OUP user menu

Gene expression profiling of substantia nigra dopamine neurons: further insights into Parkinson's disease pathology

Filip Simunovic, Ming Yi, Yulei Wang, Laurel Macey, Lauren T. Brown, Anna M. Krichevsky, Susan L. Andersen, Robert M. Stephens, Francine M. Benes, Kai C. Sonntag
DOI: http://dx.doi.org/10.1093/brain/awn323 1795-1809 First published online: 3 December 2008


Parkinson's disease is caused by a progressive loss of the midbrain dopamine (DA) neurons in the substantia nigra pars compacta. Although the main cause of Parkinson's disease remains unknown, there is increasing evidence that it is a complex disorder caused by a combination of genetic and environmental factors, which affect key signalling pathways in substantia nigra DA neurons. Insights into pathogenesis of Parkinson's disease stem from in vitro and in vivo models and from postmortem analyses. Recent technological developments have added a new dimension to this research by determining gene expression profiles using high throughput microarray assays. However, many of the studies reported to date were based on whole midbrain dissections, which included cells other than DA neurons. Here, we have used laser microdissection to isolate single DA neurons from the substantia nigra pars compacta of controls and subjects with idiopathic Parkinson's disease matched for age and postmortem interval followed by microarrays to analyse gene expression profiling. Our data confirm a dysregulation of several functional groups of genes involved in the Parkinson's disease pathogenesis. In particular, we found prominent down-regulation of members of the PARK gene family and dysregulation of multiple genes associated with programmed cell death and survival. In addition, genes for neurotransmitter and ion channel receptors were also deregulated, supporting the view that alterations in electrical activity might influence DA neuron function. Our data provide a ‘molecular fingerprint identity’ of late–stage Parkinson's disease DA neurons that will advance our understanding of the molecular pathology of this disease.

  • Parkinson's disease
  • microarray
  • laser microdissection
  • pathogenesis
  • dopamine


Parkinson's disease is a neurodegenerative disorder caused by a progressive deterioration of midbrain dopamine (DA) neurons in the substantia nigra pars compacta (SNc). The death of DA cells is associated with tremor and rigidity and results in a gradual dysfunction of the extrapyramidal motor system. The disease affects about 2–3% of individuals over the age of 65 years and there is evidence that its prevalence is higher in the male population (Cantuti-Castelvetri et al., 2007). There is currently no cure for Parkinson's disease and the underlying pathogenesis of the disease is still unknown. Two forms of Parkinson's disease are recognized: a ‘familial’ or early-onset Parkinson's disease (<10% of all patients) and an ‘idiopathic’ or late-onset Parkinson's disease (>85% of all cases) that does not appear to exhibit heritability. Overall, the pathology of Parkinson's disease is complex and is most likely a ‘consequence of an unspecified combination of genetic and environmental factors, which induce a common pathogenic cascade of molecular events’ (Maguire-Zeiss and Federoff, 2003; Miller and Federoff, 2005).

Since the first description of this syndrome in 1817 by James Parkinson, Parkinson's disease has been the subject of intense investigation to understand its pathophysiology and to develop therapeutic interventions. So far, pharmacological and surgical therapies are available and can alleviate some of the symptoms, but these interventions are associated with serious side effects and generally lose efficacy over time (Benabid, 2007; Schapira, 2007). Although research has progressed, one of the main hurdles for the development of therapeutic or preventative measures is the still limited understanding of the underlying pathophysiology of Parkinson's disease and the lack of reliable biomarkers. To a large extent, biomedical research on Parkinson's disease focuses on in vitro and in vivo disease models, as well as studies of postmortem brain. Based on the availability of more sophisticated technologies, the latter has become more prominent over the past years and has revealed novel insights in the pathogenesis of Parkinson's disease. For example, several studies have used microarray technologies on the substantia nigra of normal control and Parkinson's disease patients to assess differential gene expression profiles; data from these studies have helped to further delineate some disease-associated pathways (Grunblatt et al., 2004; Hauser et al., 2005; Zhang et al., 2005; Duke et al., 2006; Miller et al., 2006; Moran et al., 2006, 2007; Moran and Graeber, 2008). However, the array results in these studies did not entirely represent the DA neuronal profile, since large amounts of other cell populations were also included in the dissected tissue. The introduction of laser microdissection (LMD) has further refined this approach and was essential to the demonstration of a broad gender-linked difference in the gene expression profile of human substantia nigra DA neurons (Cantuti-Castelvetri et al., 2007).

In the current study, we used LMD (Benes et al., 2007) to isolate DA neurons from the substantia nigra of nine normal and 10 idiopathic Parkinson's disease patients. Using microarray-based gene expression profiling, we have analysed our data based on cluster analyses of biological functions and cellular pathways relevant to Parkinson's disease pathology and have compared the results to the published expression profiles. Our data confirm the involvement of several known molecular regulatory pathways in the pathogenesis of Parkinson's disease: these include oxidative stress-induced cell responses and dysfunction of the mitochondrial and ubiquitin-proteasome system (UPS). In particular, we found clusters of differentially expressed genes that appear to be involved in extrinsic and intrinsic signalling events in programmed cell death (PCD), as well as a prominent down-regulation of multiple members of the PARK gene family, which are associated with familial forms of Parkinson's disease. In addition, we have also noted changes in the expression of neurotransmitter and ion channel genes that suggest alterations in synaptic activity; the latter have been implicated in the modulation of survival and/or degeneration of DA neurons.

Materials and Methods

Subjects and affymetrix-based microarrays

All affymetrix-based microarrays and data about subjects are publicized at the National Brain Databank and were deposited by Dr. Francine Benes (http://national_databank.mclean.harvard.edu/brainbank). Material collection, preparation and data generation were according to previously published protocols (Benes et al., 2007). Briefly, frozen tissue blocks containing SNc from control subjects and patients with idiopathic Parkinson's disease matched for age and postmortem interval (PMI) were cut using a Microm HM 560 CryoStar cryostat (8 µm), mounted on LEICA Frame Slides with a PET-membrane (1.4 µm) and placed on a LEICA AS LMD apparatus. Since DA neurons contain neuromelanin, they could easily be visualized and collected using laser-based microdissection. Each vial into which the laser-dissected specimens fell by gravity contained a small volume of a lysis/denaturing solution to inhibit RNAse activity. An average of 300 or 700 DA neurons were collected from control subjects or Parkinson's disease patient's brains, respectively. RNA extraction was undertaken with a Qiagen RNeasy Micro Kit (Qiagen, Valencia, CA), and quality was assessed using an Agilent 2100 bioanalyser (Agilent Technologies, Palo Alto, CA). Following the manufacturer's instructions, three rounds of linear amplification of the target was carried out using the MessageAmp aRNA Amplification kit (Ambion, Austin, TX). The use of three rounds of amplification could induce degradation of RNA and potentially bias the microarray data; however, all the samples from both groups were processed in an identical fashion, making it unlikely that such bias occurred in one group to a greater degree than another. Subsequently, target labeling was performed with the Message-AMP Biotin Enhanced Kit (Ambion). Fifteen micrograms of biotinylated target RNA was fragmented and individually hybridized to the HU-133A arrays (Affymetrix, Santa Clara, CA). The microarrays were then stained with two rounds of streptavidin-phycoerythrin (Molecular Probes, Eugene, OR) and one round of biotinylated antistreptavidin antibody (Vector Laboratories, Burlingame, CA), scanned twice, and visually inspected for evidence of artefacts.

In addition to their demographic factors, the cases included in this study (Table 1) were chosen on the basis of their RNA quality using tissue pH, the 18S/28S ratio, and the Percent Present Calls for each case as described elsewhere (Luzzi et al., 2003; Benes et al., 2007).

