHQ-3’	5’-CCGACGGTTCCTGCACTTA-3’
Gadd45b	NM_008655	5’-AGGCGGCCAAACTGATGA-3’	5’-Fam-TGTGGACCCCGACAGCGTGG-BHQ-3’	5’-CATCCTCCTCTTCTTCGTCTATGG-3’
Gadph	NM_001001303	5’-TGGATCTGACGTGCCGC-3’	5’-Fam-TGGAGAAACCTGCCAAGTATGATGACATCA-BHQ-3’	5’-TGCCTGCTTCACCACCTTC-3’
Hbegf	NM_010415	5’-TGCTGCCGTCGGTGATG-3’	5’-Fam-TGAAGCTCTTTCTGGCCGCAGTGTTG-BHQ-3’	5’-ACCGGTCACCAACGCG-3’
Hmox1	NM_010442	5’-GTGATGGAGCGTCCACAGC-3’	5’-Fam-CGACAGCATGCCCCAGGATTTGTC-BHQ-3’	5’-TGGTGGCCTCCTTCAAGG-3’
Kcnj10	NM_001039484	5’-CTCCGCTCGCCGCTC-3’	5’-Fam-CAGCCACTTCACCTTCGAGCCAAGA-BHQ-3’	5’-ACTGTAATAGACCTTAGCGACCGAC-3’
Nr4a1	NM_010444	5’-GGCCACAGGGAGTGGGA-3’	5’-Fam-CCGGCTGGAGATGCCCTGTATTCAAG-BHQ-3’	5’-CGTTGCTGGTGTTCCATATTGA-3’
Pdlim4	NM_019417	5’-AGTGCACGCGCTGCG-3’	5’-Fam-ACGGGATCGTGGGAACCATTGTCA-BHQ-3’	5’-ATGGTAGAGCTTGTCTCTCGCC-3’
Pink1	NM_026880	5’-GAGCGTGGTGGCAATGG-3’	5’-Fam-TGATGGCCCCTGAGGTGTCCACA-BHQ-3’	5’-CGGCTTTGCTGTAGTCAATTACC-3’
Tnrsf12a	NM_013749	5’-GCCGCCGGAGAGAAAAG-3’	5’-Fam-TTACTACCCCCATAGAGGAGACTGGTGGAGAG-BHQ-3’	5’-GCCACACCTGGGCAGC-3’

Standards were prepared by cloning the coding sequence (CDS) of each gene into a pcDNA3 plasmid (Invitrogen) as previously described (Pattarini et al., 2007). The primers used to prepare the standards, including the restriction site used are listed in Table 2.

Table 2. List of primers used to prepare the standards for qRT-PCR.

The restriction enzyme (RE) used is indicated in each column preceding the primer sequence. The sequence recognized by the RE is underlined.

Gene Name	Enzyme	Forward primer	Enzyme	Reverse primer
Agxt2l1	HindIII	5’-CCCAAGCTTGGGATGTGCGAGCTCTATAGCAA-3’	XhoI	5’-CCGCTCGAGCGGCTCATGTCTTGAGTCTTTTGCT-3’
Apod	KpnI	5’-CGGGGTACCCCGATGGTGACCATGCTGATGTTCC-3’	EcoRI	5’-CCGGAATTCCGGTTACAGGAAGTCCGGGCAGTT-3’
Cdkn1a	HindIII	5’-CCCAAGCTTGGGATGTCCAATCCTGGTGATGT-3’	EcoRI	5’-CCGGAATTCCGGTCAGGACTTCACGGGGCTGC-3’
Comt	KpnI	5’-CGGGGTACCCCGATGCTGTTGGCTGCTGTCTC-3’	EcoRI	5’-CCGGAATTCCGGCTTCAGGGTTTTCTCTTGCA-3’
Edg3	EcoRI	5’-CCGGAATTCCGGATGGCAACCACGCATGCGCA-3’	XhoI	5’-CCGCTCGAGCGGTCACTTGCAGAGGACCCCGT-3’
Fosb	BamHI	5’-CGCGGATCCGCGGTGAAACCGACAGAGCCTGG-3’	EcoRI	5’-CCGGAATTCCGGGTTCCTTGCGGGTTTGTTTG-3’
Gadd45b	BamHI	5’-CGCGGATCCGCGATGACCCTGGAAGAGCTGGT-3’	EcoRI	5’-CCGGAATTCCGGTGGGTCTCAGCGTTCCTCTA-3’
Gapdh	BamHI	5’-CGCGGATCCGCGATGGTGAAGGTCGGTGTGAA-3’	EcoRI	5’-CCGGAATTCCGGTTCTTACTCCTTGGAGGCCA-3’
Hbegf	HindIII	5’-CCCAAGCTTGGGATGAAGCTGCTGCCGTCGGT-3’	EcoRI	5’-CCGGAATTCCGGATAGCTCAGGTCCTCCTCAGTGGG-3’
Hmox1	BamHI	5’-CGCGGATCCGCGTAGCCCAGTCCGGTGATGGA-3’	EcoRI	5’-CCGGAATTCCGGTGGGGGCCAGTATTGCATTT-3’
Kcnj10	BamHI	5’-CGCGGATCCGCGCTCCGCTCGCCGCTCCTGCC-3’	EcoRI	5’-CCGGAATTCCGGTCAGACGTTGCTGATGCGCA-3’
Nfkbia	HindIII	5’-CCCAAGCTTGGGAGCCATGTTTCAGCCAGCTG-3’	XhoI	5’-CCGCTCGAGCGGTTATAATGTCAGACGCTGGC-3’
Nr4a1	BamHI	5’-CGCGGATCCGCGTGCTAGAAGGACTGCGGAGC-3’	EcoRI	5’-CCGGAATTCCGGGGCTTAAAGGCACATGGGTG-3’
Pdlim4	HindIII	5’-CCCAAGCTTGGGATGACCCACTCGGTGACCCTG-3’	EcoRI	5’-CCGGAATTCCGGCAGCTCAGACAAGTTCCACCT-3’
Pink1	KpnI	5’-CGGGGTACCCCGATGGCGGTGCGACAGGCACTG-3’	EcoRI	5’-CCGGAATTCCGGTCATGGGGCTGCCCTCCAGGA-3’
Tnfrsf12a	BamHI	5’-CGCGGATCCGCGGCAATCATGGCTCCGGGTTG-3’	EcoRI	5’-CCGGAATTCCGGTGAATCACCACCTCGCCCCA-3’

Statistical Analysis

Statistical methods used to analyze microarray results are explained in the microarray analysis section. Statistical analysis for qRT-PCR results was performed with GraphPad Prism® version 4.03 for Windows® (GraphPad Software Inc., San Diego, CA). Results are expressed as the ratio of number of copies of a specific gene over the number of copies of glyceraldehyde-3-phosphate dehydrogenase (Gapdh). Each time point is the average of at least three animals. The temporal profile of each gene was analyzed by one-way ANOVA followed by Bonferroni’s multiple comparisons test to assess statistical significance versus respective control (time zero). Comparison between strains was performed either by two-way ANOVA followed by Bonferroni’s multiple comparisons test (> 2 groups) or by Student’s t-test (2 groups).

Results

The MPTP striatal transcriptome in C57BL/6J mice

To investigate the temporal transcriptional responses in the striatum in MPTP-sensitive strain, C57BL/6J mice were injected every two hours with either saline (as control) or MPTP (20 mg/kg) for a total of four injections. This injection schedule, sometimes referred to as the acute MPTP model, is used widely to simulate PD in mice and leads to a temporally predictable sequence of molecular and cellular events that culminate in the relatively synchronous death of SNpc neurons (Przedborski and Vila, 2003, Miller et al., 2005, Smeyne and Jackson-Lewis, 2005). Animals were sacrificed at 5 (three injections), 24 and 72 hours (both four injections) after the first dose of MPTP, the striatum removed and total RNA isolated and used for Affymetrix microarray analysis as described in Materials and Methods. These time points were chosen to span the period from the acute consequences of MPTP intoxication through times when formal perturbation of DA nerve endings and compromised SNpc neuron function are evident up to the point when DA neurons begin to die. Total RNA from each animal was loaded onto individual Affymetrix microarray chips. Experimental reproducibility can be estimated by comparing columns within a figure as well as between corresponding columns in Figure 1.

