Edited by: George E. Barreto, Pontificia Universidad Javeriana, Colombia
Reviewed by: Nibaldo C. Inestrosa, Pontifical Catholic University of Chile, Chile; Valentina Echeverria Moran, Bay Pines VA Medical Center, USA
*Correspondence: Richard H. Myers
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Parkinson's disease (PD) is a progressive movement disorder, characterized clinically by resting tremor, rigidity, bradykinesia and postural instability (Parkinson,
While neuropathological hallmarks contribute to the degeneration of the nigrostriatal dopaminergic system, the etiology, clinical heterogeneity and fundamental pathological mechanisms by which preferential neuronal loss occurs in PD are largely unknown. Monogenic forms of PD represent only a minority of cases (Hamza and Payami,
The microRNA (miRNA) profile of PD brains may offer insight into the molecular and pathological mechanisms that occur in the disease. miRNAs are short, noncoding RNAs that inhibit translation through sequence-specific binding of the 3′-untranslated region of target messenger RNAs. miRNAs post-transcriptionally regulate a set or multiple sets of genes simultaneously, and in the brain, their regulatory effects have profound effects on neuronal development, differentiation and maturation (O'Carroll and Schaefer,
Although the substantia nigra (SN) is the most affected brain region in PD, as many as 80% of dopaminergic neurons are lost by the time of death (Cheng et al.,
Sequence analysis allows for high-resolution, genome-wide quantification of miRNA molecule abundance. We compared miRNA levels between PD and controls to identify miRNA alterations and based on their miRNA profile, classified PD from non-diseased brains. Among the PD brains, we identified miRNAs associated with the presence or absence of dementia and age of motor onset, as well as identified a set of miRNAs altered in both PD and Huntington's disease, which may be relevant to the pathological processes that occur broadly across age related neurodegenerative diseases.
Frozen brain tissue from prefrontal cortex Brodmann Area 9 (BA9) for 29 PD samples and 33 control samples was obtained from the National Brain and Tissue Resource for Parkinson's Disease and Related Disorders at Banner Sun Health Research Institute, Sun City, Arizona (Beach et al.,
Control | 33 | – | – | 68.1 ± 14.8 | 15.0 ± 8.7 | 7.6 ± 0.7 |
All diagnosed PD | 29 | 66.5 ± 9.8 | 10.5 ± 6.5 | 77.6 ± 9.0 |
11.1 ± 9.7 | 7.3 ± 0.7 |
PD, Non demented | 18 | 64.1 ± 7.2 | 11.5 ± 6.4 | 76.1 ± 8.9 | 11.9 ± 9.2 | 7.2 ± 0.8 |
PD with dementia | 11 | 69.8 ± 12.2 | 9.2 ± 6.7 | 79.9 ± 9.0 | 9.9 ± 10.9 | 7.5 ± 0.5 |
Total RNA was isolated using QIAzol Lysis Reagent and purified using miRNeasy MinElute Cleanup columns. Samples were prepared using Illumina's TruSeq Small RNA Sample Prep Kit, according to the manufacturer's protocol, and sequenced on Illumina's HiSeq 2000 system with 1x51nt single-end reads at Tufts University and the Michigan State sequencing core facility.
Reads were processed and counted as described previously (Hoss et al.,
To further evaluate the differential expression patterns between PD cases and controls, unsupervised, Ward hierarchical clustering by Euclidian distance was applied using the heatmap2 function in the gplots R package (Warnes et al.,
To determine if miRNAs were associated with PD in the presence or absence of dementia, VST counts for PDN and PDD were compared using LIMMA, adjusting for age at death, and
Linear models were used to model the relationship between age of motor onset and miRNA levels among the 21 PD samples with onset data. Tests were performed genome-wide and exclusively among the set of differentially expressed miRNAs. FDR-adjusted
Finally, to determine whether overlap in miRNA alterations exist between PD and HD brain, results from PD differential expression analysis were compared to those of our previously published Huntington's disease (HD) study (Hoss et al.,
The miRNA sequence data analyzed here can be accessed from NCBI's Gene Expression Omnibus, series accession number
To identify miRNA differences in PD vs. non-disease subjects, miRNA sequence analysis was performed in prefrontal cortex (Brodmann Area 9) for 33 controls and 29 idiopathic PD samples (see Table
We used classification models to determine whether the levels of PD-related miRNAs in brain could accurately assign disease status. To select the most informative miRNAs, we filtered on effect size (LFC > 0.5 or LFC < −0.5). After filtering, 29 PD-related miRNAs were used in an unsupervised hierarchical cluster analysis. Samples clustered based on disease status with the exception of five PD which clustered with the controls, (see Figure
To assess whether miRNA differences specific to the PDD subtype were distinguishable from a generalized PD response, we performed a differential expression analysis comparing PDN to PDD using normalized VST count data from 18 PDN and 11 PDD samples. We observed no genome-wide significant (
To test whether PDD had increased miRNA alterations in comparison to PDN, we created an ordered categorical variable (controls, PDN, and PDD) and tested the association of this variable to genome-wide miRNA levels. 105 of the 125 differentially expressed miRNAs had a significant ordinal association (
We further investigated the clinical utility of these miRNA profiles for the assessment of dementia using classification analyses (WeightedXVoting; Reich et al.,
To understand whether deregulated miRNAs were specific to PD, or a general response to the neurodegenerative process, we compared miRNA that were significantly altered in PD to those significantly altered in Huntington's disease (HD). Twenty one miRNAs were found differentially expressed in both PD and HD experiments and the probability of this overlap occurring was more than that expected by chance (
Within the PD case sample, we tested the association of age of motor onset of PD with miRNA levels. Although we did not observe significance in a genome-wide analysis, restricting our study to the 125 significantly differentially abundant PD miRNAs revealed miR-10b-5p to have a significant, positive association to onset age (beta = 0.040,
In this study, we identified 125 miRNAs altered at genome-wide levels in PD prefrontal cortex using next-generation sequencing. This is the largest miRNA sequencing analysis performed in PD vs. control brain samples (29 vs. 33, respectively), the first to provide a detailed miRNA PD profile, classify brains by miRNA levels and to evaluate the relationship of miRNA levels in brain to relevant clinical features.