View this table:

Data normalization and analysis

All mRNA chips were normalized using the RMA, or MAS5 procedure in R packages from Bioconductor (www.bioconductor.org), or using GCRMA in Partek Genomic Suite (www.partek.com). For each contrast of classes, probesets were filtered based on the detection calls derived from MAS5 procedure according to the majority rule (for each probeset, in at least one of the classes in contrast it shall have majority of their detection calls as ‘P’ (present) in the samples of this class in order to be retained in the filtered probeset lists). The data from either RMA or MAS normalization for those filtered probes were subjected to SAM procedure (Tusher et al., 2001) to determine the significant gene lists based on intended false discovery rates (FDR). Student t-tests were then used to filter significant gene lists. Alternatively, two- or three-way ANOVA models were used to derive the differentiated genes from different contrasts of different treatment and phenotypes using the Partek Genomic Suite.

The enrichment analysis and pathway-level comparative analysis were performed using the in-house software WPS [(Yi et al., 2006); Yi and Stephens, unpublished results]. Briefly, Fisher's exact test was performed based on 2 × 2 contingency tables, to determine whether a gene is in a given list and whether it is associated with a pathway (gene set, term). One-sided Fisher's exact test was used to measure whether a particular Biocarta pathway (www.biocarta.com), GSEA gene set term (www.broad.mit.edu/gsea/) or a GO term (www.geneontology.org/) were enriched in a given gene list. The terms were ranked based on their Fisher's exact test P-values with the most enriched term listed at the top. To compare biological themes at the pathway, gene set and GO term level across multiple gene lists of different contrasts, these gene lists were also subjected to a pathway-level pattern extraction pipeline (Yi and Stephens, unpublished results). Briefly, after batch computation of Fisher's exact test for the gene lists, the log-transformed P-values were retrieved and combined into an enrichment score matrix for cluster analysis or pathway pattern extraction. The terms (pathways, or GO terms) of selected clusters with interests were further used to retrieve the associated genes from the original gene list. Pathways of interest were displayed along with the data in the WPS program.

The data were also analysed in Partek Genomics Suite to determine the segregation of individual samples and possible differences among control subjects and Parkinson's disease patients (Supplementary Fig. 1S). Although there was a ‘batch effect’ observed between samples from three different dates of microarray assays (Supplementary Fig. 1SA), this could be compensated by using three-way ANOVA (Supplementary Fig. 1SB). These results demonstrated that all individual samples from normal subjects and Parkinson's disease patients clustered and that there was a clear segregation between normal and disease-association attesting for high consistency and reproducibility of the data.

TaqMan® real-time PCR assay validation

Expression of 14 genes (listed below) was measured in three normal control and three Parkinson's disease samples (Table 1) by real-time PCR using TaqMan® Gene Expression Assays and the 7900HT Real-Time PCR System (Applied Biosystems, Foster City, CA). A total of 250–600 DA neurons were captured from each sample and total RNA isolated using the mirVANA™ miRNA Isolation Kit (Ambion). cDNAs were generated in a 25 µl reverse transcription reaction with 60 ng of total RNA from each sample using the High Capacity cDNA Archive Kit and protocol (Applied Biosystems, PN 4322169). The resulting cDNA was subjected to a 10-cycle PCR amplification followed by real-time PCR reaction using the manufacturer's TaqMan® PreAmp Master Mix Kit Protocol (Applied Biosystems, PN 4366127). The 10-cycle pre-amplification protocol has been shown to have 100% efficiency and introduced no bias in fold change determination in a previous study (Li et al., 2008). Four replicates per sample were assayed for each gene in a 384-well format plate. For data normalization across samples, GUSB was used as endogenous control gene. Normalization of Ct values of each gene and determination of fold differences gene expression Parkinson's disease versus control was calculated according to the 2−ΔΔCt method by Livak and Schmittgen (2001; Schmittgen and Livak, 2008). The following genes were analysed:

View this table:


There are several approaches to the analysis of microarray data (summarized in Miller and Federoff, 2005). A common way is clustering genes according to fold changes and their relevance to biological function. In the current study, we employed a three-pronged approach:

  1. Derivation of gene lists using SAM- and ANOVA-based data analysis (see Material and methods section for details);

  2. Analysis of candidate genes associated with cellular pathways relevant to Parkinson's disease pathology according to published literature; and

  3. Comparison with microarray data available from previous studies.

Because the statistical inclusionary criteria for deriving differentiated gene lists are somewhat arbitrary, we used different cut-offs and methods to generate corresponding lists of genes for similar class comparisons, and then assessed the consensus of the enrichment levels among these lists at functional pathway or gene set level (see details in ‘Material and methods’ section). We believe that the pathway-level enrichment, which considers gene sets or pathways with multiple relevant genes rather than individual genes, would be more consistent across these gene lists. Consequently, the gene sets or functional terms would be more relevant to the underlying biology represented by the class comparison: Parkinson's disease versus normal. For the more consensus pathways or gene sets (e.g. GO terms) associated genes were retrieved from the original gene lists and an example of this analysis is shown in Supplementary Fig. 2S. We found that the enrichment levels of the functional terms were highly concordant among the different gene lists. In addition, many of these lists were relevant to Parkinson's disease pathogenesis (see below) and similar to data from other published arrays (e.g. Grunblatt et al., 2004; Zhang et al., 2005; Cantuti-Castelvetri et al., 2007). A summary of the genes is presented in Supplementary Table 1S using three-way ANOVA (A3W, FDR10). This list was instrumental for additional cluster analyses using GenMAPP 2.1 (www.genmapp.org) (Doniger et al., 2003) and for generating gene clusters that are linked to Parkinson's disease pathology (see below).

Altogether, we found 465 down- and 580 up-regulated genes in the Parkinson's disease samples (Supplementary Table 1S). When the cut-off was set at greater than 1.5-fold difference, 358 out of the 465 downregulated genes fell into this group, while only 20 of the 580 upregulated genes were represented. Interestingly, the downregulated genes showed differences as high as 11.8-fold, while upregulated genes were not increased by more than 2-fold. In addition, almost all down- or upregulated genes had a strong association with neuronal function, pointing to a high stringency of the LMD collected material. A summary of the highest downregulated genes (>3-fold) with potential reference to the function of DA neurons is shown in Supplementary Table 2S. In the following, we present a detailed listing of our results according to gene groups and pathways that have been associated with the pathogenesis of Parkinson's disease.

PARK genes

Over the past decade, it has become clear that mutations in several genes are linked to familial forms of Parkinson's disease (Cookson, 2005; Moore et al., 2005). These genes are clustered in the PARK loci and, so far, PARK1 (a-Synuclein, SNCA), PARK2 (Parkin), PARK5 (UCH-L1), PARK6 (PINK1), PARK7 (DJ-1), PARK8 (LRRK2) and PARK9 (ATP13A2) have been implicated in this form of the disease (Schiesling et al., 2008). Our results demonstrate a down-regulation of PARK1, 5, 6, 7, 9 and 10 with an upregulation of the PARK10 loci-linked genes RAP1GA1 and RIMS1. Interestingly, DJ-1 was one of the highest downregulated genes (–8.55534-fold) in our entire data set (Table 2 and Supplementary Table 2S). These results are partly congruent with previously published arrays, in which down-regulation of PARK genes has also been described (Hauser et al., 2005; Moran et al., 2006, 2007; Moran and Graeber, 2008).

