Insights into Neurodegenerative Diseases using Transcriptomics Datasets
A re-analysis of the similarities and differences between Alzheimer’s disease and Parkinson’s disease.
Published on June 27th, 2023
By Axel Martinelli
What are Neurogenerative Diseases?
Neurodegenerative diseases are a group of disorders characterised by the progressive loss of structure or function of neurons (nerve cells) in the brain or peripheral nervous system. These include Alzheimer’s disease, characterised by the accumulation of abnormal protein deposits and Parkinson’s disease, caused by the degeneration of dopamine-producing neurons in the substantia nigra. No therapies are currently available for either condition.
As the global population ages, cases of neurodegenerative diseases are steadily increasing, placing a huge burden on the healthcare system as well as intense emotional toil on patients and their families.
Understanding the molecular basis of Neurodegenerative diseases is thus of vital importance to identify suitable biomarkers for an early diagnosis and to develop effective therapies. Gene expression data from patients form an important component of this process.
In this post, I have selected two transcriptomic datasets from the GEO database to analyse the similarities and differences between Alzheimer’s disease and Parkinson’s disease and to show how large dataset collections can be accessed to provide further insights of clinical relevance. The first dataset consists of the recent transcriptomic profiling of sporadic Alzheimer’s disease patients (Caldwell et al, 2022), while the second dataset is from an older study on the transcriptomics and proteomics profiles of Parkinson’s disease patients (Dumitriu et al, 2016).
The corresponding datasets were injected into the Omics Playground platform, which was run by intersecting the default choices of methods for differential gene expression analysis (limma, EdgeR and DEseq2), with cutoffs set at p<0.05 and logFC>0.5. For Gene Set Enrichment Analysis (GSEA), the camera, fgsea and gsva methods were intersected with cut-offs set at p<0.05 and logFC>0.2.
Alzheimer's disease and Parkinson’s disease downregulate similar genesets but have distinct upregulation profiles
In order to assess the reliability of the two datasets, I first compared the gene expression patterns of diseased vs. healthy tissue with the comparative profiles from more than 6000 public datasets. In both cases, the strongest correlation was consistent with the nature of the disease (Fig. 1), with the Alzheimer’s disease dataset pairwise comparison strongly correlated with the comparison of Alzheimer’s disease vs healthy tissue in study GSE95587 (Fig. 1A) and the Parkinson’s disease pairwise comparison strongly correlated with the comparison of Parkinson’s disease vs healthy tissue in study GSE68719 (Fig. 1B).
Figure 1. Identification of the most closely matched contrasts in public datasets. (A) Closest match for the Alzheimer’s disease study contrast; (B) closest match for the Parkinson’s disease study contrast.
For the clustering analysis, I resorted to the “Geneset UMAP” feature of the platform, which identifies differences in expression across samples for more than 50,000 pathways and genesets. For both datasets, the large-scale up- and down-regulation of whole genesets can be observed when comparing disease and healthy brain tissue samples (Fig. 2A-B)
Figure 2. (A) Geneset UMAP plots for the Alzheimer’s disease dataset. (B) Geneset UMAP plots for the Parkinson’s disease dataset.
GSEA reveals the strong downregulation of genesets associated with neuronal activity and brain function in both diseases, with four genesets (highlighted in orange) appearing among the top 15 most down-regulated genesets for both diseases (Fig. 3A-B). In particular, the strong correlation with genes downregulated during Alzheimer’s disease (DISEASE:Alzheimer’s disease DOID-10652 human GSE36980 sample 520 (down)) in both datasets can be noted.
Figure 3. Genesets and pathways found to be downregulated in (A) Alzheimer’s disease and (B) Parkinson’s disease brain samples compared to healthy brain tissue based on the intersection of three GSEA algorithms.
Conversely, the top 15 upregulated gensets provide a different picture for each condition. In Alzheimer’s disease, there is an increase in the expression of genes associated with vascular endothelial growth and angiogenesis (highlighted in blue, Fig. 4 A), which have been implicated in the pathogenesis of the condition (Jefferies et al, 2013). Meanwhile, the upregulation of inflammatory response was observed in Parkinson’s disease samples, as indicated by the positive correlation with S. aureus infection and prostaglandin J2 exposure expression profiles (highlighted in green, Fig. 4 B). The correlation with the upregulation of genes following Prostaglandin J2 treatment is particularly intriguing, as this prostaglandin has been proposed as a target for intervention against inflammation-induced neurodegeneration (Figueiredo-Pereira et al, 2016).
Figure 4. Genesets and pathways found to be upregulated in (A) Alzheimer’s disease and (B) Parkinson’s disease brain samples compared to healthy brain tissue based on the intersection of three GSEA algorithms.