Reduced levels of miR-133b (Kim et al.,
Several miRNAs that we report altered in PD brain may interact with PD-related genes. Monogenic forms of PD include mutations within the alpha-synuclein gene (
Glucocerebrosidase (GBA) deficiency is associated with PD (Aharon-Peretz et al.,
We were able to classify PD based on the levels of 29 miRNAs with less than a 5% error rate, although this require external validation in an independent sample to confirm. While this classification was performed using postmortem brain samples, we believe this may be relevant for PD biomarker discovery, particularly if these miRNAs are peripherally detectable. We reasonably differentiated PD subtypes (PDN/PDD) based on miRNA levels, and we observed a pattern of increased changes in the PDD samples relative to the PDN samples in the set of altered miRNAs. We observed that the majority of differentially expressed miRNAs had an ordinal relationship to controls and PD cases stratified by the presence or absence of dementia, suggesting PDD may represent a more severe alteration of the PD miRNA profile.
Our study in PD brains identified profiles of miRNAs that distinguish PD from controls, which if also observed in peripheral biofluids, such as blood or cerebrospinal fluid (CSF), could be valuable in the evaluation of PD diagnosis, prognosis, or progression. The small size (~22 nucleotides) of miRNA may allow for neuropathologically altered miRNA to cross the blood-brain barrier in exosomes (Kalani et al.,
We previously reported 75 miRNAs altered in HD prefrontal cortex (Hoss et al.,
Of the two miRNAs with discordance between PD and HD, miR-10b-5p emerged due its relationship to onset in both diseases. miR-10b-5p is markedly increased in HD in comparison to controls, and has a negative association to age of onset for HD, with higher levels of miR-10b-5p corresponding early onset age. In contrast, miR-10b-5p is significantly decreased in PD, and has a positive association to onset age, where higher miRNA levels correspond to later onset ages. In a separate Alzheimer's disease (AD) study, examining miRNA levels in prefrontal cortex, miR-10b-5p levels were significantly reduced in AD (Lau et al.,
This study provides evidence that miRNA levels are altered in PD prefrontal cortex. These changes are sufficiently consistent that diseased brains can be discriminated with high confidence from non-diseased brains based on the level of 29 miRNAs. PDD may represent a more severe profile of PD related miRNAs than PDN. 21 miRNAs changes were similar between PD and HD, with the exception of miR-10b-5p, which had opposite direction of effects to disease association and to motor onset age in the two diseases. Further characterization of miR-10b-5p in the neurodegenerative disease context is warranted to better understand if it has clinical potential as a biomarker for disease progress or to identify potential therapeutic targets.
Conceived and designed the experiments: AH, JL, RM. Performed the experiments: RM, AH. Analyzed the data: AH, AL, JL. Contributed reagents/materials/analysis tools and critically reviewed the manuscript: RM, TB. Wrote the paper: AH, AL, JL, RM.
Supported by grants from US National Institutes of Health (R01-NS076843, Characterization of the Role of Cyclin G-associated Kinase in Parkinson Disease, (R01-NS073947), Epigenetic Markers in Huntington's Disease Brain, National Science Foundation, PHY-1444389 Early-concept Grants for Exploratory Research (EAGER), U24 NS072026 National Brain and Tissue Resource for Parkinson's Disease and Related Disorders.
AH, AL, JL and RM declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. TB is a consultant for GE Healthcare and Avid Radiopharmaceuticals. The reviewer VEM and handling Editor declared a current collaboration and the handling Editor states that the process nevertheless met the standards of a fair and objective review.
We would like to acknowledge the National Brain and Tissue Resource for Parkinson's Disease and Related Disorders at Banner Sun Health Research Institute (NS072026), Sun City, Arizona, the Harvard Brain Tissue Resource Center McLean Hospital, Belmont, Massachusetts, and the Human Brain and Spinal Fluid Resource Center VA, West Los Angeles Healthcare Center, California for providing the brain samples used in these studies.
The Supplementary Material for this article can be found online at: