Narcolepsy in Parkinson's disease with insulin resistance

Background: Parkinson’s disease (PD) is characterized by its progression of motor-related symptoms such as tremors, rigidity, slowness of movement, and difficulty with walking and balance. Comorbid conditions in PD individuals include insulin resistance (IR) and narcolepsy-like sleep patterns. The intersecting sleep symptoms of both conditions include excessive daytime sleepiness, hallucinations, insomnia, and falling into REM sleep more quickly than an average person. Understanding of the biological basis and relationship of these comorbid disorders with PD may help with early detection and intervention strategies to improve quality of life. Methods: In this study, an integrative genomics and systems biology approach was used to analyze gene expression patterns associated with PD, IR, and narcolepsy in order to identify genes and pathways that may shed light on how these disorders are interrelated. A correlation analysis with known genes associated with these disorders (LRRK2, HLA-DQB1, and HCRT) was used to query microarray data corresponding to brain regions known to be involved in PD and narcolepsy. This includes the hypothalamus, dorsal thalamus, pons, and subcoeruleus nucleus. Risk factor genes for PD, IR, and narcolepsy were also incorporated into the analysis. Results: The PD and narcolepsy signaling networks are connected through insulin and immune system pathways. Important genes and pathways that link PD, narcolepsy, and IR are CACNA1C, CAMK1D, BHLHE41, HMGB1, and AGE-RAGE. Conclusions: We have identified the genetic signatures that link PD with its comorbid disorders, narcolepsy and insulin resistance, from the convergence and intersection of dopaminergic, insulin, and immune system related signaling pathways. These findings may aid in the design of early intervention strategies and treatment regimes for non-motor symptoms in PD patients as well as individuals with diabetes and narcolepsy.


Invited Reviewers
23 Nov 2020 report report report

Introduction
Parkinson's disease (PD) is characterized by its progression of motor-related symptoms such as tremors, rigidity, slowness of movement, and difficulty with walking and balance [1][2][3] . The motor difficulties associated with PD are attributed to the loss of dopaminergic neurons in the substantia nigra 1,4 . There are also non-dopamine lesions that are involved in PD that include the caudal group of intralaminar nuclei (located in dorsal thalamus), and subcoeruleus nuclei 1,2 . The main genetic cause of PD is attributed to mutation of the LRRK2 (leucine-rich repeat kinase 2) gene 5,6 .
There is also a significantly increased risk of PD among patients with a history of diabetes and PD associated motor symptoms and cognitive decline are accelerated in individuals with diabetes 7,8 ). Results from numerous studies investigating the connection between diabetes and PD indicate that there is an overlap in disease mechanisms and pathways particularly in the context of the accumulation of misfolded proteins 2,9 , defects in mitochondrial function leading to oxidative stress 10,11 , immune system activation resulting in inflammation 12-14 , reduced synaptic plasticity, and a decrease in dopamine levels 15, 16 . Dopamine signaling is also linked to circadian rhythm and sleep 17 . Studies demonstrate that dopamine levels cycle in a circadian manner in the retina, olfactory bulb, striatum, midbrain, and hypothalamus 17-20 . Disrupted sleep patterns and circadian rhythm are also associated with many neuropsychiatric and neurodegenerative illnesses including LRRK2-PD and it is often more disturbing than the motor symptoms 21-23 .
Most PD patients have daytime sleep attacks and REM sleep disorder that resemble narcolepsy associated sleep symptoms such as excessive daytime drowsiness, sleep paralysis, hallucinations 24 , and in some cases episodes of cataplexy 25 . People with narcolepsy frequently enter REM sleep rapidly, within 15 minutes of falling asleep and the muscle weakness or dream activity of REM sleep can occur during wakefulness or be absent during sleep 26 . Alleles of the HLA-DQB1 (major histocompatibility complex, class II, DQ beta 1) gene are associated with a predisposition to narcolepsy 27 , PD 28 , and Type I diabetes 29 .
Besides HLA-DQB1, the relationship between PD, narcolepsy, and IR may be in part attributed to the hypocretins/ orexins which are produced by the HCRT gene 30 . Hypocretins are neurotransmitters that are manufactured by a small number of neurons in the hypothalamus 31 . They act to stimulate target neurons and promote wakefulness while suppressing rapid-eyemovement (REM) sleep 32 . Research has shown that there is a massive loss of hypocretin neurons in patients of both PD and narcolepsy and it is hypothesized that the reduction of hypocretin may be the underlying pathogenesis of the narcoleptic symptoms in PD 24,32,33 . In addition to their role in narcolepsy and PD, hypocretins modulate glucose and insulin metabolism 33 and also play a critical role in dopamine regulation 34 . In this study we explore the connection between PD, narcolepsy, and IR using an integrative genomics and systems biology approach.