View this table:
Table 2

Genes associated with Parkinson's disease linkage (PARK loci)

PARKGene symbolGenBank IDDescriptionFold changeP-value
PARK1SNCABG260394Synuclein, alpha (non A4 component of amyloid precursor)−1.858990.00037
PARK5UCH-L1NM_004181Ubiquitin carboxyl-terminal esterase L1 (ubiquitin thiolesterase)−1.944170.00409
HIP2NM_005339Huntingtin interacting protein 2−1.221730.00218
PARK6PINK1AF316873PTEN induced putative kinase 1−2.158390.00010
PARK7DJ-1NM_007262Parkinson disease (autosomal recessive, early onset) 7−8.555340.00048
PARK9ATP13A2NM_022089ATPase type 13A2−1.377970.00432
PARK10RAP1GA1AB007943RAP1 GTPase activating protein1.421680.00045
RIMS1AF263310Regulating synaptic membrane exocytosis 11.221180.00142
RIMS3NM_014747Regulating synaptic membrane exocytosis 3−2.880550.00132

Programmed cell death

There are two major forms of apoptosis, intrinsic and extrinsic. While the intrinsic mechanisms are linked to several stress-related dysfunctions of cellular organelles, extrinsic apoptosis is mediated by death receptors. We found a striking downregulation of PINK1 and DJ-1, ATF4 as an indicator of ER stress (Ron and Walter, 2007; Burke, 2008), several clusters of genes linked to mitochondrial impairment (see below), and downstream factors that are involved in anti- and pro-apoptotic regulation, such as the bcl-2 protein family members BCL2L1 and BCL2A1, mitogen-activated protein kinase 8 (jun-kinase) interacting protein 3 (MAPK8IP3), LRPPRC and NFRKB. Strikingly, there was a consistent upregulation of the death receptors FAS, TNFRSF10B and TNFRSF21 as well as genes involved in their signalling cascade, such as TRADD, TNFAIP8, TNIP2, CFLAR, CASP8 and NFRKB indicating that extrinsic apoptosis is activated in Parkinson's disease-affected neurons (Table 3).

View this table:
Table 3

Genes associated with PCD and mitochondrial function

Gene symbolGenBank IDDescriptionFold changeP-value
Extrinsic pathway
CASP8NM_001228caspase 8, apoptosis-related cysteine peptidase1.211150.0033
CFLARAF015451CASP8 and FADD-like apoptosis regulator1.156560.0038
FASZ70519Fas (TNF receptor superfamily, member 6)1.239560.0013
LMNB1NM_005573lamin B11.183410.001
NFRKBNM_006165nuclear factor related to kappaB binding protein1.191630.0037
TNFAIP8BC005352tumor necrosis factor, alpha-induced protein 81.181750.0032
TNFRSF10BBC001281tumor necrosis factor receptor superfamily, member 10b1.369050.0005
TNFRSF21BE568134tumor necrosis factor receptor superfamily, member 211.195110.0037
TNIP2AA522816TNFAIP3 interacting protein 21.21280.0038
TRADDL41690TNFRSF1A-associated via death domain1.316970.0034
ER – associated pathway
ATF4NM_001675activating transcription factor 4−2.007550.0024
Intrinsic pathway and mitochondrial dysfunction
ABL1NM_005157v-abl Abelson murine leukemia viral oncogene homolog 1−1.495440.0034
API5NM_006595apoptosis inhibitor 5−1.184480.0036
BCL2L1AL117381BCL2-like 1−1.453050.004
BCLAF1NM_014739BCL2-associated transcription factor 1−1.390140.0012
ERCC2AI918117excision repair cross-compl. rodent repair deficiency, compl.1.221150.0021
FOXO3N25732forkhead box O3−1.870880.0004
GSTA1AL096729Glutathione S-transferase A11.20620.0041
LRPPRCM92439leucine-rich PPR-motif containing−1.959340.0015
MAPK6NM_002748mitogen-activated protein kinase 6−1.756070.0003
MAPK8IP3AB028989mitogen-activated protein kinase 8 interacting protein 3−2.294380.0036
PDCD2AA764988programmed cell death 21.263410.0042
PDCD6NM_013232programmed cell death 6−1.225360.0036
PPM1FD86995protein phosphatase 1F (PP2C domain containing)−1.437270.0024
PPP2CABC000400protein phosphatase 2 (formerly 2A), catalytic subunit, alpha−1.844780.0003
PPP2CBNM_004156protein phosphatase 2 (formerly 2A), catalytic subunit, beta−2.004120.0015
PPP5CNM_006247protein phosphatase 5, catalytic subunit1.310250.0031
PRKCAAI471375protein kinase C, alpha−1.544950.0032
SOD1NM_000454superoxide dismutase 1, soluble (ALS 1 adult)−3.399970.0017
SPHK2AA485440sphingosine kinase 2−1.763870.0039
TEGTNM_003217testis enhanced gene transcript (BAX inhibitor 1)−2.120680.0031
ATP5A1AI587323ATP synthase, H+ transport., mitochon. F1 complex, alpha 1−2.295640.001
ATP5G3NM_001689ATP synthase, H+ transport., mitochon. F0 complex, subunit C3−2.221580.0021
ATP5HAF061735ATP synthase, H+ transport., mitochon. F0 complex, subunit d−1.650630.0008
ATP5JNM_001685ATP synthase, H+ transport., mitochon. F0 complex, subunit F6−2.575950.0003
ATP5LNM_006476ATP synthase, H+ transporting, mitochondrial F0 complex, G−1.338540.0025
CA5ANM_001739carbonic anhydrase VA, mitochondrial1.12860.0022
COX5BNM_001862cytochrome c oxidase subunit Vb−1.856340.0013
COX6CNM_004374cytochrome c oxidase subunit Vic−2.040830.0044
COX7A2LNM_004718cytochrome c oxidase subunit VIIa polypeptide 2 like−2.029310.0002
COX7CNM_001867cytochrome c oxidase subunit VIIc−3.002460.0007
COX8ANM_004074cytochrome c oxidase subunit 8A (ubiquitous)−1.73930.0019
FHNM_000143fumarate hydratase−1.375150.0021
GK3PAA292874glycerol kinase 3 pseudogene1.187970.0002
GKAJ252550glycerol kinase1.291470.0046
GPD2U79250glycerol-3-phosphate dehydrogenase 2 (mitochondrial)1.230370.0013
HSPE1NM_002157heat shock 10kDa protein 1 (chaperonin 10)−1.4240.00003
IMMTNM_006839inner membrane protein, mitochondrial (mitofilin)−2.163090.0009
LARSNM_020117leucyl-tRNA synthetase1.515490.0033
LARS2D21851leucyl-tRNA synthetase 2, mitochondrial−1.454820.0012
LARS2D21851leucyl-tRNA synthetase 2, mitochondrial−1.454820.0012
MTCH1AF189289mitochondrial carrier homolog 1 (C. elegans)−2.838310.0035
MRPL15NM_014175mitochondrial ribosomal protein L15−1.347740.0014
MRPL3BC003375mitochondrial ribosomal protein L3−1.989920.0009
MRPL34AB049652mitochondrial ribosomal protein L34−1.329830.0028
MRPL40NM_003776mitochondrial ribosomal protein L40−1.273340.0012
MRPL9AB049636mitochondrial ribosomal protein L91.262860.0002
NDUFA1NM_004541NADH dehydrog. (ubiquinone) 1 alpha subcomplex, 1, 7.5kDa−2.081520.0042
NDUFA4NM_002489NADH dehydrog. (ubiquinone) 1 alpha subcomplex, 4, 9kDa−1.633210.0004
NDUFA6NM_002490NADH dehydrog. (ubiquinone) 1 alpha subcomplex, 6, 14kDa−2.43910.0013
NDUFAB1NM_005003NADH dehydrogenase (ubiquinone) 1 α/β subcomplex, 1, 8kDa−2.299080.0009
NDUFB2NM_004546NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 2, 8kDa−3.700750.002
NDUFB3NM_002491NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 3, 12kDa−2.436270.0023
NDUFB4NM_004547NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 4, 15kDa−2.020870.0011
NDUFB8NM_005004NADH dehydrogenase (ubiquinone) 1 beta subcomplex, 8, 19kDa−7.459410.0001
NDUFB11NM_019056NADH dehydrogenase (ubiquinone) 1 β subcomplex11, 17.3kDa−1.453510.0037
NDUFC1NM_002494NADH dehydrog. (ubiquinone) 1 subcomplex unknown, 1, 6kDa−1.827260.00004
NDUFS5NM_004552NADH dehydrogenase (ubiquinone) Fe-S prot. 5, 15kDa−2.8520.0029
NDUFS5NM_004552NADH dehydrogenase (ubiquinone) Fe-S protein 5, 15kDa−2.8520.0029
OATNM_000274ornithine aminotransferase (gyrate atrophy)−1.766670.0005
OAZ3AW611641ornithine decarboxylase antizyme 31.151140.0018
ODC1NM_002539ornithine decarboxylase 1−1.552640.0001
PCCBNM_000532propionyl Coenzyme A carboxylase, beta polypeptide−1.167830.0006
SDHCBG110532succinate dehydrog. complex, subunit C, int. mem. prot., 15kDa−1.984560.0012
SFXN3NM_030971sideroflexin 3−1.757340.0013
SUMO3NM_006936SMT3 suppressor of mif two 3 homolog 3 (S. cerevisiae)−2.687990.0002
TIMM17AAK023063translocase of inner mitochondrial mem. 17 homolog A (yeast)−2.595620.0021
TOMM20BG165094translocase of outer mitochondrial membrane 20 homolog (yeast)−1.092910.0007
UCRCNM_013387ubiquinol-cytochrome c reductase complex (7.2 kD)−1.949790.0004
UQCRC2NM_003366ubiquinol-cytochrome c reductase core protein II1.183310.001
UQCRHNM_006004ubiquinol-cytochrome c reductase hinge protein−2.878140.0001

Mitochondrial dysfunction and protein degradation

Inhibition of mitochondrial function and the impairment of the UPS have long been linked to Parkinson's disease pathology and are part of the intrinsic mechanisms of PCD (Bredesen et al., 2006; Gomez et al., 2007). Mitochondrial dysfunction is mainly characterized by the generation of reactive oxygen species (ROS), a decrease of mitochondrial complex I activity, cytochrome-c release, ATP depletion and caspase 3 activation. We found differential expression of multiple genes related to these signaling cascades (Table 3) and consistent with other results, downregulation was more prominent confirming reduced mitochondrial activity in Parkinson's disease (Duke et al., 2006). For example, there was downregulation of superoxide dismutase 1 (SOD1) and upregulation of glutathione S-transferase A1 (GSTA1), which are both implicated in protecting cells from ROS and the products of peroxidation (Raza et al., 2002; Martin et al., 2007), though SOD1 has recently also been shown to increase the production of toxic ROS in the intermembrane space of mitochondria (Goldsteins et al., 2008). The expression of several cytochrome c oxidase subunits was also markedly decreased as well as NADH dehydrogenase subunits and the mitochondrial mRNA-binding protein LRPPRC (Mootha et al., 2003).

Together with lysosomes, the UPS is part of the proteolytic machinery to degrade misfolded, damaged proteins, or proteins with an abnormal amino acid sequence. Defects in the proteolytic systems lead to accumulation and organization of cellular aggregates, such as Lewy bodies in the Parkinson's disease DA neurons (Olanow and McNaught, 2006). Our data demonstrate downregulation of gene clusters linked to ubiquitination (including the PARK genes HIP2, UCHL-1 and RAP1GA1, see above), chaperone function (e.g. heat shock and associated proteins), and subunits of the proteasome (Table 4). In this context, we also found decreased expression of ST13, a cofactor of heat-shock protein 70 (HSP70) that stabilizes its chaperone activity.

View this table:
Table 4

Genes associated with protein degradation

Gene symbolGenBank IDDescriptionFold changeP-value
SNCABG260394synuclein, alpha−1.8590.0003
ATP13A2NM_022089ATPase type 13A2−1.3780.0043
HSF1NM_005526heat shock transcription factor 1−1.49530.0005
HSF2BPNM_007031heat shock transcription factor 2 binding protein1.239590.0003
HSP90AA1R01140heat shock protein 90kDa alpha (cytosolic), class A member 1−5.87210.0026
HSPA8AA704004heat shock 70kDa protein 8−2.35710.0033
HSPE1NM_002157heat shock 10kDa protein 1 (chaperonin 10)−1.4240.00003
HSPH1NM_006644heat shock 105kDa/110kDa protein 1−1.59530.0038
DNAJC4NM_005528DnaJ (Hsp40) homolog, subfamily C, member 41.151860.0046
DNAJC7NM_003315DnaJ (Hsp40) homolog, subfamily C, member 7−1.802190.0002
UBBNM_018955ubiquitin B−5.94040.002
UBE1CAL117566ubiquitin-activating enzyme E1C (UBA3 homolog, yeast)−1.864470.0012
UBE2E1AL518159ubiquitin-conjugating enzyme E2E 1 (UBC4/5 homolog, yeast)−2.050430.0012
UBE3BAL096740ubiquitin protein ligase E3B1.169310.0043
USP10BC000263ubiquitin specific peptidase 101.295720.0028
USP34AB018272ubiquitin specific peptidase 34−1.94830.0032
USP34AW502434ubiquitin specific peptidase 341.200760.000008
USP47BE966019ubiquitin specific peptidase 47−2.46530.0024
UCHL1NM_004181ubiquitin carboxyl-terminal esterase L1−1.94420.004
UBA52AF348700ubiquitin A-52 residue ribosomal protein fusion product 1−2.14430.0045
SCRN1NM_014766secernin 1−2.091630.0017
CPENM_001873carboxypeptidase E−2.752120.0016
DNPEPNM_012100aspartyl aminopeptidase−1.09220.0006
ADAMDEC1NM_014479ADAM-like, decysin 11.175130.00009
PSEN2U34349presenilin 2 (Alzheimer disease 4)−2.530790.0009
HIP2NM_005339huntingtin interacting protein 2 (ubiquitin-conjugating enzyme)−1.22170.0021
PSMB4NM_002796proteasome (prosome, macropain) subunit, beta type, 4−2.36620.0046
PSMB5BC004146proteasome (prosome, macropain) subunit, beta type, 5−1.57740.0009
PSMC3AL545523proteasome (prosome, macropain) 26S subunit, ATPase, 31.13550.0013
PSMD4NM_002810proteasome (prosome, macropain) 26S subunit, non-ATPase, 4−2.23850.000009
PSMC3IPNM_013290PSMC3 interacting protein1.176770.0033
SUMO3NM_006936SMT3 suppressor of mif two 3 homolog 3 (S. cerevisiae)−2.687990.0002
AP3B2NM_004644adaptor-related protein complex 3, beta 2 subunit1.245630.0031
AP4E1AB030653adaptor-related protein complex 4, epsilon 1 subunit1.125760.0012
AP4S1BC001259adaptor-related protein complex 4, sigma 1 subunit1.311310.0034
HSPC152NM_016404hypothetical protein HSPC152−1.94570.001
GULP1AK023668GULP, engulfment adaptor PTB domain containing 11.185920.0014
ZFYVE9NM_007323zinc finger, FYVE domain containing 91.457510.0003
ATP6V0A1AL096733ATPase, H+ transporting, lysosomal V0 subunit a1−1.900860.001
ATP6V0A2AW444520ATPase, H+ transporting, lysosomal V0 subunit a21.342690.0038
ATP6V1E1BC004443ATPase, H+ transporting, lysosomal 31kDa, V1 subunit E1−3.699450.0004

Synaptic dysfunction

There was a number of deregulated genes which are involved in synaptic function and altogether there was more down- than upregulation (Table 5). In particular, expression of synaptogyrin 3 (SYNGR3) and NSF was diminished, which has also been described in a MPTP mouse model of Parkinson's disease (Miller et al., 2004). In contrast to Miller and Federoff (Miller and Federoff, 2005), we did not detect a down-regulation of the DAT-binding protein syntaxin-1A (Lee et al., 2004). However, we found down-regulation of the GABA transporter member 1 (SLC6A1), GABA receptor beta subunit 1 (GABRB1) and the GABA receptor-associated proteins (GABARAPL) 1, 2 and 3 (Table 6).

View this table:
Table 5

Genes associated with synaptic function

Gene symbolGenBank IDDescriptionFold changeP-value
Transport of peptide-containing vesicles to neuron terminal
    KIF5BBF223224kinesin family member 5B−1.593550.0041
    KIF5CNM_004522kinesin family member 5C−4.928520.0003
    KIF4ANM_012310kinesin family member 4A1.151030.0013
Vesicle reserve pool maintanance and vesicle mobilization
    SYN1H19843synapsin I−1.663030.0004
    ABLIM3NM_014945actin binding LIM protein family, member 3−1.350520.00007
    GTPBP4NM_012341GTP binding protein 4−1.532530.003
    NSFNM_006178N-ethylmaleimide-sensitive factor−3.230790.0003
    SV2ANM_014849synaptic vesicle glycoprotein 2A−2.441870.0018
    SV2BNM_014848synaptic vesicle glycoprotein 2B−2.946790.0002
    RIMS1AF263310regulating synaptic membrane exocytosis 11.221180.0014
    RIMS3NM_014747regulating synaptic membrane exocytosis 3−2.880550.0013
    CADPSNM_003716Ca2+-dependent secretion activator−1.479480.0016
    SYT1AV731490synaptotagmin I−4.132710.0026
    SYT12AK024381synaptotagmin XII1.317590.0039
    CLTANM_001833clathrin, light chain (Lca)−1.757410.0016
    CLTCNM_004859clathrin, heavy chain (Hc)−4.102730.0001
    DNM1AF035321dynamin 1−5.372610.0031
    DNM2NM_004945dynamin 2−1.207220.0039
    SYNJ2AK026758synaptojanin 21.316970.0022
Synaptic vesicle surface proteins
    SCAMP1NM_004866secretory carrier membrane protein 11.296750.0023
    STX8NM_004853syntaxin 8−1.344580.0012
    SYT1AV731490synaptotagmin I−4.132710.0026
    VAMP4NM_003762vesicle-associated membrane protein 41.192240.002
    SYN1H19843synapsin I−1.663030.0004
    VAMP8NM_003761vesicle-associated membrane protein 81.13530.0017
Proteins involved in synaptic plasticity
    SYNGR3NM_004209synaptogyrin 3−4.141380.0009
    SNCABG260394synuclein, alpha−1.858990.0003
    TUBA1AAF141347tubulin, alpha 1a−6.371570.0017
    TUBBBC005838tubulin, beta−1.720280.002
    TUBB2ANM_001069tubulin, beta 2A−11.8450.0009
    TUBB2BAL533838tubulin, beta 2B−3.786210.0018
    TUBB2CAA515698tubulin, beta 2C−3.042850.0011
    TUBB2CBC004188tubulin, beta 2C−2.385720.001
    TUBB3NM_006086tubulin, beta 3−3.848890.00003
    TUBD1BC000258tubulin, delta 11.326770.0008
    DYNC1I1NM_004411dynein, cytoplasmic 1, intermediate chain 1−3.301220.0033
    DYNLL1NM_003746dynein, light chain, LC8-type 1−2.849630.00006
    DYNLRB1NM_014183dynein, light chain, roadblock-type 1−1.865940.0045
View this table:
Table 6

Growth factors, receptors and ion-channels

Gene symbolGenBank IDDescriptionFold changeP-value
Growth factor—related transcripts
    CTGFM92934connective tissue growth factor1.2090.0029
    TGFBR3NM_003243transforming growth factor, beta receptor III1.2520.0005
    NFATC1U08015nuclear factor of activated T-cells, cytoplasmic, calcineurin-dependent 11.3860.0029
    NFATC2IPAA152202nuclear factor of activated T-cells, cytoplasmic, calcineurin-dependent 21.1860.0023
    NFKBIL2NM_013432nuclear factor of kappa light polypeptide gene enhancer in B-cells1.1490.0014
    NFRKBNM_006165nuclear factor related to kappaB binding protein1.19160.0037
    NFRKBAI887378nuclear factor related to kappaB binding protein1.32010.0016
    NGFRNM_002507nerve growth factor receptor (TNFR superfamily, member 16)1.210.0006
    NGFRAP1NM_014380nerve growth factor receptor (TNFRSF16) associated protein 1−4.310.0004
    TDGF1/3NM_003212teratocarcinoma-derived growth factor 1/31.7040.0018
    GDF3NM_020634growth differentiation factor 31.2630.003
    FGF21NM_019113fibroblast growth factor 211.0840.0037
    FGF23NM_020638fibroblast growth factor 231.3340.00003
    FGFR2M87771fibroblast growth factor receptor 21.1460.0025
    GFRA2U97145GDNF family receptor alpha 21.2730.0014
    PIK3C2GAJ000008phosphoinositide-3-kinase, class 2, gamma polypeptide1.284710.0012
    PIK3R1AI680192phosphoinositide-3-kinase, regulatory subunit 1 (p85 alpha)−1.755590.0037
    PIK3R2NM_005027phosphoinositide-3-kinase, regulatory subunit 2 (p85 beta)1.395380.0004
Neurotransmitter—related transcripts
    GABRB1NM_000812gamma-aminobutyric acid (GABA) A receptor, beta 1−2.950.0037
    GABARAPL1/3AF180519GABA(A) receptor-associated protein like 1−4.160.001
    GABARAPL2AB030710GABA(A) receptor-associated protein-like 2−1.530.0012
    GRIN2BU90278glutamate receptor, ionotropic, N-methyl D-aspartate 2B1.2670.0005
    GRM7X94552glutamate receptor, metabotropic 71.232740.0014
    DRD1X58987dopamine receptor D11.240.0045
    HTR1FNM_0008665-hydroxytryptamine (serotonin) receptor 1F1.1690.0044
    CHRNA4L35901cholinergic receptor, nicotinic, alpha 41.291010.0028
    CHRNB2NM_000748cholinergic receptor, nicotinic, beta 2 (neuronal)1.317870.0018
    SSTR4NM_001052somatostatin receptor 41.177160.0019
Ion channel—related transcripts
    KCNA10NM_005549potassium voltage-gated channel, shaker-related subfamily 101.207940.001
    KCNJ6U24660potassium inwardly-rectifying channel, subfamily J, member 6−1.503870.0034
    KCNK1U90065potassium channel, subfamily K, member 11.180890.0029
    KCMF1NM_020122potassium channel modulatory factor 1−2.095770.0021
    SCN3BAB032984sodium channel, voltage-gated, type III, beta−1.450190.0028
    SCN7ANM_002976sodium channel, voltage-gated, type VII, alpha1.172270.0004
    CACNB3U07139calcium channel, voltage-dependent, beta 3 subunit−2.696930.0036
    CLCNKA/KBNM_004070chloride channel Ka/chloride channel Kb1.419450.004
    ATP13A2NM_022089ATPase type 13A2−1.377970.0043
    ATP1B1NM_001677ATPase, Na+/K+ transporting, beta 1 polypeptide−4.960660.0004
    ATP2A3Y15724ATPase, Ca++ transporting, ubiquitous−1.590980.0009
    ATP2B2R52647ATPase, Ca++ transporting, plasma membrane 2−1.588110.0018
    ATP2C1AF189723ATPase, Ca++ transporting, type 2C, member 1−1.316860.0005
    SLC6A1AI003579solute carrier family 6 (GABA), member 1−1.678560.0008
    SLC6A2AB022847solute carrier family 6 (noradrenalin), member 21.233180.0033
    SLC11A2AF046997solute carrier family 11 (prot-coupled divalent metal ion transporters)1.186620.0031
    SLC16A3AL513917solute carrier family 16, 3 (monocarboxylic acid transporter 4)1.159390.0007
    SLC22A17NM_020372solute carrier family 22 (organic cation transporter), member 17−1.876880.0021
    SLC24A2NM_020344solute carrier family 24 (Na/K/Ca exchanger), member 21.188260.0025
    SLC24A3NM_020689solute carrier family 24 (Na/K/Ca exchanger), member 3−1.306680.0017
    SLC24A6NM_024959solute carrier family 24 (Na/K/Ca exchanger), member 61.363450.0014
    SLC34A1NM_003052solute carrier family 34 (sodium phosphate), member 11.236660.0043
    SLC35A1NM_006416solute carrier family 35 (CMP-sialic acid transporter), member A1−1.543450.0001
    SLC39A6AI635449solute carrier family 39 (zinc transporter), member 6−2.558730.0019
    SLC43A3AI630178solute carrier family 43, member 31.295240.0006

DA phenotype, survival and cytoskeleton

Interestingly, from the 1046 genes in our data set none of the ‘classical’ DA neuron-associated genes were significantly deregulated (e.g. TH, AADC, DAT, EN-1, NURR1), although there was a trend for reduced expression of TH and DAT by qRT-PCR (see below). We noticed an upregulation of a cluster of genes linked to cell survival (Table 6) indicating the activation of compensatory mechanisms in response to cell stress. These genes comprise mitogen-activated protein kinases (MAP3K3, MAP6, MAPK8IP3), growth factors (FGF21 and 23, GDF3, TDGF1/3), growth factor receptors and associated proteins (FGFR2, TGFBR3, NGFR, GFRA2, TNFRSF16, GDF3, DRD1, VDR), and other ion or neurotransmitter receptors (discussed separately below). In addition, there was downregulation of genes related to cytoskeletal maintenance (Table 5), e.g. dyneins, which are involved in the trafficking of cellular components, transport of organelles, cell–cell contact and cytoskeletal stability via interaction with β-catenins and microtubules. Strikingly, we found deregulation of microtubulin-associated genes like MAPT, MAPRE1, TCP1 [which take part in unfolding translated proteins in the cytosol, such as actin and tubulin (Stirling et al., 2007)] and multiple subunits of tubulin (Table 5), but not microtubule affinity regulating kinase (MARK1) and microtubule-associated protein (MAP2) as described elsewhere (Miller et al., 2006; Moran et al., 2007).

Ion channels and neurotransmitter receptors

Over the past years, there has been emerging evidence that survival of DA neurons depends on their unique properties of electrical activity involving Na+, K+ and Ca2+ channels and the association of mitochondrial dysfunction and ROS production with K+ and Ca2+ channel activation has been suspected as a major contributor to Parkinson's disease pathogenesis (Michel et al., 2007; Surmeier, 2007). Many molecules related to these mechanisms are dysregulated in our data set (Table 6). For example, there was striking downregulation of the Na+/K+-ATPase carrier protein (ATP1B1), which is involved in actively pumping Na+ out of and K+ into the cell plasma to maintain their electrochemical gradients. Another downregulated gene was the G protein-gated inwardly rectifying K+ channel 2 (GIRK2 or KCNJ6), which is predominantly expressed in the SNc DA neurons and has been implicated in Parkinson's disease (Kobayashi and Ikeda, 2006). In addition, the calcium channel subunit β3 (CACNB3), ATPase type 13A2 (PARK9, Table 2) and several subunits of Ca2+ transporting ATPases (ATP2A3, ATP2B2, ATP2C1) were downregulated further substantiating a deficit in organelle function and Ca2+ sequestering. Finally, our data demonstrate an upregulation of the glutamate receptors GRIN2B and GRM7 and the nicotinic cholinergic receptors α4 and β2 (CHRNA4, CHRNB2) (Table 6), which is consistent with the notion that NMDA and nicotinic acetylcholine (ACh) receptors contribute to DA neuronal survival (reviewed in Michel et al., 2007).

Validation of microarray data by TaqMan®-based real-time PCR

To validate the results from the microarray assays, we additionally performed TaqMan®-based real-time PCR on laser-microdissected cells from two new control and one new Parkinson's disease brain as well as control brain C3 and Parkinson's disease brains PD3 and PD4, which were used for the microarray analysis (Table 1). We selected the DA neuronal-specific genes tyrosine hydroxylase (TH), dopamine transporter (DAT or SNC6A3) and Girk2 (KCNJ6) (Table 6) and all PARK genes including LRKK2, which was not present on the HG-U133A Affymetrix chip. Using the 2−ΔΔCt method to determine fold differences of relative gene expression in Parkinson's disease versus control samples (Livak and Schmittgen, 2001; Schmittgen and Livak, 2008), the real-time PCR experiments largely confirmed the results from the microarrays (Fig. 1). However, we also observed high variability between samples, which prompted us to additionally analyse our results by comparing relative gene expression of individual genes using the 2−ΔCt method (Livak and Schmittgen, 2001; Schmittgen and Livak, 2008) for the real-time PCR assays and Z-scores for the microarrays after removal of batch effects (Supplementary Fig. S3). Although these data showed considerable variability of gene expression levels within each sample (Supplementary Fig. S3A) and across the sample population (Supplementary Fig. S3B), there was an overall consistency between both methodologies demonstrating a broad downregulation of PARK genes and, to some extent, also of TH and DAT in the PCR assays. The latter, however, did not reach significance in the microarrays using three-way ANOVA at FDR10%.

Figure 1

Validation of gene expression using TaqMan® real-time PCR on three control and three Parkinson's disease samples (Table 1). The following genes were selected: tyrosine hydroxylase (TH), dopamine transporter (DAT), Girk2 (KCNJ6), SNCA (PARK1), Parkin (PARK2), UCHL1 and HIP2 (PARK5), PINK1 (PARK6), DJ-1 (PARK7), LRRK2 (PARK8), ATP12A2 (PARK9), RAP1GA1, RIMS1 and 3 (PARK10). Data analysis was based on the 2−ΔΔCt method (Livak and Schmittgen, 2001; Schmittgen and Livak, 2008) and results were plotted as fold differences of relative gene expression normalized to controls.


Studying Parkinson's disease pathogenesis using microarray technology

Multiple microarray studies have compared the gene expression profiles of cells within the midbrain of normal controls with those from Parkinson's disease brains (Grunblatt et al., 2004; Hauser et al., 2005; Zhang et al., 2005; Duke et al., 2006; Miller et al., 2006; Moran et al., 2006, 2007; Moran and Graeber, 2008). These studies were based on sections encompassing substantia nigra as well as other adjacent regions such as striatum and thalamus, and therefore, contained a large amount of cells other than DA neurons. Consequently, microarray analyses on dissected tissue revealed a global set of genes that are dysregulated in Parkinson's disease, which is in agreement with an increasing conceptual view that not only the DA neurons, but also other cells within the substantia nigra and adjacent brain regions are involved in Parkinson's disease pathology (summarized in Duke et al., 2006). Altogether, these studies confirmed several cellular functions that are affected in Parkinson's disease, such as the UPS and the mitochondrial system, synapse function, DA phenotype, and cytoskeletal maintenance pointing to defects in cell communication, survival and axonal transport (Duke et al., 2006; Miller et al., 2006). However, they do not provide gene expression of single DA neurons. So far, three groups reported expression profiling on directly targeted DA neurons by laser microscopy (Lu et al., 2004, 2006; Cantuti-Castelvetri et al., 2007; Grundemann et al., 2008). Two of these groups used laser capture microscopy (LCM) with an Arcturus PixCell II instrument after quick immunostaining or ethanol fixation and methylene blue staining of the dissected midbrain tissue. This differs from our and Grundemann et al.'s approach, in which LMD was performed on unprocessed freshly cut sections and the DA neurons visualized by their neuromelanin content. In addition, the LMD-isolated neurons fell by gravity into collection tubes, in contrast to fixation of the cells on the slide matrix by LCM. We attempted to compare our results with the microarray data published by (Cantuti-Castelvetri et al., 2007), but unfortunately in this study a different Affymetrix platform with a different probe set (U133_X3P) was used (http://www.ebi.ac.uk/microarray-as/ae/). Based on our analysis criteria (three-way ANOVA, FDR10), we were not able to retrieve differential gene expression profiles as seen in our study.

It should be noted that an important parameter in the interpretation of the LMD-based microarray data refers to the integrity and status of the isolated cells. Especially downregulation of gene expression could be a result of neuronal death that is not necessarily related to a dysfunction of pathways associated with Parkinson's disease pathogenesis. Therefore, it should be emphasized that gene expression in this study should be viewed in the context of biological function and—when deregulated—in relation to a possible role in pathogenic processes that are linked to Parkinson's disease.

Deregulated gene expression as indicator for dysfunctional cellular pathways in Parkinson's disease

PARK genes

PARK proteins are associated with familial forms of Parkinson's disease and their functions have been linked to all major pathways related to Parkinson's disease pathogenesis including mitochondrial and synaptic dysfunction, protein degradation, PCD and cell survival (Moran et al., 2007; Olanow, 2007; Thomas and Beal, 2007; Burke, 2008; Schiesling et al., 2008). Although there is evidence that both forms of Parkinson's disease share common pathogenic mechanisms, it is still unclear if, and to what extent, the familial-linked PARK proteins are involved in the sporadic illness. Our data show a striking downregulation of most of the PARK genes. Since PARK1, RIMS1 and RIMS3 are involved in vesicular function, PARK2, PARK5 and RAP1GA1 with the UPS, PARK6 in mitochondrial function, PARK7 in intrinsic pathways of PCD, and PARK8 in cytoskeletal process regulation, it appears that a deregulation of these molecules might also contribute to the pathogenesis of sporadic Parkinson's disease. Thus, our data could support the view that the PARK genes might present a significant group of key factors in common pathogenetic mechanisms of both forms of Parkinson's disease (Moran et al., 2007; Thomas and Beal, 2007; Burke, 2008; Schiesling et al., 2008).

Cellular pathways involved in Parkinson's disease pathogenesis

Multiple cellular pathways have been associated with Parkinson's disease pathogenesis and one of the key mechanisms relates to processes involved in PCD. These comprise a large subset of molecules that also include some of the PARK genes, such as PARKIN, PINK1 and DJ-1 (Tatton et al., 2003; Burke, 2007, 2008; Moran et al., 2007; Olanow, 2007; Singh and Dikshit, 2007; Schiesling et al., 2008). Many of the functional aspects of these molecules stem from experimental models of Parkinson's disease and have been extensively summarized elsewhere (e.g. Olanow, 2007; Singh and Dikshit, 2007; Burke, 2008). However, there is only very little information available from Parkinson's disease patient's material other than rather controversial and mixed results from morphologic assessments (Tatton et al., 2003; Burke, 2007, 2008). Our data show a set of deregulated genes that are directly or indirectly involved in PCD confirming the current concept of apoptotic cell death of the DA neuron. Particularly interesting is the observed upregulation of genes involved in extrinsic PCD, because there have been several observations on postmortem brain tissue suggesting a role of TNF-α and FAS signalling in the neurodegeneration of Parkinson's disease (Boka et al., 1994; Mogi et al., 1996; Ferrer et al., 2000; Hartmann et al., 2001, 2002; Burke, 2007). In addition, our data show a dysfunction of both the mitochondria and the UPS, which are major contributors to PCD and Parkinson's disease pathogenesis (Duke et al., 2006). This included multiple cytochrome c oxidase and NADH dehydrogenase subunits that have been recently associated with impaired mitochondrial function in pesticide-induced Parkinson's disease (Gomez et al., 2007). Interestingly, there was a decrease of LRPPRC expression, a gene linked to the mitochondrial neurodegenerative disorder French-Canadian-type Leigh syndrome, which is caused by defects in oxidative phosphorylation (Mootha et al., 2003) and ST13, which is part of a number of marker genes (including HIP2) that have been proposed as possible biomarkers in Parkinson's disease (Scherzer et al., 2007). It should be noted that SNCA, a component of Lewy bodies, whose pathologic accumulation is caused by oxidative stress, mitochondrial dysfunction and impairment of cellular proteolytic mechanisms (Lundvig et al., 2005) was also deregulated.

There were several deregulated genes pointing to impairment of synaptic function and plasticity and some of these genes were also observed in other studies, such as SYNGR3, NSF, SV2B, SYN1, SYT1 and dynamin (Miller et al., 2006). The deregulated genes in our study belong to important mechanisms involved in maintaining synaptic function and integrity, such as a number of proteins from the SNARE complex (priming of the synaptic vesicle and synaptic vesicle surface proteins) that play a role in vesicle binding and fusion to the plasma membrane (Brunger, 2005). Other downregulated genes encode the GTPase family-associated molecules dynamin 1 and 2, which are involved in severing nascent vesicles from the membrane, receptor-mediated endocytosis, trafficking in and out of the Golgi apparatus, maintenance of mitochondrial morphology and mitochondrial-associated pathways of apoptosis (Scorrano, 2007; Ungewickell and Hinrichsen, 2007). In addition, there was striking down-regulation of genes related to cytoskeletal maintenance including MAP kinases, tubulins and dyneins, while several growth factor receptor and their signalling-associated genes were upregulated. We also found downregulation of GABA receptor and signalling-related genes supporting the previous suggestion that GABAergic synapses are reduced in the substantia nigra of Parkinson's disease resulting in a reduction of DA neuron inhibition and an increase in neurotransmission and function of the remaining functional DA neurons (see below) (Miller and Federoff, 2005). Altogether our results are consistent with other observations pointing to a functional disconnect of the striatonigral trophic signalling pathways (Miller et al., 2006).

Our data also support evidence from other investigators suggesting that survival of DA neurons depends on their unique properties of electrical activity involving Na+, K+ and Ca2+ channels. For example, Michel et al. proposed a mechanism in which the dysfunctional mitochondria and ROS trigger adenosine triphosphate-sensitive K+ (KATP) channel-mediated hyperpolarization of substantia nigra DA neurons, which renders them susceptible to degeneration (Michel et al., 2007). We found a striking downregulation of the Na+/K+-ATPase carrier protein (ATP1B1) that is involved in actively pumping Na+ out of and K+ into the cell plasma to maintain their electrochemical gradient. Mutation in this gene causes rapid-onset dystonia Parkinsonism (de Carvalho Aguiar et al., 2004). It should be noted that SOD (or SOD mimetics) can abolish the K+-mediated hyperpolarization by inhibiting ROS formation (Liss et al., 2005) and expression of SOD was markedly downregulated in our data. Also, there was downregulation of GIRK2 expression, which can cause permanent depolarization and loss of spontaneous pacemaker activity and, thus, contributes to cell death (Liss et al., 2005). Other receptors that have been implicated in the long-term survival of DA neurons are L-type Ca2+ channels, which drive their pace-making activity by sustaining low intracellular Ca2+ concentrations that are sequestered by the ER and mitochondria using ATP-dependent transporters (Surmeier, 2007). These energy-consuming processes require oxidative phosphorylation, a prominent feature of DA neurons. In combination with the generation of ROS and consecutive mitochondrial DNA damage this high metabolic rate might accelerate their ageing—including dysfunctional proteins that are directly or indirectly involved in these processes, e.g. some of the PARK genes including ATPase type 13A2 (Surmeier, 2007). Our data show a reduction in multiple calcium channel subunits including ATPase type 13A2 (PARK9) and several subunits of Ca2+ transporting ATPases adding to the overall picture of an imbalanced Ca2+ homeostasis of the Parkinson's disease DA neuron. Finally, neurotransmitters have also been implicated in the survival of DA neurons (reviewed in Michel et al., 2007). NMDA receptors seem to be involved in controlling their burst-firing mode and enhance the survival promoting effect of BDNF. However, there is also evidence that they contribute to degeneration through an excitotoxic process. Nicotinic ACh receptors protect DA neurons in vitro and in vivo against MPTP or 6-OHDA toxicity and their effects are attributed to a reduction of glutamate-meditated excitotoxicity, upregulation of trophic factors, or a rise in intracellular Ca2+ (see above). This is particularly interesting, since the ACh receptors α7, α4 and β2 have strong depolarizing activity on DA neurons consisting with the view that modulation of their excitability might support survival (Matsubayashi et al., 2004; Quik et al., 2007). Taken together, the upregulation of glutamate nicotinic cholinergic receptors in our data set contributes to the interpretation that compensatory survival mechanisms are activated in response to cell stress mediated by PCD, protein degradation, mitochondrial and synaptic dysfunction.

Insights into Parkinson's disease pathogenesis through a ‘molecular fingerprint’ identity of the parkinsonian DA neuron

Miller and Federoff postulated a model for common pathways of Parkinson's disease pathogenesis based on microarray data collection (Miller and Federoff, 2005). This model encompasses several genes that are involved in the function or dysfunction of DA neurons in Parkinson's disease model systems and postmortem brain analyses from Parkinson's disease patients. Downregulated genes are related to the DA phenotype, synaptic function, cytoskeletal stability and axonal transport, while upregulated genes refer to metabolism, protein disposal and inflammation. Among the postulated genes, we found no significant down- or upregulation of DAT, AADC, EN1, MARK-1, MAP2, DSCR1L1, HK1, ZFP162 and UNC-5. However, and also consistent with other reports, there was a downregulation of SYNGR3 (Miller and Federoff, 2005), Synaptotagmin 1 (SYT1) (Zhang et al., 2005; Moran et al., 2006), N-ethylmaleimide-sensitive factor (NSF) (Miller and Federoff, 2005; Zhang et al., 2005), UCHL-1 (Moran et al., 2007), kinesin family members (KIF5B and KIF5C) (Miller et al., 2006), and dynein-related genes (DYNC1I1, DYNLL1 and DYNLRB1) (Miller and Federoff, 2005). Although several of these genes are linked to pathways in DA pathogenesis (see above), we could not confirm the six genes in the Miller and Federoff study (Miller and Federoff, 2005), which are postulated as a highly conserved dysregulation in the three Parkinson's disease systems analysed (i.e. DAT, EN-1, HK-1, DSCR1L1, ZFP 162 and UNC-5). Given that many of their cellular functions in DA neurons are currently unknown (except of DAT and EN-1) further studies will be needed to confirm their direct or indirect involvement in Parkinson's disease pathology.

A recent publication by Moran and Graeber (2008) provided an extensive pathways analysis based on 892 dysregulated priority genes from a Parkinson's disease substantia nigra microarray data set. The authors concluded that Parkinson's disease has biological associations with cancer, diabetes, and inflammation. In addition, this study revealed prominent changes in similar cell function and disease pathways evident from our data, such as apoptosis, cell survival, cytoskeleton, signal transduction, synaptic and mitochondrial function, protein degradation and networks that are directly linked to Parkinson's disease-associated genes. These investigators also found a strong association with inflammation and, interestingly, a cluster of upregulated genes related to functions of the immune system are also present in our data set (Supplementary Table 3S). This might add further evidence to an involvement of inflammatory processes in the disease development of Parkinson's disease (Whitton, 2007). Altogether, comparison of our results with the data from these and other investigators as discussed above suggests that there are two major classes of factors involved in Parkinson's disease pathogenesis:

  1. A core of highly conserved primary (priority) factors that are major players of key pathways in the function of the substantia nigra DA neuronal phenotype; and

  2. Secondary factors that are directly or indirectly affected by (or effect) the dysfunction of the primary molecules. Dysregulation of molecules from both classes contribute to Parkinson's disease.

It is important to emphasize that mRNA data reveal information about transcriptional activation of genes, but do not tell much about actual protein levels and function. In addition, array data cannot predict if deregulated gene expression is a primary or a secondary effect of cell function. For example, a gene could be down- or upregulated by factors, such as miRNAs or transcriptional activators (or inhibitors) independent of its protein function and/or as a consequence of positive and negative feedback loops. Moreover, protein function relies on the interaction of down- and upstream factors within a pathway, i.e. downstream factors are more dependent on upstream signalling than upstream factors, which might influence a cascade of downstream events that can include multiple pathways. Thus, the consequences of deregulated gene expression are on multiple levels within a complex and dynamic interplay of factors and mechanisms. Lasermicroscopy-based microarray studies can only reveal a ‘snap-shot’ of these events. Nevertheless, our study shows that many genes associated with Parkinson's disease pathogenesis are deregulated in single captured postmortem DA neurons. This could provide a ‘molecular fingerprint identity’ of a late stage DA neuron affected by sporadic Parkinson's disease. A key aspect is the striking downregulation of PARK genes. Since their mutation-induced malfunction in the familial forms of Parkinson's disease rapidly accelerates DA neuron degeneration, the results from our study could support the view that these genes are also involved in the pathogenesis of sporadic Parkinson's disease. Our data also point to an imbalance in the neuronal homeostasis and stress characterized by factors related to high metabolic rate, neurotransmission and ion-channel activity. This stress might be part of the DA neuronal normal homeostasis and aging, but could exacerbate when there is an unfavourable imbalance. In addition, the array data suggest a disintegration of key cellular functions, such as mitochondria-associated energy metabolism, protein degradation, synaptic function and cytoskeletal integrity revealing a cellular state that is characterized by PCD. However, despite this cellular demise, some genes linked to survival mechanisms were upregulated indicating the activation of compensatory mechanisms. Finally, the lack of or relatively modest deregulation of genes important for the DA neuronal phenotype suggests that the DA neurotransmitter identity (including DA production) seems to be sustained even when the neurons are severely damaged. Altogether, it appears that the gene expression profile of late stage Parkinson's disease DA neurons is consistent with the view that Parkinson's disease is a complex disorder and that multiple factors and cellular pathways are involved in its pathogenesis.

Supplementary material

Supplementary material is available at BRAIN online.


Massachusetts’ Alzheimer's Disease Research Center (partial); Harvard NeuroDiscovery Center (partial); NIH/NINDS N5057460 (partial).


The authors want to thank Dr. Donna McPhie for reading the manuscript and providing critical comments.


  • Abbreviations:
    laser microdissection
    programmed cell death
    postmortem interval
    ubiquitin-proteasome system


View Abstract