Comparative Analysis of Alzheimer's disease and Parkinson’s disease datasets
Due to the similarities and differences between Alzheimer’s disease and Parkinson’s disease expression profiles, I accessed the “Compare datasets” feature of the platform to take a closer look. A correlation scatterplot between the gene expression profiles obtained when comparing diseases vs healthy tissue samples from both experiments indicate, as expected, a strong positive correlation (Fig. 5). Among the most correlated genes are several heat shock proteins, known to play a role in both diseases (Aridon et al, 2011; Campanella et al, 2018), as well as gene FOXJ1. FOXJ1 is a transcription factor involved in the maturation of ependymal cilia, which provide support and protection to the neurons in the central nervous system and which have been linked to Neurodegenerative diseases (Nelles and Hazrati, 2022).
Figure 5. Comparative analysis of the Alzheimer’s disease vs healthy brain samples and Parkinson’s disease vs healthy brain samples contrasts with the most strongly correlated genes highlighted, The analysis indicates a high positive correlation (r=0.631).
However, what is also interesting to analyse when comparing the two datasets are the differences. I could identify two groups of genes that showed opposed expression profiles in the two diseases (Fig. 6):
Genes significantly upregulated in Parkinson’s disease and downregulated in Alzheimer’s disease
Among the genes that were upregulated in Parkinson’s disease and downregulated in Alzheimer’s disease, I observed four inflammation-related genes (IL1B, IL1RL1, CD74,FCGR3A (CD16)). Since both Parkinson’s disease and Alzheimer’s disease are associated with inflammation, this suggests that the two diseases differ in the extent and type of inflammatory responses present. Furthermore, expression of the gene HAMP (Hepcidin antimicrobial peptide) is regulated by IL-6 and has been associated with anaemia in mice (Roy et al, 2007). It is intriguing to notice that there is actually a positive correlation between Parkinson’s disease and anaemia (Wang et al, 2021).
Gene SLC5A11, which is involved in the transport of neurotransmitters and is associated with Benign Familial Infantile Epilepsy (Roll et al, 2002), was also upregulated in Parkinson’s disease but downregulated in Alzheimer’s disease. SLC transporters like SLC5A11 have long been implicated in Neurodegenerative diseases (Ayka and Şehirli, 2020), so it is interesting to observe a divergent expression profile between Alzheimer’s disease and Parkinson’s disease.
Gene HAUS7 encodes a subunit of the augmin complex, which plays a central role in microtubule organisation in post-mitotic neurons. Its disruption impairs neurite formation (Sánchez-Huertas et al, 2016), which the data suggests may be also taking place in Alzheimer’s disease, but not Parkinson’s disease patients.
Genes significantly upregulated in Alzheimer’s disease and downregulated in Parkinson’s disease
Four genes stood out in this group. The most divergent was STUM, a gene coding for a mechanosensory transduction mediator homolog with a high level of expression in the brain. Mechanosensory transduction mediators are involved in converting mechanical stimuli into electrical signals and its downregulation may be correlated with the motor impairment observed in Parkinson’s disease patients (Moustafa et al, 2016).
SEMA3G (Semaphorin-3G) serves as a synaptic organiser that regulates synaptic plasticity and hippocampal-dependent memory (Tan et al, 2019). It has been shown that semaphorin-guided axonal growth can lead to recovery from experimental Parkinson’s disease (Díaz-Martínez et al, 2013), which points to the downregulation of SEMAG3 playing a role in the pathology of Parkinson’s disease.
EDN3 encodes a protein called endothelin-3, which is essential for development of neural crest-derived cell lineages and has been associated with congenital disorders involving neural crest-derived cells (McCallion and Chakravarti, 2001; Sánchez-Mejías et al, 2010). It is also thought to regulate neuronal cell development and proliferation (Perrine, 2007). Endothelin-1, an orthologue of EDN3, contributes to loss of endothelial functional integrity and cerebrovascular inflammation that are characteristic of Alzheimer’s disease (D’Orléans-Juste et al, 2018), suggesting a similar role for EDN3.
AMIGO2 encodes a member of a protein family involved in neuronal formation (Kuja-Panula et al, 2003) and promoting neuronal survival (Laeremans et al, 2013) that has been previously observed to be upregulated in Alzheimer’s disease (Kong et al, 2009), consistent with the results of the analysed Alzheimer’s disease dataset. While no evidence for a role in Parkinson’s disease has been exposed, its function and downregulation in the analysed dataset marks it as a potential research target.
