Global microRNA Expression Profiling of Caenorhabditis elegans Parkinson's Disease Models
- First Online:
- Cite this article as:
- Asikainen, S., Rudgalvyte, M., Heikkinen, L. et al. J Mol Neurosci (2010) 41: 210. doi:10.1007/s12031-009-9325-1
- 490 Views
MicroRNAs (miRNAs) play an important role in human brain development and maintenance. To search for miRNAs that may be involved in the pathogenesis of Parkinsons disease (PD), we utilized miRNA microarrays to identify potential gene expression changes in 115 annotated miRNAs in PD-associated Caenorhabditis elegans models that either overexpress human A53T α-synuclein or have mutations within the vesicular catecholamine transporter (cat-1) or parkin (pdr-1) ortholog. Here, we show that 12 specific miRNAs are differentially regulated in the animals overexpressing α-synuclein, five in cat-1, and three in the pdr-1 mutants. The family of miR-64 and miR-65 are co-underexpressed in the α-synuclein transgenic and cat-1 strains, and members of let-7 family co-underexpressed in the α-synuclein and pdr-1 strains; mdl-1 and ptc-1 genes are target candidates for miR-64 and miR-65 and are overexpressed in α-synuclein transgenic as well as miR-64/65 (tm3711) knockout animals. These results indicate that miRNAs are differentially expressed in C. elegans PD models and suggest a role for these molecules in disease pathogenesis.
KeywordsmicroRNAParkinson's diseaseNeurodegenerationMicroarrayqRT-PCRCaenorhabditis elegans
Parkinson’s disease (PD) is the second most common neurodegenerative disease and is characterized by the progressive loss of dopaminergic (DAergic) neurons within the substantia nigra and locus ceruleus. Cellular hallmarks of PD also include the accumulation of proteinaceous intracellular inclusions termed Lewy bodies (LB) in surviving DAergic neurons. However, some patients lack LBs which suggest that the etiology of the disease may involve a complex array of inherited and environmental factors (Tan and Skipper 2007; Lazzarini et al. 1994; Braak et al. 2003; Hardy et al. 2006). Familial studies indicate that a number of mutant proteins likely play a significant role in the disease pathogenesis. The most abundant protein in LBs is the presynaptic protein α-synuclein. Although the precise function of α-synuclein is not clear, α-synuclein may play a role in synaptic vesicle formation, regulation of dopamine (DA) biosynthesis, and in neurotransmission. Mutations within α-synuclein have been shown to increase DA neuron vulnerability, degeneration, and propensity to develop PD (Polymeropoulos et al. 1997; Farrer et al. 1999; Singleton et al. 2003). Mutations within the Parkin gene, an E3 ubiquitin protein ligase that is involved in proteasomal degradation of damaged proteins, has also been associated with an increase in risk for developing PD (Kitada et al. 1998; Shimura et al. 2000; Mata et al. 2004). Mutations within α-synuclein and parkin can result in an early onset form of the disease (Mata et al. 2004). However, the mechanisms by which both genes contribute to the pathogenesis are not clear.
MicroRNAs (miRNAs) are a conserved class of short RNAs that regulate gene expression by sequence-specific base pairing with target mRNAs (Lee et al. 1993; Reinhart et al. 2000, 2002; Ambros et al. 2003; Alvarez-Garcia and Miska 2005). miRNAs are transcribed as extended RNA precursors (pre-miRNAs) that form complex secondary structures of several hundred nucleotides (nt). Pre-miRNAs are processed in the nucleus by the microprocessor protein complex that contains the RNAse III enzyme Drosha and its cofactors that excises an approximately 70 nt long-stem loop to form the pre-miRNA. Pre-miRNAs are exported from the nucleus by a RanGTP/exportin 5-dependent mechanism and are then excised to generate mature miRNAs of approximately 22 nt in length. The mature cytoplasmic miRNA is then incorporated into the RNA-induced silencing complex that mediates inhibition of target mRNA translation or mRNA degradation (Denli et al. 2004; Kim and Kim 2007). One hundred and seventy-four Caenorhabditis elegans miRNAs are annotated in miRBase database (Welcome Trust Sanger Institute); yet, the actual number of miRNAs may be higher (Lim et al. 2003a). miRNAs have been shown to hybridize to mRNA 3′ UTRs and more recently to 5′ UTRs or coding regions to inhibit protein translation (Tay et al. 2008). The region close to the 5′ end of the miRNA is also known to be important in determining the target sequence and has been termed the seed region (Lai 2002; Lim et al. 2003a, b; Ambros et al. 2003). Members of miRNA families share close sequence identity in their seed regions even though they are often derived from separate primary transcripts (Roush and Slack 2008; Abbott et al. 2005). Gene family members also tend to inhibit common mRNAs, and a single miRNA may inhibit as many as 100 different target genes (Roush and Slack 2008; Abbott et al. 2005; Lim et al. 2005). miRNA concentrations within C. elegans may be as high as 30,000 copies per cell suggesting efficient ongoing translational gene repression (Lim et al. 2003b).
Mammalian miRNAs play a significant role in brain morphogenesis, neuronal specification, function, and maintenance (Krichevsky et al. 2003, 2006; Sempere et al. 2004; Schratt et al. 2006). Several microarray studies indicate that miRNAs regulate the expression levels of specific genes to maintain or modify cellular identity and suggests that miRNAs may have played an essential role in the development of complex tissues and organs in higher organisms (Frasch 2008; Cacchiarelli et al. 2008; Heimberg et al. 2008).
Recent vertebrate studies suggest that miRNA pathways may be playing a critical role in the development of PD. Deletion of Dicer in mice show that miRNAs are necessary for DA neuronal survival and inhibition of PD-associated cellular phenotypes. DA neuron-specific human miR-133b has also been shown to regulate maturation and function of midbrain DAergic neurons and its expression is downregulated in PD (Kim et al. 2007; Schaefer et al. 2007). Furthermore, miR-7 has been shown to repress α-synuclein mRNA expression levels and is decreased in PD-associated neurotoxin cell and rodent models (Junn et al. 2009).
In this study, we utilized NCode miRNA microarrays to evaluate miRNA expression level changes in C. elegans strains that either overexpress α-synuclein or have a functional deletion on VMAT or Parkin gene (Lakso et al. 2003; Sulston et al. 1975; Duerr et al. 1999; Loer and Kenyon 1993; Ved et al. 2005; Springer et al. 2005). Each of these strains model key genetic and/or biochemical attributes of PD and should allow us to identify miRNAs involved in PD-associated gene expression and pathogenesis. Our results indicate that two specific families of miRNAs, miR-64/65 and let-7, are differentially regulated in these genetic backgrounds. Furthermore, we identified candidate target genes and pathways. Our studies suggest that miRNA dysregulation occurs in multiple models of PD and furthermore provides molecular insight into the expression and/or maintenance of PD.
Materials and Methods
C. elegans Strains
Strains N2 (wild-type (wt)), CB1111 (cat-1; e1111), VC1024 (pdr-1; gk448) were obtained from the Caenorhabditis Genetics Center (CGC). VC1024 has a 372-bp deletion removing the 5′ UTR and coding region and RT-PCR analysis confirmed that no transcript was detectable under conditions that identified a transcript in N2 strain (data not shown). The pan neuronal promoter aex-3 overexpressing transgenic strain ls10 [aex-3::α-synuclein A53T; dat-1::gfp] was constructed and integrated as previously reported (Lakso et al. 2003). The miR-64/65 mutant allele tm3711 contains a 150-bp deletion overlapping both miR-64 and miR-65 transcripts. Nematodes were grown on OP50 Escherichia coli and nematode growth media agar as previously described (Brenner 1974).
Sample Preparations for Microarray Analysis
Microarray data are available in gene expression omnibus (accession no. GSE14899). Animals were synchronized by hatching purified eggs using 1% hypochlorite solution with 250 mM KOH, grown at 20°C and harvested at Larva 4 (L4) stage. mirVanaTM miRNA Isolation protocol (Ambion, Austin, TX) was used to isolate small (<200 nt) RNAs containing miRNAs. RNA samples were isolated as four independent biological replicates (worm populations) for each strain.
NCodeTM miRNA Labeling System (Invitrogen, Carlsbad, CA) was used to prepare small RNAs for hybridizations on NCodeTM Multispecies miRNA Microarray V2-arrays (Invitrogen). Each array contains 115 probes for annotated C. elegans miRNAs in miRBase release 13 printed as three subarrays. In addition, an array contains Alexa Fluor® Dye Control probes for normalization of differences in fluorescent signal intensities between two dye channels. RNA from wild-type (N2) strain was labeled with Alexa Fluor® 3 dye and RNA from the strain of investigation with Alexa Fluor® 5 dye. After labeling, comparable RNA samples (wt and mutant) were pooled and hybridized on arrays resulting in at least four biological replicate arrays for each strain of investigation. Arrays were scanned using ScanArray 5000 (GSI Lumonics, Bedford, MA) and ScanArray Express (Perkin Elmer, Waltham, MA) software generating two color signal intensity images.
