Thyroarytenoid muscle gene expression in a rat model of early-onset Parkinson’s disease

Abstract

Objectives:

Voice disorders in Parkinson’s disease (PD) are early-onset, manifest in the preclinical stages of the disease, and negatively impact quality of life. The complete loss of function in the PTEN-induced kinase 1 gene (Pink1) causes a genetic form of early-onset, autosomal recessive PD. Modeled after the human inherited mutation, the Pink1−/− rat demonstrates significant cranial sensorimotor dysfunction including declines in ultrasonic vocalizations. However, the underlying genetics of the vocal fold thyroarytenoid (TA) muscle that may contribute to vocal deficits has not been studied. The aim of this study was to identify differentially expressed genes in the TA muscle of 8-month-old male Pink1−/− rats compared to wildtype controls.

Methods:

High throughput RNA sequencing was used to examine TA muscle gene expression in adult male Pink1−/− rats and wildtype controls. Weighted Gene Co-expression Network Analysis was used to construct co-expression modules to identify biological networks, including where Pink1 was a central node. The ENRICHR tool was used to compare this gene set to existing human gene databases.

Results:

We identified 134 annotated differentially expressed genes (p<0.05 cutoff) and observed enrichment in the following biological pathways: Parkinson’s disease (Casp7, Pink1); Parkin-Ubiquitin proteasome degradation (Psmd12, Psmd7); MAPK signaling (Casp7, Ppm1b, Ppp3r1); and inflammatory TNF-α, Nf-κB Signaling (Casp7, Psmd12, Psmd7, Cdc34, Bcl7a, Peg3).

Conclusions:

Genes and pathways identified here may be useful for evaluating the specific mechanisms of peripheral dysfunction including within the laryngeal muscle and have potential to be used as experimental biomarkers for treatment development.

1. Introduction

Parkinson’s disease (PD) is a multisystem, degenerative disorder that affects millions of people worldwide.^1,2 Cranial sensorimotor deficits, including vocal communication changes and impairments, impact 90% of individuals diagnosed with PD.^3–6 Dysarthria, dysphonia, and dysprosody manifest early in disease progression (preclinically), prior to the hallmark nigrostriatal brain degeneration of PD, are refractory to standard pharmaceutical therapies, and significantly diminish social interactions and patient quality of life.^7–12 Despite this significant negative health impact, management of vocal dysfunction in PD is limited by our understanding of underlying early-onset disease pathology within the subsystems of voice including the resonating and respiration systems as well as within the laryngeal system including the vocal fold.

The classic pathology of PD is the progressive death of dopaminergic neurons within the substantia nigra along with the accumulation of alpha-synuclein-containing protein aggregations (Lewy bodies). These markers positively correlate to limb motor impairments and progressive cognitive decline at the time of and after a formal diagnosis. However, recent research has shown that PD is a complex disease that likely manifests in multiple systems including the enteric, autonomic, and other peripheral systems up to a decade prior to clinical diagnosis (reviewed in Shapira et al., 2017).¹³ Of interest, changes to vocalization parameters manifest early and decline in most PD patients. Vocalization impairments do not respond to standard dopamine replacement therapies,^14–16 suggesting that the mechanism of dysfunction within the vocal fold muscle differs from that of the limb. Data from human post-mortem (late-stage PD) studies have found the presence and accumulation of abnormal neuromuscular pathology (alpha-synuclein, Lewy Bodies) within the peripheral structures of the pharynx and larynx.^17–19 However, the prodromal/preclinical period associated with the onset of vocal dysfunction is a time period that is difficult to study in human subjects. Therefore, a genetic model of early-onset PD, the Pink1−/− (phosphatase and tensin homolog [PTEN]-induced putative kinase 1) rat, is useful to study early-stage tissue pathology.

Rats produce social ultrasonic vocalizations (USVs) that occur in the 50-kilohertz (kHz) range and are analogous to human vocal communication.^20–23 Rat USVs are generated on an egressive airflow, forced air flow through the vocal fold,^24,25 and modulated, in part, by the intrinsic thyroarytenoid (TA) muscles.^{23,26–28,29} Rat laryngeal anatomy consists of a dense cartilage framework containing an intralaryngeal ventral pouch important to resonatory features of the USV.³⁰ USVs are hypothesized to be produced via an air jet passing glottal and supraglottal spaces. Vocal folds are essential to this whistle-like production because several studies have demonstrated that vocal folds must be adducted for USV production to occur. This complex vocalization behavior requires the contraction of intrinsic muscles, including that of the of the TA which shortens and adducts the vocal fold, all together resulting in the physical mechanism of the egressive airflow and whistle through the larynx and vocal output.^31–34 Previous work in normal rats has shown that TA muscle activity (measured by electromyographic recordings) influence USV call duration and frequency.^25,32,35 Thus, dysfunction within the TA muscle may provide insight into the pathogenesis of early-onset vocal deficits in PD. Earlier research has shown that male Pink1−/− rats exhibit early and progressive USV deficits compared to healthy wildtype (WT) control rats.^36–38 These changes in vocalization, including decreases in intensity (decibels (dB)) and pitch (bandwidth (kHz), peak frequency (kHz)), appear independent of severe nigrostriatal dopamine loss as they do not respond to L-dopa, but may be related to significant neuropathological findings including α-synuclein in the brainstem motor regions that control vocal output.³⁹ Other work has shown that respiratory behavioral measures in Pink1−/− rats are similar to WT controls suggesting that dysfunction may be pinpointed at the level of the laryngeal source of the vocalization (e.g. the primary muscle of the vocal fold), thus providing further evidence to investigate the TA muscle.⁴⁰ For example, preliminary investigation of the muscular properties of the TA muscle in the Pink1−/− rat show increases in centralized nuclei and changes to muscle fiber profiles with decreases in muscle fiber size and increased 2L myofiber densities.⁴¹ These muscular changes are negatively correlated to vocalization intensity (unpublished data), suggesting a deficit in the peripheral muscle relates to vocal performance.

Large scale gene expression studies in post-mortem PD, Alzheimer’s, Amyotrophic Lateral Sclerosis (ALS) tissue, and normal healthy controls has led to the generation of databases of the gene expression underlying these diseases. Often though, whole genome association studies of PD are performed with late-stage or postmortem tissues. To date, there has been no examination of the gene expression changes in laryngeal muscles in the Pink1−/− rat model of early-onset PD. The goal of the present study was to identify differentially expressed genes using RNA-sequencing and bioinformatic analysis of the TA muscle of male Pink1−/− rats compared to WT controls. We hypothesized that loss of Pink1 from the TA muscle will enable us to: (1) detect changes in expression of genes that interact with Pink1, (2) catalog molecular pathways that may modulate voice and provide avenues for vocal-specific treatments, and (3) compare gene overlap between multiple degenerative disorders.

2. Materials and Methods

2.1. Rats and housing

Eight (n=4 Pink1−/−, n=4 WT) male Long Evans rats (Envigo^™ Research Labs, Boyertown, PA, USA) were aged to 8 months of age.^42,43 All rats were housed in groups of two (within like genotypes) in standard polycarbonate cages (290 mm × 533 mm × 210 mm) on a reversed 12:12 hour light: dark cycle. Food and water were available ad libitum. All experimental procedures were approved by the University of Wisconsin-Madison Animal Care and Use Committee (IACUC) and were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory animals.⁴⁴

2.2. Muscle tissue harvest and processing

Rats were deeply anesthetized with isoflurane and rapidly decapitated. The whole larynx was grossly dissected from the rat. Isolation of the TA muscle was performed as quickly as possible to avoid degradation of biological targets. The chilled larynx was prepared on the dissection surface under a dissection microscope. Using a micro-forceps and micro-scissors, the thyroid cartilage, arytenoid cartilages, and epiglottis were visualized, and TA muscles were isolated by removing them from all cartilaginous attachments. To remove the TA muscle from cartilaginous attachments, the larynx was bisected longitudinally, cutting through the thyroid cartilage vertically to bisect the region between the right and left TA at their insertion into the thyroid cartilage. The posterior larynx was then bisected between the right and left arytenoid cartilage within the back of the larynx. On each side, the TA was dissected away from the thyroid cartilage and arytenoid cartilage.⁴⁵ Individual dissected samples from the left and right TA muscles were placed in Eppendorf microcentrifuge tubes and promptly frozen at −80 until processing. Samples were processed by an experimenter blinded to genotype. Tissue was homogenized with an electric sonic dismembrator (Fisher Scientific, Hampton, NH, USA) and RNA was extracted with the Bio-Rad Aurum Total RNA Fatty and Fibrous Tissue Kit (Catalog No. 732-6830; Bio-Rad, Hercules, CA, USA) according to manufacturer’s instructions. Total RNA was measured using a Nanodrop system (Thermo Scientific, Wilmington, DE, USA) and the 28S:18S rRNA was quantified with an Agilent RNA 6000 Pico kit (Eukaryote Total RNA Pico, Agilent Technologies, Santa Clara, CA) and verified with the Agilent 2100 bioanalyzer (Agilent Technologies, Santa Clara, CA, USA).

2.3. Library construction and RNA sequencing

All RNA-sequencing procedures followed guidelines by ENCODE and were performed by the University of Wisconsin-Madison Biotechnology Center’s Next Generation Sequencing Facility with the Illumina® Total RNA-Seq TruSeq platform (Illumina Inc., San Diego, CA, USA). The Illumina Stranded Total RNA Library Prep Kit was used to remove rRNA and generate sequencing libraries as described by the manufacturer and in Kelm-Nelson and Gammie, 2020.⁴⁶ Sequencing was performed on an HiSeq 2000 high-throughput sequencing system within a single run (Illumina, Inc). After removal of adaptor sequences, contamination and low-quality reads, reads were mapped to the annotated rat (Rattus Norvegicus) genome in Ensembl.⁴⁷ Technical quality was determined using several parameters, detailed in Kelm-Nelson and Gammie, 2020. A number of differentially expressed genes were identified (raw data (RSEM), is displayed in Supplementary Table 1).

2.4. Differential gene expression analysis

Gene analysis was performed using the EdgeR Bioconductor Package, v. 3.9.⁴⁸ The p-value cutoff was set to 0.05 for significance and a Benjamini-Hochberg correction was applied to control the False Discovery Rate (FDR).⁴⁹ EdgeR results are provided in Supplementary Table 2. The RSEM approach for normalizing RNA seq data was used.⁵⁰ Raw data were uploaded to the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE151209; GSE151209), and genes were ranked according to p value and sorted by up- or down-regulation.

2.5. Weighted gene co-expression network (WGCNA) analysis

A WGCNA was used to construct co-expression networks and modules from the gene expression dataset. Data were log2(x+1) transformed, low expression genes were removed and WGCNA was run (16593 number of genes) using R software (https://www.r-project.org/).⁵¹ Correlations were raised to a soft thresholding power β of 12. Unsupervised hierarchical clustering for WGCNA included: the minimum module size of 30 genes, the signed mode, the deepSplit parameter set to 2, the mergeCutHeight parameter set to 0.15, and a threshold setting for merging modules of 0.25. Modules were visualized using Cytoscape (v3.7.1; https://cytoscape.org/).

2.6. Gene enrichment analysis

In order to identify genes that are over-represented in the data and associated with particular functions and relevance to PD, gene enrichment analysis was performed on both the differentially expressed gene dataset and the gene modules produced by the WGCNA analysis. Genes were classified by their gene ontology (GO) and gene enrichment was performed with EnrichR.⁵²

3. Results

Gene expression from the TA muscle of Pink1−/− male rats was compared to WT male rats at 8 months of age. As confirmation of the model, Pink1 was the most significantly downregulated gene. The list of the 134 up- and down-regulated protein coding (annotated) genes and their corresponding GO (verified with NCBI gene ontology based on their function, biological process, and component) are in Supplementary Table 3. Top gene ontology categories were immune, epidermal, metabolism, cell signaling adhesion and transport, as well as muscle and epithelial.

A WGCNA was used to construct the gene co-expression biological networks, with defined modules (clusters and network nodes) from the gene expression dataset searchable networks were created (Supplementary Table 4). There were 51 modules created from the WGCNA analysis. Of these modules, the Lightgreen module (arbitrary name designed by R software) was chosen for further analysis in this study because it contained the Pink1 gene as the central node with 104 interacting genes. The data are visualized using Cytoscape (Figure 1). ENRICHR was used to evaluate gene ontology and biological processes from the Lightgreen module list; several areas of enrichment were identified (Table 1) including TNF-alpha NF-kB signaling pathway, Parkinson disease, and proteasome degradation.

Figure 1. Lightgreen WGCNA module.

The interaction network of the differentially expressed genes was visualized using Cytoscape 3.3.0 software. Data were plotted using weight (level of significance) as a factor. Only annotated genes are plotted from the dataset. Pink1 (yellow) was the top gene with the highest number of significant connections. Lines represent significant correlations between two genes.

Table 1.

Gene enrichment analysis of the “lightgreen” module containing the Pink1 gene.

ENRICHR WikiPathways	p-value	Genes
TNF-alpha NF-kB Signaling Pathway: WP246	8.43E-05	CASP7; PSMD12; PSMD7; CDC34; BCL7A; PEG3
Parkinson Disease Pathway: WP3638	0.009887	CASP7; PINK1
Proteasome Degradation: WP519	0.017132	PSMD12; PSMD7
Synthesis and Degradation of Ketone Bodies: WP543	0.019104	BDH1
MAPK Signaling Pathway: WP493	0.023501	CASP7; PPM1B; PPP3R1
Alzheimer’s Disease: WP2075	0.033882	CASP7; PPP3R1

Gene Abbreviations: CASP7= Caspase 7; PSMD12= Proteasome 26S Subunit, Non-ATPase 12; PSMD7= Proteasome 26S Subunit, Non-ATPase 7; CDC34= Cell Division Cycle 34, Ubiquitin Conjugating Enzyme; BCL7A= BAF Chromatin Remodeling Complex Subunit BCL7A; PEG3= Paternally Expressed 3; Pink1= PTEN-induced kinase 1; BDH1= 3-Hydroxybutyrate Dehydrogenase 1; PPM1B= Protein Phosphatase, Mg2+/Mn2+ Dependent 1B; PPP3R1= Protein Phosphatase 3 Regulatory Subunit B, Alpha.

4. Discussion

The TA muscle, the primary muscle of the vocal fold, is integral to the anatomic positioning of the vocal fold within the larynx structure to produce the substrate for the air jet as part of the vocalization mechanism and subsequent whistle via egressive airflow.²⁵ The Pink1−/− rat, an early-onset model of inherited PD, demonstrates early and progressive disruption of social communication USVs as well as biological changes to the TA muscle.^41,53 The present study used RNA sequencing to evaluate gene expression changes and relevant biological pathways in the TA in both male Pink1−/− rats and healthy WT controls. These findings are critical to establishing links between genetic animal models for research and human disease databases.

The manageably sized list of differentially expressed (annotated) genes that was generated (134) were categorized by biological function, process, and component according to NCBI Gene Ontology (GO). Of this list, only 19 genes were down-regulated in expression, while the remaining 115 genes were up-regulated. One down-regulated gene of interest is neurotrophic receptor tyrosine kinase 2 (Ntrk2) which is involved in both central and peripheral nervous system regulation of neuron survival, proliferation, differentiation, and plasticity through MAPK and NF-κB signaling cascades (discussed below). In addition, only 4% of genes of the significantly expressed genes were muscle-related and included myosin heavy chain 7 (Myh7), myosin light chain 3 (Myl3), troponinI1 (slow skeletal type, TNNI1), and myozenin 2 (Myoz2). Several of these genes are included in the regulation of muscle contraction, development, and response to stress.

The majority of this dataset’s differentially expressed genes were categorized as immune-related (14%). Interestingly, 3% of genes were also related to cell death by apoptosis. These data expand on the ever-growing body of literature that shows that loss of function of Pink1 disrupts immune response pathways and leads to inflammation-induced neuronal cell death.⁵⁴ For instance, adenosine deaminase (Ada), up-regulated in Pink1−/− rats, is involved in a variety of immune processes including T-cell activation and B-cell differentiation as well as response to cellular stressors such as hydrogen peroxide and hypoxia. In contrast, there were also several down-regulated genes that involve immune processes including C-X-C motif chemokine ligand 9 (Cxcl9), immunoglobulin heavy constant gamma 1 (Ighg1), Immunoglobulin J chain (Jchain), myxovirus (influenza virus) resistance 1 (Mx1), and MX dynamin like GTPase 2 (Mx2). Mx1 has also been found to regulate synapse structure or activity and postsynaptic neurotransmitter receptor internalization. This is consistent with previous cell culture work that suggests that Pink1 deficiency increases pro-inflammatory gene expression as well as the presence of reactive oxygen species (ROS) such as nitric oxide (NO) in mouse glial cells.⁵⁴ While inflammation has not yet been fully researched, our data is the first to suggest that inflammatory pathways may be integral in the Pink1 model in vivo; this has been previously demonstrated in muscle cells within other models of PD.⁵⁵ In humans, symptoms of chronic laryngitis and inflammation of the laryngeal tissues include edema, hoarseness, and loss of voice. Vocal fold edema is known to impact vibration adversely, affecting the pitch. PD inflammation processes within the vocal folds may lead to supraglottal hyperadduction, which can change the structure and function of the speech musculature.⁵⁶ Future research should investigate the direct role of inflammatory pathways within the TA muscle and subsequent alterations in rodent vocalizations.

Eight genes in the kallikrein family, which have diverse physiological functions and are found in nearly all tissues and fluids, were significantly up-regulated in the TA of Pink1−/− rats. Recent literature suggests kallikreins may serve as possible biomarkers for many diseases. For example, Klk8 is an up-regulated gene of interest as it been implicated as a potential biomarker and future therapeutic target for AD.⁵⁷ Furthermore, Klk6 is highly expressed in the nervous system and has been implicated in the pathogenesis of multiple disorders including Alzheimer’s disease, multiple sclerosis, and PD. For example, recent studies have shown that Klk6 is directly involved in the degradation and turnover of alpha-synuclein.⁵⁸ Past work has shown that in Pink1−/− rats, alpha-synuclein is present in brainstem regions important for vocal function as well as in the nucleus ambiguus, where motor neuron cell bodies for speech and swallowing reside.⁵⁹ While alpha-synuclein accumulation has been found in the larynx of post-mortem (late-stage) human PD samples, this work has not been yet replicated in the Pink1−/− rat.^17,19 Increases in alpha-synuclein within the neuromuscular junction, nerves and vessels of 6 month old Pink1−/− rats have been reported in the extrinsic tongue muscle (genioglossus) and has been hypothesized to be a small, disease occurring change in the innervation of the tongue muscles.⁶⁰ A key interpretation of the current study is that changes in gene pathways (e.g. kallikrein and/or inflammatory pathways) maybe significant in this early disease model, lead to muscle pathology, and therefore, warrants further investigation.

Another goal of the present study was to use bioinformatics to highlight gene pathways within the TA. WGNCA enrichment analysis of the Lightgreen module (containing the Pink1 gene as the central node) showed 6 significant biological pathways (TNF-α NF-κB Signaling, Parkinson Disease, Proteasome Degradation, Synthesis and Degradation of Ketone Bodies, MAPK Signaling, Alzheimer’s Disease). The top significant pathway, TNF-α NF-κB Signaling, and MAPK Signaling are involved in many important biological processes including cell proliferation and apoptosis and the regulation of immune and pro-inflammatory responses. Nuclear factor-κB (NF-κB) regulates the expression of over 200 genes involved in complex inflammatory pathways;^61,62 and has been implicated in the pathophysiology of PD including alpha-synuclein accumulation.^63–65 Increased levels of downstream NF-κB signaling targets, such as pro-inflammatory factors (IL-1β, IL-6, and TNF-α), have also been found in the nigrostriatal region of postmortem PD brain tissue and CSF.^64,66–68 In addition to NF-κB signaling, evidence has shown that MAPK signaling is triggered in response to oxidative stress and may indirectly induce apoptosis of dopamine neurons via mitochondrial dysfunction.^69,70 Previous work within the Pink1−/− rat has shown that mitophagy is a common phenotype and disrupts cellular homeostasis.⁷¹ One of the up-regulated genes involved in 4 of the 6 significant pathways is Caspase 7 (Casp7), which is involved in the modulation of the cell cycle and plays a central role in apoptosis. These new data suggest that inflammation is widespread (i.e. not only central nervous system) thus, future research should directly target Casp7 and the NF-κB and MAPK pathways as a mechanism of inflammation and mitochondrial dysfunction in the Pink1−/− rat as well as their contribution to muscle cell changes within the vocal fold.

5. Conclusions

Compared to other recently generated gene datasets (i.e. Kelm-Nelson and Gammie 2020, brain tissue), there were significantly fewer differentially expressed genes between Pink1−/− and WT rats in this study. However, WGNCA enrichment databases found significant correlations between this gene dataset and human and cell lines for idiopathic as well as PARK2 and 6 genetic forms of PD. Therefore, this data validates the translatability between human and the Pink1−/− rat model. The additional overlap between PD, Alzheimer’s disease, and multiple sclerosis also suggests that the Pink1−/− animal model may be a relevant model to study common degenerative disease mechanisms. Future studies exploring this gene expression profile in addition to using drug repurposing tools will help identify and validate treatments for the modulation of PD vocal communication and swallowing deficits. In summary, this work is the first to quantify gene expression changes in the TA muscle in the Pink1−/− rat model of early-onset PD.

Supplementary Material

sm3

Supplementary Table 3. Significant gene list. Results of a differential gene expression (protein coding genes) in the TA of male rats. Significant annotated gene list with categories.

sm1

Supplementary Table 1. RSEM output. RNA-Seq data used to analyze differential gene expression using EdgeR. Genotype and rat IDs are listed to the right.

sm2

Supplementary Table 2. EdgeR output. Results of a differential gene expression. Analysis of RNA-Seq data using EdgeR. A negative logFC indicates that gene expression is down regulated in the Pink1−/− rats.

sm4

Supplementary Table 4. WGCNA data. Gene modules identified by WGCNA.