Using criteria described in the Material and Method section, we identified 107, 287 and 191 probesets as differentially expressed at 5, 24 and 72 hours, respectively (Supplemental Table 1). This target list of probesets was used to perform hierarchical cluster, Gene Ontology and Ingenuity Pathway analyses. Hierarchical cluster analysis reveals three largely discrete sets of genes whose mRNA levels change sequentially over time following MPTP administration (Figure 1 and Supplemental Table 1). At early time points (Figure 1), the levels of mRNA for a number of genes increase and then largely decline to basal values by 24 hours (Figure 1 and Supplemental Table 1). By 24 hours a distinct and larger set of mRNAs are increased (Figure 1) and then largely decline to baseline by 72 hours at which time a new set of gene expression changes are evident (Figure 1 and Supplemental Table 1). Although less in number, there were also transient decreases in some mRNAs over the same time course (Supplemental Table 1). Additional microarray data was obtained at 2 and 96 hours post-MPTP treatment. Gene expression changes seen at 2 and 96 hours were subsets of those seen at 5 and 72 hours, respectively (data not shown).

A variety of bioinformatic tools were used to analyze mRNA changes. As expected from prior studies (Duchemin et al., 1992, Doucet et al., 1996, Smith et al., 1997, Perez-Otano et al., 1998, Hunot et al., 2004), immediate early genes are prominent in the early phase following MPTP treatment and include the transcriptional regulators early growth response 3 and 4 (Egr3 and Egr4, respectively), FBJ osteosarcoma oncogene (c-Fos), FBJ osteosarcoma oncogene B (Fosb), Jun oncogene (c-Jun) and Jun oncogene B (Junb) (Supplemental Table 1). Moreover, Gene Ontology analysis revealed that transcription factors/regulators were over-represented in the early phase (20.9%) versus intermediate (5.3%) and late phases (6.3%). Similarly, Ingenuity Pathway Analysis (Ingenuity System Inc., Redwood City, CA, USA) indicated that gene expression was statistically the most over-represented function (2.28e-08 < P < 8.72e-03) in the early phase response. Examples include changes in mRNA levels for the transcriptional regulators BTB and CNC homology 1 (Bach1), B-cell translocation gene 3 (Btg3), CCAAT/enhancer binding protein (C/EBP), beta (Cebpb), Kruppel-like factor 9 (Klf9), nuclear receptor subfamily 4, group A, member 3 (Nr4a3), paired box gene 8 (Pax8), retinoid X receptor gamma (Rxrg), superoxide dismutase 2 (Sox2) and zinc finger and BTB domain containing 16 (Zbtb16).

Another major component of the early response involves genes implicated in oxidative stress and includes cyclin-dependent kinase inhibitor 1A (Cdkn1a), DNA-damage-inducible transcript 4 (Ddit4), DNA-damage-inducible transcript 4-like (Ddit4l), FK506 binding protein 5 (Fkbp5), growth arrest and DNA-damage-inducible 45 beta and gamma (Gadd45b and Gadd45g), metallothionein 2 (Mt2), nuclear factor of kappa light chain gene enhancer in B-cells inhibitor, alpha (NFkBia) and uncoupling protein 2 (Ucp2). These changes are also consistent with studies in PD and models of the disorder where evidence of oxidative stress have been reported (Kindt et al., 1987, Krueger et al., 1990, Bates et al., 1994, Hasegawa et al., 1995, Beal, 2003, Przedborski and Vila, 2003, Hald and Lotharius, 2005, Jackson-Lewis and Smeyne, 2005, Hald et al., 2007, Schapira, 2008). Other gene expression changes in the early phase represent inflammatory responses (Tnf receptor-associated factor 5 (Traf5) and interferon-related developmental regulator 1 (Ifrd1)) and steroid/stress signaling (serum/glucocorticoid regulated kinase 1 and 3 (Sgk and Sgk3), TSC22 domain family, members 1 and 3 (Tsc22d1 and Tsc22d3)).

Gene set enrichment analysis (GSEA) (Mootha et al., 2003, Subramanian et al., 2005, Subramanian et al., 2007) revealed that the intermediate phase is characterized by enrichment for transcripts implicated in cytokine signaling and inflammatory responses (Supplemental Table 1 and Supplemental Table 6). This result is consistent with many studies showing the presence of inflammatory responses in striatum in both PD and animal models thereof (Hald and Lotharius, 2005, Nagatsu and Sawada, 2005, Hald et al., 2007, Hunter et al., 2007, McGeer and McGeer, 2007, Wilms et al., 2007). Expression of genes involved in TNF family signaling such as the receptors for TNF-alpha (Tnfrsf1a) and Tweak (Tnfsf12a) are increased. Likewise, expression of genes involved in interleukin signaling pathways such as suppressor of cytokine signaling 3 (Socs3) and signal transducer and activator of transcription 1 and 3 (Stat1 and Stat3) are elevated. Besides genes involved in cytokine and chemokine signaling, many effector molecules of the inflammatory response are increased in the intermediate phase, including the complement components 1, q subcomponent, alpha and beta polypeptide and C chain (C1qa, C1qb and C1qc, respectively), and complement component 4B (C4b); Fc receptor, IgG, high affinity I (Fcgr1); cathepsin B, C, D and Z (Ctsb, Ctsc, Ctsd and Ctsz, respectively); lectin, galactose binding, soluble 1 and 3 (Lgals1 and Lgals3, respectively) and the Lgals3 binding protein (Lgals3bp). Similarly, markers of inflammatory and immune cells such as allograft inflammatory factor 1 (Aif1 a.k.a. Iba1, a microglia marker), CD antigens 44, 68, 151 and 180 (Cd44, Cd68, Cd151 and Cd180, respectively), lymphocyte antigen 86 (Ly86), macrophage scavenger receptor 2 (Msr2) and oncostatin M receptor (Osmr) change in the intermediate phase.

Also prominent in the intermediate phase are increased transcript levels for genes associated with activation of astrocytes, including glial fibrillary acidic protein (Gfap) and vimentin (Vim). We also, confirm our earlier demonstration of elevated Hmox1 expression in striatal astrocytes following MPTP administration (Fernandez-Gonzalez et al., 2000). Although not a specific marker for gliosis, the levels of S100 calcium binding proteins A6, A10, A11, A13 and A16 (S100a6, S100a10, S100a11, S100a13 and S100a16, respectively) as well as their interacting proteins, annexin A2 and A3 (Anxa2 and Anxa3, respectively) (Bianchi et al., 1992, Tokumitsu et al., 1992, Rety et al., 1999, Santamaria-Kisiel et al., 2006, Rescher and Gerke, 2008) are also increased in the intermediate phase. In addition, a number of other gene products associated with protein folding, modification and elimination, such as heat shock protein 1, B6 and 8 (Hspb1, Hspb6 and Hspb8, respectively), transglutaminase 1, K and C polypeptides (Tgm1 and Tgm2, respectively) and tissue inhibitor of metalloproteinase 1 (Timp1) are elevated. Also indicative of ongoing responses to cellular damage and oxidative stress are elevation in levels of mRNAs for apolipoprotein D (Apod), fatty acid binding protein 7 (Fabp7) and metallothionein 2 (Mt2). In addition mRNA levels of genes linked with cell death such as myeloid cell leukemia sequence 1 (Mcl1) and transmembrane BAX inhibitor motif containing 1 (Tmbim1) and macroautophagy Bcl2-associated athanogene 3 (Bag3) (Carra et al., 2007) change in the intermediate phase. Besides gene products overtly linked to inflammation, gliosis, and cellular damage and stress responses, expression of genes involved in other signaling pathways change, including bone morphogenetic protein 1 (Bmp1), BMP2 inducible kinase (Bmp2k), CD9 antigen (Cd9), heparin-binding EGF-like growth factor (Hbegf) and transforming growth factor, beta receptor II (Tgfbr2).

By 72 hours post-treatment the majority of the mRNA changes seen at 24 hours return to basal levels and a new cohort of transcripts are altered. The persistently altered mRNAs are those linked to gliosis, inflammation and oxidative stress and include, Gfap, Vim, C1qc and C4b, lymphocyte antigen 86 (Ly86), endothelin receptor type B (Ednrb), heat shock protein 6 (Hspb6), lectin, galactose binding, soluble 1 (Lgals1) and lectin, galactoside-binding, soluble, 3 binding protein (Lgals3bp), lysosomal-associated membrane protein 2 (Lamp2), legumain (Lgmn), metallothionein 1 (Mt1), S100 calcium binding protein A6 and A13 (S100a6 and S100a13, respectively), and transferrin (Trf). The same inflammation/gliosis-related mRNAs are also elevated at 96 hours post treatment indicating persistent inflammatory responses and ongoing astrogliosis in striatum (data not shown).