Genesets and evaluation
Microarray data was collected from the Allen Brain Database using the Human Brain Atlas. To obtain the data, a gene search for LRRK2, HLA-DQB1, and HCRT was performed. Each of these genes were used to query the atlas for correlates to the hypothalamus, dorsal thalamus, pons, and subcoeruleus nucleus using the dropdown menu for each of the six donor postmortem brains available in the Allen Human Brain Atlas.
Genes whose expression pattern correlated with LRRK2, HLA-DQB1, and HCRT were collected for analysis. Correlates with a range of Pearson r values from 0.6 to 1.0 were considered in the analysis (Extended data, Workbook 1 35 . The rationale was to investigate genes with a similar expression pattern in order to identify gene correlates specific and common to LRRK2, HLA-DQB1, and HCRT. Risk factor genes and genes contributing to PD, narcolepsy, and IR were obtained from OMIM, Harmonizome, and GeneWeaver. Each geneset was evaluated using Gene Ontology (GO) enrichment for clustering, pathways, and keywords using the Database for Annotation, Visualization and Integrated Discovery (DAVID, version 6.8) and the Gene Ontology databases with integrated tools for analysis. Clustering was done in DAVID using the default parameters which include medium stringency settings and a kappa similarity value of 3. The Benjamini corrected P-value was used to determine enrichment significance. The pathway enrichment was performed using KEGG and Panther pathways. The pathways were analyzed manually and evaluated based on shared themes. For the keyword enrichment, a keyword search of the DAVID functional annotation table output was used to identify genes associated with relevant traits related to LRRK2, HLA-DQB1, and HCRT function. The keywords considered were 'sleep', 'circadian', 'parkinson', 'locomotion', 'dopamine', 'behavior', 'learning', 'memory', and 'transcription factor'. Geneset overlap was assessed using Venny 2.0, an online program that compares lists of items to determine the common and unique genes between LRRK2, HLA-DQB1, and HCRT within and among each brain region (hypothalamus, dorsal thalamus, pons, and subcoeruleus nucleus).

Network analysis
The String database (version 11.0) was used to build a protein-protein interaction network (ppi for LRRK2, HLA-DQB1, HCRT and CAMK1D which was identified in this study as the only common risk factor gene associated with PD, narcolepsy

Amendments from Version 2
Version 3 of this article has been edited to correct a duplicated paragraph in the Discussion section and a broken link pointing to Extended Data Table 1. A reference has also been added to support validation of Cacna1c expression.
Any further responses from the reviewers can be found at the end of the article REVISED and IR (Results section: "Functional analysis of PD, narcolepsy and IR risk factor and related genes"). The network was constructed based on experimentally validated interactions using the medium confidence score of 0.4. The combined scores for the interactions are computed by combining the probabilities from the different evidence channels and corrected for the probability of randomly observing an interaction. First and 2nd shell interactions are included in the network. The network was exported from STRING and analyzed in Cytoscape (version 3.7). Network bottlenecks and clusters were identified with Cytoscape plugins CytoHubba (version 0.1) and MCODE (version 1.6.1), respectively. ClueGo (version 2.5.7) was used to analyze the common risk factors and contributing genes for PD, narcolepsy, and IR. The nodes in the network have been manually arranged for proper visibility. Select enriched terms are included in the network ( Figure 3A). All of the enriched terms are provided in Extended data, Workbook 5, sheet 5 36 . Legacy data sets from the Human Protein Atlas (HPA) (https://www.proteinatlas.org/) along with gene expression data from the Allen Brain Atlas were used to assess the expression patterns of key genes identified in the ppi network. To assess gene expression of the key network genes in the relevant brain regions considered in this study, we examined microarray data from the Human Allen Brain Atlas for the dorsal thalamus and subcoeruleus nucleus and the HPA for the hypothalamus and pons. Two data sources were used because the Allen Brain Atlas RNA-Seq data did not provide enough samples to evaluate the gene expression in the hypothalamus and pons for these genes. The RNA-Seq data for the dorsal thalamus and subcoeruleus nucleus are derived from either 5 or 6 postmortem donor brains of varying ages and gender. In most instances the genes are sampled multiple times using different probes. The data are expressed as Z score log2 transformed and indicate the number of standard deviations away from the mean. The data across probes was not combined for statistical analysis because it is qualitative but we calculated the percentage of reads with a positive value which measures the detection of RNA at or above the standards used in the assay. This is not to say that the negative values (Number of standard deviations away from the mean) indicate that RNA is not detected.
The HPA includes RNA-Seq data from humans, mouse and pig. Human data are derived from the Genotype-Tissue Expression (GTEx) (https://www.gtexportal.org/home/) and Functional Annotation of Mammalian Genomes 5 (FANTOM5) projects. Mouse and Pig RNA-Seq data were generated by the Beijing Genomics Institute (https://www.bgi.com/global/ in collaboration with the HPA). Assay conditions are provided in detail at the HPA (https://www.proteinatlas.org/about/assays+annotation #transcriptomics).