TIGR Spotfinder software (Saeed et al. 2003) was used to filter unreliable spots based on Otsu's method algorithm (Liao et al. 2001) and to convert "good" spots in microarray images to numerical form. Raw signal values were analyzed using mean centering to normalize signal values between channels. Signals from Alexa Fluor® Dye Control probes were excluded from normalization process due their oversaturation leaving only probe intensities from C. elegans miRNAs to use in normalization. Signal intensity ratios were calculated for each signal pair yielding fold change values and p values for each miRNAs from the strain of investigation. A cutoff of 1.4 was used to detect modestly changing miRNAs.
Quantitative Real-Time PCR
C17G1.4, left-5′-CCCGCCTCTGAGTTTGAGAA, right-5′-AAACAAGACAGAGTAGACGA; F15A8.1, left-5′-ATGCTTACACAGCGATCTAA, right-5′-TTATTCCTGCAAATGCTGCT; mdl-1, left-5′-TTGCACTGTTAACACGAGCC, right-5′-GTGAAAGCGTTTGGCAAGCT; mls-2, left-5′-CGATGATGAAGCCAACTCTA, right-5′-CAAGTGCAGCTTCTGTGCTA; nas-13, left-5′-ATTGTGAATTTCTTGCAAGA, right-5′-CTGTTTGGTTGCACATTCCA; F17E5.2, left-5′-CAAGCTATCCGCTAGCTTTA, right-5′-ATGCTGACCGCAGGGATTAC; F58H10.1, left-5′-AAGGAACACCTACCAAGCAT, right-5′-GAGAAGGATGATTTTTGAAG; ptc-1, left-5′-ATGTGGTTGTATCATTCCTG, right-5′-GAGCCGTCATGACAATAAAG; R05H11.2, left-5′-AAAGGATCCAGTTGTTTGT, right-5′-TCCACATTGATAAGGTCTTC; rgl-1, left-5′-TGACGGTGCAAATTATAAAT, right-5′-ATCGCATAGAATGGGTTGCA; syg-1, left-5′-TGATCAATCTGACGTCACAG, right-5′-TCCGGTTGCACCGAGGAAATC; W01C8.5, left-5′-ATGGCCGGAAGATGTTCATT, right-5′-GGCTCGCTGACGGCTTAGAG. C. elegans actin gene, act-1, was used as a control; act-1, left-5′-TCG GTATGGGACAGAAGGAC, right-5′-CATCCCAGTTGGTGACGATA.
Fold changes were calculated using the 2-(ΔΔCt) method (Livak and Schmittgen 2001). All oligonucleotides were purchased from Oligomer Oy, Helsinki, Finland.
Differentially Expressed miRNAs
Differentially expressed miRNAs
Quantitative Real-Time PCR of miRNAs
Microarray x ± SE (L4) (qRT-PCR x ± SE) (L4) (qRT-PCR x ± SE) (L1)
1.84 ± 0.33
1.86 ± 0.10
1.00 ± 0.14
(1.45 ± 0.66)
(1.62 ± 0.17)
(1.26 ± 0.14)
(1.11 ± 0.09)
(1.03 ± 0.18)
(3.40 ± 0.69)
0.56 ± 0.08
0.67 ± 0.06
0.94 ± 0.20
(0.39 ± 0.04)
(0.92 ± 0.07)
(1.02 ± 0.06)
(1.14 ± 0.11)
(0.93 ± 0.08)
(1.71 ± 0.20)
0.65 ± 0.09
0.64 ± 0.05
0.95 ± 0.12
(0.54 ± 0.06)
(1.01 ± 0.05)
(1.15 ± 0.02)
(1.14 ± 0.17)
(1.06 ± 0.32)
(2.12 ± 0.39)
qRT-PCR analysis of two other C. elegans PD models also showed dysregulation of these miRNAs: djr-1.1 showed overexpression of miR-64 (4.52 ± 1.25) and underexpression of miR-65 (0.62 ± 0.08); pink-1 showed overexpression of miR-64 (2.17 ± 0.42) and underexpression of miR-65 (0.61 ± 0.07; data not shown).
Quantitative Real-Time PCR of Candidate miR-64/65 Targets
4.06 ± 0.35 ***
1.25 ± 0.05 *
3.64 ± 0.68 **
18.4 ± 0.90 ***
3.16 ± 0.43 **
0.11 ± 0.01 ***
1.62 ± 0.29
1.67 ± 0.58
1.53 ± 0.16 *
1.05 ± 0.04
1.22 ± 0.13
1.48 ± 0.16 *
1.20 ± 0.19
2.17 ± 0.18 ***
1.16 ± 0.14
0.33 ± 0.05 ***
1.06 ± 0.30
4.60 ± 0.76 **
0.99 ± 0.10
2.18 ± 0.15 ***
0.68 ± 011
0.74 ± 0.17
0.54 ± 0.04 **
0.27 ± 0.02 ***
The gene targets of the let-7 family were previously confirmed using computational and biochemical methods in C. elegans (Denli et al. 2004; Abrahante et al. 2003; Hayes et al. 2006; Grosshans et al. 2005).
A large proportion of mammalian miRNAs have been shown to be brain-specific or brain-enriched that suggests complex tissue-specific gene regulation (Sempere et al. 2004; Lagos-Quintana et al. 2002). Many miRNAs act in close association with transcription factors comprising complex regulatory networks and feedback loops in an attempt to maintain optimal gene expression patterns during neuron development and maturity that includes synaptic plasticity and axon formation (Kim and Kim 2007; Krichevsky et al. 2003, 2006; Sempere et al. 2004; Schratt et al. 2006; Grosshans et al. 2005; Ferretti et al. 2008; Yu et al. 2008). The loss in expression of specific miRNAs in aged vertebrates results in neurodegeneration, and a decrease in miR-133b levels has been associated with PD (Kim et al. 2007; Schaefer et al. 2007). A recent report also indicates that miR-7 is able to repress α-synuclein expression and is downregulated in neurotoxin-associated PD models (Junn et al. 2009). To further characterize the role that miRNAs may play in the cellular pathology, we examined miRNA gene expression changes and targets in C. elegans PD models.
Twelve miRNAs are differentially expressed in α-synuclein A53T transgenic animals that show a strong uncoordinated phenotype and a loss of dopaminergic neurons (Lakso et al. 2003). The differential expression observed is unlikely caused by integration of the transgene since another integrated strain ls84 [aex-3::α-syn(A53T), dat-1::gfp] exhibits similar downregulation of miR-64 (0.77 ± 0.06) and miR-65 (0.63 ± 0.08) at the L4 stage (data not shown). Five miRNAs were observed as differentially regulated in VMAT defective cat-1 mutants. These animals exhibit a sensing defect characterized by their inability to modify feeding behavior in the presence of food (Duerr et al. 1999). The abnormal behavior is also induced by laser ablation of dopamine containing cells suggesting that this phenotype is conferred by a defect in dopaminergic cells (Sawin 1996; Duerr et al. 1999). Only three miRNAs were differentially regulated in the human parkin mutants, which has no observable movement deficit phenotypes. These differentially expressed miRNAs appear to be developmentally regulated differential miRNA expression was not observed in L1 animals.
miR-64 and miR-65 was observed to be underexpressed in L4 stage α-syn (A53T) transgenic and cat-1 mutant C. elegans. To elucidate the functional role of these miRNAs, their putative molecular targets were investigated by combining the target search results from miRBase Targets, TargetScanWorm, and PicTar. We then experimentally investigated whether some of these target candidates were over- or underexpressed in α-syn (A53T) transgenic animals in parallel with the miR64/65 genetic knockout. Two candidate gene targets mdl-1 and ptc-1 were found to be overexpressed in both α-syn (A53T) transgenic and miR64/65 knockout animals; mdl-1 encodes a basic-helix-loop helix protein (bHLH) that acts as a transcription factor similar to the vertebrate MAD transcriptional regulators. mdl-1::gfp promoter fusions are expressed in a number of different tissues, including the posterior intestine, anterior and ventral cord neurons, pharyngeal and body wall muscles, somatic gonad precursors, and hypodermal cells. ptc-1 encodes an ortholog of the Drosophila PATCHED (PTC) and human PTCH which contains a sterol sensing domain. PTC-1 is also required for cytokinesis in the germ line, and its expression and activity are confined to the germ line and its progenitors (Kuwabara et al. 2000).
The let-7 family members are observed to be differentially expressed in α-synuclein transgenic and in the human Parkin ortholog pdr-1 mutant. Members of this gene family have been shown to control the timing of fate specification of neuronal and hypodermal cells during larval development (Yokota et al. 2003). Let-7 and miR-84 have also been shown to regulate molting by controlling hypodermal seam cell division and cuticle formation during transition from L4 to adult stage by their target genes lin-41 and C. elegans hunchback hbl-1 (known also as lin-57) (Abrahante et al. 2003). Moreover, miR-48, miR-84, and miR-241 regulate the transition from L2 to L3 stage, likely by inhibiting hbl-1 activity (Abbott et al. 2005). The let-7 family target hbl-1 is also expressed in different neuronal classes, but its function in neurons remains to be elucidated. Interestingly, the expression of C. elegans homolog of Alzheimers disease beta-amyloid precursor protein apl-1 has been shown to be controlled by the let-7 family of miRNAs (Niwa et al. 2008). As beta-amyloid is also found in Lewy bodies associated with PD (Revuelta et al. 2008), the let-7 family members could provide a protective role in amyloid-beta accumulation.
We have previously profiled the gene (mRNA) expression changes from L4 α-synuclein-expressing animals (Vartiainen et al. 2006). The large number of genes dysregulated in that study, more than would be predicted based on miRNA control only, suggests that there are multiple gene regulatory mechanisms involved in the pathogenesis in the C. elegans PD models. Indeed, we find little overlap between the proposed targets of miR-58, miR-64, and miR-65 and the mRNAs dysregulated in the prior study. Another factor that remains to be resolved is determining how the PD-associated genes affect the expression of miRNAs. The chromosomal locations of differentially expressed miRNAs are spread across the genome suggesting a lack of locally coordinated regulation (Fig. 1). In addition, the location of the pdr-1 (I, 13,771332) and cat-1 mutations (X, 5,710,545) are not located close to regulated miRNAs as shown in Fig. 1, suggesting the lack of involvement of nearby chromosomal acting elements. Prior transcript profiling analysis of α-synuclein (A53T) transgenic animals revealed the downregulation of several histone coding genes (Vartiainen et al. 2006). Considering that histones are both structural components of chromatin and transcriptional regulatory molecules, our studies suggest that these proteins could possibly be involved in selectively altering miRNA expression.
Although C. elegans miR-58, miR-64, and miR-65 do not have easily identified human orthologs, elucidating the role of microRNAs in human PD is still in the early stages; miR-7 has been shown to repress α-synuclein expression, and other human miRNAs may also regulate its expression (Junn et al. 2009). In addition to miR-7, multiple alignment of human, rat, mouse, dog, and chicken 3′ UTRs from α-synuclein mRNA revealed seed sequences for miR-643, miR-61/miR-1285, miR-1183, miR-338-5p, and miR-153 (data not shown). These studies also identified microRNA seed sequences in a multiple alignment of parkin genes in the identical five species for miR-543, miR-579, miR-580, and miR1267 (data not shown) suggesting that microRNAs may be involved in the regulation of key PD associated genes. In addition, the microarray platform used contains probes for some but not all known miRNAs. There are 60 microRNAs and 20 miRNAs in the miRBase release 14 that are missing from the used microarray platform. These missing miRNAs are listed in Supplementary Table 2.
In this study, we identified differential regulation of the miR-64, miR-65, and let-7 family members in C. elegans models of PD. Analysis of their predicted targets suggest that a bHLH transcription factor and miRNA-mediated protective mechanisms are involved in the pathogenesis. Co-regulation of miRNA expression suggests that C. elegans PD models share common neuropathophysiologic mechanisms that can be observed at the post-transcriptional level.
We thank the Finnish Graduate School of Neurosciences (FGSN), the Saastamoinen Foundation, and the Sigrid Juselius Foundation for their financial support. This study was supported in part by NIH ES014459 and MHRP W81XWH-05-1-0239 to RN. We thank Petri Pehkonen and Antti Kurronen for advice in computer analysis and Markus Storvik, Vuokko Aarnio, and Julia Vistbakka for fruitful discussions. We thank Dr. Shohei Mitani (Tokyo Women’s Medical University) for providing djr-1.1 and pink-1, and miR-64/65 (tm3711) mutant strains. Some strains used in this study were obtained from the C. elegans Genetics Center, which is funded by the NIH.
SA planned and performed the experiments and wrote the manuscript. MR, LH, KL, JV, and ML performed experiments. GW planned and performed experiments and helped write the manuscript. RN provided critical reagents, guided the study, and helped write the manuscript.