In the late phase, a new cluster of gene expression changes is evident. Several immediate-early genes including early growth response 3 (Egr3) and Fos-like antigen 2 (Fosl2) are down-regulated at 72 and 96 hours. The mRNA levels for the transcription factor ets variant gene 5 (Etv5) and for brain-specific angiogenesis inhibitor 1-associated protein 2 (Baiap2), a presumptive immediate-early gene are also persistently decreased whereas levels of the transcriptional regulators activating transcription factor 6 (Atf6), nuclear receptor subfamily 2, group F, member 2 (Nr2f2) and zinc finger protein of the cerebellum 1 (Zic1) are increased. The mRNAs levels for many membrane and secreted proteins or proteins that modify the extracellular matrix also change at 72 hours and include aquaporin 4 (Aqp4), gap junction membrane channel protein alpha 1 (Gja1 a.k.a. connexin 43), myelin oligodendrocyte glycoprotein (Mog), neural cell adhesion molecule 1 (Ncam1), proteolipid protein 1 (Plp1), solute carrier family 44, member 1 (Slc44a1), secreted acidic cysteine rich glycoprotein (Sparc), secreted phosphoprotein 1 (Spp1) and tissue inhibitor of metalloproteinase 4 (Timp4). Also prominent are changes in expression of genes associated with specific neuronal subtypes and include, parvalbumin (Pvalb), potassium voltage-gated channel, subfamily Q, member 5 (Kcnq5), and the gamma-aminobutyric acid (GABA) transporter solute carrier family 6, member 11, (Slc6a11), as well as general neuronal proteins such as bassoon (Bsn) and homer homolog 1 (Homer1). Finally, the mRNAs encoding two proteins implicated in PD, alpha-synuclein (Snca) (decrease) and G protein-coupled receptor 37 (Gpr37 a.k.a. Pael-R) (increase) are altered in the late response phase. Moreover, the same changes in these two transcripts are also evident at 96 hours suggesting that the latter two are more long-lasting alterations in gene expression (data not shown).

Assessment of temporal mRNA changes by qRT-PCR

To confirm and extend the microarray data, qRT-PCR was used to assess the temporal profiles of mRNA expression of selected genes representative of early (Cdkn1a, NFkBia, Gadd45b, Fosb and nuclear receptor subfamily 4, group A, member 1 (Nr4a1)) and intermediate (Hmox1, Tnfrsf12a, endothelial differentiation, sphingolipid G-protein-coupled receptor 3 (Edg3), PDZ and LIM domain 4 (Pdlim4) and Hbegf) phase transcripts (Figure 2). Early phase mRNAs increased between 2–5 hours post-MPTP treatment and declined to baseline by 24 hours. The only exception was Gadd45b that showed a small but statistically significant increase at 24 hours. The intermediate phase response transcripts increased between 12–24 hours post-MPTP treatment and declined to baseline by 3 days. These data serve to confirm and extend the microarray analysis.

Figure 2. Quantitative assessment of mRNA changes in the striatum of MPTP-treated C57BL/6J mice.

MPTP sensitive (C57BL/6J) mice were injected at time zero with a single dose of MPTP (40 mg/kg) and sacrificed at 2, 5, 12 and 24 hr and 3, 7, 8, 9, 10 and 11 days. Levels of mRNA for candidate genes of the early (Cdkn1a, NFkBia, FosB, Nr4a1 and Gadd45b) and intermediate (Hmox1, Pdlim4, Edg3, Tnfrsf12a and Hbegf) response phases were evaluated by quantitative RT-PCR. Results are expressed as the number of copies of a specific mRNA (normalized versus the number of copies of Gapdh mRNA) versus time (hours). Data are presented as mean ± S.E.M. of 6 (control) and 3 (MPTP-treated) animals. Differences versus control (time point zero) were analyzed with one way ANOVA and Bonferroni post-hoc test and are indicated with asterisks at the top of each significantly different point (*** P<0.001, ** P<0.01, * P<0.05). Results are in accordance with those obtained by Affymetrix gene chip technology.

Brain region specificity of MPTP-induced mRNA changes

We showed previously that Hmox1 induction was confined to the striatum following MPTP treatment (Fernandez-Gonzalez et al., 2000). Therefore, we assessed whether expression of other genes detected in the initial microarray screen were also specifically altered in striatum. Animals were injected with MPTP (20 mg/kg) every two hours for a total of 4 injections, and sacrificed at 5 and 24 hours after the first injection and global mRNA levels in striatum, cerebral cortex and cerebellum assessed using Affymetrix microarray. Total RNA from each animal was loaded onto individual Affymetrix microarray chips. Experimental reproducibility can be estimated by comparing columns within a figure as well as between corresponding columns in Figure 3. A transient early phase of gene expression changes was evident in all three brain areas (Figure 3). However, the response was most prominent in striatum both in regards to the number of genes involved and magnitude of the changes.

In marked contrast to the early phase, expression of the intermediate phase response genes was essentially unique to the striatum (Figure 3). Moreover, there was not a different set of genes to those identified in striatum whose expression changed in cerebral cortex and cerebellum following MPTP treatment (data not shown). Therefore, there is a highly coordinated and stereotypical transcriptional response triggered by MPTP administration that is spatially and temporally restricted to the brain region that is the acute target of the neurotoxin. The visual pattern generated by the hierarchical cluster analysis program depends on the number and type of samples used for the analysis. Therefore, similar time points, may display visually different patterns in different figures although the genes within the clusters are identical.

The MPTP-induced transcriptome in the striatum of sensitive and resistant strains of mice

To establish the potential relevance of the mRNA changes observed in the striatum to the pathology elicited by MPTP we compared mRNA profiles in MPTP-sensitive (C57BL/6J) and - resistant (SWR) strains of mice. Animals of both strains were injected every two hours with either saline or MPTP 20 mg/kg for a total of 4 doses. Mice were sacrificed at 5 and 24 hours following the first injection, and striatal mRNA subjected to microarray analysis. Total RNA from each animal was loaded onto individual Affymetrix microarray chip. Total RNA from each animal was loaded onto individual Affymetrix microarray chips. Experimental reproducibility can be estimated by comparing columns within a figure as well as between corresponding columns in Figure 4. The early (5 hour) phase responses in C57BL/6J and SWR mice were indistinguishable (Figure 4). This suggests there is unlikely to be a strain-dependent difference in entry of MPTP into brain, consistent with a recent chemical determination of MPP⁺ levels in brains of C57BL/6J and SWR mice (Boyd et al., 2007). In contrast, the intermediate (24 hour) response was attenuated in SWR mice (Figure 4). Whereas the magnitude of mRNA changes observed in C57BL/6J mice was consistent from animal to animal the degree of change in SWR mice varied greatly between animals and with respect to individual genes. Thus, some MPTP treated SWR mice were indistinguishable from saline-treated animals whereas others showed more robust responses that for some probesets approximated levels observed in sensitive C57BL/6J mice. Although SWR mice are considered MPTP resistant, this is a relative term. In the acute MPTP model, SWR mice exhibit an approximate 14 % loss of DA (tyrosine hydroxylase-positive) SNpc neurons compared with 58% loss in C57BL/6J mice under identical conditions (Hamre et al., 1999). Thus, it is expected that if the intermediate response is linked to neuronal loss, it should be evident to some extent even in SWR mice. Furthermore, the neuronal loss in SWR mice is variable, with some animals having no apparent loss whereas others have more substantial losses. Therefore, it is possible that the SWR mice with more robust intermediate responses represent animals in which cell loss would have been more substantial if they were allowed to survive, whereas those with little or no intermediate response may represent mice that would have sustained no neuronal loss.

qRT-PCR was used to quantify mRNA levels of selected early and intermediate phase genes at 2, 5 and 24 hours post-MPTP treatment in C57BL/6J and SWR mice (Figure 5). Confirming the microarray data, there were no significant inter-strain differences in mRNA levels for the 5 hour response, with all transcripts rising statistically to the same extent in both strains. In contrast, the absolute levels of transcripts for all intermediate phase response genes were lower in the striatum of the SWR strain, but the levels of attenuation varied from gene to gene. Levels of some transcripts (e.g. Hbegf) were not significantly altered from basal values in MPTP-treated SWR mice while others (e.g. Hmox1) were only slightly increased relative to saline treatment. At the other extreme, levels of some transcripts, such as Pdlim4 were only slightly attenuated in MPTP-treated SWR mice compared to MPTP-treated C57BL/6J mice whereas others (e.g. Edg3) were about 50% of those observed in C57BL/6J mice (Figure 5).