Functional analysis of gene correlates
The cluster analysis for the LRRK2, HLA-DQB1, and HCRT gene correlates for each brain region resulted in significant enrichment categories for only the HLA-DQB1 related genesets. For LRRK2 and HCRT there are several instances in which clusters contained enrichment terms for insulin, diabetes, PD, other neurodegenerative disorders, and circadian processes but these did not achieve significance based on the corrected P value criteria. Also, of note for almost every set of correlates, there were many significant enrichment categories and corresponding genes associated with keratinocytes/keratin and olfaction. The clustering results for each set of gene correlates are listed in Extended data, Workbook 2 37 .
For the HLA-DQB1 clusters, the significant enrichment terms are: dorsal thalamus: hsa05012:Parkinson's disease (P=4.38E-06), 31 genes and hsa04940:Type I diabetes mellitus (P=0.03), 10 genes; subcoeruleus nucleus: hsa04940:Type I diabetes mellitus (P=6.13E-05), 15 genes, and pons: hsa04940:Type I diabetes mellitus, six genes (P=0.001). The other genes and enrichment categories clustering with PD in the dorsal thalamus are related to mitochondria processes such as oxidative phosphorylation and electron transport as well as Alzheimer's disease (AD) and Huntington disease (HD).

Geneset overlap
Among the sets of gene correlates for LRRK2, HLA-DQB1 and HCRT, there are 10 common genes in both the dorsal thalamus and subcoeruleus nucleus (A_32_P232747, DISC1, GABRA4, GDF11, HNRNPU, PAK2, , PFKFB2, ROCK1, , SLC9A3R2, and ZNF846, ; Figure 1B). Among the relevant genes are ROCK1, which is involved in negative regulation of neuron apoptotic processes, DISC1 associated with neuron migration, and HNRNPU involved in circadian regulation of gene expression. The ten common genes in the subcoeruleus nucleus are GLYAT, HCN4, HMGB1 HNRNPU, ITGB2,LAMP2, LOC653110, OPA3, ,PHTF2, and SLC6A6, , among which the relevant ones to this study are HNRNPU, which as mentioned above is involved in circadian regulation of gene expression and insulin signaling, ITGB2, which is associated with PD and IR, and HMGB1, which is a ligand for the RAGE receptor. Among the sets of gene correlates for LRRK2, HLA-DQB1, and HCRT, there are no common correlated genes in the hypothalamus and pons.
A detailed description of all shared genes and their associated function for each brain region is provided in Extended data, Workbook 3 (sheets 1-8) 38 . Briefly, the dorsal thalamus and subcoeruleus nucleus have the largest number of shared correlates between LRRK2, HCRT, and HLA-DQB1. Many of these genes for both brain regions are associated with neuron, insulin, and dopamine related processes. There are also several genes connected directly to PD. In sharp contrast, however, the dorsal thalamus associated correlates have many genes linked to circadian function.
In the dorsal thalamus, the relevant genes are associated with neuron function (negative regulation of neuron apoptotic process, neuron projection development and regulation, neuron differentiation, neuron fate commitment, neuron death in response to oxidative stress, neuron regeneration), circadian processes (regulation of circadian rhythm, circadian entrainment, circadian regulation of gene expression, regulation of circadian rhythm, entrainment of circadian clock by photoperiod), and insulin signaling (insulin secretion, insulin receptor signaling, insulin secretion). Other genes of interest are related to dopamine (dopaminergic neuron differentiation, regulation of dopamine uptake involved in synaptic transmission, positive regulation of dopamine secretion, Wnt signaling pathway involved in midbrain dopaminergic neuron differentiation, dopamine receptor binding, dopamine biosynthetic process, adenylate cyclase-activating dopamine receptor signaling pathway) and also behavior (locomotory, feeding, learning, memory and vocalization, response to stimulants).
In the subcoeruleus nucleus, the relevant genes are also associated with neuron function (negative regulation of neuron differentiation, dopaminergic neuron differentiation, neuron apoptotic process, negative regulation of neuron differentiation, forebrain neuron differentiation), insulin signaling (insulin secretion, insulin receptor signaling pathway, negative regulation of insulin receptor signaling pathway, positive regulation of insulin secretion, diabetes mellitus), dopamine related processes (dopaminergic synapse, dopamine biosynthetic process, dopaminergic neuron differentiation, regulation of synaptic transmission, dopaminergic dopamine biosynthetic process from tyrosine), and behavior (locomotory, vocal learning, response to fear, grooming, response to stimulants).
There are few shared correlated genes in the hypothalamus and pons. For the hypothalamus, the most pertinent genes are involved in neuron migration and circadian processes. In the pons, the relevant genes are concerned with negative regulation of neuron apoptotic processes, neuron projection, circadian regulation of gene expression, and hippocampus and pyramidal neuron development.
Geneset overlap of the correlates for LRRK2, HLA-DQB1, and HCRT was assessed within each brain region (see Extended data, Workbook 3, sheets 9-11) 38 . Of the LRRK2 correlates,1.6% were common in all 4 brain regions. Among these are genes associated with insulin (MAX, NUCKS1, PIK3R1, PTPN11), diabetes (PIK3R1) and circadian-related processes (HNRNPU, BHLHE41). Several transcription factors were also present (MAX, SKI, ATF7IP, NUCKS1, BHLHE41, NR2C2, PIK3R1). One of these, BHLHE41, acts as a negative regulator of orexin, controls circadian rhythms, and is associated with short sleep syndrome and advanced sleep phase disorder. For HLA-DQB1, 1.2% of the correlates are common in all brain regions; relevant associated themes include PD and dopamine (SLC18A1) insulin (HLA-DRB5, HLA-DOA, HLA-DQA1, HLA-DQB1) and transcription factors (FOXE3, HMGB1, LGALS9, PYCARD, SOX8, ZNF446). Only 0.2% of the HCRT correlates are common among all brain regions considered. This includes HCRT itself and an insulin associated gene, GHSR. Of note, the MOG gene, which is present among the HLA-DQB1 correlates of the subcoeruleus nucleus and HCRT correlates of the hypothalamus, is a risk factor for narcolepsy and is linked to PD.  Functional analysis of PD, narcolepsy, and IR risk factor and related genes PD, narcolepsy, and IR risk factor and related genesets were evaluated to identify a common set of genes associated with the three disorders (Extended data, Workbook 5, sheets 1-4) 36 . There were 38 shared genes between the PD and narcolepsy genesets. CAMK1D is the only gene common among the 3 genesets for PD, narcolepsy, and IR and it is a Calcium/ Calmodulin kinase that is upregulated in PD patients and is also a risk factor for Type 2 diabetes 40,41 .
Of the common PD and narcolepsy genes, several were directly associated with PD and narcolepsy behavioral phenotypes such as locomotion ( There were two common genes between the PD and IR genesets: RREB1, which is a transcription factor, and ANKFY1, which is involved in vesicle trafficking and is also implicated in Type 2 diabetes. There is one common gene between narcolepsy and IR: HLA-DQB1, which is the narcolepsy associated gene under study here. The enrichment results are visualized as a network of functionally grouped terms and pathways and listed in the accompanying bar graph ( Figure 3A, B, Extended data, Workbook 5, sheet 5) 36 . The most significant term of a given group is highlighted as the leading term in the network which is indicated by color. The most significant terms emphasized in the graph are dopaminergic synapse (ten genes, KEGG ID:04728, P=2. The PD, narcolepsy, and IR connection A protein-protein interaction network revealed the insulin connection between the LRRK2 and HLA-DQB1 networks using the multiple protein option in the STRING database ( Figure 4A, Extended data, Workbook 6, sheets 1-2) 42 . The distribution for the PPI scores for each show that the majority of the interactions fall in the high range with scores between 0.7 and 1.0 ( Figure 4B). Insulin (INS) and its receptor (INSR) are connected to HLA-DBQ1 through 1st shell interactions both of which are based on crystallographic evidence. INSR is in turn connected to CALM1, a calmodulin binding protein involved in calcium signaling and associated with diverse processes including circadian entrainment (KEGG pathway 04713). The evidence for the INSR/CALM1 interaction is based on coimmunoprecipitation, electro mobility shift, and western blot assays. Relevant interactions, scores, and references are provided in Table 1.
In the network, CALM1 bridges INSR, CAMK1D (the only common gene among the database curated genesets for PD, narcolepsy, and IR), and LRRK2. The CALM1 and CAMK1D relationship is supported by coimmunoprecipitation and filter binding and phage display assays. The CALM1/LRRK2 interaction is supported by cosedimentation, coimmunoprecipitation and genetic interference assays.
There are many proteins in the network related to insulin signaling (CACNA1C, CALM1, CALM2, CALM3, IDE, IGF1, IGF1R,  INSRR, INS-IGF2, KCNN2  Results for gene expression of the key genes in the hypothalamus, and pons, are as follows: Each of the genes with the exception of AGER were expressed in both the hypothalamus and pons for human, mouse and pig. CACNA1C expression was also confirmed at the protein level by immunohistochemistry in the mouse hypothalamus and pons/medulla (https://www.proteinatlas. org/ENSG00000151067-CACNA1C/antibody). Expression data are summarized in Extended data, Table 1. The expression pattern of the key genes in dorsal thalamus and subcoeruleus nucleus data were obtained for humans only and are as follows: HMGB1, AGER, BHLHE41 are expressed in both the dorsal thalamus and subcoeruleus nucleus. CACNA1C is expressed in the dorsal thalamus but is not expressed in the subcoeruleus nucleus. CAMK1D was not expressed in either the dorsal thalamus or subcoeruleus nucleus. The results from the gene expression data are summarized in Table 2, Data for the dorsal thalamus and subcoeruleus nucleus are provided in Extended data, workbook 7 43 .

Discussion
The aim of this study is to identify the underlying genes and pathways linking PD, narcolepsy, and IR. An integrative genomics and systems biology approach was used for the analysis of gene expression patterns of the LRRK2, HLA-DQB1, and HCRT genes which are strongly associated with each of these disorders. A comparison of the shared gene correlates for sleep, neurodegeneration, behavior, and insulin led to the identification of genes such as AGER, BHLHE41, CACNA1C, CAMK1D, and HMGB1, whose defects might be plausible for the narcoleptic-like symptoms in PD and the relationship with IR.
The ppi network of LRRK2, HLA-DQB1, and CAMK1D reveals a connection with several insulin/diabetes, circadian, and PD risk factor genes supporting our hypothesis that these three disorders have common pathogenetic processes and further supports earlier studies that have reported a relationship between these conditions. There is a great deal of evidence derived from knockout and cell based studies linking AGER, CACNA1C and HMGB1 to Parkinson's pathogenesis. CACNA1C is also associated with circadian rhythm and narcolepsy.
There is also evidence of disrupted calcium homeostasis in PD 44 . Genetic variants of CACNA1C which is a subunit of Cav1.2 Ca2+ channels. are linked to greater PD risk which is dependent on vitamin D deficiency 45 . Microglia in an induced PD model exhibited enhanced neuroinflammation and inhibited neuroprotection in the presence of a Ca2+ agonist 46 .
Degeneration of dopaminergic neurons have also been observed in microglia-specific Cav1.2 knockdown mice intoxicated with MPTP, a neurotoxin that induces PD-like symptoms 46 . CACNA1C was also expressed 3x higher in microglia treated  with agents to stimulate neuro inflammation 46 . CAMK1D, also associated with Ca2+ signaling, exhibited decreased expression levels in iPSC-derived neurons carrying the LRRK2 G2019S mutation, the most prevalent genetic cause of late onset PD 47 .
HMGB1 and AGER appear to act in concert mediating inflammatory processes that ultimately lead to neuron cell death via NF-κB signaling 48 . Studies indicate that HMGB1 is associated with autophagy dysfunction and the degeneration of dopaminergic neurons through interaction with α-synuclein thereby intensifying protein aggregation and in conjunction with RAGE, inflammation and cell death 49,50 . RAGE initiates signal transduction cascades and activates NF-κB, increases cytokine expression and also leads to the production of reactive oxygen species 51,52 . In neuronal cells expressing the G2019S LRRK2 mutant, RAGE enhanced cell death. Expression of RAGE proteins were also upregulated in the LRRK2 mutant cells 53 . RAGE is highly expressed in PD patients when compared with age-matched controls 51 and RAGE gene variants have been linked to sporadic PD in an Asian population 54 Silencing of the RAGE pathway in a mouse model of PD improved neuroinflammation which causes dopaminergic neurodegeneration in PD patients 52 . This is important because the deterioration of dopaminergic neurons in the brain is believed to play a critical role in the development of PD 60 . By the time clinical signs of PD are identified and a diagnosis is made, a large number of dopaminergic neurons have already been lost 1 . Dopaminergic neurons are also involved in promoting feeding behavior in the hypoglycemic state which is mediated by insulin receptors in the substantia nigra, indicating that dopaminergic neuronal loss may alter glycemic control 61,62 . Loss of orexin/hypocretin is also linked to binge-eating behavior, low BMR, and obesity, which is a symptom of narcolepsy 63-65 .
BHLHE41, the other gene of interest identified from the ppi network analysis, is a transcription factor associated with circadian processes. Variants of BHLHE41 have been implicated in short sleep syndrome 66 . BHLHE41 also has a role in immune function and in addition to CAMK1D, AGER is implicated in diabetes along with the narcolepsy-related gene, HLA-DQB1.
Further insight into the relationship between PD, IR and disrupted sleep patterns is evident from studies in which the repurposing of treatment for one of these diseases has been used to alleviate symptoms in another. Results from a recent clinical trial in which PD patients were treated with intranasal insulin, reported that test subjects had improved verbal fluency and motor skills and sleep related symptoms 67 . Insulin is also promising for treating AD symptoms along with growth factors and incretins (orexin) which are a current therapy for T2D 68 . Metformin another anti diabetic drug has been used to treat AD and may have promise for PD as well 69 . Current trends in Biomarkers for disease detection include neuroimaging techniques such as petscan monitoring to glucose uptake in PD patients and also monitoring of oxidative stress and cholesterol metabolism 70,71 . Melatonin, the naturally occurring hormone that controls sleep and wake cycles, was also found to be beneficial in PD 72 .
There are more than 10 million people worldwide that live with Parkinson's disease. Additional studies aimed at identifying genes and regulatory factors underlying and bridging these comorbid disorders may aid in the design of early intervention and diagnosis strategies, as well as treatment regimes for patients with PD, diabetes, and/or narcolepsy.

Conclusion
We have identified genetic signatures that link PD with its comorbid disorders, narcolepsy and insulin resistance, from the convergence and intersection of dopaminergic, insulin, and immune system related signaling pathways. The resulting genes and pathways identified here are consistent with many published findings and may aid in the design of early intervention strategies and treatment regimes for non-motor symptoms in PD patients as well as individuals with diabetes and narcolepsy.

Data availability
Source data All data underlying the results are available as part of the article and no additional source data are required.

Underlying data
All data underlying the results are available as part of the article and no additional source data are required.

Open Peer Review expertise to confirm that it is of an acceptable scientific standard.
Neurobiology Section, Division of Biological Sciences, University of California San Diego, La Jolla, CA, USA Parkinson's disease is the second most common neurodegenerative disease right after Alzheimer's disease nowadays. There is no treatment to counteract the symptoms of the disease, and most of the medications used are aimed at alleviating the symptoms. However, the use of these drugs also has disadvantages in the form of side effects, which makes the quality of life of the patients drop drastically. Lately, research on Parkinson's disease has been focused on the detection of markers that allow diagnosis of this disorder in early stages to increase the probabilities of treatments being effective. Within these types of studies, we have this article published by Chunduri et al., in which the main objective is the search for the expression of certain genes shared between Parkinson's disease, insulin resistance and narcolepsy. The authors propose to analyze the molecular bases of the comorbidity of these three disorders in order to find risk factors that could predispose and facilitate the early diagnosis of Parkinson's disease. As a result of the study, the authors have identified several genes that are related to the signaling pathways of dopamine, insulin and the immune system, among which are: CACNA1C, CAMK1D, BHLHE41, HMGB1, and the AGE-RAGE axis.
The preliminary data obtained in this study is interesting and would allow further research on the early diagnosis of this disorder, through the individualized study of each of these candidate genes, probably through knock-out or knock-down strategies in animal models. However, some parts of the article need to be supplemented with additional data, and furthermore, the discussion of the data obtained seems short and incomplete, so the authors should increase the number of bibliographic references that are currently discussing the analysis of comorbidity in Parkinson's disease, both in animal models and in humans.
Specifically, I would like the authors to address the following points: It would be appropriate to validate the expression of some of these candidate genes in the mentioned brain regions, in the tissue of healthy human or wild-type animals (e.g. mouse, rat, monkey), mainly by quantitative PCR or in situ hybridization, or alternatively by Western Blot or immunohistochemistry. It would be very interesting if the authors could also analyze the expression of these genes in tissue from patients or animal models of Parkinson's disease.

1.
It would be quite useful to validate the data obtained in humans, with the gene expression database available for mouse or monkey brain tissue, on the Allen Brain Institute website. Mouse Brain Atlas: https://developingmouse.brain-map.org/search/index Non-human primate Brain Atlas: https://www.blueprintnhpatlas.org/microarray/search 2.
It would be great if the authors could comment on why they decided to analyze specifically the comorbidity of Parkinson's disease with insulin resistance and narcolepsy. There are other disorders that are more frequent comorbidities in Parkinson's disease patients, such as hypertension, Crohn's disease, schizophrenia, restless leg syndrome, anemia (see attached references). 1,2,3 3.
Could the authors comment and discuss if some of the candidate genes found are expressed in dopaminergic neurons in the human or mouse brain (regarding the bibliography available in the field)? 4.
Once the authors are able to address all the issues I have proposed in this report, I would be willing to re-evaluate this manuscript.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Yes Are the conclusions drawn adequately supported by the results? Partly been focused on the detection of markers that allow diagnosis of this disorder in early stages to increase the probabilities of treatments being effective. Within these types of studies, we have this article published by Chunduri et al., in which the main objective is the search for the expression of certain genes shared between Parkinson's disease, insulin resistance and narcolepsy. The authors propose to analyze the molecular bases of the comorbidity of these three disorders in order to find risk factors that could predispose and facilitate the early diagnosis of Parkinson's disease. As a result of the study, the authors have identified several genes that are related to the signaling pathways of dopamine, insulin and the immune system, among which are: CACNA1C, CAMK1D, BHLHE41, HMGB1, and the AGE-RAGE axis.
The preliminary data obtained in this study is interesting and would allow further research on the early diagnosis of this disorder, through the individualized study of each of these candidate genes, probably through knock-out or knock-down strategies in animal models. However, some parts of the article need to be supplemented with additional data, and furthermore, the discussion of the data obtained seems short and incomplete, so the authors should increase the number of bibliographic references that are currently discussing the analysis of comorbidity in Parkinson's disease, both in animal models and in humans.

Reviewer 3 comment:
Specifically, I would like the authors to address the following points: It would be appropriate to validate the expression of some of these candidate genes in the mentioned brain regions, in the tissue of healthy human or wild-type animals (e.g. mouse, rat, monkey), mainly by quantitative PCR or in situ hybridization, or alternatively by Western Blot or immunohistochemistry. It would be very interesting if the authors could also analyze the expression of these genes in tissue from patients or animal models of Parkinson's disease.

Response:
We have addressed the request for validation of candidate gene expression in the mentioned brain regions above in our response to Reviewer 1 and have added the information to the Results and Discussion sections of the revised manuscript. To summarize our findings, each of the key genes is expressed in at least two of the brain regions relevant to this study.
To address the reviewer's question concerning expression of these genes in tissue from patients or animal models of Parkinson's disease, we have analyzed the gene expression data available in the NCBI GEO database specifically for Parkinson's Disease datasets to determine if any of the key genes identified in our study are differentially expressed.
Out of 24 available datasets, we found that BHLHE41 and AGER were differentially expressed as compared with controls from two distinct studies ( https://www.ncbi.nlm.nih.gov/geo/geo2r/?acc=GSE36321 and https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE19587). For BHLHE41, the study involved assessing gene expression in human neural stem cells containing the LRRK2 (G2019S) pathogenic mutation. The LRRK2 (G2019S) containing cells had higher expression of BHLHE4. AGER was differentially expressed with borderline significance in a study that examined post mortem medullary regions from brains with evidence of Parkinson's. AGER was expressed higher in brains from individuals with Parkinson's. However, due to low n values we do not have enough confidence in the data to include these results in the manuscript. The other genes of interest, CACNA1C, CAMK1D, and HMGB1 were not identified in any of the GEO gene expression datasets as being expressed differently than controls.

Reviewer 3 comment:
It would be quite useful to validate the data obtained in humans, with the gene expression database available for mouse or monkey brain tissue, on the Allen Brain Institute website. Mouse Brain Atlas: https://developingmouse.brain-map.org/search/index Non-human primate Brain Atlas: https://www.blueprintnhpatlas.org/microarray/search

Response:
Thank you for the suggestion. However, to the best of our knowledge, there is no Microarray data available for mouse at the Allen Brain Atlas but we were able to consider the in situ hybridization data (discussed above). Regrettably the Non-human primate atlas does not contain data for the relevant brain regions used in this study which is unfortunate because this is an intriguing dataset.

Reviewer 3 comment:
It would be great if the authors could comment on why they decided to analyze specifically the comorbidity of Parkinson's disease with insulin resistance and narcolepsy. There are other disorders that are more frequent comorbidities in Parkinson's disease patients, such as hypertension, Crohn's disease, schizophrenia, restless leg syndrome, anemia (see attached references).1,2,3

Response:
Initially, the focus of our study was to investigate the connection between Parkinson's and narcolepsy. Early on in the data collection and preliminary analysis, we observed that many of the gene correlates were linked to insulin resistance. On the basis of this finding we decided to expand the scope of the study.