Figure 6. Comparative analysis of the Alzheimer’s disease vs healthy brain samples and Parkinson’s disease vs healthy brain samples with genes showing the most negatively correlated expression profiles in the two diseases highlighted.
Identifying therapeutic targets through the drug connectivity map (Drug cMAP) database
While understanding the basic biology behind Neurodegenerative diseases and identifying potential biomarkers and novel therapeutic targets is important, the ability of repurposing existing drugs to provide rapid, tested and cheap therapies, or at least provide a concrete basis from which to develop such approaches, can have a more direct and significant impact on the well-being of patients. Using the Drug expression profiles collected in the Omics Playground platform, I thus attempted to identify compounds with potential therapeutic benefits for both diseases.
Using the drug connectivity map (Drug CMAP) database, the platform highlighted several protein synthesis inhibitors among the 10 most inhibitory drugs for the treatment of both Alzheimer’s disease and Parkinson’s disease (highlighted in purple in Fig. 7A-B). Misfolded proteins build up and overpower the usual processes of balance and quality control in neurodegenerative diseases. A potential approach to enhance protein quality control involves decreasing protein translation. This approach has been shown to protect neurons in a mouse model of Parkinson’s disease (Dastidar et al, 2020), indicating that there may be some therapeutic potential among this category of drugs. Indeed, a study from 2021 showed that homoharringtonine (shown as a potential inhibitor of both Alzheimer’s disease and Parkinson’s disease by the analysis) could inhibit the progression of Alzheimer’s disease (Jiang et al, 2021).
Colforsin, an adenylyl cyclase activator, also appeared on both lists (highlighted in green, Fig. 7A-B). It is thus interesting to note that adenylyl cyclase expression is decreased in the Hippocampus of Alzheimer’s disease patients (Yamamoto et al, 2000) and that a neuropeptide promoting adenylyl cyclase activity protects against cognitive decline in Neurodegenerative diseases (Solés-Tarrés et al, 2020).
Two other drugs that were identified against both diseases were panobinostat, a histone deacetylase (HDAC) inhibitor and enalapril, an angiotensin converting enzyme(ACE) inhibitor (highlighted respectively in blue and orange in Fig. 7A-B). HDACs have long been proposed as targets for Neurodegenerative diseases therapy, with inhibitors displaying neuroprotective properties (Xu et al, 2011; Shujla and Tevkani, 2020). The case for a therapeutic role for ACE inhibitors is more complex. Thus, while some studies indicate that ACE proteins levels appear lower in individuals suffering from Alzheimer’s disease because of their role in degrading amyloid-β (Kehoe et al, 2016; Guzel et al, 2022), other studies indicate an upregulation of ACE expression in connection with Alzheimer’s disease (Cuddy et al, 2020; Ding et al, 2021; Manon et al, 2023) and a beneficial effect of ACE inhibitors in animal models of Alzheimer’s disease (Lee et al, 2020; Cuddy et al, 2020).
Figure 7. Identification of drugs with inhibitory potential against Alzheimer’s disease and Parkinson’s disease based on the drug cMAP database. (A) Top 10 most inhibitory drugs against Alzheimer’s disease. (B) Top 10 most inhibitory drugs against Parkinson’s disease.
Conclusion
In this brief meta-analysis of two published datasets on Alzheimer’s disease and Parkinson’s disease, I could rapidly highlight similarities and differences between the two Neurodegenerative diseases thanks to the visualisation capabilities of the Omics Playground platform that include direct comparative analysis between two datasets. The analysis indicated a distinct inflammatory component in Parkinson’s disease and different downregulated genes linked to neuronal development in both diseases.
Furthermore, intuitive access to the drug cMAP database allowed me to pinpoint drugs with potential therapeutic properties against both diseases, including a compound, homoharringtonine, recently shown to halt progression of Alzheimer’s disease.
About Omics Playground
Omics Playground is a cloud-based bioinformatics software designed for bioinformaticians and biologists to interactively visualize and analyze RNA-seq and proteomics data from their experiments. You can easily create a basic account on Omics Playground to analyze your own data or contact us to book a live demo.
Don’t miss out on our scientific insights – subscribe to our blog today!
References
Aridon P, Geraci F, Turturici G, D’Amelio M, Savettieri G, Sconzo G. Protective role of heat shock proteins in Parkinson’s disease. Neurodegener Dis. 2011;8(4):155-68.
Ayka A, Şehirli AÖ. The Role of the SLC Transporters Protein in the Neurodegenerative Disorders. Clin Psychopharmacol Neurosci. 2020 May 31;18(2):174-187.
Caldwell AB, Anantharaman BG, Ramachandran S, Nguyen P, Liu Q, Trinh I, Galasko DR, Desplats PA, Wagner SL, Subramaniam S. Transcriptomic profiling of sporadic Alzheimer’s disease patients. Mol Brain. 2022 Oct 12;15(1):83.
Campanella C, Pace A, Caruso Bavisotto C, Marzullo P, Marino Gammazza A, Buscemi S, Palumbo Piccionello A. Heat Shock Proteins in Alzheimer’s Disease: Role and Targeting. Int J Mol Sci. 2018 Sep 1;19(9):2603.
Cuddy LK, Prokopenko D, Cunningham EP, Brimberry R, Song P, Kirchner R, Chapman BA, Hofmann O, Hide W, Procissi D, Hanania T, Leiser SC, Tanzi RE, Vassar R. Aβ-accelerated neurodegeneration caused by Alzheimer’s-associated ACE variant R1279Q is rescued by angiotensin system inhibition in mice. Sci Transl Med. 2020 Sep 30;12(563):eaaz2541.
D’Orléans-Juste P, Akide Ndunge OB, Desbiens L, Tanowitz HB, Desruisseaux MS. Endothelins in inflammatory neurological diseases. Pharmacol Ther. 2019 Feb;194:145-160.
Dastidar SG, Pham MT, Mitchell MB, Yeom SG, Jordan S, Chang A, Sopher BL, La Spada AR. 4E-BP1 Protects Neurons from Misfolded Protein Stress and Parkinson’s Disease Toxicity by Inducing the Mitochondrial Unfolded Protein Response. J Neurosci. 2020 Nov 4;40(45):8734-8745. doi: 10.1523/JNEUROSCI.0940-20.2020.
Díaz-Martínez NE, Tamariz E, Díaz NF, García-Peña CM, Varela-Echavarría A, Velasco I. Recovery from experimental parkinsonism by semaphorin-guided axonal growth of grafted dopamine neurons. Mol Ther. 2013 Aug;21(8):1579-91.
Ding Q, Shults NV, Gychka SG, Harris BT, Suzuki YJ. Protein Expression of Angiotensin-Converting Enzyme 2 (ACE2) is Upregulated in Brains with Alzheimer’s Disease. Int J Mol Sci. 2021 Feb 8;22(4):1687.
Dumitriu A, Golji J, Labadorf AT, Gao B, Beach TG, Myers RH, Longo KA, Latourelle JC. Integrative analyses of proteomics and RNA transcriptomics implicate mitochondrial processes, protein folding pathways and GWAS loci in Parkinson disease. BMC Med Genomics. 2016 Jan 21;9:5.
Figueiredo-Pereira ME, Corwin C, Babich J. Prostaglandin J2: a potential target for halting inflammation-induced neurodegeneration. Ann N Y Acad Sci. 2016 Jan;1363(1):125-37.
Guzel, O., Tayler, H.M., Miners, S., Skrobot, O.A. and Kehoe, P.G. (2021), The role of angiotensin-converting enzyme-1 (ACE-1) in Alzheimer’s disease, and its association with ACE1 genotype. Alzheimer’s Dement., 17: e053569.
Jefferies WA, Price KA, Biron KE, Fenninger F, Pfeifer CG, Dickstein DL. Adjusting the compass: new insights into the role of angiogenesis in Alzheimer’s disease. Alzheimers Res Ther. 2013 Dec 19;5(6):64.
Jiang X, Wu Q, Zhang C, Wang M. Homoharringtonine Inhibits Alzheimer’s Disease Progression by Reducing Neuroinflammation via STAT3 Signaling in APP/PS1 Mice. Neurodegener Dis. 2021;21(3-4):93-102.
Kehoe PG, Wong S, Al Mulhim N, Palmer LE, Miners JS. Angiotensin-converting enzyme 2 is reduced in Alzheimer’s disease in association with increasing amyloid-β and tau pathology. Alzheimers Res Ther. 2016 Nov 25;8(1):50.
Kong W, Mou X, Liu Q, Chen Z, Vanderburg CR, Rogers JT, Huang X. Independent component analysis of Alzheimer’s DNA microarray gene expression data. Mol Neurodegener. 2009 Jan 28;4:5.
Kuja-Panula J, Kiiltomäki M, Yamashiro T, Rouhiainen A, Rauvala H. AMIGO, a transmembrane protein implicated in axon tract development, defines a novel protein family with leucine-rich repeats. J Cell Biol. 2003 Mar 17;160(6):963-73.
Laeremans A, Nys J, Luyten W, D’Hooge R, Paulussen M, Arckens L. AMIGO2 mRNA expression in hippocampal CA2 and CA3a. Brain Struct Funct. 2013 Jan;218(1):123-30.
Lee SH, Gomes SM, Ghalayini J, Iliadi KG, Boulianne GL. Angiotensin Converting Enzyme Inhibitors and Angiotensin Receptor Blockers Rescue Memory Defects in Drosophila-Expressing Alzheimer’s Disease-Related Transgenes Independently of the Canonical Renin Angiotensin System. eNeuro. 2020 Dec 22;7(6):ENEURO.0235-20.2020.
Louise R, Manon L, Vincent E, Andréanne L, Philippe B, Cyntia T, Bennett DA, Sébastien H, Frédéric C. Higher Angiotensin I Converting Enzyme 2 (ACE2) levels in the brain of individuals with Alzheimer’s disease. bioRxiv [Preprint]. 2023 Jan 18:2023.01.17.524254.
McCallion AS, Chakravarti A. EDNRB/EDN3 and Hirschsprung disease type II. Pigment Cell Res. 2001 Jun;14(3):161-9.
Moustafa AA, Chakravarthy S, Phillips JR, Gupta A, Keri S, Polner B, Frank MJ, Jahanshahi M. Motor symptoms in Parkinson’s disease: A unified framework. Neurosci Biobehav Rev. 2016 Sep;68:727-740.
Nelles DG, Hazrati LN. Ependymal cells and neurodegenerative disease: outcomes of compromised ependymal barrier function. Brain Commun. 2022 Nov 4;4(6):fcac288.
Perrine S. Endothelin 3. In: Enna SJ, Bylund DB and Elsevier Science (Firm). 2007. Xpharm : The Comprehensive Pharmacology Reference. Amsterdam: Elsevier.
Roll P, Massacrier A, Pereira S, Robaglia-Schlupp A, Cau P, Szepetowski P. New human sodium/glucose cotransporter gene (KST1): identification, characterization, and mutation analysis in ICCA (infantile convulsions and choreoathetosis) and BFIC (benign familial infantile convulsions) families. Gene. 2002 Feb 20;285(1-2):141-8.
Roy CN, Mak HH, Akpan I, Losyev G, Zurakowski D, Andrews NC. Hepcidin antimicrobial peptide transgenic mice exhibit features of the anemia of inflammation. Blood. 2007 May 1;109(9):4038-44.
Sánchez-Huertas C, Freixo F, Viais R, Lacasa C, Soriano E, Lüders J. Non-centrosomal nucleation mediated by augmin organizes microtubules in post-mitotic neurons and controls axonal microtubule polarity. Nat Commun. 2016 Jul 13;7:12187.
Sánchez-Mejías A, Fernández RM, López-Alonso M, Antiñolo G, Borrego S. New roles of EDNRB and EDN3 in the pathogenesis of Hirschsprung disease. Genet Med. 2010 Jan;12(1):39-43.
Shukla S, Tekwani BL. Histone Deacetylases Inhibitors in Neurodegenerative Diseases, Neuroprotection and Neuronal Differentiation. Front Pharmacol. 2020 Apr 24;11:537.
Solés-Tarrés I, Cabezas-Llobet N, Vaudry D, Xifró X. Protective Effects of Pituitary Adenylate Cyclase-Activating Polypeptide and Vasoactive Intestinal Peptide Against Cognitive Decline in Neurodegenerative Diseases. Front Cell Neurosci. 2020 Jul 17;14:221.
Tan C, Lu NN, Wang CK, Chen DY, Sun NH, Lyu H, Körbelin J, Shi WX, Fukunaga K, Lu YM, Han F. Endothelium-Derived Semaphorin 3G Regulates Hippocampal Synaptic Structure and Plasticity via Neuropilin-2/PlexinA4. Neuron. 2019 Mar 6;101(5):920-937.e13.
Wang YC, Huang AP, Yuan SP, Huang CY, Wu CC, Poly TN, Atique S, Kung WM. Association between Anemia and Risk of Parkinson Disease. Behav Neurol. 2021 Jul 7;2021:8360627. doi: 10.1155/2021/8360627. PMID: 34306250; PMCID: PMC8279865.
Xu K, Dai XL, Huang HC, Jiang ZF. Targeting HDACs: a promising therapy for Alzheimer’s disease. Oxid Med Cell Longev. 2011;2011:143269.
Yamamoto M, Götz ME, Ozawa H, Luckhaus C, Saito T, Rösler M, Riederer P. Hippocampal level of neural specific adenylyl cyclase type I is decreased in Alzheimer’s disease. Biochim Biophys Acta. 2000 Dec 15;1535(1):60-8..