Figure 5. Quantitative assessment of differences between early and intermediate mRNA responses in MPTP-sensitive (C57BL/6J) and MPTP-resistant (SWR) mice by qRT-PCR.

MPTP-sensitive (C57BL/6J) and MPTP-resistant (SWR) mice were injected at time zero with a single dose of MPTP (40 mg/kg) and sacrificed at 2, 5 and 24 hr. Levels of mRNA for candidate genes of the early (Cdkn1a, NFkBia, FosB, Nr4a1 and Gadd45b) and intermediate (Hmox1, Pdlim4, Edg3, Tnfrsf12a and Hbegf) response phases were evaluated by quantitative RT-PCR. Results are expressed as the number of copies of a specific mRNA (normalized versus the number of copies of Gapdh mRNA) versus time (hours). Data are presented as mean ± S.E.M. of 3 (Ctrl, 2 and 5 hr time points) or 4 (24 hr time point) animals. Differences between strains were analyzed by two way ANOVA followed by Bonferroni post-hoc test and significance levels are indicated with letters at the bottom of each pair of columns (A indicates P<0.001, B indicates P<0.01). Differences versus control (time point zero) within a single strain were analyzed with one way ANOVA and Bonferroni post-hoc test and statistical significances indicated with asterisks on top of each column (*** P<0.001, ** P<0.01, * P<0.05). Results are in agreement with those obtained by Affymetrix gene chip technology and indicate no quantitative differences in the early response, whereas the intermediate response is attenuated in SWR mice compared to C57BL/6J mice.

These results indicate that expression of genes identified in the intermediate phase, but not the early phase are predictive of the pathological events associated with MPTP. Furthermore, some genes show more attenuation than others in the resistant strain, suggesting that they might be better candidates for being participants in the pathological response to MPTP.

Inter-strain differences in basal mRNA levels

As inter-strain differences in basal gene expression levels in striatum might contribute to MPTP sensitivity and/or the intermediate phase response we compared basal mRNA levels in striatum from SWR and C57BL/6J mice. Total RNA from each animal was loaded onto individual Affymetrix microarray chips. Experimental reproducibility can be estimated by comparing columns within a figure as well as between corresponding columns in Figure 6.

333 genes (219 higher in C57BL/6J and 114 higher in SWR) were differentially expressed between MPTP-sensitive (C57BL/6J) and MPTP-resistant (SWR) strains of mice (Figure 6 and Supplemental Table 2). The functions of the gene products involved span all GO categories, implying structural and functional differences between the striatum of the strains. Some of the transcripts (e.g. alanine-glyoxylate aminotransferase 2-like 1 (Agxt2l1), Apod and Msr2) are MPTP-responsive; others such as mitochondrial superoxide dismutase 2 (Sod2) and catechol-O-methyl transferase (Comt) may contribute to oxidative stress responses and dopamine metabolism, respectively. There may also be differences in microglia status between the strains as basal mRNA levels for C1qc and Msr2 are markedly lower in SWR mice (Supplemental Table 2). Finally, one gene, PTEN induced putative kinase 1 (Pink1) has been implicated in PD (Valente et al., 2004, Bonifati et al., 2005, Tan et al., 2006, Abeliovich, 2007) and is also lower in SWR mice.

qRT-PCR was performed to measure levels of transcripts that were higher in either SWR (potassium inwardly-rectifying channel, subfamily J, member 10 (Kcnj10)) or C57BL/6J (Agxt2l1, Apod, Comt and Pink1) mice (Figure 7). These results confirm the microarray findings and establish that there are substantial differences in basal levels of gene expression between the two strains of mice.

Figure 7. Quantitative assessment of basal gene expression in striata of MPTP-sensitive (C57BL/6J) and MPTP-resistant (SWR) mice by qRT-PCR.

Levels of mRNA for 5 candidate genes (Kcjn10, Comt, Apod, Pink1 and Agxt2l1) previously identified as being differentially expressed between MPTP-sensitive (C57BL/6J) and MPTP-insensitive (SWR) mice using Affymetrix gene chip technology, were assessed by qRT-PCR. Results are expressed as the number of copies of a specific mRNA normalized versus the number of copies of Gapdh mRNA. Data are presented as mean ± S.E.M. of five animals. Differences between strains were analyzed by Student’s t-test and significance levels are indicated with asterisks at the top of each column (*** P<0.001, ** P<0.01, * P<0.05). These data confirm the relative strain-dependent differences revealed by microarray.

The MPTP transcriptome in Bax−/− mice

As the intermediate response is attenuated or absent in SWR mice we assessed whether MPTP-resistant Bax−/− mice (Vila et al., 2001) show similar temporal mRNA responses to SWR mice. Moreover, as the Bax knockout is on an inbred C57BL/6J background we anticipate there should be fewer differences in basal gene expression between the strains. To further minimize genetic background effects we produced and analyzed both Bax−/− and Bax wild type littermates by inter-crossing Bax heterozygous animals. These mice were treated with the standard acute MPTP paradigm and striatal mRNA levels analyzed by Affymetrix and qRT-PCR at 24 hours post-treatment. Total RNA from each animal was loaded onto individual Affymetrix microarray chips. Experimental reproducibility can be estimated by comparing columns within a figure as well as between corresponding columns in Figure 8.

There are fewer (59 genes) differences in basal mRNA expression levels between Bax−/− and Bax wild type mice (Supplemental Table 3). Besides the expected loss of Bax mRNA, there was also loss of GABA-A receptor, subunit gamma 3 (Gabrg3) and the small nuclear ribonucleoprotein Snurf. As both genes lie close to Bax on chromosome 7 it is possible that the homologous recombination event that generated the Bax−/− allele has affected the structure and/or expression of neighboring genes. Of the differentially expressed genes, only the elevated levels of huntingtin-associated protein 1 (Hap1) mRNA in Bax−/− mice has overt implications for neurodegeneration.

Unlike SWR mice there was a robust intermediate response in Bax−/− mice that was qualitatively (number and identity of genes) and quantitatively (levels of mRNA) largely indistinguishable from that seen in wild type littermates (Figure 8 and Supplemental Table 4). Using qRT-PCR for selected intermediate response genes, all tested transcripts in Bax−/− mice increased to at least the same levels observed in Bax wild type littermates (Figure 9). In fact, levels of Tnfrsf12a mRNA increased to a significantly higher level in Bax−/− mice compared to wild type mice.

Figure 9. Quantitative assessment of mRNA levels in the intermediate response in Bax−/− and wild type littermate mice by qRT-PCR.

Animals of both strains were injected with either MPTP (20 mg/kg) or saline (control) every two hours for a total of 4 injections. Levels of mRNA for candidate genes of the intermediate transcriptional phase (Hmox1, Pdlim4, Edg3, Tnfrsf12a and Hbegf) were evaluated by quantitative RT-PCR. Results are expressed as the number of copies of a specific mRNA normalized versus the number of copies of Gapdh mRNA. Data are presented as mean ± S.E.M. of 6 wild type and 4 Bax−/− mice treated with saline and 5 MPTP treated mice of both strains. Differences were analyzed with one way ANOVA and Bonferroni post-test and significance values are indicated with asterisks (*** P<0.001, ** P<0.01, * P<0.05). Results are in accordance with those obtained by Affymetrix gene chip technology and show no significant difference in the expression levels of selected transcripts between wildtype and Bax−/− mice.

Discussion

We showed previously that acute intoxication of DAergic synapses in the striatum with MPTP induces Hmox1 in surrounding astrocytes (Fernandez-Gonzalez et al., 2000). Based upon these data we proposed that products of Hmox1, such as carbon monoxide and iron, constituted a feed-forward loop that could further damage nerve terminals leading to neuronal death (Fernandez-Gonzalez et al., 2000, Smeyne and Jackson-Lewis, 2005). Here we have expanded this hypothesis using a genome-wide approach to show that Hmox1 is but one representative of a large cohort of genes that undergo stereotypical temporal and spatial patterns of change in the MPTP model. We therefore suggest a scenario in which the initial damage to the DA nerve endings in the striatum elicited by MPTP, initiates a second wave of gene expression events in surrounding cells whose products provide the final coup de grace to the DA neurons. Genetic resistance to MPTP can therefore take at least two forms. In SWR mice, the coupling between the initial damage and the secondary response is disrupted. In Bax−/− mice, however, resistance is conferred by an ability of the neurons to resist both the primary and secondary insults.

The present data establish that there are stereotypical changes in striatal mRNA levels following MPTP administration that reflect a number of biological and pathological responses triggered by MPTP treatment. Whereas the transient acute changes in mRNA levels elicited by MPTP are not specific to striatum and are evident in both sensitive and resistant strains of mice, the intermediate and late mRNA responses are confined to the striatum and are coincident with the period in which dopaminergic synaptic terminals in this brain region undergo functional and structural damage and neuronal death is initiated. As such the delayed phases are predictive of MPTP sensitivity, and may contribute, or be responsive to the pathogenic mechanism of MPTP. The fact that the early phase in striatum is qualitatively (genes represented) and quantitatively (mRNA levels) indistinguishable between C57BL/6J and SWR mice suggests that peripheral metabolism and/or penetration of MPTP into the brain is not the underlying mechanism of resistance in the latter strain. Rather the data are consistent with a resistance mechanism in SWR involving limitation of nerve terminal damage and/or attenuation of the intermediate transcriptional response. In contrast, in Bax−/− mice there is a robust intermediate response implying a different mode of MPTP resistance where damage to DA synaptic terminals, inflammation and astrogliosis in the striatum is tolerated by the DAergic SNpc neurons.

The early response has two general features. First, a subset of genes is induced to similar levels in all brain regions examined. Most of these genes have been implicated in responses to oxidative stress and include Cdkn1a (Hershenson, 2004, O'Reilly, 2005), Gadd45g (Edwards et al., 2004) and FK506 binding protein 5 (Fkbp5) (Grigoryev et al., 2006) implying widespread metabolic compromise, presumptively attributable to inhibition of mitochondrial respiration by MPP⁺ (Kindt et al., 1987, Krueger et al., 1990, Bates et al., 1994, Hasegawa et al., 1995). These findings are consistent with an extensive literature showing oxidative stress and cytokine signaling responses in the MPTP model (Beal, 2003, Przedborski and Vila, 2003, Hald and Lotharius, 2005, Jackson-Lewis and Smeyne, 2005, Hald et al., 2007, Schapira, 2008). Second there is a striatum enriched component of the early response involving a larger number of genes whose levels of expression change in the order striatum > cerebral cortex » cerebellum. For example, many immediate-early genes are induced to lesser extents (e.g. Fosb, Junb) or not at all (e.g. c-Jun) in cortex and cerebellum when compared to striatum. A number of additional genes linked to oxidative stress responses such as metallothionein 2 (Mt2), uncoupling protein 2 (Ucp2), Gadd45b and NFkBia are also increased to greater extents in striatum, suggesting that this structure may experience additional metabolic stress. MPP⁺ preferentially accumulates in dopaminergic nerve terminals and elicits dumping of DA (Chiba et al., 1985, Rollema et al., 1988), a potent oxidant as well as a receptor ligand. Therefore, the striatum selective responses may be the result of dopamine receptor activation and local oxidative damage caused by dopamine. The finding of elevated Fosb in striatal neurons in the MPTP model (Perez-Otano et al., 1998) is consistent with dopamine receptor interactions mediating some of the early responses. Furthermore, the non-receptor-mediated induction of Hmox1 by dopamine (but not other catecholamines) in glia (Schmidt et al., 1999) also supports a local oxidative stress model.

Some of the early mRNA changes may also be triggered as part of the central stress response to the peripheral toxicity of MPTP. Indeed many stressors trigger expression of immediate-early genes in the brain (Morgan and Curran, 1989, Schreiber et al., 1991, Liu et al., 2001, de Kloet et al., 2005, Grottick et al., 2005). Moreover, several genes identified in the early response are known to be responsive to stress/steroid hormone signaling, including Serum/glucocorticoid regulated kinases 1 and 3 (Sgk, and Sgk3, respectively) and TSC22 domain family, members 1 and 3 (Tsc22d1 and Tsc22d3, respectively) (Leung et al., 2003) and some have been implicated in the MPTP model of PD (Schoenebeck et al., 2005, Stichel et al., 2005). Similarly, DNA-damage-inducible transcript 4 (Ddit4) which is implicated in the mTOR-signaling pathway and whose levels are increased in Parkinson brain (Malagelada et al., 2006) and in a 6OHDA cellular model of PD (Malagelada et al., 2006) increases in the early phase. Interestingly, we observed not only an increase of DNA-damage-inducible transcript 4 (Ddit4) mRNA levels, but also a decrease in mRNA for the related gene, DNA-damage-inducible transcript 4-like (Ddit4l) in the same time period (Supplementary Table 1). As these changes occur in both sensitive and resistant strains of mice, their relevance to the pathology of PD is questionable. Nonetheless, in order to understand their relevance, it will be important to ascertain whether these reciprocal changes in gene expression occur in the same cell population.

The intermediate phase is characterized by the over-representation of transcripts implicated in cytokine signaling, inflammatory responses, activation of astrocytes and responses to cellular stress and damage (Supplemental Table 1 and Supplemental Table 6). Many studies have reported the presence of inflammatory responses in striatum in both PD and animal models of the disease (Hald and Lotharius, 2005, Nagatsu and Sawada, 2005, Hald et al., 2007, Hunter et al., 2007, McGeer and McGeer, 2007, Wilms et al., 2007) as well as astrocytosis and glial activation (McGeer et al., 1988, Francis et al., 1995, Kohutnicka et al., 1998, McGeer and McGeer, 2007). Besides gene products overtly linked to inflammation, gliosis, and cellular damage and stress responses, a number of other signaling pathways are also apparently altered during the intermediate response to MPTP. For example, levels of both CD9 antigen (Cd9) and Hbegf increase. As CD9 antigen associates with and activates the membrane bound form of heparin-binding EGF (Iwamoto et al., 1994) this coordinate up-regulation is expected to increase signaling through this pathway. There may also be changes in the bone morphogenetic protein and transforming growth factors signaling pathways following MPTP treatment as there are alterations in the mRNA levels of bone morphogenetic protein 1 (Bmp1) and BMP2 inducible kinase (Bmp2k) as well as the transforming growth factor, beta receptor II (Tgfbr2).

With a few notable exceptions, by 72 hours post-treatment the majority of the mRNA changes seen at 24 hours return to basal levels and a new cohort of transcripts are altered. Prominent, amongst the persistently altered mRNAs are those linked to gliosis, inflammation and oxidative stress. As these same gene expression changes are evident at 96 hours, this provides evidence for an ongoing and persistent inflammatory response in striatum that initiates within 24 hours of MPTP treatment. Nevertheless, the majority of genes whose expression is altered at 24 hours have returned to basal levels and another cohort of mRNA changes is evident at 72 and 96 hours.

The biological functions of genes that alter uniquely in the late phase are diverse and potentially represent adaptive responses occurring in neurons and oligodendrocytes as well as astrocytes and microglia. For example, changes in parvalbumin and solute carrier family 6, member 11 (Slc6a11, a GABA transporter) expression imply alterations in striatal GABAergic interneurons (Kawaguchi et al., 1995, Marin et al., 2000) whereas changes in myelin oligodendrocyte glycoprotein (Mog), 2',3'-cyclic nucleotide 3' phosphodiesterase (Cnp) and proteolipid protein 1 (Plp1) indicate responses in oligodendrocytes (Scolding et al., 1989, Hardy and Friedrich, 1996, Jakovcevski and Zecevic, 2005). A number of mRNAs in the late phase encode membrane or secreted proteins involved in intercellular communication and extracellular matrix function such as neural cell adhesion molecule 1 (Ncam1), gap junction membrane channel protein alpha 1 (Gja1 a.k.a. connexin 43), secreted acidic cysteine rich glycoprotein (Sparc), secreted phosphoprotein 1 (Spp1) and tissue inhibitor of metalloproteinase 4 (Timp4). These responses may reflect the process of synaptic terminal elimination and remodeling as might changes in mRNA levels for the synaptic protein, bassoon (Bsn).

The mRNA levels of a number of genes identified in this analysis have been reported to change in various MPTP models. However, as the models diverge in terms of dosing regimens, brain regions studied, time courses examined as well as microarray/analytical platforms, statistical criteria and sample size used, we cannot readily make direct comparisons, although we can highlight similarities among the present analysis and previously published reports. Here we identify 443 genes of which 88 belong to the early response (68 increased and 20 reduced), 226 to the intermediate response (199 increased and 27 reduced) and 170 to the late response (108 increased and 62 reduced). Moreover, we have chosen time points when cell death in the SNpc has not yet started (early response), is in its infancy (intermediate response) or is advanced (late response). When we compared previously published results obtained in striatum of MPTP-treated mice with our dataset we observed relatively little overlap. Using cDNA microarray techniques, Grünblatt and colleagues identified 51 genes whose levels were modulated 8 days after the first MPTP treatment in the striatum (Grunblatt et al., 2001, Mandel et al., 2002), only one of which, solute carrier family 6, member 11 (Slc6a11), was detected in our analysis (elevated at 3 days post treatment). The analysis of the striatal response to MPTP performed by Miller and colleagues using Affymetrix arrays (U74A v2) revealed 178 and 716 genes modulated at 8 and 15 days, respectively, after the first MPTP injection (Miller et al., 2005). Of these genes, only 25 were modulated in a similar fashion in our study (early response Kit ligand (Kitl), metallothionein 2 (Mt2) and serine/arginine-rich protein specific kinase 1 (Srpk1); intermediate response AMP deaminase 3 (Ampd3), cytochrome b-245, alpha polypeptide (Cyba), C1qb, metallothionein 2 (Mt2) and oncostatin M receptor (Osmr); late response angiotensinogen (Agt), cathepsin S (Ctss), human immunodeficiency virus type I enhancer binding protein 2 (Hivep2), myelin oligodendrocyte glycoprotein (Mog), ribonuclease T2A and B (Rnaset2a and Rnaset2b), Rho-associated coiled-coil containing protein kinase 2 (Rock2), secreted phosphoprotein 1 (Spp1) and tropomyosin 1, alpha (Tpm1) and intermediate/late Apod, aquaporin 4 (Aqp4), C1qc, Gfap, lectin, galactose binding, soluble 1 (Lgals1), metallothionein 1 (Mt1), S100 calcium binding proteins A6 and A13 (S100a6 and S100a13, respectively) and TYRO protein tyrosine kinase binding protein (Tyrobp)). In a recent study (Chin et al., 2008) using a similar dosing paradigm (15 mg/kg every 2 hours, 4 injections) and the same Affymetrix chip used here, Chin and coworkers identified 181genes whose mRNA are changed in the striatum of C57BL/6J mice 7 days after MPTP treatment. Ten genes identified in the latter study were also detected in our analysis: 8 genes were elevated 7 days post MPTP treatment and were similarly regulated in the intermediate and late responses in our study (oncostatin M receptor (Osmr) and Serpina3n, both elevated at 24 hours; and ATP-binding cassette, sub-family A (ABC1), member 1 (Abca1), aquaporin 4 (Aqp4), Gfap, lymphocyte antigen 86 (Ly86), transcription factor 7-like 2, T-cell specific, HMG-box (Tcf7l2) and Vim elevated at both 24 and 72 hours). The remaining two genes in common were downregulated (protein tyrosine phosphatase 4a3 (Ptp4a3) and ryanodine receptor 1 (Ryr1), both decreased at 72 hours). Despite the relatively low overlap, the genes consistently identified by all studies suggests long-term changes in processes such as inflammation, astrogliosis and protein trafficking. Our analysis indicates that these processes are initiated within the first 24 hours of treatment. Therefore, strategies aimed at ameliorating damage likely need to target early events that couple the insult to the pathological responses.

The early response to MPTP treatment in striatum involves changes in expression levels of many genes implicated in transcriptional regulation (approximately 21%) and is replete in immediate-early gene transcription factors such as early growth response 3 and 4 (Egr3 and Egr4,respectively), Fos, Fosb, Jun and Junb (Supplemental Table 1). In addition to transient increases in immediate-early gene expression there are increases in transcriptional repressors such as Bach1 (Oyake et al., 1996), B-cell translocation gene 3 (Btg3) (Ou et al., 2007) and zinc finger and BTB domain containing 16 (Zbtb16) (Costoya, 2007) as well as putative activators such as Kruppel-like factor 9 (Klf9) (Zhang et al., 2003) and transient decreases in others such as retinoid X receptor gamma (Rxrg) and paired box gene 8 (Pax8). The implication is that these alterations in turn trigger subsequent changes in expression of other target genes, such as those in the intermediate response. Whereas the acute response occurs to the same extent in both sensitive and resistant strains, the intermediate response is much attenuated in SWR mice, suggesting that the first is not causative of the second. However, we cannot exclude the possibility that the early responses are essential but not sufficient to trigger the intermediate phase response. In this scenario the mechanism of resistance in the SWR strain could involve the uncoupling of the early transcriptional response from the intermediate response. The late response is also characterized by changes in expression of several transcription factors. Notably, the immediate-early gene transcription factors, early growth response 3 (Egr3) and Fos-like antigen 2 (Fosl2) that are up-regulated in the early response are actually down-regulated in the late phase. This is reminiscent of the behavior of c-Fos following seizures, where its levels first increase and then decline to below basal values, at which point the gene becomes un-responsive or refractory to re-induction by subsequent challenges with chemoconvulsants (Morgan et al., 1987). Levels of mRNA for the transcription factor, ets variant gene 5 (Etv5) are also decreased in the late phase whereas levels of the transcriptional regulators activating transcription factor 6 (Atf6), nuclear receptor subfamily 2, group F, member 2 (Nr2f2) and zinc finger protein of the cerebellum 1 (Zic1) are increased. This again implies that MPTP elicits coordinated transcriptional cascades in striatum that are correlated with pathology. Like MPTP, methamphetamine also causes damage to DAergic synapses in striatum (O'Callaghan and Miller, 1994). Using a cDNA array platform Cadet et al. (2001) showed that methamphetamine treatment elicited a rapid (2–4 hours) increase in the levels of many mRNAs. As in the MPTP model, this early component was enriched in transcripts encoding transcription factors/DNA binding proteins (33.6% and 35.4%, respectively) many of which were immediate-early genes. Although a direct comparison is limited by the differences in platform and strains of mice (CD1) used, several genes including c-Jun, c-Fos, Pax8, JunB and FosB are in common with our dataset. This suggests the early component may be part of a common striatal response to synaptic impairment/damage. The same study also reported gene expression changes at 16 hours post-treatment, a time intermediate between the 5 and 24-hour time points investigated here. Comparing their 16 hour dataset with our 24 hour dataset revealed only three gene products, Cathepsin D, GADD45 and Stat3 to be in common. However, the time differences between the studies do not enable us to conclude whether or not methamphetamine elicits the same intermediate response as MPTP.

Although we determined the temporal relationships of gene expression changes in striatum in response to MPTP our methods do not have cellular resolution thereby limiting interpretation of signaling cascades. i.e. we cannot prove that any two changes in gene expression occur in the same cell. Nevertheless, valuable information can be mined from the data regarding potential signaling pathways activated by MPTP. To identify transcriptional regulators in the early response that potentially contribute to changes in the intermediate response, we used the Molecular Signature Database (MSigDB, www.broad.mit.edu/gsea/msigdb/msigdb_index.html) (Supplemental Table 8, Supplemental Table 9 and Supplemental Table 10). The most significant transcription factor binding site associations to genes during the intermediate response include JunD, Nrf2, Stat1 and Stat3, Bach1 and Bach2, and members of the NFkB, AP-1 and E2F families. Levels of mRNAs for both STATs increased in the intermediate response and presumptively contribute to cytokine signaling associated with the inflammatory response. Levels of Bach1 mRNA increase in the early phase. Bach signaling has been implicated in regulation of Cdkn1a expression (Shim et al., 2006) that is ubiquitously and acutely up regulated by MPTP. In addition, putative Bach1 binding sites are present in several genes identified in the intermediate response, including Gfap, S100 calcium binding protein A10 (S100a10) and Tnfrsf12a. Moreover, induction of Hmox1 requires NRF2 to inactivate the transcriptional repressor Bach1 (Reichard et al., 2007). Putative Elf1 binding sites are also enriched in the intermediate phase. Elf1 is an Ets-related transcription factor that can associate with other transcription factors, such as AP-1 complexes (Bassuk and Leiden, 1995) and has been implicated in gene regulation in many contexts, most notably immune and inflammatory responses (Serdobova et al., 1997, Gallant and Gilkeson, 2006). Putative Elf-1 sites are present in genes encoding C1qa, Fc receptor, IgE, high affinity I, gamma polypeptide (Fcer1g) and Vim that increase in the intermediate response. Moreover, Vim also harbors an AP-1 site, members of which increase in both the early (e.g. Fos, Junb) and intermediate (c-Jun) responses opening the possibility for co-regulation through Elf1-AP-1 associations. An Elf1 site is also present in allograft inflammatory factor 1 (Aif1 a.k.a. Iba1) a marker for microglia (Ito et al., 1998) whose expression also increases in the intermediate response. Thus, Elf1 may contribute to both microgliosis and astrocytosis in the MPTP model.

The relative resistance to MPTP in SWR versus C57BL/6J mice is a polygenic trait (Cook et al., 2003) that could be attributable to differences in both the basal and MPTP modulated levels of gene expression as well as amino acid polymorphisms. In striatum alone 333 genes are differentially expressed between the strains (Supplemental Table 2) and several could potentially contribute to MPTP resistance. For example, superoxide dismutase 2 (Sod2) has been implicated in oxidative stress responses and Comt contributes to dopamine metabolism. Of the 333 transcripts, 12 (e.g. Agxt2l1, ApoD, Msr2) are also regulated by MPTP and belong almost exclusively to the intermediate and late phases that are attenuated in SWR mice (Supplemental Table 11). Notably, a number of the regulated genes, such as C1qc and Msr2 are likely expressed in microglia and are reduced in abundance in SWR mice even under basal conditions. Furthermore, another gene with reduced expression in SWR mice, CD34 antigen (Cd34) has been associated with microgliosis (Ladeby et al., 2005) whereas the complement antagonist Cd59a (Qian et al., 2000) that attenuates damage in experimental allergic encephalitis (Mead et al., 2004) is elevated in SWR mice. This could imply intrinsic functional differences in microglia between the strains that warrant further analysis in the MPTP model.

A previous quantitative trait loci (QTL) analysis identified a region of chromosome 1 (mptp1) that showed a significant association with the strain dependent differences in MPTP sensitivity in SWR and C57BL/6J mice (Cook et al., 2003). We identified 3 genes (Kcnj10, nuclear VCP-like (Nvl) and signal recognition particle 9 (Srp9)) in the mptp1 locus that are differentially expressed between the strains (Supplemental Table 12 and Figure 9). In addition, mRNA levels for another 6 genes in the locus alter following MPTP treatment in C57BL/6J mice (Supplemental Table 13). These genes therefore become candidates for more detailed analysis.

The results in SWR mice indicate that both the inflammatory response and gliosis seen in C57BL/6J mice is attenuated in the resistant strain. However, this is not a universal mechanism of MPTP resistance as Bax−/− mice exhibit a robust intermediate response that is qualitatively and quantitatively indistinguishable from wild type littermates. In addition, Bax−/− mice have very few intrinsic differences in their basal striatal mRNA profiles compared to wild type littermates. Of the differentially expressed transcripts, only the elevated levels of huntingtin-associated protein 1 (Hap1)mRNA in Bax−/− mice has any overt implications for neurodegeneration. Huntingtin-associated protein 1 (Hap1) can bind and sequester polyglutamine-expanded proteins such as Huntingtin thereby antagonizing aggregate formation (Li et al., 1995). Therefore, it is conceivable that this leads to greater MPTP resistance in Bax−/− mice. An alternative hypothesis is that Bax resistance lies downstream of the inflammatory response, possibly in the SNpc DA neurons themselves, making them tolerant of the insult. These possibilities could be tested using a floxed Bax allele and a Cre recombinase targeted to dopaminergic neurons.

The present data provide insight into the biological and pathological processes triggered by MPTP treatment as well as the genes and mechanisms that might contribute to sensitivity to this neurotoxin. However, the ultimate goal of the study is to identify genetic factors that contribute to PD. Analysis of our datasets for genes that have been linked to PD, identified 3 candidates: Gpr37 (a.k.a. PaelR) and alpha-synuclein (Snca) that are increased and decreased, respectively, at 72 hours post-MPTP treatment, and Pink1 that is expressed at higher levels in the MPTP-sensitive C57BL/6J strain of mice.

Polymorphism of Snca have been linked with familial (Polymeropoulos et al., 1997, Kruger et al., 1998) and sporadic forms of PD (Chiba-Falek et al., 2006, Ahn et al., 2008). Although the role of Snca in PD has been extensively reviewed (Dauer and Przedborski, 2003, Cookson, 2005, Schapira, 2006, Wood-Kaczmar et al., 2006), its physiological and pathophysiological roles remain elusive. Several functions have been ascribed to this protein such as: activation of microglia (Su et al., 2007), modulation of glutamatergic (Gureviciene et al., 2007) and DAergic release (Abeliovich et al., 2000) and regulation of NFkB signaling pathway (Yuan et al., 2007). The orphan G-protein coupled receptor, Gpr37 (Marazziti et al., 1997, Marazziti et al., 1998) is a known substrate for Parkin (Imai et al., 2003, Imai et al., 2007), another gene linked to familial forms of PD (Kitada et al., 1998). Gpr37 has been implicated in DA metabolism (Imai et al., 2007) as well as in the death of DA neurons in the SNpc (Kitao et al., 2007). Pink1 (PTEN-inducible kinase 1) is a serine/threonine protein kinase (Unoki and Nakamura, 2001) localized in the cytoplasm (Haque et al., 2008) and mitochondria (Valente et al., 2004). Mutations in Pink1 segregate with familial forms of PD (Valente et al., 2004, Albanese et al., 2005, Bonifati et al., 2005, Cookson, 2005). Pink1 has been reported to protect cells from apoptosis (Petit et al., 2005) and neurons from MPTP toxicity (Haque et al., 2008). Although its genetic elimination does not cause DA neuron loss (Zhou et al., 2007), Pink1 knockout mice display altered striatal DA release and synaptic plasticity (Kitada et al., 2007). Its lower levels in SWR mice seems at odds with the suggested pro-survival effect of Pink1 in PD (Abeliovich, 2007, Pridgeon et al., 2007).

A major question raised by this study is to what extent the risk for, and/or progression of PD may be influenced by the transcriptome of the striatum versus that of the SNpc. This issue is especially relevant as a number of gene products linked to PD are broadly expressed in the brain and it is sometimes unclear how they result in the specific pattern of neurodegeneration seen in this disease. For example, Pink1 is widely expressed in brain (Taymans et al., 2006) and how disruption of its function leads to PD is unclear. Hence changed Pink1 expression or function in striatum might be critical in PD. Alpha-synuclein and the Parkin substrate, Gpr37 are also broadly expressed in brain (Jakes et al., 1994, Zeng et al., 1997, Li et al., 2002, Kingsbury et al., 2004) as is Parkin itself (Stichel et al., 2000). However, Gpr37 and Snca are components of Lewy bodies (Arima et al., 1999, Ishizawa et al., 2003, Murakami et al., 2004, Wakabayashi et al., 2007) and it is most likely that their pathophysiological roles are exerted in SNpc DAergic neurons. To resolve the question of the pathophysiologically relevant sites of expression of susceptibility genes for PD it will be necessary to selectively alter levels of their transcripts in either SNpc or striatum by genetic means.

Supplementary Material

01. Supplemental Table 1: List of genes modulated in the striatum of C57BL/6J mice at 5, 24 and 72 hours after MPTP treatment.

Increase or decrease of signal for specific probesets are reported for each time point with red upward-pointing arrows () or blue downward-pointing arrows (), respectively. Probesets modulated at multiple time points are reported individually on separate lines. Ratio: MPTP/Saline. FDR: false discovery rate. Only probesets with FDR < 0.05 and fold change ≥ 1.5 or ≤ 0.667 are listed. Gene symbols and gene names have been simplified by removing references to locus and similar sequences, the reader should refer to the most up-to-date probeset description at the Affymetrix web site (http://www.affymetrix.com).

02. Supplemental Table 2: List of genes differentially expressed in striatum of MPTP-sensitive (C57BL/6J) and MPTP-resistant (SWR) mice.

The column marked “Higher in:” reports the strain of mice in which the basal level signal of a specific probeset is higher. Ratio: SWR/C57BL/6J. FDR: false discovery rate. Only probesets with FDR < 0.05 and fold change ≥1.5 or ≤0.667 are listed. Gene symbols and gene names have been simplified by removing references to locus and similar sequences, the reader should refer to the most up-to-date probeset description at Affymetrix web site (http://www.affymetrix.com).

03. Supplemental Table 3: List of genes differentially expressed in striatum of MPTP-resistant Bax−/− and MPTP-sensitive wild type littermate mice.

The column marked “Expression:” reports whether a specific probeset is higher (UP) or lower (DOWN) in Bax−/− versus their wild type littermate controls. Ratio: Wild Type/Bax−/−. Statistical analysis was performed using the Student’s t-test method. Only probesets with P < 0.05 and fold change ≥1.5 or ≤0.667 are listed. Probesets differentially expressed between genders were removed. Gene symbols and gene names have been simplified by removing references to locus and similar sequences, the reader should refer to the most up-to-date probeset description at the Affymetrix web site (http://www.affymetrix.com).

04. Supplemental Table 4: List of MPTP-modulated genes in the striatum of Bax−/− and wild type littermate controls.

The Table shows the mRNA profile of the MPTP-regulated genes reported in Supplemental Table 1 in Bax−/− (B) and wild type littermate controls (W) treated with saline (BS or WS, respectively) or MPTP (BM and WM, respectively). The column “S:” reports the result of the Student’s t-test for each condition: (a) P ≤ 0.001, (b) P ≤ 0.01, (c) P < 0.05. Gene symbols and gene names have been simplified by removing references to locus and similar sequences, the reader should refer to the most up-to-date probeset description at the Affymetrix web site (http://www.affymetrix.com).

05. Supplemental Table 5: Pathway analysis by GSEA at 5 hours post MPTP treatment in C57BL/6J mice.

Size: number of genes reported in the pathway by Ingenuity® at the time of this analysis. GiP: number of genes enriched at 5 hours. GiP%: percentage of enrichment. FDR: false discovery rate (cutoff was set at 0.25). Gene list: genes enriched at 5 hours (GiP).

06. Supplemental Table 6: Pathway analysis by GSEA at 24 hours post MPTP treatment in C57BL/6J mice.

The Table is divided between pathways that are enriched and those repressed by MPTP. Size: number of genes reported in the pathway by Ingenuity® at the time of this analysis. GiP: number of genes enriched at 24 hours. GiP%: percentage of enrichment. FDR: false discovery rate (cutoff was set at 0.25). Gene list: genes enriched at 5 hours (GiP).

07. Supplemental Table 7: Pathway analysis by GSEA at 72 hours post MPTP treatment in C57BL/6J mice.

The Table is divided between pathways that are enriched and those repressed by MPTP. Size: number of genes reported in the pathway by Ingenuity® at the time of this analysis. GiP: number of genes enriched at 72 hours. GiP%: percentage of enrichment. FDR: false discovery rate (cutoff was set at 0.25). Gene list: genes enriched at 5 hours (GiP).

08. Supplemental Table 8: Promoter sequence enrichment analysis by GSEA at 5 hours post MPTP treatment in C57BL/6J mice.

Each row lists the name of a known or putative cis-regulatory element (motif) conserved across the human and mouse genomes (http://www.broad.mit.edu/gsea/msigdb/index.jsp). TFs: transcription factors associated with the motif listed on that row. Size: number of genes reported to have the motif in their promoter region. FDR: false discovery rate (cutoff was set at 0.25). Gene list: genes modulated by MPTP at 5 hours and potentially regulated by the “motif” listed in the same row.

09. Supplemental Table 9: Promoter sequence enrichment analysis by GSEA at 24 hours post MPTP treatment in C57BL/6J mice.

Each row lists the name of a known or putative cis-regulatory element (motif) conserved across the human and mouse genomes (http://www.broad.mit.edu/gsea/msigdb/index.jsp). TFs: transcriptional factors associated with the motif listed on that row. Size: number of genes reported to have the motif in their promoter region. FDR: false discovery rate (cutoff was set at 0.25). Gene list: genes modulated by MPTP at 24 hours and potentially regulated by the “motif” listed in the same row.

10. Supplemental Table 10: Promoter sequence enrichment analysis by GSEA at 72 hours post MPTP treatment in C57BL/6J mice.

Each row lists the name of a known or putative cis-regulatory element (motif) conserved across the human and mouse genomes (http://www.broad.mit.edu/gsea/msigdb/index.jsp). TFs: transcriptional factors associated with the motif listed on that row. Size: number of genes reported to have the motif in their promoter region. FDR: false discovery rate (cutoff was set at 0.25). Gene list: genes modulated by MPTP at 72 hours and potentially regulated by the “motif” listed in the same row.

11. Supplemental Table 11: List of MPTP-modulated genes that are differentially expressed in the striatum of C57BL/6J and SWR mice.

Increase or decrease of signal for specific probesets are reported for each time point with red upward-pointing arrows () or blue downward-pointing arrows (), respectively. Probesets modulated at multiple time points are reported individually on separate lines. FDR: false discovery rate. Gene symbols and gene names have been simplified by removing references to locus and similar sequences, the reader should refer to the most up-to-date probeset description at the Affymetrix web site (http://www.affymetrix.com).

12. Supplemental Table 12: Genes located in the mptp1 locus on mouse chromosome 1 that are differentially expressed in C57BL/6J and SWR mice.

The “Higher in” column reports whether the basal mRNA level for individual genes is higher in C57BL/6J or SWR mice. FDR: false discovery rate. Gene symbols and gene names have been simplified by removing references to locus and similar sequences, the reader should refer to the most up-to-date probeset description at the Affymetrix web site (http://www.affymetrix.com).

13. Supplemental Table 13: Genes located in the mptp1 locus on mouse chromosome 1 that are modulated by MPTP in C57BL/6J mice.

Increase or decrease of signal strength for specific probesets are reported for each time point with red upward-pointing arrows () or blue downward-pointing arrows (), respectively. Gene symbols and gene names have been simplified by removing references to locus and similar sequences, the reader should refer to the most up-to-date probeset description at the Affymetrix web site (http://www.affymetrix.com).

Abbreviations

Agxt2l1

alanine-glyoxylate aminotransferase 2-like 1

Apod

apolipoprotein D

Bach1

BTB and CNC homology 1

Bax

BCL2-associated X protein

C1qa

complement component 1, q subcomponent, alpha polypeptide

C1qb

complement component 1, q subcomponent, beta polypeptide

C1qc

complement component 1, q subcomponent, C chain

C4b

complement component 4B (Childo blood group)

Cdkn1a

cyclin-dependent kinase inhibitor 1A

c-Fos

FBJ osteosarcoma oncogene

c-Jun

Jun oncogene

Comt

catechol-O-methyl transferase

DA

dopamine

Edg3

endothelial differentiation, sphingolipid G-protein-coupled receptor, 3

Fosb

FBJ osteosarcoma oncogene B

GABA

gamma-aminobutyric acid

Gadd45b

Growth arrest and DNA-damage-inducible 45 beta

Gfap

Glial fibrillary acidic protein

Gpr37

G protein-coupled receptor 37

GSEA

gene set enrichment analysis

Hbegf

heparin-binding EGF-like growth factor

Hmox1

hemeoxygenase-1

Junb

Jun-B oncogene

Kcnq10

potassium voltage-gated channel, subfamily Q, member 10

MPTP

1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine

MPP⁺

1-methyl-4-phenylpyridinium

Msr2

macrophage scavenger receptor 2

Nfkbia

Nuclear factor of kappa light chain gene enhancer in B-cells inhibitor, alpha

PD

Parkinson’s disease

Pdlim4

PDZ and LIM domain 4

Pink1

PTEN induced putative kinase 1

qRT-PCR

quantitative reverse transcription polymerase chain reaction

Snca

synuclein, alpha

SNpc

substantia nigra pars compacta

Tnfrsf12a

tumor necrosis factor receptor superfamily, member 12a

Vim

vimentin