Qing Wang
Department of Neurology, Zhujiang Hospital, Southern Medical University, Guangzhou, Guangdong, China This study has identified the genes such as CACNA1C, CAMK1D, BHLHE41, HMGB1, and AGE-RAGE that link PD with narcolepsy and insulin resistance and their signaling networks are connected through insulin and immune system pathways. These findings may contribute to develop new treatment strategies in PD patients as well as individuals with diabetes and narcolepsy. The results of the study are interesting.
However, the introduction and discussion are too weak. So it is suggested to cite more relevant references as follows (please see the reference list below) in order to elaborate on the mechanisms.
After my concerns above have been fully addressed, I am happy to re-evaluate this manuscript.
such previous papers? Otherwise, the result of network analysis alone seems preliminary.
In the same context, other experiments, such as immunohistochemistry, should be conducted to confirm the expression of these genes at the protein level.

2.
Is the work clearly and accurately presented and does it cite the current literature? Yes Is the study design appropriate and is the work technically sound? Yes

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? Partly
In this study, an integrative genomics and systems biology approach was used to analyze gene expression patterns associated with PD, IR, and narcolepsy to identify genes and pathways that may shed light on how these disorders are interrelated. The results showed that the PD and narcolepsy signaling networks are connected through insulin and immune system pathways. Important genes and pathways that link PD, narcolepsy, and IR were shown to be CACNA1C, CAMK1D, BHLHE41, HMGB1, and AGE-RAGE. The authors concluded that these findings might aid in the design of early intervention strategies and treatment regimes for non-motor symptoms in PD patients as well as individuals with diabetes and narcolepsy.
Overall, the results are interesting and clear. I agree that the paper may contribute to the PD research. My comments are as follows: Reviewer 1 comment: It would be much better if modification of the expression of the identified genes (knockout or overexpression), including CACNA1C, CAMK1D, BHLHE41, HMGB1, and AGE-RAGE, might result in phenotypes related to PD and/or narcolepsy in cells or animal models. Are there such previous papers? Otherwise, the result of network analysis alone seems Preliminary.

Response:
We thank the reviewer for taking the time to provide us constructive feedback on this manuscript. There is in fact a great deal of evidence derived from knockout and cell based studies linking CACNA1C, HMGB1, and AGER/AGE-RAGE to Parkinson's pathogenesis. CACNA1C is also associated with Circadian rhythm and narcolepsy. In addition, CAMK1D is a component of the calmodulin dependent calcium signaling cascade and there is also support for its role as an interactor of LRRK2, one of the key genes influencing Parkinson's. HMGB1 and AGER/AGE-RAGE appear to act in concert mediating inflammatory processes that ultimately lead to neuron cell death via NF-κB signaling. AGER/AGE-RAGE and LRRK2 are also linked as RAGE proteins are upregulated in LRRK2 G2019S mutant cells. The LRRK2 G2019S mutation is the most common genetic cause of neurodegeneration and PD. The other major gene of interest identified in the ppi network, BHLHE41, is a transcription factor associated with Circadian processes. Variants of BHLHE41 have been implicated in short sleep syndrome and this gene also has a role in immune function. BHLHE41 in addition to CAMK1D, RAGE, and HMGB1 are also implicated in diabetes.
Details of the supporting studies for the candidate genes as well as the other network genes have been included in the Introduction and Discussion sections of the revised manuscript.

Reviewer 1 comment:
In the same context, other experiments, such as immunohistochemistry, should be conducted to confirm the expression of these genes at the protein level.

Response:
We agree that the above mentioned studies could be useful to further support the results presented here. Regrettably we do not have access to wet lab facilities. Our method of research is based in bioinformatics, systems biology and integrative genomics.
However, we were able to address your question using legacy datasets from the Human Protein Atlas (https://www.proteinatlas.org/) which also contains data for gene expression in mice and pigs and the human database in the Allen Brain Atlas ( https://human.brain-map.org/microarray/search).
Results for gene expression of the key genes in the hypothalamus and pons are as follows: Each of the genes with the exception of AGER were expressed in both the hypothalamus and pons for human, mouse and pig. CACNA1C expression was also confirmed at the protein level by immunocytochemistry in the mouse hypothalamus and pons/medulla.
The expression pattern of the key genes in dorsal thalamus and subcoeruleus nucleus data were obtained for humans only and are as follows: HMGB1, AGER, BHLHE41 are expressed in both the dorsal thalamus and subcoeruleus nucleus. CACNA1C is expressed in the dorsal thalamus and is not expressed in the subcoeruleus nucleus CAMK1D was not expressed in either the dorsal thalamus or subcoeruleus nucleus. These data are have been added to the revised version of the manuscript.